Biosynthetic Potential of Phylogenetically Unique Endophytic Actinomycetes from Tropical Plants

Abstract

The culturable diversity of endophytic actinomycetes associated with tropical, native plants is essentially unexplored. In this study, 123 endophytic actinomycetes were isolated from tropical plants collected from several locations in Papua New Guinea and Mborokua Island, Solomon Islands. Isolates were found to be prevalent in roots but uncommon in leaves. Initially, isolates were dereplicated to the strain level by ribotyping. Subsequent characterization of 105 unique strains by 16S rRNA gene sequence analysis revealed that 17 different genera were represented, and rare genera, such as Sphaerisporangium and Planotetraspora, which have never been previously reported to be endophytic, were quite prevalent. Phylogenetic analyses grouped many of the strains into clades distinct from known genera within Thermomonosporaceae and Micromonosporaceae, indicating that they may be unique genera. Bioactivity testing and liquid chromatography-mass spectrometry (LC-MS) profiling of crude fermentation extracts were performed on 91 strains. About 60% of the extracts exhibited bioactivity or displayed LC-MS profiles with spectra indicative of secondary metabolites. The biosynthetic potential of 29 nonproductive strains was further investigated by the detection of putative polyketide synthase (PKS) and nonribosomal peptide synthetase (NRPS) genes. Despite their lack of detectable secondary metabolite production in fermentation, most were positive for type I (66%) and type II (79%) PKS genes, and all were positive for NRPS genes. These results suggest that tropical plants from New Guinea and the adjacent archipelago are hosts to unique endophytic actinomycetes that possess significant biosynthetic potential.

Actinomycetes are saprophytes that play a significant role in the breakdown of organic matter into more readily assimilable nutrients. They are also well-known producers of a vast array of secondary metabolites, many of which have useful applications in human and veterinary medicine and agriculture. Over the last 75 years, numerous strains of actinomycetes have been isolated from various substrates collected worldwide and have been coaxed to express their extraordinary biosynthetic capacity in culture (6, 15, 41). Soil, which contains the greatest density and diversity of actinomycetes, has been the predominant and conventional source of actinomycete strains. More recently, however, scientists have found that other terrestrial substrates, such as leaf litter (32) and plants (30, 31), are potential sources of actinomycete biodiversity.

Endophytic actinomycetes, which can be recovered from healthy, surface-disinfected plant tissues, are known to produce a variety of bioactive metabolites with antibiotic, enzymatic, and plant growth-promoting or -inhibiting activities (17). There is evidence that actinomycetes are fairly prevalent in crop or garden plants, such as wheat (11), maize (13), and Chinese cabbage (21), and some of these isolates have been shown to produce bioactivity or have potential as biocontrol agents, especially against phytopathogenic fungi. However, information is scarce on the biodiversity, tissue distribution, and biosynthetic potential of endophytic actinomycetes from wild, native plants. Castillo et al. (10) isolated several bioactive, endophytic actinomycetes from native plant species growing in southern Patagonia, but they only recovered Streptomyces spp., and the isolations were limited to stems. More recently, in an effort to find an endophyte that produces maytansinoids, Zhu et al. (46) isolated more than 160 endophytic actinomycetes from multiple tissues of both wild and cultured specimens of the native, tropical tree Trewia nudiflora. Although they reported the prevalence of actinomycetes from each tissue, they provided relatively little information on the biodiversity of the isolates, having characterized only two of the cultures as Streptomyces spp.

The area encompassing the island of New Guinea and the adjacent archipelago including the Solomon Islands is one of the few remaining major tropical wilderness areas on earth and is also considered to be an important storehouse of biodiversity (24). As such, it seems likely that this region would be a potential source of unique and diverse microorganisms, especially plant-associated bacteria and fungi. Surprisingly, however, there is relatively little information on the isolation and characterization of microbes from this region. The aim of this study was to evaluate the tissue distribution, biodiversity, and biosynthetic potential of endophytic actinomycetes from native plants of coastal tropical forests from several locations in Papua New Guinea (PNG) and Mborokua Island, Solomon Islands.

MATERIALS AND METHODS

Plant collection and storage.

In October 2003, 47 plant samples from trees, herbs, palms, ferns, and club mosses were collected outside of the Kulili plantation on Karkar Island, Madang Province, PNG. In July 2005, 18 plant samples from trees, herbs, palms, ferns, club mosses, grasses, and sedges were collected at Watunou, Milne Bay Province, PNG, and 10 plant samples were obtained from trees, herbs, palms, ferns, and Pandanus spp. collected at Deka Deka Island, Milne Bay Province, PNG. In June 2006, 38 plant samples from trees, herbs, palms, ferns, grasses, and Pandanus spp. were collected at Mborokua Island (also known as Mary Island), Solomon Islands. If possible, herbaceous and young woody plants were preferably collected, to lessen the growth and contamination of endophytic fungi. All plant samples were stored at 4°C until they could be processed. Plant samples from Karkar Island and Deka Deka Island were processed within 30 h of collection. Plant samples from Mborokua Island were stored for 3 days, and those from Watunou Village were stored for 7 days before processing. All collection sites ranged from 0.02 to 0.6 km² in size.

Isolation of endophytes.

Plant samples were separated into tissues (leaves, stems, roots, etc.) by cutting with pruning shears or a scalpel. To remove soil and organic debris, tissues were thoroughly rinsed under running tap water. Each tissue was successively cut into 2- by 2-mm pieces with a scalpel, placed inside a tea strainer, and then immersed in a series of solutions for surface sterilization. For thin or herbaceous tissues such as leaves, stems, petioles, etc., the immersions were as follows: 70% ethanol for 1 min, 50% Clorox bleach (approximately 3% NaOCl) for 3 min, and 70% ethanol for 0.5 min followed by a rinse in sterile water. The procedure was the same for thicker or woody tissues such as twigs and roots, except that they were immersed in 50% bleach for 5 min. Four pieces of each tissue sample were aseptically transferred to the surface of arginine vitamin agar (25) supplemented with 3% soil extract (18) and 100 μg ml⁻¹ cycloheximide and 50 μg ml⁻¹ nystatin (postautoclaving) in 100- by 15-mm petri dishes. The soil extract was made by mixing 100.0 g organic humus (Jolly Gardener Products Inc., Poland, ME) in 100 ml of tap water. The suspension was autoclaved at 121°C for 30 min, transferred to a 1-liter centrifuge bottle, and centrifuged at 4,000 rpm for 20 min. The supernatant was filtered through a 0.2-μm cellulose nitrate filter unit (Nalgene).

The petri dishes were incubated at room temperature (approximately 23 to 25°C) for up to 8 weeks and checked periodically for the growth of actinomycetes. Actinomycetes growing from plant tissues were transferred with sterile inoculating needles to oatmeal agar (ISP medium 3; Himedia Laboratories, India) until they were pure.

Ribotyping of isolates.

Purified isolates were ribotyped with the RiboPrinter microbial characterization system (DuPont Qualicon, Wilmington, DE) as previously described (29). The restriction enzyme PvuII was used to digest the genomic DNA. The ribotypes were compared to each other by the similarity search tool available through the RiboPrinter software.

DNA extraction.

Cultures were grown in 5 ml of tryptone-yeast extract broth (ISP medium 1) or modified Bennett's broth (10.0 g liter⁻¹ dextrose; 0.77 g liter⁻¹ beef extract, desiccated; 1.0 g liter⁻¹ yeast extract; 2.0 g liter⁻¹ NZ amine A; pH 7.3) in 25- by 150-mm glass culture tubes at 28°C, 250 rpm, for 4 to 7 days. Cells were harvested by centrifugation at 4,000 rpm, washed twice with 5 ml TE25-sucrose (25 mM Tris-HCl [pH 8.0], 25 mM EDTA [pH 8.0], 0.3 M sucrose), resuspended in 1 ml of TE25-sucrose, transferred to 1.5-ml microcentrifuge tubes, and pelleted by centrifugation at 14,000 × g. The supernatant was discarded and the cell pellet resuspended in 300 μl TE25-sucrose containing 5 mg ml⁻¹ lysozyme and 0.2 mg ml⁻¹ RNase followed by incubation at 37°C for 1 h. For complete cell lysis, 50 μl of 10% SDS was added, and the mixture was vortexed briefly. To the lysis mixture 85 μl of 5 M sodium chloride was added. For precipitation of proteins, 400 μl of phenol-chloroform-isoamyl alcohol (25:24:1; Sigma) was added, and the mixture was vortexed until thoroughly mixed. After centrifugation at 14,000 × g for 10 min at 4°C, the upper aqueous phase was transferred to a new microcentrifuge tube. To precipitate the DNA, 500 μl of isopropanol was added and the solutions were mixed by inversion. If a precipitate of DNA was visible, the solution was carefully decanted and the DNA washed with 70% ethanol. If no DNA precipitation was visible, the mixture was centrifuged at 14,000 × g, 4°C, for 15 min to pellet the DNA, which was subsequently washed with 70% ethanol. After evaporating the ethanol, the DNA was resuspended in 50 to 100 μl of 10 mM Tris-HCl, pH 8.0.

Amplification and sequencing of the 16S rRNA gene.

The nearly complete 16S rRNA gene was PCR amplified using the primers 8FPL (5′-AGAGTTTGATCCTGGCTCAG-3′) and 1492RPL (5′-GGTTACCTTGTTACGACTT-3′) (28). Reactions were performed on a Biometra TGradient thermocycler in 50-μl volumes consisting of 0.5 μl genomic DNA, 1.0 μM each primer, 2.0 μl dimethyl sulfoxide (DMSO), 21.5 μl sterile water, and 25 μl of JumpStart REDTaq ReadyMix (Sigma-Aldrich). Parameters were as follows: initial denaturation at 95°C for 4 min followed by 35 cycles of denaturation at 95°C for 60 s, annealing at 48°C for 45 s, and extension at 72°C for 90 s. A final extension was done at 72°C for 5 min. To confirm amplification of the desired product, samples were electrophoresed on 1% agarose gels in 0.5× Tris-acetate-EDTA (TAE) buffer. PCR products were purified with the DNA Clean and Concentrator-25 kit (Zymo Research) and directly sequenced using the 16S rRNA primers described above. Sequencing was performed on an ABI 3730 sequencer with the ABI Prism DNA sequencing kit and BigDye terminators version 3.1 (Applied Biosystems).

16S rRNA gene phylogenetic analysis.

Sequence data were edited and assembled with Sequencher version 4.7. The operational taxonomic units were determined by BLASTN 2.2.2 similarity searching (1) of the sequences against each other and the GenBank database. The 16S rRNA gene sequences of related reference strains and other isolates were obtained from GenBank and aligned using ClustalX 1.8.1 (35). Distance analyses were conducted with TREECON 1.3b (38). Matrices were measured using the method of Jukes and Cantor (20). Trees were inferred by neighbor joining, and the confidence of the tree topologies was assessed by 1,000 bootstrap replicates. Pairwise sequence analyses were performed with the Gap program available in the GCG Wisconsin package (Accelrys, Inc.).

Fermentation of isolates and analysis of fermentation extracts.

First-stage seeds were grown in 7 ml of ATCC medium 172 in 25- by 150-mm glass culture tubes at 28°C with agitation. After 3 to 4 days, the contents of the first stage were transferred to 30 ml of fresh seed medium per 250-ml Erlenmeyer flask, and the second stage was incubated at 28°C with agitation for 24 to 48 h. The two production media were prepared as follows: medium A (15.0 g liter⁻¹ soluble starch, 1.0 g liter⁻¹ dextrose, 1.0 g liter⁻¹ Bacto peptone, 0.1 g liter⁻¹ MgSO₄·7H2O, 0.5 g liter⁻¹ KH₂PO₄, 1.0 g liter⁻¹ CaCl₂·2H2O, 50 mM morpholinepropanesulfonic acid buffer, 5.0 g liter⁻¹ Diaion HP20; pH 7.0) and medium B (12.5 g liter⁻¹ Nutrisoy soy flour, 2.0 g liter⁻¹ NZ amine A, 12.5 g liter⁻¹ dextrose, 1.0 g liter⁻¹ CaCO₃, 1.5 g liter⁻¹ NH₄Cl, 5.0 g liter⁻¹ Diaion HP20; pH 6.8). The fermentations were performed in 500-ml Erlenmeyer flasks containing 100 ml of medium and incubated at 28°C with agitation for 7 to 12 days. The strains were fermented in duplicate for each medium (two in 100 ml of medium A and two in 100 ml of medium B).

The fermentations (400 ml) for each strain were combined into a 1-liter centrifuge bottle, and 20 g of wet Amberlite XAD-7 slurry was added to bind compounds secreted into the supernatant. After mixing for 1 h, the resin and cell pellets were collected by centrifugation or filtration and extracted twice with 200 ml of methanol (total of 400 ml). One milliliter of each extract was dried in vacuo, resuspended in 100 μl of DMSO (10×), and analyzed by liquid chromatography-mass spectrometry coupled with diode array and evaporative light scattering detectors (LC-MS-DAD-ELSD) on an Agilent 1100 as described by Wagenaar (40). The concentrated extracts were tested for antimicrobial activity against Staphylococcus aureus 375 (ATCC 3538p) and Candida albicans 54 (CA300) and for DNA-damaging activity in a biochemical induction assay (BIA) with Escherichia coli BR513 (ATCC 33312) (44). For the antimicrobial testing, 20 μl of extract was tested per well in an agar diffusion assay, and the zones of inhibition were measured after 18 h of incubation at 37°C. For the BIA, 10 μl of extract was spotted on the surface of the agar and incubated for 3 h at 37°C before the overlay was applied.

PCR amplification of PKS and NRPS genes.

Polyketide synthase (PKS) type I gene fragments were PCR amplified from genomic DNA with two different sets of degenerate primers: ACP sense (5′-GASCTSGGSYTSGACTCSCTS-3′)/KS antisense (5′-SGASGARCASGCSGTGTCSAC-3′) (19) and K1F (5′-TSAAGTCSAACATCGGBCA-3′)/M6R (5′-CGCAGGTTSCSGTACCAGTA-3′) (5). Nonribosomal peptide synthetase (NRPS) adenylation domains were amplified with the degenerate primer pair A3F (5′-ACSTCSGGCWCSACCGGCCIGCCSAAG-3′)/A8R (5′-AGCTCSAYSCGSWRGCCSCGSAYCTTSACCTG-3′) (19). A portion of the KSα from PKS type II genes was amplified with the degenerate primers KSαF (5′-TSGCSTGCTTCGAYGCSATC-3′) and KSαR (5′-TGGAANCCGCCGAABCCGCT-3′) (23). All reactions were performed in 25-μl volumes consisting of 0.5 μl genomic DNA, 2.0 μM each primer, 1.0 μl DMSO, 10.0 μl sterile water, and 12.5 μl of JumpStart REDTaq ReadyMix (Sigma-Aldrich). For all primer pairs, PCR parameters were as follows: initial denaturation at 96°C for 4 min, followed by 35 cycles of denaturation at 96°C for 60 s, annealing at 60°C for 60 s, and extension at 72°C for 60 s. A final extension was done at 72°C for 5 min. A negative control using sterile resuspension buffer was included with each set of reactions. All PCR products were electrophoresed on 1% agarose gels in 0.5× TAE buffer. The GeneRuler DNA ladder mix (Fermentas) was used to size the bands from the PCR products.

Nucleotide sequence accession numbers.

DNA sequences were deposited with GenBank under accession numbers GQ924480 to GQ924582 (see Table S1 in the supplemental material).

RESULTS

Isolation of strains and distribution in plant tissues.

A total of 113 plants, representing a variety of species and types including very young to mature trees (36%), herbs (26%), ferns (12%), palms (11%), grasses and sedges (7%), Pandanus spp. (6%), and club mosses (2%), were randomly sampled from all four collection sites. The plant samples were processed into 256 tissue samples, including leaves (39%), roots (20%), stems (15%), petioles (15%), leaf veins (5%), bark (2%), twigs (2%), prop roots (1%), and tendrils (>1%). A total of 123 actinomycetes were isolated from 37 (33%) of the plants (Table 1). They were found to inhabit many different types of plants, including spore formers, such as ferns and club mosses, and flowering plants, such as grasses, sedges, vines, shrubs, and trees. However, actinomycetes were most prevalent in palms and Pandanus spp. in the families Arecaceae and Pandanaceae, respectively.

TABLE 1.

Locations, taxonomy, and number of isolates from plants that yielded actinomycetes

Location	Sample no.	Plant sample	Family	No. of isolatesa
Kulili Plantation, Karkar Island, PNG	PNG03-01	Calyptrocalyx sp.	Arecaceae	6
	PNG03-09	Ipomoea pes-caprae	Convolvulaceae	5
	PNG03-16	Indeterminate woody vine	Unknown	1
	PNG03-17	Zingiber sp.	Zingiberaceae	8
	PNG03-22	Piper sp.	Piperaceae	2
	PNG03-26	Lycopodium sp.	Lycopodiaceae	2
	PNG03-29	Calamus sp.	Arecaceae	3
Watunou, PNG	PNG05-10	Rottboellia sp.	Poaceae	1
	PNG05-13	Cyperus sp.	Cyperaceae	1
	PNG05-14	Pyrrosia sp.	Polypodiaceae	1
	PNG05-15	Sterculia sp.	Sterculiaceae	8
	PNG05-16	Indeterminate herb	Acanthaceae	6
	PNG05-17	Syzygium sp.	Myrtaceae	3
	PNG05-18	Calamus sp.	Arecaceae	6
Deka Deka Island, PNG	PNG05-20	Alstonia spectabilis	Apocynaceae	3
	PNG05-24	Euphorbia sp.	Euphorbiaceae	1
	PNG05-26	Areca sp.	Arecaceae	2
	PNG05-27	Pandanus sp.	Pandanaceae	2
	PNG05-28	Pandanus sp.	Pandanaceae	2
Mborokua Island, SB	SOL06-03	Caryota rumphiana	Arecaceae	2
	SOL06-04	Pandanus sp.	Pandanaceae	3
	SOL06-06	Piper sp.	Piperaceae	2
	SOL06-11	Tristiropsis acutangula	Sapindaceae	1
	SOL06-13	Piper sp.	Piperaceae	9
	SOL06-14	Pandanus sp.	Pandanaceae	1
	SOL06-15	Indeterminate herbaceous vine	Ranunculaceae	5
	SOL06-16	Impatiens sp.	Balsaminaceae	3
	SOL06-17	Indeterminate fern	Unknown	7
	SOL06-18	Syzygium sp.	Myrtaceae	1
	SOL06-21	Metroxylon sp.	Arecaceae	2
	SOL06-22	Indeterminate young tree	Unknown	1
	SOL06-24	Heterospathe sp.	Arecaceae	3
	SOL06-27	Derris trifoliata	Fabaceae	1
	SOL06-33	Indeterminate young tree	Unknown	2
	SOL06-35	Pandanus sp.	Pandanaceae	10
	SOL06-36	Indeterminate herbaceous vine	Unknown	2
	SOL06-38	Pandanus sp.	Pandanaceae	5

^aThere were 123 total isolates.

Every tissue type yielded at least one isolate, indicating that endophytic actinomycetes can colonize many different tissues throughout a plant. Although leaves were the most numerous tissue sampled, they had a very low isolate-to-sample ratio (Fig. 1), yielding only 8 (6%) isolates. In contrast, 72 (59%) isolates were recovered from roots, an isolate/sample ratio of 1.4:1. Prop roots, which are adventitious roots that grow from the stems of many Pandanus spp., yielded 12 isolates from only three samples, an isolate-to-sample ratio of 4:1. Bark was sampled infrequently but had an isolate/sample ratio of nearly 1:1, suggesting that bark is also a good source of endophytic actinomycetes.

FIG. 1.

Nine different tissues were sampled from 113 plants. A total of 123 actinomycetes were isolated from the 256 tissue samples that were processed. The bar graph shows the total number of samples from each tissue and the total number of actinomycetes isolated from each tissue.

Assessment of diversity by ribotype and 16S rRNA gene analyses.

The RiboPrinter, which generates a DNA fingerprint or ribotype based on the hybridization of an E. coli-based ribosomal operon probe to DNA fragments produced by the digestion of genomic DNA with a restriction enzyme (7), has been reported to be capable of subtyping bacteria below the species taxonomic level (27). To avoid fermenting duplicate strains for the fermentation extract analyses, ribotypes were generated for all of the isolates by using the RiboPrinter and compared to each other by similarity value searches with the RiboPrinter software and by personal observation (data not shown). In general, isolates sharing a similarity value of ≥90% were considered to be identical strains. However, if two isolates from different collection sites shared ≥90% similarity, they were considered to be different strains due to geographical separation. According to these criteria, 105 isolates (85%) were determined to be unique at the strain level. The majority of unique strains were isolated from plants collected on Mborokua Island, followed by Karkar Island, Watunou, and Deka Deka Island (Table 2). Only four pairs of isolates (two Amycolatopsis spp., two Planotetraspora spp., two Streptomyces spp., and two 16S clade 1 Thermomonosporaceae) from different collection sites shared the same ribotype (data not shown), suggesting that there is very little horizontal transmission of strains across close geographic locations.

TABLE 2.

Taxonomic distribution and origins of strains that were unique by ribotyping^a

Taxonomic group	No. of unique strains by site				Total no. of unique strains for taxonomic group	Total no. of strains for which fermentation extracts were analyzed	No. of strains with fermentation extracts that were:
	Kulili	Watunou	Deka Deka	Mborokua			Productiveb	Nonproductivec
Actinosynemmataceae
Lechevalieria	0	0	0	1	1	1	0	1
Lentzea	0	0	0	1	1	1	1	0
Micromonosporaceae
Actinoplanes	0	0	0	1	1	1	1	0
Dactylosporangium	1	0	0	0	1	1	1	0
16S clade 1	0	0	0	6	6	6	2	4
Pseudonocardiaceae
Amycolatopsis	1	0	0	2	3	3	3	0
Kibdelosporangium	0	0	0	3	3	3	3	0
Pseudonocardia	0	0	0	2	2	1	1	0
Streptomycetaceae
Kitasatospora	0	0	0	1	1	1	1	0
Streptomyces	8	7	1	11	27	26	20	6
Streptosporangiaceae
Microbispora	0	1	0	0	1	1	0	1
Nonomuraea	0	2	0	0	2	2	2	0
Planotetraspora	4	5	2	6	17	13	1	12
Sphaerisporangium	2	5	6	7	20	18	12	6
Streptosporangium	0	2	0	0	2	2	2	0
Thermomonosporaceae
16S clade 1	7	0	0	5	12	8	5	3
16S clade 2	1	0	0	4	5	3	1	2
Total	24	22	9	50	105	91	56	35

^aUnique strains shared <90% ribotype similarity. Evaluation of productivity of some of the unique strains was via fermentation extract analyses.

^bStrains with fermentation extracts exhibiting bioactivity and/or LC-MS profiles indicative of secondary metabolite production.

^cStrains with nonbioactive fermentation extracts and LC-MS chromatograms lacking any indication of secondary metabolite production.

Ribotyping can distinguish between strains, but the sensitivity of this method may preclude it from classifying an isolate to the genus or species level. To determine the operational taxonomic unit for each unique strain, 16S rRNA gene sequences were generated and compared to each other and to sequences in GenBank by BLAST analysis. The results (Table 2) showed that the strains could be classified into six families and at least 17 different genera. Several groups of strains, listed as 16S clade 1 or 2, within Thermomonosporaceae and Micromonosporaceae, shared relatively low sequence identities with any type strains in GenBank or shared similar identities with several type strains and could not be definitively classified with known genera. Most of the strains grouped into Streptosporangiaceae (40%), followed by Streptomycetaceae (27%), Thermomonosporaceae (16%), Micromonosporaceae and Pseudonocardiaceae (8% each), and Actinosynemmataceae (2%). Within the families, the most frequently isolated genera were the ubiquitous Streptomyces (26%), followed by Sphaerisporangium (19%), Planotetraspora (16%), Thermomonosporaceae 16S clade 1 (11%), Micromonosporaceae 16S clade 1 (6%), and Thermomonosporaceae 16S clade 2 (5%). Eleven genera, including Actinoplanes, Amycolatopsis, Dactylosporangium, Kibdelosporangium, Kitasatospora, Lechevalieria, Lentzea, Microbispora, Nonomuraea, Pseudonocardia, and Streptosporangium, were isolated less frequently and were represented by only one to three strains each.

The sequences of several strains from this study were found to be very similar or identical to the sequences of other endophytic strains in GenBank. For example, Amycolatopsis sp. ACT-0100, isolated from the stem bark of Derris trifoliata, and PENDO-1789, isolated from the tendril of a Piper sp., both shared 100% 16S rRNA gene sequence identity to two endophytic Amycolatopsis sp., one isolated from the root of a Taxus sp. (accession number EF690253) and the other from Maytenus confertiflora (accession number FJ830616), both from China. Streptomyces sp. ACT-0092 isolated from Tristiropsis acutangula and GSENDO-0567 isolated from Cyperus sp. shared 100% and 99.8% identity, respectively, with Streptomyces sp. isolate C-2 (accession number DQ221755), isolated from Chilotrichum diffisum from Patagonia. These results demonstrate that some endophytic actinomycetes have little or no host specificity and that certain species can have vast geographic distributions.

Phylogenetic analysis of Thermomonosporaceae, Micromonosporaceae, and Streptosporangiaceae.

To determine the phylogenetic positions of the strains within Thermomonosporaceae and Micromonosporaceae that could not be classified by BLAST analysis, neighbor-joining phylogenetic trees were created with the sequences of type strains and the most closely related strains by sequence identity in GenBank. A phylogenetic tree of Streptosporangiaceae was also created to determine the phylogenetic relationships between the endophytic and type strains of Sphaerisporangium and Planotetraspora.

The topology of the Thermomonosporaceae phylogenetic tree (Fig. 2) shows that all of the endophytic strains, Actinomadura alba, Actinomadura spadix, and the strains retrieved from GenBank, form a large clade (bootstrap value, 100%) that is distantly related to all other type strains in Thermomonosporaceae. Within the large clade, the separation of the endophytic strains into two smaller clades (clade 1 and clade 2) is supported by a relatively high bootstrap value (69%). Most of the endophytic strains are located within clade 1, which also includes all of the most closely related isolates recovered from GenBank and the type strain Actinomadura spadix. All of the sequences in clade 1 share >98% identity except for A. spadix (>97% identity). Clade 2 consists only of endophytic strains from this study and the recently described Actinomadura alba (43), which was isolated from a Chinese soil. All of the sequences within clade 2 are closely related (>97.5% identity). ACT-0125, ACT-0147, and PENDO-1810, which share >99.7% identity, are most likely the same species. Clade 1 shares a mean identity of 94.9% and clade 2 shares a mean identity of 95.4% with the type species of the genus Actinomadura (A. madurae DSM 43067^T). The relatively low identities to A. madurae and the formation of a monophyletic cluster distinct from the other type strains in Thermomonosporaceae suggest that the strains within clades 1 and 2, including A. alba and A. spadix, may belong to new genera of actinomycetes within Thermomonosporaceae.

FIG. 2.

Neighbor-joining phylogenetic tree of the Thermomonosporaceae rooted with Streptosporangium fragile (X89942). Bootstrap values greater than 50% are shown at the nodes and are based on 1,000 replicates. Sequences of type strains and related isolates were retrieved from GenBank, and accession numbers appear in parentheses. Bar, 0.02 substitutions per nucleotide. *, renamed Actinoallomurus (33).

The phylogenetic tree of Micromonosporaceae (Fig. 3) shows that some of the endophytic strains form a monophyletic cluster (clade 1) supported by a high bootstrap value (84%). Clade 1 also includes the uncultured bacterium clone 1-12D (accession number EU289478), which is the most closely related sequence in GenBank. Interestingly, the 16S rRNA gene of clone 1-12D is from the microbial-enriched DNA of the metagenomic library of the tree Mallotus nudiflorus (42). All of the sequences in clade 1 share >98% identity. Moreover, clade 1 shares a mean identity of 97.7% with each of the type strains Krasilnikovia cinnamonea (JCM 13252^T), Pseudosporangium ferrugineum (JCM 14710^T), and Couchioplanes caeruleus subsp. azureus (DSM 44103^T), all isolated from soil (2, 4, 34). Since the strains in clade 1 are equally related by sequence identity to the nearest type strains of three different genera and are all plant derived, they may represent a new genus in Micromonosporaceae.

FIG. 3.

Neighbor-joining phylogenetic tree of the Micromonosporaceae rooted with Streptomyces griseus (Y15501). Bootstrap values greater than 50% are shown at the nodes and are based on 1,000 replicates. Sequences of type strains and related isolates were retrieved from GenBank, and accession numbers appear in parentheses. Bar, 0.02 substitutions per nucleotide.

The phylogenetic relationship of the endophytic strains to several type strains from each genus within Streptosporangiaceae is depicted in Fig. 4. Twenty strains form a monophyletic cluster supported by a bootstrap of 78% with several type strains of the recently described genus Sphaerisporangium (3). The identities of the endophytic strains to the type species of the genus Sphaerisporangium (S. melleum DSM 44954^T) range from 96.7% (ACT-0132) to 99.3% (PENDO2167 and PENDO2169). Interestingly, the majority of the endophytic Sphaerisporangium spp. form several distinct clusters distant from the cluster of the Sphaerisporangium spp. type strains, suggesting that they are new species.

FIG. 4.

Neighbor-joining phylogenetic tree of the Streptosporangiaceae rooted with Nocardiopsis dassonvillei (X97886). Bootstrap values greater than 50% are shown at the nodes and are based on 1,000 replicates. Sequences of type strains and related isolates were retrieved from GenBank, and accession numbers appear in parentheses. Bar, 0.02 substitutions per nucleotide.

Sixteen of the endophytic strains cluster with the sequences of the three type species of Planotetraspora; all sequences within the clade share >98% identity. PENDO-1813 and PNG05-15FR-4 cluster very closely (85% bootstrap value) with P. thailandica, and their sequences share >99.8% identity, suggesting that they are all strains of the same species. However, all of the other endophytic strains in the Planotetraspora clade are not associated with any of the type strains and may represent several closely related but new Planotetraspora spp.

Analysis of fermentation extracts.

Ninety-one strains with unique ribotypes were fermented in two different media: one nutrient rich and the other nutrient poor (see Materials and Methods). To increase the probability of capturing secondary metabolites that may be toxic to the organism and/or produced in very low quantity, Diaion HP20 resin was included in the fermentation medium. For each strain, the flasks for both fermentation conditions were combined at harvest and the cell/resin pellet extracted as one sample. To assess bioactivity, the extracts were tested against three microorganisms: a methicillin-sensitive strain of Staphylococcus aureus for antibacterial activity, C. albicans for antifungal activity, and E. coli strain BR513 for the detection of compounds such as enediynes that are toxic and/or induce activity in the BIA (14). Almost half (46%) of the extracts exhibited a zone of inhibition of ≥7 mm against at least one of the test organisms; eight extracts were inhibitory to all three organisms. For the 49 extracts that did not display any bioactivity, chemical profiles were generated by LC-MS-DAD-ELSD, and the chromatograms were analyzed for UV peaks and mass ion signals that displayed properties characteristic for secondary metabolites but that did not elute early with the medium components or very late with fatty acids and other nonpolar compounds (see Fig. S1 in the supplemental material). The chemical profiles of 14 (28%) of the 49 nonbioactive extracts contained spectra indicative of secondary metabolites.

Strains that produced a fermentation extract with bioactivity or a nonbioactive extract with an interesting LC-MS chromatogram were considered to be “productive,” because there was some indication that secondary metabolites were produced under these limited fermentation conditions. Strains that produced nonbioactive fermentation extracts with flat or uninteresting chromatograms were considered “nonproductive.” A total of 56 strains (62%) (Table 2) were considered to be productive and 35 (38%) were unproductive. The most productive taxonomic groups at the family level were the Pseudonocardiaceae (100%) and the Streptomycetaceae (78%). The Streptosporangiaceae was the least productive taxonomic group (47%) due to the poor productivity of the Planotetraspora under these fermentation conditions. However, Streptosporangium, Nonomuraea, and Sphaerisporangium were quite productive. Among the potentially new genera, Thermomonosporaceae clade 1 contained more productive than nonproductive strains, but the converse was true for Thermomonosporaceae clade 2 and Micromonosporaceae clade 1.

Detection of PKS and NRPS genes in selected strains.

The two fermentation media used in this study may not have provided the nutrients and conditions needed to stimulate secondary metabolite production in all of the endophytic strains. Therefore, all of the nonproductive non-Streptomyces actinomycetes (29 strains) were evaluated for the potential to produce secondary metabolites by the PCR detection of polyketide synthase types I and II and nonribosomal peptide synthetase genes, by using primers that have been previously reported (5, 19, 23). The data show that the majority of the strains produced a band of the expected size for PKS type I (66%) and PKS type II (79%) genes and that NRPS genes were detected in all of the strains (Table 3). Although some of the pathways encoding these genes may not be functional, these data suggest that the nonproductive strains possess the genetic capacity to produce some secondary metabolites if cultivated under the proper conditions. Even the Planotetraspora, which are a very rare taxonomic group and have not been reported to produce any natural products, show some biosynthetic potential: PKS type I (42%), PKS type II (83%), and NRPS (100%).

TABLE 3.

Distributions of PKS and NRPS biosynthetic pathways in strains that were nonproductive by fermentation extract analysis

Taxonomic group	No. of strains evaluated	No. of strains with gene for indicated biosynthetic pathway
		PKS I	PKS II	NRPS
Actinosynemmataceae
Lechevalieria	1	1	0	1
Micromonosporaceae
16S clade 1	4	2	2	4
Streptosporangiaceae
Microbispora	1	1	1	1
Planotetraspora	12	5	10	12
Sphaerisporangium	6	5	6	6
Thermomonosporaceae
16S clade 1	3	3	2	3
16S clade 2	2	2	2	2
Total (%)	29	19 (66)	23 (79)	29 (100)

DISCUSSION

A total of 123 isolates were cultured from the surface-sterilized tissues of 37 of the 113 wild, native plants randomly sampled on Karkar Island, Watunou, and Deka Deka Island in Papua New Guinea and Mborokua Island, Solomon Islands. Because these actinomycetes were isolated from live, healthy plants, we considered them to be endophytic and not pathogenic. Actinomycetes were isolated from every tissue type sampled; however, roots (especially prop roots) and bark had the highest isolate-to-sample ratios, indicating that endophytic actinomycetes are most prevalent in those tissues from the plants we sampled. The prevalence of actinomycetes in roots compared to other tissues is not uncommon. Taechowisan et al. (31) examined 5,400 each of root, leaf, and stem tissues from 36 species of plants in Thailand and recovered 212 (64%) isolates from roots, 97 (29%) from leaves, and 21 (6%) from stems. Furthermore, Verma et al. (39) recovered more than double the number of all isolates from roots (55%) than from stems (24%) or leaves (22%) from 20 different Indian lilac trees, Azadirachta indica, growing in northern India. Since roots are the site of water and nutrient uptake for plants, they are an ideal substrate for actinomycete colonization. Also, endophytic actinomycetes may easily gain access to roots through the damaged epidermal layer created when lateral roots grow out from the existing root structure. Such is the case for the most well-studied endophytic actinomycete, Frankia, which forms root nodules in nonleguminous woody plants but also exists in soil in substantial numbers (16). Once inside the roots, it is conceivable that endophytic actinomycetes other than Frankia could benefit plants by fixing nitrogen without forming nodules (37) and/or producing antibiotics or siderophores to protect against infection by soil-borne pathogens.

All of the isolates were ribotyped to the strain level, so that only unique strains were fermented and analyzed for secondary metabolite production. Subsequently, all of the unique strains were characterized by 16S rRNA gene sequence analysis to assess the diversity of the strains and assign them to a molecular taxonomic unit. We found that they could be classified into six families and 17 possible genera, indicating that a large variety of actinomycetes can be endophytic within tropical plants. Although Streptomyces was the most frequently isolated genus, the Streptosporangiaceae was the most well-represented family, which consisted of five different genera: Sphaerisporangium, Planotetraspora, Nonomuraea, Streptosporangium, and Microbispora. Planotetraspora and Sphaerisporangium were especially prevalent among rare genera of actinomycetes and were consistently represented in plants from each sampling location, suggesting that they may form a mutual relationship with tropical plants. Although several groups have reported that Microbispora, which belongs to Streptosporangiaceae, is quite prevalent in a variety of plants (13, 21, 26), this is the first report of the occurrence of Planotetraspora and Sphaerisporangium in plants. It is also worth noting that the three Kibdelosporangium strains, ACT-0112 and ACT-0113 from the roots of a Heterospathe sp. and ACT-0114 from a prop root of a Pandanus sp., are the first from this rare genus to be reported as endophytic.

Phylogenetic analyses suggest that the endophytic strains in Thermomonosporaceae may represent new genera. Morphologically, strains from Thermomonosporaceae 16S clades 1 and 2 appear to be very similar to Actinomadura, with short chains of spores in curves, hooks, spirals, and/or pseudosporangia arising from short sporophores (J. E. Janso, unpublished data). However, the topology of the phylogenetic tree and low 16S rRNA gene identities between clades 1 and 2 and the type species Actinomadura madurae suggest that clades 1 and 2, including the type strains A. spadix and A. alba, form distinct genera. This is supported by the findings reported by Zhang et al. (45), who assessed the relationship between members of Thermomonosporaceae by 16S rRNA, 23S rRNA, and 16S-23S internal transcribed spacer analyses and suggested that Actinomadura spadix (JCM 3146^T) may merit independent genus status based on its distant relationship to other Actinomadura spp. (see the Addendum). In clade 1, two other strains, GMKU 931 (accession number 429322) from a plant in Thailand and actinomycete 7501 from Casuarina equisetifolia in Mexico (37), are also endophytic. Interestingly, actinomycete 7501 has the ability to fix nitrogen and contains a nifH (nitrogenase) gene homolog with 97% identity to nifH from a strain of Frankia (IPNCe16) also isolated from C. equisetifolia (37). The prevalence of our endophytic Thermomonosporaceae strains in plant tissues from Karkar and Mborokua Islands, in conjunction with the findings of Valdes et al. (37), strongly suggests that these actinomycetes beneficially associate with plants possibly through a mutualistic relationship.

Phylogenetic analyses also indicate that several of the endophytic strains in Micromonosporaceae may represent a new genus. Micromonosporaceae clade 1 is comprised of six endophytic strains isolated from plants on Mborokua Island and the uncultured bacterium clone 1-12D from the metagenomic library derived from Mallotus nudiflorus stem barks collected in Yunnan, China (42). Similar to clone 1-12D, all of our endophytic strains were isolated from either the stem bark of an unidentified tree or the prop root bark of a Pandanus sp., suggesting that this genus may specifically inhabit tissues outside of the vascular cambium. Morphologically, the endophytic strains form pseudosporangia and produce motile spores (Janso, unpublished) similar to Couchioplanes (34). Phylogenetically, they are closely related to three genera: C. caeruleus subsp. azureus (DSM 44103^T), K. cinnamonea (JCM 13252^T), and P. ferrugineum (JCM 14710^T). Thus, 16S rRNA gene phylogeny suggests that the strains in clade 1 form a distinct genus; however, extensive chemotaxonomic data are needed to finalize their taxonomic position.

To assess the biosynthetic potential of the endophytic strains, we used a polyphasic approach, including detection of bioactivity, chemical profiling, and PCR detection of secondary metabolic biosynthetic genes. According to the results, 62% of the strains had productive fermentation extracts with either antimicrobial and/or DNA-damaging bioactivity or LC-MS chromatograms with peaks characteristic for secondary metabolites. Furthermore, PKS I and PKS II genes could be detected from the genomic DNA of the majority of strains, and NRPS genes were present in 100% of strains with unproductive fermentation extracts. The productivity of the fermentation extracts and prevalence of biosynthetic genes detected demonstrate that all of the endophytic strains from this study have the potential to produce secondary metabolites under certain growth conditions and/or requirements. The unusually high detection of NRPS genes may in part be explained by the efficiency of the primers A3F-A8R, which were originally developed to amplify the NRPS adenylation domains of the mannopeptimycin biosynthetic gene cluster (22). However, it is also possible that many if not all of the endophytic actinomycetes produce siderophores, many of which are encoded by NRPS genes (12), to scavenge iron from the nutrient-poor tropical soils or to provide sequestered iron to their plant hosts (8). In addition, genome sequencing of several actinomycetes has revealed that soil-dwelling [Streptomyces coelicolor A3(2)], marine (Salinispora tropica CNB-440), and even endophytic actinomycetes (Frankia sp. CcI3) all harbor multiple secondary metabolite gene clusters (36), many of which have not been expressed in culture. Considering that endophytic actinomycetes live in close association with their hosts, there is a real possibility that genes involved in natural products biosynthesis could be exchanged via horizontal gene transfer between microbes and plants, resulting in production of plant-derived compounds by a microbe such as the paclitaxel-producing Kitasatospora sp. isolated from Taxus baccata in Italy (9). Perhaps even more exciting is the possibility that these and other endophytic actinomycetes may produce new compounds created by the hybridization and/or shuffling of plant and microbe secondary metabolic biosynthetic genes.

In summary, this study is the first comprehensive investigation into the tissue distribution, biodiversity, and biosynthetic potential of endophytic actinomycetes within wild, native tropical plants. The relatively small number of plants that we sampled from a few locations within Papua New Guinea and Mborokua Island, Solomon Islands, harbor a significant variety of endophytic actinomycetes, many which belong to rare genera and some of which may merit new genus status. Considering that these microbes produce bioactive and chemically interesting fermentation extracts or possess detectable biosynthetic genes, they may play some role in protecting their host plants from infection by pathogens and/or positively affect plant growth and development. At the very least, endophytic actinomycetes are a promising source of new and interesting natural products that may have value in agricultural research and drug discovery programs.

ADDENDUM

During preparation of the manuscript, Tamura et al. (33) formally described a new genus, Actinoallomurus, and transferred A. spadix to Actinoallomurus as the type strain of the genus. The formation of the new genus Actinoallomurus corroborates our phylogenetic results.