Multiple strain analysis of Streptomyces species from Philippine marine sediments reveals intraspecies heterogeneity in antibiotic activities

Isolation of marine sediment-derived actinomycetes from west-central Philippines

The geographical sites identified in this study were evaluated to explore the actinomycete distribution in west-central Philippines (Fig. 1a). A total of 16 sediment cores were recovered from the 11 sampling sites and were processed in the laboratory using cultured-dependent actinomycete isolation (Supplementary Table S2). The seawater’s physicochemical conditions in all sampling sites were uniform with pH 7.0 and salinity ranging from 3.1 to 3.2. The characteristics of marine sediments and the distance of the actual collection sites identified using the given criteria varied per sampling location. The sediment characteristics vary from coarse to fine sand with mixture of broken corals and pebbles (Supplementary Table S3). Culture-dependent isolation revealed that actinomycete-like colonies and spores were observed in the minimal marine media after 30 to 60 days of incubation. Actinomycete isolates were repeatedly sub-cultured in enriched marine medium 1 (MM1) to obtain pure cultures as shown in Fig. 1b. Actinomycete growth observed in enriched media was white to gray aerial spores with brown to yellow mycelia or without diffusible pigmentations. Notably, there were strains that produced colonies with no diffusible pigmentations.

In this work, a total of 2212 pure actinomycete strains were isolated from marine sediments collected in 11 geographically distant sampling sites across the west-central Philippines (Fig. 1a). Actinomycete strains were highly abundant in Negros Occidental with 580 isolates (26%), followed by Southern Antique with 348 isolates (16%) and Southeastern Iloilo with 228 (10%). We recovered least actinomycete strains in Occidental Mindoro and South Central Visayas with only 94 and 81 strains, respectively.

Antibacterial activity profile of actinomycete strains

We assessed the antibacterial activities of actinomycete strains against a multidrug-resistant Gram-positive bacterium (S. aureus ATCC BAA-44) and three Gram-negative bacteria (E. coli ATCC 25922, P. aeruginosa ATCC 27853, and E. aerogenes ATCC 13048) using resazurin agar overlay assay and microbroth susceptibility assay as initial and confirmatory screenings, respectively. A total of 218 (9.9%) out of the 2212 actinomycete isolates have antibacterial activities in the initial screening as indicated by positive results or retained blue resazurin color in wells containing actinomycete overlaid with the test pathogens (Supplementary Fig. S2). The 218 active isolates were fermented to produce biomass for secondary metabolite extraction and to confirm their antibacterial activities by microbroth susceptibility testing. Figure 1c showed the confirmed 92 (4.1%) antibiotic-producing actinomycete strains. The majority of the strains (71 isolates) exhibited activity against Gram-positive S. aureus ATCC BAA-44. Thirty-nine (39) strains (42%) were active against E. coli ATCC 25922. Six strains were active against P. aeruginosa ATCC 27853, while all strains tested were inactive against E. aerogenes ATCC 13048 as indicated with no or less than 50% growth inhibition. Twenty-three (23) active strains targeted 2–3 test pathogens, while 69 active strains were only active against one test pathogen (Supplementary Table S4).

Phylogenetic diversity of multiple antibiotic-producing strains

The 92 active actinomycete strains were further identified and confirmed as Streptomyces species based on genomic analysis of their 16S rRNA and rpoB gene sequences. Comparison of 16S rRNA gene sequences (ranging from 1150 to 1500 nucleotides) and rpoB (700–995 nucleotides) gene sequences with their similar matches in the GenBank verified that the 92 active strains were closely related (97 to 100%) with 19 species under the genus Streptomyces. The nearly complete 16S rRNA and rpoB gene sequences were analyzed in a phylogenetic tree using maximum likelihood algorithms. The 16S rRNA and rpoB gene sequences of active strains reported in the present study were deposited in the GenBank nucleotide database (Supplementary Table S5).

Phylogenetic analysis revealed multiple strains with identical 16S rRNA gene sequences which clustered together into 13 major clusters (shown by the colored nodes in the tree) with high bootstrap values (> 90%) in the phylogenetic tree (Supplementary Fig. S3). Thirty-three strains (36%) were highly similar to S. parvulus presented in red circle. Followed by 12 strains with high similarity to S. enissocaesilis (light blue circle), 11 S. rochei strains (dark pink), six S. mutabilis strains (dark blue), five S. diastaticus strains (light green), four S. kunmingensis strains (green) and three S. geysiriensis strains (light orange). A phylogenetic analysis of rpoB gene sequence was conducted to provide a better resolution of the evolutionary relationship among strains within and between species supporting the taxonomic identity of the phylogenetically identical strains.

Similarly, the rpoB gene sequences phylogenetic tree (Fig. 2) showed 13 major clusters that were highly supported with bootstrap replicates > 90%, except for monophyletic clusters III (Streptomyces sp. strain DSD176) and X (Streptomyces sp. strain DSD1006) with low bootstrap replicates (< 90%). Active strains under cluster II and III were resolved S. diastaticus and S. geysiriensis lineages, respectively. Streptomyces sp. strain DSD176 under cluster II showed high similarity with S. geysiriensis based on its 16S rRNA gene sequences but separated from the other active S. geysiriensis strains found under cluster VII based on their rpoB gene sequences. Meanwhile, two active strains with high bootstrap support (98%) under cluster II diverged from the remaining active S. diastaticus strains in cluster VI, where cluster II being a sub-phyletic sister clade of active strains under cluster VI with similar 16S rRNA gene sequence matches in the GenBank. Interestingly, cluster VII was supported by 95% bootstrap replicates and comprised of sub-phyletic lines of 25 active strains belonging to three Streptomyces species: S. enissocaesilis (12 strains), S. rochei (11 strains), S. geysiriensis (two strains). Streptomyces parvulus (33 strains) and both S. enissocaesilis (12 strains), and S. rochei (11 strains) were the most dominant among the 19 bioactive Streptomyces species with the highest number of active strains. Interestingly, from the 92 active strains, we identified three novel species prospects: Streptomyces sp. strain DSD1006 from Northwestern Antique, Streptomyces sp. strain DSD3025 from Tubbataha Reefs and Streptomyces sp. strain DSD742 from Western Antique with 97–98% similarity with their reference match strains.

Antibiotic and anticancer variations within phylogenetic clusters

We observed varying antibiotic and anticancer strengths exhibited by multiple strains despite their similar taxonomic identities (Fig. 2). In 33 active strains that were closely related to Streptomyces parvulus, there were nine strains with more than 90% growth inhibition against a multidrug-resistant strain of S. aureus. The remaining 24 active S. parvulus strains have moderate antibacterial activities ranging from 50 to 89% growth inhibitions against S. aureus ATCC BAA-44, E. coli ATCC 25922 and P. aeruginosa ATCC 27853. Only 19 isolates have exhibited moderate cytotoxic activity ranging from 50 to 70% to ovarian carcinoma (A2780) (Supplementary Table S4). The antibacterial and anticancer activities in terms of % growth inhibition and the number of test pathogens inhibited by S. parvulus strains were shown in the outer ring heat map of phylogenetic tree. These variations in antibacterial and anticancer activities were also evident intraspecies found in other clusters in the phylogenetic tree.

Streptomyces strains nearly identical to S. rochei (11 strains), S. ennisocaesilis (12 strains) and S. geysiriensis (three strains) (Supplementary Fig. S3) revealed ambiguous phylogenetic affiliation. These intraspecies morphotypes have varying and remarkable antibacterial activities, as demonstrated in the purple-green portion of the tree and the outer ring heat map in Fig. 2. Furthermore, the intensity of antibacterial and anticancer activities and the intraspecific diversity of three dominant active Streptomyces species isolated in the west-central Philippines was presented in heat map clusters (Fig. 3a–c). In the heat map of S. parvulus active strains (Fig. 3a), seven different clusters were formed according to bioactivity against the test pathogens and ovarian carcinoma. The largest cluster formed nine strains was active against S. aureus ATCC BAA-44 and E. coli ATCC 25922 and all have cytotoxic activity ranging from 50 to 70% against ovarian carcinoma. Majority of the strains from this cluster were isolated from two neighboring sites, Negros Occidental and Negros Oriental (Fig. 1a). Streptomyces enissocaesilis (Fig. 3b) formed four clusters: the largest cluster composed of eight strains with antibiotic activity ranging from 80 to 100% growth inhibition against S. aureus ATCC BAA-44. Interestingly, no isolates from this cluster showed cytotoxic activity against ovarian carcinoma. Lastly, S. rochei strains (Fig. 3c) formed four different clusters. Two clusters formed, with four isolates each, were active against S. aureus ATCC BAA-44. One cluster, composed of DSD284, DSD1328, DSD501, and DSD1381 exhibited cytotoxic activity. The S. rochei strains from Northern Antique and Central Visayas are all found in this cluster. The differences in antibiotic and anticancer activities shown in this study demonstrate intraspecies variations within multiple strains despite they share nearly identical 16S rRNA gene and rpoB sequences. Intraspecies variations in antibiotic and anticancer activities were observed in S. diastaticus, S. kunmingensis, and S. mutabilis active strains (Supplementary Fig. S4A–C).

PCR-based detection of PKS and NRPS gene domains

We explored the PCR-based BGC profile and the extent of BGC diversity in multiple strains within and between species. The occurrence of NRPS, type-I and type-II PKS biosynthetic genes were assessed through PCR amplification approach^{37,38,39,40,41}. Visualization of strong unambiguous amplicons was shown in Supplementary Fig. S5–7. One primer set (A3F, A7R) was used to amplify adenylation (AD) domains of NRPS genes, while different primer sets were used to detect PKS type-I (KSMA-F, KSMB-R; KS-F, KS-R) and PKS type-II (KS1-F, KS1-F; KSα, KSβ; 540F, 1100R) genes in this study. Amplification homology of PKS domains by several primer sets used in this study may suggest occurrence of additional PKS gene clusters in the bioactive strains. The detection of amplified genes by the primers used in this study signified the biosynthetic potential of the bioactive strains and increased the likelihood to produce polyketides and non-ribosomal peptides with antibacterial and anticancer properties. The lack of amplification of ketosynthase (KS) and adenylation (AD) gene fragments may indicate absence of PKS and NRPS systems, or the BGC machineries responsible for the bioactivities of antibiotic-producing Streptomyces species have lower homology with the primers used.

As illustrated in Fig. 2, all 19 Streptomyces species have NRPS gene based on the positive amplification of the adenylation domain. However, we found variation in PKS type-I and -II genes between Streptomyces species. Majority of the species harbor both target regions in PKS I gene. Only Streptomyces sp. strain DSD3025 did not harbor the PKS-I gene. Strains identical to Streptomyces mutabilis do not have the KS domain conserved region targeted by KS-F and KS-R primers (Supplementary Fig. S6). Interspecies diversity was evident in the domain content of PKS-II as well. Interestingly, only four species, S. parvulus, S. griseorubens, S. carpaticus and S. xiamenensis, carried KSα and KSβ domain complex of PKS II at 500–600 bp, which is contrary to the expected 800–900 expected length of target region³⁹ (Supplementary Fig. S7B.1–B.2). This amplification of KSα and KSβ domain complex can be associated to the complexity and diversity of PKS type-II compounds. Strains with nearly identical 16S rRNA and rpoB gene sequences to S. kunmingensis, S. sedi, and Streptomyces sp. strain DSD3025 did not harbor all target domains of PKS II in their genomes. The PKS diversity may infer differences in the metabolite production, explaining the antibiotic activity variation observed between species. Multiple strain comparative analysis clearly showed that taxonomically related strains shared similar set of NRPS and PKS domain contents. Furthermore, identical strains isolated from geographically distant locations still harbored uniform NRPS and PKS domain contents.

Interestingly, we also observed variation in the amplification of PKS type I and type II genes within the strains of Streptomyces sedi and Streptomyces carpaticus (Fig. 2). The S. sedi strain DSD3018 showed the absence of PKS-1 gene KS domain region targeted by the KS-F and KS-R primers which is contrary to the other S. sedi strains, DSD3011 and DSD2987. Streptomyces carpaticus strain DSD331 harbors the 5’ portion of KSα domain of PKS-II gene; however, absence of this sequence was observed in S. carpaticus strain DSD274. These variations may suggest the presence of sequence variants in the same region of KS and KSα domains.

Isolation technique and carbon source utilization profiles

Culture-dependent techniques and carbon source composition of minimal marine media enhanced the isolation of active marine sediment-derived Streptomyces strains. We found that the heat-shock method (HSM) yielded higher recovery of active actinomycete strains (49 strains, 53%) as compared to dry stamp method (DSM) (43 strains, 47%) (Fig. 4a). However, active Streptomyces species were more diverse when processed using DSM (diversity index = 2.062) as compared to HSM (diversity index = 1.968).

The Venn diagram (Fig. 4b) showed that the 5 dominant species (> 1% abundance) can be recovered in both methods. We also noted that some species were exclusively recovered using a specific method. Eight species were exclusively recovered in DSM compared to 6 species in HSM. All active strains of S. kunmingensis, S. mutabilis, S. sedi and S. olivaceus were only recovered by HSM. Contrary, the active strains of S. carpaticus and S. harbinensis were only isolated using DSM.

Carbon source composition of marine minimal media, along with effective isolation techniques, was crucial for the isolation of antibiotic-producing Streptomyces. Among the five minimal marine media used, three carbon sources yielded high isolation rate: glucose, mannitol, and trehalose yielded nine species with 32 strains (35%), nine species with 18 strains (20%), and ten species with 18 strains (20%) respectively. However, only eight species (12 strains, 20%) and two species (12 strains, 20%) were isolated in raffinose, and starch-based media, respectively. In the contrary, high diversity indices were observed in trehalose (2.197), mannitol (2.0), and raffinose (1.979) (Fig. 5a). As expected, the starch-containing media had the lowest isolation rate and diversity as only two species (S. enissocaesilis and S. parvulus) were able to utilize a more complex carbon source.

Interestingly, co-isolation of species in different carbon sources was shown in the Venn diagram (Fig. 5b). Eight species can be recovered from at least two different carbon sources, whereas 11 species were exclusively isolated from a specific carbon source. Bioactive S. enissocaesilis and S. parvulus strains were recovered from all of the carbon sources utilized in this study. Active S. rochei were isolated in four media but not in starch-based media. In contrast, more exclusive species were isolated in trehalose with four species (S. harbinensis strains, Streptomyces sp. strain DSD3025, S. pseudogriseolus, and S. xiamenensis). Followed by glucose with three species (Streptomyces sp. strain DSD742, S. carpaticus and S. sedi), and two species each on mannitol- and raffinose-based media (Fig. 5b). The results indicated that diverse Streptomyces species preferred simple sugars-containing one or two sugar molecule as nutrient source compared to complex sugars.

Streptomyces abundance and diversity in geographical sampling locations

Bioactive Streptomyces species were widely distributed across the different sampling locations in west-central Philippines (Fig. 6a). Although Southern Antique, Negros Occidental and Negros Oriental have highest number of active strains isolated, we found that Southern Antique, Southern Iloilo, and Western Antique were the most diverse sampling sites (Fig. 6b). We have isolated the greatest number of antibiotic-producing Streptomyces species which were evenly distributed in Southern Antique. This indicates that Southern Antique is stable with many potential niches that can support highly diverse Streptomyces species.

Streptomyces parvulus emerged as the dominant antibiotic-producing species in this study. Out of the 33 active S. parvulus strains, the majority were isolated in Negros Occidental (11 strains) and Negros Oriental (ten strains). Although Negros Occidental and Negros Oriental have high abundance, its microbial community is highly dominated by one species, S. parvulus, supported by the low species richness, evenness, and diversity (Fig. 6b). This finding implies that these sites may have few potential niches that only a few species dominate. Bioactive S. enissocaesilis strains were recovered in four sampling locations only; specifically, Southern Antique (five strains), Western Antique (three strains), Southeastern Iloilo (three strains), and Northwestern Antique (one strain). Active S. rochei were isolated and evenly distributed in seven sampling locations, but were not present in Western Antique, Northwestern Antique, South Central Visayas, and Tubbataha Reefs. Notably, we observed that no bioactive S. parvulus, S. enissocaesilis, and S. rochei were isolated in Tubbataha Reefs, but antibiotic-producing S. cacaoi, S. psuedogriseolus and S. mutabilis strains were isolated only in Tubbataha Reefs marine sediments. Meanwhile, site-specific species such as S. sedi were recovered only in Occidental Mindoro. The isolation of site-specific species within genus Streptomyces can offer insight on the adaptive capacity of strains to inhibit locally coexisting resource competitors within and among these distinct locations.

Distribution of bioactive Streptomyces species at different sediment depths

We further investigated the distribution of antibiotic-producing Streptomyces strains along with the 110-cm sediment depth in different sampling sites. From the sediments that were partitioned according to depth with five categories at 25-cm increments, heterogeneous distributions of bioactive species were observed at deeper sediment with different dominant species in each depth (Fig. 6c). Although, S. parvulus, S. rochei, and S. enissocaesilis strains were ubiquitous in all depths, several species thrive abundantly in specific depths compared to other species. Streptomyces parvulus was the most dominant species in surface sediment. Meanwhile, Streptomyces rochei and S. enissocaesilis strains were more adapted in surface and sub-bottom sediments, respectively. Depth-specific Streptomyces strains were also identified as follows: Streptomyces sp. strain DSD3025 was isolated in subsurface sediments; S. mutabilis strains were abundant in the middle sediment layer; Streptomyces sp. strain DSD1006 and S. pseudogriseolus strain were recovered from sub-bottom sediments; and Streptomyces strain sp. DSD742, S. albus, S. sedi and S. xiamenensis strains were obtained from bottom sediments. High species diversity was positively correlated with increasing sediment depth, where surface sediments are known to be more prone to dispersal and wash-offs by environmental factors such as deep ocean currents⁴². Furthermore, the depth-specific species identified largely influenced the species richness in varying sediment depth.

Source: Ecology - nature.com