Genome Sequence and Composition of a Tolyporphin-Producing Cyanobacterium-Microbial Community

ABSTRACT

The cyanobacterial culture HT-58-2 was originally described as a strain of Tolypothrix nodosa with the ability to produce tolyporphins, which comprise a family of distinct tetrapyrrole macrocycles with reported efflux pump inhibition properties. Upon reviving the culture from what was thought to be a nonextant collection, studies of culture conditions, strain characterization, phylogeny, and genomics have been undertaken. Here, HT-58-2 was shown by 16S rRNA analysis to closely align with Brasilonema strains and not with Tolypothrix isolates. Light, fluorescence, and scanning electron microscopy revealed cyanobacterium filaments that are decorated with attached bacteria and associated with free bacteria. Metagenomic surveys of HT-58-2 cultures revealed a diversity of bacteria dominated by Erythrobacteraceae, 97% of which are Porphyrobacter species. A dimethyl sulfoxide washing procedure was found to yield enriched cyanobacterial DNA (presumably by removing community bacteria) and sequence data sufficient for genome assembly. The finished, closed HT-58-2Cyano genome consists of 7.85 Mbp (42.6% G+C) and contains 6,581 genes. All genes for biosynthesis of tetrapyrroles (e.g., heme, chlorophyll a, and phycocyanobilin) and almost all for cobalamin were identified dispersed throughout the chromosome. Among the 6,177 protein-encoding genes, coding sequences (CDSs) for all but two of the eight enzymes for conversion of glutamic acid to protoporphyrinogen IX also were found within one major gene cluster. The cluster also includes 10 putative genes (and one hypothetical gene) encoding proteins with domains for a glycosyltransferase, two cytochrome P450 enzymes, and a flavin adenine dinucleotide (FAD)-binding protein. The composition of the gene cluster suggests a possible role in tolyporphin biosynthesis.

IMPORTANCE A worldwide search more than 25 years ago for cyanobacterial natural products with anticancer activity identified a culture (HT-58-2) from Micronesia that produces tolyporphins. Tolyporphins are tetrapyrroles, like chlorophylls, but have several profound structural differences that reside outside the bounds of known biosynthetic pathways. To begin probing the biosynthetic origin and biological function of tolyporphins, our research has focused on studying the cyanobacterial strain, about which almost nothing has been previously reported. We find that the HT-58-2 culture is composed of the cyanobacterium and a community of associated bacteria, complicating the question of which organisms make tolyporphins. Elucidation of the cyanobacterial genome revealed an intriguing gene cluster that contains tetrapyrrole biosynthesis genes and a collection of unknown genes, suggesting that the cluster may be responsible for tolyporphin production. Knowledge of the genome and the gene cluster sharply focuses research to identify related cyanobacterial producers of tolyporphins and delineate the tolyporphin biosynthetic pathway.

INTRODUCTION

Tetrapyrrole macrocycles are known as the “pigments of life” and comprise the family of molecules including heme, siroheme, vitamin B₁₂, F₄₃₀, chlorophylls, and bacteriochlorophylls (1). The distinctions among the various tetrapyrroles reside in the nature of the substituents about the perimeter of the macrocycle, the degree and path of unsaturation in the macrocycle, and the nature of the centrally chelated metal (e.g., Fe, Mg, Co, or Ni) if present.

In 1992, Prinsep and coauthors (2–4) reported that a lipophilic extract of the terrestrial cyanobacterium Tolypothrix nodosa strain HT-58-2 reversed multidrug resistance in an in vitro cancer screen. The active anticancer agent was identified as a previously unknown type of tetrapyrrole macrocycle that they called tolyporphin A. The structure was subsequently refined by independent and lengthy chemical synthesis carried out by Kishi and colleagues (5–7). The structure of tolyporphin A is shown in Fig. 1, along with that of chlorophyll a, the chief photosynthetic pigment in cyanobacteria, for comparison. Since then, a family of tolyporphins (A to J, L, and M) which vary in the nature of the substituents in the two reduced rings has been identified (2–4, 8, 9); such substituents often are C-glycosides.

FIG 1.

Structures of tetrapyrrole macrocycles. The similarities in substitution patterns of tolyporphin A (7), chlorophyll a, and uroporphyrinogen III (a possible tolyporphin precursor) are shown in red.

Tolyporphins are remarkable for a host of reasons, including molecular structure, biological provenance, unknown physiological function, potential for pharmacological development, and likely biosynthesis via novel enzymes. These points are amplified as follows (1). Concerning molecular structure, tolyporphins differ from almost all known tetrapyrroles in the absence of carbon-containing substituents at two of the β-pyrrole positions (rings B and D), the presence of oxo groups in the reduced rings (rings A and C), and the presence of C-glycosides and hence a geminal-dialkyl group in the reduced ring. Moreover, the absence of a centrally chelated metal, while not unique, is unusual (2). Tolyporphins A to J, L, and M are bacteriochlorins (containing two reduced pyrrole rings), whereas chlorophylls, the photosynthetic pigments of cyanobacteria, are chlorins and hence contain only one reduced ring. The presence of the two oxo groups in tolyporphin A results in spectral properties more resembling those of chlorophyll a rather than bacteriochlorophyll a (10, 11), and yet the biosynthesis of bacteriochlorins generally is the province of anoxygenic photosynthetic bacteria, not cyanobacteria. Why bacteriochlorins are present in cyanobacteria is a mystery with possible evolutionary implications (3). Tetrapyrroles are known to perform myriad biological functions—ranging from catalysis to signaling to pigmentation—and are the subject of intense focus across scientific disciplines. In this regard, the identification of a strange tetrapyrrole with unknown in vivo function in an unlikely organism begs investigation. The identification of efflux pump inhibition properties (2, 4, 12) in an anticancer assay suggests that tolyporphins might provide new therapeutic lead compounds. Regardless of whether tolyporphin analogues can be developed for such applications, the structural features are compelling for synthetic and chemoenzymatic manipulation. The latter objective requires identification and exploitation of the as-yet-unknown enzymes for tolyporphin biosynthesis, the availability of which would profoundly enhance access to a class of molecules that are accessed synthetically with considerable difficulty (5–7). Only a few C-glycosyltransferases have been described (13–16), and hence the identification of such enzymes is of substantial interest and may transcend the specific objectives concerning tolyporphins (5). Finally, all established pathways to the tetrapyrrole “pigments of life” proceed via uroporphyrinogen III (Fig. 1) (17). A distinctive feature of uroporphyrinogen III is the pattern of substituents about the perimeter of the macrocycle: circumambulating from rings A to D, the substituents are AP-AP-AP-PA, where A is acetic acid and P is propionic acid. The acetic acid groups undergo decarboxylation to afford methyl (Me) groups, and the same pattern of reversed substituents in ring D is manifested in all known tetrapyrroles, as illustrated for MeX-MeX-MeX-XMe in chlorophyll a (where X describes various substituents derived from the propionic acid units). An analogous pattern of the reversed methyl-group location in ring D is observed in tolyporphin A (and all other tolyporphins), which constitutes a smoking gun for derivation via an as-yet-unknown biosynthetic pathway from uroporphyrinogen III.

The novelty of tolyporphins is not lessened by the discovery history. The HT-58-2 cyanobacterium was isolated from a soil sample in Nan Madol, Pohnpei, Micronesia (18), recently designated a UNESCO World Heritage Site (http://whc.unesco.org/en/list/1503). Upon initial isolation and in subsequent reports, the cyanobacterium has been described as a filamentous, oxygenic phototroph morphologically similar to isolates of Tolypothrix nodosa, and thus, this taxonomic nomenclature has persisted (2–4, 8). Indeed, the HT-58-2 cyanobacterium grows with an entangled, clumping growth morphology in liquid medium, thus resembling the Tolypothrix nodosa Bharadwaya group (“a knobby hairy ball of yarn”) (19). Few molecular systematic, genetic, or genomic studies have been described for this group of filamentous cyanobacteria, although the more distantly related Nostoc sp. strain PCC 7120 (Anabaena sp. strain PCC 7120) has the best-developed, model genetic system for filamentous cyanobacteria (20, 21).

In this paper, we report results concerning HT-58-2, which has been revived into culture. The studies concern first the unexpected nonaxenic composition of the culture, which has been examined by light, confocal fluorescence, and scanning electron microscopy (SEM) and by metagenomic analysis. Next, we report the determination of the complete, circular genome sequence of the HT-58-2 cyanobacterium. A significant finding concerns the identification of a putative gene cluster for the biosynthesis of tolyporphins. Taken together, these findings establish the molecular framework for investigation of tolyporphin biosynthesis and its regulation in the HT-58-2 cyanobacterial-bacterial community.

RESULTS

We use the following terminology for clarity throughout this report: “HT-58-2” refers to the original sample collection and the microbial population in total (cyanobacterium plus other bacteria, etc.) for what we show here to be a nonaxenic culture; “HT-58-2Cyano” refers to the only filamentous cyanobacterium in the sample, previously designated Tolypothrix nodosa; and “community bacteria” refers to the bacteria other than HT-58-2Cyano that are present in the culture.

Morphology of the HT-58-2 community.

Extensive reports on the structures and variations of tolyporphins (2, 3, 5–9, 12, 22) have not been accompanied by studies of the HT-58-2 microbial culture that produces tolyporphins, which we address here. The culture was first grown and extracted with an organic solvent (CH₂Cl₂-isopropanol). Examination of the extract by standard methods (mass spectrometry, absorption spectroscopy, and high-pressure liquid chromatography [HPLC] analysis [2, 3, 8, 9, 11]), along with comparison with authentic samples of tolyporphins, confirmed the presence of tolyporphins in the HT-58-2 culture. However, the appearance of bacterial colonies among the HT-58-2 green filaments grown on BG-11 agar plates showed the culture to be nonaxenic. Light microscopy of HT-58-2 grown in liquid BG-11 showed cyanobacterial filaments similar to those described for “T. nodosa” (19, 23), where branching growth nodes are observed frequently along the filaments (Fig. 2). However, the classic branching at heterocysts as seen in other Tolypothrix strains (23) is absent in HT-58-2Cyano cells. “Blebbing” at new filament branch points was observed. In BG-11o (which lacks nitrate), heterocysts were uncommon and irregularly positioned along the filament. HT-58-2Cyano cells contain granules dispersed throughout the cells upon growth in both BG-11 and BG-11o, which may be refractile cyanophycin bodies (24, 25); the genes for their synthesis have been identified in the genome (see below). Cells are contained within a membrane sheath that is curved at the termini with the presence of occasional calyptra. Apices are not tapered and consist of long cells. These images also provided the initial indication of attached and unattached bacteria among the filaments.

FIG 2.

Light microscopy of the HT-58-2 culture. A 14-day culture grown in BG-11 shows the development of branches at multiple points along a filament (A) and at a curve in a filament (B). Developing branches are indicated with arrows, and attached clumps of other bacteria are indicated with asterisks.

The morphology and composition of the HT-58-2 culture under soluble-nitrogen-replete (BG-11) and -depleted (BG-11o) conditions were examined by several microscopy methods. Representative results are shown in Fig. 3. Light microscopy images (Fig. 3A and E) show the morphology of HT-58-2Cyano filaments.

FIG 3.

Microscopic imaging of the HT-58-2 cyanobacterium-bacterial community. HT-58-2 cultures were grown in media with (BG-11) (A to D) and without (BG-11o) (E to H) nitrate. Arrows indicate examples of attached bacteria, and Het denotes heterocyst. Light microscopy images (A and E) at an ×1,000 magnification show the morphology of HT-58-2Cyano filaments. Fluorescent images of a 45-day culture show autofluorescence from HT-58-2Cyano displayed in magenta and green (B and F), whereas staining with FilmTracer FM 1-43 green biofilm cell stain (green) shows attached and nonattached bacteria. SEM images at ×1,000 (C and G) show entangled filaments, whereas images at ×7,000 (D) and ×10,000 (H) show bacteria attached to the filaments and an extracellular matrix coating the filaments. The scale is noted on each image.

Confocal fluorescence microscopy of a 45-day culture with 488-nm excitation showed autofluorescence from HT-58-2Cyano in two distinct spectral regions: 513 to 616 nm and 654 to 729 nm. The distinct autofluorescence spectra correspond to (at least) two pigment types. Moreover, the pigments were partitioned such that a given cell (even within a particular filament) contained only one of the pigment types (see Fig. S1 in the supplemental material). No autofluorescence in bacteria associated with the filaments was observed in this spectral region. Assignment of the two major autofluorescence spectral bands to particular pigments will require further study.

Cyanobacterial autofluorescence indicating the two spectral regions (513 to 616 nm, displayed in green; 654 to 729 nm, displayed in magenta) and cellular partitioning are shown in Fig. 3B and F. In these representative images of cultures grown in BG11 (Fig. 3B) and BG11o (Fig. 3F), similar variations in cell shape and pigment (spectral) partitioning were observed. The latter two images were obtained with cells stained with a biofilm stain to highlight the community bacteria (which do not give noticeable autofluorescence under the conditions examined here). Live bacteria attached to and surrounding the cyanobacterial filaments were observed (shown in green, Fig. 3B and F), regardless of growth in BG-11 or BG-11o.

Scanning electron microscopy (SEM) images clearly revealed close interactions between the cyanobacterial filaments and a bacterial community (Fig. 3C and G with expanded images in Fig. 3D and H, respectively). Diverse cell morphologies were attached and closely associated with the filaments irrespective of BG-11 or BG-11o. Several were rod-shaped and others were prosthecate, stalked bacteria. In addition, growth in BG-11o (Fig. 3H) showed the presence of an extracellular material coating the cyanobacterial filaments, with some community bacteria embedded in the sheath. The material also appeared attached to some of the community bacteria (Fig. S2). Extensive washing of HT-58-2 for SEM imaging appeared to have removed essentially all unattached bacteria, leaving only bacteria that were physically attached to the filaments.

Several approaches were used in attempts to grow HT-58-2Cyano in the absence of other community bacteria, and yet none yielded an axenic culture (data not shown). The treatments included the following: (i) multiple antibiotics, (ii) cycloserine/cycloheximide, (iii) arsenite, and (iv) anaerobic conditions (see the supplemental material). In most cases, the treatments yielded a reduction in filament biomass and a significant chlorosis, or there was an increase in culture turbidity associated with growth of other bacteria. Therefore, to date, HT-58-2Cyano could not be grown separately from one or more of the bacteria present in the associated microbial community.

Phylogenetic analysis of HT-58-2Cyano.

Using the HT-58-2Cyano 16S rRNA sequence and several related and signature filamentous cyanobacterial 16S sequences retrieved from GenBank (Table S1), two methods (SplitsTree [26] and MrBayes [27]) were used for phylogenetic tree construction. Figure 4 and Fig. S3 show that there were 5 clades with the sequences used and that highly similar sequences (identity of >99%, no gaps) grouped into one or a few clades (i.e., the Brasilonema CENA isolates). The clades of cyanobacteria evident included strains of Brasilonema, Nostoc, Rivulariaceae (Calothrix and Phyllonema), and Tolypothrix and were relatively monophyletic, although the Scytonema strains as currently classified are less so. The 16S rRNA sequence aligning the most closely with HT-58-2Cyano was strain CCIBt3568, which was classified as Scytonema (28), and yet the latter is outside the larger Scytonema clade. Both HT-58-2Cyano and CCIBt3568 aligned more closely with the large collection of CENA strains designated Brasilonema (29), rather than Tolypothrix. The HT-58-2 culture was originally designated Tolypothrix nodosa in the era prior to readily accessible genomic sequence data (2, 30).

FIG 4.

16S rRNA Bayesian inference phylogenetic analysis of HT-58-2Cyano. 16S rRNA sequences from position ACGGGTGAGT to ACCGCCCGTC (1,262 ± 2 bp) were aligned with ClustalW (79) and analyzed using Geneious with MrBayes plug-in (27) with Nostoc sp. PCC 7120 as the outgroup. Posterior probability values are shown next to the nodes. Highly similar 16S rRNA sequences (identity greater than 99%, no gaps) are represented in the same terminal branch (CENA114^a, SPC951; CENA347^b, CENA361, CENA366, CENA381, HA4187-MV1, UFV-E1, UFV-OR1; 1F-PS^c, 1f; YK-02^d, CCIBt3134; ACSSI 057^e, NIES 3756; PCC 7504^f, UTEX B 481; CENA373^g, CENA379; CENA341^h, CENA328, CENA326). Strains clustering together (clades) are color coded. Other strains are shown in dark gray. All accession numbers and taxonomic designations are available in Table S1 in the supplemental material.

Diversity in the HT-58-2 culture.

Microscopy and 16S rRNA universal primer amplicon sequencing confirmed the nonaxenic nature of the HT-58-2 culture. Therefore, HT-58-2 cultures were grown and used to produce MiSeq reads of the 16S rRNA V3 and V4 regions. These were grouped into 12 operational taxonomic units (OTUs) for BG-11 and BG-11o growth conditions. Cyanobacterial 16S rRNA sequences from the two growth conditions were comparable at 58% in BG-11 and 57% in BG-11o (Fig. 5; Table S2), all of which aligned exactly with the HT-58-2Cyano sequence, confirming that there is only one cyanobacterium in the culture.

FIG 5.

OTUs of HT-58-2 cultures grown in BG-11 or BG-11o. 16S rRNA V3-V4 regions from a 20-day culture grown in light were amplified and sequenced using Illumina MiSeq. The identity of the binned OTUs was determined by BLASTN and One Codex. The absence of nitrate (BG-11o) slightly alters the distribution of HT-58-2 community bacteria. One Porphyrobacter species comprises 97% of the Erythrobacteraceae OTUs.

Other OTUs in the cultures were dominated by the bacterial family Erythrobacteraceae, representing 35% and 24% of all OTUs in BG-11 and BG-11o, respectively. The remaining OTUs represented Sphingomonadaceae at 2.8% and 5.6%, Proteobacteria at 1.9% and 7.6%, and Alphaproteobacteria at 1.2% and 4.3%, respectively, and unknown bacteria at 1% in both BG-11 and BG-11o. The remainder represented only 0.69% and 0.32% of all OTUs, respectively. The dominant OTU within the Erythrobacteraceae family aligned with the 16S rRNA of Porphyrobacter sp. (31), which accounted for 97% of the reads within that group under both growth conditions. Additional studies of the effects of nutrients on the community bacterial population are needed, although the Porphyrobacter sp. is the dominant component associated with HT-58-2Cyano.

HT-58-2Cyano genome.

The presence of a microbial community in the HT-58-2 culture presented challenges in obtaining cyanobacterial DNA suitable for genomic sequencing. In this regard, filamentous cyanobacterial genomes are notorious for being difficult to assemble (32). Ultimately, a short growth period in the presence of 10% dimethyl sulfoxide (DMSO), followed by extensive washing of filaments with 10 to 20% DMSO and lysis of the resulting filaments under liquid nitrogen (cell preparation method 2 in Materials and Methods), yielded cyanobacterial genomic DNA that could be assembled from PacBio SMRT cell sequencing reads (finishing and closing details are summarized in Table S3). The complete HT-58-2Cyano genome is circular, comprised of 7,846,907 bp with an average GC content of 42.6%. There are 6,177 coding sequences (CDSs), 12 rRNAs (4 operons), 44 tRNAs, 6 clustered regularly interspaced short palindromic repeat (CRISPR) arrays, and 4 noncoding RNAs identified to date. The 344 pseudogenes identified require refined annotation. Figure 6 shows the map of the genome. Genes are distributed on both strands, and there is not a strong GC skew or single DNA replication origin inflection point. tRNAs are located on both strands, whereas the four rRNA operons are only on the reverse strand within the first two-thirds of the genome. Each of the rRNA operons has an above-average G+C content, although there are other localized GC-rich sites throughout. The rRNA operon sequences are all identical; the 16S rRNA matches the initial sequence obtained by PCR and used in the phylogenetic tree (Fig. 4) and is most similar to the 16S rRNA of Scytonema sp. strain CCIBt3568 (97% identity over query coverage of 99%; E value = 0.0).

FIG 6.

Map of the HT-58-2Cyano genome. The dsDNA genome contains 6,581 genes distributed evenly throughout both strands of the 7.85-Mbp genome (outer green rings). tRNAs (44) occur on both strands (first gray ring), whereas rRNA operons (4) are only on the reverse strand (second gray ring). The G+C content shows peaks at the rRNA operons and at other specific sites in the genome (first gold/purple plot). The GC skew (innermost gold/purple ring) does not show a strong strand bias that indicates major leading strands of replication or a single origin of DNA replication. The genome is oriented with dnaA on the positive, clockwise strand starting 100 bp from position 1. DNAPlotter (89) was used to generate the map.

CDS products encoded by the HT-58-2Cyano genome predict 350 genes associated with the biosynthesis of cofactors, vitamins, prosthetic groups, and pigments, including 67 associated with photosynthesis (Fig. 7). Notable highly represented cell constituent categories are those for carbohydrates (448 genes); amino acids and derivatives (355); fatty acids, lipids, and isoprenoids (185); cell wall and capsule (169); and RNA (142), whereas those for notable metabolic processes include translation (321), stress responses (151), respiration (119), and transport (116).

FIG 7.

Functional category assignments of CDSs in the HT-58-2Cyano genome. Default RAST annotation parameters were used. Diverse functions were identified, and yet proteins of unknown function (51%) were the largest category (shown as hypothetical).

There are 3,168 hypothetical genes, 51.3% of all CDSs (Fig. 7). The genome was complete and closed as determined by the following: (i) the assembly analytics, (ii) PCR and amplicon sequencing across the ends of the assembled genome, and (iii) the identification of essential “housekeeping” genes distributed throughout the large chromosome (Table S4). Some 86% of so-called housekeeping genes were identified in the HT-58-2Cyano genome, similar to the analysis conducted by Leão et al. (33) for the filamentous marine cyanobacterium Moorea.

All genes for conversion of glutamic acid to heme, chlorophyll a, and phycocyanobilin were identified, as well as most of the genes required for cobalamin biosynthesis (Table S5). The tetrapyrrole biosynthetic pathway inferred to occur in HT-58-2Cyano is diagrammed in Fig. 8. Additional genes encoding six of the required eight enzymes for conversion of glutamic acid to protoporphyrinogen IX were found in a single cluster (Fig. 9). The genes in the cluster (bp 2586793 to 2609884) include hemA, hemL, hemB, hemC, hemE, and two hemF genes, whereas the two missing genes are gltX and hemD. Also included in this cluster are genes identified as encoding a putative glycosyltransferase, cytochrome P450-like enzymes, and a flavin adenine dinucleotide (FAD)-binding protein. The presence of the tetrapyrrole and additional genes implies a role as a putative gene cluster for tolyporphin biosynthesis.

FIG 8.

Inferred tetrapyrrole biosynthesis pathway in HT-58-2Cyano. All genes shown were identified in the HT-58-2Cyano genome, and those shown in bold are also within the cluster of hem genes at bp 2586793 to 2609884. Complete pathways to chlorophyll a and phycocyanobilin have been identified, whereas the pathway to cobalamin is incompletely identified.

FIG 9.

Putative tolyporphin biosynthetic gene cluster in the HT-58-2Cyano genome. Genes (hem) for enzymes associated with tetrapyrrole biosynthesis are shown in green, other identified genes are shown in dark gray, hypothetical (hyp) genes are in light gray, and an asterisk indicates that only the domain was identified. The start and end genome coordinates of the cluster are numbered.

Well-recognized genes for nitrogen fixation processes were also identified and tabulated (Tables S6 to S8). Several of the nif and alternative nitrogen fixation pathway (vnf and anf), cyanophycin (cph), and heterocyst formation (het) genes are present in the genome. Four genes for the photoprotective orange carotenoid protein (ocp) were identified (Table S9).

antiSMASH (34, 35) analysis of HT-58-2Cyano revealed 64 possible genetic regions for pathways and secondary metabolite biosynthesis (Table S10), with apparent clusters for the biosynthesis of welwitindolinone, pseudopyronine A or B, nostopeptolide, cepacian, heterocyst glycolipids, cylindrocyclophane, lipopolysaccharides, aeruginoside, nostophycin, puwainaphycins, anatoxin, chondrochloren, shinorine, and guadinomine.

In addition to the main chromosome, one contig assembled into an HT-58-2Cyano plasmid (pHT582-1; Table S3), which was confirmed by PCR and sequencing. pHT582-1 is circular, is 37,028 bp in length (43.5% G+C), and contains 32 predicted genes. Twenty-four of these genes encode hypothetical proteins. pHT582-1 is predicted to encode ParA and ParB, integrase, AAA family ATPase, a transposase, a transposase DNA invertase, a site-specific DNA-methyltransferase, and a restriction endonuclease subunit R.

A second contig assembled as another plasmid (pHT582-2; Table S3) consists of 26,382 bp (43.6% G+C) and has 17 genes. Assembly of this plasmid presented added challenges due to the presence of a phage-like mobile element that is also found in five copies on the chromosome. Annotation of pHT582-2 revealed 9 hypothetical genes and other genes that encode mobile element proteins, ParAB, primase, DNA invertase, DNA integrase, transposase, phage tail protein, and a GTPase. Overall, the two plasmids are different but have three blocks ranging in size from 2 to 4.8 kbp of similar sequence; the rest of the plasmid DNAs share no similarity. Both plasmids exist at low copy number, and neither presents genes that appear to be involved in tetrapyrrole biosynthesis. Maps of these two plasmids are provided in Fig. S5.

DISCUSSION

Our initial long-term objectives include delineating the biosynthetic pathway of tolyporphin production, repositioning the enzymes for use in chemoenzymatic syntheses of tetrapyrroles, and understanding the in vivo function of tolyporphins. Below, we discuss the morphology of the HT-58-2 cultures and the microbial diversity as measured by 16S rRNA sequence data and finish with a brief summary of the cluster of hem and other genes that may be involved in tolyporphin production.

Morphology and phylogeny of HT-58-2.

HT-58-2Cyano is a filamentous cyanobacterium with occasional heterocyst-like cells when grown in nitrate-free medium (BG-11o), with filament morphology that is similar to that of Tolypothrix spp. wherein single filaments branch off a main filament accompanied by occasional false branching (http://www.cyanodb.cz). However, HT-58-2Cyano has an even greater morphological similarity to Brasilonema spp. (http://www.cyanodb.cz), which is consistent with our phylogenetic analysis of the HT-58-2Cyano 16S rRNA. Indeed, HT-58-2Cyano does not align with the Tolypothrix clade but aligns closely with the proposed Brasilonema clade, although the strain closest to HT-58-2Cyano is Scytonema CCIBt3568 (Fig. 4). The genus Brasilonema was described only a decade ago as a genus apart from that of Scytonema (29), and other Brasilonema species have since been identified (36). Accordingly, it appears that the name “tolyporphin” is a doubly flawed portmanteau—the strain is not likely a Tolypothrix, and the dioxobacteriochlorin macrocycle is not strictly a porphyrin. Our results recapitulate the complexity of cyanobacterial systematics (37) and suggest that a revision of the assigned “T. nodosa” nomenclature for the HT-58-2 cyanobacterium is warranted. The names for the macrocycles, however inappropriate, remain tolyporphins.

The analysis of 16S rRNA for cyanobacteria is known to have some limitations of reliability (38). Regardless, because of the absence of extensive genomic data for diverse cyanobacteria, 16S rRNA has been employed here for phylogenetic analysis of HT-58-2Cyano. The genome of HT-58-2Cyano was uploaded to AmphoraNet (https://pitgroup.org/amphoranet/) for phylogenetic analysis on the basis of 31 marker proteins (39), and yet no phylotype more specific than Nostocaceae was identified.

Fluorescence microscopy determined that there is autofluorescence in HT-58-2Cyano in two different wavelength regions: 513 to 616 nm and 654 to 729 nm. Emission in the shorter (green) wavelengths from native pigments has been reported for other cyanobacteria (40) and includes components such as R-phycoerythrin, a phycobiliprotein. Chlorophylls typically fluoresce at the longer (red) wavelengths (11). Fluorescence of these two components occurs in different cells within the filament, with no spatial overlap seen. This result indicates pigment partitioning or compartmentalized expression in HT-58-2Cyano, which has been observed in cyanobacteria within halite rocks (41). Further study is required to identify the pigment(s) producing the autofluorescence in the 513- to 616-nm wavelengths. In this regard, a recent photophysical study of tolyporphin A showed that the absorption and fluorescence spectral properties (including fluorescence spectrum and intensity) were largely unchanged across solvents of diverse polarity (11).

The presence of optically dense granules in HT-58-2Cyano is consistent with observations on other cyanobacteria (42, 43), the composition of which is generally unknown but may be refractile cyanophycin, given that the genes encoding synthesis and degradation of this nitrogen storage compound (24, 25) were identified in the genome. An extracellular material covering HT-58-2Cyano filament sheaths was clearly evident when cultures were grown in BG-11o (lacking soluble nitrogen), although it was of unknown composition. Extracellular matrix has been reported for other cyanobacteria, with Wollea saccata producing proteoglycan (44) and Nostoc verrucosum producing polysaccharide (45).

The images and variety of cell morphologies seen here in close interaction with HT-58-2Cyano filaments suggested the presence of a diverse bacterial community (Fig. 3). Such diversity is consistent with that of other nonaxenic cyanobacteria from different environments (46–50), where both free and attached bacteria have been associated with cyanobacterial filaments (47, 50). Although the explicit identity of bacteria firmly attached in the SEM images (Fig. 3) is unknown, the 16S rRNA diversity and genomic sequencing suggest that Erythrobacteraceae, and Porphyrobacter species in particular, dominate. Diverse communities associated with cyanobacteria have been reported in cyanobacteria isolated from different environments such as marine sediments (51) and aquatic environments (52). Direct attachment of bacteria to cyanobacteria producing natural products of interest has also been reported (53). Attempts to isolate HT-58-2Cyano from HT-58-2 were unsuccessful, suggesting that either HT-58-2Cyano is highly sensitive to the strategies used or is simply nonculturable in the absence of one or more community bacteria that are sensitive to the treatment used. The mechanisms of attachment and the basis for the interactions with HT-58-2Cyano filaments are unknown. Further work is needed to yield insights into these processes.

Microbial 16S rRNA diversity.

The initial determination of the community composition using an Illumina MiSeq 16S sequence survey showed that the major bacterial constituents, whether grown in BG-11 or in BG-11o, were roughly the same. A greater number of OTUs was found in the BG-11 culture than in the BG-11o culture, where the diversity with nitrogen appears to be higher.

The major components of the HT-58-2 bacterial community are also present in associations with other cyanobacteria, i.e., 88% of bacteria found with Microcystis sp. were identified as Proteobacteria (54); Erythrobacter and Porphyrobacter have been identified with the nitrogen-fixing Odular sp. (51); Sphingomonadaceae, Novosphingobium, and Sphingomonas have been found with Nostoc (48); and both Caulobacterales and Sphingomonadales have been found with the alga Botryococcus braunii (55). This diversity survey of HT-58-2 can be compared with other metagenomic studies of cyanobacteria, where predominating microorganisms are of the Alphaproteobacteria class. Porphyrobacter spp. have been observed in association with cyanobacteria (51). Further investigation is needed to determine the role of these bacteria in the growth of HT-58-2Cyano and possible production of tolyporphins. Cyanobacteria have been shown to play a role in bacterial community selection (47, 56), and vice versa, and the bacteria can play a role in modulating gene expression of the cyanobacterium (57). Perspective concerning the present metagenomic survey of the HT-58-2 cyanobacterium-microbial community will require comparable surveys of members of the proposed Brasilonema clade. Comparisons with other complex microbial communities and holobionts also are suggested (58, 59).

HT-58-2 cyanobacterial genome.

Obtaining a complete genome sequence of HT-58-2Cyano proved to be challenging. Others also have struggled to generate a complete genome from a nonaxenic cyanobacterium (50). The method described here (DMSO treatment and DNA isolation method 2) yielded sufficient high-quality, long cyanobacterial DNA to complete the genome sequence of HT-58-2Cyano.

At 7.85 Mbp, the HT-58-2Cyano genome represents one of the larger closed cyanobacterial genomes in GenBank. The genome is comparable in size to those of other filamentous nitrogen-fixing strains that were proposed to have undergone genome expansion from a non-nitrogen-fixing progenitor (60). Similarly large genomes—Nostoc punctiforme ATCC 29133 (8.2 Mbp) and Arthrospira platensis NIES-39 (6.8 Mbp) (http://genome.microbedb.jp/cyanobase)—are two examples. HT-58-2Cyano appears to be closely aligned with the proposed Brasilonema clade, and yet to our knowledge, no fully sequenced genome of a member therein has been reported. The HT-58-2Cyano genome shows no apparent GC skew, which is consistent with other cyanobacteria such as Anabaena strain ATCC 29413 (61), Arthrospira (Spirulina) platensis NIES-39 (62), Gloeobacter violaceus PCC 7421 (63), and Synechococcus elongatus PCC 6301 (64). The large number of hypothetical proteins and the 118 mobile element proteins or transposases are consistent with other cyanobacterial genomes (65–67). The presence of a type IV pilus component (PilA) and type II secretion component PulG, where PilA has also been shown to play a role in biofilm formation (68), suggests possible roles in biofilm or extracellular matrix formation by the cyanobacterium. A gene cluster for far-red light photoacclimation (FaRLiP) is known in certain cyanobacteria, particularly those that live in environments enriched in far-red light (69, 70), but was not found in the genome of HT-58-2Cyano. The exact origin of the HT-58-2 culture on Nan Madol, a sprawling archeological site, is unknown. Moreover, the present culture has been maintained under unspecified conditions for >25 years, and hence, comparison with the original strain does not appear possible.

In addition to the tetrapyrrole biosynthesis genes dispersed throughout the genome, a cluster of genes for almost all of the early steps of tetrapyrrole biosynthesis (from glutamic acid to protoporphyrinogen IX) also was observed. The cluster also contains 10 putative genes and one hypothetical gene. This putative tolyporphin gene cluster is more reminiscent of secondary metabolite biosynthesis than tetrapyrrole biosynthesis, where the latter directs singular product synthesis (e.g., enzyme cofactors) rather than a promiscuous collection of structurally related compounds as in tolyporphins A to M. The genes for tetrapyrrole biosynthesis are known to be dispersed throughout the genome in unicellular cyanobacteria (71). We examined the filamentous cyanobacteria Nostoc punctiforme PCC 73102 (accession no. NC_010628.1) and Anabaena variabilis ATCC 29413 (accession no. CP000117.1) and also found dispersal throughout the genome of tetrapyrrole biosynthesis genes. The presence of the gene cluster suggests that there could be coordinated regulation of these genes as seen for antibiotics and other secondary metabolites. The substituent pattern in tolyporphins suggests derivation from uroporphyrinogen III, although the point of biosynthetic departure for formation of tolyporphins is not known. Other genes in the cluster appear to encode proteins with glycosyltransferase, P450, and FAD-binding domains (Fig. 9), all of which could have roles in adding, modifying, and removing substituents of the tolyporphin macrocycle. The availability of the annotated genome provides a framework for more systematic searching for, and characterization of, tolyporphin biosynthesis genes.

Comparative genome analysis of HT-58-2Cyano with Nostoc punctiforme PCC 73102 and Anabaena variabilis ATCC 29413 using progressiveMauve (72) indicates that while there are conserved regions within the genome, there is considerable rearrangement in HT-58-2Cyano versus the other genomes. The conserved colinear blocks shared between the three cyanobacteria represent only small sections of the genome. No homology was observed between the three genomes in the region of bp 2586793 to 2609884 of the HT-58-2Cyano genome, which contains the putative gene cluster for tolyporphin biosynthesis. Genome comparisons with cyanobacteria in the proposed Brasilonema clade are hampered by the lack of available genome sequences.

Analysis of biosynthetic pathways for production of secondary metabolites (via antiSMASH) identified gene clusters within the HT-58-2Cyano genome that, while distinct from some sequenced cyanobacterial genomes, have similarity to secondary metabolite pathways present in other cyanobacteria. In particular, there are genes coding for production of heterocyst glycolipids (85% similarity) and those for a pederin-type polyketide compound, nosperin, initially identified from a Nostoc sp. strain associated with lichen (73). The data presented here indicate that the biosynthesis of nosperin may occur in a cyanobacterium not associated with lichen but clearly associated with a microbial community. A gene cluster for biosynthesis of shinorine (75% similarity), a mycosporine-like amino acid acting as a sunscreen in cyanobacteria (74), is also present in HT-58-2Cyano. Confirming the biosynthesis of these compounds by the HT-58-2 culture presents new avenues and challenges for natural product discovery.

Conclusions.

This study revealed a diverse community of bacteria with, and directly attached to, the HT-58-2 cyanobacterium. A metagenomic survey revealed the presence of several bacterial types, with Erythrobacteraceae (Porphyrobacter sp. in particular) dominating the community. The complexity of the HT-58-2 culture shows a cyanobacterium providing a foundational infrastructure for a diverse biological community. The genome sequence of HT-58-2Cyano indicates the potential for extensive secondary metabolite production. A cluster of hem genes colocalized with potential genes for tetrapyrrole modification was identified as a putative gene cluster for tolyporphin biosynthesis. The advances described here concerning growth conditions, cyanobacterial phylogeny, delineation of a cyanobacterium-microbial community, and the finished HT-58-2Cyano genome provide a strong foundation for investigation of the biosynthesis and function of the extraordinary tetrapyrroles tolyporphins.

MATERIALS AND METHODS

Culture source, media, and growth conditions.

Environmental sample HT-58-2 was originally isolated from soil collected at Nan Madol, Pohnpei, Micronesia, in the late 1980s (18). The sample was stored (but apparently lost) at the University of Hawaii (UH) for many years. The strain was recovered at UH by one of us (P.G.W.) in 2010. Two cryopreserved vials of HT-58-2 in 20% glycerol were shipped to North Carolina State University (NCSU), and one of them was grown in BG-11 medium (25 ml, 28°C, 150 rpm), which provides soluble nitrogen in the form of NaNO₃ (75). Green filamentous clumps were transferred to fresh BG-11 and cultured under continuous white light for stock cultures (76). BG-11 was prepared by addition of stock I (10 ml), stock II (10 ml), stock III (10 ml), Na₂CO₃ (0.02 g), stock V (1 ml), and NaNO₃ (1.5 g). For BG-11o, NaNO₃ was omitted. Stock solutions are as follows: I, 0.1 g of EDTA disodium salt liter⁻¹, 0.6 g of ferric ammonium citrate liter⁻¹, 0.6 g of citric acid-H₂O liter⁻¹, and 3.6 g of CaCl₂·2H₂O liter⁻¹; II, 7.5 g of MgSO₄·7H₂O liter⁻¹; III, 4 g of K₂HPO₄·3H₂O liter⁻¹; and V, 2.86 g of H₃BO₃ liter⁻¹, 1.81 g of MnCl₂·4H₂O liter⁻¹, 0.222 g of ZnSO₄·7H₂O liter⁻¹, 0.079 g of CuSO₄·5H₂O liter⁻¹, 0.05 g of CoCl·6H₂O liter⁻¹, and 0.391 g of NaMoO₄·2H₂O liter⁻¹. After combining stocks, the pH was adjusted to 7.5 followed by autoclaving (http://microbiology.ucdavis.edu/meeks/BG11medium.html). Agar plates and slants were prepared for midterm (1 to 3 months) preservation by addition of 1.5% agar to BG-11. The plates and slants were preserved under dim light at room temperature. For long-term preservation, HT-58-2 was suspended in BG-11 containing 20% glycerol and cryopreserved at −80°C. All HT-58-2 studied at NCSU was subcultured from this stock.

Sample inoculation and growth conditions.

An HT-58-2 30-day culture (150 to 200 ml) was centrifuged at 6,000 × g, the pellet was washed with 20 to 30 ml of fresh BG-11 and centrifuged again at 6,000 × g for 10 min, and that pellet was suspended in 10 ml of BG-11 and then homogenized to break up the HT-58-2 cell clumps. The homogenized culture was centrifuged and washed as described above, and the pellet was suspended in 10 ml of BG-11 or BG-11o; a sample of 1 to 3 ml was used to inoculate 150 to 200 ml of fresh BG-11. HT-58-2 was cultured in BG-11 or BG-11o at 28°C with shaking at 190 rpm under continuous white light (62 μmol m⁻² s⁻¹). When harvesting cells, HT-58-2 cultures grown in BG-11 were centrifuged at 34,000 × g for 20 min. Cell cultures grown in BG-11o were centrifuged for 1 h or longer to form a tight cell pellet. Supernatants were discarded, the pellet was lyophilized overnight, and cells were preserved by storage at −20°C.

Microscopy of HT-58-2. (i) Slide preparation and light microscopy.

Light and fluorescence microscopy used plain slides (25 by 75 by 1 mm) and no. 1.5 coverslips (22 by 22 mm). Light microscopy images were collected using live, unstained HT-58-2 cells (1 to 2 clumps in 15 μl of culture) under coverslips. Cells were imaged using a Zeiss Axioskop 2 Plus microscope under a 100× oil objective (N-Achoplan 100× 1.25 oil Ph3 [Zeiss]) using a Spot Pursuit camera and Spot 5.1 imaging software.

(ii) Fluorescence confocal microscopy.

HT-58-2 filament clumps were viewed live and unstained (autofluorescence) on a Zeiss Axio Imager microscope using a 63× objective with laser excitation at 488 nm with emission captured at 513 to 616 nm and at 654 to 729 nm. For stained imaging, HT-58-2 cells (1 to 2 clumps) were stained using FilmTracer FM 1-43 green biofilm cell stain (catalog no. F10317; Thermo Fisher) according to a modified version of the FilmTracer protocol with increased incubation with the stain to improve the staining of attached bacteria. Briefly, the FilmTracer stain was used at a final concentration of 1 μg ml⁻¹, with enough stain (20 to 30 μl) to cover the filament clumps, and then incubated for 1 h at room temperature protected from light. The sample was carefully washed under reduced light twice with sterile water to remove excess stain before viewing using a Zeiss LSM 710 laser scanning confocal system with a Zeiss Axio Observer Z1 inverted microscope at 63× using a 1.2-numerical-aperture (NA) W C-Apochromat objective with the following laser settings: 488-nm excitation at 13.7% and images captured at 513- to 619-nm and 654- to 729-nm emission.

(iii) Scanning electron microscopy.

HT-58-2 was cultured in BG-11 or BG-11o liquid medium for 14 days following transfer from a 30-day culture in BG-11. Three to five clumps of HT-58-2 were used for imaging. Cyanobacterial clumps were fixed in 3% glutaraldehyde in sodium phosphate buffer (0.1 M, pH 7.0) and then transferred to microporous capsules (Structure Probe Inc.). Samples were rinsed in three changes of phosphate buffer followed by dehydration in a graded ethanol series (30%, 50%, 70%, and 95% and three times at 100%), critical-point dried in liquid CO₂, mounted on carbon tabs, sputter coated with gold-palladium, and viewed at 10 kV. The method was developed at the Center for Electron Microscopy at NCSU.

Phylogenetic analysis of HT-58-2Cyano.

Phylogenetic analysis used the HT-58-2Cyano 16S rRNA sequence obtained from genomic sequence data (PacBio and Illumina MiSeq; cell separation and DNA preparation method 1, employing incubation in 2% DMSO as described below). HT-58-2Cyano and several related and signature filamentous cyanobacterial 16S rRNA sequences were identified by BLASTN and retrieved from GenBank (77, 78); short clones and uncultured entries were omitted. All sequences were trimmed from ACGGGTGAGT to ACCGCCCGTC to have 1,262 ± 2 bp, covering six 16S rRNA variable regions (V2 to V7). Sequences were aligned using ClustalW (79) with parameters as follows: gap open penalty, 15; gap extension penalty, 6.66; and weight matrix, IUB (54). The Phylip-formatted output has been deposited in figshare (https://doi.org/10.6084/m9.figshare.5296270). Phylogenetic tree construction used the neighbor-joining SplitsTree method (26) and the Bayesian inference method using the Geneious MrBayes plug-in (27). Parameters for the latter were as follows: substitution model, HKY85; rate variation, gamma; gamma categories, 4; chain length, 1,100,000; heated chains, 4; subsampling frequency, 200; burn-in length, 100,000; heated chain temperature, 0.2; and random seed, 14,922. Nostoc sp. strain PCC 7120 (BA000019) was selected as the outgroup (20, 80).

16S rRNA analysis of the HT-58-2 microbial community.

HT-58-2 cultures were grown in liquid BG-11 or BG-11o for 20 days as described above. Two hundred fifty microliters of biomass was collected along with culture medium to a final volume of 2 ml. Biomass was pelleted at 17,000 × g for 2 min, the supernatant was removed, and the pellet was frozen at −20°C until extracted. Community genomic DNA was extracted from the pellets using the PowerSoil DNA isolation kit (Mo Bio Labs, Inc.) according to the manufacturer's recommendations. The extracted DNA was quantified using the Quant-iT double-stranded DNA (dsDNA) high-sensitivity assay kit (Thermo Fisher). For Illumina MiSeq analysis, 16S rRNA amplification targeted the V3 and V4 regions to create a library using the primers according to the work of Klindworth et al. (81). PCR mixtures contained the following: 5 ng of extracted DNA in 2.5 μl of 10 mM Tris (pH 8.5), 1 μΜ (each) forward and reverse primers, and 12.5 μl 2× Q5 high-fidelity polymerase (New England Biolabs) in a final volume of 25 μl. The PCR program used was as follows: 95°C for 3 min followed by 25 cycles of 95°C for 30 s, 55°C for 30 s, and 72°C for 30 s before 7°C for 5 min and a hold at 4°C. The resulting PCR products were cleaned using AMPure XP beads according to the manufacturer's protocol. The second-stage PCR was carried out by combining 5 μl of the clean PCR product, 5 μl of Nextera X1 index primer 1 (BG-11, N709; BG-11o, N710), 5 μl of Nextera X1 index primer 2 (BG-11, S517; BG-11o, S517), 25 μl of 2× Q5 high-fidelity polymerase, and 10 μl of PCR-grade water, using the following program: 98°C for 30 s with eight cycles of 98°C for 10 s, 58°C for 20 s, and 72°C for 30 s followed by 72°C for 2 min and a 4°C hold. The resulting PCR product was cleaned using AMPure XP beads. The resulting library was quantified using the Quant-iT dsDNA high-sensitivity assay kit (Thermo Fisher). Samples were diluted to 10 nM in 10 mM Tris buffer (pH 8.5) and used for sequencing on an Illumina MiSeq instrument. The number of reads obtained was 32,095 and 27,578 from the BG-11 and BG-11o cultures, respectively. OTUs of two and fewer reads were removed from the analysis, and in this initial survey, sample replicates and a no-DNA reaction were not included. MiSeq reads were analyzed using the CLC Microbial Genomics Module (Qiagen) followed by strain identification with the One Codex platform (San Francisco, CA).

DNA purification for genome sequencing.

Using cyanobacterium-specific 16S rRNA primers (82), one clear sequence was read from the amplicon, although Sanger sequencing of universal 27F and 1492R primer amplicons yielded mixed sequence (data not shown). Such observations accrued despite using a cell separation and DNA preparation procedure (method 1) employing incubation in 2% DMSO for 2 to 5 days that by light microscopy appeared to remove obvious bacteria from the HT-58-2Cyano filaments (see Fig. S4 in the supplemental material). This DNA preparation was used with Illumina MiSeq (San Diego, CA) methods from which 300-bp paired-end reads were assembled and 16S rRNA coding sequences for HT-58-2Cyano were identified. The sole cyanobacterial 16S rRNA sequence assembled matched exactly that obtained in the amplicon sequence using cyanobacterium-specific primers [CYA106F, 5′-CGGACGGGTGAGTAACGCGTGA-3′; CYA781R(a), 5′-GACTACTGGGGTATCTAATCCCATT-3′] (82). These data confirmed the presence of only one cyanobacterial strain in the HT-58-2 culture.

(i) Cell preparation method 1 for DNA extraction.

An HT-58-2 culture (25 ml, 30 days) was homogenized at 8,000 rpm for 5 min to disrupt cell clumps, followed by centrifugation at 34,000 × g for 30 min. The cell pellet was washed in 10 ml of BG-11 and recentrifuged at 34,000 × g. The washed cell pellet was suspended in 10 ml of BG-11, of which 100 μl was inoculated into 15 ml of fresh BG-11. The cells were grown at 28°C with shaking at 150 rpm under continuous white-light illumination (62 μmol m⁻² s⁻¹). After incubation for 15 days, dimethyl sulfoxide (DMSO; 300 μl) was added to the 15-ml HT-58-2 culture. After 2 days, the DMSO-treated HT-58-2 cells were examined with optical microscopy and then centrifuged at 34,000 × g for 20 min (Fig. S2). The resulting cell pellet was vortexed in 5 ml of phosphate-buffered saline (PBS) buffer (137 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, and 1.8 mM KH₂PO₄, pH 7.4) and then centrifuged at 18,000 rpm followed by decanting of the supernatant (containing community cells); this procedure was performed three times. Finally, the washed cell pellet was vortexed with PBS buffer and filtered through a 0.45-μm filter. The filtered cell pellet was collected.

(ii) Cell preparation method 2 for DNA extraction.

A 100-ml 30-day culture of HT-58-2 in BG-11 was centrifuged at 3,000 ×g for 20 min and washed twice in fresh BG-11 to remove unattached bacteria in the culture medium. The pellet was suspended in 10 ml of BG-11 and then homogenized at 0°C for 20 min at 30,000 rpm to yield single filaments that were centrifuged at 4,000 × g for 20 min and washed twice in BG-11. The washed pellet was suspended in 30 ml of fresh BG-11.

Three cultures of HT-58-2 were established by adding 10 ml of homogenized and washed HT-58-2 biomass to 200 ml of fresh BG-11. Cultures were incubated for 10 to 14 days at 28°C with light and 190 rpm to increase biomass. After 14 days, the cells were centrifuged at 3,000 × g for 20 min and washed twice in BG-11o, and the resulting pellet was suspended in fresh BG-11o medium and incubated as before for 7 days to reduce the number of community cells present. The resulting cultures were once again centrifuged, washed and suspended in BG-11, and incubated as before for 2 days. The resulting culture was centrifuged at 326 × g for 20 min, washed twice in BG-11, pelleted at 326 × g for 20 min, homogenized to single filaments, and again washed twice in BG-11 with centrifugation at 145 × g before being suspended in 100 ml of fresh BG-11 containing 10% DMSO. Cultures were incubated at 28°C for 24 h in the light at 190 rpm. The reduction in community cells was assessed using light microscopy. DNA isolation typically was done after 2 days of incubation in DMSO (when the removal of most community cells occurred) without a large detrimental effect on HT-58-2Cyano viability (biomass/pigment assessment). Filaments were harvested by settling the culture in a centrifuge tube and removing the supernatant, followed by centrifugation at 208 × g for 20 min and washing twice in PBS containing 10% DMSO, followed by washing twice in PBS containing 20% DMSO with a 5-min incubation at room temperature between each wash. The filaments were pelleted by centrifugation at 208 × g for 20 min at 4°C. Filaments were then washed in PBS twice to remove any residual DMSO by centrifugation at 208 × g for 20 min at 4°C. Any remaining attached cells were lysed using the lysis protocol in the Qiagen Puregene Yeast/Bact kit with modification. Briefly, the pelleted cells were suspended in 5 ml of cell suspension solution with 25 μl of lytic enzyme solution, incubated at 37°C for 30 min, and centrifuged at 2,310 × g for 10 min at 4°C. The supernatant was then removed, and another 5 ml of cell lysis solution was added with 25 μl of RNase A solution followed by incubation at 37°C for 15 min. The sample was centrifuged at 2,310 × g for 10 min at 4°C, washed with PBS, and centrifuged again. This procedure afforded DNA enriched from HT-58-2Cyano.

(iii) DNA extraction.

The final cyanobacterial pellet (2 to 3 g wet weight) of HT-58-2 was carefully ground under liquid nitrogen, and DNA extraction was carried out twice using the phenol-chloroform method (83). DNA was precipitated in ethanol and sodium acetate using a method modified from that of Sambrook et al. (84) with the addition of 1 μl of 20 mg ml⁻¹ glycogen (Thermo Scientific) followed by incubation at −20°C for 18 h. DNA size and quantity were assessed using an Agilent 4200 TapeStation system.

(iv) Purity assessment of DNA.

Sequencing of a 16S rRNA gene amplicon was used to assess the purity of extracted HT-58-2 DNA samples. PCR was carried out using universal 16S rRNA primers (27F and 1492R) and the following protocol: 95°C for 10 min and 30 cycles of 90°C for 30 s, 50°C for 30 s, and 72°C for 2 min, followed by a final 5-min extension at 72°C and a hold at 4°C. Reagents were as follows: primers at 2 mM concentrations, 200 μM deoxynucleoside triphosphate (dNTP), master mix (1× concentration of 10× standard master mix), and 1 U of DNA polymerase (Apex Taq; Genesee Scientific Corp.) in a final volume of 25 μl. PCR products were precipitated with ethanol as described above and suspended in 15 μl of nuclease-free water. Amplicon (20 to 50 ng μl⁻¹) sequencing used commercially available primers 27F (5′-AGAGTTTGATCMTGGCTCAG-3′) and 1492R (5′-TACGGYTACCTTGTTACGACTT-3′).

Genome sequencing and assembly.

Using cell and DNA preparation method 2, HT-58-2Cyano DNA was sized (20 kbp) and libraries were prepared in the NCSU Genomic Sciences Laboratory followed by PacBio SMRT cell sequencing by RTL Genomics (Lubbock, TX). Reads (490,426; 9,180 average length; N₅₀ of 16,358; 4,502,006,607 bp total) were quality control processed to 80,332 reads of 6,511 average length and an N₅₀ of 7,517 and yielding a total of 523,002,363 bp for assembly. Assembly used the CLC Genomics PacBio de novo assembly pipeline (Qiagen Co.), yielding a total of 10 contigs of which four were identified with BLASTN as cyanobacterial and three as Porphyrobacter sp. These were processed and extended using the CLC Genome Finishing Module: Extend Contigs tool (Qiagen). The extended HT-58-2Cyano contigs were used as a scaffold for implementing the CLC de novo assembly map to contigs tool with Illumina MiSeq (San Diego, CA) reads obtained with DNA from preparation method 1. Contigs were assessed for inverted repeats using dot plots of alignment against self. If repeats were observed, the contig was assessed for assembly errors. Contigs with shared end homology were connected using the Geneious version 8.1.8 (85) de novo assembly tool. Correct joining of the final two large HT-58-2Cyano contigs was confirmed with a forward primer at position 2875629 (5′-CCAATCCTCAACACACCTACATTGGC-3′) and a reverse primer at position 2876779 (5′-GCTGGCTCCTACATGTAACAGTTCG-3′), generating an 1,150-bp product, and with a forward primer at position 4939194 (5′-GGTCTTGCAATAGATTTCTGGCTTCG-3′) and a reverse primer at position 4939903 (5′-GCCATTAGTTCATATTGTGTTCCAG-3′), generating a 719-bp product. Sanger sequencing of both PCR products confirmed proper contig joining. Correct closure of plasmid one (pHT582-1) contig was confirmed with the forward primer at position 27322 (5′-CTAATTGCACTTCTAGCTTTTGG-3′) and the reverse primer at position 28119 (5′-GGAGTTTGCTTATGAATTAGGTTGC-3′), generating a 797-bp product. Sanger sequencing of the resulting PCR product confirmed proper contig closure. At the time of manuscript submission, closure of plasmid 2 (pHT582-2) to the exact base pair was in progress.

Preliminary annotation was carried out using the RAST prokaryotic pipeline with frameshift analysis (86–88), and annotation for GenBank submission was done using the NCBI Prokaryotic Genome Annotation Pipeline (https://www.ncbi.nlm.nih.gov/genome/annotation_prok/). The annotated HT-58-2Cyano genome has the start of the genome 100 bp 5′ from the initiation codon of dnaA. Comparative genome analysis used progressiveMauve (72). Two autonomous plasmids identified in the HT-58-2Cyano genome sequence were assembled from the same collection of reads and were finished and annotated by the same approach.

Accession number(s).

Accession numbers for DNA sequences reported here are included under BioProject PRJNA369278 and are CP019636, CP019637, and CP019638.