Genomic basis for natural product biosynthetic diversity in the actinomycetes

Abstract

The phylum Actinobacteria hosts diverse high G + C, Gram-positive bacteria that have evolved a complex chemical language of natural product chemistry to help navigate their fascinatingly varied lifestyles. To date, 71 Actinobacteria genomes have been completed and annotated, with the vast majority representing the Actinomycetales, which are the source of numerous antibiotics and other drugs from genera such as Streptomyces, Saccharopolyspora and Salinispora. These genomic analyses have illuminated the secondary metabolic proficiency of these microbes – underappreciated for years based on conventional isolation programs – and have helped set the foundation for a new natural product discovery paradigm based on genome mining. Trends in the secondary metabolomes of natural product-rich actinomycetes are highlighted in this review article, which contains 199 references.

1 General features of Actinobacteria

The large bacterial phylum Actinobacteria is home to some of the most common soil microbial inhabitants that are responsible for degrading and recycling organic materials. This diverse group of high G + C, Gram-positive bacteria exhibits varied morphologies, physiologies, and metabolic properties that allow it to prosper in a wide range of environments. Actinobacteria are particularly noteworthy in human medicine where they encompass a number of devastating human pathogens as well as life-saving producers of antibiotics. The discovery of the antituberculosis agent streptomycin from the culture broth of Streptomyces griseus by Waksman in 1944 represented an important milestone in natural product chemotherapy¹ and heralded the beginning of targeting the genus Streptomyces and related Actinomycetales microorganisms for antibiotics.

The taxonomy of the Actinobacteria is complex, and in 1997 was classified by Stackebrandt into five orders, 11 suborders, 42 families, 110 genera and more than 1000 species (Fig. 1).² Microorganisms belonging to the order Actinomycetales are fascinatingly diverse and include human pathogens (e.g., Mycobacterium tuberculosis,³ Corynebacterium diphtheriae,⁴ Propionibacterium acnes,⁵ and Nocardia farcinica⁶), plant pathogens (Streptomyces scabies⁷ and Leifsonia xyli⁸), nitrogen-fixing symbionts (Frankia spp.⁹), terrestrial (Streptomyces spp.^10-12) and marine (Salinispora spp.¹³) sediment inhabitants, and key members of the gut flora that reside in the colon (Bifidobacterium spp.¹⁴). A notable feature of the Actinomycetales is their metabolic versatility in the production of chemically diverse and biologically active secondary metabolites.¹⁵ Streptomyces in particular produces not only the majority of the clinical antibiotics of natural origin (e.g., vancomycin, erythromycin and tetracycline),^16,17 but also gives rise to important antifungal (amphotericin B),¹⁸ anticancer (mitomycin C),¹⁹ antiparasitic (ivermectin),²⁰ and immunosuppressive (rapamycin)²¹ agents. Other members of the Actinomycetales with robust metabolic features include Corynebacterium glutamicum, which is an important industrial producer of amino acids and vitamins,²² and Rhodococcus species, which have the ability to catabolize a wide range of organic compounds.²³

Fig. 1.

Phylogenetic relatedness of Actinobacteria based on 16S rRNA sequence and the relative number of putative gene clusters for secondary metabolites. Strains with completed genomes are colored blue, those not yet published are noted with an asterisk, and those completed but with incomplete annotation are marked with double asterisks. Phylogenetic analysis of aligned sequences was done by a bootstrap method using bootstrap number 1000 and seed number 100.

2 Actinomycete genome projects

Since the completion by shotgun sequencing of the first genome in 1995, that of Haemophilus influenzae Rd,²⁴ over 700 bacterial genomes have been deposited in the NCBI database. While genome-sequencing efforts were initially biased toward pathogens, more recent sequencing projects have focused on environmental and industrially relevant bacteria. To date, 53 Actinobacteria genomes have been completed and annotated, with the vast majority being in the order Actinomycetales (Table 1). Their genomes range from 1.93 Mb (Bifidobacterium animalis subsp. lactis) to 10.15 Mb (Streptomyces scabies) in size and contain 1605 (Mycobacterium leprae) to 8983 (S. scabies) protein-encoding genes. While the majority of their chromosomes are circular, most of the larger Actinomycetales chromosomes (e.g., Rhodococcus jostii RHA1,²³ Rhodococcus opacus B4, and all Streptomyces strains²⁵) form linear structures with distinct telomeres containing unique terminal-inverted repeats that bind terminal proteins.²⁶ Although linear chromosomes, which are the predominant genetic elements in eukaryotes, have also been identified in other prokaryotes such as the Lyme disease spirochaete Borrelia burgdorferi²⁷ and the α-proteobacterium Agrobacterium tumefaciens^28,29 (the causal agent of Crown Gall disease in plants), terminal-inverted repeats and terminal proteins are absent in these chromosomes.

Table 1.

Features of Actinobacteria completed genomes^a

Microorganism	Subclass	Order	Size (bp)	GC (mol%)	Chr	Pls	Orf	%Coding	RNAs	rrn
Acidimicrobium ferrooxidans  DSM 10331	Acidimicrobidae	Acidimicrobiales	2 145 416	68	b	—	2156	90	54	2
Acidothermus cellulolyticus 11B	Actinobacteridae	Actinomycetales	2 443 540	67	1C	—	2157	89	56	1
Actinosynnema mirum DSM 43827	Actinobacteridae	Actinomycetales	8 133 898	73	b	—	7258	86	75	5
Arthrobacter aurescens TC1	Actinobacteridae	Actinomycetales	4 597 686	62	1C	2C	4041	88	73	6
Arthrobacter chlorophenolicus A6	Actinobacteridae	Actinomycetales	4 395 537	66	1C	2C	3885	89	65	5
Arthrobacter sp. FB24	Actinobacteridae	Actinomycetales	4 698 945	65	1C	3C	4146	89	69	5
Beutenbergia cavernae DSM 12333	Actinobacteridae	Actinomycetales	4 669 183	73	1C	—	4197	92	53	2
Bifidobacterium adolescentis  ATCC 15703	Actinobacteridae	Bifidobacteriales	2 089 645	59	1C	—	1631	86	70	5
Bifidobacterium animalis subsp.  lactis AD011	Actinobacteridae	Bifidobacteriales	1 933 695	60	1C	—	1528	84	59	2
Bifidobacterium animalis subsp.  lactis ATCC SD5219	Actinobacteridae	Bifidobacteriales	1 938 709	60	1C	—	1567	85	64	4
Bifidobacterium animalis subsp.  lactis DSM 10140	Actinobacteridae	Bifidobacteriales	1 938 483	60	1C	—	1566	85	63	4
Bifidobacterium longum DJO10A	Actinobacteridae	Bifidobacteriales	2 375 792	60	1C	2C	1990	86	73	4
Bifidobacterium longum NCC2705	Actinobacteridae	Bifidobacteriales	2 256 640	60	1C	1C	1727	85	70	4
Bifidobacterium longum subsp.  infantis ATCC 15697	Actinobacteridae	Bifidobacteriales	2 832 748	59	1C	—	2416	85	94	4
Brachybacterium faecium DSM  4810	Actinobacteridae	Actinomycetales	3 608 624	72	b	—	3187	89	57	b
Clavibacter michiganensis subsp.  michiganensis NCPPB 382	Actinobacteridae	Actinomycetales	3 297 891	73	1C	2C	2984	89	60	2
Clavibacter michiganensis subsp.  sepedonicus ATCC 33113	Actinobacteridae	Actinomycetales	3 258 645	72	1C	2C	2941	86	51	2
Corynebacterium aurimucosum  ATCC 700975	Actinobacteridae	Actinomycetales	2 790 189	60	1C	1C	2531	88	67	4
Corynebacterium diphtheriae  NCTC 13129	Actinobacteridae	Actinomycetales	2 488 635	54	1C	—	2272	87	69	5
Corynebacterium efficiens YS-314	Actinobacteridae	Actinomycetales	3 147 090	63	1C	—	2950	90	70	5
Corynebacterium glutamicum  ATCC 13032 Bielefeld	Actinobacteridae	Actinomycetales	3 282 708	54	1C	—	3002	87	86	6
Corynebacterium glutamicum  ATCC 13032 Kitasato	Actinobacteridae	Actinomycetales	3 309 401	54	1C	—	2993	86	80	6
Corynebacterium glutamicum R	Actinobacteridae	Actinomycetales	3 314 179	54	1C	1C	3052	86	76	6
Corynebacterium jeikeium K411	Actinobacteridae	Actinomycetales	2 462 499	61	1C	1C	2104	89	61	3
Corynebacterium kroppenstedtii  DSM 44385	Actinobacteridae	Actinomycetales	2 446 804	57	1C	—	2018	86	55	3
Corynebacterium urealyticum  DSM 7109	Actinobacteridae	Actinomycetales	2 369 219	64	1C	—	2024	89	60	3
Eggerthella lenta DSM 2243	Coriobacteridae	Coriobacteriales	3 610 731	64	b	—	3181	87	53	3
Frankia alni ACN14a	Actinobacteridae	Actinomycetales	7 497 934	73	1C	—	6711	86	63	2
Frankia sp. CcI3	Actinobacteridae	Actinomycetales	5 433 628	70	1C	—	4499	84	70	2
Frankia sp. EAN1pec	Actinobacteridae	Actinomycetales	8 982 042	71	1C	—	7191	83	59	3
Kineococcus radiotolerans  SRS30216	Actinobacteridae	Actinomycetales	4 761 183	74	1C	2C	4480	90	64	4
Kocuria rhizophila DC2201	Actinobacteridae	Actinomycetales	2 697 540	71	1C	—	2357	88	57	3
Kytococcus sedentarius DSM 20547	Actinobacteridae	Actinomycetales	2 788 433	71	b	—	2703	90	51	2
Leifsonia xyli subsp. xyli str.  CTCB07	Actinobacteridae	Actinomycetales	2 584 158	68	1C	—	2030	70	50	1
Micrococcus luteus Fleming  NCTC 2665	Actinobacteridae	Actinomycetales	2 501 097	72	1C	—	2345	88	56	2
Mycobacterium abscessus CIP  104536	Actinobacteridae	Actinomycetales	5 067 172	64	1C	1C	4920	92	50	1
Mycobacterium avium 104	Actinobacteridae	Actinomycetales	5 475 491	69	1C	—	5120	88	50	1
Mycobacterium avium subsp.  paratuberculosis K-10	Actinobacteridae	Actinomycetales	4 829 781	69	1C	—	4350	91	49	1
Mycobacterium bovis AF2122/97	Actinobacteridae	Actinomycetales	4 345 492	66	1C	—	3920	90	50	1
Mycobacterium bovis BCG str.  Pasteur 1173P2	Actinobacteridae	Actinomycetales	4 374 522	66	1C	—	3952	90	52	1
Mycobacterium bovis BCG str.  Tokyo 172	Actinobacteridae	Actinomycetales	4 371 711	65	1C	—	3947	90	50	1
Mycobacterium gilvum PYR-GCK	Actinobacteridae	Actinomycetales	5 619 607	68	1C	3C	5241	92	55	2
Mycobacterium leprae Br4923	Actinobacteridae	Actinomycetales	3 268 071	57	1C	—	1604	49	47	b
Mycobacterium leprae TN	Actinobacteridae	Actinomycetales	3 268 203	58	1C	—	1605	49	50	1
Mycobacterium marinum M	Actinobacteridae	Actinomycetales	6 636 827	66	1C	1C	5423	89	51	1
Mycobacterium smegmatis str.  MC2 155	Actinobacteridae	Actinomycetales	6 988 209	67	1C	—	6716	90	54	2
Mycobacterium sp. JLS	Actinobacteridae	Actinomycetales	6 048 425	68	1C	—	5739	92	57	2
Mycobacterium sp. KMS	Actinobacteridae	Actinomycetales	5 737 227	68	1C	2C	5460	92	57	2
Mycobacterium sp. MCS	Actinobacteridae	Actinomycetales	5 705 448	68	1C	1C	5391	92	61	2
Mycobacterium tuberculosis  CDC1551	Actinobacteridae	Actinomycetales	4 403 837	66	1C	—	4189	90	48	1
Mycobacterium tuberculosis F11	Actinobacteridae	Actinomycetales	4 424 435	66	1C	—	3950	90	48	1
Mycobacterium tuberculosis H37Ra	Actinobacteridae	Actinomycetales	4 419 977	66	1C	—	4034	90	50	1
Mycobacterium tuberculosis H37Rv	Actinobacteridae	Actinomycetales	4 411 532	66	1C	—	4402	90	50	1
Mycobacterium ulcerans Agy99	Actinobacteridae	Actinomycetales	5 631 606	66	1C	—	4160	72	50	1
Mycobacterium vanbaalenii PYR-1	Actinobacteridae	Actinomycetales	6 491 865	68	1C	—	5979	91	58	2
Nocardia farcinica IFM 10152	Actinobacteridae	Actinomycetales	6 021 225	71	1C	2C	5683	90	64	3
Nocardioides sp. JS614	Actinobacteridae	Actinomycetales	4 985 871	71	1C	1C	4645	91	55	2
Propionibacterium acnes  KPA171202	Actinobacteridae	Actinomycetales	2 560 265	60	1C	—	2297	89	54	3
Renibacterium salmoninarum  ATCC 33209	Actinobacteridae	Actinomycetales	3 155 250	56	1C	—	3507	90	51	2
Rhodococcus erythropolis PR4	Actinobacteridae	Actinomycetales	6 516 310	62	1C	2C1L	6030	91	74	5
Rhodococcus jostii RHA1c	Actinobacteridae	Actinomycetales	7 804 765	67	1L	3L	7211	91	63	4
Rhodococcus opacus B4	Actinobacteridae	Actinomycetales	7 913 450	67	1L	—	7246	91	62	4
Rubrobacter xylanophilus DSM  9941	Rubrobacteridae	Rubrobacterales	3 225 748	71	1C	—	3140	91	64	1
Saccharopolyspora erythraea  NRRL 2338	Actinobacteridae	Actinomycetales	8 212 805	71	1C	—	7198	84	66	4
Salinispora arenicola CNS-205	Actinobacteridae	Actinomycetales	5 786 361	70	1C	—	4917	85	63	3
Salinispora tropica CNB-440	Actinobacteridae	Actinomycetales	5 183 331	70	1C	—	4536	88	61	3
Streptomyces avermitilis MA-4680	Actinobacteridae	Actinomycetales	9 025 608	71	1L	2L	7577	86	89	6
Streptomyces coelicolor A3(2)	Actinobacteridae	Actinomycetales	8 667 507	72	1L	1L1C	7769	88	86	6
Streptomyces griseus subsp. griseus  IFO13350	Actinobacteridae	Actinomycetales	8 545 929	72	1L	—	7136	87	86	6
Streptomyces scabies 87.22d	Actinobacteridae	Actinomycetales	10 148 695	72	1L	—	8983	87	92	6
Thermobifida fusca YX	Actinobacteridae	Actinomycetales	3 642 249	68	1C	—	3110	85	67	4

^aAbbreviations: Chr, chromosome; Pls, plasmid; C, circular replicon; L, linear replicon; Orf, open-reading frames; RNAs, functional RNA molecules (tRNA, rRNA, tmRNA, 4.5S RNA, RNaseP; rrn, ribosomal RNA operon (16S rRNA–23S rRNA–5S rRNA).

^bnot defined.

^cThe species name in the first report was not defined but the strain has since been classified as Rhodococcus jostii.

^dThe detailed properties are not published but sequence and annotation data are available from the Sanger Centre (ftp://ftp.sanger.ac.uk/pub/pathogens/ssc/).

Homology searching of genes encoding known proteins involved in secondary metabolism reveals that Actinomycetales microbes contain a wide range of diverse secondary metabolic biosynthesis gene clusters from just a couple to well over 30 pathways (Fig. 1). Some families in particular, such as the Streptomycetaceae (genera Streptomyces and Kitasatospora), Micromonosporaceae (Salinispora and Micromonospora) and Pseudonocardiaceae (Saccharopolyspora), regularly possess more than twenty gene clusters for secondary metabolism and in doing so dedicate over 5% of their coding capacity to these processes. The natural product-rich Actinomycetales organisms generally have large genomes of >5 Mb and concentrate their secondary metabolic genes distal to the origin of replication gene dnaA (Fig. 2). Not all Actinomycetales, however, are prolific secondary metabolite producers, especially those with small genomes ranging from 2–4 Mb such as Kocuria rhizophila (2.7 Mb), which harbors just two secondary metabolic pathways.³⁰

Fig. 2.

Circular representation of the actinomycete chromosomes S. coelicolor A3(2), S. erythraea NRRL2338 and S. tropica CNB-440 oriented to the dnaA gene (top). All genomes are scaled to their relative size. The inner rings show a normalized plot of GC skew, while the center rings show a normalized plot of GC content. The outer circles show the distribution of secondary metabolite gene clusters. Biosynthetic gene clusters associated with thiotemplate-based assembly (PKS, NRPS) are depicted in red, terpene clusters in blue, and loci encoding all other secondary metabolic pathways are marked in black. Clusters whose products have been isolated are labeled accordingly.

The Tuberculosis Database (TBDB) (http://www.tbdb.org/) was recently developed as a single repository to house genome sequence and annotation data together with gene expression data.³¹ As of September 2008, TBDB hosted 28 different M. tuberculosis strains as well as other mycobacteria and bacteria from related taxa such as Streptomyces coelicolor A3(2), Streptomyces avermitilis, Nocardia farcinica IFM 10152, and Rhodococcus jostii RHA1. Furthermore, TBDB houses nearly 1500 public tuberculosis and 260 Streptomyces microarrays. This database provides a suite of analytical tools for the comparative genomic analysis of M. tuberculosis and is intended to serve as a community hub for TB research but may also facilitate genomic research in the greater actinomycete scientific community. A number of other TB databases exist, including GenoMycDB (a database for comparative analysis of mycobacterial genes and genomes)³² and MGDD (Mycobacterium tuberculosis genome divergence database), yet they are at present more limited in scope.³³ The rich secondary metabolism of mycobacteria, which is employed in the elaborate construction of its specialized lipid components, has been previously reviewed³⁴ and is thus not covered in this review article. Rather, we focus our attention on reporting and discussing trends in select natural product-rich Actinobacteria with completed genome sequences.

2.1 Streptomyces

The Streptomyces genus is the largest within the Actinobacteria, with over 900 described species (http://www.ncbi.nlm.nih.gov/Taxonomy/). Their natural product biosynthetic prowess is arguably unsurpassed amongst the microbes, and hence they are of utmost importance in the pharmaceutical industry.¹⁵ To date, four Streptomyces genomes have been completed and are publicly available (Table 1), while numerous other Streptomyces genome projects have either been completed privately or are in draft assembly (such as the 16 draft Streptomyces genome sequences assembled at the Broad Institute – http://www.broadinstitute.org/annotation/genome/streptomyces_group). These studies revealed that Streptomyces have large (~8–10 Mb), linear chromosomes²⁵ containing well over 20 secondary metabolic gene clusters, which encode the biosynthesis of polyketides by polyketide synthases (PKSs),³⁵ peptides by nonribosomal peptide synthetases (NRPSs),³⁶ bacteriocins,³⁷ terpenoids, shikimate-derived metabolites, aminoglycosides, and other natural products. In this section we highlight the individual and comparative secondary metabolic features of Streptomyces coelicolor A3(2),¹⁰ Streptomyces avermitilis MA-4680,^11,38 and Streptomyces griseus IFO 13350.¹² The fourth species with a completed genome sequence, Streptomyces scabies strain 87.22, has not yet been published but is publicly available online at http://www.sanger.ac.uk/Projects/S_scabies/.

2.1.1 Streptomyces coelicolor A3(2)

Streptomyces coelicolor A3(2) is genetically the best known representative of the genus and has served as a model strain for the study of actinomycete chemistry and biology.³⁹ Its single, linear chromosome from the derivative strain M145, which lacks the plasmids SCP1 and SCP2 of the parent strain A3(2), was sequenced and annotated in 2002 from an ordered cosmid library and contains 8 667 507 bp (72.1% GC content) coding for 7769 genes and 56 pseudogenes (Table 1).¹⁰ The linear SCP1⁴⁰ (356 023 bp; 69.1% GC content; 351 coding sequences) and the circular SCP2⁴¹ (31 317 bp; 72.1% GC content; 34 coding sequences) plasmids were later sequenced separately.

One of the main reasons why S. coelicolor A3(2) was developed as a model genetics system is its production of diffusible pigments that could serve as convenient genetic markers.⁴² At the time the genome sequence was released, the structures and biosynthetic gene clusters associated with the blue aromatic polyketide antibiotic actinorhodin (1),⁴³ the red oligopyrrole prodiginine antibiotics (2),^44,45 and the acidic lipopeptide calcium-dependent antibiotics (3)⁴⁶ were known. Furthermore, the whiE type II PKS gene cluster encoding an unknown aromatic dodecaketide associated with the gray spore pigment was established⁴⁷ as was the chemistry of the common streptomycete membrane lipid hopanoids hopene (4) and 29-(2′-hydroxyethyl)hopane.⁴⁸ The initial genomic analysis of the S. coelicolor A3(2) chromosome predicted a further 17 secondary metabolic gene clusters.¹⁰ At present there are 29 predicted chromosomal gene clusters specifying secondary metabolites and another two from the linear plasmid SCP1 (Table 2). The majority of these gene clusters reside outside the central core of the chromosome in the subtelomeric regions.¹⁰ Their associated pathway products arise from a myriad of enzymatic processes and include polyketides, nonribosomal peptides, bacteriocins, terpenoids, and other natural products, which is a hallmark of streptomycete genomes. The most thoroughly investigated S. coelicolor A3(2) pathway involves the biosynthesis of the aromatic type II PKS product actinorhodin,⁴⁹ which significantly was the target of the first genetic engineering experiment to yield an “unnatural” hybrid molecule⁵⁰ and helped establish the scientific concept of smallmolecule rational engineering through combinatorial biosynthesis.⁵¹

Table 2.

Secondary metabolite gene clusters in Streptomyces coelicolor A3(2)

No.	Cluster location	Actual or predicted producta
1	SCO0124-0129	Eicosapentaenoic acid
2	SCO0185-0191	Isorenieratene (12)
3	SCO0267-0270	Lantibiotic
4	SCO0381-0401	Deoxysugar
5	SCO0489-0499	Coelichelin (5)
6	SCO0753-0756	Bacteriocin
7	SCO1206-1208	THN (6), flaviolin (7)
8	SCO1265-1273	Aromatic polyketide
9	SCO1864-1867	5-Hydroxyectoine (14)
10	SCO2700-2701	Melanin
11	SCO2782-2785	Desferrioxamine (13)
12	SCO3210-3249	Calcium-dep. antibiotic (3)
13	SCO5071-5092	Actinorhodin (1)
14	SCO5222-5223	Albaflavenone (11)
15	SCO5314-5320	Gray spore pigment
16	SCO5799-5801	Siderophore
17	SCO5877-5898	Prodiginine (2)
18	SCO6073	Geosmin (9)
19	SCO6266	γ-Butyrolactone Scb1 (15)
20	SCO6273-6288	Hexaketide
21	SCO6429-6438	Dipeptide
22	SCO6681-6685	Lanthionine-cont. peptide SapB
23	SCO6759-6771	Hopene (4)
24	SCO6826-6827	Polyketide
25	SCO6927-6932	Lantibiotic
26	SCO7221	Germicidin (8)
27	SCO7669-7671	Aromatic polyketide
28	SCO7681-7691	Coelibactin
29	SCO7700-7701	2-Methylisoborneol (10)
30	SCP1.228c-246	Methylenomycin (18), 2-alkyl-4-   (hydroxymethyl)furan-3-   carboxylic acids (17)

^aObserved products are highlighted in bold.

The complete sequence and annotation of the S. coelicolor A3(2) genome allowed for its rational mining to identify many of the orphan pathway products.⁵² Structure prediction of the product of the trimodular NRPS associated with SCO0492 initially suggested a tripeptide siderophore⁵³ that upon subsequent isolation yielded the new tetrapeptide iron chelator coelichelin (5).⁵⁴ This seminal discovery of a novel natural product by genomic prediction helped usher in a new era of microbial “genome mining” for new chemical entities.^52,55,56 In the case of 5, this discovery not only allowed the characterization of new biosynthetic reactions in NRPS biochemistry but also led the way to the characterization of further S. coelicolor A3(2) metabolites often with very intriguing structures, biosyntheses, and biological properties. Three type III PKS gene clusters are encoded in the genome, and the products of two of these pathways, 1,3,6,8-tetrahydroxynaphthalene (THN) (6) and its auto-oxidation product flaviolin (7) by THN synthase^57,58 (SCO1206), and the germicidin (8) family of alkyl pyrones by germicidin synthase⁵⁹ (SCO7221), have been characterized, while that of SCO7671 is unknown. Terpenoid biochemistry is also prevalent in S. coelicolor A3(2), with at least five gene clusters dedicated to this type of biosynthesis (Table 2). After the completion of the genome analysis, recombinant enzymes associated with three odorous-terpenoid compounds, namely geosmin^60,61 (9), 2-methylisoborneol⁶² (10) and albaflavenone^63-65 (11), were characterized and their pathway products confirmed. Geosmin and 2-methylisoborneol are widespread not only in Actinobacteria but also in Cyanobacteria and Myxobacteria as well as fungi.⁶⁶ The remaining terpenoid pathway involves the carotenoids isorenieratene (12) and β-carotene, which were shown to be produced under blue light induction.⁶⁷ Further genome mining efforts led to the tris-hydroxamate family of desferrioxiamine (13) siderophores,⁶⁸ which are assembled by a new family of ATP-dependent oligomerization-macrocyclization biocatalysts.⁶⁹ Production of ectoine and 5-hydroxyectoine (14), compatible solutes conferring dehydration protection in the genus Streptomyces, is encoded by a set of four genes (SCO1864–SCO1867) and can be triggered upon exposure to high salinity or elevated growth temperature.⁷⁰ While no ribosomally-derived bacteriocins have been characterized from S. coelicolor A3(2), four putative gene clusters are likely (Table 2), with SCO6681–SCO6685 being very homologous to the SapB lantibiotic-like peptide from S. coelicolor J1501/pKN22.⁷¹

The biosynthesis of an unknown polyketide metabolite encoded by the cryptic modular PKS gene cluster SCO6273–SCO6288 is partially controlled by the signaling compound Scb1 (15),^72-74 which is a γ-butyrolactone structurally related to the prototypical actinomycete hormone A-factor (16) that regulates antibiotic production and morphological differentiation in S. griseus.⁷⁵ The Scb1 butenolide synthase gene scbA SCO6266 is located adjacent to the PKS gene set that it regulates. A homolog, mmfL, is located within the linear plasmid SCP1, where it facilitates the biosynthesis of 2-alkyl-4-(hydroxymethyl)furan-3-carboxylic acids (17).^76,77 These molecules specifically induce the production of the antibiotic methylenomycin (18)^78,79 and hence have been collectively termed the methylenomycin furans.⁷⁶ Since glycosylated derivatives of 17 have been described from other streptomycetes⁸⁰ and a related three-gene operon harboring mmfL is present in the S. avermitilis genome, these furans may constitute a new class of diffusible signaling molecules much like the better-known γ-butyrolactones.

2.1.2 Streptomyces avermitilis MA-4680

The draft genome sequence of the industrial microorganism S. avermitilis was published in 2001 and allowed the first genome-level glimpse into the extensive secondary metabolic dexterity of a streptomycete.³⁸ Its completed nucleotide sequence was published two years later and showed that the linear genome contains 9 025 608 bp (70.7% GC content) and encodes 7582 protein coding sequences (Fig. 3).¹¹ S. avermitilis also harbors two linear plasmids, SAP1 (94 287 bp; 69.2% GC content; 96 coding sequences) and SAP2 (ca. 200 kb).

Fig. 3.

Schematic representation of the S. avermitilis MA-4680 chromosome. Drawing details are the same as for Fig. 2.

Streptomyces avermitilis was discovered in 1977 from a soil sample in Japan and produces the polyketide macrolide avermectin (19), which is a potent anthelmintic agent extremely effective against nematodes, insects and spiders. Its semisynthetic derivative ivermectin has been in use commercially since the 1980s in human and veterinary medicine.^81,82 Prior to the availability of the genome sequence, the structure and biosynthesis of avermectin⁸³ and that of another polyketide macrolide, oligomycin⁸⁴ (20), had already been established. A further 11 PKS gene clusters (Table 3) were deduced from the genome sequence, leading to the subsequent isolation of a third group of macrolides,⁸⁵ the filipins (21), which belong to the polyene family of antifungal agents previously isolated from Streptomyces filipinensis.⁸⁶ As with S. coelicolor A3(2), S. avermitilis produces a brown spore pigment associated with the whiE-homologous type II PKS cluster spp, which upon disruption led to albino-colored spores (H. Ikeda, unpublished). Homologs of the whiE and spp clusters are prevalent in other Streptomyces microbes,⁴⁷ yet not in S. griseus, which instead produces polyketide melanins via a type III PKS pathway.⁸⁷

Table 3.

Secondary metabolite gene clusters in Streptomyces avermitilis MA-4680

No.	Cluster location	Actual or predicted producta
1	SAV_76	Terpene
2	SAV_100–101	Polyketide
3	SAV_257–259	Microcin
4	SAV_407–419	Filipin (21)
5	SAV_603–609	Peptidic siderophore
6	SAV_837–869	Non-ribosomal peptide
7	SAV_935–953	Avermectin (19)
8	SAV_1019–1025	Isorenieratene (12)
9	SAV_1136–1137	Melanin
10	SAV_1249–1251	PK-NRP hybrid
11	SAV_1550–1552	Polyketide
12	SAV_1650–1654	Hopene (4)
13	SAV_2163	Geosmin (9)
14	SAV_2267–2269	2-alkyl-4-hydroxymethylfuran-3-   carboxylic acids
15	SAV_2277–2282	Polyketide
16	SAV_2367–2369	Polyketide
17	SAV_2372–2388	Aromatic polyketide
18	SAV_2465–2467	Siderophore
19	SAV_2835–2842	Spore pigment
20	SAV_2890–2903	Oligomycin (20)
21	SAV_2990–3002	Neopentalenolactone (23)
22	SAV_3031–3032	Albaflavenone (11)
23	SAV_3155–3164	Non-ribosomal peptide
24	SAV_3193–3202	Non-ribosomal peptide
25	SAV_3636–3651	Non-ribosomal peptide
26	SAV_3653–3667	Aromatic polyketide
27	SAV_5149	Ochronotic pigment
28	SAV_5269–5272	Desferrioxamine (13)
29	SAV_5361–5362	Melanin
30	SAV_5686–5689	Microcin
31	SAV_6395–6398	5-Hydroxyectoine (14)
32	SAV_6632–6633	Non-ribosomal peptide
33	SAV_7130–7131	THN, flaviolin
34	SAV_7161–7165	Non-ribosomal peptide
35	SAV_7184	Polyketide
36	SAV_7320–7323	Vibrioferrin-like siderophore
37	SAV_7360–7362	Polyketide

^aObserved products are highlighted in bold.

In addition to the enormous diversity of PKS pathways in S. avermitilis,⁸⁸ the genome harbors secondary metabolic gene clusters for at least eight nonribosmal peptides, six terpenoids, and assorted pigments, osmolyte compounds, siderophores, and bacteriocins,⁸⁹ for a total of 37 secondary metabolic gene clusters spanning 6.6% of the genome (Table 3). Most of the secondary metabolic genes are present in the chromosomal arms rather than in the core region of the chromosome. While no peptide products have yet been characterized, the majority of the terpenoids have either been characterized or their structures deduced based on biosynthetic precedent. The triterpene hopene (4) (H. Ikeda, unpublished), its abundant precursor squalene, the carotenoid isorenieratene⁶⁷ (12), the odorous sesquiterpenoids geosmin⁹⁰ (9), germacradienol⁹⁰ (22), and albaflavenone (11) (via its precursor epi-isozizaene⁶⁵), and derivatives of the sesquiterpene antibiotic neopentalenolactone^91-93 (23) from Baeyer–Villiger monooxygenase chemistry, have all been characterized. While all the S. avermitilis terpenoid chemistry had been previously identified in other streptomycete strains, investigating the corresponding biosynthesis enzymes as recombinant proteins has been instrumental in shedding light onto the mechanisms by which these common streptomycete terpenoid molecules are assembled. In the case of 9 and 22, the gene product of SAV_2163 catalyzes their Mg-dependent formation from farnesyl diphosphate.⁹⁰

Streptomyces avermitilis produces the desferrioxamine-related siderophore nocardamine (24) under iron-limiting conditions (H. Ikeda, unpublished). However, unlike S. coelicolor A3(2), it does not produce additional siderophores via NRPS logic such as coelichelin (5), although it possesses the genetic capacity. The NRPS SAV_603 exhibits 65% homology to a myxobacterial counterpart involved in the biosynthesis of the siderophore myxochelin⁹⁴ and is associated with three putative iron(III)/siderophore transporter genes (SAV_600–SAV_602). The S. avermitilis genome also harbors another gene cluster (SAV_7320–SAV_7323) that may code for a novel streptomycete siderophore related to vibrioferrin from the Gram-negative bacterium Vibrio parahaemolyticus,⁹⁵ yet the product of this pathway as well as the majority of the S. avermitilis secondary metabolome are presently unknown.

2.1.3 Streptomyces griseus IFO 13350

Streptomyces griseus IFO 13350 was discovered in S. A. Waksman’s laboratory more than 60 years ago and is famous for producing the aminoglycoside antibiotic streptomycin (25), which was instrumental in helping save people suffering from tuberculosis.¹ Its complete genome sequence was published in 2008 and consists of a single linear chromosome of 8 546 929 bp with no plasmids (Fig. 4) and containing extremely long terminal inverted repeats of 132 910 bp each.¹² In addition to its industrial importance in producing streptomycin, S. griseus is an important model bacterium for studying extracellular signaling by a diffusible, low-molecular weight compound, the γ-butyrolactone A factor (16).^75,96

Fig. 4.

Schematic representation of the S. griseus IFO 13350 genome. Drawing details are the same as for Fig. 2.

Sequence analysis of the S. griseus genome and comparison against the other completed Streptomyces genomes suggested the existence of at least 36 genes or gene clusters associated with the biosynthesis of secondary metabolites (Table 4). As with S. coelicolor A3(2) and S. avermitilis, the majority reside outside the core region of the chromosome. In addition to streptomycin (25) whose biosynthesis had previously been characterized,⁹⁷ biosynthetic genes and gene clusters were also known for A factor (16),⁹⁸ the diffusible yellow pigmented grixazones (26),⁹⁹ an uncharacterized carotenoid,¹⁰⁰ and the 1,4,6,7,9,12-hexahydroxyperylene-3,10-quinone (HPQ) melanin (27)^87,101 via the prototype bacterial type III PKS THN synthase (RppA).¹⁰² HPQ melanin is the dominant spore pigment of S. griseus and is distinct from aromatic polyketide spore pigments associated with the whiE gene cluster of S. coelicolor A3(2) and S. avermitilis which are absent in this strain. A second type III PKS pathway involving SGR470–SGR472 was later established to producealkylresorcinol (28) and alkylpyrone (29) phenolic lipids that confer penicillin resistance on the strain.¹⁰³

Table 4.

Secondary metabolite gene clusters in Streptomyces griseus IFO13350

No.	Cluster location	Actual or predicted producta
1	SGR_54–60	Carotenoid
2	SGR_278–283	PK-NRP hybrid
3	SGR_443–455	Peptidic siderophore
4	SGR_470–472	Alkylresorcinol (28)
5	SGR_527–528	Melanin
6	SGR_574–593	Non-ribosomal peptide
7	SGR_604-611	9-Membered enediyne
8	SGR_653–656	Non-ribosomal peptide
9	SGR_810–815	PK-NRP hybrid
10	SGR_895–901	Non-ribosomal peptide
11	SGR_962–966	Hopanoid
12	SGR_1268–1269	2-Methylisoborneol (10)
13	SGR_2079	Terpene
14	SGR_2446–2447	Melanin
15	SGR_2482–2489	PK-NRP hybrid
16	SGR_2586–2598	Non-ribosomal peptide
17	SGR_3239–3288	PK-NRP hybrid
18	SGR_3845–3849	Lantibiotic
19	SGR_4238–4250	Grixazone (26)
20	SGR_4408–4421	Thiazolylpeptide
21	SGR_4747–4751	Desferrioxamine (13)
22	SGR_5285–5295	Unknown
23	SGR_5632–5635	Ectoine, 5-hydroxyectoine (14)
24	SGR_5914–5940	Streptomycin (25)
25	SGR_6065	Terpene
26	SGR_6071–6083	Polyketide
27	SGR_6177–6183	Polyketide
28	SGR_6360–6387	Polyketide
29	SGR_6619–6620	HPQ melanin (27)
30	SGR_6709–6717	Peptidic siderophore
31	SGR_6730–6742	Peptidic siderophore
32	SGR_6776–6786	PK-NRP hybrid
33	SGR_6824–6830	Isorenieratene (12)
34	SGR_6839	Geosmin (9)
35	SGR_6889	A factor (16)
36	SGR_7079–7085	Carotenoid

^aObserved products are highlighted in bold.

The majority of the S. griseus secondary metabolome is dominated by thiotemplate-based chemistry involving a total of at least nine PKS and nine NRPS gene clusters (Table 4). None of the natural product chemistry associated with the type I and type II PKS gene clusters, including the putative enediyne PKS (SGR604–SGR611) – which is similar to those common in Salinispora genome-sequenced strains¹⁰⁴ – as well as from the NRPS pathways, has been characterized to date. Amongst the NRPS gene clusters, three (SGR443–SGR455, SGR6709–SGR6717 and SGR6730–SGR6742) are associated with siderophore-dependent iron transporter genes and thus are likely involved in the biosynthesis of peptidic siderophores as seen in S. coelicolor A3(2).

Terpenoid biosynthetic pathways also predominate in the S. griseus genome and include three copies of carotenoid gene clusters, four orthologs of terpene cyclases, and a cluster for hopanoid biosynthesis (Table 4). The carotenoid gene cluster of SGR54–SGR60 is identical to that of SGR7079–SGR7085 and together they reside in the long inverted-repeat telomere regions. Orthologs corresponding to the geosmin/germacradienol and 2-methylisoborneol terpene synthases have been identified, and the production of 2-methylisoborneol (10) was experimentally confirmed.⁶⁶ The remaining terpene synthases, however, likely utilize farnesyl diphosphate for the synthesis of as-yet unknown sesquiterpenoid compounds. Further cryptic clusters associated with the biosynthesis of the siderophore nocardamine, melanins derived from l-dopamine, and bacteriocins are also predicted from the S. griseus genome.

2.1.4 Streptomyces comparative genomics

Sequence comparison of the three taxonomically distinct Streptomyces genome strains, S. coelicolor A3(2), S. avermitilis and S. griseus, suggests a conserved internal core region of approximately 6.4 Mb in which the majority of the genes are conserved and show global synteny (Fig. 5).¹² Flanking the conserved core are variable subtelomeric regions of 0.5–2 Mb in size that carry mainly nonessential variable genes that include the majority of the secondary metabolic gene clusters. This variability was also reported in the S. ambofaciens chromosome,¹⁰⁵ suggesting that this feature is common in Streptomyces linear chromosomes. Large DNA rearrangements involving gene duplication, elimination and acquisition are frequently observed in the chromosomal arms, suggesting that chromosomal instability and intermolecular recombination are important driving forces in Streptomyces evolution.^106,107 Streptomyces typically carry large linear plasmids of >200 kb that can harbor a multitude of secondary metabolic gene clusters¹⁰⁸ as in the case of pSLA2-L of S. rochei,¹⁰⁹ and their chromosomal integration¹¹⁰ may provide the immediate opportunity for the production of a suite of bioactive products.

Fig. 5.

The comparative analysis of nucleotide sequences of S. griseus IFO 13350 (top), S. avermitilis MA-4680 (middle) and S. coelicolor A3(2) (bottom) using the MURASAKI program version 1.40 (weight; 45, length; 90). Predicted core regions (blue lines) account for 6.28 Mb of S. coelicolor A3(2), 6.50 Mb of S. avermitilis and 6.39 Mb of S. griseus and are centered about the replication of origin oriC (black arrow). The subtelomeric regions (bold lines) are less conserved with regard to sequence and ortholog distribution and contain more than half of the gene clusters for secondary metabolism (colored triangles).

The three Streptomyces genomes have seven conserved gene clusters, including those for hopanoid (4) and desferrioxamine (13) siderophore biosyntheses at the core region of their respective chromosomes (Fig. 6). Each gene cassette is arranged identically in terms of size and direction and is located within very similar genomic regions. On the other hand, other examples exist in which species-specific secondary metabolic gene clusters reside in genomic islands within conserved regions, as is the case with the oligomycin (20) cluster in S. avermitilis and the streptomycin (25) cluster in S. griseus (Fig. 7). The oligomycin biosynthesis gene cluster in S. avermitilis is inserted between the conserved F0–F1 ATPase operon and the teichoic acid biosynthetic gene tagO, while the S. griseus streptomycin gene cluster is inserted between the phenylalanyl-tRNA synthetase operon and the fatty acid catabolic gene fadC. Although the origins of these secondary metabolic gene clusters are not clear, they were most likely acquired through horizontal gene transfer. The genomic island concept linking secondary metabolism with functional adaptation and acquisition has recently been explored in the marine actinomycete genus Salinispora,¹⁰⁴ and is further discussed in Section 2.3.

Fig. 6.

Conserved genomic regions amongst S. griseus IFO 13350, S. avermitilis MA-4680 and S. coelicolor A3(2), harboring (A) hopanoid and (B) desferrioxamine biosynthetic gene clusters.

Fig. 7.

Examples of “island” secondary metabolic gene clusters in Streptomyces involving (A) oligomycin biosynthesis in S. avermitilis and (B) streptomycin in S. griseus.

2.2 Saccharopolyspora erythraea NRRL2338

The filamentous soil bacterium Saccharopolyspora erythraea is used for the industrial-scale production of the polyketide erythromycin A (30), which was the first macrolide antibiotic to be used clinically.¹¹¹ This soil-dwelling actinomycete was originally identified as Streptomyces erythraeus but later assigned to the genus Saccharopolyspora based on chemotaxonomic observations.¹¹² Despite this reclassification, the biology of S. erythraea had been considered similar to that of Streptomyces microbes, even though these distinct genera belong to different suborders. The sequence analysis of the 8 212 805 bp S. erythraea NRRL2338 genome established in 2007, however, revealed considerable divergence from the streptomycetes in both gene organization and function.¹¹³ Furthermore, while all Streptomyces genomes sequenced to date are linear, the S. erythraea genome has a circular topology like most other Actinobacteria (Table 1, Fig. 2).

The S. erythraea genome sequence suggests a very active secondary metabolism, with at least 27 biosynthesis genes or gene clusters associated with a diverse range of secondary metabolites of mostly unknown chemical compositions (Table 5). Only five of these gene sets, including that of erythromycin (ery), are distributed in the predicted core region of the genome that contains most essential genes.¹¹³ Expression profiling analyses using DNA microarrays revealed the differential expression of the core and non-core genes during the different growth stages of S. erythraea in which ery and six other secondary metabolism clusters were up-regulated.¹¹⁴ The erythromycin PKS encoding genes were first sequenced in the early 1990s,^115,116 and established the paradigm for bacterial type I modular PKSs in the assembly of natural and engineered polyketides.^117,118 Further S. erythraea polyketide potential includes modular and iterative type I PKSs, such as a second modular PKS (pke) of unknown function¹¹⁹ related to alkenyl furanone biosynthesis in streptomycetes¹²⁰ and myxobacteria¹²¹ and a putative polyunsaturated fatty acid PKS pathway. While no type II PKS gene clusters typical of streptomycetes are evident in the S. erythraea genome,¹¹³ the common streptomycete type III PKS THN synthase (SACE1242–SACE1243) is associated with the flaviolin (7)-based reddish pigments particular to the S. erythraea red variant.¹²²

Table 5.

Secondary metabolite gene clusters in Saccharopolyspora erythraea NRRL2338

No.	Cluster location	Actual or predicted producta
1	SACE_0018–0028	Polyunsaturated fatty acids
2	SACE_0483–0486	Ectoine, 5-hydroxyectoine (14)
3	SACE_0712–0734	Erythromycin (30)
4	SACE_1241–1244	THN (6), flaviolin (7)
5	SACE_1304–1310	Dipeptide
6	SACE_2342–2347	Diketide
7	SACE_2618–2622	PK-NRP hybrid
8	SACE_2628–2631	Tetraketide
9	SACE_2692–2703	Pentameric siderophore
10	SACE_2864–2879	Diketide
11	SACE_3013–3017	Non-ribosomal peptide
12	SACE_3029–3039	Tetrapeptide siderophore
13	SACE_3187	Terpene (geosmin?)
14	SACE_3223–3229	Non-ribosomal peptide
15	SACE_3721–3722	2-methylisoborneol (10)
16	SACE_3976–3979	Terpene (geosmin?)
17	SACE_4025–4029	Lanthionine-containing peptide
18	SACE_4128–4145	Nonaketide lactone
19	SACE_4230–4233	Lanthionine-containing peptide
20	SACE_4275–4292	Lipopeptide
21	SACE_4302–4307	Polyketide
22	SACE_4327–4331	Hopanoids
23	SACE_4471–4478	Polyketide
24	SACE_4567–4578	Aromatic polyketide
25	SACE_4906–4909	Terpene (geosmin?)
26	SACE_5306–5309	Methylsalicylic acid
27	SACE_5532	Naphthoic acid

^aObserved products are highlighted in bold.

Also absent is a homologous desferrioxamine siderophore biosynthetic pathway common to Streptomyces and Salinispora genera. Rather, S. erythraea produces a hydroxamate-type iron-siderophore (erythrobactin)¹²³ of unknown structure and may produce two further peptidic siderophores associated with the NRPS gene clusters SACE2692–SACE2703 and SACE3029–SACE3039. Terpenoid biochemistry is also prevalent in S. erythraea, which harbors a putative hopanoid biosynthetic pathway and three copies of geosmin/germacradienol synthase genes (Table 5).⁶⁶ While their functions have not yet been verified, S. erythraea produces geosmin (9) and germacradienol (22), and phylogenetic analyses suggest that SACE3187 may be involved in their biosynthesis. S. erythraea also produces 2-methylisoborneol (10) during sporulation on solid media, and SACE3721–SACE3722 have been identified as 2-methylisoborneol synthase and geranyl diphosphate methyltransferase genes, respectively.⁶⁶

2.3 Salinispora tropica CNB-440 and Salinispora arenicola CNS-205

The genus Salinispora constitutes a discrete group of actinomycete bacteria that thrive in tropical and sub-tropical oceanic sediments.^124,125 Phylogenetically associated with the family Micromonosporaceae, the Salinispora strains fall into three major phylotypes that exhibit different geographical distribution patterns. Salinispora arenicola has the widest distribution and appears to be ubiquitous in tropical and subtropical sediments, while Salinispora tropica and “Salinispora pacifica” have been retrieved from more restricted locations.^13,126 A common feature of all Salinispora strains and what makes them of interest for biotechnological applications is the production of bioactive natural products for use in infectious disease and cancer treatments.¹²⁷

The genomes of two representative strains, S. tropica CNB-440¹²⁸ in 2007 (Fig. 2) and S. arenicola CNS-205¹⁰⁴ in 2009 (Fig. 8), were assembled and functionally annotated. Ortholog plots of the fully sequenced chromosomes revealed a high degree of conservation and synteny across the entire Salinispora genomes, even though there are also several breakpoint regions that represent genomic islands or inversions. In contrast to the linear 7.7-Mb chromosome of their relative Micromonospora chalcea,¹²⁹ the Salinispora genomes are significantly smaller with 5 183 331 bp and 5 786 361 bp, respectively, and exhibit a circular topology (Table 1). A large amount of coding capacity is devoted to the biosynthesis of secondary metabolites, at 8.8% in S. tropica and 10.9% in S. arenicola. Bioinformatic analyses revealed a total of 49 gene clusters in which the vast majority encode the biosynthesis of as-yet unknown chemical entities.¹⁰⁴ Both genomes are particularly rich in polyketide pathways, including those for multimodular type I PKSs, iterative type I PKSs, hybrid type I PKS–NRPSs, heterodimeric type II PKSs, and homodimeric type III PKSs. While thiotemplate-based chemistry involving polyketide synthases and nonribosomal peptide synthetases appears to be the predominant theme in natural product assembly, the Salinispora chromosomes also harbor genes for the production of non-NRPS derived siderophores, bacteriocins, aminocyclitols, and terpenoids.

Fig. 8.

Schematic representation of the S. arenicola CNS-205 genome. Drawing details are the same as for Fig. 2.

Altogether 19 biosynthetic gene clusters were annotated from the S. tropica genome (Table 6).^128,130 Two are associated with novel natural products, namely the hybrid PKS–NRPS salinosporamide products¹³¹ (31) and the modified enediyne sporolide polyketides¹³² (32), and a third with the previously characterized streptomycete metabolite lymphostin¹³³ (33). The analysis of the salinosporamide gene cluster gave rise to the elucidation of a new biological route of chlorination¹³⁴ associated with the biosynthesis of these unusual γ-lactam-β-lactone proteasome inhibitors and further facilitated the engineered biosynthesis of unnatural salinosporamide analogs.^135-137 Sequence analysis of the largest S. tropica secondary metabolic gene cluster, the 80 kb slm cluster, revealed that it was associated with the biosynthesis of the novel polyene macrolactam salinilactam A (34).¹²⁸ Its structure elucidation was not only aided by the predictive nature of the genome sequence but also assisted in the sequence refinement of its highly repetitive modular PKS sequence. Twelve of the S. tropica gene clusters, including that putatively assigned to lymphostin biosynthesis, are also conserved in the phylogenetically related S. arenicola.

Table 6.

Secondary metabolite gene clusters in Salinispora tropica CNB-440

No.	Cluster location	Actual or predicted producta
1	Stro0586–0610	10-Membered enediyne
2	Stro0667–0694	Dipeptide
3	Stro1012–1043	Salinosporamide (31)
4	Stro2174–2227	Glycosylated decaketide
5	Stro2340–2346	Aminocyclitol
6	Stro2428–2440	Thiazolylpeptide
7	Stro2486–2510	Aromatic polyketide
8	Stro2541–2555	Desferrioxamine (13)
9	Stro2645–2659	Yersiniabactin-like siderophore
10	Stro2691–2737	Sporolide (32)
11	Stro2757–2781	Salinilactam (34)
12	Stro2786–2813	Dihydroaeruginoic acid
13	Stro2814–2842	Coelibactin
14	Stro3042–3054	Class I bacteriocin
15	Stro3055–3066	Lymphostin (33)
16	Stro3244–3253	Carotenoid pigment
17	Stro4264–4267	Phenolic lipids
18	Stro4410–4429	Tetrapeptide
19	Stro4437–4441	Carotenoid pigment

^aObserved products are highlighted in bold.

The majority of the isolated chemistry from S. arenicola CNS-205 is conversely associated with known actinomycete metabolites, including the polyketide antibiotic rifamycin (35), the cyclic heptapeptide cyclomarin A (36), and the indolocarbazole staurosporine (37),¹²⁷ and their derivatives saliniketal¹³⁸ (38) and cyclomarazine¹³⁹ (39), which are unique to this strain. The functions of the remaining 27 biosynthetic gene clusters have not yet been reported, but represent a very diverse assemblage of secondary metabolic pathways that comprise a staggering 10.9% of the genes in the genome (Table 7).¹⁰⁴ The S. arenicola genome encodes two enediyne type I PKS gene clusters reminiscent of the actinomycete anticancer agents calicheamicin and kedarcidin; however, in both cases, their associated gene clusters are fragmented across secondary metabolic-rich genomic islands that may signify that the respective pathways are no longer functional. Although associated enediyne-derived molecules have not yet been characterized from S. arenicola, this genus is known to produce enediyne-relevant chemistry in the form of the chlorinated metabolites sporolide A (32) from S. tropica and cyanosporaside A (40) from “S. pacifica”.^132,140,141 The Salinispora genomes encode desferrioxamine and salicylate-primed NRPS siderophore biosynthetic pathways as in many terrestrial streptomycetes, yet are devoid of their characteristic terpenoid biosynthetic pathways that give rise to the “earthy” odiferous geosmin, 2-methylisoborneol and albaflavenone.

Table 7.

Secondary metabolite gene clusters in Salinispora arenicola CNS-205

No.	Cluster location	Actual or predicted producta
1	Sare0345–0367	Pentapeptide
2	Sare0478–0499	PK-NRP hybrid
3	Sare0545–0560	9-Membered enediyne – A
4	Sare0570–0573	Amino acid conjugate
5	Sare0602–0623	Class I bacteriocin
6	Sare1041–1073	Polyketide
7	Sare1240–1278	Rifamycin (35)
8	Sare1286–1288	Diterpene
9	Sare2017–2049	10-Membered enediyne – A
10	Sare2070–2081	Yersiniabactin-like siderophore
11	Sare2088–2121	9-Membered enediyne – B
12	Sare2144–2151	Amino acid conjugate
13	Sare2163–2206	10-Membered enediyne – B
14	Sare2326–2342	Staurosporine (37)
15	Sare2400–2409	PK-NRP hybrid
16	Sare2483–2491	Aminocyclitol
17	Sare2583–2595	Thiazolylpeptide
18	Sare2669–2694	Aromatic polyketide
19	Sare2728–2744	Desferrioxamine (13)
20	Sare2939–2968	Tetrapeptide
21	Sare3051–3063	Dipeptide
22	Sare3148–3163	Macrolide
23	Sare3268–3280	Class I bacteriocin
24	Sare3281–3293	Lymphostin (33)
25	Sare3471–3480	Carotenoid pigment
26	Sare4547–4569	Cyclomarin (36)
27	Sare4694–4697	Phenolic lipids
28	Sare4885–4904	Tetrapeptide
29	Sare4927–4931	Carotenoid pigment
30	Sare4932–4956	9-Membered enediyne – C

^aObserved products are highlighted in bold.

The availability of two Salinispora genome sequences has allowed the genetic dissection of conserved and dissimilar pathways that are involved in secondary metabolism. Despite a close phylogenetic relationship between S. tropica and S. arenicola, comparative genomics indicated that the majority of their biosynthetic potential is species-specific.¹⁰⁴ This result is consistent with previous fermentation studies showing that the Salinispora species produce distinct suites of secondary metabolites dependent on their geographic origin of isolation.¹⁴² Interestingly, most of the metabolic pathways that are unique to either one of the Salinispora genomes reside in genomic islands, whereas shared pathways conversely occur in the genus-specific core.¹⁰⁴ A similar picture has also emerged in the Streptomyces, as discussed in Section 2.1.4. The Salinispora genomic islands are not contiguous regions of species-specific DNA. Indeed, gene acquisition, loss, duplication, inactivation, and rearrangement have shaped their genetic makeup.¹⁴³ Surprisingly, the overall composition, evolutionary history, and function of the island genes are similar in both strains, with duplication and horizontal gene transfer accounting for the majority of genes, and secondary metabolism representing the largest functionally annotated category. It has been speculated recently that genomic islands house many of the adaptive traits that contribute to niche differentiation in bacteria.¹⁴⁴ Applying this paradigm to Salinispora would support a functional role for secondary metabolites in environmental adaptation and ultimately speciation.

2.4 Frankia sp. strain CcI3, Frankia alni ACN14a, and Frankia sp. strain EAN1pec

Frankia is a nitrogen-fixing endophyte that forms symbiotic associations with actinorhizal plants, but can also survive as a free-living soil dweller. To clarify the genetic basis for the distinct host range of different phylotypes, the genomes of three Frankia strains, ACN14a, CcI3, and EAN1pec, were sequenced in 2007.¹⁴⁵ All three chromosomes share a circular topology and do not contain any independently replicating plasmids. Correlating with host characteristics and biogeography, their sizes vary dramatically, ranging from 5.43 Mb for CcI3 to 9.04 Mb for EAN1pec (Table 1). Genes involved in secondary metabolism and regulation were found to be more common in the larger genomes of ACN14a and EAN1pec, respectively, suggesting that their acquisition has played a role in the process of host diversification and environmental adaptation.⁹ ACN14a and EAN1pec dedicate about 5% of their genomes to natural product assembly, a percentage only marginally smaller than the values that were previously reported for Streptomyces species. In contrast, the biosynthetic potential of CcI3 is significantly smaller (~3%). Analyses of codon adaptive indices have further indicated that CcI3 has fewer highly expressed biosynthesis genes than ACN14a and EAN1pec.¹⁴⁶ As secondary metabolism is generally assumed to enhance the competitiveness of bacteria, these findings may provide the molecular basis for the limited biogeographical distribution of CcI3. Notwithstanding these differences, all sequenced Frankia strains were found to be quite versatile natural product producers, harboring a multitude of different pathways, including those for polyketide,¹²⁸ non-ribosomal peptide, NRPS-independent siderophore, terpenoid, bacteriocin,⁸⁹ and aminocyclitol¹⁴⁷ biosynthesis. Homology-driven genome analysis has further revealed a unique pathway in ACN14a that potentially encodes the production of a novel phosphonate antibiotic.¹⁴⁸ A conserved cassette of six genes, homologous to those in the phosphinothricin cluster from Streptomyces viridochromogenes, is involved in the biosynthesis of phosphonoformate (Fig. 9). Even though late-pathway enzymes required for phosphinothricin biosynthesis are not encoded on the ACN14a locus, it is likely that phosphonoformate is not the end product of the Frankia pathway, as evidenced by the presence of additional biosynthetic genes. Many phosphonate-containing natural products exhibit herbicidal properties,¹⁴⁸ and the production of such a compound by a bacterial plant symbiont might be of ecological significance.

Fig. 9.

Gene organization and proposed biosynthesis of a phosphonoformate-like compound in Frankia alni ACN14a. (A) Phosphonate biosynthetic gene cluster in F. alni ACN14a. Genes that are responsible for the production of hydroxyethylphosphonate are depicted in grey (minimal cassette consisting of phosphoenolpyruvate mutase ppm, phosphonopyruvate decarboxylase ppd and phosphonoacetaldehyde dehydrogenase adh). (B) Proposed pathway to a hitherto unknown phosphonate antibiotic.

To date, the secondary metabolome of Frankia has not been explored beyond two reports concerning the production of anthraquinone- and calcimycin-type antibiotics.^149,150 The genome sequences of Frankia now certainly provide a rationale to explore the chemistry of these intriguing microorganisms.

2.5 Nocardia farcinica IFM 10152

Members of the genus Nocardia are saprophytic soil bacteria, but also include opportunistic agents of nocardiosis. Together with the two pathogens Corynebacterium and Mycobacterium, Nocardia forms a monophyletic taxon within the order Actinomycetales.¹⁵¹ Organisms within this clade share an unusual cell wall composition, characterized by the presence of a waxy cell envelope containing mycolic acids, which distinguishes them from other bacteria. In 2004, the genome of the N. farcinica clinical isolate IFM 10152 was fully sequenced, revealing a single circular 6 021 225 bp chromosome and two smaller plasmids of 184 027 bp (pNF1) and 87 093 bp (pNF2) (Table 1).⁶ Consistent with previous phylogenetic analysis, ortholog plots confirmed the close relationship to C. glutamicum and M. tuberculosis. The synapomorphy between these taxa is genetically conserved in a pathway sustaining the synthesis of mycolic acids.¹⁵² Iron mobilization is crucial for many pathogenic bacteria, and therefore it is not surprising that the production of salicylate-derived siderophores is a prevailing feature of nocardial secondary metabolism.^153-155 N. farcinica is no exception to this rule and possesses a gene set homologous to the mycobactin locus from M. tuberculosis,¹⁵⁶ involving a hybrid PKS–NRPS assembly line (NFA7610–NFA7710), as well as mechanisms of lysine modification. In addition, the N. farcinica genome contains several orphan clusters that potentially encode the production of novel natural products. Except for an ortholog of the nrps5 region in S. avermitilis (SAV6631–SAV6633), these loci are unique to N. farcinica.¹⁵² Most striking is the discrete NRPS gene NFA50330, which encodes a 14,474-aa biosynthetic enzyme. This NRPS putatively catalyzes the assembly of at least twelve amino acid monomers, but lacks a terminating thioesterase to release the maturated peptidyl chain from covalent tethering on the final thiolation domain. Therefore it is not clear whether or not the associated cluster is functional.

Even though nocardial biosynthetic pathways are strongly biased towards thiotemplate-based assembly lines, there is also genomic evidence for the production of terpenoid structures. Ishikawa and coworkers identified a putative cluster encompassing genes for the mevalonate pathway associated with isoprenoid-related genes.⁶ While no secondary metabolites have yet been characterized from N. farcinica IFM 10152, its genetic potential clearly mirrors findings from previous chemical investigations in which nocardial strains yielded a large number of structurally diverse metabolites such as anthracyclines,^157,158 angucyclines,¹⁵⁹ macrolides,¹⁶⁰ monobactams,¹⁶¹ indole alkaloids,¹⁶² and terpenoids.^163,164

2.6 Rhodococcus jostii RHA1

Rhodococcus jostii RHA1 is a versatile polychlorinated biphenyl-degrading soil microbe that has emerged as a model actinomycete for the biodegradation of a wide range of organic compounds.¹⁶⁵ Its complete nucleotide sequence was reported in 2006, revealing a linear 7 804 765 bp chromosome and three linear plasmids of 1 123 075 bp (pRH1), 442 536 bp (pRH2), and 332 361 bp (pRH3) (Table 1).²³ Although its linear chromosomal topology is similar to that of Streptomyces microbes, comparative genome analyses with other actinomycetes revealed that R. jostii is more closely related to M. tuberculosis. The genome is exceptionally rich in oxygenases (203) and ligases (192) that are predicted to contribute to its catabolic versatility in the degradation of xenobiotics. For instance, R. jostii RHA1 produces an impressive array of Baeyer–Villiger monooxygenases that catalyze diverse regio- and enantioselective reactions.¹⁶⁶ While rhodococci are known for their catabolic prowess, they are largely unknown for their secondary metabolic abilities. Although the closely related R. jostii K01-B0171 produces the anti-mycobacterial, lasso peptides lariatins A and B via a ribosomal process,¹⁶⁷ there are no reports of natural products derived from R. jostii RHA1. Surprisingly, when the R. jostii RHA1 genome was completed, two PKS, a hybrid PKS–NRPS, and 14 NRPS gene clusters were identified, suggesting the possibility of a robust secondary metabolism (Table 8).²³ Six of the NRPS genes are >25 kb in length, and one cluster (ro02318–ro02320) may be involved in the biosynthesis of an NRPS-based siderophore because of co-clustered siderophore-iron complex transporter genes. Terpene biosynthesis genes involved in common actinomycete metabolites such as carotenoid, hopanoid, and odoriferous sesquiterpenes are, however, noticeably absent.

Table 8.

Secondary metabolite gene clusters in Rhodococcus jostii RHA1

No.	Cluster location	Predicted product
1	RHA1_ro00071–00073	Non-ribosomal peptide
2	RHA1_ro00232–00235	Non-ribosomal peptide
3	RHA1_ro00429–00435	Non-ribosomal peptide
4	RHA1_ro01305–01307	Ectoine
5	RHA1_ro02207–02310	PK-NRP hybrid
6	RHA1_ro02318–02320	Peptidic siderophore
7	RHA1_ro02391–02397	Non-ribosomal peptide
8	RHA1_ro02492–02494	Non-ribosomal peptide
9	RHA1_ro04063–04066	Polyketide
10	RHA1_ro04230–04231	Polyketide
11	RHA1_ro04379–04382	Non-ribosomal peptide
12	RHA1_ro04713–04717	Non-ribosomal peptide
13	RHA1_ro05093–05103	Non-ribosomal peptide
14	RHA1_ro05430–05431	Non-ribosomal peptide
15	RHA1_ro05452–05453	Non-ribosomal peptide
16	RHA1_ro05466–05468	Non-ribosomal peptide
17	RHA1_ro06098–06103	Non-ribosomal peptide
18	RHA1_ro06663–06665	Non-ribosomal peptide

3 Actinomycete genome scanning for new natural product biosynthesis gene clusters

Since genes encoding biosynthetic enzymes, resistance proteins, and regulatory systems associated with microbial natural products are typically clustered, genome sequencing has emerged as a powerful tool to quickly identify biosynthetic gene clusters associated with known metabolites. Natural product biosynthetic gene clusters are typically identified by targeting proposed biosynthetic or resistance genes or by various in vivo strategies such as transposon mutagenesis or by complementing nonproducing mutants. Whole-genome scanning offers an alternative method that is direct and reliable, especially for those gene sets associated with natural products without specific biosynthetic reactions for which to conveniently probe. Several examples exist from the actinomycete literature, and some of these are highlighted below.

The distinctive C₅N unit in the phosphoglycolipid moenomycin A (41) facilitated a PCR-guided discovery of a three-gene operon in Streptomyces ghanaensis associated with its biosynthesis from aminolevulinate.¹⁶⁸ However, sequence analysis of flanking genes did not reveal the remainder of the moenomycin A biosynthetic gene cluster. After failing to identify the complete gene cluster by standard approaches, the genome of S. ghanaensis was shotgun sequenced and partially assembled to give 1018 contigs covering 7.4 Mb. Genome scanning for candidate genes quickly identified the main moenomycin A gene cluster of 20 open reading frames, which was subsequently sequenced in full on two overlapping cosmid clones and verified by gene disruption and heterologous expression. This split cluster differs from other carbohydrate-containing natural product gene clusters in that most of its biosynthetic building blocks are derived from primary metabolism and thereby represents an alternative paradigm for the biosynthesis of glycosylated natural products.

Sequence analysis of approximately 3.0 Mb of the two chromosomal arms of S. ambofaciens ATCC23877¹⁰⁷ (http://www.weblgm.scbiol.uhp-nancy.fr/ambofaciens/) revealed at least nine secondary metabolic gene clusters, including those for the siderophore coelichelin (5),¹⁶⁹ the structurally unknown angucycline antibiotic alpomycin,^170,171 and the pyrrole-amide antibiotic congocidine¹⁷² (42), whose structure had been determined prior to genome sequencing. Congocidine is synthesized by an atypical iterative NRPS, and its gene cluster represents the first example responsible for the biosynthesis of a pyrrole-amide natural product.

The prenylated alkaloid diazepinomicin/ECO-4601 (43), which contains a very unusual dibenzodiazepinone core that was of unknown biosynthetic origin, was discovered from two Micromonospora strains including one obtained from a marine ascidian.¹⁷³ Genome scanning of strain 046Eco-11 revealed five distinct natural product biosynthetic gene clusters, one of which spanning 50 kb and coding for 42 genes was shown to be involved in the synthesis and coupling of isoprenoid (via the mevalonate pathway), 3-hydroxyanthranilic acid, and 3-amino-5-hydroxybenzoic acid residues in the biosynthesis of 43.¹⁷⁴

Thiopeptide antibiotics such as the prototypical thiostrepton (44), a topical veterinary antibiotic from Streptomyces azureus and Streptomyces laurentii, were long suspected to be of either ribosomal or nonribosomal origin.¹⁷⁵ Attempts since the early 1990s to identify the biosynthetic gene cluster had been unsuccessful¹⁷⁶ until two groups in 2009 independently established that thiostrepton is indeed a highly modified bacteriocin.^177,178 Partial genome sequencing of S. laurentii was pursued in one study to yield 7.26 Mb on over 3600 fragments that led to the identification of the chromosomal gene tsrA encoding the thiostrepton prepeptide.¹⁷⁷ Gene inactivation of tsrA resulted in the loss of thiostrepton biosynthesis and subsequently led to the cloning and sequencing of its gene cluster from two overlapping fosmids. In addition to thiostrepton, other thiopeptide bacteriocin biosynthetic gene clusters recently identified include those for siomycin A from Streptomyces sioyaensis,¹⁷⁸ thiocillin from Bacillus cereus,^178,179 and the thiomuracins from the rare actinomycete Nonomuraea,¹⁸⁰ that collectively illuminate the extensive posttranslational modification reactions associated with these ribosomally-derived peptides.

4 Actinomycete genome mining for new natural product chemical entities

Genome sequencing has also proved to be a powerful method to quickly access the genome-wide natural product biosynthetic capacity of diverse bacteria in order to guide the discovery of new chemical entities.^{52,55,56,181-183} Various strategies have been employed to mine natural product-laden genomes, as exemplified in Section 2 of this article on completed actinomycete genomes. One of the first reports of genome mining from partial sequences of actinomycete genomes involved the characterization of enediyne polyketide chemistry in 2003.¹⁸⁴ Since many polyketides, peptides and hybrids thereof are synthesized on large, modular synthases in an ordered, assembly-line process,^185,186 their products or fragments can be predicted often with high accuracy to greatly aid in their isolation and structure elucidation.¹⁸⁷ Numerous databases and in silico computational approaches have been developed to analyze orphan pathways and assist in the identification of their products.^188-192 Examples abound from the actinomycete literature on mined polyketides, which include the 28-membered macrolide halstoctacosanolide A (45) from Streptomyces halstedii HC34,^193,194 the linear polyketide ECO-02301 (46) from Streptomyces aizunensis NRRL B-11277,¹⁹⁵ the antimicrobial glycosidic polyketide ECO-0501 (47) from the vancomycin-producer Amycolatopsis orientalis in which genomic analysis revealed at least ten genetic loci associated with secondary metabolism,¹⁹⁶ and three new 5-alkenyl-3,3(2H)-furanones such as E-837 (48) from two Streptomyces species.¹²⁰ E-837 from S. aculeolatus NRRL 18422 is identical to actinofuranone A, which was isolated by a standard approach from the marine-derived strain CNQ766,¹⁹⁷ that like S. aculeolatus also produces the type III PKS-derived napyradiomycin family of meroterpenoid antibiotics.¹⁹⁸ Analysis of the recently sequenced genome of the thermophilic actinomycete Thermobifida fusca further led to the discovery of the fuscachelin (49) family of novel NRPS-derived siderophores.¹⁹⁹

5 Outlook

Actinomycetes have long been known for their biosynthetic prowess in the production of complex natural products. In recent years, genomics has transformed the ways in which we think about these chemically prolific microorganisms and has greatly influenced the manner in which we probe their chemistry and biology. We now know that conventional natural product isolation programs vastly underestimated their biosynthetic ability, in many cases by upwards of 80–90%. As a consequence, orphan secondary metabolic gene clusters deduced from genome sequencing efforts represent a large, untapped resource of new chemical entities that promise to illuminate new biology as well as provide a novel source of drug candidates. This latter point is especially relevant to actinomycetes such as Frankia that are not known for their natural product chemistry, but on the basis of genomic analyses are predicted to be nearly as rich in natural products as the streptomycetes. One of the greatest future challenges in this field will be in the development of general methods to awaken these silent gene clusters and characterize the chemical biology of their pathway products. What is certain is that microbial genomics has made an indelible mark in our understanding of the chemistry and biology of the actinomycetes, and that future sequencing efforts will continue to stimulate the broader scientific community.

Biography

Markus Nett (born 1977) studied pharmacy, and obtained his Ph.D. on the subject of natural product chemistry under the supervision of Professor Gabriele M. König at Rheinische-Friedrich-Wilhelms-Universität Bonn. In 2007, he continued with postdoctoral research in the group of Professor Bradley S. Moore, now focussing on the engineering of biosynthetic pathways in marine actinomycete bacteria. In 2008, he gained a fellowship of the German Academic Exchange Service and became a DAAD fellow at the Scripps Institution of Oceanography, San Diego. Since January 2009, he has been head of a junior research group at the Leibniz Institute for Natural Product Research and Infection Biology (Hans-Knöll-Institute) in Jena. His research interests are in exploring the secondary metabolic capabilities of predatory bacteria, mutasynthesis and the role of natural products in microbial communication processes.

Haruo Ikeda was born in Tokyo, Japan (1954), receiving his B.S. (1977) and M.S. (1979) in pharmaceutical sciences at Kitasato University. He obtained his Ph.D. (1982) in pharmaceutical sciences from Kitasato University, where he studied biosynthesis of 16-membered macrolide antibiotics produced by Streptomyces spp. with Professor Satoshi Omura. He began studying Streptomyces genetics with Professors Sir David A. Hopwood and Keith F. Chater at the Department of Genetics, John Innes Institute, UK, as a postdoctoral fellow, and then took a faculty position in the School of Pharmaceutical Sciences, Kitasato University (1983–2002), becoming Associate Professor in 1992. Since 2002 he has been a full Professor at Kitasato Institute for Life Sciences and the Graduate School of Infection Control Sciences, Kitasato University. His research interests are in studying the biosynthesis of microbial secondary metabolites using bioinformatics and in engineering and designing the Streptomyces chromosome for the evaluation and optimization of the secondary metabolism.

Bradley S. Moore is currently Professor of Oceanography and Pharmaceutical Sciences at the Scripps Institution of Oceanography and the Skaggs School of Pharmacy and Pharmaceutical Sciences at University of California, San Diego. He was first introduced to natural product research as a chemistry undergraduate student at the University of Hawaii, where he explored the chemistry and biosynthesis of cyanobacterial natural products with the late Prof. R. E. Moore. Fascinated by the beauty and complexity of natural product structures, he went on to conduct graduate (Ph.D. 1994 in bioorganic chemistry with Prof. H. G. Floss at the University of Washington) and postdoctoral research (1994–1995 with Prof. J. A. Robinson at the University of Zürich) on the biosynthesis of bacterial natural products. Prior to moving to the University of California at San Diego in 2005, he held academic appointments at the University of Washington (1996–1999) and the University of Arizona (1999–2005). His research interests involve exploring and exploiting marine microbial genomes to discover new biosynthetic enzymes, secondary metabolic pathways, and natural products for drug discovery and development.