Antibiotics from rare actinomycetes, beyond the genus Streptomyces

Throughout the golden age of antibiotic discovery, Streptomyces have been unsurpassed for their ability to produce bioactive metabolites. Yet, this success has been hampered by rediscovery. As we enter a new stage of biodiscovery, omics data and existing scientific repositories can enable informed choices on the biodiversity that may yield novel antibiotics. Here, we focus on the chemical potential of rare actinomycetes, defined as bacteria within the order Actinomycetales, but not belonging to the genus Streptomyces . They are named as such due to their less-frequent isolation under standard laboratory practices, yet there is increasing evidence to suggest these biologically diverse genera harbour considerable biosynthetic and chemical diversity. In this review, we focus on examples of successful isolation and genera that have been the focus of more concentrated biodiscovery efforts, we survey the representation of rare actinomycete taxa, compared with Streptomyces , across natural product data repositories in addition to its biosynthetic potential. This is followed by an overview of clinically useful drugs produced by rare actinomycetes and considerations for future biodiscovery efforts. There is much to learn about these underexplored taxa, and mounting evidence suggests that they are a fruitful avenue for the discovery of novel antimicrobials.


Introduction
The phylum Actinomycetota, formerly known as Actinobacteria [66], includes unicellular and filamentous bacteria that are commonly referred to as actinomycetes.The former includes human pathogens such as Mycobacterium tuberculosis (Mtb) and the industrial amino acid producer Corynebacterium glutamicum, whereas filamentous actinomycetes are studied for their complex developmental life cycles and their production of specialised metabolites.The type genus for this phylum is Streptomyces, which belongs to the filamentous group and includes more than 1100 verified species, producing specialised metabolites that form the basis of ∼50% of clinically used antibiotics [42].Their specialised metabolites are also used as cancer therapeutics (doxorubicin, daunorubicin), antifungal compounds (amphotericin, nystatin), immunosuppressants (rapamycin, FK506) and antiparasitic agents (avermectins).The genes encoding the biosynthesis of these molecules are typically grouped together in biosynthetic gene clusters (BGCs) alongside genes involved in immunity, regulation and transportation.As discussed below, a recent survey of bacterial (meta)genomes revealed that only three percent of their BGCs have been matched to metabolites, suggesting there are many more useful pathways and chemical products waiting to be discovered [34].
A recent survey of microbial genomes reported that Streptomyces encodes the greatest chemical diversity within the Actinomycetes, but that Amycolatopsis and Micromonospora also have significant biosynthetic potential [34].As shown in Figure 1, the number of sequenced Streptomyces strains far exceeds all other filamentous actinomycete genera, with the next most common being Amycolatopsis and Micromonospora, this is discussed further below.Other genera include Pseudonocardia that is well-studied largely because of its mutualistic symbiosis with fungus-farming ants in the tribe Attini, which are endemic to South and Central America [7].The queen of an ant colony vertically transmits a single strain of mutualist Pseudonocardia to worker ants, who cultivate the bacterium on their cuticles, providing it with food and shelter in return for antibiotics; these antibiotics influence the ant microbiome and provide a means to protect themselves and their food supply from Escovopsis, a genus of parasitic fungi [101].Other filamentous actinomycetes have also been isolated from these ants, including Amycolatopsis and Streptomyces [49,78,81,101], but the stable association is typically with Pseudonocardia species [15].Additionally, species of Saccharopolyspora and Streptomyces have been isolated from fungus-growing ants in Kenya and have yielded novel antibiotics [73,79].
Evidence is accumulating that free-living soil actinomycetes have also adapted to colonise plants and animals, as observed in the well-studied Pseudonocardiaattine ant, Streptomyces-beewolf, and Streptomyces-bark beetle symbioses [80,93].Rare actinomycetes have been isolated from diverse environments, including soil, freshwater, marine animals and deep-sea sediments, and from the roots of plants and trees.In terrestrial ecosystems, plant root exudates may provide nutrients, so it is perhaps unsurprising that many species of Pseudonocardia [74], Micromonospora [104], Saccharopolyspora [76] and Streptomyces have been isolated from plant roots.In some cases, they have been shown to be highly enriched in the rhizosphere and endosphere of plant roots [71].
The marine environment has also been shown to be a rich source of rare actinomycetes [50].In fact, many of the 31 verified Saccharopolyspora species were isolated from marine environments and 16 have been described as halotolerant or halophilic [76].The obligate marine genus Salinispora is perhaps one of the most comprehensively studied for their biosynthetic and chemical potential and has been proposed as a model organism for specialised metabolite discovery [44].The potent proteasome inhibitor, salinosporamide A [32], produced by Salinispora tropica [43], is currently in phase-three clinical trials as an anticancer drug [8].Aside from the structurally diverse and novel chemistry produced by this genus, a culture collection of thousands of strains [75] has enabled a comprehensive assessment of the diversity and evolution of specialised metabolism across the Salinispora genus [28,51,108].
Despite the success stories from the genera studied in detail, many remain vastly understudied, this includes Catellatospora, of which three new plant-growth-promoting ansamacrolactams have recently been discovered [54] and Saccharothrix, where strains from the marine environment were found to produce eight novel, cytotoxic amphiphilic siderophores [84].Genome sequencing and mining of existing genomic databases combined with availability through culture collections can certainly aid the prioritisation of strains.

Chemical and biosynthetic potential
As noted above, it has been estimated that only three percent of bacterial BGCs have been matched to known products [34].Among bacteria, the Actinomycetota phylum harbours a particularly high biosynthetic potential, dedicating up to 3.0 Mb of coding capacity, over 15% of their genome, to specialised metabolite biosynthesis [3,18].Despite the biosynthetic potential of this phylum, fewer than 10% of the sequenced microbial genomes available in public repositories belong to this group [45].Furthermore, it has been estimated that currently only 30-50% of the biological diversity of this phylum is represented by sequenced strains [83].The potential of the Streptomyces genus, as mentioned previously, is reflected in the high number of BGCs with their genomes [46].In fact, a comparison of biosynthetic diversity calculated as unique gene cluster families showed that the Streptomyces genus is responsible for most of the phyla-level biosynthetic diversity [34].Another comparative analysis based on the abundance of BGCs per genome, identified the orders Pseudonocardiales, Streptosporangiales and Micromonosporales as gifted taxa, with an average of 19.8, 15.0 and 13.3 BGCs per genome, respectively, in comparison with the 21.6 calculated for the order Streptomycetales [27].However, although rare actinomycetes usually have fewer BGCs per genome, their novelty and diversity could be higher than Streptomyces.For example, a study performed with marine-derived actinomycetes showed that when gene diversity was normalised by the number of BGCs, it was higher in rare actinomycetes, with the genera Corynebacterium, Gordonia, Nocardiopsis, Saccharomonospora and Pseudonocardia representing the highest BGC diversity [77].This study also determined that phylogeny and genome size combined can be used to predict the likelihood of BGC diversity.Similarly, recent work that applied a genomic metric called biosynthetic novelty index (BiNI) to assess BGC novelty was applied to actinomycetes strains from various taxonomic and ecological backgrounds and revealed that Streptosporangium, Lentzea, Actinokineospora and Saccharothrix genera exhibited the highest novelty index [35].Similarly, a Nonomuraea sp. was recently found to harbour the largest actinomycete chromosome encoding approximately 32 BGCs [65].However, not all rare actinomycetes appear to be as talented.For example, genome sequencing of genera such as Micrococcus and Agrococcus demonstrated less than five predicted BGCs per genome [86], and whilst this does not preclude these metabolites from being useful, it does highlight the incredible variation within this group.
The frequent rediscovery of known specialised metabolites typically comes from screening multiple isolates of common Streptomyces strains [96].Representatives of the Actinomycetota, and their numbers of predicted BGC are shown in Figure 1, with Micromonospora, Salinispora, Actinomadura, Nocardiopsis, Amycolatopsis, Saccharopolyspora and Pseudonocardia identified as common 'rare actinomycete' genera across the three databases (Figure 1).The antiSMASH database (anti-SMASH DB), which provides pre-calculated anti-SMASH annotations for completed, publicly available bacterial genomes [10,11], shows that 38% of Actinomycetota BGCs belong to the Streptomyces genus (Figure 2).The prevalence of Streptomyces-encoded specialised metabolites increases when manually curated metadata, such as specialised metabolite chemical identity, are included.In the Minimum Information about a Biosynthetic Gene cluster (MIBiG) database [88], 74% of Actinomycetota BGCs belong to the Streptomyces genus.A similar situation is observed for the Natural Products Atlas (NPAtlas), a database of known microbial-specialised metabolite structures [94], where 72% of Phylogenetic tree generated by AutoMLST [1] showing genome size, number of BGCs predicted by antiSMASH and the proportion of these that are predicted to be more than 10% of a match to previously characterised clusters identified by antiSMASH (in the MIBiG database).The input to AutoMLST was 20 representitive, compleate genomes of genera discussed in this review.
specialised metabolites were isolated from Streptomyces strains.As such, even though antiSMASH DB contains a considerable number of rare actinomycete BGCs, only a very small proportion of their products has been chemically characterised (i.e.correlated MIBiG and NPAtlas database results).
Of the 45,956 Actinomycetota genomes in the NCBI database, 3234 are complete genome sequences and only 630 of these are NCBI reference sequences of rare actinomycetes [72].To highlight the biosynthetic potential of rare actinomycetes, reference genomes from diverse genera were selected, including where possible, those mentioned in this review.The abundance and phylogenetic distribution of (un)characterised BGCs harboured by rare actinomycetes is shown in Figure 2.This analysis reinforces common knowledge, that is, that Streptomyces sp.typically harbour more BGCs that have already been characterised or are highly matched to those that have, further illustrating a lower potential for novel discoveries.For example, S. venezuelae and S. coelicolor showed 24/30 and 14/21 'matched' BGCs, respectively.In contrast, the rare actinomycete Saccarothrix syringae shows a remarkable number of predicted BGCs.AntiSMASH predicted that the 10.88-mb genome contained 56 BGCs, of which 20 of these had 10% or less similarity to known BGCs.In addition, although there are strains with a lower number of predicted BGCs, for example, Gordonia insulae only encodes 16 predicted BGCs in its 5.96-Mbp complete genome, 10 of them are predicted to be novel BGCs.Whilst these types of analyses have caveats, including the assumption of accurate assembly, large-scale genome mining data have confirmed that there is extensive novel biosynthetic potential encoded in rare actinomycete genomes.

Clinically used drugs from rare actinomycetes
In total, five classes of clinically used antibacterial drugs are derived from rare actinomycetes, several of which are Chemical structures of antibacterial drugs derived from natural products of rare actinomycetes.Rifampicin is a semisynthetic molecule derived from rifamycin S, and the active pharmaceutical ingredient (API) of gentamicin and teicoplanin comprises a complex mixture of congeners, the most abundant of which are shown in each case.
on the WHO list of Essential Medicines [99] (Figure 3, Table 1).The oldest of these is the macrolide antibiotic erythromycin produced by Saccharopolyspora erythraea, introduced into clinical use over 70 years ago [59].Macrolide antibiotics, with a 14-atom ring size, inhibit protein synthesis and are particularly useful for the treatment of upper respiratory tract infections.Yet, erythromycin suffers from acid instability that is linked to its common gastrointestinal side effects and action as a motilin agonist [97].These and other limitations are overcome to some degree by a range of semisynthetic variants of which clarithromycin and azithromycin are the most successful [26].This motilin agonist activity means erythromycin is commonly used off-label for gastroparesis, and its immunomodulatory activity has led to low-dose off-label use in Southeast Asia for the treatment of diffuse panbronchiolitis [53].Intriguingly, nature has produced an acid-stable erythromycin congener called sporeamicin, which is also the product of a Saccharopolyspora species [103], but the biosynthetic steps that differentiate these pathways remain unresolved.
The glycopeptide vancomycin produced by Amycolatopsis orientalis [58] inhibits cell wall biosynthesis by binding to the lipid-II precursor.This is prescribed for lifethreatening infections by Gram-positive bacteria that are resistant to other treatments: this reserved use stems, in part, from a range of serious side effects.Glycopeptides are poorly absorbed by the gastrointestinal tract, meaning they are mostly restricted to intravenous use, except for Clostridioides difficile infection (CDI) treatment.The related glycopeptide teicoplanin, produced by Actinoplanes teichomyceticus, differs most notably in two of the seven amino acids that form its backbone, and in the glycosylation pattern [16,67].Semisynthetic modification of these and the related A40926 family of glycopeptides, produced by an Actinomadura sp., led to the development of the secondgeneration lipoglycopeptide drugs telavancin, oritavancin and dalbavancin [9].Rifamycin, first isolated from Amycolatopsis rifamycinica in 1959, is the founding member of the ansamycin family of polyketides [82].It is an inhibitor of the DNA-dependent RNA polymerase, but the metabolite itself lacks the properties required for effective pharmacological application, meaning semisynthetic modification was required to produce successful drug molecules [33].The most widely used is rifampin that was first marketed in 1968 and has become a first-line antituberculosis therapy when used in combination with other agents.The rifamycin congener kanglemycin A produced by Amycolatopsis spp.maintains potency against RNA polymerases containing rifampicin-resistant mutations and has become the subject of significant recent interest as a lead for semisynthetic optimisation [61,69].The aminoglycoside antibiotic gentamicin is produced by Micromonospora purpurera and was introduced into clinical use in 1964 [98].Aminoglycosides can be highly effective for the treatment of Gram-negative pathogen infections, in particular sepsis, and target the bacterial ribosome 30S subunit, inhibiting protein synthesis by interfering with initiation, codon fidelity and translocation.However, their use is often limited due to side effects, and gentamicin displays significant renal and ototoxicity.The pharmaceutical formulation gentamicin C comprises a complex of five congeners and there is evidence to suggest that individual components may have lower toxicity, which has led to efforts for pathway engineering [48].The mechanism of action, which interferes with codon fidelity, means gentamicin is being repurposed for the treatment of genetic diseases such as Duchenne muscular dystrophy in which mutations lead to the presence of premature stop codons leading to truncated proteins [95].
The most recently approved antibacterial drug (2011) to come from a rare actinomycete is fidaxomicin for CDI, an inhibitor of RNA polymerase that inhibits the same enzyme but acts through a different mechanism to rifamycin-based molecules [29].The active entity of fidaxomicin is the 18-membered macrolide tiacumicin produced by Dactylosporangium aurantiacum subsp.hamdenesis.Tiacumicin B was first discovered in 1986 [89] and is part of a larger group of more than 40 related rare actinomycete-specialised metabolites, the first being lipiarmycin that was discovered in 1975 from Actinoplanes deccanensis [21,68].This delay in development stems from poor oral bioavailability limiting broad-spectrum application, but this would later prove beneficial for the narrow-spectrum treatment of CDI and the resurrection of this compound class.
Beyond the approved drugs, peptide antibiotics produced by rare actinomycetes have proven very successful as leads for the treatment of CDI [70].These include the glycolipodepsipeptide ramoplanin that is a cell wall-targeting antibiotic produced by Actinoplanes ramoplaninifer [23,56].LFF-571 is a semisynthetic derivative of the thiopeptide GE2270A produced by Planobispora rosea that inhibits protein synthesis via bacterial elongation factor thermal unstable (EF-Tu) [62], and NVB302 is a semisynthetic analogue of the type-B lantipeptide deoxyactinogardine produced by Actinoplanes liguriae that binds the cell wall precursor lipid II [12].All three peptides are narrowspectrum agents for the treatment of CDI, but while ramoplanin is being fast-tracked through Phase-III clinical trials, the development of LFF-571 and NVB302 appears to have halted.Very recently, semisynthetic variants of the unusual macrolide sequanamycin produced by Allokutzneria albata were reported as candidates with utility against Mtb that overcomes the inherent macrolide resistance of Mtb [105].

Considerations for the future study of rare actinomycetes
Recent advances in genetic modification systems for actinomycetes have been the subject of several review papers [24,60], including genetic modification focussed on cloning and heterologous expression in Streptomyces species [64,102].Protocols for genetic engineering of actinomycetes, as well as vector maps, are freely available on http://actinobase.org[23,31] where you can also download a free pdf copy of the Practical Streptomyces Genetics manual, first published in 1985, and updated in 2000 [47].Genetic engineering has provided a promising avenue for exploiting the biosynthetic potential of microorganisms, but in the case of actinomycetes, most of the protocols were developed for Streptomyces species, and are not generally applicable to other genera without modifications [55].For example, while site-specific recombination systems based on bacteriophage integrases have been a fundamental tool for Streptomyces genetic engineering, mediated by phage-based vectors such as pSET152 and pMS82, the application of these tools in rare actinomycetes depends largely on the presence of the same genetic elements [31,37].Bacteriophage integration occurs between DNA sequences on the phage attachment site and the host genome on the bacterial attachment site (attB) [20].While innate attB sites are regularly found in Streptomyces genomes, they are less conserved in rare actinomycetes [4], but an attB sequence in Pseudonocardia alni, with an identity of 89% to the canonical attB sequence in S. coelicolor, facilitated its genetic engineering to produce a host mutant ( [52], p. 200).In many cases, the attB site could be also introduced before recombination and this strategy was successfully applied to produce a Salinispora tropica mutant [106,107].The yeast meganuclease I-SceI has also been introduced into Streptomyces species on an integrative phage vector and used to make targeted double-strand breaks, but this is mediated by a selfreplicating plasmid and such plasmids would need to be identified and/or developed for use in rare actinomycetes [85].
In the absence of tools to make precise gene deletions, it is sometimes possible to disrupt genes using a suicide vector approach, that is, by introducing a non-replicating vector through conjugation with E. coli.A suicide vector approach to disrupt the nystatin P1 antifungal BGC was successful in Pseudonocardia P1 [6].The disruption plasmid harboured an ∼1500-bp internal fragment of nypI polyketide synthase as well as an apramycin resistance cassette and conjugal origin of transfer, which was used to mobilise the plasmid from E. coli to Pseudonocardia P1.Integration of the disruption plasmid into nypI abolished the ability of Pseudonocardia P1 to inhibit Candida albicans in a whole-cell bioassay (Figure 4).Recent advances in Clustered Regularly Interspaced Short Pallindromic Repeats-associated protein 9(CRISPR-Cas9)-based engineering also represent an opportunity to precisely edit the genomes of rare actinomycetes, spanning multiple genera [40,91].For example, CRISPR-Cas9-based techniques have been successfully applied in Actinoplanes [100] and Micromonospora [13,19], but again this is dependent on successful conjugation and plasmid replication in these bacteria.If conjugation is not possible, then polyethylene glycol (PEG)-mediated protoplast transformation is another historically important tool in Streptomyces genetics that has successfully applied in rare actinomycetes.However, the efficiency for protoplast formation and regeneration varies widely depending on the strain and seems to be less effective than in Streptomyces species [63].
Despite evidence that vertical inheritance is critical for BGC diversity [17], it is widely accepted that novel habitats, lifestyles or ecological interactions of rare actinomycetes may influence their potential to harbor novel BGCs and thus in turn, influence the structural diversity of their produced metabolites.This functional diversification can result from both abiotic and biotic ecosystem pressures, and the latter includes drivers such as signalling, protection and defence.Even with these widely accepted views, studying chemical ecology linked to function is a challenge and there is limited evidence to pinpoint this to biosynthetic diversification.Studies showing the recruitment of BGCs as a strategy for 'sampling from the environment' before becoming more 'fixed' in a population are still in their infancy [14,108], yet they provide an exciting step forward in our understanding of ecology and BGC evolution.
In order to incorporate ecological thinking across taxa, we must first improve out ability to capture these lessstudied species.Part of the challenge of working with rare actinomycetes, is they are harder to grow and often grow more slowly under standard culture conditions compared with Streptomyces strains.For example, in our experience, Pseudonocardia strains isolated from attine ants take 2-4 weeks to grow, compared with 2-4 days for many Streptomyces species, and Pseudonocardia bacteria can typically only be cultured on solid agar.In soil, Streptomyces species have been found to be the dominant group among the total actinomycete population, although of course this could be biased based on approach [5,36].However, selective isolation of a particular rare actinomycete genus must involve techniques that enhance the growth of desirable actinomycetes (enrichment) and eliminate Streptomyces and other undesirable taxa from the isolation media [39,90].There have been several relatively recent advances in the area of isolation and culturing of previously unculturable organisms.One example that was successful for Nanohaloarchaeota was fluorescent in situ hybridisation and fluorescence-activated cell sorting to specifically isolate live cells of interest, allowing them to be cultured [38].Highthroughput dilution to extinction experiments has been effective with marine organisms by avoiding the need for artificial media whilst removing other, potentially fastergrowing microorganisms [38].The availability of metagenomes also enables scientists to specifically target organisms of interest.For example, Cross and colleagues described a method termed 'reverse genomics isolation' in which cell surface proteins were predicted from sequencing data and allowed for specific target antibodies to be generated [22].Moving forward, a range of techniques, old and new, will be required along with communication of results.One solution is the sharing of methods and protocols through community platforms such as http://actinobase.org[31].Another solution is to adapt a culturing approach that mimics their natural environment more closely, such as soil mesocosms and microcosms, as was demonstrated through the impact of bacterial community structure on the metabolome of marine sediments [92].Another strategy to elicit new chemistry is that of co-culture because it is well-known that microorganisms exist in complex multi-species/ kingdom niches and, as such, competition and interactions are commonplace.There are several excellent pioneering ecology studies focussed on such interactions, including mutualism and competition, for example, interactions of the genus Couchioplanes [57], and such approaches could be particularly useful if the goal is to elicit specialised metabolites with antibiotic potential.
In summary, rare actinomycetes are under-represented in natural product databases, they exhibit more BGC diversity than streptomycetes and their potential is often undervalued.While we have outlined some considerable experimental hurdles to working with these genera, including genetic tractability, isolation and metabolite elicitation, they harbour considerable biosynthetic potential and it is worthwhile developing tools to exploit these bacteria.Their taxonomic and ecological diversity is both a blessing and a curse and success will come from shared knowledge across strains and communities, and from perseverance.There are some taxonomic starting points for rare actinomycete natural product discovery, for example, from Pseudonocardia, Salinispora and Micromonospora species, and community data sharing will enable us to evaluate rare actinomycete genera and thus be informed about where efforts are best spent.In the future, these understudied taxa may be anything but rare.

Figure 1 Current
Figure 1

Figure 2 Current
Figure 2

Figure 3 Current
Figure 3

Figure 4 Current
Figure 4