Exploring halophilic environments as a source of new antibiotics

Abstract

Microbial natural products from microbes in extreme environments, including haloarchaea, and halophilic bacteria, possess a huge capacity to produce novel antibiotics. Additionally, enhanced isolation techniques and improved tools for genomic mining have expanded the efficiencies in the antibiotic discovery process. This review article provides a detailed overview of known antimicrobial compounds produced by halophiles from all three domains of life. We summarize that while halophilic bacteria, in particular actinomycetes, contribute the vast majority of these compounds the importance of understudied halophiles from other domains of life requires additional consideration. Finally, we conclude by discussing upcoming technologies- enhanced isolation and metagenomic screening, as tools that will be required to overcome the barriers to antimicrobial drug discovery. This review highlights the potential of these microbes from extreme environments, and their importance to the wider scientific community, with the hope of provoking discussion and collaborations within halophile biodiscovery. Importantly, we emphasize the importance of bioprospecting from communities of lesser-studied halophilic and halotolerant microorganisms as sources of novel therapeutically relevant chemical diversity to combat the high rediscovery rates. The complexity of halophiles will necessitate a multitude of scientific disciplines to unravel their potential and therefore this review reflects these research communities.

Keywords:

• Halophiles
• halophilic bacteria
• antimicrobial discovery
• actinomycetes
• natural products

Introduction

Life exists in almost all environmental niches on the earth. Our planet plays host to various unique environments, and the diversity of these ecosystems is reflected by the diversity of microbial communities that exists within them (Rothschild and Mancinelli 2001; Maheshwari and Saraf 2015). Extremophiles are organisms that survive and even thrive in bizarre and extreme ecosystems – both physical (temperature, radiation, or pressure) and geochemical extremes (desiccation, salinity, pH, oxygen species, or redox potential) (Rothschild and Mancinelli 2001). Extremophiles are some of the oldest life forms on earth. Their existence is thought to be an evolutionary adaptive response to the extreme environments that were present in the earth’s primordial atmosphere (Rothschild and Mancinelli 2001). The selective nature of these environments resulted in the evolutionary development of the unique metabolism thought to be fundamental to their survival (Maheshwari and Saraf 2015). Today, extremophiles live in habitats that will support no other form of life.

Halophiles are extremophiles that can exist in a range of salt concentrations, from freshwater habitats to the hypersaline lakes of the Dead Sea, and even the salt flats of Salar de Atacama (Rothschild and Mancinelli 2001). The world of halophilic microorganisms is highly diverse and contains taxonomic representatives from all three domains of life – Archaea, Bacteria, and Eucarya (Oren 2002).

Originally thought to be bacterial, their distinctive 16S ribosomal RNA sequences and lipid compositions revealed that archaea were a separate and distinct third domain of life (Woese and Fox 1977). Archaea are comprised of three phylogenetically distinct groups: The Crenarchaeota (mainly consisting of hyperthermophilic sulphur-dependent organisms), the Euryarchaeota (the methanogens), and the extreme halophiles. The classification of archaea is a rapidly evolving one, most recently, a superphylum Asgard was proposed whose members include Lokiarchaeota and Thorarchaeota (Zaremba-Niedzwiedzka et al. 2017), with the eventual cultivation of an Asgard archaeon Candidatus Prometheoarchaeum syntrophicum after 12 years of growth (Imachi et al. 2020). Numerous genera of archaea live and thrive in extreme niches, and therefore a significant portion of halophilic microorganisms belongs to the Archaea domain (Kessel and Klink 1980).

Among the bacteria, halophiles exist within the phyla Cyanobacteria, α-Proteobacteria, Firmicutes, Actinobacteria, Spirochaetes, and Bacteroidetes (Sorokin et al. 2006; Oren 2008). Halophilic bacteria offer a high potential for biotechnological applications due to their adaptations to extreme natural environments particularly concerning their contribution to biogeochemical processes, the formation, and dissolution of carbonates, the immobilization of phosphate, and the production of growth factors and nutrients (Rodriguez-Valera 1993).

Eukaryotic organisms contribute to a smaller percentage of life in hypersaline environments; representatives include the unicellular green algae Dunaliella (Wegmann et al. 1980), the brine shrimps Artemia salina and Artemia franciscana (Gajardo and Beardmore 2012), and the fungus Trimmatostroma salinum and Hortaea werneckii (Gunde-Cimermana et al. 2000).

Through evolution, these salt-loving organisms have adapted to the high osmotic pressures that exist within their environments and their unique biochemical and genetic characteristics have resulted in the production of a multitude of biotechnologically relevant bioactive compounds including antimicrobials (Maheshwari and Saraf 2015).

Once considered the “wonder drugs” of the twentieth century (Bud 2007), the disproportionate and indiscriminate usage of antibiotics has accelerated the emergence of clinically significant multiple drug-resistant (MDR) pathogenic strains (Ciofu et al. 1994). The dynamic nature of pathogenic microorganisms to evolve sophisticated mechanisms that inactivate antibiotics (Butler et al. 2017), and the continued global recognition that the emergence of MDR pathogens pose a serious threat to human health, have substantially emphasized the urgent need for new antibiotics that would target this emerging crisis (Butler et al. 2017). Pathogenic strains that are resistant to most, or all, available antibiotics are now routinely isolated (Brown and Wright 2016).

Microbial natural products

The beginning of the modern “antibiotic era” is often associated with Paul Ehrlich and Alexander Fleming, replacing the antibacterial sulphonamide, and the toxic arsphenamine/606. Ehrlich with his “magic bullet” argued that chemical compounds could be synthesized and would “exert their full action exclusively on the parasite harbored within the organism” (Aminov 2010). While Sir Alexander Fleming’s serendipitous discovery of penicillin in 1928 (Fleming 1929), would revolutionize medical practice, dramatically decreasing fatality rates from bacterial infections (Kovac and Sneader 2006) and in the midst of World War Two they saved millions of lives because “Thanks to Penicillin…He Will Come Home!” (Levy 2001).

Fleming’s discovery kick-started a concerted search for other antimicrobial agents. Over the next 30 years, the “Golden age” (1940–1962) of antibiotic discovery, resulted in the development of half of the clinically used antibiotics today (Figure 1). Throughout the remainder of the twentieth century, two parallel lines of drug discovery were employed; the isolation of antibiotics from natural sources and the chemical synthesis of antibiotics (Davies 2006).

Figure 1. Timeline depicting the year in which antibiotic resistance in each major class of antibiotics was first observed. The main mechanism of resistance for each antibiotic class is also included.

Microbial natural products have made an incredible contribution to the antibiotic drug discovery and development process over the last seven decades, with 70 out of 90 antibiotics marketed in the years 1982–2002 originating from natural products (Newman et al. 2003). Between 1981 and 2010, 34% of all medicines approved by the FDA were natural products or their derivatives (Newman and Cragg 2016). Synthetic antibiotics followed James Black’s famous observation that “the most fruitful basis for the discovery of a new drug is to start with an old drug” (Raju 2000).

Almost all of the antibiotics approved between the early 1960s and 2000 were iterative, semisynthetic tailored derivatives of previous natural scaffolds (Figure 2) (Von Nussbaum et al. 2006; Fischbach and Walsh 2009). The beginning of a discovery drought, with a void in antibiotic discovery since lipopeptides in 1987 (Friedman and Alper 2014), marked the end of the “Golden Age” of antibiotic drug discovery, the law of diminishing returns had forced the hands of the drug discovery sector to increasingly turn their efforts towards high-throughput screens of synthetic compounds. However, the chemical diversity offered by synthetic and combinatorial chemical libraries was not sufficient to provide the required number of quality hits for hit-to-lead programs in drug discovery and efforts were largely unsuccessful (Demain 2014).

Figure 2. Successive generations of major antibiotic classes over time. Black bonds represent the original skeletal structure while red bonds are structural changes.

The milestone of the failed synthetic approach was noted in 2007 when Payne et al. detailed the unsuccessful hits of a huge screening program by GSK (Payne et al. 2007). This effect was amplified at the turn of the century with further disengagement and withdrawal by pharmaceutical companies in antibiotic research and development, which had previously played a significant role in bringing many compounds to market (Koehn and Carter 2005).

Numerous factors prompted this withdrawal, including high rediscovery rates of known natural products and the challenges associated with the purification and structure elucidation of natural products (Clardy et al. 2006). Economically, antibiotics have a poor return on investment due to their short-term curative nature compared with chronic drug therapies (Projan and Shlaes 2004). Additionally, regulatory hurdles within clinical trials (Projan 2003; Brown 2013; Lewis 2013), increased market competition from more affordable generic drugs, and marketplace and industry consolidation (Projan and Shlaes 2004), has led to the virtual collapse of antimicrobial research and development by the pharmaceutical industry (Simpkin et al. 2017). Even within research companies, compound libraries employing inappropriate filters such as Lipinski’s rules (to attain desirable physicochemical properties for improved oral bioavailability of drugs (Lipinski et al. 1997)), ignore the principles required for targeting prokaryotes (Lewis 2013). For the companies that remain in the area, cautionary tales with antibiotics such as solithromycin (Owens 2017) demonstrate the risk of antibiotic research, and recent insolvency in smaller start-up companies such as Achaogen (Dall 2019) pose a warning that stifles antimicrobial discovery.

The emergence of AMR and the lack of novel antibiotics in the pharmaceutical pipeline has prompted a resurgence to screen natural products, which have provided and continue to provide a significant number of clinically approved drugs (Newman and Cragg 2012; Thaker et al. 2013). There is a necessity to improve the efficiency of the discovery process, and bioprospecting of antibiotics from underexplored habitats represents a largely untapped repository of compounds with unique chemical structures (Clardy et al. 2006; Mohammadipanah and Wink 2016).

Within extreme environments, natural forces already eradicate a large percentage of the competition by creating a habitat in which it is difficult to thrive. Therefore there would be a reduced necessity for the remaining microbes to utilize secondary metabolites for competition (Pettit 2011). However, recent studies have demonstrated that extremophiles are capable of producing a diverse range of secondary metabolites (Wilson and Brimble 2009). The hypersaline environment is one such niche, and the following sections explore the antimicrobial potential of each halophilic domain – the Archaea, the Bacteria, and the Eukarya.

Halophiles as a source of antibiotic chemical diversity

Halophilic and halotolerant bacteria

When compared to their terrestrial and marine counterparts, bacteria from hypersaline environments are largely under-explored as a source of antimicrobial agents. Previous studies have described the antimicrobial activity of halophilic bacteria from marine invertebrates, solar salterns, soil, and salt lakes. A predominant genus of antimicrobial-producing strains are Bacillus (Pinchuk et al. 2002), but other species including, Halomonas salifodinae (Velmurugan et al. 2013), halophilic cyanobacteria (Pramanik et al. 2011) have also been described.

Kamat et al. (2011) demonstrated the antimicrobial potential of a range of marine Bacillus spp. from the salt pans of Goa, India. In a study by Velho-Pereira, (Velho-Pereira and Furtado 2012) 100 halophilic bacteria from marine invertebrates were screened for their antibacterial activity. Forty-six of the isolates with the most potent activity belonged to the genera Chromohalobacter, Bacillus, and Corynebacterium. A further study confirmed their abundance, which revealed that amongst the 200 halophilic bacteria isolated, the dominant genus was Bacillus compromising 24% of the total isolates (Irshad et al. 2014). Sawale et al. (2014) reported similar findings with two halophilic bacterial strains, Bacillus pumilus NKCM 8905 and Bacillus pumilus AB211228 displaying antimicrobial production against Staphylococcus aureus, Bacillus subtilis, Aspergillus niger, and Aspergillus flavus.

Li et al. cultured 152 halophilic bacterial isolates from Yuncheng Salt Lake, China; 34 of the isolates were related to the phylum Firmicutes with representatives in three families Bacillaceae, Clostridiaceae, and Staphylococcaceae, while other strains belonged to the phylum γ-Proteobacteria and were identified as members of Halomonadaceae and Idiomarinaceae. Six of the isolates were capable of inhibiting both bacterial and fungal pathogens, the majority of which belonged to the Bacillaceae family (Li and Yu 2015).

However, Chen et al. (2010) reported not only the bioactivity of halophilic Bacillus, but also the potential of other moderately halophilic bacterial strains, with 23 of the 45 strains inhibiting the growth of Bacillus subtilis, one strain inhibiting Escherichia coli, and two strains inhibiting Pseudomonas fluorescens. Additionally, almost all the halophilic strains had inhibitory effects on one or several indicator fungal strains. Phylogenetic analysis revealed that only one of the strains was related to the genus of Bacillus, and 12 strains were related to Halobacillus.

Halophilic microorganisms have also been shown to possess bioactivity beyond antibacterial and antifungal activities as reported by Velmurugan et al. (2013). The organic extracts of Halomonas salifodinae MPM-TC displayed not only antibacterial activity against several aquatic bacterial pathogens (Vibrio harveyi, Vibrio parahaemolyticus, Pseudomonas aeruginosa, and Aeromonas hydrophila) but also possessed in vitro antiviral activity against white spot syndrome virus and immunostimulatory activity in prawns.

The saline ecosystem of Sundarbans’ mangrove forests serves as an example of a unique environment with the potential for antimicrobial discovery. Pramanik et al. (2011) reported the first study of antimicrobial production from obligate halophilic cyanobacteria against several human pathogens. Marine cyanobacteria are known for their ability to produce a diverse array of biologically active secondary metabolites, which have attracted significant research interest due to their potent biological activities. Some of these compounds such as cryptophycin 52 (LY355703) advanced to Phase II clinical trial, but failed due to its high neurotoxicity (Weiss et al. 2017), and curacin A was evaluated for its potential therapeutic as an antimitotic agent (Simmons et al. 2005). In addition to the antimicrobial potential of halophilic cyanobacteria, a further study isolated 54 halophilic actinomycetes with nine of the isolates showing antimicrobial activity against both bacterial and fungal test organisms (Sengupta et al. 2015).

Halophilic and halotolerant Actinobacteria

This bacterial phylum, the Actinobacteria, has proven to be a vibrant source of microbial natural products. Since the discovery of streptomycin from Streptomyces griseus by Selman Waksman’s group in 1942 (Schatz et al. 1944), the greatest number of clinically relevant antibiotics discovered have been from the phylum Actinobacteria (Baltz 2007). Actinomycetes account for 45% of all known microbial secondary metabolites (Bérdy 2005) and over two-thirds of the clinically useful antibiotics (Lam 2006; Papagianni 2012) including gentamicin, vancomycin, clindamycin, erythromycin, amphotericin, rifampicin, and tetracycline (Baltz 2007). Actinobacteria are recognized as being the richest microbial source of natural products, their unique chemical scaffolds (Bérdy 2015) has meant that they have traditionally been viewed as the most prolific group of antibiotic producers amongst the prokaryotes (Bérdy 2005). As such, this phylum is routinely screened for new bioactive structures (Baltz 2007). Actinobacteria have been the focus of many antimicrobial discovery programs, and in recent decades, isolation and exploitation of Actinobacteria for novel compounds from conventional environments have led to the perennial rediscovery of known compounds (Tulp and Bohlin 2005). It has been argued that Actinobacteria have been so extensively studied that the chances of finding novel chemistries are too low to make them viable research leads and despite years of success, this over-mining of soil bacteria has resulted in diminishing returns (Lewis 2012). Watve et al. (2001) have disputed this maintaining that only a minor percentile of all the species of Streptomyces existing in nature have been grown in culture and their mathematical models suggest that the number of antibiotics still to be discovered from Streptomyces spp. could well be above 100,000. However, this high rate of compound rediscovery in culture-based screening programs has meant that it is now crucial that new groups of Actinobacteria be pursued as sources of novel therapeutically relevant compounds (Bull et al. 2000). Actinobacteria have a ubiquitous distribution in the biosphere, including the extremobiosphere. In the same regard, there is evidence that suggests Actinobacteria isolated from the extremobiosphere will be a rich source of novel natural products (Bull 2011). These unique environments will undoubtedly harbor novel strains of actinomycetes, which are likely to yield novel metabolites. It has been nearly 50 years since the occurrence of actinomycetes in saline environments and their tolerance to high salt concentrations was first described (Tresner et al. 1968; Gottlieb 1973). Since this time, Hamedi et al. estimated 70 species of halotolerant and halophilic actinomycetes from at least 24 genera have been successfully described (Hamedi et al. 2013). Table 1 summarises currently known antimicrobial compounds from halophilic and halotolerant actinomycete.

Table 1. Antimicrobial compounds from halophilic and halotolerant actinomycetes.

Source	Compound	Structure	Reference
Streptomyces sp. strain B6921 (Sediment, Mauritius Indian Ocean)	Himalomycins		(Maskey, Li, et al. 2003)
Verrucosispora sp. strain AB-18-032 (Sediment, Sea of Japan)	Abyssomicins B–D
Micromonospora sp. strain DPJ12 (Isolated from Didemnum proliferum)	Diazepinomicin		(Charan et al. 2004)
Streptomyces sp. (NA)	Gutingimycin		(Maskey et al. 2004)
Streptomyces nodosus (NPS007994) (Marine sediment, California)	Lajollamycin		(Manam et al. 2005)
Salinispora tropica, strain CNB-392 (Marine sediment, Bahamas)	Sporolides A and B		(Buchanan et al. 2005)
Salinispora pacifica CNS103 (NA)	Cyanosporasides A and B		(Oh et al. 2006)
Streptomyces sp. strain CNH365 (Marine sediment, Santa Barbara, California)	Anthracimycin		(Jang et al. 2013)
Nocardiopsis terrae YIM 90022 (NA)	Quinolone alkaloid		(Tian et al. 2014)
Streptomyces sp. strain DB634 (Atacama Desert, Chile)	Abenquines A-D		(Schulz et al. 2011)
Streptomyces sp. strain C34 (Atacama Desert, Chile)	Chaxalactins A-C		(Rateb, Houssen, Harrison 2011)
Saccharothrix sp. SA198 (Saharan soil)	Antibiotic A4 and A5		(Boubetra et al. 2013)
Streptomyces sp. TK-VL_333 (Algeria)	4-(4-hydroxyphenoxy) butan-2-one		(Kavitha et al. 2010)
Streptomyces sp. TK-VL_333 (Algeria)	Acetic acid-2-hydroxy- 6-(3-oxo-utyl)-phenyl ester		(Kavitha et al. 2010)
Streptomyces sp. C34 (Laguna de Chaxa, Salar de Atacama, Chile)	Chaxamycins A–D (1-4)		(Rateb, Houssen, Arnold 2011)

Despite this, halophilic (and more generally extremophilic) actinomycetes remain largely underexplored for the discovery of novel bioactive secondary metabolites (Hamedi et al. 2013). The unique metabolic diversity and biotechnological potential found in halophilic and halotolerant microorganisms demonstrate that this phylum of Actinobacteria are an invaluable source to be exploited (Cragg and Newman 2013; Hamedi et al. 2013). Jose et al. have reviewed the biodiscovery potential of novel hypersaline actinomycetes and conclude that halophilic and halotolerant actinomycetes have an enormous capacity to produce a range of bioactive secondary metabolites with diverse activities (Jose and Jebakumar 2014). The discovery of taxonomically unique strains of extremophilic Actinobacteria has led to the anticipation that mining these groups could add an alternative dimension to the line of natural product drug discovery (Thumar et al. 2010) that promises to raise the prospect of discovering novel bioactive compounds (Jose et al. 2011; Hamedi et al. 2013). In this regard, isolation programs have been overhauled towards largely unexplored, unusual, and extreme environments like hypersaline environments (Hamedi et al. 2013; Jose and Jebakumar 2013). The determination towards understanding these species has led to the creation of online databases such as actinobase, which compiles information on the diversity, molecular phylogeny and biotechnological potential of actinomycetes (Sharma et al. 2012).

The presence of such activity is widely described in the literature, and numerous authors have reported the occurrence of antimicrobial-producing halophilic actinomycetes (Kokare et al. 2004; Li et al. 2005; Manam et al. 2005; Vasavada et al. 2006; Sarkar et al. 2008; Suthindhiran and Kannabiran 2009; Thumar et al. 2010). Similarly, their halotolerant counterparts have also been reported for their bioactivity (Maskey, Helmke, et al. 2003; Maskey, Li, et al. 2003; Maskey et al. 2004; Mahyudin et al. 2012).

Bioactives from halophilic actinomycetes have been previously described since the early 1990s, with the isolation of the first moderately halophilic actinomycete, Actinopolyspora mortivallis JCM7550, capable of producing nucleoside antibiotics isolated from saline soils, sampled from Death Valley, California, USA (Yoshida et al. 1991). To date, several studies have observed actinomycetes-like-halophiles in many diverse hypersaline environments from salt lakes (Swan et al. 2010; Guan et al. 2011), solar salterns (Baati et al. 2008), salt mines (Chen et al. 2007) to brine wells (Xiang et al. 2008), and particularly from salt pans (Jensen et al. 1991; Dhanasekaran et al. 2004; Roshan et al. 2013). Following the isolation of Actinopolyspora, another halophilic Actinopolyspora was isolated, on this occasion, from marine saltpans. This isolate demonstrated antibacterial and antifungal activity that was dependent on the growth medium (Kokare et al. 2004), the significance of this finding highlights the importance of utilising an array of media designed to promote the maximum production of bioactives.

Irshad et al. (2013) demonstrated that the foreshore soil of Daecheon Beach and Saemangeum Sea (Korea), was a rich source of halophilic Actinobacteria. Out of the 63 strains examined, 12 isolates exhibited pathogenic bacteria, pathogenic fungi, and yeast. Two isolates, Microbacterium oxydans, and Streptomyces fradiae displayed antibacterial activity against all pathogens tested.

Ballav et al. assessed the capacity of halotolerant and halophilic Actinomycetales from the crystallizer pond sediments of the Ribandar saltern, Goa, for their capacity to produce antibacterial secondary metabolites (Ballav et al. 2015). These isolates demonstrated antimicrobial activity against S. aureus, Staphylococcus citreus, and Vibrio cholerae.

The importance of a biologically competitive environment was emphasized by Imada et al. the study described an actinomycete with high similarity to Micromonospora globosa, whose antibacterial production was dependent on seawater (Imada et al. 2007). A previous study reported that a similar strain, M. globosa JCM 3126, did not produce any antibiotics in the presence of seawater (Shiomi et al. 1990), speculating that environmental factors that influence antibiotic production could exert strain-specific, rather than universal, effects. Many Streptomyces species are reported to secrete antibiotics against bacteria, fungi, and yeasts at higher salinity and alkaline pH (Basilio et al. 2003). For example, the alkaliphilic Streptomyces sp. AK409 secretes the antimicrobial compound pyrocoll, but only under alkaline conditions (Dietera et al. 2003). Another species, Streptomyces sannanensis RJT-1, a halophile isolated from alkaline soil, was also reported to be a potent antibiotic producer of compounds active against Gram-positive bacteria (Singh et al. 2006). Gohel et al. highlighted again the role of biologically competitive environments, with unique conditions such as pH, temperature, pressure, oxygen, light, nutrients, and salinity, in the production of novel antibiotic compounds from actinomycetes isolated from saline and alkaline habitats (Gohel et al. 2015). Another example provided by Okami et al. is the isolate Streptomyces griseus, cultivated from marine sediment. This isolate exhibited no streptomycin production, however when grown in a medium containing low concentrations of organic nutrients and a high concentration of NaCl, it produced a new ionophoric polyether antibiotic (Okami et al. 1976). While it is unclear if this production was an epigenetic trigger or an adaptive response to the saline environment, it does signify the importance of utilizing a growth medium that produces maximum bioactivity.

Amongst the most known and most extensively characterized antimicrobial compounds produced by Actinobacteria are polyketides. In 2011, a moderately halophilic Streptomyces strain JAJ06 isolated from saltpan soil in Tuticorin (India) produced a seawater-dependent antimicrobial activity against a panel of bacteria and yeast strains with high potency, e.g. MIC of 1–2 μg/mL against MRSA. Subsequent sequential analysis revealed that this compound was a polyketide (Jose et al. 2011). Zhao et al. discovered further linear polyketides, actinopolysporins A, B, and C, as well as the known antineoplastic antibiotic tubercidin from the halophilic actinomycete Actinopolyspora erythraea YIM 90600 (Zhao et al. 2011). Later Jang et al. described the noteworthy discovery of the polyketide antibiotic, anthracimycin, produced by the marine-derived Actinobacteria, Streptomyces sp. strain CNH365, which had significant activity against Bacillus anthracis (Jang et al. 2013). In 2014 Chen et al. isolated the halophilic Actinobacterium, Actinopolyspora erythraea YIM90600, from the Baicheng salt field, the only known erythromycin-producing extremophile. Additionally, the new erythromycin biosynthetic gene cluster had a high identity and similarity to Saccharopolyspora erythraea, a soil actinomycete known to produce erythromycins (McGuire et al. 1952; Chen et al. 2014).

Other antibiotic chemical classes were isolated in 2014, from the soil-associated halophile Nocardiopsis terrae strain YIM 90022. This strain produced five types of secondary metabolites, two of which, a quinoline alkaloid and N-acetyl-anthranilic acid, displayed antimicrobial activities. The quinoline alkaloid displayed antibacterial activity against S. aureus, B. subtilis, and E. coli and antifungal activity Pyricularia oryzae. N-acetyl-anthranilic acid inhibited the growth of S. aureus and E. coli and antifungal activity against Fusarium avenaceum, Fusarium graminearum, and Fusarium culmorum.

Tian et al. reported the isolation of another halophilic species Nocardiopsis gilva YIM 90087. The Actinobacteria produced three p-terphenyl derivatives with antifungal activity against some pathogenic fungi (including F. avenaceum, F. graminearum, F. culmorum, and C. albicans) and antibacterial activity against B. subtilis (Tian et al. 2013).

The high-altitude deserts of Chile, such as the Atacama, are amongst the oldest and driest deserts on the earth, long considered to be too extreme to support any form of life given their harsh conditions, high levels of UV radiation, and lack of liquid water (Gómez-Silva et al. 2008). However, despite these seemingly inhabitable conditions, many authors have reported the isolation of actinomycetes from soils and halites collected from these arid deserts (Okoro et al. 2009; Bull and Asenjo 2013). The Atacama has also yielded interesting organisms with unique biosynthetic potential such as Streptomyces leeuwenhoekii, which produces the novel antimicrobials the chaxalactins and chaxamycins (Busarakam et al. 2014).

While halophilic actinomycetes are a demonstrable resource of antimicrobials, strains such as Saccharopolyspora salina VITSDK4 from salt pan soil of the Marakkanam coast of the Bay of Bengal, displayed not only antagonistic properties to various bacterial and fungal pathogens but additionally, extracellular bioactives were found to inhibit proliferation of HeLa cells indicating potential cytotoxic activity (Suthindhiran and Kannabiran 2009).

The scaffolds of about 80% of antibiotics approved for clinical use over the past decade are derived from microbial natural products (Thaker et al. 2013). However, traditional screens of natural product extracts for compounds with antibiotic activity are faced with low hit rates and high rates of rediscovery, a characteristic of conventional discovery approaches (Thaker et al. 2013). Modern approaches should be utilised to exploit halophilic actinomycetes, given their enormous capacity to produce a range of bioactive secondary metabolites with diverse activities.

Halophilic archaea

Halotolerant and halophilic bacteria actively pump ions out of cells and produce organic osmotic solutes to maintain a low intracellular ionic concentration and prevent the “salting-out” of enzymes. Archaea, however, adopt a different strategy to achieve osmotic balance and their cytoplasm contains molar concentrations of salts, mainly KCl, which provides the necessary osmotic equilibrium (Oren 2006). Difficulties in the isolation and cultivation of archaea have meant that they have been vastly under-investigated as a source of novel bioactive compounds.

Traditionally, natural product chemistry has focussed primarily on bacteria, fungi, plants, and the marine environment as sources of bioactive compounds. By comparison, research into the extreme biosphere of archaea has largely been overlooked, since their first description and proposal as a distinct branch, a third domain, of the tree of life by Woese in the late 1970s (Woese and Fox 1977). With the unbounded rise of AMR, archaea could be an important source of antimicrobial compounds. Given that archaea survive and even thrive in unique extreme environments, the compounds they produce could also be both unique in terms of structure and pharmacology, and distinct from other clinically used antibiotics.

While traditional antibiotics production have been demonstrated from the bacterial and eukaryotic domains of life, the archaeal third domain is yet to yield such compounds. Given the similarity of archaea both to bacterial and eukarya and the co-colonization by archaea of the dynamic and competitive environmental niches inhabited by the other domains of life where antibiotic production has proven advantageous, we propose that the archaea are a potentially untapped source of novel, clinically useful antibiotic compounds.

The most widely studied antimicrobial compounds isolated from archaea are the archaeocins. Thought to endow archaea with a competitive advantage for nutrients and other resources, archaeocins are ribosomally synthesized bacteriocin-like antimicrobial peptides (AMPs) or proteins with antagonistic activity against other microorganisms (Rodriguez-Valera et al. 1982). Archaeocins fall into two phylogenetic groups: halocins produced by haloarchaea (Shand and Leyva 2007) and sulfolobicins produced by the genus Sulfolobus (Prangishvili et al. 2000). Halocins have been shown to possess activity against haloarchaea, thermophilic crenarchaea (Haseltine et al. 2001) and some bacterial strains. In comparison with bacteriocins and eucaryocins, there is much to be learnt from archaea-derived AMPs (O’Connor and Shand 2002).

The first halocin was described almost four decades ago by Rodriguez-Valera et al. (1982), – secreted by Haloferax mediterranei (Rodriguez-Valera et al. 1982). In 1994 Torreblanca et al. proposed that halocins were a near-universal feature in all haloarchaea (Torreblanca et al. 1994). However, despite this apparent abundance, only 12 archaeocins (11 halocins and one sulfolobicin) have been partially or fully characterized to date, with hundreds of archaeocins thought to exist within the haloarchaea (Shand and Leyva 2008). Only halocins H4, H6, and RI have been characterized to a protein level (Meseguer and Rodriguez-Valera 1985; Rdest and Sturm 1987; Torreblanca et al. 1989), while halocins S8, H4, and C8 are the most characterized halocins, at both a genetic and mRNA transcription level. Table 2 shows the known primer sequences for these halocins (Sun et al. 2005).

Table 2. Forward and Reverse primer sequences for Halocins S8, H4 and C8.

Halocin	Primer sequence
S8	F1 (5′-GCTTCAGCTGTCGGATTTTC-3′)
	R1 (5′-CTTCAAGTATCGCGGACACA-3′)
H4	F1 (5′-CAGTCTCGGAGCGCTTAGTT-3′)
	R1 (5′-CGGGATACTTGGGTTCTTCA-3′)
C8	C8-IG1 (5′-ACGCTGTGCGAGTAGGAGTT-3′)
	C8- IG2 (5′-GAGACGACGTTCGAGGAATC-3′)
	C8-9 (5′-CCGACGGGGTGCAGTCC-3′)
	C8-10 (5′-CGTCACCGCAACCGCTGT-3′)

Most halocin genes follow the pattern of induction during the transition into the stationary phase. Halocins, such as H4, H6, and S8, have low detectable levels during exponential growth, but as the media becomes depleted during the stationary phase, these levels increase drastically (Cheung et al. 1997). An exception to this is halocin H1, which possesses activity first detectable earlier during the mid-exponential phase (Platas et al. 1996; O’Connor and Shand 2002). Furthermore, while the activity of most of these halocins decreases again during the stationary phase, others such as HalR1 maintain a high level of activity for a prolonged time (Shand et al. 1998).

Archaeal proteins undergo secretion in either an unfolded (via the Sec pathway) or a folded conformation (via the Twin arginine transport (Tat) pathway). Halocins contain a predicted Tat signal peptide synthesized at the N-terminal that targets them for secretion from the cell, and this would suggest that they are generally secreted by the Tat pathway (Price and Shand 2000; O’Connor and Shand 2002). Halocin genes are plasmid-encoded, with typical archaeal TATA boxes found within promoter regions, although for halocin C8 the TATA box is closer to the transcription site compared to the other halocins, 18 bp rather than 22–25 bp (Sun et al. 2005). Their transcription is described as leaderless given their transcription site either coincidences with or is only a few bps upstream from the translational start codon ATG (Shand and Leyva 2007).

Halocins can be divided into two classes based on size: the smaller microhalocin peptides (<10 kDa but as small as 3.6 kDa) and the larger halocin proteins (>10 kDa up to 35 kDa) Microhalocins are cleaved from a larger precursor protein, the preproprotein (O’Connor and Shand 2002).

The hydrophobic microhalocins are much more robust than larger halocin proteins. They can be desalted without losing activity, are insensitive to organic solvents (such as acetone or acetonitrile), and are relatively insensitive to heat (O’Connor and Shand 2002). Microhalocins such as R1 can also be stored at 4 °C for long periods, retaining activity after seven years (O’Connor and Shand 2002). The larger protein halocins are much more fragile, they are generally more heat-labile, and lose their activity when desalted below 5% w/v NaCl. The addition of one or more serine proteases, typically proteinase K and trypsin, confirms the presence of mature halocins. Table 3 summarises many of the known features of several well-described halocins.

Table 3. Characterisation of known halocins, with thermal stability, salt dependency, the spectrum of activity, mechanism, and known sequences.

Halocin	Source	Size (kDa)	Thermal stability	Salt-dependence	Activity spectrum^a	Mechanism	Peptide sequence	Reference
A4	Haloarchaeon strain TuA4	<5	≥1 week at boiling	No	Broad Sulfolobus spp.	ND	ND	(Haseltine et al. 2001; O’Connor and Shand 2002)
C8	Halobacterium sp. AS7092	6.3	≥1 h at boiling	No	Broad	Cell wall	H2N-DIDITGCSACKYAAGQVCTIGCSAAGGFICGLLGITIPVAGLSCLGFVEIVCTVADEYSGCGDAVAKEACNRAGLC-COOH	(Li et al. 2003; Meknaci et al. 2014)
G1	Halobacterium strain GRB	ND	ND	ND	Broad	ND	ND	(Soppa and Oesterhelt 1989)
HA1	Haloferax larsenii HA1	∼14	<80 °C	Yes	Narrow	Rupturing cell boundary	ND	(Kumar and Tiwari 2017b)
HA3	Haloferax larsenii HA3	∼13	<80 °C	Gradual reduction at 4 °C	Narrow	Rupturing cell boundary	ND	(Kumar and Tiwari 2017a)
H1	Haloferax mediterranei Xai3	31	<50 °C	Yes	Broad	Membrane permeability	ND	(Rodriguez-Valera et al. 1982; Platas et al. 1996)
H2	Haloarchaeon strain GLA22	ND	ND	ND	Broad	ND	ND	(Rodriguez-Valera et al. 1982)
H3	Haloarchaeon strain GAA12	ND	ND	ND	Broad	ND	ND	(Rodriguez-Valera et al. 1982)
H4	Haloferax mediterranei R4	39.6 preprotein	<60 °C	Partially	Narrow	Proton flux	H2N-MSKDRDGRRTSRRGTLKKIGGFSLGALSFGAVGRTQAATGSSVTTADIAPPGPNGDPKSVQIDDKYTGAEMYGEGDFRVGLGTDLTMYPPVYRESLGNGSGGWEFDFTVCGSTACRFVDSNGDVKEDDKAKEMWWQEINFNDINQDLYSRNDSDWVGSTPADTQPEFDYTDFALARDGVTLALTALNPAMGSLALGATYFLSDMVNWIASQHEDDSSLKRKWDYDGLSGPLYADSSTYLLARDEMTSNSYESFTIDNIAVAFPEFPVRTKYYVTFTAPDDPSTQSISTLEEEGIYRVPATEVAAARPPGSRRSKSAADEMVYVADPKKFIEVEPVKNPSIPDRIYEEIEQKKKQRSRKQ-COOH	(Rodriguez-Valera et al. 1982; Meseguer and Rodriguez-Valera 1985, 1986; Meseguer et al. 1995; Cheung et al. 1997; Shand et al. 1999)
H5	Haloarchaeon strain MA220	ND	ND	ND	Narrow	ND	ND	(Rodriguez-Valera et al. 1982)
H6/H7	Haloferax gibbonsii Ma2.39	32	≤90 °C	No	Broad Sulfolobus spp. M. thermaphila	Na^+/H^+ antiporter inhibitor	ND	(Rodriguez-Valera et al. 1982; Torreblanca et al. 1989; Meseguer et al. 1995)
H17	Haloferax sp. SWI17	30–40	ND	ND	Halobacterium salinarum DSM3754	ND	ND	(Mazguene et al. 2018)
KPS1	Haloferax volcanii KPS1	ND	<80 °C	Yes	Broad E. coli, S. mutans, P. aeruginosa, B. subtilis, S. aureus	ND	ND	(Kavitha et al. 2011)
R1	Halobacterium sp. GN101	3.8	60 °C	No	Broad	ND	N2H- LQSNININTAAAVILIFNQVQVGALCAPTPVSGGGPPP-COOH	(Shand et al. 1998; Haseltine et al. 2001; O’Connor and Shand 2002)
Sech7a	Haloferax erranei AY823953	10.7	<80 °C	Yes	ND	Potentially novel	ND	(Pašić et al. 2008)
SH10	Natrinema sp. BTSH10	ND	ND	ND	ND	Cellular swelling	ND	(Pašić et al. 2008)
S8	Haloarchaeon Strain S8a	3.6 mature	≥24 h at boiling	No	Broad Sulfolobus spp.	ND	H2N-SDCNINSNTAADVILCFNQVGSCALCSPTLVGGPVP-COOH.	(Shand et al. 1999; Price and Shand 2000; Haseltine et al. 2001)

^aActivity spectrum refers to inhibition of haloarchaea unless otherwise indicated. ND: not determined.

Halocin H4

HalH4 was the first reported halocin, and was isolated from Haloferax mediterranei. It has characterized at both the protein and genomic levels. HalH4 is cleaved from a preprotein (39.6 kDa) to yield the mature HalH4 protein (34.9 kDa), and its activity against other haloarchaea has been described as archeaolytic. Halocin H4 appears to disrupt the permeability of the membrane on sensitive Hbt. salinarum cells, leading to an ionic imbalance and eventual cell lysis. Halocin H4 can be desalted to ∼0.06% w/v NaCl (10 mM) and with only a twofold loss in activity, against Halobacterium sp. (Rodriguez-Valera et al. 1982; Meseguer and Rodriguez-Valera 1986; Cheung et al. 1997).

Halocin S8

The halS8 gene has an open reading frame of 933 base pairs long, which when translated yields a 311 amino acid in length pre-pro-protein. Catalytic cleavage of the pre-pro-protein generates separate peptides, a 230 amino acid N-terminal protein, a 45 amino acid C-terminal peptide, and the 36 amino acids mature microhalocin S8. The mature halocin S8 has a molecular mass of 3.58 kDa with four cysteine residues and can form potentially two disulfide bridges. Whether the release of this active halocin from the pre-pro-protein is due to autocatalysis or the activity of a protease is unknown (Price and Shand 2000).

Halocin R1

R1 is a microhalocin produced by Halobacterium salinarum GN101 (O’Connor 2002). While initial studies suggested a molecular mass of 6.2 kDa(Rdest and Sturm 1987), it was later revealed that R1 was part of a larger preproprotein that had a mass of 29 kDa (Shand et al. 1998; O’Connor and Shand 2002). The mature HalR1 is made up of 38 amino acids and shares 71% similarity to S8 (Price and Shand 2000; O’Connor and Shand 2002), this difference is proposed to account for the differences in the spectral activity and thermostability, between the two peptides as previously reported (Shand and Leyva 2007). The sequences of these two halocins are:

–R1: lqsNINiNTAAaVILiFNQVqvgALCaPTpVsGGgPpP

–S8: sdcNINsNTAAdVILcFNQVgscALCsPTIV–GG–PvP

The mechanism of action of R1 is proposed to be archaeostatic (Rdest and Sturm 1987) when the peptide was incubated with sensitive Hbt. salinarum cells, there were no changes in cell morphology, or optical density followed seven days of incubation, but when HalR1 was removed from the cells, the culture was able to resume growth in new media. The mechanism is also thought to be dose-dependent as increasing the concentration of HalR1 resulted in a proportional increase in growth inhibition (Rdest and Sturm 1987).

Halocin H7

HalH7 (formerly known as HalH6) is produced by Haloferax gibbonsii Ma2.39 (Torreblanca et al. 1989). Halocin H7 is a stable halocin that retains its activity when desalted and is heat resistant. Initially determined to have a molecular mass of 32 kDa (Torreblanca et al. 1989), as with halocin R1, this halocin is released from a larger preproprotein, yielding a smaller 3 kDa active peptide, thus classifying halocin H7 as a microhalocin. This halocin is lytic, as sensitive cells were exposed to the protein, they would swell and eventually lyse. Further studies revealed that this mechanism was due to the action of the halocin on the Na⁺/H⁺ antiporter system on the cell membrane of both haloarchaea and mammalian cells (HEK293, NIH3T3, Jurkat, and HL-1) (Meseguer et al. 1995).

Halocin A4

Halocin A4, another microhalocin was produced by an uncharacterized haloarchaeon isolated from a Tunisian saltern. It has a molecular mass of 7.435 kDa, is hydrophobic and acidic. It demonstrates a broad spectrum of activity, compared to other halocins with activity against haloarchaea and the crenarchaeal hyperthermophile Sulfolobus solfataricus (Shand 2006). This indicates that there may be a conserved biological target between euryarchaeal halophiles and crenarchaeal hyperthermophiles, where the halocin acts (Haseltine et al. 2001).

Halocin C8

Halocin C8 is produced by Halobacterium strain AS7092 (Li et al. 2003; Sun et al. 2005) and contains ten cysteine residues and is the largest microhalocin at 7.44 kDa. Like S8, it is cleaved from a preproprotein, at the C-terminal end, by an unknown mechanism to produce the active mature peptide HalC8 and an immunity protein HalI. HalI functions to protect the producer organism against its native halocin, by binding, to and inactivating the halocin (Sun et al. 2005).

Halocin H1

H1 is a larger halocin peptide produced by Haloferax mediterranei M2a. Halocin H1 has a molecular mass of 31 kDa and is heat-labile. It loses activity if desalted and requires a minimum of 5% w/v NaCl to maintain activity. Additionally, the compounds within the media influence the production of this halocin with optimal antimicrobial activity being obtained when 0.5% of N-Z-amine E was used as a nutrient (Platas et al. 1996). This halocin displayed activity against Haloferax gibbonsii ATCC 33959, Halobacterium salinarum NRC 34002 and Haloferax volcanii NCMB 2012.

The spectrum of activity of the halocins varies for each compound, halocins H4 and H6 exhibit a narrow range of activity while others such as halocin H1 possess a wider range of activity towards numerous haloarchaea as mentioned previously (Blum 2008; Clark et al. 2008). Initially, halocins were thought to display activity confined to closely related Halobacteria sp. such as C8 (Li et al. 2003). However, the microhalocins A4 and S8 demonstrate a broader activity with cross-phylum inhibition of Sulfolobus solfataricus (Haseltine et al. 2001) and halocin R1 inhibiting the growth of Methanosarcina thermophile (Shand et al. 1999).

A relatively restricted number of studies have demonstrated the potential of archaea to interact with and inhibit the growth of bacteria. Atanasova et al. demonstrated this interaction, they screened 68 haloarchaea of different genera (Halorubrum, Haloferax, Haloplanus, Haloarcula, Halogranum, Halobacterium, Halosarcina, and Halogeometricum), and found they produced halocins that had antagonistic interactions on 22 halophilic bacteria (Halomonas, Rhodovibrio, Salisaeta, and Pontibacillus). Additionally, this study also found that halophilic bacteria from various genera could be identified as halocin producers (Atanasova et al. 2013). More recently Ghanmi et al. (Ghanmi et al. 2016), screened 35 halophilic bacterial and archaeal strains for antagonism, and three cases of cross-domain inhibition (archaea/bacteria or bacteria/archaea) were observed. Four archaeal strains inhibited the growth of several other strains (both bacterial and archaeal). Three archaeal strains were further investigated, and these strains were shown to encode and express either halocins S8 or H4, demonstrating that these halocins have both anti archaeal and antibacterial properties. Another study by Shand and Leyva (2008), screened 75 strains isolated from the Wilson Hot Springs (48 archaeal and 27 bacterial strains) for growth inhibition and found that 77% inhibited the growth of at least one other isolate. Inter-domain interactions were also present with 43 haloarchaea inhibiting halophilic members of the domain Bacteria, while seven bacteria were capable of antagonising haloarchaea. The 75 isolates possessed no bioactivity against Gram-negative bacteria, but inhibited, the growth of several Gram-positive bacteria was inhibited (Shand and Leyva 2008).

Microhalocins like KPS1 produced by Haloferax volcanii KPS1 exhibited inhibitory activity towards Streptococcus mutans, S. aureus, E. coli, and P. aeruginosa (Kavitha et al. 2011). Another microhalocin, C8 or C8-like, produced by Natrinema gari QI1, was shown to have antimicrobial activity towards both archaea and the bacterium, Micrococcus luteus, but not against any other pathogens tested (S. aureus, B. subtilis, B. cereus, and Sarcina lutea) (Quadri et al. 2016). The body of evidence demonstrating the potential of halocins as anti-bacterial agents is growing. While halocins display anti-archaeal activity, further investigation to assess their potential against Gram-positive and Gram-negative bacteria requires investigation. Microhalocins remain active after desalting and based on previous studies are likely to be a stronger candidate for antibacterial activity. Additionally in order to fully correctly assess the potential of haloarchaea supernatants extraction of microhalocins from culture supernatant requires the use of non-polar solvent extractions instead of traditional liquid-liquid extraction solvents such as ethyl acetate,

To date, few studies have addressed the mechanism of action of halocins. While it has been shown that some halocins such as A4 are cytocidal, and others like S8 and R1 are cytostatic, only the mechanism of one halocin (H7/6) has been elucidated and involved inhibition of the Na⁺/H⁺ antiporter (Lequerica et al. 2006). Microhalocin peptides have either little or no net charge and therefore are unable to interact with the negatively charged bacterial membrane in the same way many bacteriocins and eucaryocins do, suggesting that their mechanism is distinct from these other classes.

The most recently identified halocins have been HA3 (Kumar and Tiwari 2017a), H17 (Mazguene et al. 2018), and HA1(Kumar and Tiwari 2017b) however, a deeper understanding of the mechanism of action of halocins along with structural elucidation, will undoubtedly prove highly beneficial with regards to their future applications. Equally, investigation at both a protein and gene expression level will allow for the screening and discovery of novel AMPs with genomic mining.

Halophilic eukaryotes

Fungi from hypersaline environments do not require salt for viability but can tolerate salt to very high concentrations from 0 to 30% NaCl (Plemenitaš et al. 2008), a much broader range than bacteria or archaea (Plemenitaš et al. 2014). They have developed survival strategies to adapt to low water activity (a_w) such as the accumulation of compatible solutes such as glycerol, which do not interfere with the vital functions of their cellular proteins while counteracting changes in turgor pressure (Gunde-Cimerman and Zalar 2014). Only a few fungal species display true halophilic behaviour (Zalar et al. 2005), such as Debaryomyces hansenii, Hortaea werneckii, and Wallemia ichthyophaga, these fungi have been isolated from natural hypersaline environments (Gunde-Cimerman et al. 2009).

While halotolerant fungi such as Aspergillus variecolor B-17 have already proven to be useful sources of cytotoxic agents (Wang et al. 2007), extremophilic fungi have been largely overlooked as a reservoir of potential antibiotic compounds.

A previous study demonstrated that stress conditions (high NaCl concentration and low a_w) could induce the production of compounds with antibacterial activity. Methanol and acetone extracts from different fungal strains demonstrated antimicrobial activity against B. subtilis. The study revealed that low a_w induces the production of active metabolites in halophilic and halotolerant fungi from hypersaline waters (Sepcic et al. 2010).

However, other eukaryotes such as the halophilic yeasts isolated from cocoa bean pulp displayed antimicrobial activity against Salmonella spp. (Indah et al. 2015). The halotolerant microalgae Dunaliella, and its close relatives have been reported for their potential to produce a multitude of biologically relevant compounds. Of clinical relevance, Ohta et al. reported the ability of methanol extracts of D. bioculata and D. primolecta to inhibit the cytopathic effect of herpes simplex virus type 1 (Ohta et al. 1998).

Quorum sensing inhibition as an anti-virulent strategy

Within bacterial infections, an estimated 65–80% (Potera 1999; NIH 2003) are attributable to the occurrence of specialized surface-adhered bacterial communities known as biofilms (Lewis 2007). Bacterial biofilms are characterized by their highly elevated tolerance to antimicrobial challenges, with biofilms requiring up to 1000 times higher concentrations of antibiotics compared to their planktonic equivalents to obtain complete eradication (Koch and Høiby 2000). Bacterial behaviors within a biofilm are regulated, in part, by the phenomenon of quorum sensing (QS), a population density-dependent gene regulation system whereby bacteria release chemical signaling molecules and, when a threshold concentration has been achieved, express genes in a cell density-dependent manner (Bhargava et al. 2010). Bacterial pathogens employ QS to regulate virulence and QS has emerged as a promising target for anti-virulence therapeutic approaches in the past decade, and several quorum sensing inhibitors (QSI) from halophiles have been identified – as summarised in Table 4.

Table 4. Examples of QSI compounds and their corresponding activity produced from halophilic and halotolerant bacteria.

Source	Compound	Structure	Activity	Reference
Halobacillus salinus C42	N-(2′-phenylethyl)-isobutyramide		Inhibition of bioluminescence by V. harveyi BB120 Inhibition violacein production against Chromobacterium violaceum CV026	(Teasdale et al. 2009)
	2,3-methyl-N-(2′-phenylethyl)-butyramide		Inhibition of bioluminescence by V. harveyi BB120 Inhibition of violacein production against C. violaceum CV026 Inhibition of violacein production against C. violaceum ATCC 12472 Inhibition of green fluorescence protein by E. coli JB525
Bacillus cereus D28	Cyclo (L-Pro-L-Try)		Inhibition of bioluminescence by V. harveyi BB120 Inhibition of violacein production against C. violaceum ATCC 12472	(Teasdale et al. 2011)
Marinobacter sp. SK-3	Cyclo (L-Pro-L-Phe)		Inhibition of bioluminescence production in E. coli pSB401	(Abed et al. 2013)
	Cyclo (L-Pro-L-Leu)		Inhibition of bioluminescence production in E. coli pSB401 Inhibition of QS violacein production by C. violaceum CV-017
	Cyclo (L-Pro-L isoLeu)		Inhibition of bioluminescence production in E. coli pSB401 Inhibition of QS violacein production by C. violaceum CV-017
Blennothrix antharidosmum	Tumonoic acid F		Inhibition of bioluminescence by V. harveyi BB120	(Clark et al. 2008)
Lyngbya majuscula	Malyngolide		Inhibition of violacein production against C. violaceum CV017 Las QS-dependent production of elastase by P. aeruginosa PAO1.	(Dobretsov et al. 2010)
Leptolyngbya crossbyana	Honaucin A Honaucin B Honaucin C		Inhibition of bioluminescence by V. harveyi BB120	(Choi et al. 2012)

Teasdale et al. (2009) isolated two phenethylamide-based metabolites from the marine bacterium, Halobacillus salinus C42, which were capable of inhibiting a range of QS-controlled phenotypes in bioreporter strains including; violacein production by Chromobacterium violaceum; bioluminescence in Vibrio harveyi; and QS controlled Green Fluorescent Protein (GFP) production in E. coli JB525. These compounds act as antagonists in Gram-negative bacterial QS by competing with N-acyl homoserine lactones (AHL) for receptor binding. Further investigation into the widespread production of these QS inhibitory molecules, identified five isolates, either Bacillus or Halobacillus, that were capable of interfering with QS-controlled bioluminescence in V. harveyi and inhibiting violacein production by C. violaceum, with the elucidation of the QSI dipeptide cyclo-L-proline-L-tyrosine produced by Bacillus cereus D28 (Teasdale et al. 2011).

Hypersaline mats have previously been shown to be a source of QS-inhibitory compounds, following the total ethyl acetate extraction of the mat, extracts demonstrated QS-inhibitory properties against bioreporter strains Agrobacterium tumefaciens NTL4 (pZLR4) and Salmonella enterica S235, but not C. violaceum CV017 (Abed et al. 2011). Abed et al. (2013) subsequently screened the halophilic and moderately thermophilic strains from a hypersaline microbial mat for QSI molecules, resulting in the structural elucidation of four related diketopiperazines (DKPs) from Marinobacter sp. SK-3. The compounds capable of inhibiting violacein production by C. violaceum and reduced QS-dependent luminescence of the bioreporter E. coli pSB401. DKPs have previously been shown to have antimicrobial, antifungal, antiviral, and antitumor properties (Herbert 1992; Martins and Carvalho 2007) and their production has been observed across many genera, including Bacillus (Wang et al. 2010), Streptomyces (Li et al. 2006), and Pseuoalteromonas (Qi et al. 2009), and from a wide range of environments. In addition, DKPs are capable of activating QS-regulated pathways as well as inhibiting them, as observed by the haloarchaea Haloterrigena hispanica (Tommonaro et al. 2012). In Abed et al. (Abed et al. 2013), the supernatant and water extract of the cell pellet from the isolate, Haloterrigena saccharevitan SK-6, was found to possess QSI compounds capable of inhibiting violacein production in the bioreporter strain CV017.

Numerous studies have indicated the existence of QSI compounds from the marine environment, particularly from marine cyanobacteria. For example, the marine cyanobacterium, Blennothrix cantharidosmum, was found to produce several QSI tumonoic acids with tumonoic acid F being the most potent (Clark et al. 2008). Dobretsov et al. revealed that another marine cyanobacterium, Lyngbya majuscule, secreted the δ-lactone QSI compound, malyngolide. This compound inhibits the violacein production in C. violaceum CV017 and las QS-dependent production of elastase by P. aeruginosa PAO1 (Dobretsov et al. 2010).

More recently Choi et al. isolated, Leptolyngbya crossbyana, from corals on the Hawaiian coast, which produced the AHL mimetic compounds, honaucins A–C, which inhibited QS-regulated bioluminescence in V. harevyi BB120 (Choi et al. 2012).

The discovery of compounds and enzymes which interrupt QS via inhibition of signal generation, degradation of generated signals, or suppression of QS receptors, have important clinical applications (Hong et al. 2012; Tang and Zhang 2014). For pathogens like Burkholderia species, P. aeruginosa, Erwinia carotovora, and Vibrio spp., biofilm formation, toxin production, and virulence factor production are regulated by QS (Tang and Zhang 2014). QSI can be considered as an alternative antimicrobial strategy, that must be considered to control bacterial disease, by targeting the virulence factors essential for an infection rather than the viability of pathogens (Clatworthy et al. 2007). Additionally, the delayed emergence of resistance which results from imposing less selective pressure on pathogens (Clatworthy et al. 2007) makes QSI an ideal candidate as an adjunct therapy to improve antibiotic drug treatment (Busetti et al. 2014). QSI may be used as antifouling agents capable of inhibiting biofilm formation in medicine, industry, and environmental settings (Abed et al. 2013).

Important tools and techniques for antimicrobial drug discovery from halophiles

As the prevalence of AMR increases, there is a critical and urgent need to develop new technologies and strategies to combat this major global public health concern (Bakour et al. 2016). Particular focus has surrounded the development of methods that can extract novel antimicrobials from microbes uncultivable under standard laboratory conditions. “The Great Plate Count Anomaly” accounts for the 99% of microorganisms in the natural environment that cannot be isolated using conventional culture techniques (Amann et al. 1995). The development of methods that either enhance current culturing or utilise a culture-independent approach such as genome mining are indispensable if these “hidden” antimicrobial metabolites are to be investigated (Challis and Ravel 2000; Hug et al. 2018).

Genomic mining

It is now widely accepted that most of the investigated culturable microorganisms have a broader genetic capacity to produce a greater number of natural products than the number of compounds that they produce under conditions of laboratory culture (Bussi and Gutierrez 2019). Following the full genome sequencing of Streptomyces coelicolor (Bentley et al. 2002), it was clear that there is underexplored potential within bacterial genomes for drug discovery. S. coelicolor could produce 20 secondary metabolites when only three secondary metabolites were identified (Bentley et al. 2002). Within the Streptomyces genus, conservative estimates suggest that this genus alone can produce over 150,000 natural products, with less than 5% of these compounds characterized (Watve et al. 2001).

This classical approach of genome mining involves the prediction of cryptic metabolic pathways and genes that encode putatively in the biosynthesis of secondary metabolites of interest, such as biosynthetic gene clusters (BGC) (León et al. 2014; Blin et al. 2019; Zheng et al. 2019). The most typical examples of BGCs, with regards to antimicrobials, are two multifunctional mega-synthases, non-ribosomal peptide synthetase (NRPS) and polyketide synthase (PKS) (Fischbach and Walsh 2006; Gross 2007). NRPS and PKS have long been utilised by bacteria for cellular defence and offense against competitors (Czaran et al. 2002). They are prolific and generate many of the antibiotics used clinically, such as penicillin, erythromycin, vancomycin, and daptomycin among numerous others (Fischbach and Walsh 2006).

The continued enhancement of computational biology has resulted in many of these approaches carried out in silico. In silico mining has significantly advanced the analyses of microbial genome sequences, allowing the determination of BGC involved in the production of new antibiotics from different organisms (Challis 2008; Scheffler et al. 2013), with the use of sequence-based comparison software such as DIAMOND (Buchfink et al. 2015) or BLAST (Altschul et al. 1990). One of the more popular and straightforward tools used in genome mining is the antiSMASH- a web-based bioinformatics server. It allows users to identify antiSMASH-annotated BGCs from its database with the most recent version containing 6200 bacterial genomes (Blin et al. 2019).

Genome mining has led to the discovery of many antimicrobial compounds (Nikolouli and Mossialos 2012; Scheffler et al. 2013) including orfamide A (Gross 2007), salinilactam (Udwary et al. 2007), holomycin (Li and Walsh 2010), and emericellamide A, C–E (Chiang et al. 2008) (summarised in Table 5). Within halophiles, Schäberle et al. reported three strains of Enhygromyxa salina, that were producers of novel secondary metabolites (Schäberle et al. 2010). Furthermore, Meklat et al. revealed that halophilic actinomycetes, such as Nocardiopsis had a high frequency of NRPS genes (Meklat et al. 2011). Genome mining of the marine halophile, Haliangium ochraceum, identified the compound haliamide, the product of a gene cluster containing one NRPS module followed by four PKS modules. Five more PKS-NRPS gene clusters were identified in the genome of H. ochraceum, indicating that there is further antimicrobial potential (Dávila-Céspedes et al. 2016; Sun et al. 2016). Despite the success of genome mining in the discovery of many natural products, it does pose several limitations (Gross 2007). The prediction algorithms of the bioinformatic tools are based on the identification of known biosynthetic activities, and therefore truly novel biosynthetic pathways are not identified (Challis 2008; Weber 2014). In addition to unreliable predictions, if the genes are “silent”, it becomes much more difficult to produce, identify, and test the biological activity of the product. In such cases, growing the organism under many different conditions may activate the bioactive (Okami et al. 1976; Scherlach and Hertweck 2006). Genome mining methods include the genomisotopic approach, gene inactivation method, heterologous expression approach, the use of transcription activators, inactivation of a silent gene inhibitor (Scheffler et al. 2013), and the resistance-guided approach (Thaker et al. 2013).

Table 5. Structures of antimicrobials identified through genome mining from five species of bacteria.

Source	Compound	Structure	Reference
Pseudomonas fluorescens Pf-5	Orfamide A		(Gross 2007)
Streptomyces tropica CNB-440	Salinilactam A		(Udwary et al. 2007)
Streptomyces clavuligerus ATCC 27064	Holomycin		(Li and Walsh 2010)
Aspergillus nidulans R153	Emericellamide A, C–E		(Chiang et al. 2008)
Haliangium ochraceum SMP-2	Haliamide		(Sun et al. 2016)

The resistance-based mining technique is a genome-mining approach that based on the identification of self-protection mechanisms that antibiotic producers need to avoid autotoxicity. This self-resistant mechanism acts as a filter with the potential to identify antibiotic-producing bacteria and therefore could be exploited for the discovery of novel bioactive (Thaker et al. 2013). This resistance-guided approach is directly relatable to many halophilic organisms and studies by Khelaifia et al. have reported AMR in the haloarchaea Halalkalicoccus tibetensis and Natronococcus amylolyticus (Khelaifia and Drancourt 2012). A similar approach Culp et al. recently described an evolution-guided antibiotic discovery approach that combines phylogenetic divergence of BGCs along with the absence of self-resistance determinants to predict the novel glycopeptide corbomycin (Culp et al. 2020).

Another widely used method for antibiotic discovery involves metagenomic screening, which circumvents the difficulties associated with cultivation. Metagenomic screening relays on prior examination for BGCs using in silico bioinformatics tools, then a heterologous expression of selected pathways originating from uncultured microorganisms. The current problem with this approach is the discovery of novel antimicrobials requires a large reference dataset of “meta-omics” derived compounds (Hamm et al. 2019; Wang et al. 2019).

There are two main strategies utilized in the metagenomic approach – a classical functional metagenomic screening approach, and a targeted sequence-based metagenomic screening approach- Figure 3 summarises the main steps for both methods. In both cases, environmental samples are collected and environmental DNA (eDNA) is extracted and purified. In the classical approach, metagenomic libraries are created by extracting, cloning, and ligating the eDNA into shuttle vectors and transforming them into appropriate heterologous hosts, e.g. E. coli. The metagenomic library is screened for bioactivity using standard antimicrobial tests such as disc diffusions, MTT, or MIC assays. Positive clones are recovered from the metagenomic library, and the eDNA insert is sequenced. The antibacterial violacein was isolated from a soil metagenome using this approach (Brady et al. 2001).

Figure 3. Summary describing the two main approaches that used for antimicrobial discovery from metagenomes. Adapted from Hug et al. (2018).

In the targeted approach, eDNA is screened by PCR amplicons for specific sequences within BGCs. The sequence tags are evaluated for their biosynthetic origin, compared to a reference database, and biosynthetic gene clusters are reassembled in silico. A metagenomic library is generated with samples harboring the in silico reassembled BGC of interest. The proteasome inhibitor landepoxin A was identified using this targeted approach from metagenomic samples (Lok 2015). For both approaches, the novel BGCs are assembled, modified for heterologous expression, and then the bioactive natural product is isolated and structurally elucidated. Charlop-Powers et al. used a metagenomic approach and revealed that there is a rich biosynthetic diversity of clinically relevant BGCs in eDNA from New York City park soils (Charlop-Powers et al. 2016).

Enhanced culture techniques

In 2007 Coates and others, argued that the genomics route has proven to be target-rich, but has not led to the introduction of a marketed antibiotic as yet and suggested that non-culturable bacteria may be an alternative source of new antibiotics (Coates and Hu 2007). Moreover, while researchers involved in genome mining would debate this lack of value in their approach, the prospect of tapping into these non-culturable organisms offers the unquestionable opportunity to tap into a rich vein of drug discovery. Strategies used to increase bacterial diversity have ranged from soil-extract agar (Hamaki et al. 2005) to simulated environments with diffusion chambers (Kaeberlein et al. 2002) and the iChip (Ling et al. 2015), to variation of culturing environments (co-cultures or mixed cultures) (Tanaka et al. 2004) and to the role of host-associated environments (Vartoukian et al. 2010). An extensive review of these technologies is provided by Stewart et al. (Stewart 2012).

The discovery of teixobactin is probably the most significant example of a compound discovered as a direct result of using advanced isolation technologies (Ling et al. 2015). More recently, a screening of symbionts from the microbiome of entomopathogenic nematodes revealed a Gram-negative specific antibiotic termed darobactin (Imai et al. 2019). These studies demonstrate that the cultivation of microbes from unexplored territories (and not genomic mining) continues to provide the mainstay of novel classes of antimicrobials with unique natural product chemistry. Concerning extremophiles, access to their bioactives is already limited with difficulty in their cultivation using traditional laboratory conditions (Hug et al. 2018). Technologies such as that of the iChip device are likely to improve isolation efficiency and fermentation of previously uncultured bacteria and archaea from these largely unexplored habitats. Therefore, extreme environments possess an under-explored microbiome whose metabolites require interrogation to assess their antimicrobial potential.

Conclusion

We must rejuvenate the antibiotic arsenal with novel classes of antibiotics that possess unique modes of action that can overcome traditional mechanisms of drug resistance. To discover these therapeutically unique classes, we must shift our emphasis towards organisms from an equally unique environment. This review demonstrates that halophiles from all domains are a promising source of antimicrobials. Among the halophilic microorganisms, Actinobacteria, are of noteworthy interest for their unique metabolic diversity, biological activity, and biotechnological potential. The fastidious nature of extremophiles, especially haloarchaea, requires that biotechnologies such as enhanced isolation e.g. the iChip, genomic mining e.g. antiSMASH, and metagenomic discovery be further exploited to investigate the antimicrobial potential of these underexplored ecological niches. The future of natural product discovery will be the application of a combination of multi-omics approaches to improve the access of these non-culturable microbes, investigate the potential of their silent operons and reduce the high rediscovery rate of known compounds. Both genomic mining and enhanced culture techniques will prove invaluable tools for unlocking new avenues for antimicrobial natural products.

Authors contributions

TT and BG conceived the idea. TT collected literature data, created the tables and figures, and wrote the first draft of the manuscript. BG reviewed and edited the manuscript. Both authors approved the final version of the manuscript.

Acknowledgments

We acknowledge our colleagues at Queen’s University Belfast for the great discussions and feedback. Also, Simon Gibbons from the University of East Anglia for the thoughtful advices.

Disclosure statement

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Additional information

Funding

This work was supported by the Department for Employment and Learning Northern Ireland (via a PhD Scholarship to TT) and Department for the Economy [DFE2019].