Bioprospecting Sponge-Associated Microbes for Antimicrobial Compounds

Sponges are the most prolific marine organisms with respect to their arsenal of bioactive compounds including antimicrobials. However, the majority of these substances are probably not produced by the sponge itself, but rather by bacteria or fungi that are associated with their host. This review for the first time provides a comprehensive overview of antimicrobial compounds that are known to be produced by sponge-associated microbes. We discuss the current state-of-the-art by grouping the bioactive compounds produced by sponge-associated microorganisms in four categories: antiviral, antibacterial, antifungal and antiprotozoal compounds. Based on in vitro activity tests, identified targets of potent antimicrobial substances derived from sponge-associated microbes include: human immunodeficiency virus 1 (HIV-1) (2-undecyl-4-quinolone, sorbicillactone A and chartarutine B); influenza A (H1N1) virus (truncateol M); nosocomial Gram positive bacteria (thiopeptide YM-266183, YM-266184, mayamycin and kocurin); Escherichia coli (sydonic acid), Chlamydia trachomatis (naphthacene glycoside SF2446A2); Plasmodium spp. (manzamine A and quinolone 1); Leishmania donovani (manzamine A and valinomycin); Trypanosoma brucei (valinomycin and staurosporine); Candida albicans and dermatophytic fungi (saadamycin, 5,7-dimethoxy-4-p-methoxylphenylcoumarin and YM-202204). Thirty-five bacterial and 12 fungal genera associated with sponges that produce antimicrobials were identified, with Streptomyces, Pseudovibrio, Bacillus, Aspergillus and Penicillium as the prominent producers of antimicrobial compounds. Furthemore culture-independent approaches to more comprehensively exploit the genetic richness of antimicrobial compound-producing pathways from sponge-associated bacteria are addressed.


Introduction
Antimicrobial resistance (AMR) is an emerging global threat, decreasing the possibilities for prevention and treatment of infectious diseases caused by viruses, bacteria, parasites and fungi [1,2]. A global surveillance report by the World Health Organization (WHO) [2] indicated an increase of morbidity and mortality of infectious diseases due to AMR, which could lead to a world-wide economic loss of up to 100 trillion US dollars (USD) in 2050 as the result of a 2%-3% reduction in the gross domestic product (GDP) [1]. A conservative estimation is that AMR now annually attributes to 700,000 deaths globally, with a potential leap to 10 million in 2050 [1]. AMR is a response of microorganisms against antimicrobial compounds, which can arise via several mechanisms such as chromosomal mutations [1], binding site modifications [2] or horizontal transfer of genes conferring resistance [3]. For several pathogenic bacteria such as Staphylococcus aureus [4], concentration of 0.3 µg/mL [26]. The sponge-associated fungus Stachybotrys chartarum MXH-X73 produces the compound stachybotrin D (3), which exhibited anti-HIV-1 activity by targeting reverse transcriptase [27]. At EC50 concentrations from 2.73 µg/mL to 10.51 µg/mL, stachybotrin D was active not only against the wild type HIV-1 but also against several non-nucleoside reverse transcriptase inhibitor (NNRTI) resistant HIV-1 strains. Li et al. [28] reported identification of three other anti-HIV-1 compounds from Stachybotrys chartarum: chartarutine B, G, and H. Of these three chartarutine compounds, chartarutine B (4) showed the lowest concentration that resulted in 50% inhibition of HIV-1 (IC50 of 1.81 µg/mL), followed by chartarutine G (IC50 of 2.05 µg/mL) and chartarutine H (IC50 of 2.05 µg/mL), respectively.
Sponge-associated microbes have also been found to produce anti-influenza compounds ( Table  1). Zhao et al. [29] elucidated 14 new isoprenylated cyclohexanols coined as truncateols A-N from the sponge-associated fungus Truncatella angustata, and these compounds were tested in vitro against the influenza A (H1N1) virus. Truncateols C, E and M displayed bioactivity against H1N1, with truncateol M (5) being the most potent inhibitor, as shown by its IC50 value of 2.91 µg/mL. This inhibitory concentration was almost six fold lower than that of the positive control oseltamivir at 14.52 µg/mL. Truncateol M was predicted to be active at the late stage of the virus infection, likely during the assembly or release step of the virion [29] due to resemblance of the inhibition patterns observed for neuraminidase-inhibitor drugs, e.g., zanamivir and oseltamivir [30]. In addition, the presence of a chlorine atom in the chemical structure of trucanteol M is of particular interest since halogenation often enhances bioactivity of a given compound [31,32].  Sponge-associated microbes have also been found to produce anti-influenza compounds (Table 1). Zhao et al. [29] elucidated 14 new isoprenylated cyclohexanols coined as truncateols A-N from the sponge-associated fungus Truncatella angustata, and these compounds were tested in vitro against the influenza A (H1N1) virus. Truncateols C, E and M displayed bioactivity against H1N1, with truncateol M (5) being the most potent inhibitor, as shown by its IC 50 value of 2.91 µg/mL. This inhibitory concentration was almost six fold lower than that of the positive control oseltamivir at 14.52 µg/mL. Truncateol M was predicted to be active at the late stage of the virus infection, likely during the assembly or release step of the virion [29] due to resemblance of the inhibition patterns observed for neuraminidase-inhibitor drugs, e.g., zanamivir and oseltamivir [30]. In addition, the presence of a chlorine atom in the chemical structure of trucanteol M is of particular interest since halogenation often enhances bioactivity of a given compound [31,32].  [36] (´)-5-(hydroxymethyl)-2-(2 1 ,6 1 ,6 1 -trimethyltetrahydro-2H-pyran-2-yl)phenol and a known compound (Z)-5-(Hydroxymethyl)-2-(6 1 -methylhept-2 1 -en-2 1 -yl)phenol from a sponge-associated Aspergillus sp. (Table 2). Of these five substances, the compound sydonic acid (11) exhibited the lowest MIC value against Escherichia coli at 1.33 µg/mL. This is the lowest inhibition concentration against E.coli reported from a compound produced by sponge-associated microbes although the inhibition concentration is still higher than the positive control ciprofloxacin (0.21 µg/mL) ( Table 2).
Pruksakorn et al. [49] reported three prospective anti-tuberculosis compounds: trichoderin A (12), A1 and B from the sponge-associated fungus Trichoderma sp. 05FI48. Both under standard aerobic growth and dormancy-inducing hypoxic conditions, these three compounds inhibited Mycobacterium smegmatis, M. bovis BCG, and M. tuberculosis H37Rv with MIC values in the range of 0.02-2.0 µg/mL. Of these three compounds, trichoderin A was the most potent compound indicated by the lowest MIC values against those Mycobacterium strains. Additional analysis revealed that bioactivity of trichoderin A is based on its ability to inhibit adenosine triphosphate (ATP) synthesis of mycobacteria [50]. Compounds such as trichoderin A are particularly important because in many cases, pathogens such as Campylobacter spp., Helicobacter pylori, and Legionella pneumophila are difficult to treat due to the fact that they are present in a dormant state [51]. Such physiologically inactive cells highly contribute to the need for prolongued antibiotic treatments, which may lead to the emergence of resistant strains [52,53].

Antifungal Activity
The incidence rate of fungal infections has increased significantly over the past decades. This is mainly caused by clinical use of antibacterial drugs and immunosuppressive agents after organ transplantation, cancer chemotherapy, and advances in surgery [102,103]. Several fungal species that often cause human infections include Candida albicans, Candida glabrata, Cryptococcus neoformans and Aspergillus fumigatus [102,104,105]. The story becomes more complex as many of these pathogenic fungi develop resistance against available antifungal drugs, which will prolong duration of treatments [106].
Screening for antifungals is often focused on finding compounds active against Candida albicans, the prominent agent for candidiasis (Table 3). Invasive candidiasis is accounted as the most common nosocomial fungal infection resulting in an average mortality rate between 25%-38% [103]. El-Gendy et al. [107] isolated Streptomyces sp. Hedaya 48 from the sponge Aplysina fistularis and identified two compounds: the novel compound saadamycin (13) and the known compound 5,7-dimethoxy-4-p-methoxylphenylcoumarin (14) (Figure 3). Bioassays indicated that both saadamycin and 5,7-dimethoxy-4-p-methoxylphenylcoumarin displayed pronounced antifungal activity against Candida albicans with MIC values of 2.22 µg/mL and 15 µg/mL, respectively. In addition, both compounds displayed bioactivity against some pathogenic dermatophytes (skin-infecting fungi), such as Epidermophyton floccosum, Trichophyton rubrum, Trichophyton mentagrophytes, Microsporum gypseum, Aspergillus niger, Aspergillus fumigatus, Fusarium oxysporum, and Cryptococcus humicolus (Table 3). Further analysis showed that saadamycin displayed a more potent bioactivity indicated by a 3875 fold lower MIC than that of the reference compound, miconazole, whereas 5,7-dimethoxy-4-p-methoxylphenylcoumarin was around a 200 fold more potent than miconazole. mainly caused by clinical use of antibacterial drugs and immunosuppressive agents after organ transplantation, cancer chemotherapy, and advances in surgery [102,103]. Several fungal species that often cause human infections include Candida albicans, Candida glabrata, Cryptococcus neoformans and Aspergillus fumigatus [102,104,105]. The story becomes more complex as many of these pathogenic fungi develop resistance against available antifungal drugs, which will prolong duration of treatments [106].
Antifungal activity was also detected from the sponge-associated fungus Phoma sp. Q60596. The sponge-derived fungus produced a new lactone compound, YM-202204 (15) [108], which was effective against C. albicans (IC80 of 6.25 µg/mL), along with Cryptococcus neoformans (IC80 of 1.56 µg/mL), Saccharomyces cerevisiae (IC80 of 1.56 µg/mL) and Aspergillus fumigatus (IC80 of 12.5 µg/mL). Furthermore, Nagai et al. [108] showed that YM-202204 was able to block the glycophosphatidylinositol (GPI) anchor, an important structure for protein attachment in the membrane of eukaryotic cells and one of the targets in developing antifungal drugs [109,110].  Antifungal activity was also detected from the sponge-associated fungus Phoma sp. Q60596. The sponge-derived fungus produced a new lactone compound, YM-202204 (15) [108], which was effective against C. albicans (IC 80 of 6.25 µg/mL), along with Cryptococcus neoformans (IC 80 of 1.56 µg/mL), Saccharomyces cerevisiae (IC 80 of 1.56 µg/mL) and Aspergillus fumigatus (IC 80 of 12.5 µg/mL). Furthermore, Nagai et al. [108] showed that YM-202204 was able to block the glycophosphatidylinositol (GPI) anchor, an important structure for protein attachment in the membrane of eukaryotic cells and one of the targets in developing antifungal drugs [109,110].

Antiprotozoal Activity
Malaria, caused by Plasmodium spp. infections, represents the most devastating protozoal disease worldwide, and results in both mortality and economic loss, mainly in developing countries [116]. Developing drugs with a better therapeutic profile against the parasite is one of the key aims of current malaria research, which includes screening for antimalarial substances from marine organisms [117,118].
Manzamine A (16) (Figure 4), first reported by Sakai and co-workers [119] from the sponge Haliclona sp., is a promising substance against Plasmodium spp. Initially, its antitumor property was of main interest, but subsequently diverse antimicrobial activities such as: anti-HIV, antibacterial, and antifungal were identified from the compound [120]. Currently the antimalaria properties of manzamine A are considered its most promising bioactivity. Manzamine A was shown to inhibit P. falciparum D6 and W3 clonal cell lines that are sensitive and resistant against the antimalarial chloroquine [121], with IC 50 values of 0.0045 and 0.008 µg/mL, respectively [122]. Furthermore, in vivo screening by Ang et al. [116] showed that manzamine A at concentration of 0.008 µg/mL inhibited 90% growth of the parasite Plasmodium berghei that causes malaria in rodents. In addition, Rao et al.
Isolation of manzamine A from several other sponge species [120] raised the hypothesis that it was of microbial origin [123,124]. Hill et al. [125] confirmed this hypothesis by isolating Micromonospora sp. M42 as the microbial producer of manzamine A from the Indonesian sponge Acanthostrongylophora ingens. A series of analyses using molecular-microbial community analysis, and Matrix Assisted Laser Desorption Ionization-Mass Spectrometry (MALDI-MS) corroborated that indeed the strain Micromonospora sp. M42 synthesizes manzamine A [126,127]. Considering the therapeutic potential of manzamine A for treating malaria and leishmaniasis, Micromonospora sp. M42 could be a sustainable provider of the substance, because the "Sponge Supply Problem" has been overcome [127]. Moreover, identification of several manzamine-derivatives e.g. manzamine E, F, J, and 8-hydroxymanzamine A, from marine sponges which displayed antibacterial, antifungal and antiprotozoal activity [122,124], could also lead to isolation of associated microbial producers in the future.
Scopel et al. [129] isolated two sponge-associated fungi, namely Hypocrea lixii F02 and Penicillium citrinum F40 ( Table 4) that were active against the protozoal parasite Trichomonas vaginalis, which causes trichomoniasis, a sexually transmitted disease [130]. Culture filtrates of both isolates inhibited T. vaginalis ATCC 30236 and fresh clinical isolates, including the metronidazole-resistant TV-LACM2, with MIC values of 2.5 mg/mL. Further observation indicated that culture filtrates of these two fungi had no haemolytic effect against mammalian cells, which is one of the important criteria to further develop anti-protozoal drugs [129].

Antimicrobial Compounds from Sponge-Associated Microbes: What We Learned So Far
Bioprospecting is the effort to discover natural compounds with therapeutic and biological applications [140]. In line with this definition, sponge-associated microbes offer a huge potential as the source of antimicrobial substances as shown by many microbial isolates being reported to inhibit pathogenic reference strains in vitro and to synthesize active substances against one or several groups of infectious agents. Based on our review, antimicrobial compounds produced by sponge-associated microbes with the most pronounced bioactivity include: 2-undecyl-4-quinolone, sorbicillactone A, stachybotrin D and chartarutine B against HIV-1; truncateol M against H1N1 M; YM-266183, YM-266184, kocurin, mayamycin, sydonic acid, naphthacene glycoside SF2446A2 and trichoderin A against a variety of bacterial strains; saadamycin and YM-202204 against fungi; manzamine-A against malaria; and valinomycin against Trypanosoma. In this case the most pronounced activity is solely based on reported inhibition data and does not yet take potential side effects into account. Therefore the most promising compounds may be ones that have higher IC50 values, but cause less side effects. As these data are not available for the majority of the reported compounds, we have focused on the most potent compounds.
Sponge-associated bacteria and fungi are the two groups of microorganisms that have been found to produce antimicrobial compounds ( Figure 5). The large majority of the antimicrobial compounds found in sponge-associated microbiota is produced by bacteria (90%), while fungi account for approximately 10% of the compounds reported. Sponge-associated bacteria derived antimicrobial compounds were found from 35 genera (Figure 5B). At a higher taxonomic level, these 35 bacterial genera can be classified into the four phyla Actinobacteria, Proteobacteria, Firmicutes and Cyanobacteria with percentages of 48.8%, 36.6%, 11.4% and 0.4% respectively. In contrast, sponge-associated fungi that have been found to produce antimicrobials are affiliated solely to the phylum Ascomycota.

Antimicrobial Compounds from Sponge-Associated Microbes: What We Learned So Far
Bioprospecting is the effort to discover natural compounds with therapeutic and biological applications [140]. In line with this definition, sponge-associated microbes offer a huge potential as the source of antimicrobial substances as shown by many microbial isolates being reported to inhibit pathogenic reference strains in vitro and to synthesize active substances against one or several groups of infectious agents. Based on our review, antimicrobial compounds produced by sponge-associated microbes with the most pronounced bioactivity include: 2-undecyl-4-quinolone, sorbicillactone A, stachybotrin D and chartarutine B against HIV-1; truncateol M against H1N1 M; YM-266183, YM-266184, kocurin, mayamycin, sydonic acid, naphthacene glycoside SF2446A2 and trichoderin A against a variety of bacterial strains; saadamycin and YM-202204 against fungi; manzamine-A against malaria; and valinomycin against Trypanosoma. In this case the most pronounced activity is solely based on reported inhibition data and does not yet take potential side effects into account. Therefore the most promising compounds may be ones that have higher IC 50 values, but cause less side effects. As these data are not available for the majority of the reported compounds, we have focused on the most potent compounds.
Sponge-associated bacteria and fungi are the two groups of microorganisms that have been found to produce antimicrobial compounds ( Figure 5). The large majority of the antimicrobial compounds found in sponge-associated microbiota is produced by bacteria (90%), while fungi account for approximately 10% of the compounds reported. Sponge-associated bacteria derived antimicrobial compounds were found from 35 genera (Figure 5B). At a higher taxonomic level, these 35 bacterial genera can be classified into the four phyla Actinobacteria, Proteobacteria, Firmicutes and Cyanobacteria with percentages of 48.8%, 36.6%, 11.4% and 0.4% respectively. In contrast, sponge-associated fungi that have been found to produce antimicrobials are affiliated solely to the phylum Ascomycota. nitrogen-containing heterocyclic compound with a wide range of biological activities [67,144], and several studies from terrestrial environments and chemically synthesized phenazines have been reported as antiviral [145], antibacterial [146], and antimalaria [147]. Moreover, this group of compounds is attractive for therapeutic application since their structures are relatively small and hence can easily reach tissues and organs [67,148].  Streptomyces is the most prominent genus as indicated by 30% of sponge bacteria-derived compounds. Streptomyces has become a main target for screening for bioactive compounds both from terrestrial and marine environments due to the high diversity of secondary metabolites they produce [141,142]. Of the many sponge-associated Streptomyces isolates reported, Streptomyces sp. HB202 and Streptomyces sp. RV15 are of particular interest in term of producing antibacterial compounds. Streptomyces sp. HB202, isolated from the sponge Halichondria panicea has been documented to produce three antibacterial substances: mayamycin, streptophenazine G and K, which are mainly active against Gram positive pathogenic bacteria (Table 2). Streptomyces sp. RV15, on the other hand, produces the compound naphthacene glycoside which up to now is the only anti-Chlamydia reported from sponge-associated microbes [46]. In addition, the report on crude extract inhibition of Streptomyces sp. RV15 against S. aureus and E. faecalis [82] may give a hint to discover other antibacterial substances from this strain. Streptomyces sp. Hedaya48 is currently the most potent sponge-associated bacterial isolate for antifungal activities with the production of saadamycin and 5,7-dimethoxy-4-p-methoxylphenylcoumarin [107]. In addition, isolation of the anti-Trypanosoma and anti-Leishmania compounds valinomycin, staurosporine and butenolide from Streptomyces sp. 43, 21 and 11 [128], affirms Streptomyces as the currently most prominent producer of antimicrobial substances from sponges.
Pseudovibrio follows as the second most prolific bacterial genus isolated from sponges (20%) with respect to antimicrobial activities. Reports on Pseudovibrio spp. are concentrated on antibacterial activity and are mainly based on screening of crude extracts. Up to now, tropodithietic acid is the only antibacterial compound that has been identified from Pseudovibrio [72]. Although representing a lower percentage of the sponge-associated bacteria found to produce antimicrobials than Streptomyces and Pseudovibrio, 9% of the currently known bioactives was found to be produced by sponge-associated Bacillus spp., with activities against viruses, bacteria and fungi. Bacillus cereus QNO3323 is currently the most prominent antimicrobial producer from this genus with the very potent thiopeptides YM-266183 and YM-266184 that are active against Gram positive bacteria.
Sponge-associated Ascomycota found to produce antimicrobials can be further classified into 12 genera. Of these 12 fungal genera, Aspergillus (30%) and Penicillium (23%) are currently the two most prominent groups of sponge-associated fungi reported as antimicrobial producers. This finding is not suprising since both Aspergillus and Penicillium are known prolific producers of secondary metabolites from other sources [143]. Aspergillus versicolor [58] and an unidentified Aspergillus sp. isolated from the sponge Xestospongia testudinaria [48] showed a strong antibacterial activity as indicated by potent inhibition of pathogenic bacteria. The antimicrobial activities found from sponge-associated Penicillium spp. are particular remarkable as it is the only fungal genus that is found to produce antivirals, antibacterials antifungals and antiprotozoals. Penicillium chrysogenum [26] and Penicillium sp. FF01 [57] are to date the most promising sponge-associated Penicillium isolates for which anti-HIV activity (sorbicillactone) and antibacterial activity (citrinin) were reported, respectively. Sponge-derived Stachybotrys spp. are only known for antiviral activity, particularly against HIV and enterovirus 71 (EV71), and there are no reports of other antimicrobial activities. Generally, although the number of produced antimicrobials is outnumbered by those of sponge-associated bacteria, sponge-associated fungi should be considered as an important reservoir of antimicrobial compounds.
Phenazine is a nitrogen-containing heterocyclic compound with a wide range of biological activities [67,144], and several studies from terrestrial environments and chemically synthesized phenazines have been reported as antiviral [145], antibacterial [146], and antimalaria [147]. Moreover, this group of compounds is attractive for therapeutic application since their structures are relatively small and hence can easily reach tissues and organs [67,148].
6.2. Discovering Antimicrobial Compounds from Sponge-Associated Microbes: From Culture-Dependent to Culture-Independent Methods Isolation of antimicrobial producers provides a valuable basis for assessing the biotechnological potential of sponge-associated microbes. In a wider perspective, however, only a small fraction of this sponge-microbial community has been isolated under laboratory conditions leaving the majority resistant to in vitro growth with current cultivation approaches [15,149,150]. Several studies have focused on improving cultivability of sponge-associated microbes. Some of the approaches include using low nutrient media [151], floating filter cultures [152], employing different carbon sources, e.g., lectin [153], sponge extracts [152], and in situ cultivation using a diffusion growth chamber [154]. Furthermore, flow-cytometry and density gradient centrifugation have been applied to separate sponge cells from their associated bacteria to enrich the inoculum [155,156]. Additionally, co-cultivation through mixing of two or more microbial isolates in vitro [157] is an approach proposed to discover more natural compounds from sponge-associated microbes. The idea behind co-culture lies in the fact that many biosynthetic gene clusters found in microorganisms remain cryptic under standard laboratory conditions, and co-cultivation might provide a possibility to activate these silent genes [158,159]. As an example, the co-culture by Dashti et al. [98] of the sponge-associated Actinobacteria, Actinokinespora sp. EG49 and Nocardiopsis sp. RV163, resulted in isolation of the antibacterial compound 1,6-dihydroxyphenazine, which was not found from the individual isolates. However, even if the cultivability of sponge-associated microbes is improved, there is a long way ahead to reach a point that we will be able to isolate and routinely cultivate 50% of the microbes that are found in sponges. At the same time, the advance of genetic and molecular studies has resulted in the development of tools to study genes, transcripts and proteins by directly analyzing environmental DNA, RNA and proteins, thus bypassing cultivation procedures [157]. In relation to screening for antimicrobial activity, metagenomics has been applied to identify antimicrobials of uncultivated microorganisms from terrestrial environments, such as the antimycobacterial nocardamine, the putative antibacterial activity of terragines A-E [160], violacein that is active against S. aureus, Bacillus sp. and Streptococcus sp. [161] and a polyketide with activity against the yeast Saccharomyces cerevisiae [162].
Two main metagenomic approaches, functional screening and sequence homology-based methods, are generally distinguished [163]. Functional screening relies on detection of the metabolic activities of metagenomic library clones without requiring any prior sequence information [163][164][165]. Gillespie et al. [9] applied function-based metagenomics with E. coli as expression host, to identify the antibiotics turbomycin A and B from a soil sample. MacNeil et al. [166] identified the antimicrobial indirubin by constructing a BAC (bacterial artificial chromosome) library in E.coli. Yung et al. [167] reported two hydrolytic enzymes from fosmid clones CcAb1 and CcAb2, which were derived from a metagenome of the sponge Cymbastela concentrica using E. coli as the host. Both fosmid clones inhibited the growth of Bacillus sp. with an inhibition diameter of 20 mm, and clone CcAb1 showed additional inhibition of S. aureus and an Alteromonas sp. with diameters of inhibition of 50 mm and 60 mm, respectively. Further phylogenetic analysis showed that active genes encoding for these enzymes were of microbial origin [167]. He et al. [168] constructed a fosmid library of the sponge Discodermia calyx using E. coli as the host and identified antimicrobial activity of the enzyme 3-hydroxypalmitic acid against B. cereus and C. albicans. In addition, using the same approach He et al. [169] observed an active clone, pDC113, that displayed a clear inhibition zone against B. cereus. Subsequently, 11 cyclodipeptides were identified from this clone. Generally, it can be stated that although a number of antimicrobials have been discovered through functional screening of metagenomic libraries from sponges, the expression of large gene clusters such as those encoding (polyketide synthase( PKS) and (non-ribosomal peptide synthetase (NRPS) is still a difficult hurdle to take. Several key elements need to be considered to achieve successful expression of biosynthetic gene clusters; namely mobilizing the biosynthetic pathway into a suitable vector, selecting an appropriate heterologous host and stably maintaining the gene clusters in the host [170]. The size of many of these gene clusters requires the use of cloning vectors that can accept large inserts, such as fosmids, or BACs if the required insert size is over 100 kb [171]. Selection of heterologous expression systems in particular is a crucial factor before applying functional metagenomics to identify antimicrobials, because expression hosts are microbes as well and especially clones that express genes encoding for enzymes involved in production of antimicrobials may therefore be non-viable. Ongley et al. [170] pointed out some considerations in selecting an expression host such as relatedness to the native producer, availability of genetic tools and precursors, a high growth rate, and suitability for fermentation at a large scale. E. coli, the most commonly used expression host, has limitations for expressing parts of metagenomes because, e.g., of the sheer size of some gene clusters, genes with deviating codon usage, incompatible regulatory elements, lack of biosynthesis precursors or unavailability of posttranslational modifications [165,172]. Therefore, in order to make screening for antimicrobials through metagenomic libraries more efficient, it is of utmost importance to diversify the suite of expression hosts used. Several non-E.coli hosts, such as Agrobacterium tumefaciens, Bacillus subtilis, Burkholderia graminis, Caulobacter vibrioides, Pseudoalteromonas haloplanktis, Pseudomonas putida, Ralstonia metallidurans, Rhizobium leguminosarum, Streptomyces avermitilis, S. albus, Pseudomonas putida, Sulfolobus solfataricus, Thermus thermophilus, Thiocapsa roseopersicina and Saccharopolyspora sp. have been developed and should be more seriously considered as expression hosts when performing metagenomic screenings for antimicrobials [165,172,173].
Sequence-based screening, on the other hand, requires information on the sequence of genes involved in the production of a natural product as guidance to search for similar sequences in a sequenced metagenomic library or scaffolds reconstructed from direct metagenomic sequencing [165]. Homology-based screening is suitable to identify a compound with highly conserved biosynthesis pathways, e.g., those mediated by PKS and NRPS [174]. Piel and colleagues [175][176][177][178][179] applied this method, and identified the antitumor polyketide onnamide from uncultivated bacteria of the sponge T. swinhoei. Sequence-based screening was applied by Fisch [180] to unravel the complete pathway of the polyketide psymberin that was found to possess a potent antitumor activity, from uncultivated sponge-associated microbes. By sequence-based screening of metagenomic libraries, Schirmer et al. [181] reported diverse polyketide gene clusters in microorganisms from the sponge Discodermia dissoluta. The development of techniques that yield longer read lengths, such as Pacific Biosciences (PacBio) RS II SMRT (Single Molecule Real-Time) sequencing technology, in which a single read can be extended over 10 kbp [182], can be instrumental in increasing the accuracy in assembling large gene clusters. Application of PacBio for secondary metabolite gene clusters has been reported by Alt and Wilkinson [183], who identified the 53,253 bp genomic fragment encoding the transacyltransferase (trans-AT) polyketide synthase (PKS) from a marine Streptomyces sp responsible for the production of the antibiotic anthracimycin (atc). Furthermore, using Streptomyces coelicolor as heterologous expression host, the authors confirmed production of anthracimycin [183]. Furthermore, single cell analysis by combining cell separation and fluorescence-assisted cell sorting (FACS) could be a strategy to overcome the complexity of the microbial community in sponges since this method can be used to select for genomes from microbes that are present in low abundance in the sponge leading to a simplified reconstruction of secondary metabolite gene clusters present in these bacteria [184]. This strategy has been applied by Wilson et al. [185] for resolving the gene clusters encoding the machinery needed for the production of the polytheonamides produced by the candidate genus Entotheonella from the sponge Theonella swinhoei.
Inspired by these examples, homology-based screening could be further exploited to identify biosynthesis gene sequences that could lead to the identification of novel antimicrobial substances from Nature's excessive diversity. Moreover, application of homology-based screening can benefit from publicly available metagenomic sequencing data and prediction tools for analyzing biosynthesis gene clusters, e.g., AntiSMASH (Antibiotics and Secondary Metabolite Analysis Shell) [186,187]. Application of sequence-based screening, however, is limited by the fact that the found sequences need to be related to known compounds, inherently limiting the potential for novelty. Furthermore, information on gene sequences is no guarantee that the acquisition of a complete gene pathway has been obtained [188]. Therefore, sequence-based methagenomics should ideally be complemented by chemical analysis to confirm whether the predicted compound exists and is fully functional (Figure 6).

Conclusion and Outlook
Sponge-associated microbes already offer a rich source of potent antimicrobial compounds against viruses, bacteria, protozoa and fungi, and currently available compounds are predominantly active against HIV-1, H1N1, nosocomial Gram positive bacteria, Escherichia coli, Plasmodium spp, Leishmania donovani, Trypanosoma brucei, Candida albicans and dermatophytic fungi. Streptomyces, Pseudovibrio, Bacillus, Aspergillus and Penicillium are the microbial genera associated with sponges from which potent antimicrobial compounds are most frequently isolated. However, none of the antimicrobial compounds highlighted in this review have been succcesfully marketed as pharmaceuticals. To clearly translate bioactivity of these important compounds it is crucial to further unravel their mode of actions and measure their level of toxicity, since the majority of these studies has been focused on in vitro bioassays and elucidation of the chemical structures only.
The known versatility of antimicrobial activities found in sponge-associated microorganims could easily be expanded even without considering additional sponge sampling campaigns. Bioactivity screens of identified compounds or undefined sponge extracts is often restricted to a specific antimicrobial activity. The selection, for instance, relies on the specific research activities of the groups involved in isolating the microbes [117]. Consequently, it is probably safe to assume that other potent antimicrobial properties from many sponge isolates and their bioactive compounds remain undetected. Therefore, known antimicrobial compounds and producer strains are a valuable source for additional antimicrobial activities screenings using different target types (viruses, bacteria, fungi, protozoa and beyond). In addition, sponge-derived strain collections that comprise isolates Figure 6. General overview of the strategies used to discover antimicrobial compounds from sponge-associated microorganisms.

Conclusions and Outlook
Sponge-associated microbes already offer a rich source of potent antimicrobial compounds against viruses, bacteria, protozoa and fungi, and currently available compounds are predominantly active against HIV-1, H1N1, nosocomial Gram positive bacteria, Escherichia coli, Plasmodium spp, Leishmania donovani, Trypanosoma brucei, Candida albicans and dermatophytic fungi. Streptomyces, Pseudovibrio, Bacillus, Aspergillus and Penicillium are the microbial genera associated with sponges from which potent antimicrobial compounds are most frequently isolated. However, none of the antimicrobial compounds highlighted in this review have been succcesfully marketed as pharmaceuticals. To clearly translate bioactivity of these important compounds it is crucial to further unravel their mode of actions and measure their level of toxicity, since the majority of these studies has been focused on in vitro bioassays and elucidation of the chemical structures only.
The known versatility of antimicrobial activities found in sponge-associated microorganims could easily be expanded even without considering additional sponge sampling campaigns. Bioactivity screens of identified compounds or undefined sponge extracts is often restricted to a specific antimicrobial activity. The selection, for instance, relies on the specific research activities of the groups involved in isolating the microbes [117]. Consequently, it is probably safe to assume that other potent antimicrobial properties from many sponge isolates and their bioactive compounds remain undetected. Therefore, known antimicrobial compounds and producer strains are a valuable source for additional antimicrobial activities screenings using different target types (viruses, bacteria, fungi, protozoa and beyond). In addition, sponge-derived strain collections that comprise isolates that tested negative for antimicrobial activity at first may have done so, because the compound of interest is not produced under standard laboratory conditions. Exposure of these strains to potential microbial targets may lead to recovery of bioactivity that would otherwise go unnoticed.
Ideally, researchers who isolate microbes from sponges will deposit them to publicly available culture collections so that laboratories with complementary expertise and interests could benefit and screen the deposited isolates for different antimicrobial activities. This will make exchange of materials and knowledge that can be obtained much more efficient. Importantly, a fair agreement on intellectual property rights needs to be established for translating this into reality. Lastly, the revolutionary advance of next generation sequencing technologies combined with more diversified heterologous expression systems ( Figure 6) are expected to open up the large unexplored reservoir of antimicrobials produced by yet uncultivated sponge-associated microbes.