1. Introduction
Cyanobacteria belong to an ancient group of photosynthetic prokaryotes that present a very wide range of cellular strategies, physiological capacities, and adaptations that support their colonization of very diverse microenvironments worldwide. As a consequence, cyanobacteria occur in varied and often extreme habitats and are then able to settle in diverse biotopes (e.g., marine, terrestrial, freshwater, thermal springs) [
1,
2,
3]. They are also well known for the production of a wide variety of bioactive natural products, including some potent toxins (e.g., microcystins, anatoxins, and saxitoxins) [
2,
3]. Due to the remarkable capability of cyanobacteria to proliferate and form toxic blooms that induce potential human health consequences [
4], numerous studies have been conducted to develop tools for the monitoring of such blooms [
5,
6] or effective strategies for the mitigation of their overgrowth [
7]. On the contrary, certain cyanotoxins could also constitute a promising opportunity for drug development such as certain cancer therapies [
8].
Two main aspects known as the chemical diversity and the related bioactivity have to be considered when considering the application potential of natural products produced by cyanobacteria. The chemical diversity of metabolites produced by these organisms has been well described and about 15 reviews have been already published in the past 20 years, dealing with their structural and chemical diversity [
9,
10,
11,
12,
13,
14] or their respective biosynthetic pathways [
15,
16]. Beyond the notorious harmful effects of cyanotoxins, other cyanobacterial natural products show a wide range of bioactivities that could be potentially useful for diverse applications [
17,
18,
19,
20,
21]. So far, among the existing reviews related to the diversity of cyanobacterial metabolites, only one has addressed the relative taxonomical positions of the different producing strains [
9]. A few taxa appear to be especially prolific producers of a large set of metabolites, while others still remain to be investigated. Recent genomics approaches and genome sequencing have been important steps in the elucidation of the pathways implicated in the biosynthesis of natural products. Their wide structural diversity has been described as a consequence of the numerous biosynthetic pathways developed by cyanobacteria in order to produce these metabolites [
15]. Most of the active cyanobacterial molecules are considered as being produced either through the non-ribosomal peptide (NRP) or the hybrid polyketide-NRP biosynthetic pathways [
10], or by the ribosomal synthesis of pro-peptides that are post-translationally modified (RiPP). Previous genome analysis demonstrated that the diversity of the known metabolites is merely a fraction of the true metabolic potential of cyanobacteria [
15]. Concerning bioactivity, cyanobacteria have long been a source of molecules with a potent nutritional property [
18]. Aztec civilizations consumed cyanobacteria (
Spirulina) in their routine diet [
22], and Chadian populations still use them as one of their substantial food sources [
23]. Besides nutritional and probiotic purposes [
13,
21], cyanobacteria are well-known as an important source of metabolites with technological applications in the biotechnical or pharmaceutical fields, which lead to an increase in interest in these research realms [
10]. Most bioactivities described to date are the antibacterial, antifungal, anti-cancerous, immunosuppressive, anti-inflammatory, and anti-tuberculosis activities that have the potential to be used in fields such as pharmacology, cosmetology, agriculture, the food industry, or as biofuel [
17]. Cyanobacteria cells represent a sustainable resource for biotechnology due to their photosynthetic, N-fixation, and autotrophic capacities [
17,
18,
24]. Due to the current increase in their pharmaceutical value and in their application prospects for use in medicine or biotechnology, the exploration of uncovered cyanobacterial taxa constitutes a promising strategy to efficiently explore the chemical diversity of their bioactive compounds.
The present review globally and systematically describes current knowledge on the biological activities described for cyanobacterial natural products, and, thanks to the construction of a specific and freely available molecular database, regroups all information described so far concerning the chemical structures, the producing organisms, and the various bioactivities of all the different cyanobacterial metabolite families. This original material allows us to depict, from data based on exhaustive literature, which kinds of bioactive metabolite are potentially produced by the different cyanobacterial taxa. In this case, the producer organisms were considered at different taxonomic levels (family, order, and genus) and are referenced according to their original habitats (freshwater, marine, and others). The chemical diversity is described with respect to the different kinds of bioactivity and the potential links between them are questioned, according to their potential or effective molecular mechanisms of action. A specific focus on 47 cyanobacterial compounds presenting beneficial bioactivities is detailed and discussed regarding their potential in pharmaceutical, cosmetical, biotechnical, and agricultural applications, which opens new perspectives on the discovery of novel and potent bioactive cyanobacterial molecules.
2. Methods for Dataset Construction
A database was constructed using different search engines, notably PubMed and Google Scholar. The keywords used were “cyanobacteria,” “metabolite” or “natural product,” “beneficial” and “activity,” or “biological properties.” The database was first based on reviews and further completed with recent publications dealing with the isolation of new compounds from cyanobacteria.
The main entries into the database were the names of the metabolites. To avoid bias in the counting of metabolites, we stored all the data for each molecule and its variants as a “family.” In fact, there are still no molecular classification references for a natural product description. As discussed by Janssen [
25], there is no standardized naming system along cyanobacterial metabolites, as in natural product discovery in general, that could induce an underestimation of the real diversity of natural products and to hide the potential link between their chemical structures, biosynthetic pathways, and evolution routes. Thus, such a valuable classification of cyanobacterial metabolites is still needed, notably in the current context of genomic and metabolomic development.
In our database, metabolites were grouped and classified based on different criterion, initially selected by different authors [
13,
15,
25]. First, they were classified according to their biosynthetic pathways based on the genomic data reviewed by Dittmann et al. (e.g., microcystins, cryptophycins, and aeruginosins) [
15]. Secondly, when biosynthetic information was not available, metabolites were classified, according to their structural homology, as proposed by Boudreau et al., [
26], Janssen [
25], and Chlipala et al. [
13], supposing that they might be sharing at least a similar, if not the same, biosynthetic route (e.g., kulolide-family, aerucylamides, and cyanopeptolins). In most cases, metabolite variants have a few differences occurring on a few residues and have conserved the specific structure of their metabolite family. For example, the cyanopeptolin-like family, which contains, so far described, 139 variants, is comprised of a core structure of six amino acid residues and a variable side chain containing between 1 and 3 residues. The sequence of the amino acids in the core structure is usually composed with: Thr − [Leu|Arg|Tyr] − Ahp − [Ile|Phe|Thr|Leu|Val] − N-Me[Tyr|Phe] − [Val|Ile] (see
Supplementary Data S1). Some amino acids are variable (in brackets) and some others are identical in the large majority of the variants, notably the 3-amino-6-hydroxy-2-piperidone (Ahp), and the threonine (Thr) that support the side chain and close the cycle with an ester bond linkage (
S1) [
13].
The data collected were then classified depending on the chemical class of the compound, the chemical structure, and the strain producing the metabolites with all the taxonomic information (species, genus, family, and order), in accordance with Komarek et al. (2014) [
27]. In addition, we compiled the demonstrated activities for the purified compounds. Fourteen classes of activity were mostly tested through the literature: lethality (against brine shrimp, and other small invertebrates), neurotoxicity, hepatotoxicity, dermal toxicity, cytotoxicity, anti-inflammatory activity, antioxidant activity, antiviral, antibacterial, antifungal, antialgal, antiprotozoal, serine protease inhibition, and other types of enzyme inhibition.
Additionally, 670 publications were analyzed, dating from the 1970s until today (April 2019). Around 1630 unique molecules have been reported so far and were grouped in 260 families of metabolites (see
Supplementary Data S2). To validate the knowledge depth of our work, a rarefaction curve of the number of molecule families was constructed using the number of analyzed publications (
Figure 1).
3. Taxonomy of the Producing Strains
The 260 families of molecules were attributed to cyanobacteria at their different taxonomic levels (order, family, and genus) (
Figure 2). Some families of compounds can be produced by different strains and, thus, occur at different taxonomical levels. For example, microcystins are produced by various strains belonging to seven different genera, five families, and three orders.
The Oscillatoriales produces the largest number of metabolite families (153 families, 46.5%). The strains belonging to the Nostocales are also considerable producers of metabolites with 98 families (29.7%). The other main producers are the strains belonging to Chroococcales and Synechococcales, which exhibit, respectively, 34 and 31 described molecule families (10.3% and 9.4%). It is interesting that, except for these four orders, the others (i.e., Pleurocapsales, Chroococcidiopsales, Gloeobacterales, and Spirulinales) remain weakly represented in the database: less than five families of metabolite have been reported so far for all of them.
Some metabolites have been isolated from cyanobacterial assemblages without accurate identification of the producer organisms. For these cases, the authors identified the genera of the two dominant cyanobacteria of the assemblage but could not accurately determine which one of them produces which molecule [
28,
29,
30,
31,
32,
33,
34,
35,
36,
37,
38,
39,
40]. Tidgewell et al. (2010) [
9] also identified the prevalence of marine cyanobacterial products within Oscillatoriales and Nostocales with 58% and 24% of the isolated molecules, respectively. Within Oscillatoriales, members of the genus
Lyngbya, and, notably,
Lyngbya majuscula produce the highest number of metabolites. This benthic genus is widely spread through the tropical marine ecosystem and has been widely studied because of its toxicity and implication in many dermatitis cases around the world [
41,
42]. A number of studies have been conducted on the
Lyngbya genus, and a high number of new metabolites have been described. In fact,
Lyngbya is, to date, the most productive genus of bioactive cyanobacterial compounds (
Figure 2B). Recent studies showed that
Lyngbya is polyphyletic [
27,
43] and using polyphasic approaches,
Lyngbya has been split into four new genera:
Moorea [
44],
Okeania [
45],
Limnoraphis [
46], and
Microseira [
47]. Some marine strains previously identified morphologically as
Lyngbya majuscula and
Lyngbya sordida were, therefore, renamed as
Moorea producens, and some strains of
Lyngbya bouillonii were renamed to
Moorea bouillonii on the basis of molecular and phylogenetic analyses [
44]. In the same way, some freshwater strains morphologically identified as
Lyngbya wollei were separate from the
Lyngbya genus and described as
Microseira wollei after analysis of their phylogenetic position [
47].
According to this information, we decided to present the number of metabolite families produced by the
Lyngbya and the
Moorea genera together (reported as
Lyngbya-Moorea in
Figure 2B), given that the majority of families isolated from
Lyngbya species were reported to be from
Lyngbya majuscula (46 of 78 described from all the
Lyngbya) or from
Lyngbya spp. strains sampled from tropical marine environments (22 of 78), as described for the
Moorea genus and were possibly misidentified with regard to this newly described genus [
44].
At the family level, the main producers of known bioactive compounds belong to Oscillatoriaceae (30.3%, producing 122 families of compounds), followed by Nostocaceae and Microcoleaceae (17.2% and 10.9% for 69 and 48 molecule families, respectively) (
Figure 2A). At the genus level (
Figure 2B),
Lyngbya-Moorea exhibits the highest number of isolated compounds (85 families of metabolites representing 20.6%), in accordance with the perceived richness of production for the
Lyngbya genus due to its polyphyletic status [
48].
Nostoc is the second most prolific genus of bioactive compound families with 50 isolated families so far (12.1% of the total number of families of metabolites). The other most important genera are
Anabaena,
Oscillatoria, and
Microcystis (with 32, 31, and 27 families of molecules, respectively, representing 7.8%, 7.5%, and 6.6%) (
Figure 2B).
When looking at the habitats of these cyanobacteria, a large number of compounds were isolated from marine environments (148 families of metabolite in the database, which means 53% of the families of metabolites) in comparison to the number of strains isolated from freshwater environments (77 families of metabolites, 27.6%) (
Figure 2B). However, this difference might be at least partly due to the high number of compounds isolated from the marine species
Lyngbya majuscula-Moorea producens (49 families of molecules, 18.8% of the families in the database) and to the existence of various research programs focused on marine species (e.g., the Panama International Cooperative Biodiversity Group, ICBG).
Overall, we observed that diversity at the genus level is important, as illustrated by the 90 different genera present in the database. Moreover, 65 different genera have been reported to produce less than four molecules (
Figure 2B). We also noticed that five molecules were isolated from
Lyngbya/
Schizothrix assemblages and five others from unidentified strains of cyanobacteria (
Figure 2B). Thus, at the genus level, the diversity of producers is large with a high number of genera studied (90 different genera). Nevertheless, these genera generally belong to the same orders (e.g., Oscillatoriales, Nostocales, Synechococcales, and Chroococcales) while some orders were not studied. For example, among the Pleurocapsales order, only four genera have been reported to produce metabolites. As a result, the covered diversity appears not to be exhaustive and can still be increased.
According to Shih et al. (2013) [
49], the genomic potential of cyanobacteria to produce secondary metabolites is high with more than 70% of the studied strains presenting non-ribosomal peptide synthase (NRPS) or polyketide synthase (PKS) gene clusters in their genomes. In particular, they identified one strain belonging to the
Fischerella genus (
Fischerella sp. PCC 9339) that exhibits 22 NRPS/PKS clusters in its genome. On the contrary, only five compound families have been isolated from the genus
Fischerella so far and are listed on the present database. Moreover, it is interesting to note that, among the 126 strains analyzed by Shih et al. (2013) [
49], only 14 were formally reported to produce characterized metabolites.
In addition, the best producer genus,
Lyngbya-Moorea, remains rarely studied at the genomic level: four genomes are available in the Genbank database and another three are available on the Microscope platform [
50]. Considering the number of compounds isolated from the
Lyngbya-Moorea genus (85 compound families), most of the links between the identified molecules and the responsible biosynthetic gene clusters remain to be characterized. We also compared our collected data with those reported by Dittman et al. (2015) [
15] in order to determine when the isolated molecule families are linked with a specific and identified biosynthetic gene cluster. This review showed that less than 20% of the molecule families from the database are associated with specific identified biosynthetic gene clusters. Thus, the biosynthetic pathways of a large majority of compounds is still unknown as well as the regulation mechanisms controlling their production. Therefore, these observations highlight part of the remaining possibilities for the discovery of new molecules, gene production, and biosynthetic pathways.
4. Chemical Diversity and Bioactivity of Natural Products from Cyanobacteria
Each of the 260 families of compounds was classified by chemical classes and bioactivity (
Figure 3 and
Figure 4). The 260 families of compounds were classified by their chemical classes, and 10 different classes were listed: alkaloids, depsipeptides, lipopeptides, macrolides/lactones, peptides, terpenes, polysaccharides, lipids, polyketides, and others (
Figure 3). Of the 260 metabolite families, 66 belong to the peptide class. Together with the depsipeptide and lipopeptide classes, they represent 133 families of compounds (51%) derived from peptides. This is not surprising, regarding the diversity of biosynthetic pathways described in cyanobacteria: NRPS (non-ribosomal peptide synthase), PKS (polyketide synthase) and RiPPs (ribosomally synthesized and post-translationally modified peptides) with the ability to produce a wide range of metabolites and notable peptides [
15] (
Figure 3).
Fourteen major activities have been listed from the literature (lethality, neurotoxicity, hepatotoxicity, dermaltoxicity and cytotoxicity, anti-inflammatory, antioxidant, antiviral, anti-microalgal, antibacterial, antifungal, and antiprotozoal activities as well as protease and enzyme inhibition activities). Cytotoxic activity against various cell lines is the most frequently detected type of bioactivity with up to 110 families of the 260 listed. On the other hand, lethality and the antibacterial activities have been detected for 54 and 43 compound families, respectively (
Figure 4).
The number of compounds displaying each tested activity is shown in
Figure 5. The activities of molecules have been tested against different targets ranging from a specific cellular mechanism to an entire organism. For example, the inhibitory activity of proteases and other enzymes was shown to target enzymatic processes when the lethality and antimicrobial activity were tested against whole organisms. The lethality tests were generally realized against small invertebrates such as the brine shrimp crustacean
Artemia salina, the gastropod mollusk
Biomphalaria glabrata, and the crustacean
Thamnocephalus platyurus. The present analysis confirms preceding observations (i.e., that cytotoxicity is the most commonly detected activity, followed by lethality and antibacterial activity). Some activities were detected only for a restricted number of compounds: dermaltoxicity concerned only two families of metabolites (aplysiatoxins and lyngbyatoxins) [
51,
52], hepatotoxicity was observed for three families (cylindrospermopsins, microcystins, and nodularins) [
53,
54,
55], antioxidant and anti-inflammatory activities were observed for four (carotenoids, chlorophylls, mycosporine-like amino acids, and phycocyanins) [
56,
57,
58,
59], and seven metabolite families (coibacins, honaucins, aeruginosins, malyngamides, phycocyanin, scytonemin, and tolypodiol) [
60,
61,
62,
63,
64,
65,
66], respectively. Nevertheless, there are only a few examples of these activities being tested by authors in comparison with cytotoxicity and lethality, which have been investigated far more regularly. In terms of anti-inflammatory activity, all seven tested molecules cited above were positive for this type of activity, and 53% of the studied molecule families have been tested for cytotoxic activity, while only 2.7% have been tested for anti-inflammatory activity. In parallel, some of these metabolite families can exhibit more than one activity. In fact, a total of 362 activities have been detected for the whole of the 260 metabolite families.
Focusing on the chemical classes, it appears that there is no specific indication that one chemical class exhibits specific activities with regard to other classes. The results from the review showed that the polysaccharide class presents only two tested activities (enzyme inhibition and antiviral activity), but only three types of polysaccharides isolated from cyanobacteria have been observed so far (calcium spirulan, cyclodextrins, and iminotetrasaccharide) [
67,
68,
69]. Five chemical classes (the alkaloids, the depsipeptides, the lipopeptides, the macrolides, and the peptides) seem to present a remarkably large set of activities. When comparing the number of detected activities with the number of molecules belonging to each chemical class, the most bioactive molecules were shown to be the alkaloids, the lipopeptides, and the polyketides, which exhibit respectively 2.2, 1.9, and 1.8 activities per molecule on average.
These observations highlight a bias in the bioactivities searched from the isolated molecules. First, reported activities were those that researchers decided to test. Thus, the metabolite bioactivity profile could be underestimated because of the number of tests realized and remains the main limitation for the description of the potential applications of the bioactive molecules. In addition, there is still no consensus concerning the dose and dilution threshold that should be considered for each individual bioactivity test. In some cases, the concentration difference, used to determine if two distinct molecules are active, is important. For example, odoamide [
70], which is a cyclic depsipeptide member of the aurilides family, and scytoscalarol [
71], a sesterterpene, have both been described as being “cytotoxic.” However, their respective IC
50 values appear to be very different: 26.3 nM against HeLa S3 human cervical cancer cells for odoamide and 135 µM against Vero cells for scytoscalarol, which represents a concentration difference of 500 times between their respective inhibition potentials. Furthermore, tests can be realized against several cell lines and strains presenting different sensitivity responses, which limit the comparison between results.
With 10 chemical classes and 14 types of bioactivity, the cyanobacterial metabolites are diverse and highly active. However, half of the families of metabolites listed in the database are peptides or peptide derivatives. This could be due to the importance of the peptide biosynthetic pathway (NRPS, PKS, and RiPPs) or the extraction methods used, which might favor peptide extraction. We did not observe a link between chemical classes and activities, but this observation must be considered carefully with regard to the low number of investigated molecules in some classes (i.e., polyketides, polysaccharides, and terpenes). The most frequently detected activity for cyanobacterial metabolites is cytotoxicity (42% of the metabolite families), whereas antioxidant or anti-inflammatory activities were detected for only 1.5% and 2.7% of the families. This imbalance is due to the frequency at which tests were carried out. In fact, cytotoxicity was tested for 53% of the molecules, while anti-inflammatory activity was only tested in 2.7%. This observation may reflect the research inclination to find new pharmaceutical compounds, notably cytotoxic compounds that are usable in cancer therapy, and suggests the potential for the discovery of new activities for application in other fields.
6. Conclusions
In this review, all available information concerning the beneficial activities of natural products of cyanobacteria was gathered. To write this review, a molecular database of the various families of metabolites isolated from cyanobacteria was constructed from the systematic analysis of 670 articles. The derived database represents 260 families of metabolites. It groups various types of information concerning the taxonomy of producing strains, the respective chemical classes, the origin strain habitats, and the tested/demonstrated activities for each member of the family, together with the related full references.
According to this review, from the above 300 different genera of cyanobacteria (referenced by the taxonomy published by Komarek et al. in 2014) [
27], 90 have, so far, been reported to produce bioactive metabolites. Some of them have been shown to produce a high number of compounds, such as those from the genus
Lyngbya-Moorea, which includes 85 families of metabolites isolated so far. However, the
Lyngbya genus is a polyphyletic group and its taxonomy position is under revision. This number might be re-evaluated and distributed within distinctive new genera. The genomes of the producing strains are not available in the majority of cases, whereas Shih et al. (2013) demonstrated the large genomic potential of numerous cyanobacteria thanks to the biosynthetic pathways of metabolites highlighted by genome mining analyses [
49]. Therefore, the potential for the discovery of new natural molecules and new biosynthetic pathways from cyanobacteria still remains very important and needs to be systematically explored.
Cyanobacterial metabolites belong to 10 chemical classes (including peptides, alkaloids, terpenes, and lipids), where most of the families of metabolites are peptide derivatives (above 50% of the families). Fourteen different types of activities can be distinguished for cyanobacterial metabolites (e.g., antimicrobial, lethality, cytotoxicity, and antioxidant). The large majority of the components are cytotoxic (110 families), whereas some activities have only been tested rarely, and their occurrence appears to be weakly demonstrated. Globally, no clear correlation has been observed between chemical classes and the specificity of the respective types of bioactivity. Further studies are needed in order to precisely understand the mechanisms of action of cyanobacterial metabolites, which potentially links bioactivity with structural features in order to support the new hypothesis on the biological function of the production of these components for organisms.
Lastly, 47 metabolites isolated from cyanobacteria that present remarkable interest for diverse fields of application were investigated further in the present literature review. For example, hassallidins, which show specific antifungal activity without antibacterial activity, and scytonemin, which has anti-inflammatory properties with no cytotoxicity, were detailed. These metabolites are potentially useful for the development of new concrete applications for cyanobacterial natural products and illustrate the interest in cyanobacteria as a prolific source of bioactive molecules.