In vivo and in silico screening for antimicrobial compounds from cyanobacteria

Abstract Due to the emerging rise of multi‐drug resistant bacteria, the discovery of novel antibiotics is of high scientific interest. Through their high chemodiversity of bioactive secondary metabolites, cyanobacteria have proven to be promising microorganisms for the discovery of antibacterial compounds. These aspects make appropriate antibacterial screening approaches for cyanobacteria crucial. Up to date, screenings are mostly carried out using a phenotypic methodology, consisting of cyanobacterial cultivation, extraction, and inhibitory assays. However, the parameters of these methods highly vary within the literature. Therefore, the common choices of parameters and inhibitory assays are summarized in this review. Nevertheless, less frequently used method variants are highlighted, which lead to hits from antimicrobial compounds. In addition to the considerations of phenotypic methods, this study provides an overview of developments in the genome‐based screening area, be it in vivo using PCR technique or in silico using the recent genome‐mining method. Though, up to date, these techniques are not applied as much as phenotypic screening.


| INTRODUCTION
The excessive use of antibiotics over the past decades has led to the rise of multi-drug resistant (MDR) bacteria, making it one of the substantial problems faced by the modern health care system. Due to increased resistance, effective treatment becomes more and more complicated with the available, common antibiotics. Therefore, new treatments have to be brought onto the market, discovering new antibacterial substances, a key factor in the fight against the widespread of MDR bacteria (Laxminarayan et al., 2013;With, 2015).
Even though the pharmaceutical industry has made great advances in synthetic chemistry regarding the development of new, bioactive substances against a wide variety of pathogens, this technology still has its limitations: many natural products have highly complex structures that are too complicated and too expensive to produce on an industrial scale. In addition, natural sources offer a high diversity of substances, from which only a small part has been discovered so far. Therefore, the screening and isolation of bioactive compounds as new therapeutic substances remains an important aspect of research (Ahmad & Aqil, 2020;Lahlou, 2013).
In terms of bioactive compounds, cyanobacteria are a promising source of new, undiscovered substances. Cyanobacteria are photoautotrophic microorganisms that occur in many different environments, such as freshwater, seawater, and fields, leading to a high These extracts are then used for antibacterial activity assays. The following chapter deals with common cultivations, extraction conditions, and antibacterial activity assays, but also gives a brief outlook on less prevalent methods. An overview of cyanobacterial extracts with antibacterial properties and their respective method of cultivation, extraction, and activity assay are given in Table 1.

| Enhanced production of antimicrobial compounds by varying cultivation parameters
Environmental samples can be screened directly by using them for extraction and a subsequent antimicrobial activity assay (Deyab et al., 2019). However, if an interesting compound is detected larger amounts of biomass are often required for the extraction and further characterization of the unknown substance. Therefore, the natural consortium can be cultivated in special bioreactors imitating the natural habitat, or the cyanobacteria have to be isolated. However, for further investigations, high biomass productivity and high production of antimicrobial compounds are required. The cultivation parameters of this step can differ greatly (see Table 1). Temperature is normally chosen between 20°C and 30°C and the light intensity in the reviewed literature ranges from 7 up to 100 μmol Photons/(m²s) (Belhaj et al., 2017;Lamprinou et al., 2015;Montalvão et al., 2016). In some instances, a constant light source, and in some instances a day/night cycle of different lengths were simulated (see Table 1). Cultivation is commonly conducted as photoautotrophic cultivation submerged in standard media such as BG-11 with or without nitrogen (Rippka et al., 1979) or Z8 (Kotai, 1972). In general, the cultivation conditions likely reflect default methods for the cultivation of cyanobacteria and no specific strategy designed to optimize the production of antimicrobial compounds. Exceptions are, for example, the cultivation of the terrestrial cyanobacterium Nostoc sp. (formerly Trichocoleus sociatus) in an aerosol-based photobioreactor, leading to a substantial increase of the antimicrobial activity in comparison to submerged cultivation . The exposure of cyanobacterial cultures to UV-B radiation leads to a decreased minimum inhibitory concentration (MIC) of the resulting crude extract (Fatima et al., 2017). One parameter of particular interest is the cultivation time until harvest for the antibacterial activity assay since the content of an antimicrobial compound can change over-cultivation (Chetsumon et al., 1993). For cyanobacterial cultures, comparatively long cultivation times are common. The cultivation duration varied between 4 and 200 days. The duration of 150-200 days described by Lamprinou et al. (2015) was stated to be necessary for the production of sufficient biomass. However, a very low light intensity of 7 μmol Photons/(m 2 s) was used, which likely led to a low growth F I G U R E 1 Schema of the commonly used procedure for the screening of antibacterial compounds from cyanobacteria, LLE, liquidliquid extraction; EPS, extracellular polymeric substances.
T A B L E 1 Overview of the antimicrobial activity of cyanobacterial extracts, as well as extraction parameters (fraction of the cultivation, solvent, and special properties of the extraction), antimicrobial activity assay, and cultivation parameters (culture temperature/media/duration/and light intensity/light-dark-rhythm)  . Nevertheless, the tolerable exposure intensity differs greatly between different cyanobacteria and needs to be taken into account (Lamprinou et al., 2015). Besides the light intensity and other cultivation parameters, the phase of harvesting the biomass varies within the literature. In many papers biomass from the exponential phase was used (Elshouny et al., 2017;Konstantinou et al., 2020;N. Padmini et al., 2020), which is reached after different cultivation durations, depending on the growth speed of the corresponding cyanobacteria. Hamouda Ali and Doumandji explicitly stated that biomass was harvested before reaching the exponential phase, namely after 5-6 days (Hamouda Ali & Doumandji, 2017). Figure 2 gives an overview of the different cultivation parameters that can influence the production of antimicrobial compounds.

| Extraction
One of the difficulties in extracting an unknown substance is choosing the most suitable extraction solvent without knowing the properties of the compound, such as polarity, and so on. A good solvent for the extraction of antimicrobial activity preferably has a relatively low boiling point, to simplify removal, and does not interfere with the subsequent activity assay, since residues of the solvent may remain in the dried extract. Throughout the literature, a large spectrum of polar and nonpolar solvents, as well as their mixtures are used for the extraction of antimicrobial substances, like methanol, acetone, ethyl acetate, ethanol, petroleum ether, chloroform, isopropanol, and water (see Table 1). Since the substances to compared water, isopropanol, and methanol for extraction and tested the activity of these extracts against Staphylococcus leopoliensis (Fatima et al., 2017). The MIC of the methanol extract was around 50% lower than that of the isopropanol or water extract.
Interestingly, the methanol extract worked against all tested bacteria strains (E. coli, S, aureus, K. pneumoniae, P. aeruginosa, and E. aerogenes), while the aquatic extract only inhibited the growth of E. coli, S. aureus, and E. aerogenes. Thus, it can be assumed that more than one active substance is produced in this case (Fatima et al., 2017 and also shows to be one of the most efficient solvents regarding the antimicrobial activity of the resulting extract. In general, polar solvents seem to be more suitable for the extraction of bioactive compounds (Barboza et al., 2017;Esquivel-Hernández et al., 2017).  (Lamprinou et al., 2015;Strieth et al., 2017).
The concept of using the supernatant for extraction is not well established in the screening of cyanobacteria, although it is already used more frequently in other areas (Moradi et al., 2019;Thomas Hoffmann et al., 2018). This extraction type is based on the assumption that an antimicrobial substance, which is produced as a defense mechanism, can also be secreted (Alkotaini et al., 2013; R. A. Mogea et al., 2015). In general, extraction using the supernatant can be done by liquid-liquid extraction or solid-phase extraction (

| Antimicrobial activity assay
A good activity assay is crucial for a successful in vivo screening for antimicrobial substances. Ideally, an assay is cheap, easy, has fast/ high-throughput, and has high sensitivity as well as reproducibility.
Furthermore, it needs to be ensured that no compounds of the extract are interfering with the assay itself (Hadacek & Greger, 2000).
The antimicrobial activity of an extract or substance can be determined using several different assays, with the most common being the agar diffusion and microdilution assay.
For the agar diffusion assay, a culture of a bacterial test strain (e.g., E. coli) is prepared and uniformly spread on an agar culture plate.
The extract is then applied to the plate with a disk (disk diffusion test) or wells are punched into the agar and filled with extract (well diffusion test) (Bonev et al., 2008). After incubation of the agar plates, they can be examined for an inhibition zone around the discs or wells, where an antimicrobial compound diffusing into the agar would inhibit bacterial growth. The antibacterial activity of the extract can then be described using the size of the inhibition zone, with a larger inhibition zone corresponding to a higher antibacterial activity (Bonev et al., 2008 which hinders the comparison of inhibition zones between different papers, is the high variance in the amount of used extract, as well as the varying extract concentration and concentration of the antimicrobial compound within the crude extract. As an alternative to the agar diffusion assay, inhibition can also be examined using well plate-based assays, in which the inhibition is usually anti-proportional to an increase in the optical density of a bacterial test strain. Alternatively, a well plate test can be conducted as a resazurin assay, in which resazurin is enzymatically reduced to resorufin by hydrogenases using NADH/NADPH as co-substrate and causing a shift of fluorescence wavelength (Präbst et al., 2017). The resazurin assay is proclaimed to have an advantageous sensitivity compared to optical density-based tests (Palomino et al., 2002). If the bioactive substance is applied in a variety of concentrations, the assay is called microdilution and the inhibition can be described by the MIC, describing the lowest concentration inhibiting visible bacterial growth. Sometimes the inhibition is additionally stated using the minimum bactericidal concentration (MBC), which describes the lowest concentration needed to kill a bacterium. To obtain the MBC, the respective bacteria are sub-cultured after performing an inhibition assay to obtain the capacity of reproduction (Owuama, 2017). Alternatively, the antibacterial activity can be described using an 'inhibition percentage', which is based on positive (commercial antibiotics) and negative controls (buffer or media). In comparison to an agar diffusion assay, a microdilution assay has the advantage of commonly describing the MIC, in which the concentration is directly implied, reducing variations between different working groups. In addition, a microdilution assay can be carried out in a well plate, allowing a significantly higher throughput than an agar method.
The conditions for the assay vary in a similar way to the agar diffusion assay with different incubation times (overnight up to 24 h) and incubation temperature (25°C-37°C). Furthermore, optical density can be measured at different wavelengths (Costa et al., 2015;Levert et al., 2018).
Even though there are a variety of assays available, most of the time agar diffusion or microdilution assay measuring the optical density is used, since these methods are already well established in most laboratories. Even though the inhibition zone assay has drawbacks like its expenditure of time, low accuracy, and detection limit, it is a simple, cheap, and robust method that can be carried out in practically every laboratory since little specific equipment is required (Osato, 2000).
No matter which test is chosen different parameters can influence the results: • The time point at which the antimicrobial substance is added.
• Time and temperature of diffusion of the antimicrobial substance.
• Inoculum concentration of test strains.
• Incubation time before measurement.
• Co-extracted compounds can disturb especially fluorescence or colorimetric assays.
• Amount of antimicrobial compounds.
• Purity of antimicrobial compounds.
Every bioactivity assay has advantages, disadvantages, and needs to be chosen based on the laboratory equipment. The biggest issue when comparing the achieved results with the literature is that most of the researchers use the method and parameters that are ing to a MIC of 2.5 mg/ml; one of 12 mm to a MIC of 0.16 mg/ml, and one of 15 mm to a MIC of 0.08 mg/ml (Belhaj et al., 2017). Although this approximation needs to be viewed with caution as the inhibition zone assay is also dependent on the diffusion rates of the compound, which are highly determined by the polarity of the substance (Ncube et al., 2008 (Dolman, 1943; FDA, 2020). Additionally, S. aureus and many bacteria from the genus Pseudomonas have known strains that are resistant to commonly used antibiotics (Köck et al., 2010;Pang et al., 2019). In response to that, some activity assays are testing the antibacterial activity of the extract against antibiotic-resistant strains like Vancomycin-resistant E. faecium (VRE) and Methicillin-resistant S. aureus (MRSA). Even against these, some extracts from cyanobacteria were able to achieve an inhibiting effect (Lamprinou et al., 2015). well as E. coli and P. aeruginosa (Fatima et al., 2017). In general, the type of bacteria used for antimicrobial assays may also depend on the location of the laboratory since the handling of pathogenic strains is controlled by national laws, dealing with the prevention and control of infectious diseases.

| GENOMIC APPROACHES FOR THE SCREENING
Due to the phenotypic nature of traditional screening methods, they rely on the synthesis of a sufficient amount of antibacterial components during cyanobacterial cultivation to be able to detect it in a subsequent inhibition assay. Since cyanobacteria grow rather slowly, this can lead to a long cultivation time before an activity assay is possible (Lamprinou et al., 2015;Niveshika et al., 2019;Pham et al., 2017). In addition, cultivation conditions have a high impact on the production of secondary metabolites. As a consequence, promising candidates for new antibiotics might be neglected due to unsuited cultivation conditions, leading to a decreased production of secondary metabolites. Therefore, the interest in genome-based screening as an addition to the phenotypic screening of cyanobacteria has increased in recent years (Micallef, D'Agostino, Al-Sinawi, et al., 2015;Micallef, D'Agostino, Sharma, et al., 2015;Singh et al., 2010). This interest was mainly promoted by the fact that the availability and accessibility of genome data have highly improved. In combination with the creation of new bioinformatics tools, this has generated many new options for screening (Corre & Challis, 2007;Levasseur & Pozzobon, 2020;Shiha et al., 2013). In general, genomic methods can be divided into molecular biological methods, using for example polymerase chain reaction (PCR) for the detection of DNA sequences in vivo, or genome mining approaches in which genomic data are analyzed in silico.

| Properties of antibacterial gene clusters
For the discovery of new bioactive substances based on genomic properties, significantly more information than for the execution of an antibacterial test is needed. It is, therefore, crucial to examine data about similar substances and their related biosynthesis from literature. There are several reviews about cyanobacteria dealing with the properties of already isolated and characterized substances and their corresponding bioactive activities (Agrawal et al., 2017;Tan & Phyo, 2020). Cyanobacteria are described to synthesize a range of antibacterial substances from different substance classes: alkaloids, depsipeptides, lipopeptides, macrolides/lactones, peptides, terpenes, polysaccharides, lipids, polyketides, and others (Swain et al., 2017). A majority of these bioactive substances are described to be peptide-

| Screening using genome mining and PCR
In general, most of the secondary metabolites are synthesized via bioactive gene clusters (BGC) (Naughton et al., 2017). These gene clusters often contain highly conserved sequences within a substance family, such as the adenylation modules of the NRPS or LanC, which is involved in the modification of lantibiotics (Mayer et al., 2001;Shiha et al., 2013). A conserved sequence refers to a nucleotide sequence with a very high homology across different species (Sarkar et al., 2011). The in silico screening for BGC is commonly called genome mining, which is described as the process of deriving information over an organism or its synthesized products through the analysis of genomic data and can be used for "predicting and isolating natural products based on genetic information without a structure at hand" (Ziemert et al., 2016). Genome mining can be done using a variety of different approaches. If the genome sequence of cyanobacteria is known (accession e.g., via NCBI (https://www.ncbi. nlm.nih.gov/), with up to date 500 complete genome sequences) it can be analyzed using web-based genome mining tools. One wellknown tool is the "Antibiotics and Secondary Metabolite Analysis SHell," commonly known as antiSMASH (Weber et al., 2015). This tool allows to identify gene clusters within a nucleotide sequence, as well as comparing them to known biosynthetic gene clusters (BGCs) to determine the gene cluster type as well as predict a possible STRIETH ET AL.  (Sandiford, 2017). In this way, cyanobacteria from a genome database can be screened regarding their possession of genomic sequences for the production of specific secondary metabolites. An If the genome of cyanobacteria is not sequenced, analysis can also be conducted in vivo by PCR. PCR is used to detect gene sequences within the genome through specific short nucleotide sequences called primers, which bind to complementary sequences and allow amplification of the DNA segment between forward and reverse primer by a DNA polymerase. There is also the possibility of designing a degenerated primer, which is a mixture of primers with highly similar sequences but substitution of different bases at some points of its sequence, making it possible to detect conserved regions of biosynthesis clusters in vivo (Sarkar et al., 2011). For example, this method was carried out by Ehrenreich et al., who examined isolated cyanobacteria for the presence of NRPS/PKS gene clusters to compare them with the cytotoxicity of the strains (Ehrenreich et al., 2005). Additionally, PCR products can be sequenced and used for further in silico analysis. This approach was used by Micallef et al. to close potential gaps in the nucleotide sequences (Micallef, D'Agostino, Sharma, et al., 2015) Even if these approaches offer many new possibilities, they should be seen as an addition to phenotypic tests and are not capable of replacing them completely. For example, PCR can be used to detect NRPS gene clusters, which can lead to the synthesis of an antibacterial peptide. However since around 70% of the cyanobacteria contain a corresponding gene cluster, this information alone does not guarantee an antibacterial activity (Neilan et al., 1999).
Hence, further investigations of antibacterial substances after the first molecular biological or genome mining approaches are crucial.
The approaches are commonly coupled with a subsequent activity assay or isolation and analysis of the compound using mass spectrometry (MS) and nuclear magnetic resonance (NMR) to determine its structure (Mohimani et al., 2014;Sigrist et al., 2020).
However, in silico methods have the advantage that the substance leading to a subsequent phenotypic hit is known, which greatly facilitates the purification. Partly, promising gene sequences are cloned into host bacteria like E. coli for a heterologous expression of the target molecule. The resulting extracts can then be screened using inhibition assays (Shi et al., 2019;Shih et al., 2013;Singh et al., 2010). However, it must be noted that nonphenotypic methodologies for the identification of bioactive substances in cyanobacteria are up to date a very small share compared to phenotypic screenings. Even today, genome mining in cyanobacteria is more of a promising outlook than a technique that is solidly

ACKNOWLEDGMENTS
This project is financially supported by DFG (STR1650/1-1) and the federal state of Rhineland-Palatinate (iProcess intelligent process development-from modeling to product). Open access funding enabled and organized by Projekt DEAL.

CONFLICTS OF INTEREST
The authors declare no conflicts of interest.

ETHICS STATEMENT
None required.

DATA AVAILABILITY STATEMENT
Not applicable.