Antibacterial Activity and Characterization of Bacteria Isolated from Diverse Types of Greek Honey against Nosocomial and Foodborne Pathogens

: It has been suggested that microorganisms present in honey are a potential source of antimicrobial compounds. This study aimed to isolate and characterize bacteria from 46 Greek honey samples of diverse botanical and geographical origin and to determine whether these bacteria demonstrate antibacterial activity against ﬁve important nosocomial and foodborne pathogens. In total, 2014 bacterial isolates were obtained and screened for antibacterial activity. Overall, 16% of the isolates inhibited the growth of Staphylococcus aureus , 11.2% inhibited the growth of Pseudomonas aeruginosa and Acinetobacter baumannii , 10.2% inhibited the growth of Salmonella Typhimurium and 12.4% of the isolates affected the growth of Citrobacter freundii . In total, 316 isolates that inhibited the growth of more than two of the tested pathogens were grouped by restriction fragment length polymorphisms (RFLP) analysis of the 16S rRNA gene amplicon. Fifty of them were identiﬁed by 16S rRNA gene sequencing. The majority, 62% of the isolates, belonged to the genus Bacillus . Only 10% of the isolates were identiﬁed as Gram-negative bacteria. Furthermore, in several bacterial isolates, genes encoding polyketide synthases and nonribosomal peptide synthetases that catalyze the biosynthesis of secondary metabolites which might contribute to the exerted antimicrobial activity, were detected. This study demonstrates that honey microbiota exerts antimicrobial activity and is a putative source of secondary metabolites against important nosocomial and food pathogens that warrants further investigation.

For centuries honey has been utilized not only as a food, but also as a medicine and food preservative [12,13]. Recent studies confirmed its beneficial bioactivity on human A total of forty-six Greek honey samples from various geographical and botanical origins were provided by individual beekeepers and beekeeping associations. Each sample was assigned to a unique reference number. Details regarding botanical source, geographical location, and harvest date were recorded (Table S1). The identification of the botanical source of honeys was based on the flora availability during the harvest season, the location of the apiary and, in several cases, palynological analysis. Honey samples were stored at room temperature in dark conditions.
Aerobic mesophilic bacteria: 100 µL of each suspension were plated in triplicate onto standard plate count agar (PCA) (Condalab, Madrid, Spain) [7,8] or Luria Bertani (LB) agar (Lab M, Heywood, UK) [12] and then incubated aerobically at 30 • C for 5 days. After incubation, in order to obtain pure bacterial cultures, single colonies were transferred onto new PCA Petri dishes using the successive subculturing method. Pure cultures were stored as glycerol stocks (MH broth +20% glycerol) in 96-well plates at −80 • C, until further analysis.
Aerobic spore-forming bacteria: each diluted sample was subjected to heat activation at 80 • C for 10 min and cooled at room temperature for another 15 min [8,18]. Then, 100 µL of each sample were spread in triplicate onto the selective Bacillus cereus medium (BCM) (Lab M, Heywood, UK) and incubated under aerobic conditions at 30 • C for 48 h. Pure cultures, obtained as described above, were stored as glycerol stocks (MH broth +20% glycerol) in 96-well plates at −80 • C, until further analysis.
Lactic acid bacteria (LAB): A 100 µL aliquot of each diluted sample was spread in triplicate onto the selected media, Man Rogosa and Sharpe (MRS) agar (NEOGEN, Heywood, UK) and incubated anaerobically in an AnaeroJar AG25 with the AnaeroGen Atmosphere Generation system (Oxoid, Basingstoke, UK) at 30 • C for approximately 4 days. Resazurin sodium salt (Alfa Aesar, Thermo FisherGmbH, Kandel, Germany) (30 mg/lt) was used as an indicator for anaerobic conditions (turns from blue to pink). Each colony was subcultured anaerobically onto MRS agar supplemented with 1% calcium carbonate (CaCO 3 ) (Scharlau, Barcelona, Spain) [32]. A clear zone around a colony indicates hydrolysis of CaCO 3 due to organic acid production [34]. Colonies that exhibited a clear zone were picked, cultured onto eppendorfs filled to the top with Thioglycolate Broth (Scharlau, Barcelona, Spain) +20% glycerol and stored at −80 • C, until further analysis.

Dual Culture Overlay Assay
The antibacterial activity was tested by dual culture overlay assay as previously described [28,36] with few modifications. A replicate of each 96-well plate was transferred onto PCA square Petri dishes (120 × 120 mm, Aptaca Spa, Canelli, Italy) by using a microplate replicator (Boekel Scientific, Pennsylvania, PA, USA) and incubated at 30 • C for 24 h. Then, each pathogen (approximately 9 × 10 4 cfu/mL) was mixed into soft Nutrient Agar (Biolab, Budapest, Hungary) containing 0.75% (w/v) agar, keeping it liquid at 42 • C. The mixture of soft agar and bacterial cells was poured as a thin overlayer on top of the plates with the overnight-cultivated replica of the bacterial isolates. The plates were incubated at 37 • C for 24 h and inhibition zones were observed. Each bacterial isolate that exhibited a clear inhibition zone was tested again, in triplicate, in order to confirm the antibacterial activity.

Grouping and Identification of Bacteria Exerting Antibacterial Activity
From the glycerol stock each isolate was plated on PCA agar and incubated aerobically at 30 • C for 24 h. A single colony was obtained and cultured in MH broth aerobically at 30 • C, overnight. Pure broth cultures were used for genomic DNA extraction with the ExtractMe Genomic DNA Kit (Blirt, Gdańsk, Poland). Universal primers 27F (5 -AGAGTTTGATCMTGGCTCAG-3 ) [37] and 1492R (5 -GGTTACCTTGTTACGACTT-3 ) [38] (Eurofins Genomics, Germany) were used to amplify the 16S rRNA gene by PCR. The reaction mixture contained: 1 U FastGene Taq DNA Polymerase (NIPPON Genetics, Tokyo, Japan), 1× PCR buffer A, 25 pmol of each primer, 1 mM dNTPs, 3 µL DNA template, and deionized sterile water to a final volume of 50 µL. The thermal cycler Primus 25 (PEQLAB Biotechnologie, Erlangen, Germany) was used in the following PCR conditions: initialization at 95 • C for 3 min, followed by 35 cycles of denaturation at 95 • C for 30 s, annealing at 50 • C for 30 s, and elongation at 72 • C for 2 min. A final elongation step at 72 • C for 2 min was added.
Restriction fragment length polymorphisms analysis (RFLP) of 16S rRNA amplicons was conducted as a simple prescreening for dereplication and grouping of the bacterial isolates. Two µL of each PCR product (approximately 1400 bp), were incubated at 37 • C for 1 h, with 10 U of RsaI and HinfI endonucleases and 10 × buffer K (Bioron, Römerberg, Germany) [5,18,25]. The analysis was performed by electrophoresis in a 3% agarose gel (Invitrogen, Paisley, UK) by standard electrophoresis. Following RFLP analysis, bacterial isolates from each group were phenotypically characterized by Gram staining [39].
Amplicons of representative isolates of each group demonstrating antibacterial activity were purified using the NucleoSpin Gel and PCR clean-up kit (Macherey-Nagel, Düren, Germany) and then directly sequenced via Sanger dideoxy termination method by Macrogen Europe (Amsterdam, The Netherlands). Chromas version 2.6.6 software (Technelysium Pty Ltd., South Brisbane, Australia, www.technelysium.com.au) (accessed on 21 June 2021) was used to check the quality of the obtained sequencing results. Sequences were assembled into a single sequence via MEGA X version 10.1.6 software [40] and Gene Runner version 6.5 software (www.generunner.net accessed on 21 June 2021) and subjected to a BlastN (Megablast) (https://blast.ncbi.nlm.nih.gov/Blast.cgi accessed on 21 June 2021, BLAST, NIH) search in the 16S rRNA Database-GENEBANK to identify sequences with the highest similarity [3,5,18]. Obtained sequences were deposited to GenBank under the following accession numbers: MW700012-MW700061.

NRPS and PKS Gene Screening
Identified bacterial isolates were further screened for the presence of biosynthetic gene clusters of nonribosomal peptide synthetases (NRPS) and type-I polyketide synthases (PKS-I). The oligonucleotide primer set NRPS-1F (5 -GCSTACSYSATSTACACSTCSGG-3 ) and NRPS-1R (5 -SASGTCVCCSGTSCGGTAS-3 ) (Eurofins Genomics, Ebersberg, Germany) was used to amplify partially the adenylation (A) domain (approximately 750-800 bp) and the primer set PKS-IF (5 -GTGCCGGTNCCRTGNGYYTC-3 ) and PKS-IR (5 -CGATGGAYCCNCARCARYG-3 ) (Eurofins Genomics, Ebersberg, Germany) for the amplification of partially KS domain (approximately 600-750 bp) [32]. PCR reactions were carried out in 50 µL final volume, using Primus 25 thermal cycler (PEQLAB Biotechnologie, Erlangen, Germany). The reaction mixture for NRPS A domain contained 1.5 U FastGene Taq DNA Polymerase (NIPPON Genetics, Tokyo, Japan), 1× PCR buffer B, 25 pmol of each primer, 1 mM dNTPs and 5 µL DNA template. The amplification conditions were the following: initialization at 95 • C for 3 min, followed by 35 cycles of denaturation at 95 • C for 30 s, annealing at 59 • C for 2 min and elongation at 72 • C for 4 min with a final elongation step at 72 • C for 1 min. For the KS domain the reaction mixture contained 1 U Taq DNA Polymerase (Invitrogene, UK), 1× PCR buffer, 25 pmol of each primer, 1 mM dNTPs, 1.5 mM MgCl 2 , and 5 µL DNA template. The initialization at 94 • C for 3 min was followed by 40 cycles of denaturation at 94 • C for 30 s, annealing at 60 • C for 1 min and elongation at 72 • C for 30 s. The final elongation step was performed at 72 • C for 5 min. Sequencing was performed via Sanger dideoxy termination method by CeMIA SA (Larissa, Greece). Sequencing data were subjected to a BLASTN (https://blast.ncbi.nlm.nih.gov/Blast.cgi, BLAST, NIH) (accessed on 21 June 2021) search and to BLASTX (E value: 1 × 10 −5 ) in order to identify the closest known homologues and determine the level of homology of the amino acid sequence.

Isolation of Bacteria from Honey
In total, 2014 bacterial isolates were obtained from 46 Greek honey samples of various geographical and botanical origin, utilizing four different culture media (PCA, LB agar, BCM and MRS agar) (Table S1). Overall, 695 bacterial isolates were obtained on PCA, 851 on LB agar, 370 on BCM, and finally 98 isolates were obtained by using MRS agar. Bacterial isolates were obtained from all honey samples no matter the botanical source, geographical origin, and the date of harvest. The only exception was honey sample 42, which was tested only for aerobic spore-forming bacteria on BCM.

Antimicrobial Activity of Bacterial Isolates
All bacterial isolates (2014 in total) were screened by dual overlay assay, in order to determine whether they inhibited the growth of five important human and foodborne pathogens: S. aureus, P. aeruginosa, S. Typhimurium, A. baumannii, and C. freundii (Table 1). A clear inhibition zone (>5 mm) against a given tested pathogen was considered as positive antibacterial activity. Overall, 323 (16%) isolates were found to inhibit the growth of S. aureus, 225 (11.2%) isolates inhibited the growth of P. aeruginosa, 206 (10.2%) isolates inhibited the growth of S. Typhimurium, while 225 (11.2%) isolates and 249 (12.4%) isolates affected the growth of A. baumannii, and C. freundii, respectively. Collectively, 296 (14.7%) isolates inhibited the growth of at least one of the tested pathogens, 167 (8.3%) isolates inhibited the growth of at least two of the five tested pathogens and 111 (5.5%) honey bacterial isolates affected the growth of at least three of the five tested pathogens. Forty (1.99%) isolates exerted antibacterial activity against at least four out of five tested pathogens, whereas 21 isolates (1.04%) inhibited the growth of all five pathogens. These results demonstrated that 31.5% (635/2014) of the obtained bacterial isolates exerted antibacterial activity against at least one pathogen.

Grouping and Identification of Bacterial Isolates Exerting Antibacterial Activity
The bacterial isolates (316) that exerted antibacterial activity against at least two of the tested pathogens were selected for further analysis. The 16S rRNA gene of all these isolates was amplified by PCR and the amplicons were subjected to RFLP analysis. Overall, the 316 bacterial isolates were distributed in 19 groups (Table S2). Representative isolates from each group were further subjected to Gram staining. In total, 50 bacterial isolates were identified by 16S rRNA gene sequencing.
Sequencing of the 16S rRNA gene revealed that 90% of the 50 bacterial isolates were classified as Gram-positive bacteria whereas only 10% as Gram-negative. Thirty-one out of 50 strains (62%) belonging to the genus Bacillus were classified into eight species (Table 2). B. safensis was the most frequently identified species (7 out of 31 isolates, 22.6%) followed by B. pumilus (5/31, 16.1%), B. subtilis (4/31, 12.9%), and B. cereus (2/31, 6.5%). Four isolates were identified as B. licheniformis, Priestia megaterium (former B. megaterium), B. paramycoides, and B. vallismortis respectively. Classification at species level within the Bacillus group was not possible for 9 isolates due to the high similarity of 16S rRNA gene sequence to more than one species. Other isolates were identified as Lysinibacillus fusiformis, Microbacterium imperiale, Micrococcus yunnanensis, Paenibacillus sp., Paenibacillus profundus, Terribacillus saccharophilus, and Terribacillus sp. (2% for each phylotype). Seven isolates were identified as Staphylococcus spp. (14%). Specifically, five isolates were identified as S. arlettae, S. epidermidis, S. hominis, S. pasteuri, and S. warneri respectively, and two isolates were identified as S. cohnii. Regarding the Gram-negative isolates, four (8%) were identified as Pseudomonas sp., P. coleopterorum, P. fulva, and P. stutzeri and one as Acinetobacter lwoffii ( Table 2). -   Furthermore, all 50 isolates were screened for the presence of NRPS and PKS-I biosynthetic gene clusters. In fourteen isolates, NRPS genes were detected; six isolates were tested positive for PKS gene clusters whereas in 13 isolates both NRPS and PKS gene clusters were detected. Twenty-six out of 31 (83.9%) bacterial isolates identified as Bacillus spp. Twelve NRPS amplicons and 11 PKS amplicons were sequenced, in order to validate the PCR result (Tables 3 and 4). Ten out of 12 amplicons were indeed identified as NRPS gene clusters. The identification of the other two amplicons was not possible due to low quality of sequencing data. In most cases there was an agreement regarding the genus classification and sequencing results, except for amplicons A1 and A7 that were amplified from bacterial isolates belonging to Bacillus spp., whereas the NPRS amplicons shared homology with NRPSs of P. moteilii. Amplicons A10, A20, and A46 from the strains B. safensis (CTA10), B. safensis (CTA20), and Bacillus sp. (CTA46) were identified as NRPSs that were implicated in surfactin biosynthesis, while B69 from the strain Paenibacillus sp. shared homology with the subunit B of gramicidin synthase.
Furthermore, the sequencing data verified the PCR results for the 11 tested PKS amplicons. Six amplicons exhibited homology with ketosynthases, (A5, A9, A10, A20, A125, B11), 3 were identified as PKSs (A109, B31, B69), while amplicon A16 was identified as amino acid adenylation domain-containing protein. In all cases there was an agreement regarding the classification at genus level and the sequencing results. Interestingly, sequencing data regarding A9 amplicon of B. subtilis CTA9 strain suggest that bacillaene is biosynthesized by this strain.

Discussion
The antimicrobial activity exerted by honey is one of its most well-established bioactivities described in many studies [11]. Honey affects the growth and survival of microorganisms via multiple mechanisms [14,41]. Low pH, high sugar concentration, high osmolarity and the presence of antibacterial substances such as hydrogen peroxide and phytochemicals in honey provide a hostile environment for bacteria [42]. Consequently, honey is expected to contain a limited number and diversity of microorganisms [14]. Nevertheless, honey is not sterile since it is a natural product that is not processed [3,5,12,28]. Several studies have reported isolated microorganisms from honey samples produced in various geographical locations [4,7,12,18,43]. Interestingly, the number of microorganisms in honey ranged widely, from 0 to several thousand colony forming units (CFUs) per gram, depending on the sample and its freshness [8,9,18]. Sinacori et al. [18] reported low bacterial load in 33 out of 38 honey samples harvested in southern Italy. Similar results were reported by other researchers in Argentina [8], Morocco [15], Poland [12], Saudi Arabia [44], and Mexico [7]. Fernández et al. [27] studied the microbiological quality of honey from Argentina and their results were comparable to those reported in previous studies conducted in Argentina as well as other parts of the world [8,12,18,28,45]. Bacteria have also been isolated from stingless bee honey [4,23,42]. Moreover, there are few studies that have described the microbial communities of honey using culture-independent methods like next generation sequencing (NGS) [17,[46][47][48].
Honey bacteria exerting antibacterial activity have been successfully isolated by several research groups [6,12,42,[49][50][51][52]. There are also reports on the antibacterial activity exerted by various bacterial isolates from stingless bee (Heterotrigona itama) honey [34,42]. Lee et al. [28] screened more than two thousand bacterial strains isolated from six US domestic honeys and two manuka honeys against seven bacteria and one mold but identification of bacterial isolates was not conducted. Recently, Pajor et al. [12] investigated the antimicrobial potential of 163 bacteria isolated from honey samples produced in Poland.
A number of studies have demonstrated high antibacterial activity of Greek honeys against major nosocomial and foodborne pathogens [1,13,35,53]. However, this is the first study aiming to isolate and characterize bacteria from diverse types of Greek honeys exhibiting antibacterial potential. In total, 2014 bacterial isolates were screened against five important human and foodborne pathogens (S. aureus, P. aeruginosa, S. Typhimurium, A. baumannii, and C. freundii). Three of them belong to the "ESKAPE" pathogens group (E. faecium, S. aureus, K. pneumoniae, A. baumannii, P. aeruginosa, and Enterobacter spp.). The World Health Organization (WHO) included ESKAPE in a list of pathogens for which new antimicrobial compounds are urgently needed. A. baumannii and P. aeruginosa are listed in the critical priority list of pathogens, whereas S. aureus (MRSA) is in the high priority group [54][55][56]. Furthermore, multidrug-resistant C. freundii is of major concern, often implicated in nosocomial infections [57,58]. This is the first study regarding the antibacterial activity of honey bacterial isolates against C. freundii and A. baumanii.
Moreover, in this study a higher number of bacterial isolates (2014) was screened against five pathogens, compared to relevant studies [4,42,51,52]. The most susceptible pathogen was S. aureus, followed by C. freundii, P. aeruginosa, and A. baumannii. In contrast, S. Typhimurium was the least affected tested pathogen. A possible explanation might be that Gram-negative bacteria possess an outer membrane which often contributes to resistance to a wide range of antibacterial factors (for instance, loss of porins located in outer membrane). In agreement with our study, Pajor et al. [12] reported that 20.3% of the honey isolated strains inhibited the growth of S. aureus ATCC 25923, while activity against P. aeruginosa ATCC 27857 was rarely observed.
In previous studies, aerobic spore-forming bacteria belonging to Bacillus spp. and related genera (Brevibacillus, Lysinibacillus, Paenibacillus) have been identified as the most common bacteria present in honey [5,59]. Furthermore, bacteria belonging to Bacillus genus have been often isolated from the honeybee gut [60][61][62]. Similarly, in this study, 62% of the identified isolates belonged to the genus Bacillus. Malika et al. [15] detected Bacillus spores in most of the honey samples from Morocco, while Pomastowski et al. [3] reported that regardless of the geographical and botanical origin, the main bacteria in honeys were spore-forming Bacillus spp., which is in agreement with our data.
Nevertheless, to the best of our knowledge, this is the first time that B. paramycoides, L. fusiformis, M. imperiale, M. yunnanensis, P. profundus, and T. saccharophilus were isolated from honey. Paenibacillus strains isolated from various ecological niches synthesize a wide range of antibiotics (polymyxins, polypeptins, gavaserin) [51,64]. P. profundus strain Sl 79, a deep sediment bacterium, exerted a remarkable inhibitory activity against both Grampositive and Gram-negative bacteria. It is known to produce isocoumarin and a new linear glyceryl acid-derived heptapeptide that exhibited antibacterial activity against S. aureus, S. epidermidis, B. subtilis, and E. faecium [64,65]. On the other hand, M. yunnanensis was previously detected in bee gut [61,66] but not in honey. Micrococcin was the first thiazolyl peptide isolated from a Micrococcus strain [67]. Sponge-derived M. yunnanensis F-256,44 was reported to produce kocurin, an antibiotic of the thiazolyl peptide family [67,68]. Moreover, Ravi Ranjan et al. [69] isolated M. yunnanensis strain rsk5, which produced a novel antibacterial compound. This compound exerted activity against antibiotic-resistant S. aureus and differs from other characterized secondary metabolites. T. saccharophilus, as its name suggests, is a sugar-loving bacterium [70]. A study reported that a T. saccharophillus strain isolated from the gut of an edible Indian freshwater fish could be applied as a probiotic in fish feed formulations [71]. Another study reported the production of copious amounts of exopolysaccharides by T. saccharophilus strain PS-47. Exopolysaccharides are secondary metabolites, used in the pharmaceutical and food industries [72]. Microbacterium spp. were previously isolated from the gastrointestinal tract of adult worker honeybees [62]. A study reported that M. imperiale (MAIIF2a) reduced fungal symptoms by Fusarium solani in cassava roots, indicating its antimicrobial potential [73].
The antibacterial activity of B. paramycoides and L. fusiformis against important human and food pathogens has been described for the first time in this study. However, these bacterial species are known for their bioremediation potential [74][75][76]. Moreover, B. paramycoides was reported as a plant growth-promoting rhizobacterium [77], while L. fusiformis strain C250R produced a novel extracellular thermostable protease designated as SAPLF, which could be applied in pharmaceuticals, food, and biotechnology industries [78].
In this study, 14% of the characterized bacterial isolates were identified as Staphylococcus spp. Previous studies have reported the presence of S. epidermidis, S. pasteuri, and S. hominis in honey [3,22,79]. Moreover, all Staphylococcus spp. identified in this study, except S. arlettae, have been described as part of the honeybee gastrointestinal microbiome [3,61,62,66,80,81]. S. arlettae is commonly found in the skin and nares of poultry and goats [82].
Surprisingly, none of the bacterial isolates grown on MRS agar that exerted antimicrobial acitivity in this study was identified as lactic acid bacteria. However, all four isolates were identified as Staphylococcus spp., one of them (S. hominis) previously isolated using MRS agar [83].
Only 10% of the characterized bacterial isolates were identified as Gram-negative bacteria, namely Pseudomonas sp., P. coleopterorum, P. fulva, P. stutzeri, and A. lwoffii. Pseudomonas spp. and Actinetobacter spp. are often present in soil [7]. However, Veress et al. [84] isolated for the first time the bacterium A. lwoffii from a Transylvanian honey sample, suggesting that it was derived from the honeybee microbiota. Pseudomonas spp. have been identified both in honey and the digestive tract of honeybees [9,10,17,22,62]. The Pseudomonas genus is very diverse and comprises 272 species, organized in several major taxonomic groups that are widely distributed in diverse niches [85]. Some of these strains exert biotechnologically interesting metabolic potential (denitrification, degradation of aromatic compounds, nitrogen fixation) [86][87][88][89]. P. fulva is a widespread environmental species exhibiting high biodegrading potential. It is reported to degrade neonicotinoid insecticides that contribute to honeybee colony collapse disorder [86], as well as a pyrethroid insecticide, reported to be accumulated in honeybee wax samples [87]. On the other hand, P. coleopterorum was detected in the gut microbiota of some plant-related insects [90].
Bee products have been suggested as a rich source of beneficial bacteria that might be potential biocontrol agents against bee pathogens [4,28,42]. Several Bacillus strains produce compounds against pathogenic and food-spoilage bacteria [2,51]. Moreover, B. subtilis, B. amyloliquefaciens, B. pumilus, and B. licheniformis are considered as probiotics and antibiotic producers. Indeed, B. subtilis has already been used as a probiotic supplement for human and animal diets [23,42].
LABs commonly isolated from honey produce secondary metabolites, like bacteriocins and lipopeptides [6,42,91,92]. However, we did not identify any LABs exhibiting antibacterial activity in the present study, presumably due to the relatively small number of isolates grown on MRS agar.
Magdalena et al. [51] have targeted seven antibacterial peptide biosynthetic genes present in two B. velezensis honey isolates using specific primers. However, in our study the universal primer set NRPS-1F-NRPS-1R for the amplification of adenylation (A) domain was used, thus targeting a much broader range of NRPS biosynthetic clusters.
PKS gene clusters were detected in B. safensis, Bacillus sp., and S. cohnii strains. Type I PKSs are large multifunctional enzymes, implicated in pharmacologically important metabolites, most often detected within prokaryotes [93,98]. PKs were detected in B. subtilis, B. amyloliquefaciens, B. methylotrophicus, B. atrophaeus, B. laterosporus, and Paenibacillus sp. [93]. Within the B. subtilis group, three types of antimicrobial PKs are produced, including bacillaene, difficidin, and macrolactin [98]. Moreover, Zhang et al. [99], reported the partial isolation of a gene cluster from Staphylococcus lentus encoding a PK exerting broad-spectrum activity against fungi and bacteria. To the best of our knowledge, this is the first time that PKS encoding genes have been detected in bacteria isolated from honey exerting antimicrobial potential.
Twelve NRPS amplicons and 11 PKS amplicons were sequenced, in order to validate our detection approach. Three out of the 12 NRPS amplicons were implicated in the synthesis of surfactin. Surfactin is a cyclic lipoheptapeptide produced by various strains of Bacillus subtilis, which alters membrane integrity and permeability acting against several microorganisms [51,92,101].
Furthermore, it has been reported that Paenibacillus strains, isolated from various environments, produce a wide range of antibiotics [64,92]. Three Paenibacillus species were detected in honey, P. alvei, P. polymyxa, and P. larvae [3,28,92,102]. Polymyxins produced by P. polymyxa have been used as a last-resort treatment of multidrug-resistant Gram-negative bacteria infections. They also exhibit strong antifungal activity [93].
Finally, our sequencing results regarding the PKS amplicon from Bacillus subtilis CTA9 strain suggest the presence of a PKS implicated in bacillaene biosynthesis. Bacillaene was firstly isolated from B. subtilis strains, but since then has been reported to be produced by several other Bacillus spp. It is a hybrid PK synthesized by a PKS-I and NRPS operon and it has demonstrated a strong activity against various bacteria and fungi [92,93,98]. Taking together all these data, it is evident that most of the bacterial isolates in this study could be potent producers of secondary metabolites including antibiotics, bacteriocins, and enzymes that might lead to future biotechnological applications.

Conclusions
In the present study, 2014 bacterial isolates were obtained from 46 Greek honey samples of diverse botanical and geographical origin. All bacterial isolates were tested against five important nosocomial and foodborne pathogens. Overall, 31.5% of the isolates exerted antibacterial activity against at least one pathogen. Gram-positive S. aureus was the most susceptible of the five tested pathogens, followed by C. freundii, P. aeruginosa, and A. baumanii. S. Typhimurium was found to be the least susceptible pathogenic bacterium. Fifty isolates that inhibited the growth of at least two tested pathogens were further characterized. Most of them (62%) were identified as Bacillus spp. Nevertheless, six identified species were isolated from honey samples for the very first time. Moreover, the detection of NRPS and PKS (PKS-I) biosynthetic gene clusters in several isolates is reported, underlining the high biosynthetic potential of honey bacteria. Collectively, the results presented in this study demonstrate that the honey microbiome is an untapped source of antimicrobial metabolites against multidrug-resistant nosocomial and food pathogens that warrants further investigation and bioprospecting.