Former investigations were able to associate different steps of the phenazine biosynthetic pathway with the corresponding genes, e.g., phzC, phzD, phzE and phzF. The transformation from chorismate to 2-amino-2-deoxyisochorismic acid (ADIC) is necessary for the formation of the core structure of phenazines and is catalyzed by the enzyme PhzE. Thus, PhzE is a key enzyme in phenazine biosynthesis and the corresponding gene phzE is suitable for primer design. Sequences from the phenazine biosynthetic pathway for Alpha-, Beta- and Gammaproteobacteria, Actinomycetes and Firmicutes are available at the homepage of the National Centre for Biotechnology Information (NCBI) and known from literature [29,30]. To ensure the inclusion of only true phenazine sequences, oligonucleotide primers were constructed only from those genes known to be involved in the biosynthesis of corresponding chemical substances. Two conserved sites occurred within the alignment of phzE sequences (Figure 2), which had a distance to each other to produce fragments of an appropriate length. The degenerated primers phzEf (5′-GAA GGC GCC AAC TTC GTY ATC AA-3′) and phzEr (5′-GCC YTC GAT GAA GTA CTC GGT GTG-3′) were designed to amplify a highly conserved stretch of the phzE gene of approximately 450 bp. The comparison of the oligonucleotide sequences from designed phzEf and phzEr primers with known phenazine genes verified this stretch as highly specific for phzE genes. Because the basic phenazine gene cluster including the phzE gene is highly conserved and derivatization of the basic phenazine structure are made at a later stage in the biosynthesis, the constructed phzE primers are expected to detect genes of a large variety of different phenazine structures and are appropriate to search for unknown bacteria producing novel phenazines.

Genes belonging to the phenazine biosynthetic pathway were present in approximately 10% of the bacterial strains analyzed. PCR results of 13 (8%) out of 164 bacterial strains and four reference organisms were positive in regard to the presence of phzE gene fragments (Figure 3, Tables 1 and 2). The investigated bacteria comprised different bacterial phyla, namely Actinobacteria (76), Bacteroidetes (2), Firmicutes (28) and Proteobacteria (62) (Table 1).

Corresponding gene fragments were detected in 11 strains of Actinobacteria, one strain of Firmicutes and two strains of the Alphaproteobacteria. All sequences were similar to known phzE gene sequences in a range from 65% similarity (phzE of strain LB151 to phzE of P. chlororaphis, AAF17499) to 95% similarity (phzE of strain AB108 to phzE of gene from S. cinnamonensis, CAL34110) (Table 2). Regarding the environmental isolates none of the strains within the Bacteroidetes, Beta- as well as Gammaproteobacteria could be shown to contain phzE in PCR amplification. This was unexpected, because among the 36 gammaproteobacterial isolates 18 Pseudomonas strains were examined and our PCR approach was performed with primer sequences largely based on sequences from Pseudomonas strains known as producers of phenazines [16,31,32]. The suitability of our primer set to detect phenazine genes in Pseudomonas species was further demonstrated by performing a database search that matched perfectly several phenazine genes, e.g., P. chlororaphis (L48339), Pseudomonas sp. M18 (FJ494909), P. aeruginosa (FM209186, CP000744, CP000438, AE004091, AF005404). Anyhow, a study based on phzF sequences exhibited a hit ratio of 100% including 51 Pseudomonas strains [21]. Therefore, all 18 pseudomonads from our study exhibiting negative results using phzE primers were subjected to a genetic approach with phzF primers. While the PCR-amplification of phzE and phzF gene fragments of the control type strains was positive (Table 2), amplification of the investigated isolates failed. As an additional control experiment, crude extracts of six Pseudomonas strains were analyzed by HPLC-UV/MS. Because of the distinctive properties of phenazine UV-absorption spectra the presence of phenazine metabolites was out of question. Additionally, for another study all natural products from two of the investigated Pseudomonas strains were isolated and chemically identified. No phenazines were detected. We conclude that the Pseudomonas strains analyzed in this study lack genes for phenazine production and are unable to produce phenazines. In agreement with this, the only known marine phenazine producing Pseudomonas species is P. aeruginosa [33–35], synthesizing almost always pyocyanin. In contrast, different marine streptomycetes are known for production of variable phenazine structures [9]. Streptomyces strains in this study are the most productive group as well. While Brevibacterium, Bacillus and Pelagibacter were known as marine phenazine producers [36–38], this is the first time that representatives of the genera Micromonospora, Kiloniella and Pseudovibrio were identified as marine phenazine producers as well.

To demonstrate the synthesis of phenazines in all phzE positive strains, cultures of these strains were extracted and analyzed by HPLC-UV/MS analyses. 14 out of 17 of these strains were able to produce one or more substances with molecular masses and UV-spectra similar to known phenazines (Table 3, Figure 4a–c). In S. cinnamonensis DSM 1042^T the production of endophenazines A–C (Figure 5) and phenazine-1,6-dicarboxylic acid [15] could be demonstrated (Figure 3a). The metabolite chlororaphin was discovered from Pseudomonas chlororaphis subsp. chlororaphis DSM 50083^T. 2-hydroxy-phenazine (Figure 5) and phenazine-1-carboxylic acid were produced by Pseudomonas chlororaphis subsp. aureofaciens DSM 6698^T and Pseudomonas chlororaphis subsp. aurantiaca DSM 19603^T. In addition, the presence of senacarcin A (strain Streptomyces sp. HB117), saphenyl ester D, aestivophoenin C and a derivative thereof (strains Streptomyces sp. HB122 and HB291) as well as phencomycin methyl ester and 1-carboxymethyl phenazine from strain Streptomyces sp. LB129 (Figure 3b) were identified.

All environmental isolates producing phenazines (6%) were marine Streptomyces sp. or Micromonospora sp. strains. Most of these strains produced both known phenazines and phenazines which did not show any accordance to a database entry. In total, 22 known phenazines were identified. In the case of strain Streptomyces sp. HB202 (Figure 4c), the production of streptophenazines A-H was verified using NMR spectroscopic analyses [39]. The large number of Streptomyces strains containing phzE genes is in good agreement with previous reports describing streptomycetes as a rich source for phenazines [9,15,39,40].

In nine of the culture extracts a total of 13 different substances showed typical phenazine like UV-absorption spectra, but gave no hit in the databases concerning UV and mass data. This indicates the presence of unidentified and possibly new natural phenazine products which warrant further investigation.

For some of the identified phenazines interesting biological activities were reported. Senacarcin A is known for its activity against Gram-positive bacteria and tumor cell lines [45] and aestivophoenin C has antioxidative activity and acts as a neuronal cell protecting substance [44]. Interesting bioactivities of phenazines are also known from the marine Streptomyces sp. strain HB202, which produced several streptophenazines with activity against Gram-positive bacteria [39].

We expect that investigation of other so far unidentified phenazines from marine Actinobacteria is a remunerative challenge. Interestingly, phenazines were not detected in culture extracts of phzE positive strains of Alphaproteobacteria and Firmicutes. Though, all bacteria containing a phzE phenazine gene fragment have the capability to synthesize the phenazine core structure, proof of gene fragments from a biosynthetic pathway does not give evidence of the integrity of corresponding gene cluster. Additionally, the expression of a gene cluster under conditions used is not warranted. Therefore, it is most likely that the cultivation conditions used were not appropriate for the production of some of the phenazines and have to be modified for the selected strains by our genetic approach in further studies.

166 bacterial strains used in this study were of diverse phylogenetic affiliation and were isolated from Halichondria panicea (HB strains) [46] and Saccharina latissima (synonym Laminaria saccharina) (LB strains) [47] collected at the Kiel Fjord, Germany, and also from different sponges collected from the Adriatic Sea near Rovinj, Croatia (AB strains). The strains belong to six different phylogenetic groups (Table 1). Additionally, type strains known to produce phenazines were used as positive controls: Streptomyces cinnamonensis DSM 1042^T, Pseudomonas chlororaphis subsp. chlororaphis DSM 50083^T, Pseudomonas chlororaphis subsp. aureofaciens DSM 6698^T, and Pseudomonas chlororaphis subsp. aurantiaca DSM 19603^T. For S. cinnamonensis DSM 1042^T phenazine gene sequences and the production of different endophenazines and PCA (phenazine-1-carboxylic acid) have been demonstrated [15]. P. chlororaphis subsp. chlororaphis produced chlororaphin [48], P. chlororaphis subsp. aureofaciens and P. chlororaphis subsp. aurantiaca produces 2-hydroxy-phenazine [42] and phenazine-1-carboxylic acid [41], respectively.

For identification of the strains 16S rRNA gene sequence analyses were carried out according to Thiel et al. 2007 [49]. Comparison of the 16S rDNA sequences was performed using the EMBL nucleotide database available at the European Bioinformatics Institute homepage using the Basic Local Alignment Search Tool (nucleotide blast) [50] and the Ribosomal Database Project (RDP) database [51].

For the primer construction, amino acid sequences and nucleotide sequences of different phzE genes were retrieved from the European Bioinformatics Institute homepage and aligned using the program CLUSTAL_X [52]. Nucleotide sequences were deduced from amino acid sequences. The alignment was analyzed manually. The following phzE sequences were used for primer design: Streptomyces cinnamonensis (AM384985/CAL34110/68793…70757; putative 2-amino-2-desoxy-isochorismate synthase), Pseudomonas chlororaphis PCL1391 (AF195615/AAF17499/4873…6786; phenazine-1-carboxamide), Pseudomonas aeruginosa PAO1 (AF005404/AAC64488/3294…5177; pyocyanin), and Pseudomonas aeruginosa PAO1 (AE004091/AAG07601/4716660…4718543/AAG05292/2073555…2075438; phenazine biosynthesis protein PhzE). Primers (Table 4) were synthesized by MWG (Ebersbach, Germany). In order to check the specificity of the primers, the sequences of phzEf and phzEr were compared with sequences from the EMBL database using the Basic Local Alignment Search Tool (http://blast.ncbi.nlm.nih.gov/Blast.cgi). This comparison revealed 100% identity of the primers with corresponding sites of phenazine biosynthesis genes. Since Ashenafi et al. (2008) [53] reported that the anthranilate synthase (SvTrpEG) of Streptomyces venezuelae has a high degree of amino acid sequence similarity to the phenazine biosynthetic enzyme PhzE, the corresponding nucleotide sequence (AF01267) was compared with the phzE primers using the bl2seq tool (http://blast.ncbi.nlm.nih.gov/Blast.cgi). No significant similarity was found indicating that false positive results are excluded.

The amplification reactions were carried out in a final volume of 25 μL. Taq DNA Polymerase (New England BioLabs, Ipswich, UK; MA, 5 U reaction⁻¹) with the ThermoPol Buffer Kit (New England BioLabs, Ipswich, UK; MA, USA) was applied. Primers phzEf and phzEr were deployed in a 10 μM concentration. 1 μL of a preparation containing each deoxynucleoside triphosphate at a concentration of 2.5 mM was used. 10 to 50 ng DNA of all strains used in this study was employed as template.

The amplification of the phzF gene sequence of the pseudomonads used in this study was performed using puReTaq Ready-To-Go polymerase chain reaction Beads (Amersham Biosciences, Uppsala, Sweden) with the primers Ps_up1 and Ps_low1 [21]. Cycler conditions for both PCR experiments were as follows: Initial denaturation: 94 °C for 120 s followed by 36 cycles of primer annealing at 54.7 °C (phzE) and 57 °C (phzF), respectively, for 60 s; primer extension at 72 °C for 120 s and denaturation at 94 °C for 60 s. A final extension of 72 °C for 420 s was performed. All PCR reactions were conducted in a T1 thermocycler (Whatman Biometra^®, Göttingen, Germany). Results of the amplifications were checked on a 1.5% agarose gel stained with ethidium bromide. DNA sequencing was done according to Wiese et al. [47]. The comparison of the phzE and phzF fragments, respectively, was done in the EMBL nucleotide database available at the European Bioinformatics Institute homepage using the Basic Local Alignment Search Tool (blastx) [50].

For 1 L cultures the supernatants were separated from the cell mass pellets by centrifugation at 4.700 × g for 20 min and extracted separately. Cells were homogenized by addition of 150 mL 96% EtOH and using Ultra-Turrax (IKA, Staufen, Germany) at 13,000 rpm for 30 s. The extracts were dried in vacuo and redissolved in MeOH for further analyses. Supernatants and the other cultures were extracted with EtOAc by homogenization with the help of Ultra-Turrax at 16,000 rpm for 30 s, also dried in vacuo and redissolved in MeOH for further analyses.

Reversed phase HPLC experiments were performed using a C₁₈ column (Phenomenex Onyx Monolithic C18, 100 × 3.00 mm) applying an H₂O (A)/MeCN (B) gradient with 0.1% HCOOH added to both solvents (gradient 0 min 5% B, 4 min 60% B, 6 min 100% B; flow 2 mL/min) on a VWR Hitachi Elite LaChrom system coupled to an ESI-ion trap detector (Esquire 4000, Bruker Daltonics). Dereplication of substances was realized by comparison of MS and UV data obtained by HPLC-UV/MS analyses used data from the Antibase [57] and the Chapman & Hall/CRC Dictionary of Natural Products databases [58]. For endophenazines A and B, 2-hydroxy-phenazine and phenazine-1-carboxylic acid structure was confirmed by ¹H NMR analysis.

The nucleotide sequence data reported in the present study were deposited in the GenBank nucleotide sequence database under the accession numbers HM460698 (AB049), HM460699 (AB108), AJ849545 (YIM 90018), AM231308 (YIM 36723), GQ863906 (HB117), GQ863907 (HB122), GQ863918 (HB202), GQ863921 (HB253), GQ863922 (HB254), GQ863926 (HB291), AM749667 (LB066), AM913982 (LB114), AM913952 (LB129), AM913970 (LB150) and AM913971 (LB151) for 16S rRNA and HM460700-HM460715 for phzE gene fragments.