Many freshwater cyanobacteria produce blooms which may be toxic to fauna, flora, and humans (Chorus and Bartram, 1999) and raises the importance of making rapid diagnosis of toxin content and/or the toxic potential of these blooms. General phycocyanin probes can be easily deployed in a natural environment to rapidly detect total cyanobacteria concentrations (Brient et al., 2008), while toxin concentrations are measured by standard HPLC or mass spectrometry based techniques (Lawton and Edwards, 2008). Although these latter methods are time consuming, they remained references for the calibration of rapid biochemical (ELISA and PPIA) assays. Potentially toxic species are often identified by biomolecular (PCR) assays (Carmichael and An, 1999; Pearson and Neilan, 2008; Humpage et al., 2012) but they involve DNA extraction which destroys cells and colony organization. Identification of cyanotoxin producing species requires additional time consuming microscopic isolation, purification, and cultivation. Thus a rapid assay, which simultaneously provides the identity and the activity of cyanotoxin producing species, still remains a challenge.

The mcyA gene has been frequently used in many molecular based approaches, as it is involved in the ribosomal production of microcystine (MC) synthetase, responsible for MC production, through non-ribosomal pathways (Tillett et al., 2000). Many studies have shown a transcriptional regulation of mcyA gene in accordance with changes in toxin concentrations, although discrepancies have been described since early studies which suggested a post-transcriptional regulation of mcyA gene expression (Schawbe et al., 1988; Meissner et al., 1996). Very recently, a whole cell Tyramid Signal Amplification-Fluorescent in situ Hybridization (TSA-FISH) assay was developed to target mcyA mRNA transcription as a proxy of MC-synthetase production (Zeller et al., 2016). This assay could verify an up-regulation of mcyA-mRNA synthesis by light, and show similar spot-like fluorescent labeling as the rpob-mRNA probe, where the rpob gene is involved in RNA polymerase synthesis (Gaget et al., 2011). That study demonstrated the potential interest to distinguish toxic from non-toxic organisms, by combining the precision of molecular based technique with species identification, as cellular integrity was not destroyed. However, direct measurement of MC toxin by standard HPLC methods was not provided, and it's in situ application remained to be demonstrated.

The aim of the present study was to demonstrate that (i) mcyA-mRNA TSA-FISH assay could label MC producing laboratory strains, as well as cells from the natural environment and (ii) identification of MC producing species could be easily acquired simultaneously by the use of specific identity MICR3-16S rDNA probe; when (iii) toxin concentrations were verified by standard HPLC and/or LC-MS/MS protocols.

In this study, we successfully applied a simple TSA-FISH protocol (Table S1) to detect specifically in cultures and the natural environment MC-producing cells. Positive control of TSA-FISH assay, both in culture and the natural environment, was ensured by the eubacterial domain specific probe EUB338 (Figures 1, 2) which brightly labeled cells with fluorescein fluorochrome (FITC). Presence of Microcystis genus was specifically targeted by the MICR3 probe (Figures 1, 2B1,B2, 3A, 4A,B) while the Chroococcale Woronichinia spp., the closest phylogenetic none specific genus, were not targeted (Figure 4A thin arrow, Figure 4B thick arrow), nor the most distantly related filamentous Nostocale Dolichospermum mendotae (Figures 2B1,B2 arrow). TSA-FISH assay is a widely used technique for the specific detection and quantification of cyanobacteria (Schonhuber et al., 1999; West et al., 2001; Biegala and Raimbault, 2008). However, the detection of messenger RNA has long been believed to be challenging (Pernthaler and Amann, 2004) because of low copy number of specific mRNA copies per cell. Trcek et al. (2012) and more recently (Zeller et al., 2016) successfully labeled single mRNA within microorganisms, including M. aeruginosa strains. Both studies showed the precision of in situ hybridization approaches using either multiple monolabeled DNA probe (DIG-FISH) or a simple TSA-FISH assay, respectively. In addition, once fixed with PFA (paraformaldehyde) all proteins are crosslinked which prevent RNA degradation from endogenous RNAse. Both observations justified the use in this study of similar TSA-FISH protocol to target whole cell rRNA and mRNA.

We further demonstrated in this study that MCYA probe, specific for MC-synthetase messenger RNA, only targeted toxic Chrooccocales cells belonging to three different species of Microcystis. These species are M. aeruginosa from cultivated strains (Figure 1, Table 1), Microcystis viridis and Microcystis wesenbergii from Vioreau reservoir (Figures 2C1, 4C, Table 1). On the contrary, the non-MC-producing M. aeruginosa PCC 9432 strain (Figure 1, Table 1) and the Microcystis flos aquae from Vioreau reservoir were not targeted (Figure 2C1 arrow, Table 1). In addition, the MCYA probe labeled mcyA mRNA as many small spots within P. rubescens toxic strain (Figure 5, Table 1), as well as the highly toxic bloom of Planktothrix isothrix from Drennec lake (Figure 4D arrow, Table 1), while filaments from the non-MC-producing P. agardhii PCC 7805 were not labeled by MCYA (Figure 5, Table 1). Unexplained uniform green fluorescence could also be seen in some filaments from this non-toxic PCC 7805 strain.

According to our results and those from Zeller et al. (2016) it is interesting to note that two kinds of intracellular labeling could be observed in MC-synthetase producing cyanobacteria, (i) small dots spread within the cytoplasm and (ii) densely packed cytoplasmic labeling, similar to the one of 16S rRNA probes (EUB338 and MICR3) from both Chrooccocale and Oscillatoriale species in cultures or in situ. Moreover, a focus on some M. aeruginosa cells from the PCC 7806 strain, cultivated in this study, shows a clear rRNA and mcyA-mRNA co-localized-labeling in the surrounding of densely packed plasmids (Figures 3A,B). McyA-mRNA labeled plasmids from PCC 7806 M. aeruginosa were also found extracellularly (Figure 3C, arrows). Previous TSA-FISH or DIG-FISH studies (Metcalf et al., 2009; Gan et al., 2010) on mcyA genes fluorescent detection have also shown some intracellular variability in accordance with toxin measurements. However, the lack of controls and information on time exposure during micrographs acquisition prevent from a clear understanding of presented micrographs.

The variations in mcyA-mRNA labeling observed in this study could be explained by the plasticity of microbial genome including that of toxic cyanobacteria (Figure 6) (Toussaint and Merlin, 2002; Tooming-Klunderud et al., 2008) where gene clusters are either integrated in chromosomal DNA with fewer mRNA transcripts, or can be multiplied in several hundreds of plasmids to rapidly synthetize metabolite through the production of intron-deprived mRNA (Schumann, 2008). We hypothesize that ribosomes targeted by MICR3 probe are either spread in the cytoplasm, or densely packed around the plasmids. This process could explain both the small dots in some Microcystis strains (Zeller et al., 2016) and Oscillatoriales (Figure 4D) and the co-localization of rRNA and mRNA Microcystis strains PCC7806 from this study (Figures 3A,B). Interestingly toxic strains have previously been shown to contain more extrachromosomal DNA than non-toxic ones (Schawbe et al., 1988), or to produce abundant mcyA PCR product (Valerio et al., 2009). All these results are coherent with the theory of horizontal gene transfer to explain sporadic distribution of mcy gene cluster in cyanobacteria (Kalaitzis et al., 2009). Figure 3C (arrows) also shows extracellular release of plasmids in M. aeruginosa, which is a well-known process in bacteria (Matsui et al., 2003) to respond to biotic or abiotic stresses (Paul and David, 1989; Alonso et al., 2000).

This study demonstrated that an easily accessible TSA-FISH protocol could be used as a diagnostic assay to characterize both the potential of a cyanobacterial bloom to be toxic by identifying Microcystis spp. with the specific MICR3 probe, and by simultaneously labeling all MC-producing species with MCYA probe. In addition, this powerful tool will help to further develop ecological and biomolecular studies on toxic cyanobacteria as visual localization of gene expression can be monitored.

LB contributed to the different sections of this manuscript. NB aquired most TSA-FISH results. MP contributed to TSA-FISH results aquisition. MR did the specificity test and contributed to the results section. MB contributed to the manuscript. PZ, OP, and AM designed the FISH probes. IB designed the project and contributed to all sections of the manuscript.