Standard curves were established by conducting serial dilutions (1.2–1.2 × 10⁶), of cyanobacterial 16S rRNA, Microcystis 16S rRNA, mcyA and mcyB gene cell equivalents/mL of genomic DNA extracted from the M. aeruginosa M6 culture. For cyanobacterial 16S, Microcystis 16S, mcyA and mcyB, highly significant linear curves between the amount of starting DNA (in cell equivalent numbers) and the threshold cycle values (Ct) were obtained. The efficiencies and curve parameters obtained for the different standard curves are summarized in Table 2.

The melting temperature of the different qPCR products of the samples were similar to those obtained with the standard strain (Table 2), thus demonstrating the reliability of the real-time PCR amplification. The variation of the number of both cyanobacteria and Microcystis spp. followed the same pattern throughout the sampling period progressively increasing after the first sampling date (9.5 ± 5.2 × 10¹ and 1.1 ± 0.5 × 10² cell equivalents/mL, respectively), reaching the maximum values at August 12 (2.7 ± 0.1 × 10⁶ and 2.1 ± 0.2 × 10⁶ cell equivalents/mL, respectively), and then progressively declining until the last sampling date (7.2 ± 4.0 × 10³ and 9.7 ± 5.8 × 10² cell equivalents/mL, respectively) (Figure 1). Both the number of potentially MC-producing cyanobacteria and MC-producing Microcystis varied in a similar way throughout the entire sampling period, presenting its maximum value at August 12 (2.5 ± 0.7 × 10⁵ and 2.8 ± 0.9 × 10⁵ cell equivalents/mL, respectively), and the minimum at July 1 (1.2 ± 0.8 × 10² and 1.2 ± 0.8 × 10² cell equivalents/mL, respectively).

The data obtained by qPCR, were also used to evaluate the dynamics of the proportion of toxic genotypes inside the Microcystis spp. population (expressed by the ratio between the number of mcyB genes in cell equivalents/mL and the number of Microcystis 16S rRNA genes in cell equivalents/mL) throughout the sampling period (Figure 2).

Within the Microcystis spp. population, the percentage of potentially MC-producing genotypes varied from around 2% to higher than 100% of the cells. The percentage of toxic genotype progressively and markedly decreased from the first sampling date, at which all the cells of Microcystis seem to present the mcyB gene, and reaching a minimum of around 2% at August 26. Then, the proportion of toxic genotype increased again, constituting around 50% of the entire population in the last sampling date. This dynamics matched the one observed for the mcyA cell equivalents number (data not shown).

MC quantified by ELISA varied between 2.46 (July 1) and 6.97 μg MC-LR eq/L (August 12). MC values followed approximately the same pattern of the variation as toxic Microcystis genotypes (mcyB) densities, except in the last sampling date (Figure 3). The highest value of MC was determined for the same date as the highest number of mcyB cell equivalents (August 12). Curiously, at this date, we can also observe the lowest MC quota per toxic genotype (mcyB) (Figure 4). The MC concentrations were always above the WHO guideline values for drinking water (1 μg MC-LR/L) for all the sampling dates.

MC concentration had a significant positive correlation (r = 0.79; P < 0.05) with the abundance of Microcystis mcyB and of cyanobacterial mcyA cell equivalents (Table 3).

Furthermore, strong positive correlations were found between the abundance of Microcystis mcyB cell equivalents and the abundance of cyanobacterial mcyA cell equivalents (r = 1.00; P < 0.001), and between the abundances of Microcystis 16S rRNA cell equivalents and cyanobacterial 16S rRNA cell equivalents (r = 0.96; P < 0.05). The proportion of toxic genotypes in the Microcystis population correlates negatively, but not significantly, with MC concentration (r = −0.61). Moreover, the proportion of toxic genotypes showed a negative significant correlation with both Microcystis and cyanobacteria abundances (r = −0.89 and −0.86, respectively; P < 0.05). Concerning the influence of limnological parameters on the overall MC level, no significant correlations between MC concentrations and pH, DO, water temperature or conductivity were detected (Table 3).

The qPCR technique is currently the only quantitative tool available to determine the proportion of potential toxin-producing genotypes in environmental water samples. Although some previous reports have showed a close relationship between the occurrence of certain toxic genotypes (containing toxin genes) and MC net production [38], and that it is possible to infer MC concentrations from Microcystis cell numbers [39], correlations between MC levels and toxic genotypes numbers are often difficult to obtain [26,40]. The work presented here aimed to help clarify these contradictory observations. In this work, the proportion of toxic Microcystis genotypes varied between approximately 2% and around 100% of the entire Microcystis population. These values are above others previously reported between toxic and non-toxic Microcystis strains, such as 10% in Lake Kasumigaura (Japan) [41], 34% in Lake Kasumigaura [42], 16% in Lake Grand-Lieu (France) [43], 45% in Lake Wannsee and Lake Pehlitzsee (Germany) [44], ranging from 1 to 38% [39]. More recently, proportions of mcyB genotypes ranging from 6% to 93% of the M. aeruginosa population were reported in Grangent reservoir, France, with the highest proportions appearing at the beginning and after the decline of the bloom [45]. Our results also revealed an inverse relationship between density and frequency of toxic genotype and the highest Microcystis cell densities, with the former being detected 4–6 weeks before and 8–10 weeks after the latter (Figures 1 and 2). Indeed, significant negative correlations were obtained between mcyB genotype proportions and cyanobacteria and Microcystis cell densities (Table 4). Nevertheless, the highest toxic genotypes numbers and MC concentration values were, in fact, found when the highest Microcystis cell densities were observed. McyB is specific to Microcystis producing MC and mcyA is present in all cyanobacteria genera that produce MC. By using these two markers it can be determined whether the MC is mainly produced by Microcystis, which was the case, or whether other cyanobacteria are responsible for the toxin production.

Detection of MC-producing genotypes during non-bloom season (winter), has also been previously reported [26], however, no MC production could be detected. This raised the possibility that toxic cyanobacteria do not produce MC during times of low growth or low metabolism (during the winter), although mcy genotypes may be present [26,39]. In the present work, MCs were always detected, with the highest MC cell quotas being observed simultaneously with the lowest Microcystis numbers and vice versa (Figures 2 and 4). This may suggest that MC production may play an important role when conditions do not favor Microcystis growth. Previous data also showed a negative relationship between cyanobacterial biomass and average MC cell content during the development of cyanobacterial blooms [45].

On the other hand, the observed decrease in toxic genotypes proportion as the bloom developed suggests that favorable environmental conditions for Microcystis growth lead to the selection of non-toxic genotypes. In fact, it was previously shown that, under environmental conditions favorable for cyanobacterial growth, the fitness of non-MC-producing strains is greater than that of MC-producing ones [34,47]. Similarly, a negative correlation between cell abundance and the proportion of potentially MC-producing genotypes was observed during a P. agardhii bloom [48].

Considering the influence of environmental factors on MC levels and on Microcystis population dynamics, in the present work, no significant correlations were found between MC concentration and any of the limnological parameters measured. The regulation of the internal dynamics of MC-producers populations is predictable to be affected by a large array of environment conditions [49]. For instance, the influence of nitrate concentrations on the relative dynamics of mcy genotypes was shown within the M. aeruginosa communities of Lake Mikata [24]. Furthermore, it was already suggested that grazers play an important role on this dynamics. Due to selective grazing, depending on whether the cell is producing toxin or not, grazers may affect the total biomass of cyanobacteria and Microcystis, as well as the relative abundance of Microcystis genotypes [26,32]. Other authors suggested that cyanophages may be involved in the interaction between microcystin-producing and non-microcystin-producing M. aeruginosa populations [49].

The results now obtained confirm that one of the most known problems of qPCR methodologies is the difficulty in converting the gene quantities to quantities of cells that carry these target genes. This can distort the enumeration of cyanobacterial cells in natural samples by qPCR. In fact, as described previously in other studies [26], some samples showed an abundance of potentially toxic Microcystis higher than the total Microcystis abundance and in other cases, the abundance of Microcystis cells appeared >100% of the total abundance of cyanobacteria. These results may be due to the use of the 16S rRNA gene as a target for quantification of both cyanobacteria and Microcystis due to the variable copy number of 16S rDNA operons or their sequence heterogeneity in cyanobacterial cells in natural populations [26,47]. Moreover, some bias can be introduced by many mechanical factors, such as sample handling, DNA extraction, qPCR standards preparation and PCR conditions [50].

The present work showed, nevertheless, that qPCR can be used to easily and quickly estimate the toxicity of environmental samples. In fact, toxic genotypes densities correlate well with MC concentrations, suggesting that they could be used to predict MC toxicity risk. This work also showed that as blooms develop, the proportion of toxic genotypes decreases, and that at low densities most of the Microcystis genotypes are toxic. These observations indicate that conditions favoring Microcystis growth lead to the selection of non-toxic genotypes and, on the other hand, that even at low cell densities there is high potential MC threat. This raises greater concern on the presence of potential MC-producing cyanobacteria in temperate water reservoirs, particularly considering long term exposure effects, making it imperative to develop more efficient monitoring methods.

Torrão reservoir is located on the Tâmega River, in the north of Portugal, being used for water consumption and recreational purposes. This reservoir presents periodic dominance of toxin-producing cyanobacteria, including Microcystis aeruginosa and Aphanizomenon flos-aquae [37]. Surface water samples (0–0.3 m depth) were collected from Marco station (Torrão reservoir), where a water treatment plant is located, every two weeks between July 2009 and September 2009 and then monthly until the end of October. The collected samples were immediately transported cooled to the laboratory. For DNA extraction, volumes of 40 to 2000 mL of well mixed water samples were filtered through GFC filters (Whatman, Kent, UK) in triplicate, and the filters stored at −80 °C. For ELISA analysis, 15 mL of water samples were collected and stored at −20 °C. Limnological parameters were measured in situ namely, dissolved oxygen (DO), measured with the oxygen meter Oxi 320 (WTW, Weilheim, Germany), temperature, conductivity and pH, measured with the MultiLine P3 pH/LF (WTW, Weilheim, Germany) instrument.

DNA was extracted from the biomass retained in the frozen GFC filters, using a commercially available Kit, PureLink Genomic DNA Mini Kit (Invitrogen, Carlsbad, CA), following the manufacturer’s instructions for gram negative bacteria. The extracted DNA was suspended in 100 μL of elution buffer (pH 8.0) (Invitrogen, Carlsbad, CA) and stored at −20 °C. The reproducibility of the extraction method was evaluated by spectrophotometric quantification of the DNA obtained from the filter replicates. The qPCR reactions were performed using Bio-Rad cycler IQ5 (Bio-Rad, Hercules, CA) with 5 μL of DNA from the standard or sample, 0.4 μM of each primer (0.2 μM for MCYA-CD1F/CD1R and MCYB-2959F/3278R primer pairs) and 1X IQ SYBR Green Supermix (Bio-Rad, Hercules, CA), in a final reaction volume of 25 μL. The CYA 359F/781R primer pair [51] was used to amplify a 422-bp fragment of the 16S rRNA gene common to all cyanobacteria. The Micr 184F/431R [52] was used to amplify a 248-bp fragment of the 16S rRNA gene, specifically from Microcystis cells (Table 4).

The mcyA CD1F/mcyA CD1R primer pair [22] was designed to amplify a 297-bp fragment of the mcyA gene from MC-synthesizing Microcystis and Planktothrix strains and was previously proved suitable to detect MC-producing cells from different genera, including Anabaena, Microcystis and Planktothrix (Table 1). The mcyB 2959F/mcyB 3278R primer pair [46] was used to amplify a 320-bp fragment of the mcyB gene, which is present in MC-producing strains of Microcystis. The thermal cycle conditions are summarized in Table 5. The generation of products was monitored after each extension step by measuring the fluorescence intensity of the double-stranded DNA-binding SYBR Green I dye. All the samples were amplified at least in triplicate. Cycle threshold (Ct) was determined by the fit point method using Bio-Rad IQ5 software (ver. 2.0.148.60623) (Bio-Rad, Hercules, CA). The number of cyanobacterial cells was determined from the Ct obtained according to the regression equations of the external standards. To determine the melting temperatures of the amplification products of the standards and samples, fluorescent melting curve analysis was performed after the PCR by gradually increasing the temperature from 55 °C to 95 °C at a rate of 0.5 °C/cycle. Fluorescence intensity data was collected continuously, and converted to melting peaks using Bio-Rad software. The presence or absence of the amplified product was determined by a melting curve analysis and by subjecting the reaction products to 1.5% (w/v) agarose gel electrophoresis. Ethidium bromide stained bands were visualized on a UV transilluminator (Cleaver Scientific, Warwickshire, UK) and digitally recorded using MicroDOC gel documentation system (Cleaver Scientific, Warwickshire, UK).

A cyanobacterial strain of M. aeruginosa (M6) (reference strain) was used for the production of a standard curve. This was based on predetermined cell densities and was established by relating the known DNA concentrations (in cell equivalents) to the Ct values of the diluted samples. 1.5 mL of a suspension containing 2.37 × 10⁷ cells of M. aeruginosa (M6) suspension (determined by a direct microscopic count) were collected by centrifugation, and seven dilutions of template DNA ranging from 1:10 to 1:10⁶ (equivalent to 1.2 × 10⁶ cells and 1.2 cells, respectively) were prepared from the DNA extract to serve as external standards of the real-time PCR.

The Envirogard^® MC Plate Kit (Strategic Diagnostic Inc., Newark, NJ) was used to quantify MC in the water samples collected. This assay uses antibodies against MC-LR, the most common MC, binding also with other MC variants such as MC-RR, MC-YR, or with NOD [53]. Fifteen milliliters of whole water samples were frozen and thawed, and then sonicated on ice for 5 min at 50 W to facilitate cell lysis. The resulting solutions were then applied to the above mentioned ELISA kit following the manufacturer’s instructions. The absorbance was measured using a Synergy™ HT Multi-Detection Microplate Reader (BioTek^® Instruments Inc., Winooski, VT) using BioTek’s Gen5™ Data Analysis Software. Values are presented in μg MC-LR equivalents/mL. Previous work has shown that ELISA is efficient for the quantitation of total MC in this reservoir, taken into account the MC chemical profile [54].

The nonparametric Spearman Rank Correlation coefficient was calculated as a measure of correlation between all possible pairs of variables. Statistical analysis was performed using the GraphPad Prism version 5.01 for Windows (GraphPad Software, San Diego, CA) statistical analysis software. Correlations showing P value <0.05 were considered significant in this analysis.