The description of the biosynthesis gene clusters for cyanotoxins allowed for the development of PCR assays to detect sequence targets within selected genes, which served to identify potentially toxic strains. These assays were soon adapted for a qPCR format, enabling the quantification of toxin genes. This tool had an expected application for monitoring toxic blooms and also for studying the dynamics of fluctuations in potentially toxic to non-toxic genotypes in response to environmental factors. A methodological description of the qPCR method is not included here, but can be found in Martins and Vasconcelos [46
], where its use in studying cyanobacterial population dynamics is also described, and in Kim et al.
], where its application for quantitatively investigating environmental microbial communities is discussed. Here, we focus on the use of qPCR to assess the toxicity of cyanobacterial blooms that produce MC, CYN or STX, since these toxins are prevalent in freshwater and have been addressed with molecular methods. The geographical distribution of the studied environments is represented in Figure 1
. A chronological overview of the literature available on qPCR for these cyanotoxins is shown in Figure 2
3.1. qPCR for Detecting Microcystin Genes
Pioneer studies that used qPCR to quantify MC-producing genotypes appeared in 2002 and 2003. The first attempt involved a Taq nuclease assay to detect microcystin synthetase genes (mcyA
]. This assay was able to discriminate between the toxic and nontoxic strains of Microcystis
; additionally, a preliminary test was made with cyanobacterial environmental samples, but no data on parallel MC quantification were reported.
Later, Kurmayer et al.
] proposed a different method for quantifying toxic genotypes by combining primers to quantify the total Microcystis
cells (with the intergenic spacer region of the phycocyanin operon as the target) and MC-producing genotypes (mcyB
gene as the target). In this study, relative quantification was based on serial dilutions of template DNA and not on a conventional qPCR assay. In using this tool, the authors concluded that higher mcyB
-to-phycocyanin ratios corresponded to higher MC cell quotas and that variations in the proportion of MC-producing genotypes could explain the MC net production in Microcystis
colonies of different sizes.
During the same year, Kurmayer and Kutzenberger [50
] developed a Taq nuclease assay that also targeted the phycocyanin locus and the mcyB
gene, but by using new primes/probes. The assay was validated with laboratory strains of Microcystis
and also with samples from Microcystis
blooms. This study concluded that cell quantification by PCR correlated significantly with cell numbers as determined by microscopic counting, and the investigators applied this method to following seasonal shifts in MC genotype proportions in a lake. However, toxin concentrations were not measured, and correlations between mcyB
quantification and toxicity were not evaluated.
Also in 2003, Vaitomaa et al.
] applied qPCR by using a set of primers to target mcyE
, which was designed to distinguish Microcystis
strains. The mcyE
copy numbers correlated positively with the MC concentrations, and it was postulated that this method could be valuable for monitoring hepatotoxic blooms. However, it should be noted that a positive correlation between the MC concentration and mcyE
copy number depended on whether the mcyE
data were considered individually for Microcystis
, or in combination (as the sum of mcyE
copy numbers for both), and it also depended on whether the samples from the two lakes were considered individually or in combination.
From that year on, a number of studies that combined qPCR and MC quantification to analyze toxic cyanobacterial blooms appeared (Figure 2
), and these tools are still proposed for risk assessment. These studies are shown in Table 1
. This set of papers was the result of searches that were performed in PubMed, Science Direct and Google Scholar using the keywords “microcystin” and “qPCR”, and some papers were also recovered from the reference list of other studies. Studies dating from 2003 to 2015 are listed. Key aspects related to the nature of the studied samples, PCR design and the possible correlation between the MC concentrations and qPCR results or proxy parameters, such as chlorophyll-a and cell density, are included. In gathering these data, our primary intent was to estimate the degree to which the mcy
gene copy numbers, as assessed by qPCR, correlated positively with the MC concentration. The sample fraction used for MC extraction and the detection method or the MC concentration were also listed to investigate if these factors would influence the correlation between qPCR results and toxin quantification. The target mcy
gene that was selected for qPCR was included to provide an overview of the most common assay designs.
From the studies considered here, 22 reported a consistent positive correlation between the MC concentration and mcy
gene copy number, and this correlation was not observed in 11 reports. Unfortunately, some studies were inconclusive in this respect, because this correlation was not tested [49
]. There were also some cases in which the mcy
copy numbers were determined, but the MC concentration was reported as the cell quota; therefore, the possible correlation could not be evaluated. In considering the studies that tested the correlation between the MC concentration and mcy
gene quantification, we asked if certain methodological factors could influence the results.
The most commonly-used target mcy
genes were mcyA
. Primers are described for different regions of these genes, and they are generalists for any MC producer or specific for Microcystis
. In approximately 80% of the cases in which mcyE
was quantified, positive correlations were found with the MC content. For the other cited mcy
genes, this correlation was found in approximately 60% of the cases. In principle, any mcy
gene would be equally reliable for evaluating the presence of an mcy
cluster and the number of toxic cells, because they are single copy genes. Even in studies that measured more than an mcy
gene simultaneously, quantification resulted in similar patterns, but did not match completely for different mcy
]. This finding can be attributed to different primer efficiencies when targeting mcy
genes, or because of the presence of incomplete mcy
clusters. In more recent studies, there is a tendency to choose target genes that code for enzymes that are involved in the first steps of cyanotoxin biosynthesis, with the aim of increasing reliability, avoiding genes that code for tailoring enzymes.
Considering the sample fraction from which MC was obtained, 60% of these studies used particulate material, and almost 40% measured the total MC (only one study measured dissolved MC). The quantification of mcy
gene copies correlated better with MC concentrations when the total toxin concentration was considered (in 80% of the cases) than when only the particulate fractions were used (in 60% of the cases). Although MC is an intracellular compound, it can be released in water by cell lysis, and it can then be degraded by heterotrophic bacteria or adsorb to particulate matter. Thus, its extracellular concentrations are usually low [85
]. However, from a risk assessment perspective, it is important to consider the total MC once significant concentrations can be found dissolved in water. This is especially true under conditions involving some mitigation actions to reduce cyanobacterial bloom density, which can include chemical control and/or the natural senescence of a bloom [54
In relation to the method used for MC quantification, 32% of reports considered in our search used ELISA; 26% used HPLC; 18% quantified the MC by PPIA; 13% used LC-MS/MS; and 11% used more than one method. MC quantification by ELISA seems to correlate better with the mcy gene content because a positive correlation was found in all studies in which the immunoassay was used. When other methods were applied for toxin quantification, the percentage of studies that reported a positive correlation was lower (63% for HPLC, 50% for LC-MS/MS, 29% for PPIA). However, there is no consensus about which analytical method can be considered the standard for determining the MC concentration, because they vary in selectivity and sensitivity. Therefore, it is not possible to conclude that ELISA would be better for MC quantification in combination with mcy gene copy numbers. In fact, LC-MS/MS has been increasingly applied to quantify MC because of its superior specificity.
In summary, although methodological factors may favor finding a positive correlation between mcy qPCR data and MC contents, from the examination of this limited number of studies, it is not possible to conclude that a single technical aspect could be critical.
Due to the fact that approximately one-third of the available reports did not identify this type of correlation, a question about the actual utility of qPCR for the risk assessment of toxic cyanobacterial blooms naturally emerges. As more studies appear, this conclusion can be re-evaluated, but the actual picture does not justify the conclusions found in many papers regarding the promise of qPCR for monitoring bloom toxicity [52
However, for the same set of studies listed in Table 1
, when the correlation between the MC concentration and chlorophyll-a or number of cyanobacterial cells was tested, it was positive in 84% of the cases. Thus, the degree to which qPCR is advantageous in comparison with these traditional monitoring approaches should be questioned during risk assessments of cyanobacterial blooms.
qPCR for Determining the Microcystin Toxic Genotype Proportion
In some studies, the mcy
copy number was directly related to MC concentrations, while in others, qPCR was designed to quantify both mcy
genes and a housekeeping gene (16S rRNA or cpc
) to account for the total cyanobacteria or Microcystis
abundance, and the percentage of toxic genotypes is reported (Figure 3
). Thus, a possible correlation between the percentage of toxic genotypes and MC concentrations was tested. The results are contradictory in this regard. In some cases, the MC content showed a clear correlation with the proportion of potentially toxic cells [9
]. However, in other studies, this finding was not observed [55
]. In some cases, even if the mcy
copy numbers alone showed a significant positive correlation with the MC concentration, the percentage of toxigenic Microcystis
did not show this correlation.
The translation of gene copy numbers into cyanobacterial cell numbers is prone to errors, according to numerous studies. Toxin genes are present in single copies in the genome of sequenced strains. Consequently, in principle, the quantity of these genes should never outnumber the 16S rRNA gene (a multiple copy gene) count or the cell concentration, as determined by microscopy. However, several studies reported toxin gene numbers that were greater than the cell numbers [10
]. When another gene is quantified to estimate the total cyanobacteria density (a 16S rRNA gene or cpc
locus, for example), inconsistent results between the gene copy numbers and total cyanobacteria cell densities are commonly found [10
]. This inconsistency results in a cumulative source of errors that is reflected in the mcy
/16S rRNA gene ratio.
There are multiple reasons for errors during the quantification of gene copy numbers by qPCR; there is natural variability in the gene copy numbers (16S rRNA) in the genomes of different species, polyploidy, the presence of an unknown number of target gene copies per cell because of ongoing chromosome replication, the presence of environmental DNA from dead cells that serve as PCR templates, a loss of cells when processing samples for DNA extraction, the incomplete recovery of DNA from cells, the presence of enzyme inhibitors or the differential selectivity of primers because of genetic variation in the target regions of different strains and species. All of these issues are more complicated in the case of 16S rRNA quantification because it is a multiple copy gene. However, errors in cyanobacterial cell counting by microscopy are also possible (particularly for filamentous cyanobacteria and for picoplanktonic species). Thus, when qPCR gene abundance is compared to cell counts, deviations from both approaches are expected.
3.2. qPCR for Detecting Cylindrospermopsin Genes
In comparison with microcystin, the number of studies that apply qPCR to monitor CYN-producing blooms is reduced. These studies are listed in Table 2
and are briefly discussed below.
In 2008, Rasmussen et al.
] described the first qPCR assay for detecting CYN-producing cyanobacteria, more specifically, C. raciborskii
. It consisted of a duplex Taq nuclease assay to target a pks
genetic determinant (the cyrC
gene, which was known at the time as the C. raciborskii
homologue of the A. ovalisporum aoaC
gene) and the rpoC1
gene (a RNA polymerase gene). The rpoC1
gene target sequence was specific for detecting C. raciborskii
, and the toxin gene target sequence was common to diverse CYN-producing cyanobacteria. In environmental samples, the qPCR was specific, with positive results for all samples with C. raciborskii
cell densities above 103
. Although the detection of the toxin gene by qPCR was always consistent with positive results for CYN by LC-MS/MS, no correlation was observed between the cyrC
copy numbers and toxin concentration. It was postulated that this finding occurred because a significant proportion of CYN is in the extracellular fraction and can be heterogeneously distributed in the water body, and during that study, only the intracellular fraction was considered.
In samples in which C. raciborskii was the only CYN-producing species, the cyrC and rpoC1 copies were approximately 1:1, which was consistent with the fact that both genes are present in single copies in the C. raciborskii genome. This finding also indicated that all of the detected strains were potentially toxic. It was concluded that the method was sensitive and rapid for detecting potential CYN-producing cyanobacteria in field samples.
A further evaluation of the usefulness of this qPCR assay was performed with samples from three subtropical reservoirs in Australia where C. raciborskii
blooms were registered [87
]. Relations between the total cyanobacteria, C. raciborskii
cell density, CYN concentrations and qPCR results were analyzed. In addition to the duplex Taq nuclease assay developed by Rasmussen et al.
], a single qPCR was performed to quantify the total cyanobacteria. Thus, the cyrC
copy numbers were normalized by C. raciborskii
cell counts, and this value was related to the CYN cell quotas derived from toxin quantification. A positive correlation between the cyrC
cell quota and the CYN cell quota was found, and the authors concluded that a qPCR analysis of cyrC
in combination with the C. raciborskii
cell count could be used to estimate the intracellular CYN concentration in field samples. These investigators also found that the spatial and temporal variations in cyrC
cell quotas in these reservoirs suggest that the toxicity of C. raciborskii
blooms might result from the relative abundance of strains with different CYN cell quotas. However, it could also result from variations in the relative abundance of potentially toxic strains because the cyrC
copy numbers were lower than the rpoC1
copy numbers, at variable levels. However, in this study, qPCR was not effective as an alternative method for quantifying cyanobacteria, given that a poor linear relation was observed between the gene copy numbers (both rpoC1
and 16S rRNA) and C. raciborskii
A qPCR method that was similar to that reported by Orr et al.
] was proposed by Moreira et al.
], but no information about the correlation between the cell density and 16S rRNA or rpoC1
copy numbers was reported, and validation with field samples and a comparison with toxin concentrations was not possible.
The original idea of Rasmussen et al.
] to apply qPCR to near real-time monitoring of CYN in the field was tested by Marbun et al.
] in Taiwan’s reservoirs. This assay was tested with field samples, from which five samples were positive for C. raciborskii
and two for CYN. In these cases, the microscopic cell counts and total toxin concentrations were in accordance with the qPCR results. All of the steps were performed on-site within 4 h after sampling, indicating that qPCR could be applied for the rapid on-site detection of toxic C. raciborskii
A different qPCR assay using the cyrA
gene as a target was able to detect different CYN-producing genera. It was included as part of a TaqMan-based multiplex qPCR assay reported by Al-Tebrineh et al.
], which also included an estimation of total cyanobacteria through the quantification of 16S rRNA genes. This assay was applied to samples from a mixed cyanobacterial bloom [8
]. The temporal variation of cyrA
copy numbers over three sites indicated that CYN-producing cells initially dominated the bloom and then declined. The toxin concentration correlated positively with cyrA
A TaqMan-based qPCR assay for quantifying A. ovalisporum
CYN production was developed by Campo et al.
]. This assay was able to discriminate CYN-producing A. ovalisporum
strains from other Nostocales, such as C. raciborskii
and A. bergii
. This approach was tested using field samples, for which the presence of CYN-producing A. ovalisporum
was demonstrated. In three samples, the quantification of cyrJ
gene copy numbers showed a positive correlation with extracellular CYN concentrations. The copy numbers of rpoC1
were higher than the cyrJ
copy numbers in one sample, suggesting the presence of non-toxic A. ovalisporum
cells. In spite of the limited number of samples, the authors concluded that the qPCR assay was sensitive and specific for the quantification of potential CYN-producing A. ovalisporum
in environmental samples.
In a recent study, qPCR was applied to track shifts in the proportion of toxic and nontoxic strains of C. raciborskii
in response to nutrient availability during a bloom [91
]. Upon testing the effects of different nutrient concentrations on toxicity, a high correlation was reported between the cyrA
/16S rRNA ratio and the calculated CYN cell quotas, indicating that shifts in the relative proportion of toxic and non-toxic strains could be considered a major cause of variation in bloom toxicity.
Another application of qPCR for detecting CYN producers was reported as part of a multiplex reaction [56
] (see the multiplex section below). In this case, the cyrC
copy numbers were highly correlated to the CYN concentrations, rpoC1
copy numbers and Cylindrospermopsis
cell density, indicating that C. raciborskii
was the CYN producer in these samples, and a high proportion of potentially toxic cells was present.
Given that the original proposal was intended to employ a qPCR assay to detect CYN-producing cyanobacteria, the primary objective was to use it as a molecular approach to estimate the potential toxicity of blooms rapidly. Thus, the critical factor in the usefulness of this method is whether the cyr
gene copy numbers reflect the CYN concentrations in water samples. Most studies listed here have tested this possibility, and in some cases, a positive correlation was found between the toxin concentration and cyr
gene copy number. However, it should be noted that this approach resulted from the use of different methodologies to quantify the CYN (Table 2
). CYN is known to be released from cells, and a significant proportion can persist when dissolved in water [93
]. Thus, to achieve an accurate estimation of toxin concentrations in water, it is necessary to combine the particulate and dissolved fractions or to convert the concentration to a cellular quota when only the intracellular fraction is analyzed.
The combination of a cyr gene with a marker gene (rpoC1) to assign toxin production to a certain species was found to be useful for both C. raciborskii and A. ovalisporum blooms. Although a consistent correlation between the cell concentrations and the rpoC1 gene copy number was not always met, the ratio between the cyr and rpoC1 numbers could be a good indicator of the proportion of potentially toxic strains in the population.
3.3. qPCR for Detecting Saxitoxin Genes
The characterization of the STX biosynthesis gene cluster in different species of cyanobacteria made it possible to develop a qPCR assay to quantify potentially STX-producing cells [95
]. This assay was based on the detection of the sxtA
gene. Primers were designed by considering the low variability of sxtA
gene sequences in A. circinalis
isolates. As an internal control, a pair of primers was included to target a 16S rRNA gene sequence that is conserved in all cyanobacteria. The sxtA
primers were able to generate specific products from other STX-producing species (L. wollei, Aphanizomenon
sp. and C. raciborskii
), but qPCR resulted in lower amplification efficiencies and different melt curve profiles compared to the results for A. circinalis
qPCR assay was then applied to 13 samples of cyanobacterial blooms that were collected from diverse ecosystems in Australia (Table 3
). A microscopy analysis revealed that A. circinalis
was the dominant species in these samples. The STX concentrations determined by HPLC correlated positively with the sxtA
copy numbers. The qPCR-inferred STX concentrations resulted in higher values than those that were directly measured by HPLC. Another discrepancy was observed in some samples in which the estimated cell density based on sxtA
copy numbers gave higher values than microscopic cell counts. This study indicated that the sxtA
copy numbers could be used to estimate the potential toxigenicity of toxic A. circinalis
blooms. However, for other STX-producing cyanobacteria, this assay may require further optimization. Although this optimization can pose a problem when establishing a general method for detecting STX-producing cells, this sequence variability could be explored to develop a more discriminatory qPCR assay that would be able to distinguish STX-producing cyanobacterial species in environmental sample.
Later, the same group proposed another approach to sxtA
detection as part of a multiplex qPCR assay for targeting the genes of several cyanotoxins, as described below in the multiplex qPCR section [96
]. In this case, a new set of primers/probes for sxtA
was designed on the basis of a target sequence that was conserved among four different genera. This assay was then applied to samples from mixed blooms that were occurring in an Australian river, in which both A. circinalis
and C. raciborskii
were reported [8
]. The sxtA
gene was detected in almost all samples, and the copy numbers indicated temporal variations in the amount of STX-producing cells. The qPCR results were consistent with the STX concentrations that were determined by ELISA.
It is evident from the data available in the literature that the use of qPCR to monitor STX-producing cyanobacteria in the environment is scarce. Although the sxtA copy numbers were shown to correlate with the toxin concentration, a greater number of studies is needed to evaluate the applicability of this method to monitor water samples, particularly in samples containing other species besides A. circinalis.
3.4. Multiplex qPCR for Cyanotoxins
Almost all studies that pertain to the molecular detection and quantification of cyanotoxin genes in environmental samples are focused on a single target gene and evaluate the potential toxicity of only one class of toxins. However, one species of cyanobacteria may produce different classes of cyanotoxins, and during cyanobacterial bloom events, a diverse composition of species with different toxic profiles can be found [6
]. If the strength of qPCR is that it serves as a rapid and simple protocol to monitor water supplies for the presence of potentially toxic cyanobacteria, this method should ideally simultaneously detect and quantify the most common cyanotoxins. This motivation led to the development of multiplex qPCR assays.
Multiplex qPCR is an effective solution to save time, samples and costs, but the design of a successful assay depends on the choice of compatible primers and probes and the optimization of reaction conditions to obtain the required sensitivity and specificity for each of the combined targets. This assay includes TaqMan probes that carry different reporters with distinct fluorescent spectra at the 5’ end and a quencher group at the 3’ end.
The first described multiplex qPCR for cyanotoxins targeted four different toxin biosynthesis gene clusters simultaneously, namely microcystin, nodularin, cylindrospermopsin and saxitoxin, as well as the 16S rRNA gene as an internal control [96
]. The selected cyanotoxin targets were mcyE/ndaF
, which were all based on sequences with at least 90% identity among the strains of different cyanobacteria genera. The specificity of the multiplex qPCR was validated by testing 51 toxic and non-toxic cyanobacterial strains. Once the proposed method was shown to be specific, sensitive and reliable, the next step was to apply it to environmental samples.
This strategy was soon used to evaluate the toxigenicity of mixed cyanobacterial blooms occurring in the Murray River, Australia [8
]. A. circinalis
, M. flos-aquae
and C. raciborskii
were detected by microscopic analyses and with samples collected from different sites that varied in their community composition and cyanotoxin profiles. Therefore, this study constituted an ideal situation to test the applicability of the multiplex assay. The toxicity assessment with ELISA revealed low STX concentrations in some samples and no toxin in others, low concentrations of CYL in all samples and no MC. A temporal analysis indicated that a bloom was initiated with a higher proportion of CYN-producing cells that were gradually substituted for STX-producing cyanobacteria. A positive correlation was observed between the toxin gene copy numbers and cyanotoxin concentrations as determined by ELISA. The authors concluded that the consistency between the qPCR data and the toxin concentrations for the three tested toxins supported the applicability of the multiplex assay to bloom risk assessment. This use of a single reaction to detect and quantify biosynthesis genes for the major cyanotoxins found in environmental samples allows for the characterization of toxigenic cyanobacterial assemblages, as well as the monitoring of dynamic changes in toxigenic profiles of complex blooms.
Multiplex qPCR-based monitoring was also performed in a reservoir in Macau, China, where abundant populations of Microcystis
were present and MC and CYL were detected some years before [56
]. A total of 72 water samples were tested. The multiplex assay was designed to quantify the total C. raciborskii
gene), CYN-producers in general (cyrC
gene), total Microcystis
16S rRNA gene) and MC-producers (mcyA/B/C/D/E/G/J
genes). This study reported the successful application of multiplex qPCR in water samples, confirming the value of this method for estimating CYN-associated risk, but also showing that in the case of mcy
genes, the assessment of toxin levels by qPCR may be uncertain.
Although only a few studies have applied multiplex qPCR to environmental samples, the potential use of this method as a tool for monitoring water supplies was anticipated. This approach has many advantages for use in field sample analysis as follows: high sensitivity, broad dynamic range allowing the quantification of very scarce or highly abundant targets, a high throughput capacity, cost-effectiveness and fast results. The disadvantages of multiplex qPCR are related to its laborious optimization steps. These steps are required because of the competition between primers or between targets, cross-oligo interactions and difficulties in quantifying the gene targets in complex samples containing predominantly background DNA [96
]. The determination of absolute 16S rRNA gene copy numbers can also be problematic with qPCR because of the variability in the number of these genetic loci in the genomes of different cyanobacteria species. Nevertheless, the relative quantification of multiple genes can be of special interest when monitoring spatial-temporal changes in the toxigenic composition of complex blooms.
3.5. Advantages of qPCR for Estimating Cyanotoxin Concentrations in the Environment
Although we question the use of qPCR for estimating the toxicity of cyanobacterial blooms, this method is valuable for the study of toxic cyanobacteria population dynamics. Considering that qPCR is a fast, simple and cost-effective method and is easily applicable for the analysis of multiple samples, it is a suitable choice for exploring temporal and spatial variations in the relative abundance of toxic strains, improving our understanding of the dynamics of cyanobacterial blooms and their relations to environmental factors.
It is possible to further improve the simplicity and speed of the assay by implementing protocols that are compatible with crude cell extracts (which is not possible with other analytical methods) and with portable PCR equipment, configuring the assay to a near real-time detection test.
For the assessment of complex bloom samples, qPCR is extremely valuable because many cyanotoxin genes can be investigated simultaneously, while the analysis of different toxins by other analytical methods requires the use of different methodologies (LC-MS/MS or ELISA kits).
In addition, the combination of generalist and specific primers during qPCR permits the identification of the cyanobacteria genus or species responsible for producing the toxin in mixed blooms and to follow variations in their relative contributions over time or space.
Because of its high sensitivity (a detection of less than 102
gene copies per mL), qPCR is advantageous over microscopic examination for detecting toxin-producing cyanobacteria when they are present in minor concentrations. For example, as reported by Lee et al.
] in Lake Vancouver, although the Microcystis
sp. abundance rarely exceeded one percent of the total cyanobacteria and was rarely detected in microscopic counts, the qPCR results indicated that the majority of the Microcystis
population contained the mcyE
gene, and the MC concentrations repeatedly exceeded WHO guidelines for drinking water. In addition, microscopy-based monitoring focuses on biomass increases as indicative of risk, but the MC content can be high even when the cyanobacterial abundance is low (for example, prior to bloom proliferation), and in this case, PCR can detect potentially toxic strains and can be important as an early monitoring tool.
3.6. Limitations of qPCR for Estimating Cyanotoxin Concentrations in the Environment
Since the development of early assays, qPCR was considered a promising tool for monitoring potential cyanotoxin producers in field samples, and the basic premise for its applicability is that a cyanotoxin gene copy number has a positive correlation with cyanotoxin concentrations in the samples.The result depends on accurate measurements of gene contents [47
] and also on the method selected for cyanotoxin analysis [3
]. In qPCR design, absolute quantification of a target gene is based on a standard curve relating known concentrations of a standard DNA to threshold cycle values. The calibrating DNA can be genomic DNA from an isolated strain, a plasmid carrying the target gene or purified amplicons. This curve is then used for extrapolating the concentration of the target sequence in an experimental sample. Absolute quantification based on this calibration assumes that both standard DNA and test samples are amplified with equal efficiency, which may not be true since environmental samples consist of a mix of DNA templates, abundant background DNA and may contain polymerase inhibitors. Depending on the nature of the sample, the extracted DNA can contain contaminants that result in PCR inhibition, affecting reproducibility and sensitivity, leading to false negative results. False negatives can be prevented by including universal targets as internal controls, such as rRNA genes, or by spiking standard targets in amplification reactions. These are possible causes of inconsistencies between cell counts and copy numbers of housekeeping genes. Another cause may be the incomplete recovery of DNA from environmental samples. The efficiency of DNA recovery can be different for different cyanobacterial species and also depends on the extraction method employed.
Quantification can also be adversely affected by copy number variation of target genes or genomes per cell in different species present in environment samples. Toxin genes are present in single copies in the genome of sequenced strains; hence, the quantification of these genes should never outnumber the cell concentration, in principle. However, several studies reported toxin gene numbers that were greater than the cell counts. Possible reasons considered here included errors in cell counting or natural variation in the cellular copy numbers of target genes. In cases where 16S rRNA gene counts are included to estimate the number of cyanobacterial cells, this kind of error is magnified due to copy number variation. The copy number of the 16S rRNA gene varies between one and 15 per genome among bacteria [47
]. Microbial communities are composed of diverse species with different numbers of 16S rRNA genes per genome, and the sequences of these multiple 16S rRNA genes on a genome can be distinct. Thus, these variations hinder the use of this gene to infer the abundance of cells. To overcome these problems, many studies use single copy genetic markers, such as the rpoC1
gene or the cpc
locus. However, these markers are still underrepresented in reference databases as compared to 16S rRNA gene sequences [47
A qPCR assay depends on the choice of primers and probes to target the sequence of interest. First, the design of primers/probes is laborious, mainly for TaqMan assays and for multiplex applications. Primers/probes should ideally recognize target sequences in every toxic strain or species. However, because primer/probe design is based on the alignment of a limited number of reference sequences, it is not known which proportion of toxic cells will fail to be detected because of sequence variability in toxin genes. Thus, the use of generic primers/probes can lead to unequal amplification efficiencies for different species or genera, and this situation can result in the underestimation of toxic cells by qPCR. As the number of sequences in the databases increases, the design of primers/probes should be updated and optimized to improve coverage by detecting new variants of the target gene.
Another recognized problem is the presence of mutated, rearranged or partially deleted versions of toxin biosynthesis clusters in some strains and that the proportion of strains containing non-functional clusters in natural samples is not known. Because the target regions chosen in PCR assays correspond to very short sequences (approximately one hundred base pairs), they can result in an undetermined level of false positives.
An important consideration is that the toxin gene abundance informs the investigator about the potential of the population for toxin production, and analytical methods, such as LC-MS/MS or ELISA, determine the actual toxin concentrations. Thus, incongruent results could be related to intrinsic variations in toxin production by different strains and/or regulated changes in toxin biosynthesis in response to environmental conditions.
Another potential cause for disagreement between LC-MS/MS or ELISA results and qPCR is that the latter is sensitive enough to detect minor populations of potentially toxic cells, which do not produce toxin amounts above the limit of detection of the analytical methods. In addition, some toxin variants can be missed depending on the protocol for LC-MS/MS or the chosen ELISA kit.
The choice of the sample fraction that is used when analyzing toxin concentrations is a highly variable issue in studies dealing with environmental samples, and they vary from intracellular (converted or not to a cell quota), particulate or extracellular material or else the total water sample. This variation makes the comparison of different studies and conclusions about correlations difficult for the quantification of toxin genes and toxin concentrations. Considering MC measurements, as indicated in Table 1
, most studies analyze the particulate fraction, either alone or in combination with the dissolved content. However, in cyanobacterial cells, a large amount of MC is complexed with proteins and is lost in standard methanol extraction [2
]. This is also a possible cause of discrepancy between the quantification of toxin genes and MC toxin concentrations.
The above-mentioned factors are the most commonly-recognized causes of inconsistency between toxin concentrations and qPCR results. It is clear that multiple technical and biological issues contribute to this situation, which limits the use of the molecular assay by itself for risk management at present. The general picture drawn from the literature is that traditional analytical methods and PCR should be combined to assess the presence of cyanotoxins in environmental samples.