Introduction

Analytical techniques targeting 16S rDNA or functional genes were widely used for microbial quantification. These include hybridization-based techniques, such as membrane hybridization (Raskin et al. 1995) and fluorescence in situ hybridization (FISH) (Okabe et al. 1999) as well as polymerase chain reaction (PCR)-based techniques, such as denaturing gradient gel electrophoresis (DGGE) (Muyzer et al. 1993) and cloning-sequencing (Zhang and Fang 2001; Zhang et al. 2003a). Hybridization methods, which have detection limits in the order of 105 DNA/RNA copies or greater, are in general less sensitive. They can thus only be used for environmental samples of relatively high microbial concentrations. PCR-based methods, on the other hand, are capable of detecting DNA/RNA at low concentrations. However, the precision of PCR-based methods may be compromised due to a number of factors. These may include reagent depletion, competition of amplicons with primers, and the loss of polymerase activity as the number of amplification cycle increases (Schneegurt and Kulpa 1998). To overcome such deficiency, a new technique known as quantitative real-time PCR (qRT-PCR) has emerged for the detection and quantification of microorganisms at low concentrations.

The qRT-PCR method monitors the amount of PCR products of DNA as they are amplified in real time (Higuchi et al. 1993; Heid et al. 1996; Gibson et al. 1996). From the change of PCR product concentration throughout the amplification cycles, the initial concentration of the target DNA/RNA can be estimated. The first qRT-PCR instrument was commercialized in 1997 (DeFrancesco 2003) soon after the concept was demonstrated in 1993 (Higuchi et al. 1993). Since then, qRT-PCR was widely applied in medical research (Klein 2002; Ginzinger 2002). This article reviews the principle of qRT-PCR, fluorescent reagents and quantification procedures, and its applications for the quantification of microorganisms in environmental samples.

Principle

During PCR, the target DNA sequence is amplified over a number of denaturation–annealing–extension cycles. In a conventional PCR, only the final concentration of the amplicon may be monitored using a DNA-binding fluorescent dye. However, in the qRT-PCR, the concentration of the amplicon is monitored throughout the amplification cycles using a group of new fluorescent reagents. These reagents bind with the amplicon without causing damage at the end of each cycle so that amplification may continue to proceed. The fluorescence intensity emitted during this process reflects the amplicon concentration in real time.

Figure 1a are the plots of fluorescence intensity throughout the amplification cycles in whole cell PCR, using a series of five standard solutions as examples. These solutions contained Microcystis aeruginosa PCC 7820 at varied concentrations from 4.6×103 to 4.6×107 cells/ml. The threshold cycle C t is defined as the cycle at which the fluorescence intensity crosses over a level where the amplification enters a logarithmic growth phase. C t is inversely proportional to the log value of the initial concentration of the target, as illustrated in Fig. 1b for M. aeruginosa PCC 7820. Based on the standard curve like Fig. 1b, concentration of the target DNA in an environmental sample can be estimated from its C t measurement in qRT-PCR.

Fig. 1
figure 1

Quantification of Microcystis aeruginosa PCC 7820 using qRT-PCR. a Relative fluorescence intensity of five standard solutions of PCC 7820 throughout amplification cycles where Ct represents the threshold cycle number. b The standard curve for qRT-PCR measurement of PCC 7820

Full size image

Figure 1a also shows that as the number of PCR cycle further increases, the amplification efficiency reduces gradually. This is due to a number of factors including the depletion of reagents, the inhibition caused by the increased product concentration, and the gradual loss of polymerase activity. Thus, amplicon concentration and the corresponding fluorescence intensity beyond a certain number of amplification cycles cannot be used to estimate the initial concentration of the target DNA (Schneegurt and Kulpa 1998). Figure 1a shows that the fluorescent intensities of the five standard solutions after 50 cycles were nearly the same despite the differences of their initial concentrations.

Detection reagents

Fluorescent reagents used for qRT-PCR can be classified into three categories: (1) dyes that bind with double-stranded DNA (dsDNA) (Hernández et al. 2004); (2) DNA-sequence specific probes, including TaqMan probe (Haugland et al. 1999), molecular beacon (Briones and Raskin 2003), and dual hybridization probe (Glazer and Mathies 1997; Reina et al. 2005); and (3) DNA sequence-specific primers including Amplifluor primer (Hernández et al. 2004), scorpion primer (Taveau et al. 2002), Light Upon eXtension (LUX) primer (Donia and Pana 2005), and universal template (Zhang et al. 2003b). Among these fluorescent reagents, SYBR Green I, TaqMan probe, and molecular beacon, are the most widely used in qRT-PCR applications of environmental samples.

dsDNA-binding dyes

During the extension phase in the PCR cycle, dsDNA are synthesized from the denatured single-stranded DNA. The fluorescence intensity of a dsDNA-binding dye increases as a result when the dye binds with the increased amount of amplified dsDNA. Such an increase can be monitored in real time (Morrison et al. 1998). For qRT-PCR applications, the fluorescence intensity of the dye has to be substantially increased when it binds with the dsDNA from its autofluorescence intensity. For example, the fluorescent intensities of ethidium bromide, i.e., the first dsDNA-binding dye used for qRT-PCR (Higuchi et al. 1993), increase 25 times after binding with dsDNA; whereas the intensity of SYBR Green I, i.e., the most commonly used dsDNA-binding dye (Nadkarni et al. 2002; Stubner 2002; López-Gutiérrez et al. 2004), increases 200 times. Other dsDNA-binding dyes used include BEBO (Bengtsson et al. 2003) and Thiazole Orange (Benveniste et al. 1996).

There are three most notable advantages of the dsDNA-binding dye: (1) it is simple to use without the complicated requirement for the design of a probe; (2) it can be used in conjunction with any PCR primer set; and (3) it is less costly compared to hybridization probes, such as TaqMan probe and molecular beacon. This technique, on the other hand, has its limitations (Sharkey et al. 2004) resulting from: (1) its inability to discriminate different dsDNA segments because the dye binds with all dsDNA indiscriminately; (2) the formation of primer dimmers, which may affect its detection sensitivity; and (3) the mistaken amplification of nontarget DNA segments. Thus, an accurate primer design and the optimization of the PCR conditions are crucial when DNA-binding dyes are used.

Table 1 contains the list of primers used in SYBR Green I qRT-PCR for environmental samples. It should be noted that most primer pairs might produce unspecific products when applied to complex environmental samples of high microbial diversity. This may partly invalidate applications of qRT-PCR using SYBR Green I as detection reagent. Thus, an additional melting curve analysis is often required to ensure the quality of the amplified product (Zhang and Fang 2005; Bischoff et al. 2005).

Table 1 Primers and probes used in qRT-PCR related to environmental microbiology
Full size table

TaqMan probe

The TaqMan probe is a linear oligonucleotide with a 5′ end fluorophore (called reporter) and a 3′ end quencher. The fluorescence emitted by the reporter is greatly dimmed by the nearby quencher, which disperses energy as heat rather than fluorescence via a fluorescence resonance energy transfer (FRET) mechanism. Figure 2 illustrates the role of TaqMan probe in a qRT-PCR amplification cycle. After DNA denaturation, TaqMan probe is hybridized with the target DNA segment prior to annealing. During the subsequent extension phase, Taq DNA polymerase hydrolyzes the TaqMan probe bound on the target DNA strand, cleaving nucleotides from the probe and releasing the reporter into the solution. This results in a drastic increase of reporter’s fluorescence intensity due to its disconnection with the quencher on the probe. As PCR proceeds, the fluorescence intensity increases proportionally with the amount of the amplicons (Holland et al. 1991; Heid et al. 1996).

Fig. 2
figure 2

The role of TaqMan probe in a qRT-PCR amplification cycle (reporter ○, quencher □)

Full size image

Reporter fluorophores commonly used include FAM (6-carboxyfluoroscein), TET (tetrachloro-6-carboxyfluorescein), JOE (2,7-dimethoxy-4,5-dichloro-6-carboxyfluorescein), VIC, Cy3, Cy5, Texas Red, and HEX (hexacholoro-6-carboxyfluorescein). There are a number of quenchers available, the most common of which being TAMRA (6-carboxytetramethylrhodamine). Two new quenchers were recently introduced to replace TAMRA: minor groove binding (MGB) quencher and black hole quencher (BHQ). They have lower fluorescence background, higher sensitivity, higher hybridization stability, better resolution of fluorescence spectrum, longer storage life, plus higher reproducibility and specificity (Wilson et al. 2005). However, it should be noted that TaqMan technique, as compared to SYBR Green I, is more costly, more difficult in probe design, and limited to the detection of shorter PCR product, normally less than 150 bps.

Molecular beacon

Like the TaqMan probe, molecular beacon is also a fluorescent oligonucleotide with a 5′ end reporter and a 3′ end quencher. However, unlike TaqMan probe, molecular beacon has a special “hairpin” structure as illustrated in Fig. 3. As a result, the fluorescence emitted by the reporter is greatly suppressed by the quencher due to their close proximity (Tyagi and Kramer 1996). Upon hybridization with the target denatured DNA segment during annealing, the reporter’s fluorescence intensity increases as its distance from the quencher increases. The fluorescence intensity thus becomes the indicator of the target DNA concentration. As temperature increases during the subsequent extension, molecular beacon detaches from the DNA segment, retaining its hairpin structure, and can rebind to the target DNA segment in the next cycle.

Fig. 3
figure 3

The role of molecular beacon in a qRT-PCR amplification cycle (reporter ○, quencher □)

Full size image

In addition, molecular beacon has a very high specificity, i.e., it requires a perfect hybridization with the target due to the high thermal stability of its hairpin structure. It is able to discriminate DNA sequences that differ by a single nucleotide substitution. It is thus suitable for mutation analysis and for single nucleotide polymorphism detection. On the other hand, design of a molecular beacon and its detection conditions are very demanding.

Molecular beacon has been used for gene expression (Taveau et al. 2002), single-nucleotide polymorphism detection (Mhlanga and Malmberg 2001), and pathogen detection (Belanger et al. 2003). Its application to environmental sample analysis is still very limited (Harms et al. 2003). Similar to TaqMan probe, molecular beacon is costly and often used for the detection of short PCR product. However, due to its complex configuration and required binding property, molecular beacon is more difficult to design than the TaqMan probe.

Target DNA sequences

16S rDNA and functional genes are the most common target DNA sequences for microbial quantification using qRT-PCR. Although other DNA sequences, such as 5S rDNA, 23S rDNA, and 16S-23S rDNA interspacer region can also be used as targets, their applications are very limited so far.

16S rDNA

16S rDNA contains conserved and variable regions according to their genetic stability. Based on variations of DNA sequences in these regions, organisms can be classified into different taxonomy levels. Organisms within a domain often share the same DNA sequence in the most conserved region, whereas species of the same genus may be discriminated from DNA sequence in the variable regions (Woese et al. 1990; Skovhus et al. 2004). qRT-PCR probes and primers may be designed either basing on the more than 150,000 16S rDNA sequences available in the genetic databases, such as GenBank, RDP, and ARB, or by modifying the 1,200 primers and FISH probes available in the probeBase (http://www.microbial-ecology.net/probebase/).

Functional genes

Functional genes (Wagner and Loy 2002) shared by microorganisms of similar physiological characteristics can also be used as target in qRT-PCR. Currently, more than 14,000 DNA sequences are known for more than 100 functional genes. There are many functional genes that may be used for environmental applications. These include those of nitrification/denitrification (Purkhold et al. 2000; López-Gutiérrez et al. 2004; Okano et al. 2004), nitrogen fixation (Gruntzig et al. 2001; Qiu et al. 2004), carbon dioxide fixation (Wu et al. 2001), polymer degradation (Wu et al. 2001), dissimilatory sulfate reduction (Wagner et al. 1998), methane oxidation (Maria et al. 2002), methane production (Hales et al. 1996), organic contaminant degradation (Rhee et al. 2004), metals reduction (Wu et al. 2001), and algal toxin production (Foulds et al. 2002; Rinta-Kanto et al. 2005).

Quantification

Standard curve

For a qRT-PCR analysis, a standard curve like Fig. 1b is needed for each target 16S rDNA or gene. Such a curve may be constructed from serial dilutions of a given source, which could either be cells, which are extracted genomic DNA of a pure culture or plasmids with the target DNA insert, or PCR-amplified DNA segments. Cells in the serial standard solutions can be quantified by plate counting or microscopic counting. Genomic DNA, plasmids, and PCR product may be quantified by spectrophotometry at 260 nm or by kits using dsDNA-binding dyes such as PicoGreen.

The precision of microbial quantification using qRT-PCR relies on the assumption that the environmental sample and the standard solutions share the same PCR efficiency. It is thus crucial to check the PCR efficiency in the analysis of both standard solutions and environmental samples (Devers et al. 2004). For a PCR process,

$$N_{n} = N_{o} \times {\left( {{1 + \eta } \mathord{\left/ {\vphantom {{1 + \eta } {100\% }}} \right. \kern-\nulldelimiterspace} {100\% }} \right)}^{n}$$

where Nn is number of amplified target at the end of nth cycle of amplification, No the initial number of target, and η the PCR efficiency (Abell and Bowman 2005). At 100% efficiency, two DNA segments are produced from a target segment in each PCR cycle. The value of threshold cycle Ct can then be expressed as:

$$C_{t} = {{\left( {\log N_{t} - \log N_{o} } \right)}} \mathord{\left/ {\vphantom {{{\left( {\log N_{o} - \log N_{t} } \right)}} {\log {\left( {{1 + \eta } \mathord{\left/ {\vphantom {{1 + \eta } {100\% }}} \right. \kern-\nulldelimiterspace} {100\% }} \right)}}}} \right. \kern-\nulldelimiterspace} {\log {\left( {{1 + \eta } \mathord{\left/ {\vphantom {{1 + \eta } {100\% }}} \right. \kern-\nulldelimiterspace} {100\% }} \right)}}$$

where N t is the number of amplified target after the threshold cycle C t. The slope (S) of a standard curve of C t vs log N o, as illustrated in Fig. 1b, equals to \({ - 1} \mathord{\left/ {\vphantom {{ - 1} {\log {\left( {{1 + \eta } \mathord{\left/ {\vphantom {{1 + \eta } {100\% }}} \right. \kern-\nulldelimiterspace} {100\% }} \right)}}}} \right. \kern-\nulldelimiterspace} {\log {\left( {{1 + \eta } \mathord{\left/ {\vphantom {{1 + \eta } {100\% }}} \right. \kern-\nulldelimiterspace} {100\% }} \right)}.}\) The slope of an ideal PCR amplification is thus equal to −3.32 (i.e., −1/log 2). In practice, a reliable standard curve should have a R 2 value of more than 0.95 and a slope between −3.0 and −3.9 corresponding to PCR efficiencies of 80–115%. The standard cure in Fig. 1b shows a slope of −3.41 corresponding to an efficiency of 96% and a R 2 of 0.997.

Detection limit, dynamic range, and factors affecting quantification

The dynamic range refers to the range in which the target quantity is linear with C t in the standard curve (van der Velden et al. 2003). Probes, primers, and PCR conditions should be optimized not just for a low detection limit but also for a broad dynamic range (Panicker et al. 2004; Ponchel et al. 2003). Under the most optimum conditions, SYBR Green I and TaqMan probe may have a detection limit as low as two cells per PCR sample and a dynamic range of 107 (Bassler et al. 1995: Chen et al. 1997).

The precision of qRT-PCR analysis may be affected by PCR inhibitors present in the environmental sample. Ideally, each sample should be serially diluted and tested to determine PCR efficiency to see if there are any PCR inhibitors present (Stubner 2002; Harms et al. 2003). The effect of the PCR inhibitor may be reduced by conducting the test at higher degrees of dilution (Stubner 2002; Harms et al. 2003). Careful handling and storage of sample, primers, probes, and enzymes are crucial to the accuracy of microbial quantification. In addition, the DNA extraction efficiency, which may vary from batch to batch, has a significant effect on the reliability of qRT-PCR (Dionisi et al. 2003).

Applications in environmental samples analysis

Early qRT-PCR applications were mainly for the detection and quantification of pathogens, such as Salmonella spp. (Chen et al. 1997), toxigenic Escherichia coli (Oberst et al. 1998), Stachybotrys chartarum (Haugland et al. 1999), Ehrlichia spp. (Leutenegger et al. 1999), Listeria monocytogenes (Bassler et al. 1995; Nogva et al. 2000), and Vibrio cholerae (Lyon 2001). More recently, applications of qRT-PCR were extended to environmental samples’ analysis as summarized in Table 1.

Eubacteria and Archaea

As shown in Table 1, a universal PCR primer set and a universal probe were designed for prokaryote, i.e., Eubacteria and Archaea (Takai and Horikoshi 2000). In addition, four probes/primers were designed for Eubacteria (Stubner and Meuser 2000; Dionisi et al. 2003; Takai and Horikoshi 2000; Suzuki et al. 2000) and two for Archaea (Takai and Horikoshi 2000; Suzuki et al. 2000). Nadkarni et al. (2002) showed that the microbial quantity in an anaerobic sample measured by qRT-PCR using a universal probe was 40-fold greater than that measured by the conventional culture methods.

Subgroups in Archaea

In addition to the qRT-PCR primer/probe set for the whole domain of Archaea, several primer/probe sets were developed, which are specific for different Archaea subgroups (Table 1). Using these sets, Sawayama et al. (2006) showed that Methanosarcina spp. and Methanobacterium spp. immobilized in the polyurethane foam in a reactor were 7.6×109 and 2.6×108 DNA-segment/ml, respectively, about 1,000 times higher than those in the original anaerobically digested sludge.

Methanotrophic bacteria

Methane oxidation in the environment is mostly accomplished by methanotrophic bacteria. As shown in Table 1, primer sets were developed to target the pmoA gene, which encodes the α subunit of particulate methane monooxygenase of four methanotroph groups: Methylococcus, Methylobacter/Methylosarcina, Methylosinus, and Methylocapsa (Kolb et al. 2003). Detection limits were between 10 and 100 cells per PCR sample. qRT-PCR analysis of soil samples spiked with these four bacteria recovered only 20% of Methylosinus, but almost all the added cells of the other three groups.

Nitrifying bacteria

Ammonium oxidation by autotrophic ammonia-oxidizing bacteria (AOB) is a key process in agricultural/natural ecosystems and wastewater treatment (Jordan et al. 2005). Using qRT-PCR based on 16S rDNA, the AOB concentration in a fertilized soil was found as 6.2×107 cells/g, three times higher than that in the unfertilized soil (Hermansson and Lindgren 2001). AOB population in soil may also be quantified by qRT-PCR targeting the ammonia-monooxygenase gene (amoA) (Okano et al. 2004) with a detection limit of 1.3×105 cells/g.

qRT-PCR was also used to monitor the nitrifying bacteria population dynamics in wastewater treatment plants (Hall et al. 2002; Harms et al. 2003; Limpiyakorna et al. 2005). In 12 activated sludge samples collected over a year, the mean concentrations were 4.3±2.0×1011 cells/l for Eubacteria, 3.7±3.2×1010 cells/l for Nitrospira, 1.2±0.9×1010 cells/l for all AOB, and 7.5±6.0×109 cells/l for N. oligotropha-like AOB (Harms et al. 2003). Ammonia-oxidizing rates were estimated as 7.7–12.4×10−12 mol/(h·cell), comparable to the 0.5–25.0×10−12 mol/(h·cell) found in soil (Okano et al. 2004). The qRT-PCR standard curves for Nitrospira 16S rDNA, AOB 16S rDNA, and N. oligotropha-like amoA were linear over six orders of magnitude and that of Eubacteria 16S rDNA was linear over four orders of magnitude.

Denitrifying bacteria

Denitrification is performed by phylogenetically diverse groups of bacteria. It is thus difficult to design a universal primer/probe set for denitrifiers based on 16S rDNA. Recently, qRT-PCR was developed to quantify denitrifier based on functional genes, such as gene narG encoding the α subunit of the membrane-bound nitrate reductase (López-Gutiérrez et al. 2004) and gene nirK encoding nitrite reductase (Sonia et al. 2004). The qRT-PCR standard curves were linear over seven orders of magnitude and with a sensitivity of 100 DNA target segments per PCR sample. The results also showed that narG quantity ranged between 5.08×108 and 1.12×1011 DNA-segment/g-soil (López-Gutiérrez et al. 2004), far higher than the nirK quantity of 9.7×104 to 3.9×106 DNA-segment/g-soil (Sonia et al. 2004).

Sulfate-reducing bacteria

Sulfate-reducing bacteria (SRB) is a group of polyphyletic microorganisms. Due to the well-established phylogenetic database of 16S rDNA, many qRT-PCR primer/probe sets specific for different classification levels (down to species) are available for SRB, as shown in Table 1. The qRT-PCR analysis for gram-positive SRB Desulfotomaculum lineage 1 had the detection limit of 100 DNA target segment per PCR sample, equivalent to 106 DNA-segments/g-soil (Stubner 2002), while the detection limits for gram-negative SRB were between 2×105 and 4×103 DNA-segments/g-soil (Stubner 2004). In addition, qRT-PCR primer set for dissimilatory sulfite reductase gene was also developed (Tang et al. 2004).

Cyanobacteria

Microcystis is one of the most common cyanobacteria that may cause serious algae bloom. As shown in Table 1, quantification of Microcystis was accomplished by the qRT-PCR utilizing primer/probe sets specific for Microcystis 16S rDNA, and microcystin synthetase genes of mcyB (Nonneman and Zimba 2002) and mcyD (Foulds et al. 2002; Rinta-Kanto et al. 2005). Results showed that Microcystis concentration in the bloom area varied from 2×103 to 4×108 cells/l (Rinta-Kanto et al. 2005). The quantification result was valuable in the identification of the blooming sources (Rinta-Kanto et al. 2005).

Pollutant degrader

A methyl tert-butyl ether (MTBE) degrading bacteria PM1 was detected using the TaqMan qRT-PCR. The detection limit was 2,000 cells/l in pure culture or 1.80×105 cells/l in a mixture of PM1 and Escherichia coli. The quantification results indicated that increases in PM1 concentration corresponded to the rate of MTBE removal (Hristova et al. 2001). qRT-PCR analyses for hydrocarbon-degrader bssA gene, encoding benzylsuccinate synthase (Beller et al. 2002), and trichloroethylene-degrader mmoC gene, encoding a soluble methane monooxygenase (Kikuchi et al. 2002), had the same detection limit of five DNA target segments per PCR sample. The dynamic ranges of qRT-PCR analysis were over seven orders of magnitude for the former gene and five for the latter.