OUP user menu

Mapping of picoeucaryotes in marine ecosystems with quantitative PCR of the 18S rRNA gene

Fei Zhu, Ramon Massana, Fabrice Not, Dominique Marie, Daniel Vaulot
DOI: http://dx.doi.org/10.1016/j.femsec.2004.10.006 79-92 First published online: 1 March 2005


A quantitative PCR (QPCR) assay based on the use of SYBR Green I was developed to assess the abundance of specific groups of picoeukaryotes in marine waters. Six primer sets were designed targeting four different taxonomic levels: domain (Eukaryota), division (Chlorophyta), order (Mamiellales) and genus (Bathycoccus, Micromonas, and Ostreococcus). Reaction conditions were optimized for each primer set which was validated in silico, on agarose gels, and by QPCR against a variety of target and non-target cultures. The approach was tested by estimating gene copy numbers for Micromonas, Bathycoccus, and Ostreococcus in seawater samples to which cultured cells were added in various concentrations. QPCR was then used to determine that rRNA gene (rDNA) copy number varied from one to more than 12,000 in 18 strains of phytoplankton. Finally, QPCR was applied to environmental samples from a Mediterranean Sea coastal site and the results were compared to those obtained by Fluorescent in situ hybridization (FISH). The data obtained demonstrate that Chlorophyta and more specifically Mamiellales were important in these waters, especially during the winter picoplankton bloom. The timing of major abundance peaks of the targeted species was similar by QPCR and FISH. When used in conjunction with other techniques such as FISH or gene clone libraries, QPCR appears as very promising to quickly obtain data on the ecological distribution of important phytoplankton groups. Data interpretation must take into account primer specificity and the varying rRNA gene copy number among eukaryotes.

  • Coastal ecosystems
  • Ecology
  • Fluorescent in situ hybridization
  • Picoplankton
  • Prasinophytes
  • Quantitative PCR

1 Introduction

Photosynthetic picoplankton (i.e. the fraction of the plankton composed of cells less than 2–3 μm in size) plays a critical role in marine ecosystems. Although its contribution to photosynthetic biomass (as estimated for example from chlorophyll a) is more important in open ocean oligotrophic waters (typically 80%[1]), it is far from negligible in coastal waters, outside of the bloom season (e.g. up to 87% in the English Channel off Roscoff [2]). While the prokaryotic component of picophytoplankton is reduced to two genera Synechococcus and Prochlorococcus[3], photosynthetic picoeukaryotes are much more diverse and can belong to a variety of divisions (classes): Chlorophyta (Prasinophyceae), Heterokontophyta (Pelagophyceae, Bolidophyceae), Haptophyta (Prymnesiophyceae) etc. The use of molecular methods and, in particular, the analysis of the 18S rRNA and psba genes has allowed to investigate in great details the composition of natural communities [46]. In particular, prasinophytes have been shown very recently to be one of the key picophytoplankton group in marine waters [2]. They are composed of at least seven distinct clades, some of which correspond to established orders such as the Mamiellales or the Prasinococcales, while other do not contain any described species [7]. The importance of this group had already been suspected based for example on the relative abundance of chlorophyll b in some marine waters [8]. In coastal areas, the order Mamiellales that contains in particular the three typical picoplanktonic species Micromonas pusilla[9], Bathycoccus prasinos[10], and Ostreococcus tauri[11], appears specifically important as revealed by culture studies, optical and electron microscopy observations [1214], as well as more recently by 18S rDNA analyses [2,7]. Therefore, it has become critical to assess the abundance of members of this group at different spatial and temporal scales in order to estimate their importance and to determine under which conditions they thrive. Since cells from genera such as Bathycoccus and Ostreococcus are too small to be identified by optical microscopy, the use of antibodies or of molecular markers constitute the only practical approach for their detection. Monoclonal antibodies provide a good choice to detect certain species or ecotypes [15] but they are expansive to produce and they cannot be applied to higher taxonomic levels since their specificity is always narrow. Probes targeted against 18S rRNA are much more flexible since they can be tailored (in theory at least) to target any taxonomic level from the ecotype to the division. Fluorescent in situ hybridization (FISH) with either mono-labeled probes or after amplification [16] can be used to count eukaryotic picoplankton by epifluorescence microscopy [2]. However, this method is quite tedious and, although automation of microscopy analysis [17] or detection by flow cytometry [18] can increase significantly sample throughput, alternative methods are required.

In this context, quantitative PCR (QPCR, see [19] for a review) appears as particularly promising. Its principle consists in amplifying a given gene (in the present case 18S rDNA) with primers specific of one taxonomic group and monitoring product formation in real-time by fluorescence. The later can be induced by a probe labeled with a fluorochrome and a quencher which is released as the probe binds to the product (Taqman approach) or more simply, by a dye, such as SYBR Green I, binding to double stranded DNA as it is formed during the PCR. The number of gene copies in the initial sample is deduced from the number of PCR cycles (cycle threshold or CT) required to cross a certain fluorescence level. The major advantages of this approach are its linearity, its sensitivity, and the speed at which a large number of samples can be processed. Its use in environmental microbiology is rapidly expanding and it has been recently applied to aquatic environments to obtain large scale estimates of the abundance of major bacterial groups [20], of certain important bacterial groups such as Pseudoalteromonas[21], and of specific Synechococcus ecotypes [22].

In the present paper, we introduce six primer sets to be used with QPCR targeting four different eukaryotic taxonomic levels: domain (Eukaryota), division (Chlorophyta), order (Mamiellales) and genus (Bathycoccus, Micromonas, and Ostreococcus). The primers were first tested on cultures, then used to determine the number of rRNA gene copies in a variety of microalgal strains, and finally applied to coastal samples.

2 Materials and methods

2.1 Cultures

A variety of cultured strains were chosen to test primer specificity, to serve as templates for the optimization of QPCR assays, and to estimate the number of rDNA copies per genome (Table 1). Marine photosynthetic strains were grown in K media [23] in tissue culture flasks (Sarstedt, Orsay France) at 20 °C, under a light–dark cycle of 14–10 h, at 100 μEin m−2 s−1 light intensity (except for RCC 341 that was grown at 10 μEin m−2 s−1), and harvested in early stationary growth phase. Cells of Escherichia coli were cultivated in LB media (Sigma, L'Isle d'Abeau Chesnes France) at 37 °C for 12 h.

View this table:

Strains used in this study

DomainClassRCCaSpeciesStrain name
EukaryaPrasinophyceae113Bathycoccus prasinosCCMPb 1898
Prasinophyceae116Ostreococcus tauriOTTH0595
Prasinophyceae614Ostreococcus tauriOTTH0595-Genome
Prasinophyceae114Micromonas pusillaCCMP490
Prasinophyceae450Micromonas pusillaCCMP489
Prasinophyceae434Micromonas pusillaBL122
Prasinophyceae136Prasinococcus capsulatusCCMP1407
Prasinophyceae234Tetraselmis sp.MIN 008-15m B
Chlorophyceae1Chlamydomonas concordiaPLY491
Cryptophyceae20Rhodomonas salinaCCMP322
Prymnesiophyceae362Emiliania huxleyiPROSOPE_115
Bacillariophyceae76Thalassiosira weissflogiiCCMP1336
Bacillariophyceae436Thalassiosira sp.BL_77
BacillariophyceaeNitzschia closteriumROS97005
Eustigmatophyceae92Nannochloropsis salinaCCMP527
Chrysophyceae21Ochromonas distigmaCaen
Pelagophyceae100Pelagomonas calceolataCCMP1214
Pelagophyceae341Pelagomonas calceolataPROSOPE_63
Dictyochophyceae382Mesopedinella arcticaPROSOPE-2
Dinophyceae89Akashiwo sanguinea
Dinophyceae88Amphidinium carteraeCCMP1314
Dinophyceae291Prorocentrum minimum
Dinophyceae303Prorocentrum nuxVillF I 50m
Bacteriaγ-ProteobacteriaEscherichia coliK12
Cyanophyceae27SynechococcusWH 7803
  • aRCC: Roscoff Culture Collection.

  • bCCMP: Provasoli-Guillard National Center for Culture of Marine Phytoplankton.

2.2 Marine samples

Water samples were taken from the Mediterranean Sea off the Spanish coast (Blanes: 41° 40′N, 2° 48′E) from 20 March 2001 to 23 October 2002. Up to 10 L of seawater were collected in sub-surface with a Niskin bottle and transferred through a 200 μm nylon mesh to a black bottle. Samples were pre-filtered through a 3 μm Nuclepore membrane (Whatman International Ltd, Maidstone, England) to separate picoplankton, and subsequently filtered through a Sterivex unit (Millipore, Billerica, MA, USA) with a peristaltic pump. The Sterivex units were directly filled with 1.8 ml lysis buffer (40 mM EDTA, 50 mM Tris–HCl, 0.75 M sucrose) and stored at −80 °C until DNA extraction.

2.3 DNA extraction

For cultured strains, DNA was extracted by the CTAB protocol [24] with some modifications. Cultures were harvested by centrifugation for 10 min at 8000g. 3% (w/v) CTAB (preheated to 60 °C) was added to the pellet and gently swirled to disperse cells. Filter samples were thawed and then treated. Samples were incubated at 60 °C for 30 min in a water bath with occasional vortexing. Nucleic acids were extracted once with an equal volume of chloroform–isoamyl alcohol (48:2). After centrifugation for 10 min at 12,000g, the aqueous phase was transferred to a clean Corex tube and 2/3 volume of cold isopropanol was added to precipitate the nucleic acids at room temperature for several hours. Nucleic acids were recovered by centrifugation, washed once with 70% ethanol (v/v) and re-suspended in water prior to storage at −80 °C. For Mediterranean Sea natural samples, the DNA extraction was performed as described by Massana et al.[25], and genomic DNA extraction of E. coli was performed as described by Sambrook et al.[26].

2.4 Primer design

Eight individual primers (Table 2) were used to target different taxonomic levels within eukaryotes and used as six sets composed of a forward and a reverse primer (Table 3). Design strategy is detailed in Section 3. All primers were optimized using the Primer Express software (PE Biosystems) in order to obtain a theoretical melting temperature around 58–59 °C. The theoretical specificity of the primers (Tables 3 and 4) was verified with the ARB software [27] obtained from http://www.arb-home.de. Probes were tested against a database containing more than 28,000 complete or partial SSU rDNA sequences from both eukaryotes and prokaryotes. We started from the aligned SSU rDNA database released by the ARB team in June 2002 (http://www2.mikro.biologie.tu-muenchen.de/download/ARB/data/ssujun02.arb) and we added more than 1800 new sequences, either publicly available or unpublished.

View this table:

Primers used in this study

NameaSequence (5′- > 3′)Length (bp)GC (%)T m (°C)Minimum number of mismatches to non-target groupsReferencesb
EUK345fAAGGAAGGCAGCAGGCG176559Four mismatches to prokaryotesThis work
EUK499rCACCAGACTTGCCCTCYAAT205558Five or more mismatches to prokaryotesThis work
EUK528fCCGCGGTAATTCCAGCTC186158Five or more mismatches to prokaryotes[53]
CHLO02rCTTCGAGCCCCCAACTTTC195858One central mismatch to sequences of Chlorarachniophyceae, Cercozoa and some Apicomplexa[40]
PRAS04rCAAGCGTAAGCCCGCTTT185658Two central mismatches to some Cryptophyceae[2]
BATHY03rACCACGATGACTCCATGTCTCA225059Five or more mismatches to nontarget groupsThis work
MICRO04rCGCGTCCTCTACAGGAAGTTG215759Perfect match to Micromonas RCC 434. Two mismatches to Micromonas CCMP 489 and 490, Mantionella and Mamiella[2]
OSTRE02rAGTAACCACGGTGACTAAGTGGC235258Four mismatches to Bathycoccus[2]
  • af: Forward, r: Reverse.

  • bReference in which the primer location appeared initially. Primers have been adapted for quantitative PCR constraints.

View this table:

Primer sets used in this study with optimal reaction conditions and linearized plasmid used as a standard

SetTarget groupPrimer forwardPrimer reverseAmplicon size (bp)StandardAmplicon Tm (°C)Primer-dimer Tm (°C)MgCl2 (mM)Annealing time (min)
  • The optimal concentrations for both forward and reverse primers were 400 nM. The annealing-extension and the detection temperatures for all primer sets were 60 and 77 °C, respectively.

View this table:

Specificity of primer sets

nGelC TnGelC TnGelC TnGelC TnGelC TnGelC T
Bathycoccus prasinos0+160+250+190+225+405+34
Micromonas RCC1140+160+220+195+362+235+35
Micromonas RCC4500+180+220+165+352+255+38
Micromonas RCC4340+210+250+175+350+185+40
Ostreococcus tauri0+190+280+205+365+400+22
Prasinococcus capsulatus0+150+275+325+355+405+40
Nannochloris sp.0+160+254405+405+405+40
Rhodomonas salina0+202392+/−255+405+395+40
Emiliania huxleyi0+152355+325+365+405+40
Thalassiosira weissflogii0+192405+355+355+405+40
Nannochloropsis salina0+262305+315+405+405+40
Amphidinium carterae0+182305+355+405+405+40
Ochromonas distigma0+172395+335+405+405+40
Pelagomonas calceolata0+182295+335+405+405+40
Escherichia coli5+405+405+275+405+405+40
Synechococcus sp.5+405+405+285+405+405+40
Maximum CT for target organisms262820221822
Minimum CT for non-target organisms402925353934
  • The six primer sets (Table 3) were tested against a range of cultures (Table 1). For each primer set, primer specificity was tested in silico using the ARB software and the number of mismatches to the 18S rDNA sequence of the species determined (first column, n; 5+: five or more mismatches). The second column (Gel) provides a binary result from conventional PCR based on whether a band was detected by agarose gel electrophoresis (+ means that a band is clearly present, − means that no band is observed and +/− means that a faint band is visible). The third column (CT) indicates the threshold cycle (fluorescence level = 0.3) for quantitative PCR using 1 ng genomic DNA of each organism (see Section 2). The difference between the maximum number of cycles necessary to detect a target organism and the minimum number of cycles to detect a non-target (last two lines) provides an indication of the specificity of the primer set.

2.5 QPCR assays

Reactions were performed in a final volume of 12.5 μl using SYBR PCR core reagents (Applied Biosystems, Courtaboeuf France) and containing 1.25 μl SYBR Green PCR buffer, 200 μM each of dATP, dCTP, and dGTP, 400 μM dUTP, 0.25 U of AmpErase uracyl N-Glycosylase (AmpErase UNG), and 0.05 U of Platinum Taq DNA polymerase (Life Technologies, Cergy Pontoise France). All reactions were performed with optical tubes (Applied Biosystems). First, 2.5 μl of template was delivered into the tubes followed by 10 μl of master mix. The tubes were sealed with optical caps (Applied Biosystems). All reactions were performed with an ABI 5700 sequence detection system (Applied Biosystems) programmed with a soak step of 2 min at 50 °C, allowing AmpErase UNG to hydrolyze PCR amplicons possibly carried over from previous reactions. An enzyme activation step (94 °C, 2 min) followed the initial soak step. Forty cycles of 15 s of denaturation at 94 °C, annealing-extension at 60 °C with the times listed in Table 3, and 25 s of data collection at 77 °C were performed. All data were analyzed using Sequence Detection System v 1.3 software (Applied Biosystems).

2.6 Optimization of QPCR assays

2.6.1 Annealing-extension time

The different primer sets amplified products with different size (Table 3) and therefore required different annealing-extension time to amplify the target sequences. We tested different annealing-extension times for each primer set and chose the minimal annealing-extension time that allowed samples to reach CT in a suitable cycle range (15–35).

2.6.2 Melting curve analysis

Unlike Taqman, SYBR green I binds all double-stranded DNA without specificity. The dye can bind both PCR products and primer-dimers. The latter will affect the accuracy of the results if co-amplified with the PCR products. The dissociation curve from 65 to 95 °C was measured after the last QPCR cycle as detailed in the ABI 5700 software manual and the melting temperature (Tm) of both primer-dimers and specific PCR products was obtained (Table 3). In order to suppress fluorescence readings caused by the generation of primer-dimers, we set the temperature of the detection step above the Tm of primer-dimers but approximately 3 °C below the Tm of the specific PCR product.

2.6.3 MgCl2 and primer concentration

In order to obtain the highest amplification efficiency of QPCR assays, MgCl2 concentrations (ranging from 1 to 9 mM) and primers concentrations (ranging from 200 to 1200 nM) were tested for all primer sets. We chose the minimal concentration of MgCl2 and primers that allowed consistent amplification of the templates at different concentrations and reasonably high amplification efficiencies, estimated from the slopes of the standard curves empirically generated by QPCR assays (Table 3).

2.7 Primer specificity analysis

The specificity of the primer sets was first checked by standard PCR on a iCycler (Bio-Rad, Marnes-la-Coquette, France) using Promega (Madison, USA) PCR reagents. In a final 50 μl volume, reaction mixtures contained 5 μl of buffer, 3 mM of MgCl2, 50 μM each of dATP, dCTP, dGTP, and dTTP, and 400 nM of primers. One nanogram of genomic DNA extracted from the organisms listed in Table 4 was used as template. The following PCR steps were performed: a soak step of 5 min at 94 °C, 35 cycles of 30 s of denaturation at 94 °C, 30 s of annealing with gradient temperature from 55 to 65 °C, and 1 min of extension at 72 °C. Ten microlitres of the reaction product were run in 2% (w/v) agarose gel stained with ethidium bromide. Specificity was further verified by QPCR with 1 ng of template following the optimized protocol defined above. A fixed threshold (0.3) was used for CT computation in the Sequence Detection System software.

2.8 Calibration with plasmids

Three linear plasmids from cloned 18S rDNA for Micromonas RCC 114 (pMIC), Bathycoccus RCC 113 (pBAT), and Ostreococcus RCC 614 (pOST) were constructed to be used as standards for QPCR assays. The full 18S rRNA gene from representative strains for each genus was amplified with universal eukaryotic primers [28] and cloned using the TopoTA cloning kit (Invitrogen, Cergy Pontoise, France), according the instruction of the kit. Plasmids were extracted using Flexiprep kit (Amersham, Orsay, France). Linearized plasmids were produced from supercoiled plasmid by digestion with the restriction endonuclease NotI (New England Biolabs Beverly, MA, USA) according to the manufacturer's protocol, and purified with phenol–chloroform extraction. The concentration of genomic DNA from linear plasmids was measured fluorometrically with SYBR Green I (Molecular Probes, Leiden, The Netherlands) using λ phage DNA as a standard according to the manufacturer's protocol. The number of copies in the standards was calculated using the following formula: Embedded Image where a is the plasmid DNA concentration (g/μl), 5681 is the plasmid length, including the vector (3931 bp) and inserted PCR fragment (average 1750 bp), 660 is the average molecular weight of one base pair, and 6.022 × 1023 is the molar constant.

The pMIC plasmid was used for optimization of reaction conditions and calibration for all primer sets except for BAT and OST, for which pBAT and pOST were used respectively (Table 3). We observed that the use of pMIC, pBAT and pOST for calibration to analyze the same environmental sample with primer set MAM yielded different estimates. In order to be able to compare the concentrations of the different genera estimated by QPCR, we corrected the copy numbers of pBAT and pOST relative to pMIC (see Section 3 for details). This correction only concerned the BAT and OST primer sets.

2.9 Artificial mixture experiment

Cells of M. pusilla (RCC 114), B. prasinos (RCC 113), and O. tauri (RCC 614) were cultivated as described above for six days. Cell counts were obtained with a FACSsort flow cytometer (Becton Dickinson), following the protocol described by Marie et al.[29]. Known numbers of the cells from the three species were added to 1 L seawater samples collected off Roscoff on 1st July 2003. Samples filtration and DNA extraction were performed as described before. Estimates of 18S rRNA gene concentration of M. pusilla, B. prasinos and O. tauri were detected by primer set MIC, BAT, and OST, respectively, using pMIC, pBAT, and pOST as standards, respectively. Concentrations calibrated with the latter two were corrected as explained above.

2.10 Number of rRNA gene copies per genome

The number of rRNA gene copies per genome was estimated in 18 strains. Cells of the selected strains were cultivated for four days and cell numbers were counted by flow cytometry as described above. Two slightly different protocols were used. For a first set of strains, (RCC 1, 76, 92, 113, 114, 116, 362, 434, 436) different volumes of culture (1, 4, 7, 10, 20 ml, respectively) were filtered onto a 47 mm Supor-450 0.45 μm filter (Pall Gelman, New York, USA) and DNA was extracted from each filter using the CTAB protocol as described above. Genomic DNA was resuspended in 100 μl water and diluted 100-fold for QPCR assay. For a second group of strains (RCC 20, 81, 88, 89, 100, 234, 291, 303, 382), a more efficient protocol was used. A single volume of culture (8 ml) was filtered and extracted. Then four dilutions of the extracted DNA (corresponding to 1-, 10-, 100-, 1000-fold dilutions) were used for QPCR. The number of rDNA copies per sample were estimated with the EUK primer set using pMIC as a standard and plotted against the estimated cell number in each DNA sample. The slope of the regression provided an estimate of the number of rDNA gene copies for this strain.

2.11 QPCR analysis of natural samples

Four different plasmid concentrations (from 20 to 20,000 copies μl−1) were used to construct a standard curve. Environmental samples were diluted 10-fold for QPCR assays and run in triplicate for each primer set. 2.5 μl DNA samples (corresponding to 200–1900 pg DNA per reaction tube) were used for each reaction. The original concentration of targeted 18S rDNA (copies ml−1) was computed as follows: Embedded Image where 12.5 is the volume of QPCR reaction solution (μl), c is the 18S rDNA concentration estimated by QPCR (copies μl−1), 2.5 is the volume of the DNA sample in the reaction (μl), 10 corresponds to the sample dilution, b is the volume into which the seawater DNA was resuspended initially (μl), and a is the volume of seawater (ml) from which DNA was extracted.

2.12 FISH analysis of natural samples

FISH was used to determine the cell abundance of specific groups with the following probes: EUK1209R + CHLO01 + NCHLO01 (picoeucaryotes), CHLO02 (Chlorophyta), MICRO01 (Micromonas), BATHY01 (Bathycoccus), and OSTREO01 (Ostreococcus) as described by Not et al.[2].

3 Results

3.1 QPCR primer design, optimization and validation

In this work, we opted for the SYBR Green I approach [19], instead of the TaqMan approach, since the former requires only two primers for each target and therefore relies only on two conserved sites. For all targeted groups except eukaryotes (i.e. Chlorophyta, Mamiellales, and the three genera), we used the same primer on the 5′ side adapted from the internal sequencing primer Euk528f (Table 2). On the 3′ end, we modified slightly existing FISH probes (Table 2). For eukaryotes, no universal primer could be designed downward of Euk528f; therefore we designed two new primers upstream of the region targeted for the other groups (Table 2). In silico analysis of primer specificity (Tables 2 and 4) using the “probe match” function of the ARB software against a large database of SSU rDNA sequences (see Section 2.4), indicated that some primers were highly specific of the group they targeted and had at least four or five mismatches to non-target organisms (OSTRE02r or BATHY03r), while other displayed lower specificity (e.g. CHLO02r). Reaction conditions were then optimized for the six primer sets obtained by varying MgCl2 and primer concentration, annealing time, and temperature for fluorescence reading (Table 3). The optimal MgCl2 concentration fell into a range (3–5 mM) similar to what has been used in previous studies using QPCR to assess bacteria abundance in marine samples [30,31].

For all primer sets, except EUK, the Tm of specific PCR products was 5 °C or more above that of primer dimers, and in excess of 80 °C (Table 3). When the primer sets were used on environmental samples (see below), the melting curves became complex, because of the sequence diversity of the amplicons induced complex melting curves, as observed previously [32]. For the EUK primer set, the difference between the Tm of the primer dimers and that of the PCR products was only 3 °C. The best compromise appeared to perform detection for all primer sets after the annealing-extension step of the cycling program at 77 °C (25 s), although in the case of the EUK primer set, it was impossible to exclude completely primer dimer contribution from the PCR product fluorescence signal.

The next step was to test primer sets on a large number of cultured organisms representing a wide taxonomic range. Non-target cultures displayed from one to more than five mismatches to specific primer sets (Table 4). When tested by conventional PCR, primer sets EUK, CHL, MIC, BAT, and OST were specific of their target groups, without cross-amplification with any of the tested templates (Table 4). Primer set MAM targeting the order Mamiellales (prasinophytes) displayed faint cross-reactivity with one non-target strains, RCC 20 (Rhodomonas salina) which had only two mismatches to PRAS04r.

The use of QPCR allowed us to estimate more precisely the extent of cross-reaction between primer sets and cultured strains (Table 4). For a given primer set, there was considerable response variation for targeted strains. As an example, CT (the first PCR cycle for which fluorescence crosses a pre-determined level) varied from 16 to 26 for eukaryotes with primer set EUK. However, non-target strains, i.e. prokaryotes, displayed a much lower CT (40). The difference between the maximum CT for target strains and the minimum CT for non-target strains gives an idea of the specificity of the primer set (last two lines in Table 4). According to this criterion, primer set EUK and those targeting genera were clearly highly specific, in contrast to primer sets targeting intermediate taxonomic levels. Interestingly, the reverse primer MICRO04r matches perfectly the 18S rDNA sequence of RCC434 but has two mismatches with the two other Micromonas strains (RCC114, RCC450) that belong to different Micromonas clades [7]. This is evidenced by the slightly higher CT observed for the latter two. However, the MIC primer set still allowed to clearly distinguish between Micromonas and the other non-target species tested (Table 4).

3.2 Quantification aspects

In contrast to methods such as FISH that provide unequivocal data in terms of cell abundance, QPCR is inherently a relative method that requires internal references. Although it is possible to use genomic DNA from target organisms, a better choice appears to be linearized plasmids containing cloned 18S rRNA genes [31]. We used three different plasmids (pMIC, Micromonas; pBAT, Bathycoccus; pOST, Ostreococcus). pMIC was used to calibrate all primer sets except BAT and OST, for which we used pBAT and pOST respectively. The gene concentration in the plasmid solution was estimated by spectrofluorometry (see Section 2.8). We observed that identical estimated concentrations of pMIC, pBAT and pOST used with primer sets targeting the three plasmids (MAM and EUK) responded differently. Therefore, gene concentrations estimated based on pBAT (primer set BAT) and pOST (primer set OST) were corrected using pMIC as the reference, i.e. multiplied by 0.32 and 0.56, respectively.

Primer sets MIC, BAT, and OST were used to detect cells of cultured strains added to a natural seawater sample (Fig. 1). The number of estimated 18S rDNA copies was linearly related to the number of cells added to the sample, demonstrating that QPCR could be reliably used in natural samples. It should be noted that the slope of the relationship between gene copies and cell numbers is larger than one, reflecting the fact that eukaryote genomes contain in general more than one copy of the rDNA gene. In order to better interpret field data, it is critical to know how this number varies. This is difficult to achieve by conventional means and QPCR offers a practical solution [33]. We selected 18 algal strains representing 17 species belonging to different phylogenetic groups and ranging in size from 0.8 to 60 μm, for which we estimated the number of 18S rDNA copies using the EUK primer set. Copy number ranged from 1 in the picoplanktonic species Nannochloropsis salina (Eustigmatophyceae) to more than 12,000 for the large dinoflagellate Akashiwo sanguinea and was highly correlated with cell length (Fig. 2).


Cell addition experiment. Known concentration of cultures from Micromonas pusilla, Bathycoccus prasinos and Ostreococcus tauri were added to Roscoff seawater. After DNA collection and extraction, 18S rDNA copy numbers were measured in each sample by QPCR using primer sets MIC, BAT, and OST, respectively. The x axis corresponds to numbers of added cells estimated by flow cytometry and the y axis to 18S rDNA copies measured by QPCR. Straight lines represent regression for each species.


Correlation between rDNA copy number estimated by QPCR and cell length from 18 strains of phytoplankton.

3.3 QPCR analysis of natural samples

The six primer sets were then used to estimate rDNA gene copy numbers in a temporal series of picoplankton (<3 μm) samples from the coastal Mediterranean Sea for which clone libraries data were available [34]. We also performed FISH analyses on the same samples. A clear annual cycle is observed with a picoplankton bloom taking place in late winter. This picoplankton bloom is much longer than typical diatom blooms that last only a couple of weeks [35]. This picoplankton bloom is dominated by Chlorophyta and more specifically by the two Mamiellales genera Micromonas and Bathycoccus (Fig. 3). Average eukaryotes 18S rRNA gene copy numbers from QPCR and cell abundance from FISH fell into very similar ranges. There was a particularly good match between average values for eukaryotes around 7000 cell ml−1 for FISH or copies ml−1 for QPCR (Table 5). A good correspondence was also observed for the contribution of each group to eukaryotes (%) except for Micromonas, for which QPCR provided 4-fold lower estimates of their contribution than FISH. A detailed analysis of the time series (Fig. 3) demonstrated that QPCR estimates of the different groups were very coherent with FISH data. For example, the picoeukaryote bloom occurring in February is clearly visible with QPCR. Three of the major peaks of Chlorophyta contribution (September 2001, April and September 2002) corresponded to peak contribution observed by FISH (Fig. 3(b)). Although for both Mamiellales and Micromonas estimates of QPCR and FISH fell into different ranges, seasonal variation was quite similar for both estimates (Fig. 3(c) and (e)). Finally, for the two primer sets targeting the genera Bathycoccus and Ostreococcus, the timing of the maximal contribution (March 2002 for Bathycoccus and October 2001 for Ostreococcus) were identical for QPCR and FISH (Fig. 3(d) and (f)).


Temporal change of picoplankton at Blanes (Mediterranean Sea) obtained by QPCR and FISH from 20 March 2001 to 23 November 2002. (a) Picoeucaryotes (cell ml−1 for FISH or copies ml−1 for QPCR); (b) Chlorophyta (percentage of picoeukaryotes); (c) Mamiellales (percentage of picoeukaryotes); (d) Bathycoccus (percentage of picoeukaryotes); (e) Micromonas (percentage of picoeukaryotes); (f) Ostreococcus (percentage of picoeukaryotes).

View this table:

Comparison of average picoplankton abundance obtained by FISH and QPCR expressed as percentage of total eukaryotes at a Mediterranean Sea coastal site (Blanes) between 20 March 2001 and 23 October 2002

Eukaryotes (cell ml−1) or (copies ml−1)aChlorophyta (% of euks)Mamiellales (% of euks)Bathycoccus (% of euks)Micromonas (% of euks)Ostreococcus (% of euks)
  • aFISH data are expressed as cell ml−1 and QPCR data as copies ml−1.

4 Discussion

4.1 Primers targeting taxonomic groups

In order to address the ecology of groups such as the prasinophytes, it is important to be able to determine the distribution of different taxonomic entities. At a broad level, one may want, for example, to distinguish between ecosystems dominated by green vs. brown algal classes (e.g. prasinophytes vs. prymnesiophytes), since the groups have deeply divergent phylogenetic histories as well as different physiological responses [36]. However, it is also critical to be able to analyze the distribution of certain genera or species. A typical example will be Micromonas which seems to dominate eukaryotic picoplankton in temperate nutrient-rich waters [2]. At an even finer level, one may want to analyze what are the ecological niches of specific ecotypes: recent physiological, genetic and genomic data on genera such as Prochlorococcus or Ostreococcus demonstrate that many ubiquitous picoplankton species display considerable ecotypic variability [3739]. While it is relatively easy to design probes and primers targeting the domain level or the species level, this is considerably more tricky for intermediate levels such as those of the division (e.g. Chlorophyta) or class (e.g. Prymensiophyceae). Only a few regions in the 18S rRNA gene are suitable for such probes and these probes have only few mismatches (one or two) to many non-target sequences. For example, only two somewhat imperfect FISH probes have been designed for Chlorophyta: CHLO01 and CHLO02 [40,41]. QPCR adds more constraints than FISH since two primers are needed that must be located closely enough since PCR fragments in excess of 400 base pairs result in poor amplification efficiency [19]. Both primers must have nearly identical melting temperature (Tm) around 60 °C. In our specific case, it was impossible to find two specific primers meeting these conditions for some of the target groups. Therefore, one universal primer was used on the 5′ end and specificity was achieved through the reverse primer on the 3′ end. Among the six primer sets used in this study, three appear very satisfactory and highly specific of their target (EUK, BAT, OST), while those targeting intermediary levels (CHL, MAM) have more difficulty to distinguish targets from non-target groups and corresponding data should be evaluated more carefully. Primer set MIC is a bit special since the reverse primer does not match exactly two of the three existing Micromonas clades [7], while in practice it allows a clear distinction between target and non-target organisms (Table 4).

4.2 Quantification issues

With cell-based techniques such as FISH, quantification issues are straightforward since it is easy to determine whether a cell is labeled or not, especially when amplification techniques such as TSA are used [42]. However, with QPCR a number of points must be considered when acquiring and interpreting the data.

QPCR is performed on DNA extracted from samples usually filtered on a porous membrane. The extraction procedure must be performed under very uniform conditions in order to achieve maximum efficiency and recovery. An alternative approach could be to concentrate of microbial cells by centrifugation and to resuspend them in a lysis buffer [43]. However picoplankton can be difficult to centrifuge, although procedures reducing cell loss have been developed recently [18]. With respect to calibration, it would be best to use an internal standard (for example adding a known amount of target cells to the sample, see Fig. 1). However, this is not really feasible in practice when targeting very broad taxonomic levels and working on natural samples since standards would have to be added at sampling time. In our case, we followed the approach of Suzuki et al.[31] using linearized plasmids containing cloned 18S rDNA genes. The advantages of this approach are that standards are easily produced in unlimited amounts and can be generated from genetic library clones for uncultivated taxa, that are very common in picoplankton [4]. Using plasmids (which contain a single copy of the gene) also overcomes the problem of multiple rDNA copies per genomes that would occur if one were using genomic DNA as a standard. One problem we encountered was that plasmids containing genes from three different organisms (Micromonas, Ostreococcus, Bathycoccus) gave different results when used to calibrate data acquired with the same primer set (in a range from about 1–3) probably due to errors in plasmid quantification or maybe PCR efficiency. As a consequence, field data were corrected using the Micromonas plasmid as a reference.

A major complication with PCR compared to cell-based approaches is that the rDNA gene occurs in multiple copies. Although this problem is present with prokaryotes for which copy number varies between 1 (for example in the typical photosynthetic picoplankton genus Prochlorococcus[44]) and 13 (for example in Bacillus cereus[45]), this range is much wider in eukaryotes. For example, Prokopowich et al.[46] reports a range between 35 and 19,300 in animals and between 150 and more than 25,000 in plants. In protists and especially photosynthetic ones, the number of copies can be very small. For example in the unicellular alga Cyanidioschyzon merolae only three copies of the rrn operon are present on different loci with no repetition [47] while in the apicomplexan Cryptosporidium parvum, five copies are found [48]. The range of rDNA copies in 18 strains of microalgae that we estimated by QPCR (Fig. 2) agrees with previous data [46] but is wider since it expands all the way down to one copy for Nannochloropsis. The number of rDNA copies in Ostreococcus, which was used as a control in each PCR plate assayed, was found to vary between 1.7 and 3.4 (Fig. 2), which is coherent with an estimate of four copies arranged in a single cluster based on genome sequencing (Moreau, H., personal communication). Our estimates for nanoplanktonic dinoflagellates are also coherent with values around 1000 estimated for Alexandrium minutum[43]. Large dinoflagellates, such as A. sanguinea, however have rRNA gene copy numbers that reach the highest values observed for plants. The good correlation between rDNA copies and size (Fig. 2) is not surprising. On the one hand, it is well-known that genome size (or DNA content) is related to size (the so-called C-value paradox [49]) as demonstrated for phytoplankton [50]. On the other hand, rDNA copy number appears to be correlated to genome size [46]. The correlation we obtained is somewhat tighter than for plants and animals [46].

For the prasinophyte genera targeted in the present paper, copy numbers were equal to 8.27, 4.09 ± 0.11 (n= 2), 2.7 ± 0.9 (n= 3) for Bathycoccus, Micromonas and Ostreococcus, respectively, while for the other two picoplanktonic species examined, Pelagomonas and Nannochloropsis, it was equal to 1.33 and 0.96, respectively. Therefore, the range for picoplankton is quite small and similar to what is found in prokaryotes. In fact, the good correlation between rDNA gene copies and cell size suggests that QPCR data could be used as a better proxy for biomass than FISH data that only provides cell abundance estimates. Another way to circumvent the problem of multiple rRNA gene copies would be to start from rRNA and to perform a reverse-transcription step prior to QPCR [51], since rRNA levels are representative of cell activity [52].

4.3 Application of QPCR to natural samples

Three caveats must always be kept in mind when interpreting QPCR estimates of natural samples. First, QPCR is exponential by nature and therefore has an inherent precision of 1.4-fold (20.5, assuming that one can detect changes corresponding to half a PCR cycle). Second, as discussed above, the issue of rDNA copy number is also critical. For example, disrupted cells of larger protists or gametes of multicellular organisms passing through the pre-filter used to separate picoplankton (usually 3 μm) could seriously bias estimates, since a single cell would have as much impact as 1000 or 10,000 picoplankton cells. Third, the specificity of the primer sets might not be perfect and therefore, some non-target species may be included in a given estimate.

When applied to an annual series from a coastal site characterized by a strong seasonal cycle, some of the major features of the picoplankton community were very well recovered. Absolute QPCR estimates and FISH abundance agreed very well for eukaryotes (Table 5). This was in fact quite unexpected because of multiple bias possible in DNA extraction and QPCR calibration. The range observed of contributions for the other taxonomic groups also matched well (Table 5). In particular, taxonomic ranking was clearly respected: estimates of QPCR rDNA copy increased logically from genera to order to division and up to domain. Temporal variation was also very well recovered for most of the taxa. FISH and QPCR data agreed extremely well for Bathycoccus with two recurrent peak contribution in March 2001 and February 2002. Similarly, the increased importance of Mamiellales in May and October of 2001 and in February and November of 2002 is also seen in the QPCR data.

However some discrepancies also appeared. The sum of the QPCR estimates for the three genera Ostreococcus, Bathycoccus and Micromonas only amounts to 40% of the estimated Mamiellales contribution (Table 5), suggesting the existence of other Mamiellales genera in the samples. However, no other Mamiellales sequences, besides those of these three genera, were recovered from four clone libraries analyzed at the Blanes site [7]. In the same way, the difference between the Mamiellales and Chlorophyta contributions (that should correspond to Chlorophyta belonging to other prasinophyte orders or to other Chlorophyta classes such as the Trebouxiophyceae) disagrees with the clone library data in which very few Chlorophyta besides the three genera mentioned above have been found [7]. These discrepancies could be explained by the lower specificity of the MAM and CHL primer sets, which may amplify some non-targeted groups and therefore overestimate the target groups. In fact, clone libraries constructed using primers EUK328f and CHLO02r (Viprey, M., personal communication) yielded besides Chlorophyta sequences, other sequences belonging to the Chrorarachniophyceae or even to the alveolates. It should be noted that the former group has only one mismatch with CHLO02r (Table 2). Another discrepancy is that the relative contribution of the Micromonas genus is much lower when estimated by QPCR than with FISH. In this case, primer specificity cannot be suspected, and the explanation is probably to seek in the effect of rDNA copy numbers. Clone libraries from the Blanes sites were rich, particularly, in sequences from uncultivated alveolate groups [34] which may correspond to organisms that have large genomes and therefore a high number of rDNA copies. Their strong contribution to the total eukaryotic rDNA pool would artificially decrease the relative the contribution of genera such as Micromonas.

Once primer and reaction conditions have been optimized, QPCR appears as a very promising technique to map the distribution of plankton taxonomic groups at large temporal and spatial scales, especially because many samples that can be processed in a short time. It will be best applied to monitor specific picoplankton species since the range of picoplankton rRNA gene copy number is quite restricted and it is relatively easy to design primer sets that are highly specific of narrow taxonomic groups. However in more difficult cases, i.e. when targeting broader taxonomic groups or when looking at a wider size range, the application of QPCR may still yield invaluable information that cannot be obtained by any other available technique. For example in the present study, we demonstrated the importance of Mamiellales during the winter Mediterranean Sea plankton bloom, extending the observations recently reported in the English Channel [2].

In conclusion, it should be emphasized that we will only gain a deep understanding of marine ecosystems by combining as many techniques as possible and QPCR should not be used alone but in conjunction with other molecular biology approaches such as, FISH, clone libraries, or DGGE, as well as with more classical tools such as pigment analyses.


The work was supported by the following programs: PICODIV (EVK3-CT-1999-00021), BIOSOPE (CNRS INSUE), PICMANCH (Région Bretagne), GenoMer (Région Bretagne et Département du Finistère). FZ and FN benefited from fellowships from the French Ministry of Education and Research.


  1. [1]
  2. [2]
  3. [3]
  4. [4]
  5. [5]
  6. [6]
  7. [7]
  8. [8]
  9. [9]
  10. [10]
  11. [11]
  12. [12]
  13. [13]
  14. [14]
  15. [15]
  16. [16]
  17. [17]
  18. [18]
  19. [19]
  20. [20]
  21. [21]
  22. [22]
  23. [23]
  24. [24]
  25. [25]
  26. [26]
  27. [27]
  28. [28]
  29. [29]
  30. [30]
  31. [31]
  32. [32]
  33. [33]
  34. [34]
  35. [35]
  36. [36]
  37. [37]
  38. [38]
  39. [39]
  40. [40]
  41. [41]
  42. [42]
  43. [43]
  44. [44]
  45. [45]
  46. [46]
  47. [47]
  48. [48]
  49. [49]
  50. [50]
  51. [51]
  52. [52]
  53. [53]
View Abstract