Flow Cytometric and Microscopic Analysis of the Effect of Tannic Acid on Plant Nuclei and Estimation of DNA Content

Abstract

• Background and Aims Flow cytometry (FCM) is extensively used to estimate DNA ploidy and genome size in plants. In order to determine nuclear DNA content, nuclei in suspension are stained by a DNA-specific fluorochrome and fluorescence emission is quantified. Recent studies have shown that cytosolic compounds may interfere with binding of fluorochromes to DNA, leading to flawed data. Tannic acid, a common phenolic compound, may be responsible for some of the stoichiometric errors, especially in woody plants. In this study, the effect of tannic acid on estimation of nuclear DNA content was evaluated in Pisum sativum and Zea mays, which were chosen as model species.

• Methods Nuclear suspensions were prepared from P. sativum leaf tissue using four different lysis buffers (Galbraith's, LB01, Otto's and Tris.MgCl₂). The suspensions were treated with tannic acid (TA) at 13 different initial concentrations ranging from 0·25 to 3·50 mg mL⁻¹. After propidium iodide (PI) staining, samples were analysed using FCM. In addition to the measurement of nuclei fluorescence, light scatter properties were assessed. Subsequently, a single TA concentration was chosen for each buffer and the effect of incubation time was assessed. Similar analyses were performed on liquid suspensions of P. sativum and Z. mays nuclei that were isolated, treated and analysed simultaneously. FCM analyses were accompanied by microscopic observations of nuclei suspensions.

• Key Results TA affected PI fluorescence and light scatter properties of plant nuclei, regardless of the isolation buffer used. The least pronounced effects of TA were observed in Tris.MgCl₂ buffer. Samples obtained using Galbraith's and LB01 buffers were the most affected by this compound. A newly described ‘tannic acid effect’ occurred immediately after the addition of the compound. With the exception of Otto's buffer, nuclei of P. sativum and Z. mays were affected differently, with pea nuclei exhibiting a greater decrease in fluorescence intensity.

• Conclusions A negative effect of a secondary metabolite, TA, on estimation of nuclear DNA content is described and recommendations for minimizing the effect of cytosolic compounds are presented. Alteration in light scattering properties of isolated nuclei can be used as an indicator of the presence of TA, which may cause stoichiometric errors in nuclei staining using a DNA intercalator, PI.

INTRODUCTION

Flow cytometry (FCM) is a powerful technique that was originally developed to count blood cells (Shapiro, 2004). With improvements to equipment and methodologies, FCM has been adapted to many areas of biology including plant sciences (Doležel, 1997). Since the introduction of FCM to plant studies, estimation of DNA ploidy levels and determination of nuclear genome size have been the two most frequent applications (Bennett and Leitch, 2005). Owing to its ease, rapidity and accuracy, FCM has been an attractive alternative to traditional methods such as Feulgen microspectrophotometry (Doležel and Bartoš, 2005). When analysing nuclear genome size, many authors have noted its variation within species (Price and Johnston, 1996a; Rayburn et al., 1997, 2004; Ellul et al., 2002). However, the occurrence of this phenomenon and its extent remains a matter of discussion, as contradictory results have been obtained by different authors when analysing the same materials (Greilhuber and Obermayer, 1997; Obermayer and Greilhuber, 1999; Price et al., 2000; Suda, 2004).

Reliable estimates of nuclear DNA content require proportionality between the digitized fluorescence signal and DNA content. This depends on several factors such as stoichiometry of dye binding to DNA, accessibility of DNA to the fluorochrome, fluorescence absorption and linearity of the instrument amplification system (Bagwell et al., 1989). The accessibility of nuclear DNA to fluorochromes has recently been a topic of major concern as it was found that cytosolic compounds could interfere with fluorescent staining of nuclei in suspension (Noirot et al., 2000, 2002, 2003, 2005; Price et al., 2000; Pinto et al., 2004; Walker et al., 2006). These observations indicated that FCM can produce flawed data if the effect of cytosolic compounds is ignored.

Price et al. (2000) speculated that inhibitors that decrease dye fluorescence of nuclei were common in plants. The authors did not point to a specific compound but suggested involvement of one or more of the numerous secondary metabolites. Noirot et al. (2000), working with coffee trees, revealed negative effects of cytosol on accessibility of DNA to propidium iodide (PI), and showed that cytosolic compounds could bias nuclear DNA content estimates by up to 20 %. More recently, Noirot et al. (2003) identified two compounds that influenced PI fluorescence of petunia nuclei: caffeine and chlorogenic acid (a precursor of polyphenols). Whereas caffeine increased PI accessibility to petunia DNA, chlorogenic acid significantly decreased petunia nuclei fluorescence.

Tannic acid (TA) is a common phenolic compound, frequently accumulated in various tissues of plants belonging to diverse taxonomic groups, especially woody species. Several authors have claimed that TA might be responsible for stoichiometric errors in genome size estimations: Greilhuber (1986) showed that in DNA cytophotometry, tannins could interfere with the Feulgen reaction to an extent that makes it worthless as a quantitative method; using FCM, Favre and Brown (1996) and Zoldoš et al. (1998) encountered difficulties in measuring nuclear DNA content in Quercus species and suggested that they were due to the presence of TA. In addition, Loureiro et al. (2005) speculated that tannins were responsible for higher DNA content estimates obtained for ex vitro leaves of Quercus suber as compared with in vitro material.

The main objectives of this study were: (1) to test the effect of TA on PI fluorescence and light scattering properties of nuclei of Pisum sativum and Zea mays; (2) to evaluate the effect of different nuclear isolation buffers on the interaction of TA with nuclei in suspension; (3) to evaluate the kinetics of the interaction of TA with nuclei samples; (4) to test if nuclei of two different species, P. sativum and Z. mays, are affected similarly by TA when isolated, processed and analysed simultaneously; (5) to identify diagnostic patterns that can be used to recognize the action of cytosolic compounds; and (6) to propose recommendations on how to minimize the negative effects of cytosolic compounds on estimation of nuclear DNA content using FCM.

MATERIAL AND METHODS

Plant material

Plants of garden pea (Pisum sativum L.) ‘Ctirad’ (Fabaceae) and maize (Zea mays L.) ‘CE777’ (Poaceae) were grown from seeds in a greenhouse at 22 ± 2 °C, with a photoperiod of 16 h and a light intensity of 530 ± 2 µmol m^–2 s⁻¹.

Sample preparation

Nuclei suspensions were obtained after chopping approximately 350 mg of leaf tissue according to Galbraith et al. (1983). Four of the most popular nuclear isolation buffers (Loureiro et al., 2006) were used to prepare samples: Galbraith's buffer (Galbraith et al., 1983), LB01 (Doležel et al., 1989), Otto's buffers (Doležel and Göhde, 1995) and Tris.MgCl₂ (Pfosser et al., 1995). β-Mercaptoethanol was not included in LB01, as this study intended to compare basic buffer formulas without any additives that could modulate the action of tannic acid. Otto I and Otto II buffers were mixed in a 1 : 2 ratio. Nuclear suspension was filtered through a 50-µm nylon filter and RNase A (Fluka, Buchs, Switzerland) at a concentration of 50 µg mL⁻¹ was added to each sample. Samples were kept on ice until analysis.

Staining of nuclei

The effect of PI concentration and staining time was analysed in order to determine a saturating concentration to be used in subsequent experiments. Four millilitres of nuclear suspension was prepared from approximately 350 mg of pea leaves. The homogenate was divided into eight aliquots (0·5 mL each) and PI (Fluka) was added to achieve the following concentrations: 10, 25, 50, 75, 100, 150, 250 and 350 µg mL⁻¹. Samples were analysed 0, 5, 10 and 60 min after incubation with PI. Each sample was measured for exactly 2 min. In all further experiments, nuclei were stained with 150 µg mL⁻¹ of PI (optimal concentration).

Flow cytometry measurements

Nuclear samples were analysed using a Coulter Epics XL (Coulter Electronics, Hialeah, FL, USA) flow cytometer. The instrument was equipped with an air-cooled argon-ion laser tuned at 15 mW and operating at 488 nm. PI fluorescence was collected through a 645-nm dichroic long-pass filter and a 620-nm band-pass filter. Prior to analysis the instrument was checked for linearity with Flow-Check fluorospheres (Beckman-Coulter, Hialeah, FL, USA). The amplifier system was set to a constant voltage and gain throughout the experiments. The results were acquired using the SYSTEM II software version 3.0 (Coulter Electronics) in the form of six graphics: fluorescence pulse integral (FL); FL vs. fluorescence pulse height (FLPH); forward scatter (FS) vs. side scatter (SS), both in logarithmic (log) scales; SS in log scale vs. FL; FL vs. time; and SS in log scale vs. time. FS is proportional to particle size. However, owing to its response similarity with SS, it was not considered during data evaluation. The coefficients of variation (CV) were calculated for SS (SS-CV,%) and FL (FL-CV,%).

Fluorescence microscopy

Nuclear suspensions were evaluated in a Nikon Eclipse 80i fluorescence microscope (Nikon Corporation Nikon Instech Co., Kanagawa, Japan) using the G-2A filter cube. Digital photographs were taken using a Leica DC 200 digital camera (Leica Microsystems AG, Wetzlar, Germany).

The effect of tannic acid on P. sativum

Effect of different concentrations

Seven millilitres of nuclear suspension was divided into 14 aliquots (0·5 mL each), which were treated with TA (Fluka) made in H₂O at: 0·0 (control), 0·25, 0·30, 0·40, 0·50, 0·75, 1·00, 1·25, 1·50, 1·75, 2·00, 2·25, 2·50 and 3·50 mg mL⁻¹. A 0·25-mL aliquot of TA solution was added to each sample (final concentrations: 0·0, 0·083, 0·100, 0·133, 0·167, 0·250, 0·333, 0·417, 0·500, 0·583, 0·667, 0·750, 0·833 and 1·167 mg mL⁻¹) and left to incubate for 15 min on ice. The sample was then stained with PI for 5 min on ice and analysed for 2 min. The experiment was replicated three times for each buffer.

Effect of incubation time

For each nuclear isolation buffer, a different concentration of TA was chosen based on the results of previous experiments. The nuclear suspension (4 mL) was divided into four 1-mL aliquots. Three TA treatments were applied (TA1, TA2 and TA3): TA1 and TA2 treatments consisted of adding 0·5 mL of TA solution to the sample. In the TA1 treatment, three 0·5-mL aliquots were stained with PI for 5 min and analysed immediately. In TA2, samples were incubated with TA for 15 min prior to staining with PI. TA1 and TA2 samples were analysed for a period of 2 min. In TA3, three 0·33-mL aliquots were taken, stained by PI for 5 min and analysed for 60 s. Measurement was then halted and 0·16 mL of TA solution was added to the sample. The measurement was resumed in order to follow the kinetics of the addition of TA. The order in which the samples were measured was randomized within treatments to avoid a systematic error. This experiment was replicated twice for each buffer on different days.

Simultaneous analysis of P. sativum and Z. mays nuclei

This experiment was designed to determine if nuclei from two different species are affected by TA in the same way. Two experiments were carried out exactly as the two experiments above, except that instead of using 350 mg of pea leaf tissue, 175 mg of Z. mays leaves and 175 mg of P. sativum leaves were chopped simultaneously in a Petri dish. In the second experiment, a particular concentration of the TA solution was chosen for each buffer.

Statistical analyses

Statistical analyses were performed using a two-way ANOVA (SigmaStat for Windows 3.1, SPSS, USA). When treatments were significantly different, a Holm–Sidak multiple comparison test was used for pair-wise comparison. To calculate the optimal PI concentration, logarithmic and hyperbolic regression analyses were performed (GraphSight v.2.0.1, Belarus). A one-way ANOVA was performed to analyse differences among concentrations of PI and incubation times with this fluorochrome.

RESULTS

The effect of PI concentration and staining time

The effect of PI concentration on nuclear fluorescence was computed for each staining period and for each buffer. The logarithmic fitting, despite not being perfect, provided better regression coefficients than other methods, such as hyperbolic fitting (data not shown). These results show that the addition of 150 µg mL⁻¹ of PI was saturating (Fig. 1). The increase of nuclear fluorescence was not statistically significant at higher PI concentrations. For practical convenience, a 5-min staining period was chosen (Fig. 2). Statistically significant differences were observed only for the 60-min incubation period with PI in nuclei isolated with Tris.MgCl₂ buffer. In this case, nuclei fluorescence decreased.

Fig. 1.

Effect of propidium iodide (PI) concentration (μg mL⁻¹) on fluorescence intensity (mean channel number) of Pisum sativum nuclei isolated with four different nuclear isolation buffers: LB01 (squares), Tris.MgCl₂ (crosses), Galbraith's (diamonds), Otto's (triangles). The logarithmic fitting with the respective coefficient of regression of the values obtained with Otto's buffer are also presented (dashed line).

Fig. 2.

Effect of incubation time (min) with propidium iodide (PI) on fluorescence intensity (mean channel number) of Pisum sativum nuclei isolated with four different nuclear isolation buffers: LB01 (squares), Tris.MgCl₂ (crosses), Galbraith's (diamonds), Otto's (triangles). The asterisk indicates a statistically significant difference between incubation times according to the multiple comparison Holm–Sidak test at P ≤ 0·01.

The effect of TA on P. sativum nuclei

Effect of different concentrations

A detailed description of the action of TA will focus on nuclei isolated with Tris.MgCl₂—a buffer on which nuclei were least affected (Table 1, Fig. 3). In the absence of TA in nuclei suspension, relatively low SS-CV (30·00–40·00 %) and FL-CV (<2·50 %) values were obtained (Table 1). Up to a TA concentration of 0·50 mg mL⁻¹, no significant effect on SS, SS-CV, FL or FL-CV was detected. At 0·50 mg mL⁻¹ TA, an increase in SS and SS-CV was observed. At a concentration of 0·75 mg mL⁻¹ and in particular at 1·00 mg mL⁻¹, the presence of two new populations of particles was detected on the SS vs. FL cytogram (arrows on Fig. 3F, I). These particles were characterized by high SS and SS-CV values (Table 1). As will be shown later, similar populations of particles characterized by high SS and SS-CV were also observed with other buffers albeit at different TA concentrations (Table 2). Hereafter, this phenomenon is termed the ‘tannic acid effect’.

Table 1.

Analysis of SS, SS-CV, FL, FL-CV (mean and standard deviation of the mean) of particles in liquid suspensions prepared from Pisum sativum leaf tissue with four nuclear isolation buffers (Galbraith's, LB01, Otto's and Tris.MgCl₂) and exposed to different concentrations of tannic acid (TA)

	Galbraith's buffer				LB01 buffer
TA (mg mL−1)	SS (channel)	SS-CV (%)	FL (channel)	FL-CV (%)	SS (channel)	SS-CV (%)	FL (channel)	FL-CV (%)
Control	4·61 ± 0·290a	40·71 ± 1·887a	211·0 ± 3·81a	2·88 ± 0·476a	8·43 ± 0·580a	37·66 ± 7·955a	210·6 ± 4·60a	2·33 ± 0·103a
0·25	10·06 ± 6·739a	92·91 ± 39·858ab	210·0 ± 8·95a	3·01 ± 0·842a	23·63 ± 15·519a	91·18 ± 24·770b	215·3 ± 4·07a	2·64 ± 0·327a
0·30	26·53 ± 9·151a	138·62 ± 1·146bc	205·7 ± 10·81a	3·20 ± 0·635a	44·40 ± 8·879ab	120·13 ± 4·246b	217·3 ± 1·22ab	3·71 ± 0·544ab
0·40	59·50 ± 28·705b	149·55 ± 16·200c	207·5 ± 10·75a	3·98 ± 1·631ab	71·03 ± 14·490b	116·72 ± 1·991b	226·8 ± 5·17b	3·78 ± 0·887ab
0·50	67·63 ± 10·504b	131·30 ± 9·893c	206·1 ± 11·43a	4·71 ± 1·206ab	87·43 ± 20·144b	106·13 ± 3·482b	228·1 ± 5·10b	4·98 ± 0·725b
0·75	78·87 ± 8·411b	126·09 ± 1·209c	186·7 ± 18·76a	5·52 ± 1·959ab	141·37 ± 46·012b	101·56 ± 3·583b	227·9 ± 3·35b	4·34 ± 0·872b
1·00	86·70 ± 1·311b	121·59 ± 7·796c	150·8 ± 30·78b	8·27 ± 4·880b	160·35 ± 27·648b	98·06 ± 3·246b	227·6 ± 0·28b	4·80 ± 0·757b
1·25	No pea nuclei peaks were distinguishable				No pea nuclei peaks were distinguishable
1·50	–	–	–	–	–	–	–	–
1·75	–	–	–	–	–	–	–	–
2·00	–	–	–	–	–	–	–	–
2·25	–	–	–	–	–	–	–	–
2·50	–	–	–	–	–	–	–	–
3·50	–	–	–	–	–	–	–	–
	Otto's buffer				Tris.MgCl2 buffer
TA (mg mL−1)	SS (channel)	SS-CV (%)	FL (channel)	FL-CV (%)	SS (channel)	SS-CV (%)	FL (channel)	FL-CV (%)
Control	7·36 ± 0·318a	35·41 ± 3·764a	208·1 ± 9·41a	1·48 ± 0·046a	13·33 ± 0·961a	34·29 ± 2·402a	194·0 ± 9·37acd	2·21 ± 0·854a
0·25	8·33 ± 0·401ab	34·99 ± 3·168a	201·7 ± 13·57a	1·50 ± 0·097a	12·97 ± 1·021a	34·96 ± 2·330a	211·9 ± 10·87abc	2·41 ± 0·435a
0·30	9·19 ± 1·098ab	39·97 ± 6·820a	198·5 ± 16·06a	2·17 ± 0·583a	13·07 ± 0·945a	35·51 ± 3·115a	213·4 ± 8·24abc	2·38 ± 0·297a
0·40	10·70 ± 0·100ab	44·35 ± 6·376a	200·6 ± 9·11a	1·73 ± 0·104a	13·70 ± 1·709ab	36·75 ± 2·627a	214·2 ± 5·54ab	2·47 ± 0·340a
0·50	14·27 ± 1·464ab	53·03 ± 10·260a	197·7 ± 8·46a	1·91 ± 0·166a	22·23 ± 15·667ab	57·92 ± 38·559ab	214·9 ± 12·33ab	2·46 ± 0·269a
0·75	59·23 ± 18·451bc	103·35 ± 11·866bc	194·1 ± 6·72a	2·79 ± 0·630a	27·00 ± 15·186ab	75·17 ± 27·019ab	216·1 ± 13·14ab	3·01 ± 0·252a
1·00	84·37 ± 34·480bc	93·94 ± 22·399bc	185·4 ± 10·79ab	4·92 ± 1·038b	32·70 ± 9·009ab	91·91 ± 11·383b	213·9 ± 6·02ab	3·15 ± 0·577a
1·25	108·13 ± 34·872bc	75·35 ± 24·578ca	163·9 ± 18·77b	5·51 ± 0·631b	45·70 ± 16·838b	91·11 ± 11·383b	228·8 ± 16·30b	3·24 ± 1·52a
1·50	No pea nuclei peaks were distinguishable				30·70 ± 7·808ab	89·89 ± 15·428b	203·8 ± 7·53ab	3·05 ± 0·520a
1·75	–	–	–	–	34·63 ± 10·571ab	95·95 ± 15·298b	195·2 ± 15·37a	3·34 ± 1·09a
2·00	–	–	–	–	34·97 ± 10·702ab	91·62 ± 14·44b	191·1 ± 2·14a	3·35 ± 0·46a
2·25	–	–	–	–	29·07 ± 6·521ab	91·14 ± 18·766b	181·6 ± 3·75c	3·28 ± 0·921a
2·50	–	–	–	–	29·33 ± 6·300ab	82·81 ± 18·860b	168·6 ± 11·40cd	3·45 ± 0·681a
3·50	–	–	–	–	No pea nuclei peaks were distinguishable

SS and FL are shown as channel numbers and SS-CV and FL-CV in per cent. The incipient concentrations for the TA effect are highlighted in bold type. Means followed by the same letter are not significantly different at P ≤ 0·05 according to the Holm–Sidak multiple comparison test.

Fig. 3.

The effect of tannic acid (TA) on propidium iodide (PI) fluorescence and light scattering properties of Pisum sativum nuclei in suspension. Nuclear suspensions were obtained with Tris.MgCl₂ buffer and incubated for 15 min with TA at: 0·00 mg mL⁻¹ (A–C), 0·75 mg mL⁻¹ (D–F), 1·00 mg mL⁻¹ (G–I), 1·75 mg mL⁻¹ (J–L) and 2·50 mg mL⁻¹ (M–O). The results are displayed as forward scatter (logarithmic scale, FS log) vs. side scatter (logarithmic scale, SS log) cytograms (A, D, G, J and M), PI fluorescence intensity histograms (B, E, H, K and N) and SS log vs. PI cytograms (C, F, I, L and O). Arrows indicate new populations that appeared after the addition of particular TA concentrations.

Table 2.

Concentration of tannic acid (TA, mg mL⁻¹) at which the TA effect (significant change in SS and SS-CV) occurs, at which an increase or decrease in fluorescence intensity is observed and at which no nuclei peaks can be discriminated from the debris background

Buffer	Analysis of P. sativum			Simultaneous analysis of Z. mays and P. sativum
	SS and SS-CV	FL (loss, −; gain, +)	No peaks visible	SS and SS-CV	FL (loss, −; gain, +)	FL ratio	No peaks visible
Galbraith's	0·25–0·30	0·75 (–)	1·25	0·50–0·75	0·75 (–)	0·50–0·75	1·25
LB01	0·25–0·30	0·40 (+)	1·25	0·30	0·40 (+)	0·40	0·75
Otto's	0·75	1·00 (–)	1·50	1·00	0·30 (–)	n.s.	1·25
Tris.MgCl2	1·00	2·25 (–)	3·50	1·75	2·25 (–)	1·75	3·50

n.s., non-significant changes were observed at P ≤ 0·05.

At up to 0·75 mg mL⁻¹ TA, no clear effect on FL was observed. In two of three replications, the presence of nuclei with slightly higher fluorescence was detected. Microscopic analyses of samples revealed a tendency to precipitation, and in addition to single nuclei (as observed at lower concentrations of TA; Fig. 4A, B) some nuclei were observed to which weakly fluorescent particles were attached (Fig. 4C, D). This apparently resulted in higher FL and consequently higher FL-CV values (3·00–3·25 %). At the same time, a second subpopulation of particles with higher SS but lower FL started to appear. Microscopic observation of samples revealed the presence of a new subpopulation consisting of clumps of weakly fluorescent particles that were not attached to nuclei, thus having relatively high optical complexity and lower FL than nuclei (Fig. 4C, D).

Fig. 4.

Micrographs of nuclei suspensions prepared with Tris.MgCl₂ buffer. (A) Bright-field image of control without tannic acid (TA, scale bar = 20 μm). Magnification ×400. (B) Fluorescence image of control without TA (scale bar = 20 μm). Magnification ×400. (C) Bright-field image after addition of TA (scale bar = 10 μm). Magnification ×1000. (D) Fluorescence image after addition of TA (scale bar = 20 μm). Magnification ×400. n, nuclei; c, chloroplasts; up, unspecific particles. In contrast to control samples, clumps of various particles (up) can be observed after the addition of TA. Fluorescence images were over-exposed to observe particles with lower fluorescence.

With increasing concentrations of TA (1·25, 1·50, 1·75 and 2·00 mg mL⁻¹), the ‘tannic acid effect’ became more pronounced as the number of events present in the new subpopulations (characterized by high SS and SS-CV) increased continuously and the proportion of ‘clean’ nuclei declined (Fig. 3 J, M). No significant decrease in fluorescence of PI-stained nuclei was observed up to a TA concentration of 1·75 mg mL⁻¹ (Fig. 3K). Only at 2·00 mg mL⁻¹ TA did a significant decrease of nuclei FL of about 15·0 % occur. At this point, the proportion of ‘clean’ nuclei in the suspension was low (<10·0 %). At concentrations of 2·25 and 2·50 mg mL⁻¹ TA, the fraction of ‘clean’ nuclei in the suspension continued to decrease, and it was difficult to discriminate nuclei peaks from the background debris. This was clearly due to the presence of the two new subpopulations of particles (Fig. 3 M–O). Fluorescence of the remaining ‘clean’ nuclei was considerably lower compared with controls. In samples treated with the highest concentration of TA (3·50 mg mL⁻¹), no peaks representing pea nuclei could be discriminated from the background.

Other nuclear isolation buffers

When compared with the Tris.MgCl₂ buffer, the remaining three buffers that were used to isolate P. sativum nuclei were more affected by TA, and lower concentrations of this compound were sufficient to induce the TA effect (Tables 1 and 2).

In Galbraith's buffer, 0·25 mg.mL⁻¹ TA caused a doubling of SS values with SS-CV of 92·0 %, a value observed with the Tris.MgCl₂ buffer only at 0·75 mg mL⁻¹ TA. As in Tris.MgCl₂, nuclei fluorescence was only affected at higher concentrations of TA (0·75 and 1·00 mg mL⁻¹). At those concentrations, FL of nuclei was lower by 11·5 and 28·5 %, respectively, when compared with the untreated control, and FL-CV values were above 5·0 %. For a TA concentration of 1·25 mg mL⁻¹, peaks of pea nuclei could not be distinguished from the background (Table 1).

The use of LB01 buffer gave results similar to those obtained with Galbraith's buffer, as nuclei were affected with the addition of the same concentration of TA (0·25 mg mL⁻¹). In addition, no peaks were visible after the addition of 1·25 mg mL⁻¹ TA. Nevertheless and contrarily to what was observed for all the other buffers, nuclear fluorescence increased with increasing TA concentration. A significant fluorescence increase was obtained after the addition of 0·30 mg mL⁻¹ TA. At 0·50 mg mL⁻¹ TA, a 7·7 % difference, as compared with control, was observed (Table 1).

Nuclear suspension in Otto's buffer was less affected by TA than in Galbraith's and LB01 buffers, but more than in Tris.MgCl₂. The TA effect was observed at a concentration of 0·75 mg mL⁻¹, as in the Tris.MgCl₂ buffer. However, in contrast to what was observed with other three buffers, PI fluorescence decreased even with the addition of TA at the lowest concentration. Nevertheless, a statistically significant decrease of fluorescence was observed only at 1·00 mg mL⁻¹ TA. At this concentration the FL-CV increased significantly, reaching values of 5·0 %. At a TA concentration of 1·25 mg mL⁻¹, nuclei fluorescence decreased by 21·0 %, and this was the highest TA concentration at which nuclei peaks could be distinguished (Table 1).

Effect of incubation time

Following the concentration tests performed with each buffer, a particular concentration of TA was chosen (Table 3) to test whether incubation time was of importance. In Galbraith's and LB01 buffers the time of incubation had no marked effect; in both TA1 and TA2 experiments, the TA effect occurred, and extremely high SS and SS-CV values were obtained. Although statistically significant differences between TA1 and TA2 were observed in SS and SS-CV, in practice, nuclei from both experiments were severely affected by the addition of TA (Table 3). The FL-CV value was also higher in these treatments compared with the control value. As expected, and similarly to the results obtained in the concentration tests, nuclei fluorescence decreased significantly after the addition of 0·50 mg mL⁻¹ TA in Galbraith's buffer. By contrast, in LB01, fluorescence increased significantly with the addition of the same TA concentration. In the TA3 experiment, an immediate increase in SS and SS-CV was observed in both buffers when adding TA after running the sample for 60 s.

Table 3.

The effect of incubation time with tannic acid (TA) on plant nuclei

	Galbraith's buffer (ci = 0·50 mg mL−1)				LB01 buffer (ci = 0·50 mg mL−1)
Experiment	SS (channel)	SS-CV (%)	FL (channel)	FL-CV (%)	SS (channel)	SS-CV (%)	FL (channel)	FL-CV (%)
Control	4.34 ± 0·677a	57·53 ± 17·587a	201·3 ± 7·94a	2·36 ± 0·210a	6·25 ± 3·275a	37·79 ± 3·440a	220·1 ± 5·86a	2·68 ± 0·420a
TA1	39·25 ± 17·514b	149·36 ± 12·592b	191·5 ± 5·85b	4·27 ± 2·048b	55·60 ± 27·70b	127·20 ± 9·12b	235·5 ± 5·96b	4·03 ± 1·050b
TA2	49·67 ± 22·554c	131·95 ± 9·263c	196·9 ± 3·07c	3·77 ± 1·149a	61·92 ± 31·867c	113·67 ± 4·166c	233·0 ± 2·24b	4·60 ± 0·907b
TA3bta	8·51 ± 4·468a	65·53 ± 0·963a	199·8 ± 10·92a	2·60 ± 0·626a	4·59 ± 2·257a	48·93 ± 2·017a	225·6 ± 6·84a	2·55 ± 0·428a
TA3ata	30·14 ± 33·660b	103·08 ± 32·271b	188·8 ± 6·33a	3·24 ± 1·210a	40·15 ± 21·116b	133·82 ± 16·520b	243·5 ± 3·94b	3·49 ± 1·601b

	Otto's buffer (ci = 0·75 mg mL−1)				Tris.MgCl2 buffer (ci = 1·00 mg mL−1)
Experiment	SS (channel)	SS-CV (%)	FL (channel)	FL-CV (%)	SS (channel)	SS-CV (%)	FL (channel)	FL-CV (%)
Control	2·55 ± 0·244a	34·30 ± 2·521a	220·3 ± 3·53a	1·64 ± 0·201a	14·77 ± 0·462a	31·80 ± 1·744a	217·5 ± 6·60a	2·43 ± 0·067a
TA1	4·44 ± 0·181b	43·01 ± 8·187b	169·0 ± 6·89b	2·23 ± 0·253b	14·27 ± 0·289a	31·33 ± 0·652a	194·2 ± 3·47b	3·11 ± 0·247b
TA2	43·81 ± 8·187c	60·06 ± 3·926c	171·3 ± 6·66b	2·30 ± 0·184b	35·17 ± 7·051b	94·16 ± 5·179b	221·7 ± 4·68a	3·18 ± 0·217b
TA3bta	2·12 ± 0·222a	45·59 ± 5·720a	224·1 ± 2·83a	1·79 ± 0·271a	13·80 ± 0·265a	31·70 ± 0·721a	226·1 ± 3·44a	2·48 ± 0·127a
TA3ata	9·99 ± 6·372b	81·70 ± 23·239b	179·8 ± 6·63b	2·25 ± 0·356b	16·17 ± 0·603b	30·70 ± 0·722a	228·5 ± 0·42a	2·42 ± 0·095a

SS, SS-CV, FL and FL-CV (mean and standard deviation of the mean) were analysed in liquid suspensions prepared from P. sativum leaf tissue with four nuclear isolation buffers (Galbraith's, LB01, Otto's and Tris.MgCl₂), exposed to TA at a particular initial concentration (c_i), and incubated for different periods of time (control—no incubation with TA; TA1—incubation with TA for 5 min; TA2—incubation for 20 min with TA). In TA3, samples were run for 60 s without TA (TA3bta) and, after a pause, TA was added to samples (TA3ata). SS and FL are shown as channel means and SS-CV and FL-CV in per cent. Means followed by the same letter are not significantly different according to the Holm–Sidak multiple comparison test at P ≤ 0·05.

In Otto's buffer, the time of incubation was an important factor, as statistically significant increases of SS and SS-CV were observed in TA2 as compared with TA1. Interestingly, a decrease in fluorescence was observed in both experiments. The change in SS, SS-CV and FL occurred immediately after the addition of TA.

In Tris.MgCl₂, a similar result to that obtained with Otto's buffer was verified, i.e. the TA effect was observed only after 15 min of incubation. Compared with data for control nuclei, SS remained the same but FL was lower by 10·0 % in the TA1 experiment. This was not the case for TA2, in which both SS and SS-CV increased significantly but the fluorescence intensity of nuclei was similar to the values obtained in the control. No significant differences were detected in any SS properties, FL and FL-CV in TA3.

Simultaneous analysis of P. sativum and Z. mays nuclei

To test if two different species were affected in the same way by TA, nuclei from Zea mays and Pisum sativum were isolated, treated and analysed simultaneously. In order to compare the response of the two species, three ratios were computed: SS ratio (Z. mays mean SS/P. sativum mean SS), SS-CV ratio (Z. mays mean SS-CV/P. sativum mean SS-CV) and FL ratio (Z. mays mean FL/P. sativum mean FL).

Effect of different concentrations

In Galbraith's buffer, the addition of 0·40 mg mL⁻¹ TA increased the Z. mays SS-CV. The increase was only statistically significant at 0·75 mg mL⁻¹ TA. With increasing concentrations of TA, the SS and SS-CV from both species increased similarly and the FL ratio remained constant until a TA concentration of 0·75 mg mL⁻¹. Thereafter, a statistically significant increase in the FL ratio was observed. This was due to a more pronounced decrease of fluorescence observed in P. sativum nuclei than in Z. mays. As in the single species analysis, no nuclei peaks could be observed after addition of 1·00 mg mL⁻¹ TA (Table 4).

Table 4.

Analysis of Zea mays SS-CV (%), SS ratio, SS-CV ratio and FL ratio (mean and standard deviation of the mean) of particles in liquid suspension prepared from Pisum sativum and Z. mays leaf tissue with four different buffers (Galbraith's, LB01, Otto's and Tris.MgCl₂) and exposed to different concentrations of tannic acid (TA)

TA (mg	Galbraith's buffer							LB01 buffer							Otto's buffer							Tris.MgCl2 buffer
	mL−1)	Z. mays SS-CV (%)	SS ratio	SS-CV ratio	FL ratio	Z. mays SS-CV (%)	SS ratio	SS-CV ratio	FL ratio	Z. mays SS-CV (%)	SS ratio	SS-CV ratio	FL ratio	Z. mays SS-CV (%)	SS ratio	SS-CV ratio	FL ratio
Control	37·86 ± 9·359a	0·808 ± 0·1742a	1·002 ± 0·3772a	0·600 ± 0·007a	37·78 ± 6·578a	0·748 ± 0·0830ab	0·953 ± 0·0868a	0·613 ± 0·002a	41·00 ± 5·735ab	0·572 ± 0·1237a	1·041 ± 0·2013a	0·598 ± 0·0098a	38·66 ± 0·714a	0·766 ± 0·1347a	1·096 ± 0·1054a	0·602 ± 0·0039a
0·25	40·17 ± 8·892a	0·806 ± 0·0950a	1·020 ± 0·2176a	0·600 ± 0·011a	59·40 ± 22·558a	0·732 ± 0·1080a	1·004 ± 0·1468a	0·614 ± 0·003a	39·32 ± 5·645a	0·573 ± 0·1144a	1·108 ± 0·2313a	0·593 ± 0·0045a	39·17 ± 0·623a	0·812 ± 0·1763ab	1·020 ± 0·1098a	0·608 ± 0·0062ab
0·30	43·01 ± 11·288a	0·835 ± 0·1544a	1·106 ± 0·2255a	0·604 ± 0·010a	101·64 ± 25·449b	0·852 ± 0·2038ab	1·162 ± 0·0915a	0·618 ± 0·001a	38·73 ± 2·475a	0·754 ± 0·3479a	0·958 ± 0·0589a	0·592 ± 0·0012a	37·59 ± 2·209a	0·797 ± 0·1545a	0·988 ± 0·0993a	0·604 ± 0·0043a
0·40	70·56 ± 47·319a	0·869 ± 0·1943a	1·084 ± 0·1857a	0·605 ± 0·006a	132·74 ± 6·432b	0·922 ± 0·0479bc	1·156 ± 0·1418b	0·633 ± 0·003b	40·95 ± 5·327ab	0·594 ± 0·0657a	1·118 ± 0·2333a	0·591 ± 0·0036a	38·07 ± 2·425a	0·776 ± 0·1280a	0·997 ± 0·1007a	0·609 ± 0·0058ab
0·50	75·37 ± 58·957a	0·810 ± 0·1538a	1·032 ± 0·3321a	0·607 ± 0·009ab	114·46 ± 6·174b	1·070 ± 0·1294c	1·190 ± 0·1335b	0·646 ± 0·009b	43·59 ± 5·069ab	0·642 ± 0·0272a	1·099 ± 0·1008a	0·594 ± 0·0020a	38·44 ± 2·615a	0·777 ± 0·1208a	1·027 ± 0·0178a	0·609 ± 0·0034ab
0·75	149·57 ± 14·284b	0·724 ± 0·0697a	1·089 ± 0·0858a	0·618 ± 0·006b	No peaks were distinguishable							53·79 ± 15·875ab	0·628 ± 0·1287a	0·874 ± 0·1634a	0·592 ± 0·0064a	38·57 ± 2·116a	0·772 ± 0·1207a	1·001 ± 0·0525a	0·612 ± 0·0039ab
1·00	145·91b	0·955a	1·112a	0·659c	–	–	–	–	66·17 ± 7·353b	0·894 ± 0·0613a	1·000 ± 0·0337a	0·600 ± 0·0252a	39·82 ± 1·575a	0·759 ± 0·1212a	1·075 ± 0·1378a	0·612 ± 0·0012ab
1·25	No peaks were distinguishable							–	–	–	–	No peaks were distinguishable							43·04 ± 9·322a	0·753 ± 0·1144a	1·060 ± 0·0373a	0·614 ± 0·0033ab
1·50	–	–	–	–	–	–	–	–	–	–	–	–	55·00 ± 27·740ac	0·792 ± 0·1758a	1·128 ± 0·0627a	0·616 ± 0·0021ab
1·75	–	–	–	–	–	–	–	–	–	–	–	–	92·86 ± 17·295bc	0·887 ± 0·1224bc	1·164 ± 0·1466a	0·622 ± 0·0051b
2·00	–	–	–	–	–	–	–	–	–	–	–	–	91·790 ± 5·388bc	0·828 ± 0·0464bc	1·141 ± 0·0145a	0·615 ± 0·0086ab
2·25	–	–	–	–	–	–	–	–	–	–	–	–	88·21c	0·888c	1·168a	0·614ab
2·50	–	–	–	–	–	–	–	–	–	–	–	–	82·46c	0·865bc	1·120a	0·612ab
3·50	–	–	–	–	–	–	–	–	–	–	–	–	No peaks were distinguishable

Three ratios were calculated: SS ratio (Z. mays mean SS/P. sativum mean SS), SS-CV ratio (Z. mays mean SS-CV/P. sativum mean SS-CV) and FL ratio (Z. mays mean FL/P. sativum mean FL). The incipient concentrations for the TA effect are highlighted in bold type. Means followed by the same letter are not significantly different according to the Holm–Sidak multiple comparison test at P ≤ 0·05.

In LB01 buffer, the TA effect occurred after treatment with 0·30 mg mL⁻¹ TA. With increasing TA concentrations, both SS and SS-CV ratios increased. This was due to a more rapid increase of these parameters in Z. mays than in P. sativum. The FL ratio also increased with increasing concentrations of TA, but in line with the results obtained with P. sativum alone, the increase was due to a fluorescence gain that was higher in Z. mays than in P. sativum (Table 4).

Otto's buffer was the only one in which the ratios between SS, SS-CV and FL in Z. mays and P. sativum did not change significantly with changing concentrations of TA. With regard to the FL ratio, nuclei from both species lost a similar amount of fluorescence. The TA effect occurred after the addition of 1·00 mg mL⁻¹ TA, and even at 1·25 mg mL⁻¹ TA it was not possible to discriminate between nuclei peaks and background noise (Table 4).

In Tris.MgCl₂ buffer, the TA effect occurred after the addition of 1·75 mg mL⁻¹ TA. At this concentration a statistically significant increase was observed in the SS ratio, with the increase in SS being more pronounced in Z. mays than in P. sativum. At this concentration also the FL ratio increased significantly, because P. sativum nuclei lost more fluorescence than those of Z. mays. No nuclei peaks could be observed when 3·50 mg mL⁻¹ TA was added to the samples (Table 4).

Effect of incubation time

For Galbraith's buffer, a significant increase in Z. mays SS-CV was observed, but no significant differences were detected in the SS, SS-CV and FL ratios, either in TA1 or in TA2. The addition of TA after running samples for 60 s (experiment TA3) confirmed these results (Table 5).

Table 5.

Analysis of Zea mays SS-CV (%), SS ratio, SS-CV ratio and FL ratio (mean and standard deviation of the mean) of particles in liquid suspension prepared from Pisum sativum and Z. mays leaf tissue with four different buffers (Galbraith's, LB01, Otto's and Tris.MgCl₂) exposed to a specific initial concentration (c_i) of tannic acid (TA) and incubated for different periods of time

	Galbraith's buffer (ci = 0·50 mg mL−1)				LB01 buffer (ci = 0·40 mg mL−1)				Otto's buffer (ci = 1·00 mg mL−1)				Tris.MgCl2 buffer (ci =2·00 mg mL−1)
Experiment	Z. mays SS-CV (%)	SS ratio	SS-CV ratio	FL ratio	Z. mays SS-CV (%)	SS ratio	SS-CV ratio	FL ratio	Z. mays SS-CV (%)	SS ratio	SS-CV ratio	FL ratio	Z. mays SS-CV (%)	SS ratio	SS-CV ratio	FL ratio
Control	38·22 ± 4·079a	0·811 ± 0·0789a	1·143 ± 0·1850a	0·612 ± 0·0017a	36·57 ± 3·795a	0·670 ± 0·0254a	1·009 ± 0·0681a	0·617 ± 0·0052a	31·61 ± 3·261a	0·617 ± 0·0206a	1·113 ± 0·0790a	0·607 ± 0·0013a	39·60 ± 4·122a	0·652 ± 0·0227a	1·119 ± 0·0452a	0·603 ± 0·0151a
TA1	78·98 ± 37·786b	0·798 ± 0·1410a	1·100 ± 0·0914a	0·617 ± 0·0032a	106·43 ± 13·787b	0·773 ± 0·0712b	1·202 ± 0·0544b	0·624 ± 0·0028b	51·99 ± 7·125b	0·674 ± 0·0107b	1·218 ± 0·0885a	0·601 ± 0·0069a	66·75 ± 30·047a	0·725 ± 0·1369a	1·167 ± 0·0819a	0·632 ± 0·0044b
TA2	56·60 ± 21·179c	0·760 ± 0·0197a	1·226 ± 0·2146a	0·616 ± 0·0025a	112·15 ± 16·235b	0·807 ± 0·1304b	1·187 ± 0·0936b	0·624 ± 0·0041b	78·16 ± 15·923c	0·709 ± 0·0418c	1·152 ± 0·0851a	0·599 ± 0·0181a	91·00 ± 5·005b	0·779 ± 0·0428a	1·279 ± 0·0978b	0·634 ± 0·0048b
TA3bta	27·72 ± 12·054a	0·965 ± 0·1064a	1·045 ± 0·1038a	0·613 ± 0·0103a	39·68 ± 8·771a	0·805 ± 0·0488a	1·158 ± 0·0525a	0·612 ± 0·0033a	37·74 ± 4·924	0·629 ± 0·0518	1·105 ± 0·1463	0·613 ± 0·0053	41·23 ± 2·128a	0·665 ± 0·0368a	1·165 ± 0·0476a	0·604 ± 0·0135a
TA3ata	29·51 ± 13·619a	1·049 ± 0·1270a	0·988 ± 0·1480a	0·614 ± 0·0079a	86·27 ± 21·721b	0·798 ± 0·0548a	1·067 ± 0·0861b	0·611 ± 0·0021a	No peaks were distinguishable				72·56 ± 9·478b	0·951 ± 0·1197b	1·110 ± 0·0937a	0·606 ± 0·0212a

Control, no incubation with TA; TA1, incubation with TA for 5 min; TA2, incubation for 20 min with TA. In TA3, samples were run for 60 s without TA (TA3bta) and, after a pause, TA was added to samples (TA3ata). Means followed by the same letter are not significantly different according to the Holm–Sidak multiple comparison test at P ≤ 0·05.

In LB01, the addition of 0·40 mg mL⁻¹ TA was sufficient to induce the TA effect in both TA1 and TA2 experiments, significantly increasing the SS, SS-CV and FL ratios. Interestingly, SS, SS-CV and FL ratios were not affected in TA3 (Table 5).

In Otto's buffer, a longer incubation time with TA led to a higher SS-CV in Z. mays and affected nuclei SS values from both species differently. As expected from the concentration tests, no significant differences were detected in the SS-CV and FL ratios. An unexpected result was obtained in TA3 as no peaks could be distinguished immediately after the addition of TA (Table 5).

For Tris.MgCl₂, the addition of 2·00 mg mL⁻¹ TA caused an increase in the Z. mays SS-CV, in both TA1 and TA2. This was statistically significant only for TA2. The SS ratio was not affected in either experiment and the SS-CV ratio was only affected when samples were incubated for 15 min with TA. In TA1 and TA2, the FL decreased in both species, although, as the decrease was higher in P. sativum, the FL ratio increased. In TA3, a significant increase was only detected for the Z. mays SS-CV and SS ratio (Table 5).

The results obtained with all the buffers and in all experiments were reproducible, and no statistically significant differences were obtained between replicates.

DISCUSSION

The negative role of cytosolic compounds on estimation of genome size using FCM in plants has been formally studied only recently (Noirot et al., 2000, 2002, 2003, 2005; Price et al., 2000; Walker et al., 2006). Most of the studies were performed using crude tissue homogenates without analysing specific effects of individual chemical species. The only exception was the work of Noirot et al. (2003), which showed that the presence of chlorogenic acid (a precursor of polyphenols) induced a significant decrease in fluorescence of petunia nuclei. These authors also evaluated the effect of caffeine and found that it caused a significant increase on PI fluorescence.

The negative effect of TA, a common phenolic compound, which is frequently accumulated in various plant tissues, was discovered by Greilhuber (1988) who observed that tannins caused stoichiometric errors in Feulgen staining by limiting the access of Schiff's reagent to DNA. The author suggested interaction of tannins with nuclear proteins. In contrast, Price et al. (2000) found that soluble tannins isolated from sunflower leaves did not inhibit PI fluorescence of pea nuclei isolated in Galbraith's buffer, despite increasing the variance of the peaks. The results herein do not confirm this conclusion and show a strong effect of TA on relative fluorescence intensity of PI-stained nuclei. Furthermore, we show for the first time that a cytosolic compound can change light scattering properties of particles in a tissue homogenate.

An increase in SS and a high SS-CV were found to be diagnostic for the presence and action of TA, and the term ‘tannic acid effect’ is given to this phenomenon. The effect is due to the appearance of two new populations of particles in the nuclear suspension, which are characterized by particular side scatter and fluorescence intensity properties. In samples prepared from tissues that are relatively free of cytosolic compounds, a homogeneous population of nuclei with relatively low-scatter CV values (<50·00 %) should be observed on a cytogram of FS vs. SS. These results imply that, in addition to analysing FL and presenting FL histograms, researchers should also analyse and present the FS vs. SS in log scale and/or SS in log scale vs. FL cytograms, with the respective CV values obtained for each parameter. When the TA effect is observed, estimates of genome size may be compromised and the results should be interpreted with caution. As the TA effect usually occurs before a change in nuclei fluorescence is observed, it appears to be a safe parameter to identify ‘problematic’ samples.

The increase in nuclei SS observed at higher concentrations of TA was most probably due to the occurrence of precipitates induced by TA. These and the complex structure of TA led to nuclei aggregation. Greilhuber (1986) observed that fixed tannins adhered tenaciously to cellular structures, particularly to chromatin, strongly interfering with the Feulgen reaction. Other studies also described the effect of TA on protein precipitation (Giner Chavez et al., 1997) and membrane aggregation (Simon et al., 1994).

The TA effect describes the occurrence of two new and distinct populations of particles in a tissue homogenate. The first consists of nuclei to which weakly fluorescent particles were attached. This population differed from the population of single nuclei by slightly higher FL and higher SS values. The second population was characterized by higher optical complexity and lower FL. This population did not include nuclei and comprised various clumps of weakly fluorescent particles. Thus, at the time the TA effect occurs, three subpopulations of particles are detected by FCM: single nuclei, aggregates of nuclei with unspecific particles, and aggregates of diverse particles devoid of nuclei.

The TA effect was described in this study based on the use of a chemically defined compound. Unfortunately, little is known about the amount of TA naturally occurring in leaves and the amount of TA released after nuclear isolation procedures. Some tissues or plants may differ 100-fold in their polyphenol content. A range of treatments was therefore used from low concentrations with no detectable effect to very high concentrations that made the analyses of DNA content impossible. It will be important to prove that a similar effect is observed when analysing samples from plant tissues containing cytosolic compounds.

Price and Johnston (1996b) analysed DNA content in sunflower, which was later shown to contain compounds interfering with DNA staining (Price et al., 2000). The authors observed increased variation in DNA peaks on FL histograms and the presence of elevated levels of background debris. In order to improve the resolution of histograms of DNA content, they suggested a gating to discriminate populations of particles with optical properties different from nuclei. However, the results here indicate that even fluorescence from apparently normal populations of nuclei may be affected. Furthermore, Pinto et al. (2004) and Loureiro et al. (2005) working with two woody plants rich in phenolic compounds, Eucalyptus globulus and Quercus suber, respectively, obtained similar FS and SS patterns. In both species, it was difficult to obtain satisfactory results for mature field leaves as FL histograms were very similar to those obtained in this work. It was also clear that nuclei from target and standard species lost a great deal of fluorescence.

Previous studies on the effect of cytosolic compounds did not analyse the effect of nuclear isolation buffer (Noirot et al., 2000, 2002, 2003; Price et al., 2000). It was established here that different nuclear isolation buffers gave samples variable resistance to the negative effect of TA. However, the reasons underlying these differences are not clear. Nuclear suspensions made in LB01 and Galbraith's buffers were more susceptible to TA than those in Otto's and Tris.MgCl₂ buffers. Despite being vulnerable to lower concentrations of TA, nuclei isolated with LB01 maintained or even gained fluorescence (although this may be due to the presence of nuclei with attached fluorescent particles). This may be explained by the presence of spermine, which seems to be a better chromatin stabilizer than MgCl₂ used in both Galbraith's and Tris.MgCl₂ buffers. Otto's buffers do not contain a chromatin stabilizer and the presence of citric acid, which improves chromatin accessibility (Doležel and Bartoš, 2005), does not prevent the continuous loss of fluorescence. It is also possible that the higher tendency to clumping in LB01 and Galbraith's buffers in the presence of TA was due to the lower concentration of Triton X-100, which is included to disperse chloroplasts and decrease the tendency of nuclei and cytoplasmatic debris to aggregate (Doležel and Bartoš, 2005). Otto's buffer contains the weakest detergent, Tween 20, despite being at the same concentration as Triton X-100 in the Tris.MgCl₂ buffer. One possibility may be to use non-ionic detergents at higher concentrations to release chlorophyll from plastids and decrease fluorescence of debris.

In order to avoid the effect of antioxidants, LB01 buffer was used without β-mercaptoethanol. Because other buffers do not contain antioxidants, the results were not influenced by these compounds. Other precautions were also taken to obtain reproducible results. Following Noirot et al. (2000), a greater amount of plant tissue was chopped in appropriate volumes of nuclear isolation buffer to minimize possible intersample variations resulting from chopping time and intensity and material quantity. Suspensions of plant nuclei thus obtained were then divided into aliquots and subjected to different treatments. In addition, the staining was done with PI at saturating concentrations.

The findings of this study are relevant to standardization procedures. Internal standardization is commonly regarded to be the best procedure for genome size estimation, avoiding errors due to instrument instability and variation in sample preparation and staining. With the aim to eliminate the negative effect of cytosolic compounds, Price et al. (2000) and Noirot et al. (2000, 2003, 2005) recommended internal standardization, the rationale being that the standard and sample nuclei are influenced to the same extent. However, there is no substantial proof that cytosolic compounds affect staining of nuclei from different species in the same way. When measuring DNA content in Gossypium, Hendrix and Stewart (2005) used Galbraith's buffer to prepare nuclei suspension and obtained different effects when rice was used instead of barley as internal standard, usually with an increase in fluorescence. The results here show that with the exception of Otto's buffer, nuclei from P. sativum were more affected by TA than those of Z. mays, leading in some cases to more than a 9·0 % stoichiometric error.

In order to counteract the negative effect of cytosolic compounds on DNA staining, it is important to ascertain whether the interaction with DNA occurs immediately or if it develops with time. The results of the current study clearly show that the reaction is very rapid. This explains the results of Noirot et al. (2003, 2005), who were not able to eliminate the negative effect of cytosol by centrifugation, dilution and heat treatments of tissue homogenates.

The results demonstrate the importance of nuclear isolation buffer for accurate estimation of genome size. In addition to testing various buffers, selection of tissues with lower or no phenolic compounds should enable unbiased estimations (Suda, 2004). If no tissue and buffer combination provides acceptable results, the inclusion of buffer additives counteracting the negative action of cytosol is imperative. The most frequently used compounds are: β-mercaptoethanol, a powerful and popular reducing agent, but which is forbidden for health reasons in many laboratories; metabisulfite, a more user-friendly antioxidant that is being used instead of β-mercaptoethanol; and PVP-10, a commonly used tannin-complexing agent. Preliminary results with these protectants show that PVP-10 can suppress the TA effect (data not shown). If the TA effect continues to be present even after using additives, the quantity of plant material and chopping intensity should be reduced.

In conclusion, the present study has shown that cytosolic compounds, such as TA, interact with nuclei, potentially causing errors in estimation of nuclear DNA content. It is recommended that the presence of the TA effect is checked by analysing light scattering properties of particles in tissue homogenates. If the TA effect is observed, other buffers should be tested for tissue homogenization as well as various buffer additives. Although there are no ideal solutions, these recommendations, and the results presented here, should contribute to obtaining unbiased estimates of nuclear genome size using FCM.