SEARCH

SEARCH BY CITATION

Keywords:

  • flow cytometry;
  • genome size;
  • DNA content;
  • bryophytes;
  • propidium iodide;
  • buffer

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. LITERATURE CITED

Flow cytometry (FCM) is commonly used to determine plant genome size estimates. Methodology has improved and changed during the past three decades, and researchers are encouraged to optimize protocols for their specific application. However, this step is typically omitted or undescribed in the current plant genome size literature, and this omission could have serious consequences for the genome size estimates obtained. Using four bryophyte species (Brachythecium velutinum, Fissidens taxifolius, Hedwigia ciliata, and Thuidium minutulum), three methodological approaches to the use of FCM in plant genome size estimation were tested. These included nine different buffers (Baranyi's, de Laat's, Galbraith's, General Purpose, LB01, MgSO4, Otto's, Tris.MgCl2, and Woody Plant), seven propidium iodide (PI) staining periods (5, 10, 15, 20, 45, 60, and 120 min), and six PI concentrations (10, 25, 50, 100, 150, and 200 μg ml−1). Buffer, staining period and staining concentration all had a statistically significant effect (P = 0.05) on the genome size estimates obtained for all four species. Buffer choice and PI concentration had the greatest effect, altering the 1C-values by as much as 8% and 14%, respectively. As well, the quality of the data varied with the different methodology used. Using the methodology determined to be the most accurate in this study (LB01 buffer and PI staining for 20 min at 150 μg ml−1), three new genome size estimates were obtained: B. velutinum: 0.46pg, H. ciliata: 0.30pg, and T. minutulum: 0.46pg. While the peak quality of flow cytometry histograms is important, researchers must consider that changes in methodology can also affect the relative peak positions and therefore the genome size estimates obtained for plants using FCM. © 2010 International Society for Advancement of Cytometry

Much of the research on nuclear DNA content in plants includes observations of genome size variation between species possessing different morphological and ecological features (1–5). In plant groups where genome size variation is relatively constrained, it is essential to ensure that any observed differences in genome size are not due to technical errors. If comparisons between species are made using incorrect estimates, correlative relationships might be incorrectly inflated or obscured altogether.

Flow cytometry (FCM) has been a common tool for determining the nuclear genome size of plants since the 1980s. While methods such as Feulgen microdensitometry and Feulgen image analysis are still used to determine DNA content in plants, FCM became the prominent method used across the discipline after Galbraith et al. (6) developed a simple chopping method to release nuclei from plant tissue. Several studies have tested the efficacy of FCM and found that FCM produces reliable and rapid genome size estimates consistent with those obtained using Feulgen methods (7, 8). Improvements to the use of FCM for genome size estimation have also been documented, including the need for an intercalating fluorochrome (9, 10) and internal standardization with a plant of known DNA content (11–13). While most research groups now use propidium iodide (PI) and internal plant standardization to determine genome size using FCM, there are still many variations in other aspects of the methodology that have the potential to influence genome size estimates.

Various nuclei isolation buffers are routinely used in FCM, and it has been recognized that the choice of buffer can affect the quality of data for different plant species (14). More importantly, Loureiro et al. (14) observed that the choice of buffer also affected the relative fluorescence values, resulting in different genome size estimates, although this result was discussed only briefly. Additionally, Doležel et al. (8) observed variation in genome size estimates of the same plant obtained from four separate laboratory groups. Although different brands of FCM instruments and different buffers were used, the variation in genome size estimates was attributed to the instrument type (8). While both studies noted the effect of sample preparation methodology on C-value estimates and recommended further studies, empirical studies of this type are still lacking.

Stain concentration and staining duration are other aspects of FCM methodology that vary among research groups. Barre et al. (15) found that insufficient propidium iodide concentrations resulted in biased DNA content estimates in Coffea species. They determined that the optimum PI concentration for Coffea was 333 μg ml−1, a level much higher than most researchers use. Similarly, Loureiro et al. (16) demonstrated that increasing PI concentration resulted in increased fluorescent intensity of the samples until a saturating level of 150 μg ml−1 PI was reached, while Greilhuber et al. (17) reported that PI concentrations between 50 and 150 μg ml−1 are suitable. Other protocols suggest that 50 μg ml−1 PI is typically a suitable concentration, with the caveat that experimental determination of the appropriate concentration is required (18). Staining duration was reported to have no effect on fluorescent intensity by Michaelson et al. (7) or by Barre et al. (15), but Doležel et al. (18) stated that shorter staining duration often produces better results.

If buffer choice and staining regiment can affect genome size estimates as these studies have indicated, it becomes important to quantify such effects in order to avoid erroneous conclusions. This is particularly important when observed variations may be small but significant, as in studies of intraspecific variation or variation within higher level taxa with constrained genome sizes. Indeed, some controversial reports of intraspecific variation in genome size have been refuted by repeat analyses (19), and differences in intraspecific genome size have been suggested to be due to “technical artifacts” (20). This is critically important, as observed intraspecific variation in genome size has led to theories of genome plasticity, but may have been supported by false estimates (see Ref.19 for a complete discussion).

Given the range in methodology available for FCM determination of plant genome size, a researcher has several options to choose from. When approaching a new group of organisms to analyze for genome size, it has been recommended to test several different buffers and staining techniques before starting (18, 17, 21). The performance of the technique will depend on the type of plant tissue being analyzed and the presence of cytosolic compounds (14). However, in a review of the literature on FCM use in plant DNA research, Loureiro et al. (22) found that there was a geographical correlation with buffer choice and they suggest that this may be primarily related to the historical biases of a laboratory or researcher. As well, most papers do not provide justification for buffer choice or staining technique.

We have initiated a large-scale project examining genome size in bryophytes. As a first step we are following the recommendation to determine the optimum FCM buffer and staining techniques for bryophytes. This is critical as bryophytes have shown constrained genome size estimates up to this point (23). Bryophytes represent an example of a group where accurate estimates of genome size are particularly necessary to allow conclusions to be drawn from comparisons of genome size with ecological and physical traits. Incorrect estimates could lead to erroneous conclusions about patterns of genome size in the context of evolution and diversity in bryophytes. The objective of this study was to quantify the effects of FCM methodology, specifically buffer choice and staining conditions, on the accuracy and quality of genome size estimates for bryophytes.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. LITERATURE CITED

Plant Material

Four moss species from different families were collected in Preservation Park in Guelph, Ontario, Canada [43°30′41″N 80°13′20″W]: Brachythecium velutinum (Hedw.) B.S.G., Fissidens taxifolius Hedw., Hedwigia ciliata (Hedw.) P. Beauv. and Thuidium minutulum (Hedw.) B.S.G. As the bryophytes surveyed thus far have an average 1C-value of 0.51 pg (24), Raphanus sativus L. ‘Saxa’ (1.11 pg/2C) was selected for use as a DNA content standard from a list of recommended standards for plant flow cytometry (18). R. sativus seeds were obtained from the Laboratory of Molecular Cytogenetics and Cytometry (Olomouc, Czech Republic), and were grown in the University of Guelph Phytotron.

Sample Preparation

Preliminary analyses determined the appropriate amount of tissue needed to obtain sufficient nuclei counts and good quality peaks. For the bryophytes, 10 mg (±0.2 mg) of air-dried tissue was used for each estimate, and 50 mg (±2.0 mg) of fresh R. sativus tissue from the youngest leaves was used. Dried bryophyte tissue was used as preliminary studies showed that identical peaks were obtained from the four moss species even after several weeks (or months in some cases) of drying (for example see Fig. 1). General methodology followed Galbraith et al. (6) and Doležel et al. (18). The bryophyte and R. sativus tissues were cochopped in 1.5 ml of cold buffer on ice using a single-edged razor blade for each sample. The resulting homogenate was filtered through a 30-μm mesh filter and a final volume of 1 ml was obtained. RNase A (Sigma) was added at a concentration of 50 μg ml−1 and PI (Sigma) was used according to the methodology outlined below, and samples were incubated on ice.

thumbnail image

Figure 1. Tissue quality of Hedwigia ciliata at time zero (A) and 10 months later (B). [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Download figure to PowerPoint

Experimental Design

Effect of buffer.

The effect of nuclei isolation buffer on genome size estimates was tested for nine buffers across all four bryophyte species. Buffers were selected from the list of 10 most commonly used noncommercial buffers for plant flow cytometry (22), with the addition of two recently developed buffers (25): Baranyi's (20), de Laat's (26, modified as in27), Galbraith's (6), General Purpose (25), LB01 (28), MgSO4 (29), Otto's (30), Tris.MgCl2 (31), and Woody Plant (25). Five replicates of each moss species were analyzed on two separate days, for a total of ten independent replicates. PI was used at a concentration of 50 μg ml−1 based on literature recommendations, and samples were stained for 20 min. Inhibition effects were tested for each buffer and species combination, on each day of flow cytometry. This was completed according to the protocol described by Price et al. (12).

Effect of staining duration.

To test the effect of staining duration, LB01 buffer was selected based on the relatively high quality output obtained in the buffer test. Each sample was analyzed at the following times: 5, 10, 15, 20, 45, 60, and 120 min. A larger volume of sample (2 ml rather than 1 ml) was required to allow analysis of each sample at the seven different times. The larger sample was prepared by increasing the tissue amount to 20 mg moss and 100 mg of R. sativus, and by chopping in 3 ml of buffer. Samples were stained with 50 μg ml−1 of PI according to literature recommendations and each sample was incubated on ice between runs. Three replicates were completed for each species.

Effect of stain concentration.

Using the results from the buffer test and stain duration test, LB01 was chosen as the buffer for these experiments and 20 min as the staining duration. The PI concentrations tested were: 10, 25, 50, 100, 150, and 200 μg ml−1. Three replicates were completed for each moss at each stain concentration, except for T. minutulum at 150 μg ml−1 as there was not enough tissue.

Flow Cytometric Analyses

Samples were analyzed on a Partec CyFlow SL (Partec, Münster, Germany) equipped with a blue solid-state laser tuned at 20 mW and operating at 488 nm. The instrument was calibrated before each use using 3 μm calibration beads (Partec, Münster, Germany). The parameters recorded for each sample included orange fluorescence (FL2: 590 nm ± 25), red fluorescence (FL3: 630 nm), forward scatter (FSC) and side scatter (SSC). These parameters were observed alone and in combined histograms including: FL2 vs. FSC, FL2 vs. SSC, FL2 vs. FL3, and FSC vs. SSC.

For each peak of interest (sample and standard), the nuclei number and coefficient of variation (CV) were obtained using the gating function in FloMax Software by Partec (Version 2.52, 2007). Debris was removed by manually drawing polygon gates around regions of interest on the scattergram of PI fluorescence vs. side scatter, as shown in Figure 2B. To determine the peak mean and CV, regions were placed around the peaks of interest on the PI fluorescence histogram (Fig. 2C). While most genome size literature recommends CV's below 5% or 3% and over 1000 nuclei analyzed, it was expected that some of the methodology used here would not produce optimal results. Additionally, the small size of bryophyte nuclei can be near the resolution capacity of flow cytometers (32). For each histogram, the nuclei count and CV for each peak of interest were used to calculate relative standard error (RSE) using the formula: RSE = SE/mean, which can be calculated as peak CV%/√(peak nuclei count). The ability to consider both the number of nuclei measured as well as the CV is helpful when data might otherwise be ignored due to low nuclei counts or high CV values.

thumbnail image

Figure 2. Example of gating procedure using data from T. minutulum. Ungated histogram (A), scattergram with manually drawn polygon gate (B), histogram with polygon gate applied to remove debris and regions drawn to determine peak location and CV (C). (Gating procedure completed with FloMax Software by Partec Version 2.52, 2007). [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Download figure to PowerPoint

The nuclear genome size of the mosses was estimated using the standard R. sativus 2C-value = 1.11 pg (9), according to the formula: Moss 1C-value (pg) = (Moss peak mean/R.sativus peak mean) × 1.11 pg. The 1C-value of the moss was calculated because the leafy gametophyte tissue used for testing is haplophasic in mosses (i.e., contains one complete chromosome complement).

Statistical Analyses

The buffer data was analyzed using a mixed model analysis of variance (ANOVA) with buffer as a fixed variable and date as a random variable. To analyze the staining period data, a mixed model ANOVA with compound symmetry covariance structure was used, with time as a fixed repeated variable. Stain concentration was analyzed using a general linear model with concentration as the fixed variable. If the ANOVA results were significant (P = 0.05), means within each study and for each species were separated using Tukey's HSD post hoc test (P = 0.05). In all cases, model residuals were examined to verify that the assumptions of homogeneity of variance and normality were met. Data was analyzed using SAS 9.1 (2002–2003).

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. LITERATURE CITED

Buffer Test

Within each species, buffer had a significant effect on 1C-value (P < 0.05) and there was significant variation in the genome size estimates obtained from the nine buffers (Table 1). While absolute differences in 1C-value estimates were small (0.022–0.026 pg/1C), the largest percent differences ([maximum estimate − minimum estimate]/mean) were substantial. Estimates varied between the maximum and minimum values by 6% for B. velutinum (Otto's and Tris.MgCl2), 8% for F. taxifolius (Woody Plant and MgSO4), 8% for H. ciliata (Woody Plant and MgSO4/Galbraith's), and 5% for T. minutulum (General Purpose and Galbraith's). The pattern in estimates produced by the different buffers was not consistent across species, although Otto's buffer, Woody Plant buffer and General Purpose buffer all tended to produce estimates above the mean, and Galbraith's buffer usually produced estimates below the mean. The buffers LB01, modified de Laat's and Tris.MgCl2 produced estimates that were close to the mean values across all four species.

Table 1. The effect of buffer type on the 1C-value estimates of four moss species
BufferBrachythecium velutinumFissidens taxifoliusHedwigia ciliataThuidium minutulum
Mean 1C-value (pg) ± SEMean RSE (%) sampleMean RSE (%) standardMean 1C-value (pg) ± SEMean RSE (%) sampleMean RSE (%) standardMean 1C-value (pg) ± SEMean RSE (%) sampleMean RSE (%) standardMean 1C-value (pg) ± SEMean RSE (%) sampleMean RSE (%) standard
  1. Letters shared (within each moss) indicate no significant differences in mean 1C-value (Tukey's HSD test P = 0.05). SE, standard error of the mean; RSE, relative standard error of the mean, calculated as peak CV%/√ (peak nuclei count).

Baranyi's0.455 ± 0.002ab0.190.080.331 ± 0.001ab0.230.100.290 ± 0.001ab0.100.100.432 ± 0.002ab0.200.14
de Laat's (modified)0.450 ± 0.001ac0.130.110.327 ± 0.003cd0.170.090.280 ± 0.001c0.090.080.428 ± 0.000cd0.130.11
Galbraith's0.434 ± 0.003d0.090.070.325 ± 0.002c0.180.090.274 ± 0.002d0.080.060.419 ± 0.002e0.090.05
General purpose0.445 ± 0.002c0.060.090.332 ± 0.001ab0.170.070.286 ± 0.000a0.070.050.441 ± 0.001f0.090.06
LB010.438 ± 0.001d0.080.060.329 ± 0.003ad0.140.060.279 ± 0.001c0.070.050.426 ± 0.001d0.070.04
MgSO40.436 ± 0.002d0.100.100.324 ± 0.004c0.140.060.274 ± 0.002d0.070.050.430 ± 0.001ac0.080.06
Otto's0.459 ± 0.003b0.210.060.344 ± 0.002e0.190.060.291 ± 0.001b0.110.080.435 ± 0.002b0.160.08
Tris.MgCl20.433 ± 0.001d0.090.100.333 ± 0.002b0.180.060.280 ± 0.002c0.060.050.433 ± 0.002ab0.100.07
Woody Plant0.452 ± 0.002a0.190.080.349 ± 0.001f0.190.110.297 ± 0.002e0.110.100.429 ± 0.002acd0.200.12
Mean0.444 ± 0.003  0.333 ± 0.003  0.283 ± 0.003  0.430 ± 0.002  

The nine buffers also produced variation in the quality of the data for each of the four species (Table 1). Two measures were used to quantify quality: SE of the mean C-value and RSE. A larger SE of the mean indicated that the 1C-value estimates produced were less consistent over the ten replicates, whereas mean RSE reflected the quality of the individual histograms (combining the CV and number of nuclei obtained for each peak). The buffers LB01, modified de Laat's and General Purpose produced low SE of the mean across the four species. Low RSEs were obtained from the buffers LB01, MgSO4, and General Purpose (with sample CVs ranging from 5–7% for all four species). Woody plant buffer and Baranyi's buffer both produced the highest RSE values for the four moss species (sample CVs ranged from 6–9% for all four species).

A range in quality of data was also evidenced in the FSC vs. PI fluorescence and SSC vs. PI fluorescence scatterplots (Fig. 3). Buffers producing poor quality histograms tended to have significantly larger amounts of debris. Buffers generating high RSEs produced histograms with less compact clusters of nuclei, indicating that the buffers affected the relative size estimate (FSC) and relative surface complexity estimate (SSC) of the nuclei as well as the fluorescence. The inhibition tests did not show any clear pattern of inhibition, as the shift in the standard peak location was no different than that observed between separate runs.

thumbnail image

Figure 3. Ungated histogram and scattergram outputs for Fissidens taxifolius using LB01 buffer (A) and Woody Plant buffer (B). [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Download figure to PowerPoint

Stain Tests

Staining duration had a negligible effect on 1C-value estimates (Table 2), though there were significant differences between estimates (P < 0.05). The estimates showed a maximum change of 0.014 pg over 120 min (Table 2). While there was a statistically significant difference between estimates, the maximum percent difference was 4% for F. taxifolius and H. ciliata and 3% for T. minutulum. Histogram quality was generally poorest at the 5 minute staining period for all mosses except for T. minutulum, which showed little change in RSE until there was a sudden significant increase in RSE after 120 min of staining (data not shown).

Table 2. The effect of staining period with propidium iodide on the 1C-value estimates of four moss species
Staining duration (min)Brachythecium velutinumFissidens taxifoliusHedwigia ciliataThuidium minutulum
Mean 1C-value (pg) ± standard errorMean 1C-value (pg) ± standard errorMean 1C-value (pg) ± standard errorMean 1C-value (pg) ± standard error
  1. Letters shared (within each moss) indicate no significant differences in mean 1C-value (Tukey's HSD test P = 0.05).

50.446 ± 0.001a0.312 ± 0.002a0.286 ± 0.001a0.427 ± 0.002a
100.450 ± 0.000b0.314 ± 0.003a0.288 ± 0.000ab0.429 ± 0.002ab
150.448 ± 0.002c0.314 ± 0.004a0.288 ± 0.001ab0.430 ± 0.001ab
200.450 ± 0.001b0.314 ± 0.003a0.290 ± 0.001b0.432 ± 0.002bc
450.451 ± 0.000b0.317 ± 0.005a0.294 ± 0.001c0.437 ± 0.001cd
600.452 ± 0.001d0.319 ± 0.006ab0.293 ± 0.002c0.436 ± 0.000c
1200.453 ± 0.001d0.324 ± 0.009b0.298 ± 0.000d0.441 ± 0.003d
Mean0.450 ± 0.001 0.316 ± 0.002 0.291 ± 0.002 0.433 ± 0.002 

Stain concentration had a significant effect (P < 0.05) on the genome size estimates generated for the four moss species, with genome size estimates increasing with PI concentration (Table 3). The largest percent differences in the 1C-value estimates for each moss was 11% for B. velutinum, 10% for F. taxifolius, 12% for H. ciliata, and 14% for T. minutulum. All mosses appeared to have a saturating PI concentration at 100 or 150 μg ml−1, as indicated by the lack of statistically significant differences between estimates after these concentrations. Histogram quality at the different staining concentrations varied slightly. The lowest and highest concentrations (10 and 200 μg ml−1) generally had the poorest quality histograms (highest RSE, data not shown), whereas the data quality was best at 50 and 100 μg ml−1.

Table 3. The effect of propidium iodide concentration on the 1C-value estimates of four moss species
PI Conc. (μg/ml)Brachythecium velutinumFissidens taxifoliusHedwigia ciliataThuidium minutulum
Mean 1C-value (pg) ± Standard ErrorMean 1C-value (pg) ± Standard ErrorMean 1C-value (pg) ± Standard ErrorMean 1C-value (pg) ± Standard Error
  1. Letters shared (within each moss) indicate no significant differences in mean 1C-value (Tukey's HSD test P = 0.05).

100.414 ± 0.004a0.299 ± 0.002a0.266 ± 0.001a0.398 ± 0.002a
250.439 ± 0.001b0.313 ± 0.004ab0.279 ± 0.001b0.419 ± 0.001b
500.451 ± 0.002bc0.314 ± 0.003ab0.281 ± 0.000bc0.431 ± 0.002c
1000.452 ± 0.000c0.321 ± 0.001bc0.288 ± 0.000c0.447 ± 0.003d
1500.461 ± 0.004c0.326 ± 0.003bc0.298 ± 0.002d
2000.461 ± 0.003c0.332 ± 0.006c0.300 ± 0.002d0.457 ± 0.003d
Mean0.446 ± 0.007 0.318 ± 0.005 0.285 ± 0.005 0.431 ± 0.010 

Following the stain tests, the effect of buffer on genome size estimates was retested at the optimal PI concentration (determined in this study) of 150 μg ml−1. As the buffer tests were completed using 50 μg ml−1 according to literature recommendations, it was necessary to see if the increased PI concentrations changed the effect of buffer on genome size estimates. Three buffers that produced different estimates for H. ciliata (Galbraith's, LB01 and General Purpose) were retested using PI at a concentration of 150 μg ml−1. The higher concentration of PI resulted in higher genome size estimates for all buffers tried (Galbraith's: 0.320 ± 0.002 pg, LB01: 0.298 ± 0.004 pg, General Purpose: 0.343 ± 0.007 pg). Additionally, the proportional increase in the genome size estimates when using the higher concentration of PI was not consistent across buffers: the increase was greater for Galbraith's and General Purpose buffers than it was for LB01.

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. LITERATURE CITED

This study is the first to quantify the effects of common variants in flow cytometry practices on genome size estimates in a taxonomic group with constrained genome sizes. The use of flow cytometry to estimate genome size in plants is common practice, with a strong body of literature to support its use, and several publications that include specific recommendations to determine the appropriate methodology through careful preliminary testing (21, 18, 17). Many recent publications however, do not indicate that these preliminary tests have been carried out. As evidenced by this study, preliminary tests are essential to confirm that accurate genome size estimates are being obtained.

In this study, buffer choice resulted in up to a 8% difference in genome size estimates as well as affecting both the quality and consistency of the data. As shown in Figure 3, buffer choice strongly influenced FSC and SSC profiles (reflecting nuclei size and granularity) indicating the effect of buffer on the consistency of nuclei structural properties. The SSC histograms do not shown the same “tail effect” observed due to the effect of tannic acid as described by Loureiro et al. (16), but the buffers that produced the poorest quality histograms generally produced genome size estimates with the highest variability (SE of the mean) and also deviated the most from the mean estimate across all buffers. Some buffers, however, produced good quality histograms but resulted in genome size estimates that were also different from the mean estimate and had a higher SE of the mean, such as Galbraith's (Table 1).

Reasons for the variation in genome size produced by buffer might be attributed to the ability of the different buffers to counteract the effect of various cytosolic compounds (14, 17, 25). Bryophytes are known to have various terpenoid and aromatic compounds, though information for individual species is scarce (33). As the tests for inhibition in this study were inconclusive, one might rule out the role of cytosolic compounds. However, given the scale at which the genome size estimates were being compared, the peak shift responsible for the change in estimate would be small, and difficult to observe.

From the results, LB01 appears to be the best buffer for genome size estimation in bryophytes using flow cytometry. LB01 produced the best quality histograms (lowest RSE) and also produced consistent genome size estimates (lowest SE of the mean). Upon examination of the range in genome size estimates acquired for each of the four moss species, LB01 generally produced an estimate that was near the mean of the range. LB01 contains β-mercaptoethanol, which is an antioxidant that acts as a reducing agent by binding free radicals. Similarly, the MgSO4 buffer also produced high quality data, and also contains a reducing agent dithiothreitol (DTT). The reducing action of these compounds can ameliorate the effect of interfering compounds such as tannins or other phenolic compounds (17).

While the effect of PI staining duration on the genome size estimate was statistically significant, the absolute difference in the estimates was quite small. Stain uptake by the nuclei appears to be very rapid, with consistent genome size estimates and data quality occurring after 5 min of staining. In this study, the small DNA content of the bryophytes may have facilitated this process, and it should not be assumed that all plants react similarly to staining period.

Different PI concentrations also affected the genome size estimates for the four bryophyte species. The need for concentrations between 50 and 100 μg ml−1 of PI to reach saturation has been noted by various authors (15–17). The results in this study are similar, but the direct effect on genome size estimates has been demonstrated here. Changes in PI concentration resulted in up to a 14% difference in 1C-value (Table 3). It is interesting to note that the relatively high concentration of 100 or 150 μg ml−1 of PI was needed to reach saturation, even in organisms with such small DNA contents. This indicates that there were factors altering the uptake of the stain into the moss and radish nuclei, though the inhibition tests and the scatter plots did not indicate patterns of inhibition as found in other studies (16). While actual mechanisms are not clear, the role of secondary metabolites binding the PI is likely, reducing the amount of PI available to bind to the DNA. Inhibition of the stain binding to the DNA seems less likely, as a point was reached where the genome size estimates no longer changed with increased PI concentration. Regardless, this study indicates that higher PI concentrations are necessary to insure that genome size estimates are accurate for bryophytes. The higher concentration of PI did not alter the histogram quality.

Another factor that could contribute to slight variation in genome size estimates is the presence of microchromosomes (m-chromosomes). M-chromosomes have been reported in 20% of bryophytes, including the four families used in this study (34). The presence of m-chromosomes could account for some of the variability in the C-value estimates, but would increase the variance of all samples (regardless of buffer or stain treatment).

While this study has shown that methodology can be a source of error in estimating genome size, it also highlights the overall accuracy and efficacy of this method when used correctly. Interestingly, the 1C-value for Fissidens taxifolius estimated by Voglmayr (23) was 0.33 pg, which is the same as the estimate produced in this study, even though a different buffer and staining concentration were used. While this is encouraging, it should not deter researchers from being cautious when comparing estimates produced by different methodology.

Additionally, this study has produced three new genome size estimates for bryophytes. The methodology using LB01 buffer and a PI concentration of 150 μg μl was found to produce the most accurate data, therefore these are the 1C-value estimates reported here. The three new estimates are: Brachythecium velutinum: 0.46 pg, Hedwigia ciliata: 0.30 pg, and Thuidium minutulum: 0.46 pg. Similar to other genome size estimates for bryophytes (23), these new species have relatively small genomes.

This study highlights the importance of thorough methodological development prior to obtaining genome size estimates, as the methodology employed will necessarily vary for different study organisms. A shift in approach in which researchers pursue methodology that produces consistent and accurate genome size estimates, and not only “high quality” histograms is desirable, as these two factors may not necessarily be related. In this study, while the highest quality histograms (lowest RSE) usually corresponded with the genome size estimates that were closest to the mean and had the lowest SE of the mean, other high quality histograms did not follow this pattern. The discrepancies obtained from different buffer and staining practices may explain observed within species variation that has been reported incorrectly (19). As well, methodology that produces consistent results is especially relevant when making comparisons between species with small differences in DNA content. A lack of consistency will result in incorrect interpretations and correlations of genome size estimates to other environmental or morphological features.

Acknowledgements

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. LITERATURE CITED

Many thanks go to Paul Kron for feedback and consultation, to the Laboratory of Molecular Cytogenetics and Cytometry (Olomouc, Czech Republic) for Raphanus sativus seeds, and to the Ashton Statistical Laboratory at the University of Guelph for statistical assistance. The authors are also grateful to three anonymous reviewers for their constructive feedback.

LITERATURE CITED

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. LITERATURE CITED