Genome size is known to exhibit interspecies differences, but also to vary between populations within a given species and even between individual cells within an organism. Major differences have often been reported and attributed to differences in measurement conditions, in internal controls of genome size, and in the stains used. Flow cytometry using intercalating dyes is the most attractive method for measuring genome size.
We estimated relative genome size of nuclei from heads of Drosophila melanogaster adult males using a FACScalibur flow cytometer and propidium iodide.
We have shown that the genome size estimates depended on the temperature and humidity of the rearing medium and decreased with age in adult flies. There were large differences in genome size estimates between the vials in which the flies were maintained, but only slight variations within the vials, supporting the idea that the size estimate depends on the fly rearing conditions. Changes in the temperature of the solution of head nuclei analyzed by the cytometer also influenced the genome size estimate.
The reasons for inter-species differences in genome size is one of the major questions in biology, and has been under investigation by many researchers for many years. It is now recognized that this variation depends mainly on amplifications, deletions, and divergences of various repetitive sequences that are either distributed along the chromosomes or concentrated mainly in the heterochromatin (1–9). Relationships between genome size estimates and various physiological traits (rate of development, cell size, etc.) (10, 11) and environmental conditions (altitude, latitude, temperature, etc.) (12, 13) have suggested that natural selection could be involved in regulating genome size, although a change in size is primarily due to the genome's tolerance for repeated sequences (5) and its ability to mobilize them (14). Intra-species variation in genome size has also been shown to be associated with environmental conditions in plants, and seems to involve mobilization of some transposable elements in plants (15), Drosophila (14, 16), and pocket gophers (17). Any stressful conditions known to mobilize transposable elements, such as UV light, temperature, breeding conditions, etc. (18), are therefore factors potentially able to influence genome size. Walbot and Cullis (19, 20) thus propose that the mobilization of repeated DNA sequences may be a strategy for adaptating to a changing environment. It has even been reported that the amount of repeated elements in individual seeds depends on their position on the flowering head in Helianthus (21), although these differences seem to be very slight (6).
One important point in genome size measurement is that it is estimated by methods that are sensitive to chromatin structure, which modifies the accessibility of the DNA to the dye used. DNA contents estimated by flow cytometry, microspectrometry, or feulgen staining are usually correlated (6), and flow cytometry is increasingly viewed as an accurate method for estimating genome size if standard operating conditions are used (22). A correlation between estimates obtained using different methods is not in itself sufficient, however, to eliminate the possibility of an artifact, because DNA accessibility depends on complex interactions between intercalating or nonintercalating fluorochromes and DNA and various proteins. Variation in DNA accessibility due to the state of condensation of the chromatin has been reported in plants (21, 23), during spermatogenesis (24) and in frozen and thawed human spermatozoa (25), in mouse thymocyte nuclei (26), in ethanol-fixed cells (27), in dividing and stationary Euglena cells (28), during differentiation of Friend leukemia cells (29), and in DNAse I treated HeLa nuclei (30, 31). The exact impact of external environmental conditions on the genome size estimate therefore needs to be determined before a protocol can be defined that would permit reliable and comparable genome size estimates between individuals and populations within a given experiment and between different experiments. We investigated whether rearing conditions (temperature, humidity) and the conditions to which the extracted head nuclei were submitted before the DNA content was estimated were able to influence genome size estimate determined by flow cytometry in Drosophila melanogaster. We showed that the genome size estimate depended on the temperature and humidity of the larval rearing conditions, on the age of the adult flies, and on the temperature of the solution in which the fly nuclei were maintained before their introduction into the cytometer.
MATERIALS AND METHODS
Genome Size Estimation
The flies used in the experiment came from a natural population of Drosophila melanogaster captured in Canton (China) and maintained in the laboratory by small mass mating. Nuclei were extracted from five heads of freshly hatched Drosophila males. The heads were crushed in a small, siliconized Eppendorf vial containing 200 μL of labeling solution (0.1-g trisodium citrate, 0.01-mL Triton X100, 0.05-mg RNAse-A, water UHQ 100 mL) with 1 μg/mL propidium iodide (Aldrich) (14). The tetraodon (Tetraodon nigroviridis) blood used as the internal standard (0.8 pg) for diploid genome size estimate, but used here for relative fluorescence intensity determination of the fly nuclei, was labeled first with the fluorescent dye 5-6-carboxyfluorescein diacetate succinimidyl ester (CFSE, Molecular Probes, Leiden, The Netherlands) at 2 μg/mL in PBS for 15 min, at 37°C. The reaction was blocked by adding an equal volume of cold fetal calf serum, and the vial was placed in ice. Twenty microlitres of stained tetraodon blood was then added to the solution of fly head nuclei. The final mixture thus contained similar amounts of blood cells and fly nuclei, as estimated using a Thoma nucleus-counting cell. The mixture was then incubated for 10 min in ice, and filtered across 140-micron and then 30-micron nylon meshes. Six hundred microliters of initial labeling solution was then added to achieve an appropriate dilution. The resulting solution was analyzed on a FACScalibur flow cytometer (Becton Dickinson, San Jose, CA.) fitted with an argon laser at 488-nm wavelength. About 10,000 nuclei were analyzed, with an average rate of 500–800 events/second for each determination of the diploid genome size of the flies. The diploid tetraodon genome is usually estimated to be around 0.8 pg, but we found significant variations around this value for different fishes, suggesting that genome size is, in fact, an individual characteristic. We therefore used the blood from a single fish for each set of measurements. This means that the estimated genome size of the flies was in fact a relative estimate, depending on the fish blood used as the internal control. Hence, all comparative data of genome size estimate must be obtained within a given experiment, and absolute values of estimates of the genome size of flies from different experiments in the literature are not comparable. We thus determined the relative fluorescence intensity (the ratio of fly fluoresence intensity by tetraodon fluorescence intensity) instead of the absolute genome size of the fly nuclei.
In all experiments in which different sets of flies were to be compared (for example, three temperatures: A, B, and C, with k repetitions: 1, 2, …, k), we added the chemicals to the eppendorf vials in the following order: vial A1, then B1, then C1; followed by vial A2, then B2, then C2; and so on, until all k repetitions of the A, B, and C sets of treated flies had been dealt with. The same protocol was thus used in other independent repetitions of the same experiment. This eliminated any bias due to possible cytometer drift over time, and to variation in the exposure time of the chemicals, which could have happened if the A, B, and C sets of flies were processed successively. Although we did not observe any drift of the measurement with the propidium iodide during a single experiment, we did observe some drift of the fluorescence of the CFSE of the tetraodon blood. An example of a flow cytometry analysis of a sample of fly head nuclei and tetraodon blood is shown in Figure 1.
Effects of Temperature and Humidity
The Drosophila melanogaster flies used in our experiments were initially reared at 17°C under usual medium conditions. Two kinds of experiments were done to test the influence of temperature: 1) in seven independent experiments the flies were reared at either 17°C or 25°C; 2) in three independent experiments the flies were reared at 17°C, 21°C, 25°C, 29°C, or 31°C. In each experiment, 50 freshly laid eggs were collected and placed in vials, which were immediately placed at the temperatures at which the entire development from the eggs to adulthood was to take place. Six independent vials were considered for each temperature.
For the humidity experiment, larvae were raised either in vials from which all excess water was removed (low humidity) or in vials to which water was added (high humidity). Six independent vials were considered for each humidity level.
The flies were reared at 25°C and the relative fluorescence intensity was estimated from head nuclei of five- to six-hour-old adult males, and from six-, 10-, and 20-day-old males. Sets of five heads from each of six different vials were analyzed per age.
Effects of the Temperature of the Nuclei Solution
In the usual protocols for genome size estimation, the fly nuclei are crushed in a labeling solution containing propidium iodide (14). Tetraodon blood is then added, and the solution is kept on ice. The time during which the solution of head nuclei is at room temperature may vary with the number of samples to be analyzed. To check the influence of the temperature of this solution on relative fluorescence intensity, six sets of five heads of newly hatched flies raised at either 17°C or 25°C were crushed in 100 μL labeling solution without propidium. The solution plus nuclei was then separated into two 50-μL aliquots, which were kept at 17°C and 25°C, respectively, for 30 min. Then, 100 μL of labeling solution plus propidium, and 20 μL of the solution with tetraodon blood were added. The mixture was vortexed, placed on ice for 15 min, filtered as above, and finally 300 μL of labeling solution plus propidium was added. The resulting solution was analyzed on the cytometer.
The relative fluorescence intensities determined at different temperatures, humidity levels, ages of the flies, and temperature of the nuclei solution were compared by either one-way or two-way analysis of variance, depending on the protocol used and the number of conditions and repetitions.
Effects of Rearing Temperature and Humidity
Figure 2 shows the effects of temperature when the flies were reared at 17°C, 21°C, 25°C, 29°C, or 31°C. There was a significant global temperature effect (F = 4.29, P = 0.0047). In two experiments, there was a tendency for genome size to peak at around 21°C, but this tendency was less evident in the third experiment. Moreover, although the curves for the first two experiments had the same global form, the genome size values at the extreme temperatures could differ considerably and, for instance, the value at 17°C could be either greater or lower than the value at 29–31°C. This was clearly illustrated by comparing the seven experiments in which relative fluorescence intensity was estimated at 17°C and 25°C (Fig. 3). In five of the seven experiments, the relative fluorescence intensity at 17°C was lower than that at 25°C, and the differences were significant in three experiments. In the two experiments in which the relative fluorescence intensity at 17°C was higher than at 25°C, the differences were also significant (data not shown). Values of relative fluorescence intensity differed between the independent repetitions (F = 21.52, P < 0.001), suggesting that changes in cytometer measurement conditions, or uncontrolled external conditions, were important factors in determining the genome size estimate value.
We observed a high variance in the relative fluorescence intensity value between vials within a given experiment (see Figs. 2 and 3). This could have resulted from variation in the composition of the diet, especially in the water content, which is difficult to control. It should be noted that at 31°C, we added water to the plugs of the vials to prevent the premature death of the flies. Hence, to find out whether humidity interfered with temperature on relative fluorescence intensity, we reared flies at 25°C under “low humidity” (most of the humidity was removed from the vials) and “high humidity” (water was added to the medium) conditions. The results of two independent experiments clearly show that the relative fluorescence intensity was lower under wet conditions, as seen in Figure 4 (F = 8.46, P = 0.009). We thus conclude that differences in the humidity of the rearing medium could account for the differing estimates of genome size for a given temperature. This could also explain the decrease in the genome size estimate at high temperatures, notably 31°C, when more water was added to the vials.
Effects of Age
Figure 5 shows the relative fluorescence intensity of males analyzed just after hatching, and after six, 10 and 20 days. The relative fluorescence intensity decreased significantly with the age of the flies (F = 18.0, P < 0.0001).
Effect of the Temperature of the Solution of Fly Nuclei
Table 1 shows that the temperature of the head nuclei solution greatly influenced the relative fluorescence intensity, which was higher when head nuclei were maintained at 17°C rather than 25°C, whereas the values were not statistically different when the flies were reared at 17°C or 25°C (F= 0.22, P = 0.65 for 17°C; F = 0.29, P = 0.60 for 25°C). The relative fluorescence intensities when the head nuclei solution was maintained at 17°C were, moreover, lower than those obtained under the conditions of the usual protocol (vials immersed in ice) (F = 20.56, P < 0.001 for flies reared at 17°C; F = 20.38, P < 0.001 for flies reared at 25°C).
Table 1. Relative Fluorescence Intensity of Head Fly Nuclei Maintained in Solution at Various Temperatures
Rearing temperature of the flies
Head nuclei maintained in solution
The standard deviation is shown in parentheses.
Measurements of the genome size of species have often been determined from a single individual. It is becoming more and more evident, however, that genome size varies according to the population and even the individual concerned, and depends on various exogenous factors, the main effects of which seem to be the mobilization and increase in the copy number of transposable elements (2, 7, 32). Various correlations between genome size and latitude, altitude, temperature, light (9, 12, 13, 33), developmental rate (5), etc., are well documented, especially across species. Because the environment in which an organism grows seems to influence DNA content (6), it is concluded that genome size variation is an adaptation to changing environment. With the increasing use of flow cytometry based on intercalating fluorochromes, some of the correlations observed between genome size estimates and environmental factors may be artifacts, because DNA accessibility to the fluorochrome depends on chromatin structure, particularly its sensitivity to decondensation (23). Some slight fluctuations between individuals within populations and between populations may thus result, at least in part, from this kind of artifact, whereas differences between species are more likely really to involve natural selection and correspond to changes in genome size due to a change in the amount of DNA. We show here in Drosophila that the temperature and humidity level during development, the age of the adult flies measured, and the specific conditions of the buffers used for cytometric genome size estimation all directly influence the estimated genome size. We also checked the influence of the temperature at which the adult flies were maintained before measurement and the effects of ethyl ether anesthesia (data not shown). In both cases, some experiments showed statistically significant effects on genome size estimate, but the tendency was not reproducible, suggesting that various uncontrolled factors were interacting, sometimes in an antagonistic way, to influence the genome size estimate.
Drastic changes in the amount of transposable elements (TEs) are unlikely to be involved in such fluctuation in genome size, even though it is thought that organisms which develop at a slower rate tolerate more repeated DNA sequences (5, 34). This could be true for differences between species, but can hardly be applicable to immediate changes in genome size between individuals, depending on the temporary environmental conditions to which they are submitted. This does not mean, however, that transposable elements play no role in accounting for the differences in genome size between individuals and between populations. The TE content clearly differs between populations and species as a result of selection, and so may be related in some way to environmental conditions (14, 17). It is hard to imagine that a rapid change in TE content could be produced simply by changing the temperature at which the development of the flies takes place, the humidity of the medium, or the age of the flies, although such a possibility cannot be entirely ruled out by our experiments, and needs to be tested. A 14-fold increase in Tc1 somatic excision has indeed been reported during the lifespan of the nematode, Cænorhabditis elegans (35). It should be noted, however, that when our flies had food containing high levels of humidity, they had a longer developmental time (24 h more than under low humidity conditions), but their genome size estimate was smaller, which was contrary to what was expected. Similarly, there was no direct clear tendency towards a smaller or greater genome size estimate when development occurred at 17°C or 25°C. Moreover, the change in genome estimate depending on the temperature of the baths in which the nuclei were maintained before their DNA content was measured using the cytometer strongly suggests that what had changed was not the DNA content per se, but its estimation based on the intensity of fluorescence.
Fluctuations in genome size estimates of the kind observed in our experiments can thus simply be explained in terms of changes in the topology of chromatin, which modifies the accessibility of DNA to fluorochromes, especially to propidium iodide, which is often used in flow cytometry, as shown in various experiments (23, 26, 30, 31). Changes in DNA accessibility after a temperature treatment have been indeed reported for ethanol-fixed cells of rat thymocytes (27). The release of histones was blamed for the change in the binding of mithramycin as a result of a change in the three-dimensional structure of the chromatin (27). In Euglena cells, a change in fluorescence intensity was attributable to the exposure of chromatin-binding sites for ethidium bromide, which was related to the presence or absence of basic proteins (28). A decrease in DNA accessibility to different dyes, especially the DNA intercalators that unwind DNA, was observed during erythroid differentiation of leukemia cells. This effect disappeared after extraction of nuclear proteins using HCl (29). Histone H1 extraction increased the fluorescence intensity of isolated HeLa nuclei, because the binding sites had a higher affinity to the propidium iodide (30). Hence, the stainability of the DNA seems to depend greatly on nuclear decondensation (25) and the packaging with DNA-binding proteins, which also contributed to the difference in DNA stainability between spermatozoa from fertile and infertile men (36).
We must therefore be cautious when using flow cytometry to estimate DNA content, because many endogenous or exogenous factors, as well as the conditions to which the nuclei were submitted before being introduced into the cytometer (e.g., temperature, and pH of buffers ), may greatly influence the genome size estimate. Hence, before assigning a genome size value to a species, we need measurements for more than one population. In addition, the genome size estimate also depends on an appropriate standard of known genome size, which is a characteristic of every individual, in this case of the tetraodon fish, and of the environmental conditions to which the organisms to be analyzed were submitted. Strict controls and precautions are thus necessary before concluding that genome size is correlated with any specific environmental conditions.
We thank P. Nardon and M. Ghosh for comments and for reviewing the English text, C. Fisher for her gift of Tetraodon blood, and C. Lœvenbruck for valuable assistance.