Adaptation prevents the extinction of Chlamydomonas reinhardtii under toxic beryllium

Phytoplankton organisms and culture conditions

Chlamydomonas reinhardtii Dangeard (wild-type strain Chla) wild-type mesophilic Chlorophyceae was obtained from the Universidad Complutense de Madrid algae culture collection. This strain was isolated from a freshwater pond in Doñana National Park, Andalucía, Spain. The isolation site was not polluted by any detectable metal or radiation.

The strain was cultured axenically in 20 mL of BG-11 medium (Sigma–Aldrich Chemie, Taufkirchen, Germany) into 100 mL cell culture flasks (Greiner, Bio-One Inc., Longwood, NJ). The temperature and light conditions were 22 °C and 80 mol m⁻² s⁻¹ over the wavelength (ranging from 400 to 700 nm) under continuous illumination.

The clonal strain was maintained in a mid-log exponential growth by serial transfers of 100 cells every fifteen days to fresh BG-11 medium.

Toxic effect of beryllium on C. reinhardtii

The toxicity of beryllium was evaluated with a dose–effect relationship. The reagent used for the experiment was a beryllium sulfate tetrahydrate (BeSO₄ ∙ 4H₂O) with 98% purity level (Alfa Aesar, Haverhill, MA, USA). The beryllium salt was diluted in BG-11 filtered medium and subsequently re-diluted to obtain concentrations of 0.8, 4, 8, 16, 33 and 66 mg/l. For each dose and the unexposed control, three replicates of the Chla wild-type strain in the exponential growth rate phase were inoculated with 1.5 × 10⁵ cells/mL. The relationship between growth rate (m) and different dosage levels up to 28 days was analyzed. The number of cells was counted in Uriglass settler chambers (Fast read; Biosigma, Venice, Italy) with an inverted microscope with a coupled fluorescence lamp (Axiovert 35; Zeiss, Oberkochen, Germany). Both short-term (up to 96 h) and long-term (up to 28 days) actions were evaluated to assess the acute and multi-generational toxicity. Multi-generational effect are the changes across several generation of exposure to Be.

Adaptation capability of C. reinhardtii to beryllium

A modified fluctuation analysis for liquid media (López-Rodas et al., 2001) was performed based on the pioneering approach of Luria & Delbrück (1943). This experiment evaluated the capability of microorganisms (in our case, C. reinhardtii) to resist to a high dose of a stressor (beryllium) after a massive die-off of sensitive cells. Moreover, it allows for differentiation between evolutionary forces: if the resistance to beryllium results from the appearance of a spontaneous mutation prior to exposure (pre-selective adaptation) or from the adaptation of the strain caused by an environmental stress (post-selective adaptation) by acclimatization or post-adaptive mutations.

A schematic diagram of the modified fluctuation analysis is shown in Fig. 1. Two different sets of experiments, one acting as treatments (Set 1) and the other as controls (Set 2), were performed in gas exchange plastic tubes with BG-11 medium. All the cells inoculums were from the same parental population isolated from a single cell (clonal propagation), to prevent spontaneous mutation accumulation.

First, in Set 1 (treatment trials), 95 replicate tubes were inoculated with a low population density (N₀ = 10² cells), which ensured the absence of pre-existing mutants. The number of replicate tubes was calculated prior to the experiment to obtain suitable precision according to Li et al. (1983). These replicates were allowed to propagate until a high population density was reached (N_t = 10⁵ cells); during the cell division process (which spanned an average of 11.68 generations), several random mutations affecting the sensitivity to resistance arose at different times. At the end of the propagation process, fresh medium with the selective agent (33 mg/l of beryllium, which was the no-viability dose identified in the dose–effect assays) was added to all of the replicates.

For Set 2 (control trials), at the end of the growing phase of Set 1, 47 replicates (half of the parallel tubes in Set 1) with N₀ = 10⁵ cells from the same parental population as Set 1 and with selective medium containing 33 mg/l of Be were established.

Both sets of trials were maintained under the same culture conditions at 22 °C with continuous light for 90 days. After this period, all of the tubes were blindly counted (thereby ensuring that one resistant cell could propagate to a sufficient extent as to be detected).

From the fluctuation analysis, according to Luria & Delbrück (1943), three mutually exclusive possible outcomes could result:

1. Post-selective adaptation: If resistant variants arise by acclimatization or post-adaptive mutation, each cell presents the same probability of adapting in the selective medium such that the number of resistant cells should be similar among all of the replicates. Therefore, the inter-culture variation will follow a Poisson model (a variance/mean ratio similar to 1 and similar between Sets 1 and 2), presenting low variation in cell number.

2. Pre-selective adaptation: If resistant variants arise prior to exposure via random mutations that emerge spontaneously in the population, each parallel tube presents a unique probability per cell division of generating a resistant variant. Some parallel cultures will achieve resistant variants earlier than others during the propagation and will propagate resistant progeny. The tube-to-tube variation will be high in Set 1 and will not follow a Poisson model (a variance/mean ratio higher than 1, with a significant difference between Set 1 and Set 2). Set 2, acting as a control, will present low inter-culture variation. Consequently, if resistant variants arise by a priori mutations, then some replicates in Set 1 should show no resistant cells and others show high densities of resistant cells, whereas in Set 2, all of the replicates should show resistant variants.

3. No adaptation: If no resistant variants arise, then no living cells should be detected in any of the replicates of Set 1 or Set 2.

Fluctuation analysis also enables the estimation of the mutation rate (µ), i.e., the rate of appearance of resistant cells to the selective agent. From the proportion of replicates that show no resistant cells (P₀), we can estimate µaccording to the following equation: ${\displaystyle \text{µ}={\mathrm{Log}}_e{P}_0∕\left(N_t-N_0\right)\left(\text{Luria \& Delbrück},1943\right)}$ where Log_e is the natural algorithm, P₀ is the proportion of replicates in Set 1 that show no resistant cells at the end of the experiment, N₀ is the cell number at the beginning of the experiment and N_t is the number of cells at the end of the experiment.

Characterization of the beryllium-resistant population

Population characteristics, such as the growth rate (m) or coefficient of selection (s) of a population, facilitate an improved understanding of population dynamics. The Be-resistant variants from five different parallel tubes from Set 1 were randomly chosen and maintained under selective conditions for 60 days to establish a Be-resistant Chlamydomonas sp. strain. Subsequently, a small population of the resistant strain was cultured in fresh medium without Be and characterized under non-selective conditions.

The growth rate (Malthusian component of fitness (m)) is the rate of change of the number of individuals in a population in a time interval, and it was calculated with the following equation: ${\displaystyle N_t=N_0{e}^{mt}\left(\text{Crow \& Kimura},1970\right)}$ where t = 7 days and N₀ and N_t represent cell density at the beginning and end of the experiment.

The coefficient of selection or the selective elimination rate (s) results from the comparison of two alleles of different fitnesses and is an estimate of the relative fitness of the less-favored allele. Phytoplankton s values calculated in natural populations range between 0.03 and 0.1 depending on species (Fox, Nelson & Mccauley, 2010) and between 0.46 and 0.84 under selective conditions (López-Rodas, Maneiro & Costas, 2006). It is calculated by the following equation: ${\displaystyle s=1-\left(m^r-m^w\right)}$ where m^r and m^w are the Malthusian fitnesses under non-selective pressure of the beryllium-resistant population and the wild-type parental population.

When the mutation from wild-type cell to resistant variant occurs repeatedly in the population, new mutants will appear each generation. If the fitness of the mutant allele is lower than the wild-type allele, the resistant alleles will be eliminated from the population by chance or natural selection; however, the population will always have a number of mutant alleles that persevere. In these cases, the equilibrium of mutation-selection (q) indicates the number of resistant mutants as a function of the selective pressure present in the population and can be estimated as the balance between µ and s according to the following equation by Kimura & Maruyama (1966): ${\displaystyle q=\text{µ}∕\left(\text{µ}+s\right).}$ Phytoplankton q values under selective conditions range between 1.2 resistant cells (Marvá et al., 2014a) and 6,500 resistant cells (López-Rodas, Maneiro & Costas, 2006) per million cells in the population.

Transmission electron microscopy (TEM)

To assess the possible intracellular morphological differences between Chlamydomonas sp. Be-resistant cells and the wild-type parental cells, we performed a TEM analysis. Before the fixation of the two samples, the wild-type cells in BG11 medium and the Be-resistant cells in BG11 medium with 33 mg/l of beryllium were rinsed three times with a saline phosphate buffer by centrifugation at 4000 rpm for 5 min. The resulting pellet was preserved in EM fixative (2.5% glutaraldehyde in sodium cacodylate-sacharose buffer, pH 7.2, at 4 °C for 24 h) and post-fixed in a 2% osmium tetroxide (OsO₄) buffer for one hour at 4 °C, dehydrated in an increasing series of acetone, and embedded in Spurr resin. Afterwards, the samples were sectioned in 80-nm-thick sections with a LKB 2088 ultramicrotome and collected on copper grids. All of the reagents were purchased from Sigma-Aldrich Chemical Company (St. Louis, MO, USA).

The TEM images were obtained at 100 kV with an electron microscope, JEOL JEM-2010 (Jeol Ltd., Tokyo, Japan).

Morphometric and stereological variation

Morphometric analysis was conducted to quantitatively characterize cell ultrastructural organization and size. By means of stereology, we performed a morphometric analysis comparing the ultrastructure and cell volumes of the wild type and resistant variants of Chlamydomonas sp.

TEM ultrastructure micrographs were fabricated from each ultrathin section at 12000x. The number of micrographs of each organelle was estimated with Williams’s progressive mean technique (Williams, 1977) with a ±5% confidence limit, corresponding to approximately 25 micrographs in each case. The stereological and morphometric methodologies have been previously described in detail (Costas et al., 1988a; Costas et al., 1988b; Costas et al., 1992). The procedures used to calculate the micrograph profile areas were the point-counting procedure and planimetry. (Weibel & Bolender, 1973) (a schematic representation of the point-counting procedure is provided in Fig. S1).

The morphometric parameters studied were as follows:

1. Absolute volumes (V_x): ${\displaystyle V_x=\frac{{Vv}_x}{{Nv}_x}}$ where Vv_x is the relative volume of the component x and Nv_x is the numerical density of the component x.

2. Vv_x = volume density or relative volume of the component x, which indicates the relative volume of the component x with respect to the total cell volume. The volume density x was calculated as the relationship between the sectional areas and the reference in the micrograph ratio according to the principle of Delesse (Weibel & Bolender, 1973).

3. Nv_x = numerical density of the component x or the relative numerical density of the x component, which indicates the number of components of x per µm³. It is calculated as ${\displaystyle {Nv}_x=\frac{\left(K{Na}^{3∕2}\right)}{\left(\beta {Vv}_x\right)}}$ where Na is the numerical profile density and K and β are the coefficients of form, which estimate the diameter of dispersion (K) and the mean axial ratio (β) of the profile population.

   The coefficient of form β was calculated by using a normogram after determining the mean axial ratio. (Weibel & Bolender, 1973). ${\displaystyle K={\left(\frac{M_3}{M_1}\right)}^{\frac{3}{2}}}$ where M₃ and M₁ are the mean diameter of each profile.

4. The coefficient of form of the diameters of the axis (C.F.d): ${\displaystyle \text{C.F.d}=\frac{d}{D}}$ where d is the transverse axis and D is the longitudinal axis.

5. The coefficient of form for circular profiles (C.F.c): ${\displaystyle \text{C.F.c}=\frac{4\pi \text{Area}}{{\left(\text{Perimeter}\right)}^2}.}$ For the stereological study, the Chla wild-type and Chla Be-resistant cells were divided into five cellular compartments: nucleus (n), chloroplast (chl), mitochondria (m), storage products (s) (including vacuoles, inclusions and droplets) and total cells (c).

Imaging-PAM fluorescence valuation

PAM fluorometers emit pulse-modulated light to excite the fluorescence of chlorophyll a to evaluate the photosynthetic performance of algae and plants.

A Maxi-Imaging-PAM chlorophyll a fluorometer (Heinz Walz GmbH, Effeltrich, Germany) was used to measure the chlorophyll fluorescence of Chlamydomonas sp. This device is a high-resolution fluorometer that measures sample areas (selected areas of interest (AOI)), allowing the assessment of algae suspensions in multiwell plates. The Imaging-PAM analyzes the images of fluorescence emission captured by a couple device LED array and a filter-covered CCD camera synchronized with the light emission (the Imaging-PAM specifications are described in detail in Ralph et al. (2005).

To quantify the fluorescence values, we blindly counted the cells of the Chlamydomonas sp. wild-type strain and the Be-resistant strain with Uriglass settler chambers. In 48-well plates, we filled 15 wells of each strain with 5 × 10⁶ cells. Three wells of each strain were exposed to concentrations of 4, 8, 16, 33 mg/l of Be, and three unexposed replicates of each strain were used as controls. Measurements were taken the first 16 h and at 24, 48 and 72 h after exposure. We adapted the cells in the multiwell plate to darkness 30 min before the measurements.

Then, images of the dark-adapted state were recorded, with an image of the dark-adapted minimum fluorescence state (F₀) before the application of an excitation pulse (2700 µE m⁻² s⁻¹ provided by the blue LED ring at 470 nm) providing the dark-adapted maximum fluorescence state (Fm). Subsequently, light-adapted images were induced during a cycle of 10 measurements by the application of a saturation pulse (SP) (2700 µE m⁻² s⁻¹ blue LED light for 800 ms). During the cycle, the current fluorescence yield (Ft) in the AOI with a very low actinic light (2 Hz modulated blue excitation light of 470 nm and intensity of 0.5 µE m⁻² s⁻¹) was continuously measured to ensure minimum fluorescence (steady state). Images of the light-adapted estate were recorded immediately before the SP (the current fluorescence (Ft)) and at the end of the SP (quantifying the value of maximal fluorescence (F′m)). From the F′m and Ft values, the effective PS II quantum yield (Y(II)) was calculated (Genty, Briantais & Baker, 1989): ${\displaystyle Y\left(II\right)=\left(F^{\prime }m-Ft\right)∕F^{\prime }m}$ Phytotoxicity was quantified with the values of Y(II) of the control and the study samples as the relative inhibition of Y(II): ${\displaystyle Inh.\left(\text{\%}\right)=\left[\left(Y\left(II\right)\text{control}-Y\left(II\right)\text{sample}\right)∕Y\left(II\right)\text{control}\right]\times 100\text{\%}.}$ Values above 10% of inhibition of the Y(II) with regard to the control were considered phytotoxic.

The recorded digital images were analyzed with analytical software (Imaging-WIN; Walz, Effeltrich, Germany), and the numerical values of the fluorescence parameters of chlorophyll a were obtained.

Statistical analysis

The non-parametric Mann–Whitney U test (two-tailed) was performed to detect the significance of ultrastructural differences between wild-type sensitive Chlamydomonas sp. cells and Be-resistant mutant Chlamydomonas sp. cells. All of the statistical analyses were performed according to Siegel (1956). A X² goodness-of-fit test at p < 0.01 was performed to compare the differences in the variance/mean ratio between Set 1 and Set 2.

Toxic effect of beryllium to Chlamydomonas reinhardtii

Beryllium was found to be toxic to Chlamydomonas sp. (viability is shown in Fig. S2). In less than 24 h at the lowest dose of 0.8 mg/l, the viability decreased to less than half of the control viability. Moreover, no detectable living cells at 24 h were detected in the replicates with 33 and 66 mg/l Be, and no viability was observed at 28 days (viability was directly measured by counting fluorescence-emitting cells with an inverted microscope with a coupled fluorescence lamp). Based on these results, the dose of 33 mg/l Be was selected for the fluctuation analysis.

Furthermore, we obtained striking results at doses below 33 mg/l: following one week of exposure, the viability differences between the control and the exposed populations decreased. This phenomenon was most pronounced at 16 mg/l on the 14th day, as no living cells were observed during the first week at this dose, but during the second week, the population began to recover. This recovery phenomenon at the lower doses may be explained by the plasticity of the population and, specifically at 16 mg/l, by the propagation of the resistant cells, which might rescue the population from extirpation.

Adaptation capability of Chlamydomonas sp. to beryllium

Our first aim was to determinate if Chlamydomonas sp. was capable of adapting to 33 mg/l of Be and, if so, to determinate the nature of the adaptation (pre-selective adaptation, by means of existing mutations prior to the exposure, or post-selective, by means of plasticity or novo mutations).

Although Chlamydomonas sp. was apparently unable to survive at 33 mg/l (ostensibly undergoing a population die-off) after one month of exposure, the fluctuation analysis yielded a different outcome. At the beginning of the exposure, massive cell die-offs occurred in both Set 1 and Set 2. However, after 90 days of incubation under the same conditions, several living cells persisted in different tubes from Set 1 and in all of the tubes from Set 2 due to the proliferation of beryllium-resistant variants (Table 1). Variation in the resistant cell density between parallel replicates in Set 1 (from 0 to over 10⁶) was observed, indicating the presence of fluctuation (variance/mean ratio > 1 (p < 0.01 according to a X² goodness-of-fit test)). The strong fluctuation present in Set 1 indicates that Be-resistant variants rose prior to exposition by means of pre-selective adaptation. In Set 2, i.e., the control trials, the number of cells did not show large variation between replicates (from 10³ to over 10⁵), indicating weak fluctuation compared with that of Set 1 (p < 0.01 according to a X² goodness-of-fit test). The low level of fluctuation in Set 2 indicates that the large level of inter-tube fluctuation in Set 1 must be the result of processes other than sampling error.

The mutation rate calculated a posteriori of beryllium-resistant Chlamydomonas was 9.61 × 10⁻⁶ cells per division.

Population dynamics of the beryllium-resistant organisms

The growth rate of beryllium mutants cultured in BG-11 medium without beryllium was lower than that of the wild-type strains used as parental populations. The Malthusian component of fitness was 0.4129 for the wild-type strain and 0.3344 for the beryllium-resistant strain, yielding a selection coefficient (s) of 0.9215. This s value is one of the lowest calculated under selective pressure (López-Rodas, Maneiro & Costas, 2006). The frequency (q) of beryllium-resistant cells in the population was calculated as the mutation-selection balance, resulting in 10.42 mutant cells per million of wild-type cells in the parental population cultured without beryllium selective pressure. This q value is low compared with the range of calculated balances (López-Rodas, Maneiro & Costas, 2006). This frequency result explains the generation of beryllium-resistant cells with each generation in large populations of the wild-type strain and their sustained elimination by natural selection.

Morphometric and stereological variation: from macroscopic to TEM analysis

The first detectable phenotypical difference was observed macroscopically by comparing the behavior of the population in the culture flasks in liquid medium. Beryllium-resistant strains were distributed in aggregates at the bottom of the cell culture flasks, whereas the wild-type Chlamydomonas presented a uniform distribution in the culture medium. Chlamydomonas reinhardtii is known for its ability to form palmelloid colonies as a response to environmental changes (Iwasa & Murakami, 1968; Iwasa & Murakami, 1969; Lurling & Beekman, 2009). The optic microscopy valuation suggested such behavior (Fig. S3). According to the optic microscopic valuation, Chlamydomonas sp. Be-resistant cells presented a smaller size than did the cells of the parental population (Fig. 2) and did not show any active motion, suggesting that the flagellum was possibly damaged. The TEM images consistently showed an increased presence of starch granules within the chloroplasts (Fig. 2), which is related to a change in carbon storage characteristics that occurs in response to stress.

The stereological analysis of cell absolute and relative volume and organelle densities are summarized in Table 2. The table quantitatively characterizes the main ultrastructural features of wild-type and Be-resistant Chlamydomonas sp. The strains showed differences in ultrastructural organization. The absolute volumes of the Be-resistant Chlamydomonas sp. cells were significantly lower than those of the wild-type strain at p < 0.001, supporting the qualitative differences evaluated by optic microscopy. The relative volumes of the chloroplast and mitochondria were higher in the Be-resistant Chlamydomonas sp. strain, whereas the storage product’s relative volume was lower relative to the wild-type strain (p < 0.01). There was also a significant increase (p < 0.01) in the numerical densities of the chloroplasts and mitochondria of the Be-resistant Chlamydomonas sp. strain. No significant differences were found between the two strains in the relative volume of the nucleus.

Imaging-PAM fluorescence

The Y(II) of both wild-type and Be-resistant Chlamydomonas sp. strains was measured in triplicate at different Be doses for up to 72 h. The Y(II) of the unexposed wild-type and Be-resistant strains was approximately 0.7 and 0.6, respectively, showing close to 15% less quantum yield in the resistant strain than in the wild-type strain. A dose- and time-dependent decrease was observed in the wild-type strain. The Be-resistant strain showed stable values of Y(II) at the different dose regimes with slight variations over time. However, the unexposed replicates of the Be-resistant strain (control replicates) showed a significant increase in Y(II) at 72 h. This result can be explained by the cultivation of the Be-resistant strain in fresh culture medium without Be. A graphical representation of the Y(II) values is provided in Fig. 3.

The extent of inhibition compared with control, as represented kinetically, indicated the time- and dose-dependent performance of wild-type and Be-resistant Chlamydomonas sp. strains (Fig. 3). All of the wild-type populations showed phytotoxicity at 24 h at all dose levels. The levels of inhibition in this strain increased with time, reaching as high as 40% at 72 h at 33 mg/l of Be. The inhibition curves for the Be-resistant strain showed negative values to a value of 0.05%, indicating that no inhibition occurred for 48 h. At 72 h, the inhibition values in the resistant strain increased up to 15% (Fig. 3). This result might be the consequence of the increase in the Y(II) value of the control replicates due to the non-selective conditions of the medium.