Introduction
Vegetative desiccation tolerance is the ability of organisms to recover from extreme cellular water loss without forming specialized dormant structures (Oliver et al. 2000). Some desiccation-tolerant plants can survive only short periods of desiccation or require desiccation to onset slowly. Others, including some species of algae, bryophytes, and lycophytes, restore their physiological activity even after rapid or lengthy desiccation (e.g., Alpert & Oliver 2002). Vegetative desiccation tolerance is thought to be one of the key traits in the evolution of early land plants as they evolved from aquatic ancestors (Ligrone et al. 2012), and considerable research has focused on advancing our understanding of this capacity in mosses, lycophytes, and their immediate algal relatives. This research showed that multiple cellular processes are affected by desiccation including carbon, protein, and lipid metabolism, cell cycle, stress response pathways, and signaling activity (e.g., Holzinger et al. 2011; Yobi et al. 2012; Gao et al. 2017).
Our current understanding of the mechanisms and evolution of vegetative desiccation tolerance in land plants is mostly based on the comparison of a single terrestrial lineage of land plants (embryophytes) with streptophyte green algae (Farrant et al. 2009; Holzinger et al. 2011), a combined divergence of which spans 900+ MY (Leliaert et al. 2012). Unlike the land plants, which evolved on land once, green algae show multiple origins of terrestriality, including lineages inhabiting extreme environments, such as deserts (Lewis & Lewis 2005; Cardon et al. 2008; Rindi et al. 2009), providing multiple opportunities to understand and characterize the array of adaptations that may facilitate desiccation tolerance. In deserts the surface of the soil can be covered by soil crusts, a complex community of microorganisms, fungi, and bryophytes that together act as ecosystem engineers by powering nutrient cycles, preventing erosion, and enhancing water holding capacity, as well as influencing the composition of plant communities (Evans & Johansen 1999; Belnap & Lange 2001; Song et al. 2017). Of the bryophyte members of desert soil crusts, many possess cellular mechanisms allowing them to withstand desiccation events and persist in their environment (e.g., Proctor & Smirnoff 2000; Gao et al. 2017). The specific mechanisms of desiccation tolerance in chlorophyte green algae are not as well studied, but species isolated from desert soil crusts can, unlike their aquatic relatives, recover even after long and extreme desiccation (Gray et al. 2007).
Desert green algae arose multiple times during the diversification of Chlorophyta, often in clades containing both aquatic and terrestrial species (Lewis & Lewis 2005; Büdel et al. 2009; Fučíková et al. 2014). For example, Tetradesmus G.M. Smith (Sphaeropleales, Chlorophyceae, Chlorophyta), a common planktonic freshwater genus of unicellular or colonial green algae, contains several terrestrial species that are closely related to the aquatic species (Lewis & Flechtner 2005; Mikhailyuk et al. 2019; Terlova & Lewis 2019).
Given the irregularity of precipitation in terrestrial habitats, we hypothesize that desert, but not aquatic, Tetradesmus species tolerate desiccation in a vegetative state. We tested this by evaluating the recovery of physiological activity upon rehydration after desiccation. The rate of desiccation and the final value of relative humidity (RH) impact the time required for recovery of desiccation-tolerant plants (e.g., Bartoškova et al. Nauš 1999). Thus, our second hypothesis is that two variables, the rate and maximum intensity of desiccation impacts the extent of recovery ofTetradesmus . To test these hypotheses, we subjected six species of Tetradesmus (four terrestrial desert and two aquatic) to desiccation under three different scenarios (RH ~5% over 2-5h, 65% over 8h, 80% over 16h) followed by rehydration. We estimated photochemical yield of photosystem II (ΦPSII) from measurements of chlorophyll fluorescence during the desiccation-rehydration cycle to characterize the level of cell physiological activity. To account for possible within-species variation, which has been demonstrated for terrestrial streptophytic algae (Donner et al. 2017), we used multiple strains per species when possible.
Desiccation and osmotic stress cause permanent damage to the ultrastructure of desiccation-sensitive cells (e.g., Cheng et al. 2017), however reversible changes in ultrastructure are also known for desiccation-tolerant organisms (Wu et al. 2012; Li et al. 2014; Holzinger et al. 2015). These changes include decrease of the cytoplasm volume, following expansion or undulation of cell wall, and degradation of thylakoid membranes. Therefore, our third hypothesis is that cell ultrastructure will change in both aquatic and desert species after desiccation, and that the change is permanent in the aquatic algae, but reversible upon rehydration in the desert lineages. To test this hypothesis, we compared cell ultrastructure of selected aquatic and desert species in hydrated, desiccated, and rehydrated states using transmission electron microscopy. We also tested the integrity of the plasma membrane under osmotic stress caused by a sorbitol solution, which mimics desiccation, using a vital fluorescent dye and confocal laser scanning microscopy.
Materials and Methods
Study organisms
The response to desiccation and rehydration was studied in 13 strains ofTetradesmus belonging to 6 species: aquatic T. obliquus , T. sp. (listed at CCAP as ”T. raciborskii ” strain CCAP 276/35), temperate soil species T. dissociatus , and desert algae T. deserticola , T. adustus , T. bajacalifornicus . All algae were grown in liquid KSM medium (Clear & Hom, personal communication) under a 12:12 h light:dark cycle (photon flux density ~200 \(\mu\)mol m-2s-1) at 22˚C. Mixing of the cultures was achieved by orbital shaking at 0.4 rad\(\mathrm{/}\)sec.
DNA extraction, amplification, and sequencing
To confirm the phylogenetic placement of T. obliquus strain UTEX 72, we obtained the sequences of three DNA loci (tuf A,rbc L, and ITS2) and analyzed them with data from the other available species. Cells from culture aliquots of the strain UTEX 72 were concentrated, frozen, and then mechanically disrupted. Genomic DNA was extracted using the ZymoBIOMICS DNA Miniprep Kit. The tuf A gene was amplified with the primer pair tufAF – tufA.870r (Hall et al. 2010, Famá et al. 2012). For amplification of the rbc L gene we used the primer pair M35 – M650r (McManus & Lewis 2011). The ITS region was amplified using the primer pair ITS1 – ITS 4 (White et al. 1990, Hall et al. 2010). Standard PCR protocol was carried out with GoTaq Green Master Mix (Promega Corporation, Madison, WI, USA) according to manufacturer’s recommendations. Prior to sequencing, amplification products were purified with Exosap-IT Express (Life Technologies Corporation, Carlsbad, CA, USA). DNA sequencing was performed by Eurofins Scientific with the same pars of primers used for amplification reactions. Consensus sequences of three genes were obtained from forward and reverse sequences using Geneious 10.2.2 (https://www.geneious.com) and deposited to NCBI with accession numbers MT270139 for tuf A, MT270138 for rbc L, MT270137 for the ITS 2 region.
Phylogenetic analysis
A three-gene concatenated dataset used in the analysis, included sixTetradesmus species, some with multiple strains, and selected taxa from related genera (Supplements Table S1). ITS2 was aligned using sequences together with the secondary structure (inferred by homology prediction) using ITS2 database (Schultz et al. 2006, Koetschan et al. 2012).
Substitution model and parameter values for the phylogenetic analysis were selected with Partitionfinder2 (Lanfear et al. 2017) using algorithms greedy (Lanfear et al. 2012) and PhyML (Guindon et al. 2010). The Akaike Information Criterion (AIC, Akaike 1998) was used to select the best model. Bayesian Interference (BI) was carried out with MrBayes 3.2.6 (Ronquist & Huelsenbeck 2003) available on the CIPRES Science Gateway (Miller et al. 2010). The concatenated dataset was partitioned by gene and by codon position (for protein-coding genes). The HKY+I, F81, and HKY+G models were chosen for the first, second, and third codons of tuf A respectively, GTR+G was applied to all three codons of rbc L, and SYM+G model was selected for ITS 2. The analysis included two separate MCMC runs, each composed of four chains. Each MCMC chain ran for 2,000,000 generations, sampling trees every 100 generations. Upon completion, the runs were compared using Tracer v. 1.7 (Rambout et al. 2018) and the first 25% of generated trees were discarded as burn-in. A 65% majority-rule consensus topology and posterior probabilities were then calculated from the remaining trees.
Maximum likelihood (ML) analysis of the concatenated dataset was carried out with partitioning by genes. Model GYR+I+G was implemented with following parameters: nucleotide frequencies A=0.30920532, C=0.16801733, G=0.21847705, T=0.3043003; substitution rates: AC=0.40628109, AG=1.567514, AT=2.0983752, CG=0.60725568, CT=5.5017757, GT=1.000000; Pinvar=0.56788907; Gamma shape = 0.86898379. The ML tree was produced from the 1000 bootstrap pseudo-replicates using 65% majority rule. ML analysis and the bootstrap were performed using PAUP* V4.0a (Swofford 2003).
Desiccation and rehydration procedures
Desiccation experiments were carried out using desiccation chambers previously described by Karsten et al. (2014), illustrated in Fig. S1. Different levels of relative humidity (RH) in the chamber were achieved by adding one of three different desiccants to the chambers: 100 g of CaSO4 (W. A. Hammond DRIERITE Co. LTD, Xenia, OH, USA) achieving ~5 % RH, a solution of 33g LiCl in 100 ml of dH2O for 65% RH, and a saturated solution of KCl in dH2O (100 ml) for 80% RH.
Algal cell suspensions (50 \(\mu\)l, approximately 150,000 cells corresponding to chlorophyll concentration 30–40 mg ml-1) were placed onto glass fiber filters (Whatman, Maidstone, United Kingdom) in replicates of four. Three filters at a time were then positioned on a perforated metal grid inside a desiccation chamber containing the appropriate desiccant. RH levels in the chambers were monitored using PCEMSR145S-TH mini data logger (PCE Instruments, Meschede, Germany). The chambers were kept under dim light of ~10 \(\mu\)mol photons m-2s-1 at 22˚C. Measurements ofΔ F/Fm’ of PSII (ΦPSII) in T. obliquus, T. deserticola, T. adustus , and T. bajacalifornicus were taken using a PAM 2500 chlorophyll fluorometer (Heinz Walz GmbH, Effeltrich, Germany), the light probe was adjusted outside of desiccation chamber in 12 mm distance from the algal samples. Strains of T. dissociatusand T. sp. “raciborskii ” (CCAP 276/35) were acquired later. Experiments involving these species, along with a previously measured aquatic and terrestrial strain, followed the same protocol except that their chlorophyll fluorescence was recorded using a Junior PAM (Heinz Walz GmbH, Effeltrich, Germany).
Measurements of chlorophyll fluorescence were taken every 10 min during the desiccation period (PAM settings: measuring light 3, saturation pulse 6, actinic light 3). The desiccation was assumed complete when the mean of measured \(\Delta\)F/Fm’ for the algae on all filters reached zero. The samples were then rehydrated by adding 50 \(\mu\)L of the KSM growth medium to each replicate on the filter and the desiccant in a chamber was replaced with 100 ml of tap water to achieve higher RH (\(\sim\)96%) after which the measurements were resumed at the same intervals.
Cell structure during desiccation and rehydration
Transmission electron microscopy (TEM). Samples for TEM were prepared following the protocol described in Holzinger et al. (2015). Cells were fixed in 2.5% glutaraldehyde in cacodylate buffer (pH 6.8) for 1.5 h, washed in cacodylate buffer, then postfixed in 1% osmium tetroxide solution in cacodylate buffer overnight at 4°C. The samples were dehydrated in increasing graded ethanol solutions and embedded in Spurr’s resin (Sigma-Aldrich, St. Louis, MO, USA). Ultra-thin sections were prepared, then viewed using FEI Tecnai 12 G2 Spirit BioTWIN TEM microscope.
Confocal laser scanning microscopy (CLSM). To test whether the plasma membrane of cells is fragmented during osmotic stress, we subjected cell suspensions of one representative species from each habitat to 4M sorbitol and used the vital fluorescent dye FM 1-43 (green biofilm cell stain, Invitrogen Ltd. Paisley, UK) following the protocol described in Holzinger et al. (2011). Two aliquots of each selected species (aquatic T. obliquus strain UTEX 72, temperate soilT. dissociatus , desert T. deserticola strain SNI-2) were subjected to the osmotic stress in 4M sorbitol solution for 1 h (cells under osmotic stress), after which sorbitol was replaced with deionized H2O in one of the aliquots (rehydrated cells). Control samples served as hydrated samples. All samples were then exposed to 20\(\mu\)M Film Tracer FM 1-43, prepared from 20mM stock solution in deionized water for 30 min prior to examination with Nikon A1R Spectral confocal microscope (Nikon Inc, Tokyo, Japan). Samples were excited with the argon laser beam at 488 nm, emission was collected at 500-550 nm (false color green) and at 575–625 nm (false color red).
Statistical Data analysis with R
Data analysis and visualization were carried out with R (raw data and full code are available at DRYAD, doi:10.5061/dryad.sqv9s4n1t). We first calculated mean effective photosynthetic yield values from four individual measurements of each strain at each time point, then applied hierarchical cluster analysis to assign the ”hydrated”, ”desiccating”, and ”desiccated” physiological states to the measurements. The time of rehydration (corresponding to the ”rehydrated” physiological state) was recorded during data collection.
To compare recovery across species for each desiccation mode, we calculated recovery indices as the ratio of the mean effective photosynthetic yield value 10 min or 12 h after rehydration to the control hydrated value. We then used hierarchical cluster analysis of Euclidean distances to compare the recovery indices among the strains of desert and aquatic algae. The number of clusters was validated using gap statistics.
Results
Phylogenetic analysis
The tuf A and rbc L sequences obtained from T. obliquus strain UTEX 72 were very similar to these of the strain UTEX 393 and the ITS2 sequence of these strains were identical (see Table S1 for accession number of all sequences used in the analysis). Phylogenetic analyses of the concatenated data set showed that these two strains are grouped into the same clade with high support. Phylogenetic analyses conducted using BI and ML resulted in identical tree topologies (Fig. 1), which agrees with previously published phylogenies ofTetradesmus (Fletchner et al. 1998; Lewis & Flechtner 2005; Terlova & Lewis 2019). Species isolated from desert soil crusts do not form a single clade, and instead are dispersed across the tree, with at least one desert species having a strongly supported sister relationship to an aquatic species (Fig. 1).
Desiccation and Rehydration
The effective quantum yield of photosystem II (ΦPSII) of hydrated cells was similar in all species included in this study (approximately 0.6). The desiccation response pattern was also similar in all species and for all desiccation modes used. Specifically, ΦPSII remained constant for a considerable amount of time, but then dropped to zero within 30–80 min following the initiation of desiccation (Fig. 2). The loss of ΦPSII followed similar dynamics in desert and aquatic species of Tetradesmus , indicating that the algae are unable to prolong physiological activity under desiccation stress. A striking difference between the aquatic and desert Tetradesmuswas demonstrated upon rehydration (Fig. 3a), as only strains of terrestrial algae restored their photosynthetic capacity after desiccation (full recovery took 30–180 min). By contrast, the aquatic species showed no recovery of ΦPSII even after 12 h rehydration (with one exception, as discussed below). The response to rehydration varied among terrestrial species depending on the intensity of desiccation (Figs. 3b, 4). Rapid desiccation to \(\sim\)5% RH (2-3 h) appears to cause the most severe damage. Initially upon rehydration desert species exhibit a short-term recovery of the photosynthetic activity but did not recover permanently. Less severe desiccation to 65% RH (8 h) and 80% RH (16 h) resulted in long term recovery of all terrestrial species, however the initial level of photosynthetic activity was reached by some strains and not others as described below.
After 10 min of rehydration (Fig. 4a), the most striking difference was between the aquatic species (no photosynthetic activity with an exception of one aquatic strain) and desert species (high photosynthetic activity). Cluster analysis revealed four clusters. All desert species belonged to the same cluster when desiccated at 5% RH (Fig. 4a). Under milder conditions (~65% RH) T. bajacalifornicusexhibited ΦPSII values similar to those in the hydrated state (Fig. 4a), the aquatc taxa did not recover, and T. deserticola and T. adustus demonstrated 45–75% of their maximum ΦPSII. Under the mildest treatment (i.e., 80% RH, 16 h) one strain of the aquatic T. obliquus (UTEX 72) remained inactive, whereas another (UTEX 393) recovered some photosynthetic activity (Fig. 4a). Among desert species, T. bajacalifornicus again exhibited higher levels of recovery, with separate strains of T. deserticola and T. adustus grouped in the same cluster.
After 12 h of rehydration (Fig. 4b) the effect of intense desiccation onTetradesmus cells became more apparent, as neither aquatic nor the desert algae maintained photosynthetic activity under desiccation at 5% RH. Cluster structure of responses from the medium-rate desiccation (Fig. 4b, 65% RH) was similar to that of the 10 min rehydration (T. deserticola and T. adustus were clustered together, all strains reached ~75% recovery; T. bajacalifornicus was separated with the highest 100% recovery). Under the mildest desiccation at 80% RH, a single aquatic strain (UTEX 393) recovered 75% of its hydrated ΦPSII level, whereas another strain (UTEX 72) failed to recover even after 12 h rehydration (Fig. 4b). The responses of the soil alga T. dissociatus and the aquatic species T. sp. “raciborskii” to desiccation and rehydration followed the general trends described above (these data are presented in Supplementary materials, Fig. S2). Desiccation of these species was carried out separately from the other species, and their chlorophyll fluorescence was measured using a different instrument (Junior PAM, Heinz Walz GmbH, Effeltrich, Germany). The Junior PAM operates with a single fiber optic, and thus yielded lower values for the hydrated cells, however overall patterns displayed by these species agree with original data collected with the PAM 2500. To make these data comparable to the measurements taken for the rest of the species, a strain of aquatic T. obliquus (UTEX 393) and dessert T. bajacalifornicus (ZA 1-7) were desiccated at the same time.
Transmission electron microscopy (TEM) of cells in different hydration states
The cells of two investigated species (T. obliquus , UTEX 393 andT. deserticola , EM2-VF30) had a single cup-shaped chloroplast with a pyrenoid surrounded by several starch grains of different sizes. Typical for green algae, thylakoids are single or in stacks of 3­–4, and are never in larger stacks, contrary to what can be seen in embryophytes. Golgi bodies are composed of 6–8 cisternae with attached vesicles, and numerous mitochondria could be seen in the cytoplasm (Fig. 5). Desiccated cells do not show striking differences in their ultrastructure compared to the controls. The protoplasm did not shrink, thylakoid membranes were intact, and cytoplasm was not denser than in fully hydrated cells. Accumulation of electron dense globules, known as plastoglobuli (50–100 nm) was found inside the chloroplast of both species mainly under desiccated and rehydrated conditions (Fig. 5). Cellular membranes of T. obliquus appeared to have more contrast but did not show noticeable damage (Fig. 5b). No definite changes in cell ultrastructure were noticed upon rehydration in either investigated species. Thylakoid membranes of both species appeared intact (Fig. 5). The cytoplasm of T. obliquus was denser compared to the hydrated cells, but the plastoglobuli were smaller compared to the desiccated cells (Fig. 5). Some rehydrated cells of T. deserticola contained a larger number of small vacuoles, and chloroplast contained numerous plastoglobuli (Fig. 5b).
CLSM visualized differences in membrane integrity
The fluorescent vital stain FM 1-43 did not penetrate the plasma membrane of the aquatic T. obliquus in the hydrated state (Fig. 6a), whereas cells exposed to 4M sorbitol showed clear plasma membrane damage and rehydrated cells contained inner membranes stained with FM 1-43, indicating a severely damaged plasma membrane (Fig. 6b-c). In contrast, FM 1-43 did not penetrate the plasma membranes of the desertT. dissociatus and T. deserticola in either of the physiological states (Fig 6d-f and g-i, respectively), indicating preservation of membrane integrity.