Transcriptomic Analysis Reveals the Roles of Detoxification Systems in Response to Mercury in Chromera velia

Heavy metal pollution is an increasing global concern. Among heavy metals, mercury (Hg) is especially dangerous because of its massive release into the environment and high toxicity, especially for aquatic organisms. The molecular response mechanisms of algae to Hg exposure are mostly unknown. Here, we combine physiological, biochemical, and transcriptomic analysis to provide, for the first time, a comprehensive view on the pathways activated in Chromera velia in response to toxic levels of Hg. Production of hydrogen peroxide and superoxide anion, two reactive oxygen species (ROS), showed opposite patterns in response to Hg2+ while reactive nitrogen species (RNS) levels did not change. A deep RNA sequencing analysis generated a total of 307,738,790 high-quality reads assembled in 122,874 transcripts, representing 89,853 unigenes successfully annotated in databases. Detailed analysis of the differently expressed genes corroborates the biochemical results observed in ROS production and suggests novel putative molecular mechanisms in the algal response to Hg2+. Moreover, we indicated that important transcription factor (TF) families associated with stress responses differentially expressed in C. velia cultures under Hg stress. Our study presents the first in-depth transcriptomic analysis of C. velia, focusing on the expression of genes involved in different detoxification defense systems in response to heavy metal stress.


Introduction
Pollution with heavy metals is a major environmental concern worldwide because of their diffusion, persistence, and toxic effects. Heavy metals are released in the ecosystems during volcanic eruptions and by anthropogenic activities, such as mining, fossil fuel burning, agriculture, and waste management. Heavy metals are not degraded by organisms, but can be absorbed by plants and algae, thus entering the food chain. Among heavy metals, mercury (Hg) is particularly toxic, since it can accumulate at a high level, especially in fishes, and cause serious diseases, such as the infamous Minamata outbreak in Japan in the 1950s [1]. In the environment, Hg is present in different forms, i.e., as elemental Hg, ionic Hg 2+ , or generated as methyl-Hg. Regardless of the form, a large part of Hg ends up in the oceans, with estimates of about 290 million moles currently released [2]. Hg quickly apparatus, which seems to be well adapted to high photosynthesis-related oxidative stress [52][53][54]. Photosystem I (PSI) of C. velia non-canonically contains two permanently attached SOD molecules and several protein subunits with no sequence homology to known function [55]. Such organization of photosystems could provide a framework for the efficient and adaptable photosynthesis observed in C. velia [52,55].
This study aims to unravel the physiological responses of the marine protist C. velia to toxic Hg 2+ exposure. Due to the double role of C. velia as a model for the study of apicomplexan parasites and microalgae, the knowledge acquired by this study might be of interest for both environmental and medical purposes. We tackled the response with a physiological, biochemical, and transcriptomic approach, to identify the molecular mechanism of the stress response in C. velia. RNA-Seq is an efficient technology for transcriptomics studies [56]. In recent years, RNA-seq has proven useful for the identification of functionally related genes and their expression patterns in algal species responding to abiotic stress [57][58][59][60].
In this study we: (i) Characterize the physiological response of C. velia cultures to different concentrations of Hg 2+ , over time, assessing the toxicity effects and modulation of different ROS and RNS species; (ii) present the first deep transcriptome analyses for C. velia; (iii) identify differentially expressed genes (DEGs); (iv) investigate the molecular regulatory network of detoxification systems in C. velia; and (v) identify TFs involved in response to Hg exposure. This work provides a basic knowledge of the mechanisms used by Chromera to cope with heavy metal toxicity.

Chromera velia Culture and Treatment
C. velia was cultivated in F2 medium [49] made with fresh seawater collected in Sicily, Italy, filtered through Miracloth and pasteurized at 80 • C for 7 h. Minerals and vitamins were added at double concentration, compared to the original recipe. Cells were kept at 25 • C in polypropylene flasks under artificial light (30-50 µmol PAR/m2/s) with a day/night photoperiod 16 h/8 h. Subculturing was scheduled every two weeks, when cells were in the mid-exponential growth phase (Supplementary Figure S1) at approximately 6.5 × 10 7 cells/mL, by diluting 30 mL of inoculum in 300 mL fresh medium.
For ROS, RNS, cell viability, and chlorophyll measurements, cells were treated 10 days after subculturing, by adding 0, 0.3, 0.9, and 3 mL of a 10 mM HgCl 2 stock solution to 300 mL cultures, to yield 0, 10, 30, and 100 µM final concentrations, respectively. Measurements were performed 3, 6, 24, 48, and 72 h after treatment. Depending on the experiment, we performed three or four experimental rounds. In each experimental round, each treatment was run in three independent replicates. t-test was performed on all the analyses, and significant differences were scored for p < 0.05.
For the transcriptomic analysis, 300 mL of 10 day-old cultures were treated with either 0.9 mL HgCl 2 (30 µM final concentration) or an equal amount of pure water (mock) for 6 h. Cells were then harvested by filtration, weighted, and immediately frozen in liquid nitrogen until RNA extraction. The experiment was run in triplicate (three treated samples, three mock-controls).

Cell Count and Chlorophyll Quantification
To measure cell growth, we counted the cells present in the culture at a binocular microscope (200×) with the aid of a Burker slide chamber. Values are reported as the mean of 12 squares.
For chlorophyll quantification, we measured chlorophyll a, the only chlorophyll present in C. velia [54]. We followed the protocol described in [61], with some modifications. Briefly, 10 mL culture was centrifuged at 2000 rcf for 2 min to collect the cells in the bottom of a plastic tube. Cells were suspended in 1.5 mL of absolute ethanol in a plastic microtube and ground using a mortar and pestle with 100 mg of sand. Chlorophyll was then extracted by incubating the samples at 60 • C for 30 min in agitation (150 rpm). Quantification was performed in the Synergy H1 reader (Biotek, Winooski, VT, USA) by reading absorbance at 649 and 645 nm, and by applying the formula: Chlorophyll a (µg/mL) = 13.7 A 665 − 5.76 A 649 [62].

ROS and RNS Quantification
ROS and RNS were quantified by fluorimetric analysis using specific fluorescent dyes. In particular, 2,7-dichlorodihydrofluorescein diacetate (H 2 DCF-DA) is an intracellular marker that becomes fluorescent when oxidized [63]; dihydroethidium (DHE) is a specific marker for the superoxide anion (·O 2− ) [64]; 4-Amino-5-methylamino-2 ,7 -difluorofluorescein diacetate (DAF-FM-DA) is an intracellular specific marker for nitric oxide (NO) [65]; Aminophenyl fluorescein (APF) is a marker for peroxynitrite (ONOO − ) and hydroxyl radical (·OH), since the dye is unable to discriminate between the two molecules [66]. Dyes were dissolved in DMSO to a stock concentration of 10 mM (H 2 DCF-DA, DHE, APF) or 5 mM (DAF-FM-DA). All dyes were from Cayman Chemicals, USA. Two mL of culture was deposited in a well in a transparent 12-well polypropylene plate (Greiner Frickenhausen, Germany) and brought to pH 7.5 by adding 20 µL 10 mM Tris buffer (final concentration 100 µM). Then, 2 µL of dye was added to the culture, with final dye concentration of 10 µM (H 2 DCF-DA, DHE, APF) or 5 µM (DAF-FM-DA). Plates were incubated, in the dark, at 25 • C for 30 min on agitation (100 rpm). Fluorescence was then measured at a Synergy H1 reader (Biotek, USA) with a bottom reader mode and gain set to 80 (H 2 DCF-DA and APF) or 100 (DHE and DAF-FM-DA) and bandwidth of 9 nm. Due to the possible formation of cell clumps, which could affect the homogeneity of the fluorescence readout, the measurement was made in 21 different points on the well surface and averaged ("area scan" mode). Excitation and emission wavelengths for each dye were 495/525 Ex/Em for H 2 DCF-DA; 495/515 Ex/Em for DAF-FM-DA and APF; for DHE, we used 405/570 nm Ex/Em instead of the commonly used 480/580 nm Ex/Em because the former setup was proved to be more selective in detecting O 2− , rather than the unspecific oxidized byproduct 2-OH-ethidium [64].
For extracellular H 2 O 2 quantification, we used the xylenol orange protocol described in [67], with some modifications. Briefly, 2 mL of culture were filtered through a chromatographic column (Poly-Prep; Bio-Rad, Hercules, CA, USA) to separate cells from the growth medium. An aliquot of 500 µL of the flow through was added to an equal volume of assay reagent (500 mM ferrous ammonium sulfate, 50 mM H 2 SO 4 , 200 mM xylenol orange, and 200 mM sorbitol) and incubated for 45 min in the dark. The H 2 O 2 -mediated oxidation of Fe 2+ to Fe 3+ was determined by measuring the A 560 of the Fe 3+ -xylenol orange complex. All reactions were carried out at least in duplicate, and their reproducibility was checked.
2.4. RNA Isolation, cDNA Library Construction, and Sequencing TRI Reagent (Sigma-Aldrich, St. Louis, MO, USA) was used to extract total RNA from the six individual samples. Approximately 1.5 µg of the extracted RNA was used for complementary DNA (cDNA) library construction and subsequent Illumina sequencing. The six cDNA libraries (Cvel_cont1, Cvel_cont2, Cvel_cont3, Cvel_mer1, Cvel_mer2, and Cvel_mer3) were generated using NEBNext UltraTM RNA Library Prep Kit for Illumina (NEB, USA) following the manufacturer's instructions [68]: (i) Poly-T oligo-attached magnetic beads were used to purify mRNA; (ii) fragmentation was carried out under high temperature in NEBNext; (iii) M-MuLV Reverse Transcriptase (RNase H −) was used to synthesize first strand cDNA using random hexamer primer.; (iv) the cDNA fragments were isolated and purified with AMPure XP system (Beckman Coulter, Beverly, USA) after synthesis second cDNA strand and adaptor ligation; (v) then, the final cDNA library was created based on PCR amplification, and library quality was assessed on the Agilent Bioanalyzer 2100 system (Santa Clara, CA, USA). The resulting per cDNA library was sequenced using the Illumina NextSeq 550 platform (San Diego, CA, USA) following the manufacturer's recommendations at the Central European Institute of Technology of Masaryk University (Brno, Czech Republic).

Transcriptome Analysis
The RNA-Seq reads were assembled as previously described [68,69]. The assembled unigenes were annotated by BLAST search against protein databases, including NCBI non-redundant protein, Protein family (Pfam), Swiss-Prot, Protein sequence analysis and classification (InterPro), Kyoto Encyclopedia of Genes and Genomes (KEGG), and Gene Ontology (GO) with an E-value cutoff of 10 −5 . The best-aligning results from Nr, InterPro, and Swiss-Prot databases were selected as coding region sequences of unigenes. If the results from two databases conflicted with each other, a priority order of Nr and Swiss-Prot was considered. Meanwhile, Trinotate was used to archive all functional annotation data derived from the analysis of transcripts and integrate them into an SQLite database [70]. Also, the GO functional annotation of unigenes was gained using an in-house script, and GO functional classification was obtained to classify the possible functions of the unigenes based on Nr annotations using REViGO [71].

Quantification of Gene Expression Levels
RSEM software package was used to estimate the gene expression levels [72] for each sample: (i) The clean reads mapped to the de novo reference assembled transcriptome; (ii) the reads count was obtained for each gene; (iii) the gene expression level was normalized to FPKM (fragments per kilobase of transcript per millions fragments) based on the number of reads count. Differential expression analysis of the control and Hg-treated samples was performed using the edgaR R package (3.24.3) [73]. The threshold for differential expression was adjusted to p-value < 10 −3 using the Benjamini-Hochberg method.
Amino acid sequences of all five detoxification-related enzymes: (Heme oxygenase (HO); Superoxide dismutase (SOD); NADPH-cytochrome P450 reductase (P450R); Cytochrome P450 (CYP); and Catalase-peroxidase (katG)) from V. brassicaformis were retrieved from the Cryptosporidium Genomics database [76], whereas plants, green and red algae, alveolates (apicomplexans and ciliates), stramenopiles (oomycetes and diatom), metazoans, fungi, kinetoplastids, amoebozoans, parabasalids, and diplomonads sequences were downloaded from NCBI GeneBank and specialized databases. We also considered the available bacteria and viruses amino acid sequences in NCBI GeneBank, with particular emphasis placed on those prokaryotes that could serve as a donor of genes transferred during endosymbiotic events, such as cyanobacteria (sharing an ancestor with chloroplasts) and α-proteobacteria (sharing a common ancestor with mitochondria). Sequences were aligned using MAFFT version 7 software [85]. Ambiguously aligned sequences and gaps were excluded for further analysis. Alignments were tested using Prottest [86] to choose the appropriate model for amino acid substitution. Maximum likelihood method using RAxML [87] and IQ-TREE v1.6.10 [88] were used to construct phylogenetic trees.

RT-qPCR Analysis
The quantitative reverse transcription PCR (RT-qPCR) analysis was performed using three biological and three technical replicates of total RNA. First strand cDNA fragments were synthesized using the SuperScript IV RT (Invitrogen, Carlsbad, CA, USA). RT-qPCR was performed on the LightCycler 480 system (F. Hoffmann-La Roche Ltd., Basel, Switzerland) using a TP SYBR 2× Master Mix (Top-Bio, Prague, Czech Republic) in 25 µl reaction mixtures. The PCR reaction protocol was 94 • C for 10 min, 55 cycles of 94 • C for 10 s, 60 • C for 10 s, and 72 • C for 10 s. The fluorescence was measured via a 65-95 • C melting curve. The specific primers for RT-qPCR were designed using Geneious prime software [89] (Supplementary Table S2A). The relative expression level of the selected genes using the Actin gene as the internal control gene was calculated using ratio = 2 −∆∆CT .

Data Deposition
The six datasets of RNA-sequencing are available at the NCBI Short Read Archive (SRA) with the GenBank accession No.: PRJNA529782. The database archives all functional annotation data derived from the analysis was deposited into Mendeley Data with DOI: 10.17632/pkn2f6dzt3.1.

Mercury Toxicity
In conditions of nutrient starvation, C. velia, like many other microorganisms, enters in a dormant state, and it is less susceptible to environmental triggers [50]. Since the aim of this study was to study the physiological responses to Hg 2+ , we first characterized the growth curve of C. velia culture in our experimental conditions, to determine the best time for treatments. After subculturing, the culture experienced about three days of the initial lag phase (Supplementary Figure S1). By approximately day four, it entered into the exponential growth phase, that lasted for about two weeks. After that, if not subcultured, nutrients were depleted, and the culture rested in a stationary phase. In our previous experience with plant cell cultures and cadmium, we observed that treatments in the 50-150 µM range were able to trigger physiological responses and eventually cell death [20]. Therefore, for initial screening, we selected similar range of HgCl 2 concentrations (10, 30, 100 µM) and times of treatment (from 3 to 72 h) to study the physiological response of C. velia under different acute stress conditions, and to select the best condition for subsequent transcriptomic analysis. First, we measured cell growth and chlorophyll content, as indicators of Hg toxicity. Treatment with 10 µM Hg 2+ did not affect cell growth ( Figure 1A). At 30 µM and 100 µM treatment, we observed a dose-dependent reduction in cell number, although not statistically significant with p > 0.5 in all cases. The modest effect in cell growth was not surprising, considering the slow growth of C. velia cultures during the three days of treatment (Supplementary Figure S1, Figure 1A control).

ROS and RNS Response
Since it is well known that treatment with heavy metals, such as cadmium, often triggers ROS and RNS production, which is causally linked to toxicity [20], we wondered whether C. velia cultures treated with Hg 2+ similarly experienced oxidative and nitrosative bursts. We then measured the general level of intracellular oxidative stress, as reported by H2DCF-DA, and quantified the production of specific ROS and RNS compounds, namely extracellular hydrogen peroxide (H2O2), superoxide (O 2-), nitric oxide (NO), and peroxynitrite and/or hydroxyl radical (ONOO − / OH). Being ROS and RNS signaling molecules, as well as downstream events during cell death, we extended our analysis from 3 to 72 h. Surprisingly, we observed that the intracellular oxidative environment, measured as H2DCF-DA fluorescence, decreased when cultures were treated with high concentrations of Hg 2+ (30 and 100 μM) already within few hours after treatment ( Figure 2A). A similar pattern was observed when extracellular H2O2 was measured ( Figure 2B). H2O2 decreased at high concentrations of Hg 2+ . However, it significantly increased shortly after treatment with the lowest concentration (10 μM, 3 h, and 6 h). The decrease in H2DCF-DA fluorescence after exposure to high Hg 2+ concentrations had been already observed in other studies. In particular, the oxidative burst following treatment of Medicago sativa root cells with 30 μM Hg 2+ was modest and transient, occurring within the first hour of treatment [23]. At later times, H2DCF-DA fluorescence was markedly lower than untreated cells [22], a situation that reflects what we observed in C. velia. Similarly, another study observed decreased H2O2 levels 24 h after treatment of Conversely, chlorophyll content showed a marked reduction when cultures were treated with Hg 2+ , in a dose-dependent fashion ( Figure 1B). In untreated cultures, chlorophyll content increased slightly over time, a result in agreement with the higher number of cells counted in the same time frame. In cultures treated with 30 and 100 µM Hg 2+ , chlorophyll content was sharply reduced in all time points, indicating a toxic effect of Hg 2+ at these concentrations and time points of treatment. A similar effect of high Hg 2+ concentrations on chlorophyll content had been observed on the microalgae Coccomyxa subellipsoidea [26] and Chlamydomonas reinhardtii [8,90], and on cucumber cotyledons [91], confirming that Hg 2+ is toxic for plants and algae at high concentrations.

ROS and RNS Response
Since it is well known that treatment with heavy metals, such as cadmium, often triggers ROS and RNS production, which is causally linked to toxicity [20], we wondered whether C. velia cultures treated with Hg 2+ similarly experienced oxidative and nitrosative bursts. We then measured the general level of intracellular oxidative stress, as reported by H 2 DCF-DA, and quantified the production of specific ROS and RNS compounds, namely extracellular hydrogen peroxide (H 2 O 2 ), superoxide (O 2− ), nitric oxide (NO), and peroxynitrite and/or hydroxyl radical (ONOO − /·OH). Being ROS and RNS signaling molecules, as well as downstream events during cell death, we extended our analysis from 3 to 72 h. Surprisingly, we observed that the intracellular oxidative environment, measured as H 2 DCF-DA fluorescence, decreased when cultures were treated with high concentrations of Hg 2+ (30 and 100 µM) already within few hours after treatment (Figure 2A). A similar pattern was observed when extracellular H 2 O 2 was measured ( Figure 2B). H 2 O 2 decreased at high concentrations of Hg 2+ . However, it significantly increased shortly after treatment with the lowest concentration (10 µM, 3 h, and 6 h). The decrease in H 2 DCF-DA fluorescence after exposure to high Hg 2+ concentrations had been already observed in other studies. In particular, the oxidative burst following treatment of Medicago sativa root cells with 30 µM Hg 2+ was modest and transient, occurring within the first hour of treatment [23]. At later times, H 2 DCF-DA fluorescence was markedly lower than untreated cells [22], a situation that reflects what we observed in C. velia. Similarly, another study observed decreased H 2 O 2 levels 24 h after treatment of M. sativa roots with high (40 µM) Hg 2+ concentration [24]. Leaves showed increased H 2 O 2 production [24]. A different pattern in H 2 O 2 production was also observed for two fern species under the same experimental conditions [21], suggesting that the molecular events triggered by Hg 2+ exposure are not universal for any species and cell type.
Differently from what observed for H 2 O 2 , O 2− levels increased when cells were treated with high concentrations of Hg 2+ , especially 100 µM ( Figure 2C). O 2− production appeared to be a late event since, within the first day of treatment, levels were similar to control. A late (24 h) O 2− induction had also been shown in M. Sativa leaves incubated with Hg 2+ [24] confirming that C. velia responses are similar to higher plant systems.
NO production did not change dramatically after Hg 2+ exposure, though it was significantly lower at 24 h for all treatments, and at a later time for 10 µM ( Figure 2D). A decrease in NO after Hg 2+ treatment was observed also for the microalgae C. subellipsoidea [26]. Interestingly, NO was produced massively in Arabidopsis cells treated with another heavy metal, cadmium. Moreover, NO preceded and was required for subsequent H 2 O 2 production and programmed cell death induction [20]. The mechanism of toxicity triggered by mercury and cadmium likely follow different pathways.
Finally, we used the dye APF to quantify ONOO − and/or ·OH, since the marker is unable to discriminate between the two reactive molecules. Also, in this case, fluorescence did not vary much among treatments, although it slightly increased at 24-48 h for 30 and 100 µM treatments ( Figure 2E). ONOO − forms by reaction of NO with O 2− . Since NO levels did not vary during treatment, it was not surprising that ONOO − levels were likewise stable.
Due to the limited, nonsignificant effect of treatment with Hg 2+ at the lowest concentration (10 µM) on chlorophyll content and most of ROS/RNS production, we decided to use 30 µM as the best treatment condition for subsequent transcriptomic analyses. However, to focus on the early signaling event, and to avoid late toxicity effects, we collected cells early, 6 h after treatment. Asterisks indicate significant differences compared to the corresponding control value, for p < 0.05.

Sequencing Output and Assembly
RNA samples from C. velia cultures treated for 6 h with 30 µM Hg 2+ and controls were used for Illumina Genome Analyzer deep sequencing. Approximately 165 and 148 million raw reads were generated for control (Cvel_cont) and Hg 2+ -treated (Cvel_mer) cultures, respectively. After cleaning, read numbers reduced to 162 and 145 million, respectively. In total, we generated 313 million raw reads and 307 million cleans (Table 1). Among all the clean reads, more than 95% had Phred-like quality scores at the Q20 level (an error probability of 1%). The three sets of clean reads per treatment were de novo assembled into one de novo reference transcriptome with the "Trinity" program. After assembly, the six sets of clean reads were mapped to the reference transcriptome. Approximately 134 million (Cvel_cont) and 126.5 million (Cvel_mer) reads were mapped to the reference transcriptome, which accounted for 52.51% and 87.02% of the total clean reads for Cvel_cont and Cvel_mer cultures, respectively (Table 1).

Functional Annotation
The NCBI non-redundant (NR) database was searched using all the assembled unigenes using BLASTX tool with a cut-off E-value of 10 −5 [68]. In addition, all unigenes were annotated by aligning with the other five public databases, including Protein family (Pfam), Cryptosporidium Genomics database (CryptoDB), a manually annotated and reviewed protein sequence database (Swiss-Prot), Protein sequence analysis and classification (InterPro), and Gene Ontology (GO) databases with a cut-off E-value of 10 −5 . Analyses showed that 52,404 and 52,388 unigenes (58.32% and 58.3% of all unigenes) were annotated in the CryptoDB and InterPro databases, respectively, while only 11,173 unigenes (12.43% of all unigenes) were annotated in the Nr database due to the absence of genome and EST information for C. velia. Furthermore, 16,625 unigenes (18.5% of all unigenes) were annotated in Swiss-Prot database; 3456 (3.85% of all unigenes) unigenes were annotated in the Pfam, and 1005 (1.12%) were annotated in GO databases ( Table 2). Unigenes amino acid sequences were also searched for transmembrane helices using TMHMM [92] and signal peptide using SignalP v4.0 [78]. Out of total 89,853 unigenes, 1461 (1.62%) were predicted as transmembrane proteins while 1055 (1.17%) contained signal peptide pre-sequence (Table 2). Of the 89,853 unigenes, 68,713 (76.47%) returned at least one match at the E-value < 10 −5 ( Table 2).

Differential Expression Analysis
The read counts for unigenes were quantified based on RPKM (reads per kilobase per million) normalized matrix, to facilitate the comparison of mRNA levels. Differentially expressed genes (DEGs) (p-value < 0.001 and log2 (fold change) >2) were defined as genes that were significantly enriched or depleted in Hg 2+ -treated cultures compared to control (Supplementary Table S2B). Based on the six sample's log10RPKM, hierarchical clustering of the DEGs was performed to observe the overall gene expression pattern. The green bands represent the up-regulated genes, while the red bands identify down-regulated genes (Figure 3). Two groups of DEGs with specific expression patterns were outlined from the clustering. In total, 1239 DEGs were found. Hg 2+ -treatment resulted in up-regulation of 1070 DEGs and down-regulation of only 169 DEGs (Figure 3 and Supplementary Table S2B).

ROS-Antioxidant Defense System Related Genes
Mercury tolerance of algae and its association with the different detoxification defense systems was reported [44,93,94]. C. velia is one of the most remarkable endosymbiotic algae, which showed a strong ability to survive in a wide range of environmental conditions [54,95,96]. However, little is known about the molecular mechanisms of heavy metal tolerance in C. velia.
Based on deep sequencing of the six cDNA libraries and the GO analysis, a total of 145 ROSrelated unigenes were identified (supplementary Table S3A). Among them, 117 (80.7%) unigenes were shared by all six libraries, and 12 unigenes were specifically expressed in a single library (Cvel_cont2). Additionally, two unigenes (DN73858_c0_g1 and DN85649_c0_g1, encoding for Kynurenine 3-monooxygenase and indole-3-acetic acid amido synthetase, respectively) were only expressed in control cultures (supplementary Table S3A

ROS-Antioxidant Defense System Related Genes
Mercury tolerance of algae and its association with the different detoxification defense systems was reported [44,93,94]. C. velia is one of the most remarkable endosymbiotic algae, which showed a strong ability to survive in a wide range of environmental conditions [54,95,96]. However, little is known about the molecular mechanisms of heavy metal tolerance in C. velia.
Based on deep sequencing of the six cDNA libraries and the GO analysis, a total of 145 ROS-related unigenes were identified (Supplementary Table S3A). Among them, 117 (80.7%) unigenes were shared by all six libraries, and 12 unigenes were specifically expressed in a single library (Cvel_cont2). Additionally, two unigenes (DN73858_c0_g1 and DN85649_c0_g1, encoding for Kynurenine 3-monooxygenase and indole-3-acetic acid amido synthetase, respectively) were only expressed in control cultures (Supplementary Table S3A). Based on the unigenes FPKM values, the expressions of 145 ROS-related genes in control and Hg 2+ -treated cultures were summarized in Supplementary Table S3A and Figure 4. Of them, 125 unigenes (86.2%) showed similar expression patterns in cultures under Hg 2+ treatment (Cvel_mer) compared to controls (Cvel_cont), while 20 unigenes showed differential expression (DE) patterns (16 up-and four down-regulated) (Figure 4).
Superoxide dismutases (SOD) catalyze the disproportionation of O 2− to O 2 and H 2 O 2 as the first defense line of the cell against ROS [104]. Since ·O 2− is a precursor of several other highly reactive species, control of this free radical concentration by SOD constitutes an essential protective mechanism [105]. The SOD genes isoform FeSOD found has been reported from only a few microalgae [106]. Moreover, the activation of specific SOD isoforms can serve as a biomarker for cells that contains a high level of·O 2− due stress response [107,108]. Subsequently, the enzyme catalase (CAT) catalyzes the production of H 2 O from H 2 O 2 , while ascorbate peroxidase (APX) reduces peroxides (H 2 O 2 and organic hydroperoxides) to H 2 O or the corresponding alcohols, respectively, using ascorbate as an electron donor [44,109]. Catalase Peroxidase (KatG) is a dual catalytic activity enzyme, acting both as catalase and as a peroxidase. Evolutionarily, KatG contains two peroxidase-like domains, an N-and a C-terminal domain. Its C-terminal domain has no known catalytic activity [110]. KatG probably evolved by the fusion of two copies of a rudimentary peroxidase gene (which probably also gave rise to the ascorbate and cytochrome c peroxidases) [111]. KatG sequences are present in the genomes of Archaea, Bacteria, and some protists [112]; its C-terminal domain has no known catalytic activity [110]. Even though it shares the same EC number with monofunctional CATs, they have no sequence similarity, which suggests their different evolutionary origins [113].
Four SOD genes found in this study had been previously annotated in the C. velia genome, with two being MnSODs, and two plastid FeSODs (SOD1/2) [55] (Supplementary Table S1). Seven V. brassicaformis SOD loci were identified, three of them being MnSODs, and four FeSODs. Interestingly, we identified two plastids KatG in C. velia but none in V. brassicaformis nor other protists (Apicomplexa, Diplomonadida, and Parabasalidea). Possibly, their absence is compensated by other peroxidase class I enzymes (cytochrome c peroxidase (CcP) and ascorbate peroxidase (APx)) [112] (Supplementary Table  S1). In this study, FeSOD (DN97564_c1_g1 and DN110975_c0_g1) and MnSOD (DN100508_c0_g1) were down-regulated by Hg 2+ in C. velia (Supplementary Table S3A). Another unigene encoding MnSOD (DN109194_c0_g1) was unaffected by Hg. Both FeSOD and MnSOD showed lower expression levels in Hg 2+ -exposed cultures of C. velia than those of control cultures (Supplementary Table S3A and Figure 4). This result is in agreement with the highest ·O 2− production observed when cultures were treated with high concentrations of Hg 2+ ( Figure 2C). The gene for another water-water cycle enzyme, catalase-peroxidase (katG) (DN112103_c0_g1) was up-regulated while the other katG unigene (DN79540_c0_g1) was unaffected (Supplementary Table S3A, and Figure 4). Again, these results are in agreement with the observed ROS production, since H 2 O 2 levels decrease after Hg 2+ treatment (Supplementary Table S3A,B), probably as a result of the increased H 2 O 2 scavenging capacity. The unigene encoding FER3 (DN112480_c0_g3) was not affected by Hg (Supplementary Table S3A). Ferritin (FER) is an iron-storage protein, has been reported to seize Fe 2+ , and to prevent the formation of ·OH [18,114,115]; its overexpression improved abiotic stress tolerance [115,116]. In our experimental system, we did not observe variation in ·OH levels following Hg 2+ treatment ( Figure 2E), a trend that correlates with the unchanged expression of its gene. Other ROS-related genes were identified in this study, although we do not have enough biochemical data to correlate their expression to ROS modulation. Glutathionylation-related enzymes (thioredoxin), encoded by unigene (DN107579_c0_g1) and other redox-related enzymes (flavin reductase (NADH), NADH dehydrogenase (plastoquinone), and kynurenine 3-monooxygenase) encoded by unigenes DN94003_c0_g1, DN99597_c0_g1, and DN7632_c0_g1, respectively, were up-regulated (Supplementary  Table S3A and Figure 4).

Xenobiotics Detoxification-Related Genes
A total of 126 of xenobiotics metabolism-related unigenes were identified in C. velia (Supplementary  Table S3B (Figure 4).
Ten classes of cytochrome P450 systems (CYPs) have currently been assigned [117]. Mixed-function oxidases (MFO) class II includes most eukaryote CYPs localized to the endoplasmic reticulum and some other membranes via an N-terminal anchor and associated with a separate NADPH-cytochrome P450 reductase (P450R). The cytochrome b5 system is attributed to this activity [117][118][119]. Also, CYPs show protein-protein interactions with other CYPs to form homomeric and heteromeric complexes; these complexes can often have significant effects on CYP-mediated oxidation of substrates [120]. The analysis of the functional consequences of CYP−CYP homomeric interactions found that the complex model fits best to either a quaternary (CYP−P450R−CYP−P450R), trimeric (CYP−CYP−P450R), or dimeric (CYP−CYP) complex model, where only the binary (CYP−P450R) complex was active [120,121]. Some xenobiotic compounds are metabolized directly by phase II enzyme systems (e.g., glutathione S-transferases (GSTs) and UDP-glucoronyl transferases (UDPGTs) [37], while others are metabolized prior by the action of phase I enzymes. No evidence for either CYPs complexes or multixenobiotic resistance (MXR) presence exists in phytoplankton or macroalgae.
Genes encoding for biotransformation phase-I-related enzymes, including mixed-function oxidases (MFO) class II CYP (cytochrome P450 (CYP), cytochrome P450 reductase (P450R), and cytochrome b5 (Cyt b)) system, were identified in this study. One P450R unigene (DN11036_c0_g1) identified in our transcriptome analysis of C. velia (Supplementary Table S3B) had been previously annotated (Cvel_14639) [75]. Other CYP system-related enzymes were identified; 19 and 16 unigenes encoded for CYP and Cytb, respectively. Interestingly, six unigenes encoded for the binary homomeric (CYP−P450R) complex (Supplementary Table S3B). Similar to what we found in C. velia, the relative species V. brassicaformis has only one ER-P450R (Supplementary Table S1) and 19 CYP enzymes [75]. Phase-II-related enzymes (aldose reductase, glutathione S-transferase, and UDP-glucuronosyltransferase) were also identified in our study. Twenty and nine unigenes encoded for glutathione S-transferase and UDP-glucuronosyltransferase, respectively (Supplementary Table  S3B). We also found two different multixenobiotic resistance (MXR) transporters (MATE and ABC transporters), phase-III-related enzymes (Supplementary Table S3B). This result is especially surprising since MXR had never been described in algae before. Chloroplastid MATE efflux protein and toxin extrusion protein were encoded by six unigenes each, while the ABC transporter was encoded by four unigenes (Supplementary Table S3B). The diversity within the ABC transporter protein family, including chromerids, has been discussed before [122], while here we report the first identification of MATE transporters in chromerids. Xenobiotics detoxification systems showed different expression patterns (Supplementary Table S3B). Genes for the phase I enzymes (CYP, Cyt b, and bifunctional P-450/NADPH-P450 reductase) were up-regulated, highlighting the importance of the CYP homomeric complex (bifunctional P-450/NADPH-P450 reductase) in C. velia xenobiotics metabolism; while both up-regulation (aldose reductase and UDP-glucuronosyltransferase) and down-regulation (glutathione S-transferase) patterns were found for phase II enzymes. MXR transporters-related unigenes did not change expression after Hg 2+ treatment (Supplementary Table S3B).

Heavy-Metal Stress Biomarkers
Fifty-seven heavy-metal stress biomarkers unigenes were identified in C. velia (Supplementary  Table S3C). Forty-five unigenes (78.9%) were shared by all six libraries, and 12 unigenes were specifically expressed in a single library (Cvel_cont2) ( Table S3C). According to the unigenes' FPKM values, the expressions of the identified biomarkers did not change in 40 unigenes (88.9%), and only five were up-regulated (Figure 4).
Heme oxygenase (HO), glutathione synthase (GS), glyoxalase (GLY), and stress proteins (or heat shock proteins (HSPs)) are all involved in Hg 2+ response [44,[123][124][125][126]. Our transcriptomic analyses identified one unigene for chloroplastic HO (DN93254_c0_g1), three for GS and GLY each, and 50 for HSP (Table S3C). The differential expansion analysis showed that only one GS unigene (DN110990_c0_g1) and four HSP70 kDa (DN113480_c11_g1, DN20133_c0_g1, DN113480_c11_g1, and DN113480_c14_g1) were up-regulated after Hg 2+ treatment, while the rest of the unigenes were unaffected (Figure 4 and Supplementary Table S3C). The observed increase in GS transcript was especially interesting, since glutathione is the substrate for the synthesis of phytochelatins, the major mechanism for heavy metal chelation and detoxification in plants [127]; hence, it makes sense that C. velia responded to exposure to Hg 2+ by inducing its main detoxification system.

Phylogenetic Analyses
To provide a phylogenetic framework for interpreting our results, we constructed individual alignments for 236 FeSODs and 357 MnSODs proteins and concatenated these into a single alignment for phylogenetic analysis. Three V. brassicaformis MnSODs (Vbra_1709, Vbra_19404, and Vbra_9701), cluster with green algae and are closely related to other stramenopiles, confirming their origin in the eukaryotic nucleus. The same applies to two sequences for MnSOD of C. velia (Cvel_4244 and Cvel_26697) (Supplementary Figure S2). The rest of the chromerid FeSODs cluster within all eukaryote FeSODs, suggesting their origins in a eukaryotic nucleus. C. velia (Cvel_3019) and V. brassicaformis FeSODs sequences (Vbra_13639, Vbra_10461, and Vbra_9665) formed a sister group together with apicomplexans, while plastid C. velia (Cvel_7136) and mitochondrion V. brassicaformis (Vbra_23058) FeSODs sequences formed a sister group together with kinetoplastids and amoebozoan Entamoeba dispar, respectively. MnSODs and FeSODs showed two disconnected branches, in agreement with previous studies [55] (Supplementary Figure S2). Phylogenetic analysis based on alignments of 138 KatG proteins revealed that C. velia had duplicated KatGs (Cvel_14436; Cvel_23495), originating in a eukaryotic nucleus because it branched together with stramenopiles including oomycetes, haptophyte, cryptophyte and chlorophytes (Supplementary Figure S2). This observation agrees with previous studies proposing the transfer of the KatG enzyme among eukaryotic organisms as a horizontal gene transfer (HGT) event [112,113,128].
Nineteen CYP genes and only one P450R gene were found in each chromerid (C. velia and V. brassicaformis) (Supplementary Table S2). Targeting to organelles was predicted for all these unigenes. P450R was predicted to be targeted to the ER in both organisms (Supplementary Table S2). They also originated in a eukaryotic nucleus, because of their position in the clade composed of all eukaryotes, forming a sister group together with apicomplexan (Supplementary Figure S2). For CYPs, C. velia had eight genes for the ER, one for the plastid, four for the cytoplasm, and six for the nucleus. V. brassicaformis had 10 genes for the ER, two for the plastid, one mitochondrial, four for the cytoplasm, and two for the nucleus (Supplementary Table S2). Our phylogenetic analysis showed that the ER Chromera (DN54179_c0_g1) and the plastid Vitrella (Vbra_11766) CYPs sequences clustered with bacteria/organelles, with planctomycetes bacteria as their closest relatives. The rest of chromerid CYPs sequences originated in a eukaryotic nucleus, because of their position in the clade composed of all eukaryotes (Supplementary Figure S2).
Two HO isoforms (HO1 and HO2) were identified in V. brassicaformis, while the isoforms HO3 and HO4 were exclusively present in green plants [129]. V. brassicaformis cytoplasmic HO2 (Vbra_20700) originated in a eukaryotic nucleus, because of the position in the clade composed of all eukaryotes, forming a sister group together with apicomplexan. The other mitochondrion V. brassicaformis HO1 (Vbra_4541) and the only nuclear C. velia HO1 (DN93254_c0_g1) sequences clustered with green plants and algae within the same eukaryotic clade, suggesting their origins in a eukaryotic nucleus (Supplementary Figure S2).

Transcription Factors in Relation to Hg Stress
The assembled transcriptome showed that 56 unique TFs were differentially expressed under Hg 2+ stress in C. velia (Tables 2 and 3). TFs could trigger the activation of the ROS-related genes expression and protect cells against oxidative damage [48,[130][131][132] (Table 3). Moreover, several heavy metal-induced TFs respond to other abiotic stresses [133], suggesting that they may be crucial for the adaptation of their growth to specific environments. The TFs diversity and characteristics revealed some clues consistent with the hypothesis of endosymbioses with green and red algal symbionts in the evolutionary history of haptophytes and stramenopiles [134,135]. TF genes encoding AP2/ERF, MYB, zinc finger, and basic leucine zipper (bZIP) showed a considerable enhancement of expression in brown algae (Ectocarpus siliculosus) in response to heavy metal (copper) stress [136]. The regulation of specific members of TF families in E. siliculosus supports the hypothesis that these genes may intersect oxidative stress-related transcriptional networks [136]. Among the differentially expressed TF genes in C. velia, there were three classes (NAC, WRKY, and bZIP) involved in biological process (BP), and the rest of the TF genes were involved in cellular function process (MF) (Supplementary Table S2B). It has been previously demonstrated that MYB TF control the expression of peroxidases and ROS homeostasis, which modulates leaf cell expansion and final organ size [137]. One MYB TF (DN58459_c0_g1) was differentially expressed in C. velia, which up-regulated under Hg-stress. This result implies that the induction or repression of MYB TFs may be participating in the ROS homeostasis controlled by peroxidases. Three ERF (DN51129_c0_g1, DN105323_c0_g1, and DN107621_c0_g1) were up-regulated; while, two WRKY TFs (DN75769_c0_g1 and DN103312_c0_g1) were up-regulated and one WRKY TFs (DN77499_c0_g1) down-regulated in C. velia (Supplementary Table S2A). The results suggest that the up-regulation of WRKY and ERF TFs may be involved in the mechanism of ROS scavenging for C. velia. WRKY TFs are involved in response to abiotic and biotic stresses, including heavy metals [114,138]. The overexpression of ZmWRKY4 in maize has elevated the expression of SOD and APX under Cd stress [48]. Moreover, the activities of SOD, POD, GST, and GPX were enhanced due to the overexpression of ThWRKY7 in the transgenic plant Tamarix hispida [131]. ERF TFs could improve plant ROS tolerance by regulating the transcription of ROS metabolic enzymes [132,139]. A recent study reported that overexpression of TERF1 up-regulates the transcription of NtGPX, which may play a vital role in the regulation of ROS scavenging under Cd stress [132].
Hg exposure significantly down-regulated some TFs, including NAC (DN95341_c0_g1) and ZFPs (DN102045_c0_g1, DN102561_c0_g1, DN105281_c0_g1, and DN27427_c0_g1) (Supplementary Table  S2B). A recent study found that overexpression of ThbZIP1 increased the content of both soluble sugars and proteins, and it also enhanced the activity of both POD and SOD under stress conditions [47]. In Arabidopsis, the ZFP family were shown to positively regulate heavy metal tolerance by the regulation of the GSH1 expression [140]. Overexpression of the NAC TF family improved the osmolytes contents and reduced the H 2 O 2 and ·O 2− contents under low temperature, which contribute to improving chilling stress tolerance in transgenic tobacco [141]. Overexpression of TaNAC69-1 enhanced the expression of stress up-regulated genes such as glyoxalase protein family in transgenic wheat, suggesting that TaNAC69 is included in regulating the drought tolerance [142]. Therefore, it seems that the changes in ROS levels observed in our study, for example the increase in ·O 2− levels, may ultimately depend on the modulation of TFs expression.

Validation of the DEGs by Real-Time RT-PCR Analysis
To evaluate the reliability and validity of our transcriptome in the identified DEGs, a total of 28 DEGs were chosen and validated by RT-qPCR analysis. As shown in Figure 5, all DEGs were differentially expressed between control cultures (Cvel_cont) and cultures under Hg 2+ treatment (Cvel_mer). The expression patterns of Flavin reductase (NADPH), katG 2, and GS were significantly up-regulated by Hg exposure, and FeSOD1 and FeSOD2 were down-regulated. CYP and bifunctional P-450/NADPH-P450 reductase were strongly up-regulated by Hg stress in C. velia. Generally, RT-qPCR results were comparable with the RNA-Seq-based gene expression patterns (Supplementary Table S2B). However, MnSOD, thioredoxin, and Glutathionyl-hydroquinone reductase did not show consistent expression levels between RT-qPCR and Illumina sequencing data ( Figure 5). The discrepancies may result from different sensitivity of the two techniques.

Conclusions
This is the first study on the responses of the alveolate alga C. velia to Hg 2+ treatment. By combining physiological, biochemical, transcriptomic, and bioinformatical analyses, we were able to uncover several pathways activated in response to the heavy metal stress. In general, we observed that C. velia behaves like higher plant species in terms of ROS production, with a

Conclusions
This is the first study on the responses of the alveolate alga C. velia to Hg 2+ treatment. By combining physiological, biochemical, transcriptomic, and bioinformatical analyses, we were able to uncover several pathways activated in response to the heavy metal stress. In general, we observed that C. velia behaves like higher plant species in terms of ROS production, with a reduction in H 2 O 2 and an increase in O 2− levels at high concentrations of Hg 2+ while RNS does not seem to play a role. This is also the first transcriptomic analysis of C. velia altogether, an important organism for phylogenetic studies, and a model microalga. In total, 122,874 transcripts and 89,853 unigenes were successfully annotated in databases.
Moreover, the 1239 differentially expressed unigenes in C. velia were useful to postulate the molecular pathways involved in Hg 2+ response, such as different detoxification defense systems and TFs. For the mechanism of ROS scavenging in response to Hg stress, the POD and glutathionylation pathways were involved, which was regulated by TFs such as ERF and WRKY. The water-water cycle and PrxR/Trx pathways might be associated with its sensitivity to Hg 2+ , which was regulated by TFs, including NAC and ZFP. Moreover, the xenobiotics detoxification system shows high activity in response to Hg stress.
Importantly, variation in expression of genes involved in ROS detoxifications generally matched with our biochemical observation of ROS levels, corroborating the validity of our dual approach. Overall, our data indicate that the response of C. velia to Hg 2+ involves changes in various aspects of cell metabolism, signaling, and transport.