DNA Damage Response Pathways in Dinoflagellates

Dinoflagellates are a general group of phytoplankton, ubiquitous in aquatic environments. Most dinoflagellates are non-obligate autotrophs, subjected to potential physical and chemical DNA-damaging agents, including UV irradiation, in the euphotic zone. Delay of cell cycles by irradiation, as part of DNA damage responses (DDRs), could potentially lead to growth inhibition, contributing to major errors in the estimation of primary productivity and interpretations of photo-inhibition. Their liquid crystalline chromosomes (LCCs) have large amount of abnormal bases, restricted placement of coding sequences at the chromosomes periphery, and tandem repeat-encoded genes. These chromosome characteristics, their large genome sizes, as well as the lack of architectural nucleosomes, likely contribute to possible differential responses to DNA damage agents. In this study, we sought potential dinoflagellate orthologues of eukaryotic DNA damage repair pathways, and the linking pathway with cell-cycle control in three dinoflagellate species. It appeared that major orthologues in photoreactivation, base excision repair, nucleotide excision repair, mismatch repair, double-strand break repair and homologous recombination repair are well represented in dinoflagellate genomes. Future studies should address possible differential DNA damage responses of dinoflagellates over other planktonic groups, especially in relation to possible shift of life-cycle transitions in responses to UV irradiation. This may have a potential role in the persistence of dinoflagellate red tides with the advent of climatic change.


Introduction
Cellular DNA is continuously challenged with intracellular or exogenous agents that cause DNA damages. To preserve DNA integrity during cell division, cells have developed integrated signaling cascades linking DNA damage responses (DDRs) to cell-cycle transition, making judgement call as to damage repair per proliferation versus growth postponement, life-cycle transitions or cell death. DNA repair systems are generally conserved in nucleosomal eukaryotes.
Dinoflagellates are a major phytoplankton group in aquatic ecosystems, contributing significantly to ocean primary productivity and carbon cycling [1]. Members of the group are infamous for causing harmful algal blooms (red tides) [2], which may cause mortality or physiological impairment either due to toxin production, oxygen depletion or physical clotting of gills attributed to high biomass [3]. Proliferation of many dinoflagellates is favored with prolonged surface water stratification with the increase in seawater temperature, and global warming was predicted to expand the distribution of harmful algal blooms into higher latitudes [4]. An active cell death called paraptosis, different from apoptosis, was also observed in Amphidinium carterae under culture senescence and darkness [5], which implicated strong solar energy could differentially modulate dinoflagellate production relative to other eukaryotic groups. In addition to their ecological profoundness, dinoflagellates behold the

Direct Reversal of DNA Lesion
Alkylating agents-widely distributed reactive chemicals in intracellular and extracellular environments-react with DNA and produce various kinds of modifications on the DNA bases and backbone, leading to structure alterations and functional disruptions [44][45][46]. The alkylation attack on DNA mainly occurs at the ring nitrogen (N) and extracyclic oxygen (O) atoms of the DNA bases [46,47]. O 6 -methylguanine (O 6 -meG), a major deleterious base adduct produced from the reaction with the O 6 position of guanine, will elicit a mispair with thymine during DNA duplication, causing the transition mutation of G:C to A:T. The O 6 -alkylguanine-DNA alkyltransferase, the Ada protein in E. coli and MGMT/AGT protein in mammalians, is responsible for direct repair of this type of lesion (Figure 2A). During repair process, alkyl group of O 6 -meG is transferred to Cys residues of MGMT protein, leading to MGMT protein's inactivation and degradation [48]. N 1 -methyladenine (N 1 -meA) and N 3 -methylcytosine (N 3 -meC) are two other kinds of lesions occurred in the exposed DNA base of single-stranded DNA or replication fork. ALKB proteins, members of α-ketoglutarate/iron (II)-dependent dioxygenases, are involved in reversing these types of DNA lesions through oxidative dealkylation of the alkyl groups from N 1 -meA and N 3 -meC, leading to hydroxylmethylated products and the subsequent release of formaldehyde and the repaired base ( Figure 2B) [47,49,50]. The orthologues of MGMT and ALKB protein are found in dinoflagellates transcriptomes ( Table 2). The modified base N6-methyladenine (6mA), the most abundant modified base in eukaryotic RNAs [51,52], is also present naturally in DNA of dinoflagellates chromosomes, which was reported to accounting for 2-3% of the total nucleotides [53,54]. 6mA, a major type of modification in bacteria, is associated with restriction-modification system and discrimination between original and newly produced DNA [55]. 6mA is also conserved in eukaryotic genomes and linked with gene regulation events [56][57][58][59][60]. Both bacterial and human AlkB protein have activities for demethylating 6mA [60,61]. It remains to be further determined if dinoflagellate AlkB protein has activities towards 6mA. Alkylating agents-widely distributed reactive chemicals in intracellular and extracellular environments-react with DNA and produce various kinds of modifications on the DNA bases and backbone, leading to structure alterations and functional disruptions [44][45][46]. The alkylation attack on DNA mainly occurs at the ring nitrogen (N) and extracyclic oxygen (O) atoms of the DNA bases [46,47]. O 6 -methylguanine (O 6 -meG), a major deleterious base adduct produced from the reaction with the O 6 position of guanine, will elicit a mispair with thymine during DNA duplication, causing the transition mutation of G:C to A:T. The O 6 -alkylguanine-DNA alkyltransferase, the Ada protein in E. coli and MGMT/AGT protein in mammalians, is responsible for direct repair of this type of lesion ( Figure 2A). During repair process, alkyl group of O 6 -meG is transferred to Cys residues of MGMT protein, leading to MGMT protein's inactivation and degradation [48]. N 1 -methyladenine (N 1 -meA) and N 3 -methylcytosine (N 3 -meC) are two other kinds of lesions occurred in the exposed DNA base of single-stranded DNA or replication fork. ALKB proteins, members of α-ketoglutarate/iron (II)-dependent dioxygenases, are involved in reversing these types of DNA lesions through oxidative dealkylation of the alkyl groups from N 1 -meA and N 3 -meC, leading to hydroxylmethylated products and the subsequent release of formaldehyde and the repaired base ( Figure 2B) [47,49,50]. The orthologues of MGMT and ALKB protein are found in dinoflagellates transcriptomes ( Table 2). The modified base N6-methyladenine (6mA), the most abundant modified base in eukaryotic RNAs [51,52], is also present naturally in DNA of dinoflagellates chromosomes, which was reported to accounting for 2-3% of the total nucleotides [53,54]. 6mA, a major type of modification in bacteria, is associated with restriction-modification system and discrimination between original and newly produced DNA [55]. 6mA is also conserved in eukaryotic genomes and linked with gene regulation events [56][57][58][59][60]. Both bacterial and human AlkB protein have activities for demethylating 6mA [60,61]. It remains to be further determined if dinoflagellate AlkB protein has activities towards 6mA.   Photoreactivation, regarded as the most efficient and error-free pathway for reversal of UV-induced DNA damage, is present in bacterial, fungi, animals (except the placental mammals) and plants [62]. In plants, photoreactivation is the major and preferred pathway responsible for UV-induced lesions repair [63,64]. Photoreactivation depends on a single light-activated enzyme called photolyase to recognize and repair the photoproducts cyclobutane-pyrimidine dimers (CPDs) or 6-4 pyrimidine-pyrimidine photoproducts (6-4PPs) at lesion sites [65]. CPD photolyases and (6-4) photolyases, specific for the repair of CPDs and 6-4PPs respectively, belong to the cytochrome/photolyase family (CPF) and share similar biochemical activities on their substrates [66]. Those photolyases directly bind to damaged DNA substrates and rely on light-dependent reduction of their cofactor flavin adenine dinucleotide (FAD) to mediate lesion repair [67,68]. The activated FADHthen passes an electron to the lesions for the cleavage of the covalent bonds within pyrimidine dimers. In addition, Cry-DASH proteins, another group in CPF, are also suggested to be a group of single-stranded DNA photolyases [69][70][71]. They could bind to damaged sites of ssDNAs and have a preference for the repair of CPDs. Our bioinformatic analysis indicated putative orthologues are present in dinoflagellates ( Table 2). Dimerization of 5-hydroxymethyluracil (5hmu), the major UV-induced adduct, is theoretically impossible; as 5hmu replaces substantial amount, but not all, genomic thymines, photoreactivation repair in LCCs is likely compartmentalized. BER and MMR pathways deal with non-bulky lesions in DNA, which are glycosylase-dependent and mismatch base-pair dependent respectively. NER pathway copes with the bulky DNA lesions, which is TFIIH protein complex-dependent. Althouth detailed mechanisms are different among these three excision repair pathways, they share some similar components of DNA resynthesis, including PCNA and DNA polymerase δ.

Base Excision Repair (BER)
BER rectifies a wide range of DNA damages that modify non-bulky bases such as DNA oxidation from reactive oxygen species (ROS) attack, hydrolysis, deamination and alkylation [72][73][74][75]. Impaired DNA bases are identified and removed by DNA glycosylases, generating an abasic (apurinic-apyrimidinic, AP) site in DNA. The dinoflagellates have both mono-functional and bi-functional glycosylases ( Figure 3 and Table 3). The glycosylases targeting uracil and its derivatives including UNG [76], SMUG1 [77], MBD4 [78], TDG [77,78] and NTH1 [79], are of special interest. These genes were reported to have activities against 5hmu. The modified base 5hmu is a natural component of the DNA of dinoflagellates, which could replace 12-70% of thymine in genomes of dinoflagellates [53,54,80]. The existence of UNG, MBD4 and NTH1 implicates dinoflagellates must develop a mechanism to distinguish between the damage-induced and endogenous 5hmu, or a specific compartmentalization mechanism. The AP-site generated by mono-functional DNA glycosylases is targeted by AP-endonuclease (APE1), which then produces a single nucleotide nick, leading to the 3 OH and 5 deoxyribosephosphate(dRP) terminal in the DNA backbone. DNA polymerase Polβ is then engaged to remove the 5 -dRP group and produce 3 OH and 5 P ends. Alternatively, the bi-functional DNA glycosylases could cut the phosphodiester bond of the AP-site directly through its AP lyase activity and also create a single nucleotide nick. The nick is further converted into 3 OH and 5 P ends by extra enzymatic activities such as APE1 or polynucleotide kinase (PNKP). Later, the DNA polymerase Polβ and ligase LIG3/XRCC1 complex or ligase LIG1 is involved in gap-filling DNA synthesis and ligation sequentially. Additionally, the long-patch BER pathway is used to cope with 2-12 nucleotides lesions, in which DNA polymerases δ and ε (Polδ, Polε), FEN1, PCNA and DNA ligase I are involved [81,82]. For these steps, LIG3 and XRCC1 were not found in dinoflagellates ( Figure 3 and Table 3). No genes equivalent to LIG3 were found in many eukaryotes including most plants and budding yeast [83]. Pol λ and LIG1 were functionally replaced by Polβ and LIG3 in plant Arabidopsis thaliana [84]. A similar mechanism could be adopted by dinoflagellates.
Microorganisms 2019, 7, x FOR PEER REVIEW 9 of 39 synthesis and ligation sequentially. Additionally, the long-patch BER pathway is used to cope with 2-12 nucleotides lesions, in which DNA polymerases δ and ε (Polδ, Polε), FEN1, PCNA and DNA ligase I are involved [81,82]. For these steps, LIG3 and XRCC1 were not found in dinoflagellates ( Figure 3 and Table 3).
No genes equivalent to LIG3 were found in many eukaryotes including most plants and budding yeast [83]. Pol λ and LIG1 were functionally replaced by Polβ and LIG3 in plant Arabidopsis thaliana [84]. A similar mechanism could be adopted by dinoflagellates.

Nucleotide Excision Repair
Nucleotide Excision Repair (NER) is a multi-step process that recognizes and removes bulky structurally unrelated DNA lesions, such as CPDs and 6-4PPs induced by UV irradiation, endogenous oxidative DNA damage, and DNA adducts formed by environmental and chemical mutagens [85][86][87][88][89][90]. NER consists of two pathways: the global genome NER (GG-NER) and the transcription-coupled NER (TC-NER). GG-NER occurs in all regions of genomic DNA, while TC-NER is dedicated to repairing lesions that block the transcription of active genes. The two pathways adopt different strategies to initiate DNA damage recognition but share the same sets of proteins in the later stages of the repair process. In mammalian cells, with the help of UV-damaged DNA-binding complex composed of DDB1 and DDB2 subunits, XPC-Rad23B complex (with CETN2) is responsible for the recognition of the DNA lesions in GG-NER pathway [87,91]. Stalled RNA polymerase activates the TC-NER pathway and recruits CSA and CSB proteins to remove them at lesion sites [92,93]. TFIIH protein complex is involved in the next steps of the GG-NER and TC-NER pathways for DNA lesion verification, helix unwinding and incision. TFIIH comprises of 10 subunits and can be divided into a core complex consisting of XPB, XPD, p8, p34, p44, p52 and p62, and a cyclin-activated kinase (CAK) complex containing MAT1, CDK7 and cyclin H [94][95][96]. The helicase and ATPase activities of XPB and XPD are required for unwinding the DNA into a bubble-like structure at the lesions. Later, the pre-incision complex consisting of XPA, RPA, and XPG is formed at the damaged DNA sites, which help to recruit endonucleases. Incision action is thought to be initiated by the 5 incision complex ERCC1-XPF, followed by 3 incision of XPG, creating a single-strand DNA break. DNA polymerase δ, ε, and κ (Polδ, Polε and Pol κ), with the assistance of PCNA and RFC, are required for the subsequent DNA repair synthesis. DNA ligase LIG1 or LIG3/XRCC1 complex are involved in the final gap filling after DNA repair synthesis.
Except DDB2, XPA, the subunits of TFIIH complex including GTF2H1 and GTF2H5, and the RPA3 subunits of RPA complex, the other proteins involved in NER pathway were identified in dinoflagellates ( Figure 4 and Table 4). XPA, an intrinsically unstructured protein, is a crucial component of the pre-incision complex for DNA damage recognition [97]. No putative orthologues were found in plants, although they have the ability to remove UV photoproducts in NER dependent pathway [98]. Only three subunits of TFIIH complex were found in L. polyedrum in an analysis of basal transcriptional factors [99]. The absence of putative XPA, RPA3 and GTF2H1 orthologues were also reported in Trypanosomatids [100].      Mismatch Repair DNA mismatch repair is a DNA repair pathway specializing in repairing mismatched bases during DNA replication, process mismatch-containing heteroduplex during recombination, and correct DNA lesions resulting from either endogenous or exogenous sources [101][102][103][104]. In bacteria, MutS and MutL homodimers are responsible for the initial DNA mismatch recognition, followed by the endonuclease MutH's nick on the newly synthesized strand. The nicked strand is further excised by the exonuclease and re-synthesized by DNA polymerase [105,106]. Similar eukaryotic mechanisms are also applied for the recognition and repair of DNA mismatches. MutS homologs-membered heterodimers MSH2-MSH6 (MutSα) or MSH2-MSH3 (MutSβ) are used to scan DNA mispair. MutSα has a preference of binding single base mismatches and 1 or 2 base insertions or deletions (indels), while MutSβ is capable of recognizing both base mispairs and small/large indels [101]. The MutL homologs also form heterodimers MLH1-PMS2 (MutLα, also known as MLH1-PMS1 complex in budding yeast cells), which are then recruited to form a ternary complex with MutS homologs to initiate the incision at lesion sites. PCNA protein is thought to activate the endonuclease activities of MutLα and help to distinguish the two DNA strands. Exonuclease 1 (Exo1) subsequently cleaved the nicked mispaired strands and the resulting gap is filled by DNA polymerase Polδ and DNA ligase LIG1. Except MSH3, which is also absent from the apicomplexan Plasmodium [107], all the other orthologues are present in dinoflagellates ( Figure 5 and Table 5). Detailed mechanism without heterodimer MutSβ awaits further functional investigation. Microorganisms 2019, 7, x FOR PEER REVIEW 21 of 39

DNA Double-Strand Breaks Repair
DNA double-strand breaks (DSBs) are one of the most cytotoxic chromosome lesions in which breaks occurs in both parallel strands of duplex DNA. DNA DSBs could be caused by many endogenous and exogenous insults including chromosome replication errors, unintended breakage by nuclear enzymes, reactive species during normal metabolism, ionizing radiation and certain chemical drugs [108]. It has been estimated that DSBs happened spontaneously at a frequency of about one per 10 8 base pair per cell cycle for many organisms [109]. Additionally, a single unrepaired DNA DSB could lead to cell unviability [110].
Two major distinct pathways are involved in the repair of DSBs: non-homologous end joining (NHEJ) and homologous recombination (HR). NHEJ could occur in the whole cell cycle and particularly during the G1 phase. HR is mainly restricted to in the mid-S and G2 cell-cycle stages, when the nascent sister chromatids are available for recombination repair template.

Non-Homologous End Joining
Non-Homologous End Joining (NHEJ) refers to a type of DNA DSB repair that requires little or no sequence homology (≤4 nucleotides) to re-join two DSB ends [111,112]. Binding the DSB ends with the heterodimer Ku protein of Ku70 and Ku80 serves as a loading platform to promote recruitment of other NHEJ proteins [113,114]. Compatible DNA termini such as blunt-ends could be directly repaired with the binding of XRCC4-DNA ligase IV to those termini. For non-compatible ends, the DNA-dependent protein kinase catalytic subunit (DNA-PKcs) is engaged in a complex with DNA nuclease Artemis to process these ends [115,116]. In the complex with DNA-PKcs, Artemis acquires endonuclease activity on the DNA hairpin, and both the 5' and 3' DNA overhangs. Other proteins including aprataxin, APLF and PNK could also execute its function in the modification of these non-compatible ends [117][118][119][120]. Pol X family DNA polymerase, Pol4 in budding yeast [121], PolL and PolM in mammalian cells [122,123], are involved in the nucleotide synthesis during the processing of the ends. In addition, end processing is a repetitive progress, of which modifications could be performed for multiple rounds, leading to a small sequence deletion or insertion [124]. Finally, the ligase IV complex including XRCC4-LIG4 and XLF is responsible for the ligation of the reconstituted compatible ends [125][126][127].
Key players involved in NHEJ, such as Ku70, Ku80, DNA-PK and LIG4 were identified, but orthologues of XLF, XRCC4 and Artemis were not found in dinoflagellates transcriptomes (Table 6). Fission yeast XRCC4 gene is highly divergent from human orthologues and has only been found in recent years [128], whereas XLF orthologues were not found in plants genomes [129]. The lack of full NHEJ pathway was reported in apicomplexan Plasmodium, with the absence of Ku70, Ku80, DNA-PK and LIG4 [130].

Homologous Recombination
Homologous Recombination (HR) refers to a high-fidelity DSB repair mechanism with the use of a homology sequence as the repair template. In HR, extensive 5 to 3 resection of one DNA strand is required at first to produce 3 -OH ended ssDNA tails after DSB formation. DNA termini resection is initiated by the combined action of Mre11-Rad50-NBS1 (MRN) complex and nuclease CtIP, creating a short stretch of ssDNA. Extensive resection is then carried out by additional nucleases and exonucleases including CtIP, DNA2, BLM and EXOI [131,132]. The length of the extensively resected ssDNA tail could range from hundreds to thousands of nucleotides [131]. The RPA protein then binds to the resulting resected ssDNAs to protect it from nucleolytic degradation and formation of secondary structures. In order to proceed to recombination, recombinase Rad51, the central player of HR, is needed to be loaded onto the RPA-coated ssDNA. Mediator proteins, such as Rad52 in yeast and BRCA2 in mammalian cell, are responsible for promoting the Rad51 nucleofilament formation through the displacement of RPA. Other proteins including Rad51 paralogs and Rad54 help to stabilize the Rad51 filaments [133]. Additionally, negative regulators, e.g., DNA helicase Srs2 in yeast, recQ5, BLM and FANCJ in mammalian cells, could suppress Rad51 function via the disassembly of Rad51-ssDNA filaments [134]. Following nucleofilament formation, the recognition of homology sequences and strand invasion ensues, generating a structure called displacement loops (D-loops), primed for DNA repair synthesis from the invading 3'-end ssDNA [135].
In canonical Double-Strand Break Repair (DSBR), the other ssDNA end pairs with the displaced template strand and forms a double Holiday junction (DHJ). The resolution of DHJs could lead to either cross-over or non-crossover recombination products [136]. Alternatively, DHJs formation is inhibited in the synthesis dependent strand annealing (SDSA) mode of HR, in which the D-loops are disrupted after the limited DNA synthesis from the invading 3 -end ssDNA. The displaced 3 -end ssDNA then recombine with the complementary strand of the second 3 -end ssDNA tail, followed by repair DNA synthesis, leading to the formation of non-crossover products. SDSA is the preferred recombination pathway during mitosis [134].
HR pathway, members of which are also important for meiosis, is uninvestigated at the mechanistic level in dinoflagellates. The essential components of HR including Rad51, MRN complex (MRE11-NBS1-RAD50) and RPA (except RPA3 subunit) could be found in their transcriptomes ( Table 7). The mediator protein Rad52 was absent in Arabidopsis thaliana, Drosophila melanogaster, Caenorhabditis elegans, Plamodium [130] and Trypanosomatids [100], as in the case for dinoflagellates. A putative homolog BRCA2 was found in Symbiodinum, which could function as mediator for Rad51. Other putative orthologues such as RMI2, DNA2, SLX4 and EME1 were absent possibly due to the low sequence conservation in dinoflagellates. The recombinase RAD51 responsible for catalyzing the homology search and strand change is the essential protein in the HR pathway [137,138]. Comparative analysis with other eukaryote RAD51 orthologues showed that it contains the canonical Walker A and Walker B motif of the RECA/RAD51 superfamily ( Figure 6). A phylogenetic tree constructed with selective eukaryotic orthologues exhibited dinoflagellate RAD51 orthologues forming an individual clade ( Figure  S2). The prokaryotic RecA orthologues, which are involved in maintaining the integrity of chloroplast and mitochondria genome [139,140], are also present in S. minutum and L. polyedrum transcriptomes.

Translesion Synthesis
Translesion Synthesis (TLS)-a highly conserved pathway which repairs DNA damages associated with replication-directly mediates duplication of damaged DNA in the template strand that was blocked [141][142][143]. TLS is mediated by the specialized DNA polymerases that could tolerate DNA lesions. These specialized DNA polymerases, including DNA polymerase REV1, Rev3L, PolH, PolI and PolK, have active sites that accommodate impaired or distorted templates [143][144][145]. Execution of TLS involved the switching of canonical polymerases with TLS polymerases, which is thought to be conducted by the interaction with PCNA and Rev1 [146,147]. The low fidelity of TLS polymerase, orthologues of which are found in dinoflagellates transcriptomes (Table 8), makes the repair progress intrinsically error prone [148].

Translesion Synthesis
Translesion Synthesis (TLS)-a highly conserved pathway which repairs DNA damages associated with replication-directly mediates duplication of damaged DNA in the template strand that was blocked [141][142][143]. TLS is mediated by the specialized DNA polymerases that could tolerate DNA lesions. These specialized DNA polymerases, including DNA polymerase REV1, Rev3L, PolH, PolI and PolK, have active sites that accommodate impaired or distorted templates [143][144][145]. Execution of TLS involved the switching of canonical polymerases with TLS polymerases, which is thought to be conducted by the interaction with PCNA and Rev1 [146,147]. The low fidelity of TLS polymerase, orthologues of which are found in dinoflagellates transcriptomes (Table 8), makes the repair progress intrinsically error prone [148]. DNA interstrand cross-link (ICL), forming covalent bond between two opposite strands of DNA, can be generated from several sources including bi-functional alkylating agents (such as nitrogen mustard), by-products of lipid peroxidation, abasic sites, and natural psoralens [149]. ICLs prevent complimentary DNA strands separation and thus will impose damages at DNA replication and transcription, making it one of the most toxic DNA damages. In eukaryotes, ICL repair occurs through different mechanisms for non-dividing (G1 phase) and dividing cells (S or G2/M phase) [150][151][152]. However, both mechanisms share similar steps, which include nuclease-mediated detachment from one DNA strand, coupled with TLS polymerase-dependent synthesis across the ICL-containing DNA region, rendering a complete DNA template to finish the repair.
Fanconi anemia is a rare genetic disease associated with the mutation of one of the 19 known FANC genes [153]. In cooperation with NER, TLS and HR pathway, the FANC proteins play important roles in signaling and repair of the replication-dependent ICLs [152,154,155]. ICLs recognition is mediated through binding of FANCM to the damaged sites, which function as a landing platform for the recruitment of heptameric FANC core complex (FANCA, FANCB, FANCC, FANCE, FANCF, FANCG and FANCL). The FANC core complex further interacts with many other proteins including other FANC proteins and repair factors to repair the ICLs. It should be mentioned that the full Fanconi anemia pathway genes could to be only found in mammals but not in other organisms. In the yeast Saccharomyces cerevisiae and the plant Arabidopsis thaliana, a partial Fanconi pathway associated with FANCM was used to repair the ICLs [156,157]. Surprisingly, none of the FANC core complexs, FANCM, and FANCM accessory factors MHF1 and MHF2, were identified in dinoflagellates transcriptomes (Table 9), although we are not certain if their levels at vegetative life cycles may be too rare for mRNA isolation. The Snm1 family of widely conserved proteins, encoding metallo-b-lactamases (MBL) with nuclease activities, is believed to function in the end procession of the detached ICLs [158,159]. Snm1 deletion in budding yeast showed a hypersensitivity to ICL reagents psoralen and nitrogen mustard [159]. Mammalian cells have three SNM1 orthologues, SNM1A, SNM1B and SNM1C [154], although only SNM1A is thought to be the functional homolog. Both SNM1 and SNM1B were identified in dinoflagellates transcriptomes (Table 9).

Conclusions and Perspectives
Eukaryotic DNA damage responses (DDR) evolved from-and build on-prokaryotic counterparts, with modifications for nucleosomal accessibility; whether there is a reversal of evolution in dinoflagellates would not only be of interest in evolutionary biology, and potentially bears a biotechnology element for up-scale synthetic biology. We explored the presence of putative orthologues of DDRs genes in three dinoflagellate transcriptomes. We found most of orthologues of DNA damage checkpoint signaling networks, DNA repair pathways including direct reversal of DNA lesion, photoreactivation, BER, NER, mismatch repair, and NHEJ, homologous recombination repair and translesion synthesis, with one exception of Fanconi pathway required for repair of interstrand crosslinks. We speculate that this may be attributed to different forms of DNA damage in anisotropically-aligned supercoiled domains.
Our current knowledge of dinoflagellate DDR is still in its infancy, with a paucity of investigations addressing the specific effect on cell-cycle control, which will directly impact dinoflagellate productivity, and could potentially overlap with the effect of photo-inhibition. The present study represents a starting point for further research in the molecular biology of dinoflagellate DNA damage responses. How DDR coordinates with organelle genome damage processing, in which nuclear genomes seem to have disproportionate representation (especially plastid minicircle) will need to be addressed. In nucleosomal eukaryotes, core histones have major roles in DDRs [160,161]. Their substantially lower expression levels, as well as the lack of nucleosomal nuclease resistance pattern, will have major implications to the molecular mechanism of DNA repair.