Dinoflagellate genomics landscape and incredible inter-species dissimilarities

To date, 28 dinoflagellate genomes have been sequenced (with two sequenced more than once), mostly from the family Symbiodiniaceae (Table ​(Table1).1). Sequenced taxa other than Symbiodiniaceae included Amoebophrya sp. ex Karlodinium veneficum [9], Amoebophrya ceratii [10, 11], Amphidinium gibbosum [12], Hematodinium [20], Polarella glacialis [21], and Prorocentrum cordatum (formerly P. minimum) [22]. These taxa cover a wide range of environmental conditions and different lifestyles, despite the obvious bias toward small end of the dinoflagellate genome size spectrum. Species from the Symbiodiniaceae are endosymbionts of tropical corals or their free-living counterparts. Amoebophrya species are intracellular parasites, infecting other dinoflagellates. Amphidinium gibbosum and P. cordatum form harmful algal blooms and A. gibbosum is toxigenic. P. glacialis thrives in the Antarctic and Arctic, can survive temperature shock to 15ºC [26] and DNA barcoding suggests that this and closely related species occur in the temperate estuary Long Island Sound [27].

Most of these genome assemblies were based on Illumina Hi-Seq technology and hence are largely fragmentary. However, using PacBio to generate long sequence reads and Hi-C to establish linkages between scaffolds, chromosome level assemblies have been achieved for part of the F. kawagutii genome (>120Mbp; [19]) and the genomes of Symbiodinium microadriaticum [28] and Breviolum minutum [29].

The number of chromosomes varies vastly between species, ranging from several to more than 200 [30]. By microscopic count of DAPI-stained chromosomes and telomere end count (divided by 2), F. kawagutii has between 7 [31] and 11 chromosomes [18]. In sharp contrast, there are 94 chromosomes in S. microadriaticum [28] and 91 in B. minutum [29]. While the number for S. microadriaticum is comparable to a previous microscopic estimate (100; [32]), that for B. minutum exceeds five-fold the microscopic count of ~17 (Fig. ​(Fig.11 in [13]). This inter-species disparity is remarkable given that the genome size among these species differs by no more than 2-fold. Equally striking are their compensating differences in chromosome length. The chromosome length is markedly shorter in S. microadriaticum (16Kb to 16.6 Mb, median 6.6 Mb; [28]) and in B. minutum (median 6.7 Mb, longest~11 Mb; [29]) than in F. kawagutii (maximum>120 Mb; [19]). These data suggest an interesting evolutionary trend of chromosome number decrease and chromosome length increase from the basal S. microadriaticum to later-diverging F. kawagutii [33]. Based on the existing data, the small non-parasitic dinoflagellate genomes such as those of Symbiodiniaceae contain approximately 1–3 Gbp DNA distributed in 10–100 chromosomes of 100−10 Mbp each; the largest dinoflagellate genome contains approximately 250 Gbp DNA likely distributed in 200–1000 chromosomes of 1.25 Gbp-250 Mbp each.

Genome sequencing combined with fluorescent in situ hybridization (FISH; Fig. ​Fig.2A,2A, B) revealed that the telomere sequence in F. kawagutii consists of repeats of TTTAGGG, depicted as (TTTAGGG)_n, which is identical to the typical plant telomere (Richards and Ausubel 1988), as is also true for other dinoflagellates [34, 35]. The length of the telomere tract was estimated at 25–80 Kbp in Karenia papilionacea and Crypthecodinium cohnii [35] but is unclear in other species. Using non-denaturing fluorescent in situ hybridization, Cuadrado et al. (2022) [36] revealed bipolar distribution of clustered multiple foci of telomeric DNA: one in the nucleolus organization area and the other peripherally along the convex side of the nucleus, suggesting that the telomeres change their positions during cell division.

Open in a separate window

Fig. 2

Distribution pattern of telomeres in a tertiary endosymbiosis dinoflagellate and schematic of TADs in dinoflagellate chromosomes and associated gene distribution and expression machinery. A Chromosomes of Karenia brevis stained with fluorescent in situ hybridization of telomere probe. B DNA DAPI stain (Blue) and telomere FISH (red) in F. kawagutii. C TAD organization highlighting gene distribution and gene expression machinery. Reproduced with permission from Cuadrado et al. 2019 [37] (A), Lin et al. 2015 [18] (B), and Lin et al. 2021 [33] (C)

Dinoflagellate genomes are highly divergent. This is reflected in the genomes of Symbiodiniaceae, among which a>98% dissimilarity is consistently found by mapping reads from one genome to the assembly of the other genome [18, 23]. Whether this extreme divergence reflects rapid genome evolution in dinoflagellates in general or is due to severe isolation between Symbiodiniaceae species by their facultative symbiotic lifestyle remains to be investigated.

Topologically associated domains (TAD)

Dinoflagellate genomes form topologically associated domains (TADs; Fig. ​Fig.2C),2C), a feature widely found in eukaryotic genomes [28, 29]. TAD is a genomic region where DNA sequences within the same TAD interact with each other more frequently than with sequences outside of it [38]. Between the two species that have been examined, S. microadriaticum has about 450 TADs with a median size of 1.8 Mb, whereas B. minutum contains 580 TADs with a median size of 5.7 Mb. These are markedly larger than mammalian TADs (~0.18 Mb) and bacterial CIDs (~0.17 Mb), and do not show interactions with more distant regions of the chromosome in the Hi-C maps.

In these two genome assemblies, genes are distributed along the chromosomes as an alternating series of unidirectional clusters [28, 29]. TADs show two types of strand switch regions (SSR) (Fig. ​(Fig.2C).2C). One, where values of the DNA contact probability plot are closest to the diagonal axis in Hi-C interaction maps, corresponds to the end of transcription (a convergent strand switch region, cSSR). The other, where regions of the TADs show DNA interactions furthest from the diagonal corresponds to the initiation of transcription (a divergent SSR, dSSR). Roughly half the genes in the genome are found in groups of at least nine, all sharing a common direction. TAD boundaries are also associated with decreased GC content.

The dinoTAD structure suggests that transcriptional promoters might be found in the central region of the TADs where genes are oriented outward in both directions (Fig. ​(Fig.2C).2C). These sites could represent histone attachment points, reconciling the conservation of histone sequences with their paucity [39].

The average interaction frequency between two regions of the chromosome decreases when the distance between them increases. The trend in S. microadriaticum [28] suggests that DNA in this species is organized as linear rods. Comparison to similar analyses with other genome organizations shows that although dinoflagellate nuclei presumably use viral nuclear protein (DVNP) and bacterium-originated histone-like proteins more than histones (due to their higher expression levels), the chromosome organization appears more like eukaryotic chromosomes, with the highest similarity to mitotic chromosomes in metaphase when the chromosomes are highly condensed [28, 33].

Marinov et al. (2021) [29] and Nand et al. (2021) [28] noted that when the transcription was blocked, the TADs disappeared, revealing transcription’s critical role in maintaining the TADs in dinoflagellate chromosomes. This aligns with earlier findings on RNA’s importance in chromosomal DNA structure [40], though recent data suggest transcription activity may be more crucial than RNA in stabilizing domain structure.

Mechanisms of genome expansion: Duplication, (retro)transposition, and repetitive DNA

Dinoflagellate genome sizes vary widely among species, from <0.1 Gbp in the parasitic taxon Amoebophrya, <3 Gbp for most coral symbionts in the family Symbiodiniaceae, to ~250 Gbp for free-living taxa [10, 41, 42]. There is no phylogenetic trend in the vast genome size variation, however (Fig. ​(Fig.3A).3A). For example, the coral symbiotic lineage Symbiodiniaceae has the smallest genomes among documented dinoflagellates despite its later-diverging position in the phylogenetic tree [43]. The large genome sizes and wide range among dinoflagellates suggest multiple independent events of genome expansion in evolution. Alternatively, genome contraction might have occurred occasionally, independent of symbiosis lifestyle [17].

Open in a separate window

Fig. 3

Evolutionary dynamics of dinoflagellate genome size showing no phylogenetic trend (A) and double peak in 21-mer profile of Polarella glacialis showing evidence of genome duplication (B). A Vertical arrow indicates phylogenetic trend from the dinoflagellate sister group perkinsids, basal dinoflagellate Oxyrrhis, to later diverging taxa. Size of oval in the middle illustrates genome size (not to scale) as depicted by numbers shown on the right. B Reproduced from Stephans et al. 2020 [21] under the terms of the Creative Commons CC BY license

Whole genome duplication (WGD) is the most efficient way to achieve chromosome multiplication and genome expansion [44, 45]. The immense dinoflagellate genomes might have resulted from WGD. However, K-mer profile and scaffold synteny (proxy of chromosome synteny) analyses for F. kawagutii did not show any signs of WGD [18]. Nevertheless, the genome of Polarella glacialis (~3 Gbp) is about double the size of that in Symbiodiniaceae [21]. Its k-mer profile, based on sequencing reads, shows double peaks characteristic of diploid genomes [21], suggesting a genome duplication (Fig. ​(Fig.3B).3B). “Diploid”-like chromosome pairs have been reported for Pyrocystis lunula as well [46]. There is a good chance that WGD is widespread in large dinoflagellate genomes. However, the original interpretation of these observations as diploid is not consistent with the traditional view that dinoflagellates are haploids.

Besides WGD, genome expansion can result from processes like the proliferation of mobile DNA elements, including transposons and retrotransposons. The role of transposable elements in remodeling genomes has been demonstrated in plants [47]. Indeed, there is evidence that transposons [10] and retrotransposons [48] play roles in dinoflagellate genome expansion. The role of retrotransposition is corroborated by the prevalent gene paralogs and DinoSL (complete or truncated) at the upstream region of genes, which account for >20% of total genes [49], made possible by the large families of reverse transcriptase and integrase genes in the genomes [18, 28]. In contrast, transposition is rare in the small genome (<0.1 Gbp) of the parasitic species Amoebophrya ceratii [10]. González-Pech et al. (2021) [23] compared genome features of S. tridacnidorum and S. natans and postulated that retrotransposition might be a major driving force for the two genomes to diverge.

Stephens et al. (2020) detected a substantial proportion of long terminal repeat (LTR) elements (~12%) in P. glacialis genomes [21]. Transposable elements (such as LTRs) can comprise up to 80% of plant genomes and are induced by genome shock and polyploidization, leading to genome restructuring [50]. The abundance of LTRs in P. glacialis and their role in genome restructuring may explain the difference in genome size between populations. In S. microadriaticum, heat stress-activated Ty1-copia-type LTR retrotransposons were detected, and recent expansion was documented [51], suggesting proliferation of retrotransposable elements for adaptation to hot habitats. The survival of retroposed genes may be promoted by their methylation [52], as DNA methylation is thought to silence and promote transposition [53]. Similar TE domestication by epigenetic regulation seems common in eukaryotes [54], and methylation of transposable elements is presumably a major driving force of genome expansion in all eukaryotes [55].

The abundance of repetitive DNA also contributes to genome size variation. Using FISH, Cuadrado et al. (2022) [36] revealed clusters of dense trinucleotide repeats in the bean-shaped nuclei of Karenia selliformis and K. mikimotoi. Larger dinoflagellate genomes tend to contain more repetitive elements [21]. For instance, repetitive elements account for 23.8% of the 0.1156-Gbp Amoebophrya A120 genome, 13.1% of the 0.1160-Gbp Amoebophrya A25 genome, 27.9% of the ~1.10-Gbp Symbiodinium microadriaticum genome, and 16% of the ~1.2-Gbp Fugacium kawagutii genome [14]. For P. glacialis, with genome sizes of ~3 Gbp, the level of repetitive elements increases to ~68% [21]. Crypthecodinium cohnii has a~7 Gbp genome [56], with 55–60% composed of repetitive DNA [57]. A survey of the ~115 Gbp Alexandrium ostenfeldii genome indicated that repetitive DNA accounted for ~58% of the genome [58]. Overall, these results suggest that repetitive elements significantly contribute to genome size evolution in dinoflagellates.

Genomic innovation by rampant horizontal gene transfer (HGT)

Horizontal gene transfer (HGT) might be a significant driver of gene innovation in dinoflagellates [59]. HGT is widespread in eukaryotes, accounting for 0.1-6.4% of total genes, depending on the species [60]. While the full extent of HGT in dinoflagellates awaits the sequencing of large-genome species, current data provide ample evidence. Using EST data, Norsenko and Bhattacharya (2007) [61] identified 16 proteins acquired through ancient HGTs in the common ancestor of the genera Karenia and Karlodinium, and one protein from a more recent HGT. Eight of these proteins appear to have resulted from independent HGTs in several eukaryotic lineages, with many involved in energy metabolism. Most of these genes were transferred directly from bacteria, but some originated from another chromalveolate that had acquired them from bacteria. Some HGT-derived genes serve essential functions in dinoflagellates and warrant further discussion.

Genome sequencing for both the Antarctic (CCMP1383) and the Arctic (CCMP2088) strains of Polarella glacialis, compared with other dinoflagellate genomes, revealed putative bacteria-to-dinoflagellate HGT for ice-binding domain-containing proteins [21]. Several bacterium-originated genes were identified in the plastid genomes of Pyrocystis lunula and Ceratium horridum, including FtsY, Rpl28, Rpl33, Yef16, and Ycf24 [62]. F. kawagutii genome analysis also revealed 56 putative bacterium-originated HGT genes (0.12% of total gene models), of which 22 are flanked by eukaryotic genes on at least one side [18]. These included the erythromycin esterase superfamily gene, radical SAM protein, and Phosphoadenosine phosphosulphate (PAPS) reductase. Erythromycin esterases (IPR007815) disrupt erythromycin by hydrolyzing the macrolactone ring, thereby conferring antibiotic resistance (InterPro). Radical SAM proteins (IPR007197) catalyze diverse reactions, such as unusual methylations, isomerisation, sulfur insertion, ring formation, anaerobic oxidation, and protein radical formation in biosynthesis and biodegradation pathways of DNA precursors, vitamins, cofactors, antibiotics, and herbicides (InterPro). PAPS reductase (IPR002500) is part of the adenine nucleotide alpha hydrolases superfamily, which includes N-type ATP PPases and ATP sulfurylases; this enzyme uses thioredoxin as an electron donor for the reduction of PAPS to phosphor-adenosine-phosphate (PAP) (InterPro).

Loss of nucleosome and gains of nuclear proteins: A nuclear drama like no other

Chromosome structure is influenced by DNA binding proteins, cations, and RNA. Mg²⁺ and Ca²⁺ are important in DNA configuration [85–87]. RNA may also be involved in DNA packaging [40, 87]. Abundant RNA binding proteins were detected in the extracts of purified chromatin by high throughput proteomic analysis in Lingulodinium polyedra [88].

Dinoflagellate nuclei have a 10-fold lower nuclear protein to DNA ratio (1:10) than typical eukaryotes (1:1) and lack nucleosome (Rizzo 1991) despite presence of core histones [39, 68]. Recent studies suggest low amounts of nucleosomes may occur [89, 90]. Dinoflagellates have acquired alternative proteins through HGT. One of these proteins is the major basic nuclear protein (MBNP), the coding gene of which is one of the most highly expressed genes in dinoflagellates [2]. Two related proteins have abundantly reported: histone-like proteins (HLP) and bacterial DNA binding protein (HU)-like protein (Hcc) [91, 92]. Phylogenetic analysis (Fig. ​(Fig.4)4) suggests that MBNP, HLP, and HCc belong to the same protein family, thus warranting a unifying name. MBNP appears to be the most suitable to represent this gene family.

Open in a separate window

Fig. 4

Phylogenetic tree of major basic nuclear protein (MBNP), histone-like protein (HLP), and bacterial DNA binding protein (HU)-like protein (Hcc). The tree is rooted with bacterial DNA binding protein (HU). The lack of monophyletic separation indicates that these three proteins belong to the same protein family under different names. Species name abbreviations: Abor, Alcanivorax borkumensis; Acar, Amphidinium carterae; Bmin, Breviolum minutum; Cbur, Coxiella burnetiid; Ccoh, Crypthecodinium cohnii; Cgor, Cladocopium goreaui; Fkaw, Fugacium kawagutii; Gcat, Gymnodinium catenatum; Kven, Karlodinium veneficum; Nsci, Noctiluca scintillans; Pgla, Polarella glacialis; Pmin, Prorocentrum minimum; Ppis, Pfiesteria piscicida; Pshi, Prorocentrum shikokuence; Rbal, Rhodopirellula baltica; Sacu, Scrippsiella acuminata; Smic, Symbiodinium microadriacticum; Snat, Symbiodinium natans; Stri, Symbiodinium tridacnidorum

Another acquired protein is dinoflagellate/Viral nucleoproteins (DVNP). DVNP acquisition coincided with nucleosome disappearance early in dinoflagellate evolution [83]. Heterologous expression of a DVNP in Saccharomyces cerevisiae showed that DVNP expression inhibited growth, diminished histones, disrupted nucleosomes, and impaired transcription, detrimental effects that were relieved by histone depletion [93]. These results suggest that histone abandonment might be due to the ‘colonization’ of the ‘invading’ viral protein.

Dinoflagellates appear to have lost histone H1 [94]. HCc exhibits about 10% similarity to H1, implying that HCc might be a functional replacement of H1 [95]. The similarity could cause misidentification of HCs as H1, which can explain reports of H1 presence [89]. Stringent in silico studies suggest that H1 is absent in dinoflagellates [94]. Experimental evidence is required to nail the issue.

Gene structure: Unidirectionality and intron dynamics

Unlike typical eukaryotes, dinoflagellate genes are dominantly unidirectionally arranged in the genome, forming co-oriented clusters [13, 18, 22]. This characteristic is widespread in dinoflagellates, including basal Amoebophrya spp [11]. Such organization might facilitate transcription in unison and in foci on certain areas of the TADs [28, 29, 33].

Contrary to previous belief, dinoflagellate genes are not intron-depleted. Even in streamlined parasitic Amoebophrya ceratii genomes, 67-71% of genes contain introns, with 64–99% in Symbiodiniaceae species (Table ​(Table1).1). The ~6.4 Gbp (A) gibbosum genome has fewer introns per gene but longer total intronic regions compared to smaller dinoflagellate genomes. Average intron lengths vary: 335–345 bp in Amoebophrya spp., 505, 517, and 893 bp for S. microadriaticum, (B) minutum, and F. kawagutii, respectively [11]. Notably, a 26,372 bp intron was detected in (C) cohnii [96].

Intron-exon boundaries in dinoflagellates also differ from typical eukaryotes [11]. Besides the canonical GT-AG boundary, dinoflagellates use GCIGA-AG (25–74% in Symbiodiniaceae; <3% in Amoebophrya), and other boundaries (up to 67% in Amoebophrya). Interestingly, 99.98% of introns in Amoebophrya strain AT5 have canonical GT-AG splice sites, while over 60% in strains A25 and A120 have non-canonical splice sites.

Intron creation may be driven by introners (specialized transposable elements). Analysis of five dinoflagellate genomes revealed: (1) evidence of historical intron creation, (2) recently active introners in four of five species, (3) dynamic intron gain and loss, with higher gain rates than most eukaryotes, and (4) In P. glacialis, 12,253 introns from recent introner insertion, with 15 distinct families accounting for at least 100 introns each [97].

Data to date show that dinoflagellates use standard codons. However, exceptions have been reported, mostly in the stop codon. In Amoebophrya sp. ex Karlodinium veneficum, as demonstrated in dynein heavy chain genes, all three stop codons are inferred to code for amino acids [9]. UAA and UAG appear to encode glutamate whereas UGA either serves as the stop codon or encode tryptophan.

High gene content, complex organization, and abundant “dark” genes

High gene content

A predictive correlation between protein-coding genes and genome size was developed in Hou and Lin (2009) [98]: log (gene number)=LN (-46.200+22.217*log[genome size]), where genome size is in Kbp. This empirical model predicts 38,188 to 87,688 protein-coding genes for dinoflagellate genomes of ~2 to 250 Gbp. However, recent genome data show higher gene numbers than predicted in most cases (Table ​(Table2),2), suggesting dinoflagellate genome expansion disproportionately favors gene increase compared to other eukaryotes.

Table 2

Comparison of identified gene number with predicted gene number based on Hou and Lin (2009) [98]

Species	Genome size (Mbp)	Detected gene no.	Model-predicted gene no.	% deviationa
Parasitic
Amoebophyra ceratii AT5.2	120	19,925	15,834	26%
Amoebophyra sp. A25	116	28,091	15,656	79%
Amoebophyra sp. A120	115.5	26,441	15,633	69%
Hematodinium sp.a	4800	29,510	42,412	-30%
Symbiotic
Breviolum minutum	616	41,925	25,830	62%
Cladocopium goreaui	1,200	45,322	30,710	48%
Durusdinium trenchii	670	30,000	26,419	14%
Fugacium kawagutii	935	45,192	28,828	57%
Symbiodinium microadriacticum CCMP2467	808	49,109	27,758	77%
Free-living
Effrenium voratum RCC1521	1,200	32,108	30,710	5%
Effrenium voratum CCMP3420	1,300	39,878	31,328	27%
Effrenium voratum CCMP421	1,100	32,615	30,046	9%
Polarella glacialis CCMP1383	2,980	58,232	38,147	53%
Prorocentrum cordatum	4,750	85,849	42,316	103%

Open in a separate window

^a(detected-modeled)/modeled x 100

Shared and unique genes and gene families

Current genome sequencing efforts for dinoflagellates have biased toward species with smaller genomes, precluding identification of phylum-wide common genes. Nevertheless, comparative genomic analyses still provide interesting insights into differential dynamics and diversification of dinoflagellate genomes. Analysis of 15 Suessiales genomes found 555,682 predicted protein sequences, with 42,539 forming gene families [23]. Only 2,500 (5.88%) have homologs in all 15 isolates, indicating most gene families are lineage-specific. Within Symbiodiniaceae, 3,445 unigenes are shared by eight species, representing core genes enriched in 219 KEGG pathways (Fig. ​(Fig.55).

Open in a separate window

Fig. 5

Shared and unique genes in Symbiodiniaceae genomes. A Core genes (3,445 after de-redundancy) in proportion of total genes in each species. B Unique genes in each species. Darker colors indicate a larger number of genes

Gene expression and regulation: Roles of cold shock proteins, microrna, and alternative splicing

Dinoflagellates are generally known to have limited transcriptional regulation (5–27% of genes; for review see Lin 2011 [2]). However, significant transcriptional responses to nutrient changes have been reported [111, 112]. For example, alkaline phosphatase gene expression and enzyme activity increase under phosphate limitation [112–115]. Nitrogen deficiency elicits significant upregulation of acquisition genes and downregulation of nitrogen associated protein gene (NAP50) [116]. In addition, Rubisco, antenna protein, and some cell cycle protein genes show strong diel dynamics of transcript abundance [72, 108]. Earlier studies indicated that several percentage of genes exhibited circadian rhythms of transcript abundance at 2- or more fold variations [117]. These findings suggest that dinoflagellates may have more dynamic transcriptional regulation than previously thought, particularly in response to environmental changes and daily cycles.

Cold shock domain (CSD)

CSD-containing proteins (CSPs) were initially identified in Escherichia coli during cold shock stress and are now known to occur in major groups of organisms. In prokaryotes, CSPs act mainly as RNA chaperones and can regulate transcription by binding to the gyrA promoter. In eukaryotes, CSPs usually contain additional domains and play roles in cold stress response, nutrient limitation, and growth. In plants, CSPs are essential for freezing tolerance and regulate translation during cold stress, seed germination, and flowering [121].

CSPs are ubiquitous in dinoflagellates. In L. polyedrum, CSP genes were the only gene family with higher transcriptional levels than in other protists [109], suggesting their significant role. Some LpCSPs contain only the CSD domain while others have additional domains, but neither showed cold induction [122]. In F. kawagutii, at least eight of 121 identified transcription factors (TFs) contain CSD (Skav204535, Skav204536, Skav217190, Skav218283, Skv218284, Skv223430, Fkav224338, Skv231215), suggesting that CSPs may be the functional domains of TFs in dinoflagellates.

CSPs were identified in Scrippsiella acuminata, showing increased expression in resting cysts [123]. Prorocentrum cordatum has two CSD genes that respond significantly to low temperatures [124]. Dinoflagellate CSPs contain a ZF domain and a glycine-rich motif besides the conserved CSD, with varying ZF and GR order between species. This organization resembles bacterial and plant CSPs, suggesting possible bacterial origin through HGT.

DNA and RNA binding assays showed dinoflagellate CSD proteins prefer binding RNA over DNA, and single-stranded over double-stranded DNA, without sequence specificity [118]. This suggests they are unlikely to act as transcription factors but may be involved in DNA unwinding. Some CSPs contain calcium-regulated mRNA-binding or helicase ATP-binding domains in certain species (Fig. ​(Fig.66).

Open in a separate window

Fig. 6

Organization of cold shock domain (CSD) in various proteins in dinoflagellates. Amphidinium, A. massarti, data source MMETSP0689; CSD2, domain PS51857; CSD8, domain SM00357; CmR, Calcium regulated mRNA-binding domain (IPR052069); Gymnodinium, G. catenatum, data source MMETSP0784-20121206; Smic, Symbiodinium microadriaticum, data source CAE7767176.1 dbp2, Helicase = helicase superfamily ATP-binding domain; Sym_CCMP2592, Symbiodinium strain CCMP2592; Fkaw38550, Fugacium kawagutii, BEST = bestrophin

MicroRNA (miRNA)

The scarce and weak transcriptional regulation in dinoflagellates suggests other mechanisms must exist. MicroRNA repertoires have been documented in the transcriptomes of Symbiodinium microadriaticum [125] and Prorocentrum shikokuence [102, 114] and the genome of F. kawagutii [18].

Functional microRNA machinery is evidenced by effective RNA silencing (Zhang and Lin 2019). A. gibbosum genome analysis suggests microRNA involvement in regulating responses to nitrogen and phosphate starvation [12].

In silico analyses of F. kawagutii miRNAs indicated potential influence on coral host gene expression [18]. The genome contains a putative RNA transporter homologous to Systemic RNA Interference Deficiency–1 (SID-1) in nematodes and mammals. Close relationship between Symbiodiniaceae and cnidarian homologs suggests possible HGT (Fig. ​(Fig.7A).7A). Potential target genes of symbiodiniacean miRNA were identified in coral hosts, showing broad interactive protein network (Fig. ​(Fig.7B)7B) and GO terms (Fig. ​(Fig.77C).

Open in a separate window

Fig. 7

Evidence that Symbiodiniaceae miRNA may interact with coral host to modulate host gene expression. A Phylogenetic tree of double-stranded RNA channel (SID-1). The close relationship between the homologs in Symbiodinium and the cnidarians suggests horizontal gene transfer between the two lineages. B Protein interaction network and metabolic pathways predicted to be impacted by miRNA. C GO categorization of predicted miRNA target genes in both F. kawagutii and coral

These findings suggest Symbiodiniaceae microRNA may cross over into the invertebrate host and modulate gene expression, potentially impacting symbiosis mechanisms like nutrient uptake and photosynthate translocation. While not experimentally demonstrated, this is consistent with findings in other symbiotic and pathogenic systems. Vibrio fischeri produces noncoding small RNA that controls host gene expression in squid light organs [126, 127]. Pathogen-to-host miRNA transfer silences host immunity genes in plants [128].

Post-transcriptional regulation

Ribosome-protected mRNA profiling revealed pulses of synchronous translation of gene clusters throughout the diel cycle [129], indicating circadian or light/dark modulation of gene expression. Pearson correlation analysis of rRNA count versus mRNA count shows that on average, translation of an mRNA requires 3.6 ribosomes (Fig. ​(Fig.8).8). Deviations from this may indicate above or below average translational efficiency, with exceptionally high translation activity occurring at LD20 (four hours before light-on).

Open in a separate window

Fig. 8

Correlation between ribosomal rRNA and ribosome-bound mRNA abundances in Lingulodinium polyedra. Data from Bowazolo et al. (2022) [129]

A study of 11 core dinoflagellate transcriptomes revealed diversity in the eIF4E family [130]. Each species contained 8–15 (average 11) different eIF4E family members, phylogenetically segregated into three clades. Dinoflagellate eIF4E-1 likely contains at least one translational initiation factor. eIF4E-2 lacks mRNA cap-binding residues and may have a non-essential regulatory role. eIF4E-3 genes share attributes with metazoan Class I eIF4E. The three clades were further divided into nine subclades, six containing members from all eleven species, suggesting ancestral duplication.

Powerhouse of RNA editing

RNA editing is a gene modification process where single or multiple nucleotides in a gene are altered during or after transcription. This can be identified by comparing the nucleotide sequence of a gene with its corresponding mRNA. RNA editing occurs in mRNA, tRNA, and rRNA, involving mitochondrial, plastid, and nuclear genes [131].

In dinoflagellates, RNA editing was first documented in mitochondrial protein-coding gene transcripts, showing extensive and mechanistic diversity [132]. Mitochondrial mRNA editing has increased in the percentage of impacted nucleotides throughout dinoflagellate evolution [27]. Most nucleotide changes are synonymous and do not affect the encoded proteins but tend to increase the GC content of the impacted genes. This may enhance translation efficiency since dinoflagellate mitochondria import cytoplasmic tRNA for translation, which may not function well with AT-rich mitochondrial genes [27]. This explanation, however, does not apply to rRNA editing.

The field of RNA editing in dinoflagellates has advanced significantly in the last decade. mRNA editing of mitochondrial protein-coding genes has been extended to basal, parasitic dinoflagellates such as Amoebophrya [9]. Furthermore, editing was also found in chloroplast-encoded transcripts, including mRNA and rRNA [133, 134]. For coral symbiotic Symbiodiniaceae, RNA editing was detected in in-hospite populations [135]. Interestingly, mitochondrial gene transcripts are more frequently edited than plastid genes [135]. RNA editing occurs not only in typical peridinin-containing dinoflagellate plastids but also in Chl b-containing plastids in Lepidodinium spp. and fucoxanthin-containing tertiary plastids of haptophyte origin in Karlodinium veneficum [136–138]. Whether plastids in the Dinotom group, such as Durinskia kwazulunatalensis with diatom-origin plastids, also undergo RNA editing remains unclear. This, along with the accelerated evolution of plastid protein-encoding genes in peridinin- and fucoxanthin-containing species, indicates a significant host effect [139].

Furthermore, mRNA editing of nucleus-encoded genes has also been detected. For instance, one of the saxitoxin biosynthesis genes, sxt4, was found to undergo editing in Alexandrium pacificum [140]. Additionally, using high-throughput sequencing, mRNA editing has been detected in nuclear genes. Up to 10,486 and 69,953 putative RNA editing sites were identified in the nuclear genomes of the coral symbiont D. trenchii CCMP2556 and the free-living bloom-forming species P. cordatum CCMP1329, respectively [22]. These events included all 12 possible types of RNA edits, with more transitions than transversions and a dominance of A-to-T transversion in noncoding regions, many of which were condition-specific. Furthermore, A-to-T editing within untranslated regions appears to be associated with the upregulation of edited genes under heat stress [15]. If experimentally verified, the high number of RNA editing events suggests that RNA editing may be a more influential molecular mechanism regulating gene expression in dinoflagellates than previously thought.

New insights on dinoflagellate chloroplasts

Nucleomorphs, relic nuclei of eukaryotic endosymbionts, have been found in some dinoflagellate chloroplasts. Previously known only in the phylum Cryptophyta [141], a recent study discovered nucleomorphs in two newly described dinoflagellate species with green chloroplasts [142]. These chloroplasts, like those in Lepidodinium spp., originate from prasinophytes, although Lepidodinium spp. do not contain nucleomorphs.

Typical dinoflagellate chloroplast genomes are composed of minicircles, each containing one or two genes (occasionally none), totaling 16–17 genes per plastid genome [143]. In contrast, plastid genomes in Durinskia baltica and Kryptoperidinium foliaceum are large macrocircles, similar to typical algal plastids, and more similar to those of diatoms [144]. Chloroplast proteins in dinoflagellates evolve faster than in other algal lineages, indicating a strong host effect on plastid evolution [139].

Dinoflagellates have conserved photosystem cores (PSI and PSII) but varied light-harvesting complexes (LHCI and LHCII). Dinoflagellates have 12 subunits in PSI, ten of which are shared by all photosynthetic organisms, and 24 in PSII, 20 of which are shared by all photosynthetic organisms (Fig. ​(Fig.9).9). Dinoflagellate LHCs consist of a thylakoid intrinsic Chl a-Chl c2-peridinin-protein complex (acpPC) and a water-soluble peridinin-Chl a-protein (PCP) complex [145, 146]. Peridinin is the major accessory pigment (carotenoid) unique to dinoflagellates, which is used to define typical dinoflagellates. PCP exists in two forms, differing in absorption and fluorescence properties [147, 148].

Open in a separate window

Fig. 9

Comparison of dinoflagellate photosystem (PS) components with counterparts in other algal lineages. A PSI. Reproduced from Lin et al. 2024 [149]. B PSII. Information source: Fromme et al. 2001 [150], Pi et al. 2018 [151], Qin et al. 2019 [152], Gisriel et al. 2022 [153], Su et al. 2019, 2022 [154, 155], Zhao et al. 2023 [156], Li et al. 2024 [82]

Single-particle electron microscopy of PSI-LHCI supercomplexes in Breviolum minutum revealed 25 LHCs, including LHCF and LHCR families, and indicated a role in photoprotection via nonphotochemical quenching [157]. Recent cryo-EM studies identified novel PSI subunits (PsaT and PsaU) and suggested enhanced electron transport and energy quenching efficiency [82, 158].

Photochemical experiments have shown high quenching efficiency in dinoflagellates, such as symbiodiniacean species. Under stress conditions, Cladopodium cells activate a “super-quenching” mechanism, transferring excess excitation energy to PSI and converting it into heat, potentially triggering symbiont expulsion or bleaching [159].

Adaptation to symbiotic lifestyle

Comparative genomic analyses reveal that genes involved in nutrient uptake, photosynthate transport, stress response, and infection are present and expanded in symbiodiniacean dinoflagellates [18, 24]. In the genomes of B. minutum and F. kawagutii, genes important for mutualism, such as those related to sugar and fatty acid metabolism, oxidative stress, transport, photosynthesis, and adhesion, have undergone duplication during two major retroposition events (Fig. ​(Fig.10;10; [49]). Comparison of symbiodiniacean genomes with those of other algae, the land plant Arabidopsis thaliana, and the parasite malaria Plasmodium falciparum identified genes unique to Plasmodium and Symbiodiniaceae, such as those involved in cell recognition (e.g. merozoite surface protein 1) and genes with unknown functions (e.g. uncharacterized protein PF11_0213, protein dpy-19 homolog 1) [18]. In Plasmodium, merozoite surface proteins are crucial for red blood cell invasion (Beeson et al. 2016).

Open in a separate window

Fig. 10

Two episodes of retroposition discovered in two Symbiodiniacea genomes coincident with two major periods of symbiosis evolution. A) Fugacium kawagutii. B) Breviolum minutum. a, first episode led to expansion of sugar and fatty acid metabolism, oxidative stress response, and transport. b, second episode that promoted photosynthesis and adhesion. Based on Song et al. 2017 [49]

The F. kawagutii genome encodes 16 putative circumsporozoite protein (CSP)-coding genes and a CSP-like gene with a forkhead domain, potentially regulating vesicle-mediated transport and CSP secretion. In the chlorophyte symbiont Symbiochlorun hainanensis, a CSP-like protein is upregulated by acidification and heat stress [160]. Protein secretion via the T2S system, encoded by at least 12 genes, is a major virulence mechanism in bacterial infections and may participate in bacteria-eukaryote mutualism [161–163].

UV protection is crucial in coral-Symbiodiniaceae mutualism. Mycosporine-like amino acids (MAAs) prevent UV damage, initially believed to be synthesized only by the basal lineage (clade A) of Symbiodiniaceae [164]. Genome analyses confirmed the presence of MAA biosynthesis genes in S. tridacnidorum and absence in later-diverging lineages like F. kawagutii and Cladocopium [18, 25], although an improvement of the genomes later revealed partial MAA biosynthesis genes in F. kawagutii [19]. Furthermore, the MAA gene cluster was found to be present in D. trenchii, in addition to Symbiodinium, but absent in B. minutum, consistent with the original notion of ancestral synthesis ability with secondary loss [16].

Insights into mutualism were gained from green Noctiluca scintillans, which harbors prasinophyte Protoeuglena noctilucae as endosymbionts. This mixotrophic dinoflagellate forms massive blooms in the Arabian Sea, feeding on phytoplankton and producing ammonium for symbiont photosynthesis [165, 166]. During blooms, symbiont photosynthesis is active, but cell division is suppressed, suggesting the host promotes symbiont photosynthesis while inhibiting symbiont proliferation to maximize photosynthate translocation [166].

Adaptation to parasitic lifestyle

Amoebophrya species are intracellular parasites of marine dinoflagellates, radiolarians, ciliates, and other Amoebophrya strains. The genomes of Amoebophrya ceratii strain AT5.2 and Amoebophrya sp. strains A25 and A120 are significantly streamlined, with sizes (~120 Mbp) much smaller than other dinoflagellates (~1-250 Gbp). They have few transposable elements, short introns and intergenic regions, and a limited number of gene families, encoding 19,925, 28,091, and 26,441 genes, respectively [10, 11]. These genomes are gene-rich, with 26%, 27%, and 80% more genes than predicted for AT5.2, A25, and A120, respectively, comparable to some other dinoflagellates (7-89% above model prediction; Table ​Table22).

Most genes in all three species (56%, 63.7%, and 59%) lack functional annotation in public databases, suggesting some may be specific to parasitism. The intron density (about 1.3, 1.47, and 1.42 introns per kb of coding sequence in AT5.2, A25, and A120, respectively) is typical of dinoflagellates, but the introns are mostly non-canonical, including repeated introns with highly variable splicing motifs. The canonical GT-AG splice sites account for only 34.02% and 30.41% in strains A25 and A120, respectively, with the majority (>65%) being various other boundaries. In these strains, 11-30% of the non-canonical introns contain 8–20 nt inverted repeat motifs and 3–5 nt direct repeat motifs, spreading similarly to transposable elements (TEs).

High gene block synteny (>49%) but low gene sequence similarity (<50%) was observed among the Amoebophrya strains, suggesting strong selection pressure for protein changes due to parasitic interaction with different hosts. All three strains exhibit organelle reduction, including loss of the plastid and potential loss of mitochondrial genome and functions, indicating adaptive evolution. This is consistent with findings in Hematodinium, another parasitic dinoflagellate, which infects shellfish and other organisms [18].

Transcriptome sequencing of Hematodinium revealed 29,510 unique ORFs with no evidence of nucleus-encoded plastid-targeted protein genes or plastid gene homologs. The same BLAST workflow revealed 94 putative mitochondrial proteins. Genes involved in fatty acid, lysine, and tetrapyrrole synthesis lack organelle-targeting signal peptides, suggesting these pathways occur in the cytosol. Interestingly, the type I fatty acid synthase (FAS) responsible for cytosolic FA synthesis appears to have taken the form of plastid type II FAS by fusing multiple protein genes [20].

Highlight of newly found genes and physiologies

The rapidly growing dinoflagellate genomic data constitute a rich resource for exploring unique physiologies or metabolic features of this unique group of protists. The many functional unknown genes predicted from their genomes will also serve as a target of future studies. A few of these genes, which are highly interesting but have not received adequate attention, deserve some highlight here. They are involved in acquisition and storage of nutrients, defense (toxin production), response to stress, and life cycle.

Macrophage Migration Inhibitory Factors (MIF)

MIFs are cytokines and chemokines essential for the vertebrate immune system, mediating innate immune responses [197]. These proteins are conserved across the tree of life, serving various cellular functions [198]. In free-living organisms, MIFs activate immune responses, promote immune cell proliferation, and inhibit p53-mediated apoptosis [199]. Parasitic species produce multiple MIF proteins to modulate host immune responses, facilitating infection [200–203]. A homolog of MIF (PmMIF) was identified in the cyanobacterium Prochlorococcus marinus, showing high structural homology with mammalian MIFs but lacking the Cys-X-X-Cys motif necessary for oxidoreductase activity [204].

A MIF-like protein was detected in L. polyedrum (LpMIF), but instead of a soluble single-domain protein as a typical MIF is, LpMIF is a transmembrane protein with a cytosolic domain [205]. The MIF domain is localized in cell wall-associated membranes, vesicular bodies, and membranes of extracellular vesicles accumulating at the secretory pores of the cells (Fig. ​(Fig.11A,11A, B). Tblastn analysis using human MIF as query against the MMETSP database (E-value e-20 cutoff) yielded 92 hits, 28 from dinoflagellates including P. shikokuense, Alexandrium spp., Symbiodiniaceae species, Karenia brevis, Karlodinium veneficum, and Heterocapsa triquetra. Phylogenetic analysis shows it is closely related to a prasinophyte homolog within a clade containing cryptophytes (Fig. ​(Fig.11C).11C). Most dinoflagellate hits align with human MIF (hsMIF) from position 2 to 115, with some being shorter due to incompleteness (Fig. ​(Fig.1111D).

Open in a separate window

Fig. 11

Macrophage migration inhibitory factors (MIF) detected in dinoflagellates. A Schematic of MIF structure. B Cell surface localization of MIF in L. polyedra (green fluorescence pointed by arrows). From Jaouannet et al. 2020 [205]. C Phylogenetic tree showing clustering of dinoflagellate MIF with homologs from prasinophytes and other lineages. D Alignment of dinoflagellate MIFs with humans MIF showing high similarity. Homsa, Homo sapiens; Alemo, Alexandrium monilatum; Fugka, Fugacium kawagutii; Linpo, Lingulodinium polyedra; SymbB, Breviolum sp.; Symmi, Symbiodinium microadriaticum

Pfam search confirms these dinoflagellate MIF sequences contain the MIF functional domain. KEGG analysis indicates they are phenylpyruvate tautomerase (EC 5.3.2.1), the second member of the MIF superfamily, involved in Tyrosine and Phenylalanine metabolism. Under biotic stress, L. polyedra shows decreased MIF gene expression and reduced protein levels on the cell surface, suggesting this MIF may be involved in intercellular communication [205] or sensing of environmental stimuli.

I wish to thank my students for assistance in various ways, particularly Xiaoyu Wang, Yifan Gu Jiamin He, Jingtian Wang, and Richard Antosca for gathering information, preparing tables and figures, and organizing the references.