Background

Cryptophytes are unicellular bi-flagellate algae found in marine, brackish, and freshwater environments the world over. Photosynthetic and osmotrophic cryptophytes have been described; phototrophic species contain plastids with chlorophyll a and c and phycobilins as accessary pigments. Beyond their ecological significance, cryptophytes are of considerable evolutionary interest by virtue of the fact that they contain four distinct DNA-containing compartments: a host-derived nucleus and mitochondrion and an endosymbiont-derived plastid and a “nucleomorph.” Nucleomorphs are the remnant nuclei of algal endosymbionts and provide direct evidence for the phenomenon of secondary endosymbiosis, a process whereby a photoautotrophic eukaryote is engulfed and retained by a heterotrophic one [1, 2]. A wide array of eukaryotic algae are known to have acquired their plastids by secondary (or tertiary) endosymbiosis. In addition to cryptophytes, this includes the haptophytes, ochrophytes (plastid-bearing stramenopiles), chlorarachniophytes, and some dinoflagellates [3, 4]. In most such algae, the DNA in the endosymbiont-derived nucleus has been lost or transferred to the host nucleus during the course of endosymbiont integration. However, cryptophytes (excluding Goniomonas) and chlorarachniophytes represent a fascinating exception. Comparative genomics has revealed that the cryptophyte plastid and nucleomorph are derived from a red algal endosymbiont, whereas the chlorarachniophyte endosymbiont comes from a green alga [5, 6]. Interestingly, another example of green alga-derived nucleomorphs has recently been discovered in two different dinoflagellate lineages, although compared to cryptophytes and chlorarachniophytes, little is known about their genome biology and evolution [7, 8].

The nucleomorph genomes of cryptophytes and chlorarachniophytes have reduced dramatically to ~1 megabase pairs (Mbp) or less in size and contain only a few hundred genes spread across three chromosomes. As noted above, genome reduction has resulted in most of the nucleomorph genes being lost or transferred to the host nucleus, intergenic spacers have been streamlined, and almost all the repetitive DNA presumed to have been present in their algal progenitors has been eliminated. To date, four cryptophyte nucleomorph genomes have been sequenced, the 550.5-kilobase-pair (Kbp) genome of Guillardia theta [9], the 571.4-Kbp genome of Hemiselmis andersenii [10], the 702.9-Kbp genome of Chroomonas mesostigmatica [11], and the 485.9-Kbp genome of the secondarily non-photosynthetic species Cryptomonas paramecium [12]. The number of predicted protein-coding genes ranges from 466 in C. paramecium to 505 in Ch. mesostigmatica. A substantial proportion of the protein-coding genes in the cryptophyte nucleomorph genomes are hypothetical in nature. These hypothetical genes are composed of (i) cryptophyte nucleomorph-specific ORFs, or “nORFs,” meaning that they have conserved homologs in other cryptophyte nucleomorph genomes but not in other known genomes, and (ii) “nORFans,” genes that show no obvious sequence-based homology to any gene in known databases, nucleomorph-derived or otherwise. The number of conserved nORFs predicted in sequenced cryptophyte nucleomorph genomes is presently as follows: 196 in G. theta, 181 in H. andersenii, 186 in Ch. mesostigmatica, and 186 in C. paramecium. The overall proportions of nORFans were found to be 155 (32%) in G. theta, 74 (16%) in H. andersenii, 94 (19%) in Ch. mesostigmatica, and 133 (29%) in C. paramecium [11].

Members of the genus Cryptomonas are of particular interest in that they provide an opportunity to study the loss of photosynthesis over short evolutionary timescales and how this impacts genome biology. Phylogenetic analysis of plastid and nucleomorph genes has revealed that three different non-photosynthetic Cryptomonas lineages are closely related to different photosynthetic species [13,14,15,16,17], suggesting that members of the genus Cryptomonas have lost the ability to photosynthesize on several different occasions (Fig. 1, Supporting Information Fig. S5). Unfortunately, genomic sampling is presently sparse; only one nucleomorph genome from a non-photosynthetic Cryptomonas species has been sequenced, that of C. paramecium CCAP977/2a [12]. To rectify the situation, we have sequenced four nucleomorph genomes from closely related strains and species within Cryptomonas and carried out a comprehensive 5-way comparative genomic analysis. Our results provide a window on fine-scale nucleomorph genome variation within the genus and allow us to ascribe predicted functions to previously unknown ORFans by virtue of their presence in large syntenic blocks, as well as to identify recent examples of pseudogenization of photosynthesis-related genes. Overall, our data improve knowledge of the set of nucleomorph protein-coding genes predicted to still be functioning in non-photosynthetic cryptophytes.

Fig. 1
figure 1

Schematic phylogeny of cryptophytes based on nucleomorph small subunit ribosomal RNA gene sequences with a focus on members of the genus Cryptomonas (modified from Fig. S5). The five species whose nucleomorph genomes were compared herein are marked with arrows. Non-photosynthetic species are marked with open circles

Full size image

Results and Discussion

Cryptomonas nucleomorph genomes: size and structure

Nucleomorph genomes were sequenced telomere to telomere for two colorless Cryptomonas strains and two brown-colored strains. An overview of the characteristics of these new genomes relative to the previously published nucleomorph genome of C. paramecium CCAP977/2a [12] is provided in Table 1. All four genomes are comprised of three chromosomes with total sizes ranging from 485.8 Kbp (C. paramecium KR) to 659.1 Kbp (Cryptomonas sp. CCAC1634B) (Fig. 2). Including C. paramecium CCAP977/2a, the five genomes contain between 411 (C. paramecium KR) and 504 (C. curvata CCAP979/52) predicted protein-coding genes. Between one and six spliceosomal introns were predicted, consistent with the low number of such introns in cryptophyte nucleomorph genes in general, and in Cryptomonas genes in particular [9,10,11,12] (Table 1; note that the orf80 intron originally predicted by Tanifuji et al. [12] corresponds to a region of the nucleomorph genome now designated as a trnE pseudogene). The percent coding capacity ranges from 84 to 87%: 50–56% of the genome for protein-coding genes with predictable functions, 3–5% for RNAs, 28–30% for hypothetical ORFs, and 13–16% for intergenic sequences. Between 17 and 19 nucleomorph-encoded, plastid-associated genes are found in the three colorless Cryptomonas species, whereas 31 plastid-associated genes reside in the genomes of the two photosynthetic C. curvata strains.

Table 1 Overview of nucleomorph genome sequences from five Cryptomonas species
Full size table
Fig. 2
figure 2

Physical maps of five Cryptomonas spp. nucleomorph genomes. The figure shows syntenic chromosomes aligned species by species. Recombined regions between different chromosomes are highlighted with blue and orange dashed boxes. A duplicated region containing five genes (BRSK, nol10, pab2, trf, and orf (CPARA_1gp179)) on chromosomes 2 and 3 of Cryptomonas species CCAC1634B is marked with a blue background box. Representative examples of obvious sequence conservation between single “large” hypothetical ORFs and two or more smaller hypothetical ORFs are highlighted with red brackets and arrowheads. Fragmented pseudogenes in the genomes of one or both non-photosynthetic strains are marked with asterisks and highlighted purple to match their intact counterparts in photosynthetic species

Full size image

In the two colorless C. paramecium strains, sub-telomeric rDNA operons (5S-18S-5.8S-28S) were found on both ends of chromosome 3, but only 5S rDNA genes reside on one end of chromosomes 1 and 2 (Table 1, Fig. 2 and Supporting Information Figs. S1-S3). The other colorless Cryptomonas sp. CCAC1634B and the two brown-colored strains of C. curvata were found to have complete sub-telomeric rDNA operons on both ends of all chromosomes. The telomere sequence in C. paramecium KR is GA15-18 (similar to the GA9 found in C. paramecium CCAP977/2a [12]), (GA)4GT in C. curvata KR, and (GA)3GT in C. curvata CCAP979/52. The telomeric repeats of Cryptomonas sp. CCAC1634B were found to be much more complex than those of the other Cryptomonas species analyzed herein: T(GTA)3AG6AGA(AG)6G3AG5. This is interesting given that GAn telomeric repeats are found in much more distantly related cryptophyte nucleomorphs: GA17 in H. andersenii and GA14 in Ch. mesostigmatica, whereas the telomeric repeat in G. theta is [AG]7AAG6A [9,10,11]. For reference, sequenced nucleomorph genomes in chlorarachniophytes have telomere sequences as follows: [TCTAGGG]n in Bigelowiella natans, Lotharella oceanica, and Lotharella vacuolata, and [TCCTGGG]n in Amorphochlora amoebiformis [18,19,20].

Highly conserved genome structure in Cryptomonas nucleomorph genomes

The newly sequenced nucleomorph genomes show a high degree of structural conservation relative to the previously published genome of C. paramecium CCAP977/2a (Fig. 2). All three chromosomes can, for the most part, be aligned gene for gene. Of particular note is the fact that single hypothetical ORFs in one genome were sometimes broken into 2–6 separate ORFs in another genome (Fig. 2, red brackets and arrow heads in chromosomes 1 and 3). The nucleomorph genome of the non-photosynthetic Cryptomonas sp. CCAC1634B has a much greater degree of gene content overlap with that of the photosynthetic species C. curvata than with the non-photosynthetic C. paramecium strains CCAP979/2a and KR (Fig. 2, blue genes; see below).

Protein-coding genes with predicted functions

Of the 287–329 protein-coding genes with predicted functions in the five Cryptomonas nucleomorph genomes, most are involved in transcription, translation, DNA metabolism and cell cycle control, RNA metabolism, protein folding, protein degradation, and mitosis (Supporting Information Table S1). Nine genes (cpeT-like, fkbp, fkbp-like, hcf136, hlip, met, rubredoxin, sut, and tha4) were found only in the photosynthetic species C. curvata (Fig. 2, green genes). Nineteen genes were found to be shared between the colorless strain Cryptomonas sp. CCAC1634B and the photosynthetic strains of C. curvata (bre1-like, cdc2-like, crm, gyrA, gyrB, hat, hira, kin(mps1), psf2, ranbpm, rps21, rrp40, rrp41, rrp44, smc2, smc4, spt5, U5snRNP (20Kda), and ubc2) but missing in the C. paramecium strains (Fig. 2, light blue lines). The only gene content differences between Cryptomonas sp. CCAC1634B and two strains of the photosynthetic species C. curvata are the absence of the following nine genes with plastid-associated functions: cpeT-like, fkbp, fkbp-like, hcf136, hlip, met, rubredoxin, sut, and tha4 (Figs. 2 (light green lines) and 3).

Fig. 3
figure 3

Predicted functional proteins inferred from complete nucleomorph genomes of five Cryptomonas strains. The graph shows the proportion of predicted cryptophyte nucleomorph-specific functional protein-coding genes (blue), plastid-associated genes (green), and the total length of functional protein-coding genes

Full size image

Plastid-associated genes

The two nucleomorph genomes of the photosynthetic, brown-colored C. curvata strains were found to contain the same set of 31 plastid-associated genes (i.e., genes for plastid-targeted proteins) found in Ch. mesostigmatica, H. andersenii, and G. theta. Interestingly, the nucleomorph genomes of the non-photosynthetic species C. paramecium and Cryptomonas sp. CCAC1634B have lost many photosynthesis-related genes, but nevertheless still retain 16 plastid-associated genes found in all other cryptophyte species (Figs. 2 and 3 and Supporting Information Table S1) (these genes are clpP1, clpP2, cpn60, dnaG, engA, ftsZ, gidA, gidB, iap100, rpoD, rps15, secE, sufD, tic22, and two ORFs (orf152, orf826)). The cpeT-like, hfc136, hlip, met, rub, and tha4 genes, as well as four ORFs (orf177, orf243, orf268, and orf336), have been lost from the genomes of all three non-photosynthetic strains analyzed herein (Fig. 2, green and dark green genes), while cbbX and ycf20 show differential presence/absence patterns (cbbX is found only in C. paramecium CCAP977/2a, whereas ycf20 is missing in Cryptomonas sp. CCAC1634B; Fig. 2). That said, Cryptomonas sp. CCAC1634B has lost 11 protein-coding genes associated with photosynthesis; these genes are also absent in the two C. paramecium genomes with the exception of two cbbX and ycf20 genes in each strain (Fig. 2, green and dark green genes). Clear homologs of orf177, orf243, orf268, and orf336 are also found in the photosynthetic cryptophytes Ch. mesostigmatica, H. andersenii, and G. theta; while clearly conserved, their functions remain mysterious (Supporting Information Table S1).

Given their presence in non-photosynthetic species, cbbX (a photosynthesis-associated gene; see below) and the plastid DNA replication genes gyrA and gyrB are worthy of particular mention. DNA Gyrase (gyrA and gyrB), which is involved in DNA replication and the relaxation of DNA supercoiling, is important for plastid DNA replication [21, 22]. The gyrA and gyrB genes are encoded in the nucleomorph genomes of almost all cryptophytes, with the exception of the colorless species C. paramecium. And while both cbbX and gyrA/gyrB are absent in the nucleomorph genome of C. paramecium KR, the gyrA/gyrB genes persist in the colorless species Cryptomonas sp. CCAC1634B (Figs. 2 and 4).

Fig. 4
figure 4

Presence-absence of key plastid-associated genes in the plastid and nucleomorph genomes of photosynthetic and non-photosynthetic Cryptomonas species

Full size image

CbbX is a red-algal type ATPase enzyme involved in the activation of RuBisCO (Ribulose 1,5-bisphosphate carboxylase/oxygenase) and may serve as a molecular chaperone of RuBisCO subunit assembly. The cbbX, rbcL, and rbcS genes are arranged as an operon in the plastid genomes of red algae and cryptophytes. The plastid genomes of ochrophytes typically have this arrangement as well (i.e., rbcL-rbcS-cbbX), although the cbbX gene has moved to a different position in the plastid genomes of studied Bacillariophyceae [23, 24]. In the unicellular red alga Cyanidioschyzon merolae strain 10D, cbbX is present in both the plastid and nuclear genomes, while the RuBisCO operon (rbcL-rbcS-cbbX) is located only in the plastid genome. In cryptophytes, two distinct types of cbbX genes are present in the nucleomorph and plastid genomes (Figs. 2 and 4). Molecular phylogenetic analyses reveal that the nucleomorph-encoded cbbX of cryptophytes branches together with some α-proteobacterial cbbX sequences, not with the plastid-encoded cbbX group [25]. Interestingly, whereas the canonical RuBisCO operon is present in the plastid genome of the colorless species C. paramecium CCAP977/2a [26], cbbX is missing from both the plastid and nucleomorph genomes of the very closely related strain C. paramecium KR, while the rbcL and rbcS genes are retained (as in other cryptophytes, cbbX is also present in the C. paramecium CCAP977/2a nucleomorph genome; Fig. 4). In contrast, the plastid RuBisCO operon and nucleomorph cbbX gene are missing in the colorless species Cryptomonas sp. CCAC1634B (Fig. 2 [16]). This is similar to the situation in the colorless diatom Nitzschia spp. [27, 28] and Spumella-like flagellates (chrysophytes) [29], both of which have completely lost rbcL-rbcS and cbbX in their plastid genomes. In the euglenophytes, rbcL resides in the plastid genome but rbcS has been transferred to the nuclear genome. The non-photosynthetic euglenophyte Euglena longa still retains rbcL in its leucoplast genome, which has been shown to give rise to a very low abundance of rbcL protein [30, 31]. Beyond these examples, it should be noted that the rbcL gene has been found in the plastid or nuclear genomes of other secondarily non-photosynthetic organisms as well, including parasitic land plants [32], heterotrophic stramenopiles [33], and the heterotrophic dinoflagellate Crypthecodinium cohnii [34]. The functional significance of rbcL gene retention despite the loss of photosynthesis is, in most cases, unclear.

The same uncertainty applies to ycf20, the protein product of which is associated with nonphotochemical quenching and thermal dissipation [35]. This gene, which is found broadly across photosynthetic organisms including cyanobacteria, algae, and plants, resides in the plastid genome of red algae and most cryptophytes, but is absent in photosynthetic genera such as Cryptomonas, Rhodomonas, and Teleaulax [15], as well as non-photosynthetic species within Cryptomonas [16]. Interestingly, a ycf20-like gene is also present in the nuclear genome of the red alga Cyanidioschyzon merolae [36] and, as shown here and elsewhere, in the nucleomorph genomes of some but not all cryptophytes ([9,10,11,12], Fig. 4). At the present time, it is difficult to make sense of the patchy distribution of ycf20 other than to say that its function is not essential in at least some photosynthetic and non-photosynthetic organisms.

Synteny analysis allows functional assignment of hypothetical ORFs

A substantial proportion (28–30%) of the predicted protein-coding genes in the five Cryptomonas nucleomorph genomes are hypothetical ORFs; based on their sequence, they cannot be assigned a function. These so-called nORFs generally show substantial sequence similarity to predicted protein-coding genes in the H. andersenii and Ch. mesostigmatica nucleomorph genomes but are noticeably less similar to those of the more distantly related species G. theta (Supporting Information Tables S2 and S3). Although the colorless species lost many genes in their nucleomorph genomes, the nORFs in the colorless Cryptomonas sp. CCAC1634B are very similar to those of the photosynthetic C. curvata, but rather less similar to those of the colorless C. paramecium strains. The high degree of sequence similarity and the more conserved nature of the C. curvata ORFs allowed us to assign predicted functions to five previously hypothetical protein-coding genes in the genomes of non-photosynthetic C. paramecium and Cryptomonas sp. CCAC1634B. The “newly discovered” protein-coding sequences are the kinetochore protein (nuf2), mRNA splicing factor (sf3b3-like and sf3b1-like), retinoic acid receptor alpha (rarA), and cell division cycle 5 (cdc5-like) (see below).

Conserved and unique hypothetical ORFs

Curiously, while the two colorless C. paramecium strains have almost identical predicted gene sets and gene order across their three nucleomorph chromosomes (Fig. 2), they have different numbers of nORFs. Many single nORFs in C. paramecium KR were found to be fragmented into multiple ORFs (between 2 and 6) in C. paramecium CCAP977/2a, resulting in the CCAP977/2a strain having a total of 45 more nORFs than C. paramecium KR (Fig. 5, Supporting Information Tables S2-S3). ORF fragmentation was also apparent in a comparison of the two C. curvata genomes, although to a much lesser degree (see below). The non-photosynthetic Cryptomonas sp. CCAC1634B shares 158 nORFs with the photosynthetic C. curvata species; none are shared exclusively between the three colorless strains (Fig. 6a, gray). Twenty-one nORFs are shared exclusively between the two C. paramecium strains and 10 between C. curvata KR and CCAP979/52 (Fig. 6a, green). A total of 109 nORFs were found to be conserved in all five strains of the genus Cryptomonas (Fig. 66a, red).

Fig. 5
figure 5

Hypothetical proteins inferred from complete nucleomorph genomes of five Cryptomonas strains. The graph shows the number of cryptophyte nucleomorph-specific hypothetical protein-coding genes (nORFs) and the total length of these nORFs

Full size image
Fig. 6
figure 6

Distribution of hypothetical proteins inferred from complete nucleomorph genomes of five Cryptomonas strains and other photosynthetic cryptomonads. a Venn diagram showing the extent to which hypothetical proteins overlap between the five Cryptomonas species/strains analyzed herein. b Venn diagram showing the number of cryptophyte nucleomorph-specific hypothetical protein-coding genes shared between all five Cryptomonas strains and some or all of the nucleomorph genomes of Chroomonas mesostigmatica (Cm), Guillardia theta (Gt), and Hemiselmis andersenii (Ha)

Full size image

Extending beyond the genus Cryptomonas, 17 nORFs were found to be shared across all eight sequenced cryptophyte nucleomorph genomes (Fig. 6b), whereas 78 such ORFs were shared between all five Cryptomonas strains and another cryptophyte (Ch. mesostigmatica, or H. andersenii, or G. theta). The remaining 31 nORFs were shared among members of the genus Cryptomonas (Fig. 6b, yellow). Only 10 hypothetical ORFs were genuine nORFan genes in C. curvata species, as defined previously [10, 37], meaning they show no obvious sequence-based homology to any gene in any known genome, including nucleomorph genomes. The biological significance of the nORFs and nORFans in the cryptophyte nucleomorph genomes analyzed herein is unclear (see below).

Gene loss and pseudogenization

The high degree of synteny across the nucleomorph genomes of Cryptomonas spp. allowed us to assign putative functions to a handful of previously hypothetical proteins. It also made it possible to identify instances of gene loss and pseudogenization. As noted above, we identified numerous cases in which a single large hypothetical ORF in one nucleomorph genome was in the same genomic location as one or more smaller — and demonstrably homologous — ORFs in the genome of one or more of the four other Cryptomonas nucleomorph genomes in our dataset. For example, we found one hypothetical ORF in C. paramecium KR that was syntenic with six small ORFs in C. paramecium CCAP977/2a, each with conserved amino acid sequences and adding up to approximately the same length as the single ORF in the KR genome (Fig. 2, Supporting Information Figs. S1-S3 and Tables S2-S3). It is not clear whether such examples of ORF fragmentation represent instances of pseudogene formation, though it is interesting that an earlier RNA-Seq-based analysis of nucleomorph genomes revealed that the vast majority of nucleomorph genes, including nORFs and nORFans, are transcribed into mRNA, including those of C. paramecium CCAP977/2a [38].

Even among protein-coding genes with discernable functions, examples of “broken” ORFs were detected. The slightly less divergent nature of the C. curvata genes relative to those of the other Cryptomonas strains was particularly useful in this regard. For example, three adjacent ORFs in C. paramecium CCAP977/2a occupied the same syntenic position to the mRNA splicing factor gene sf3b3-like in nucleomorph chromosome 2 of other Cryptomonas species (Fig. 2, marked purple). Interestingly, the sf3b3-like gene is present as a single ORF in C. paramecium KR. The sf3b3-like gene was also detected in Ch. mesostigmatica [11]. Similarly, there were some smaller ORFs occupying the same syntenic position in the nucleomorph genome of C. paramecium CCAP977/2a as the splicing factor gene sf3b1-like in five Cryptomonas strains (Fig. 2, marked purple) and Ch. mesostigmatica [11]. A total of five genes (nuf2, sf3b3-like, sf3b1-like, rarA, and cdc5-like) were inferred to have been pseudogenized by single-base deletions or stop codon-causing mutations in C. paramecium CCAP977/2a (Supporting Information Fig. S4).

Gene synteny and recombination

The five nucleomorph genomes analyzed herein show evidence of inter-chromosomal recombination between chromosomes 1 and 3, specifically in their sub-telomeric regions (Fig. 2, blue and orange dashed boxes). These events presumably took place after the species and strains diverged from one another. For example, the gene content and order in the sub-telomeric region of one end of chromosome 1 of the two C. paramecium strains is almost identical to one end of chromosome 3 in the other Cryptomonas species (albeit with additional gene losses in C. paramecium) (Fig. 2, blue dashed boxes). At the same time, Cryptomonas species CCAC1634B has duplicated copies of BRSK, nol10, pab2, trf, and orfCPARA_1gp179 on chromosomes 2 and 3 (Fig. 2, highlighted in blue background; Fig. S2).

The degree of synteny between the three previously sequenced nucleomorph genomes (Ch. mesostigmatica CCMP1168 [11], G. theta [9], and H. andersenii CCMP644 [10]) is low compared to that seen within the genus Cryptomonas. We did nevertheless identify gene recombination events between the three Cryptomonas species examined herein (Fig. 7). Whereas within-species gene order conservation is largely the same in C. paramecium and C. curvata (Fig. 7a, b), gene order is substantially re-arranged between the species (Fig. 7c–e), including between the two colorless species C. paramecium and Cryptomonas sp. CCAC1634B, which lost photosynthesis independently (Fig. 7c).

Fig. 7
figure 7

Gene order conservation within and between three Cryptomonas species (five strains). a, b Circos plots showing within-species gene order differences. c–e Circos plots showing the degree of gene order conservation between species. f Circos plot showing gene order variation between all three species. The center shows syntenic gene blocks between pairs of species

Full size image

Conclusions

Together with those of chlorarachniophytes, the nucleomorphs of cryptophyte algae have long been considered the “smoking guns” of secondary (i.e., eukaryote-eukaryote) endosymbiosis [5, 6, 39, 40]. The first nucleomorph genome to be sequenced, that of the cryptophyte G. theta, was published in 2001 [9] and hailed as a nuclear genome in miniature. The 550.5-Kbp G. theta nucleomorph genome contained ~500 densely packed protein-coding genes, surprisingly few of which encoded proteins that were obviously plastid-targeted (a mere 30 in total). Three additional cryptophyte nucleomorph genomes have since been sequenced, i.e., those of H. andersenii [10], Ch. mesostigmatica [11], and C.s paramecium [12]. Comparative genomic investigations of these data underscore the fact that nucleomorph genes are primarily “house-kee**” in nature, i.e., encoding proteins involved in core eukaryotic cellular processes such as transcription, translation, and protein folding/turnover. At the same time, however, only ~50% of the predicted genes in these genomes can be assigned a predicted function based on sequence similarity alone — nucleomorph genes and genomes are highly divergent relative to their counterparts in the red algae from which they evolved.

Our study is the first examination of nucleomorph genomes from multiple strains and species within a single cryptophyte genus, i.e., Cryptomonas. This genus is of particular interest by virtue of the fact that, on at least three occasions, its members have lost photosynthesis [13, 16, 17]. The previously sequenced nucleomorph genome of the non-photosynthetic C. paramecium CCAP977/2a [12] is ~486 Kbp in size — the smallest cryptophyte genome sequenced thus far. To this single data point, we have added the genomes of two more colorless heterotrophs (C. paramecium strain KR and Cryptomonas sp. CCAC1634B) and two genomes of the phototroph C. curvata (strains CCAP979/52 and KR). Our five-way comparative investigation within Cryptomonas spp. revealed a mix of conserved and highly variable nucleomorph genomic features. While chromosome-scale synteny was readily apparent across all five genomes (and very high within species), numerous inter-chromosomal rearrangements were apparent, and telomeric repeats were found to be surprisingly variable, even between closely related strains of the same species. The nucleomorph genome of the non-photosynthetic Cryptomonas sp. CCAC1634B was found to be much more similar to the genomes of the two photosynthetic C. curvata species than to those of the non-photosynthetic strains of C. paramecium. However, all three colorless strains examined herein have roughly the same number of plastid-associated genes in their nucleomorph genomes, and it is not clear why the C. paramecium genome is substantially smaller than those of the other examined species. Interestingly, a fine-scale comparison of the KR and CCAP977/2a strains of C. paramecium revealed the presence of numerous fragmented and degraded ORFs, suggesting that genome reduction is ongoing in this species. Determining the extent to which nucleomorph-to-host-nucleus gene transfer has facilitated genome reduction will rely on the availability of nuclear genome sequence data from both photosynthetic and secondarily non-photosynthetic cryptophytes. At the same time, more fine-grain comparisons of the patterns of genome evolution seen in the nucleomorph genomes of non-photosynthetic Cryptomonas species to those in the plastid genomes of the same organisms will be important. Based on the data currently in hand ([16] and herein), common trends are readily apparent, including genome reduction, instances of expected and unexpected gene losses, and pseudogene formation. The extent to which these common patterns are a consequence of the loss of photosynthesis and/or somehow contribute to it is an open question.

Combined with BLAST-based sequence comparisons, investigation of genome synteny allowed us to assign putative functions to a handful of previously hypothetical nucleomorph genes in Cryptomonas strains and species. This is similar to how the sequence of the “large” nucleomorph genome of Ch. mesostigmatica [11] made it possible to ascribe functions to nORFans in other cryptophytes and to show ORF degeneration “in action.” However, it remains the case that many nucleomorph genes within the genus Cryptomonas are still either nORFans or nORFs (i.e., nucleomorph-specific conserved hypothetical proteins). Together with detailed protein structure-based investigations such as those recently carried out by Zauner et al. [41], we will need many more nuclear and nucleomorph genome sequences from within and beyond the genus Cryptomonas, and from diverse red algae as well, if we are to have a complete understanding of the nucleomorph “parts list,” and how nuclear and nucleomorph gene products interact in the nucleomorph, plastid, and periplastidial compartment of cryptophyte cells. Given their propensity to lose photosynthesis, deep genomic sampling of members of the genus Cryptomonas should be particularly revealing.

Methods

Cell culturing and DNA extraction

Clonal cultures of two Cryptomonas species were established from single cells isolated manually from natural habitats by glass pipetting: C. curvata KR (FBCC300012D), from Cheongyang, Korea (36° 30′ N, 126° 47′ E), and C. paramecium KR from freshwater, Daejeon, Korea (36° 21′ 57″ N, 127° 20′ 20″ E). The strains have been deposited in, and are available from, the Freshwater Bioresources Culture Collection at the Nakdong-gang National Institute of Biological Resources and the Protist Culture Collection, Department of Biology, Chungnam National University, Korea. The two cultures were grown in AF-6 medium [42] with distilled water and were maintained at 20°C under a 14:10 light:dark cycle with 30 μmol photons m−2 s−1 from cool white fluorescent tubes. Cultivation of C. curvata CCAP979/52 and Cryptomonas sp. CCAC1634B was carried out as described [16].

Genomic DNAs were extracted from C. paramecium KR and C. curvata KR (FBCC300012D) using the QIAGEN DNEasy Blood Mini Kit (QIAGEN, Valencia, CA, USA) following the manufacturer’s instructions. DNA extractions for C. curvata CCAP979/52 and Cryptomonas sp. CCAC1634B were done using a standard SDS-phenol/chloroform extraction method. For C. curvata CCAP979/52, organelle DNA-enriched fractions (i.e., plastid, mitochondrion, and nucleomorph) were purified as described previously [11].

Genome sequencing and assembly

For C. paramecium KR and C. curvata KR (FBCC300012D), Illumina-based next-generation sequencing was carried out using the MiSeq and HiSeq platforms (Illumina, San Diego, CA, USA). Amplified DNA was fragmented and tagged using the NexteraXT protocol (Illumina), indexed, size selected, and pooled for sequencing using the small amplicon targeted resequencing run, which performs paired end 2 × 300 bp or 2 × 100 bp sequencing reads, according to the manufacturer’s recommendations. C. curvata CCAP979/52 organellar DNA and total genomic DNA of Cryptomonas sp. CCAC1634B were subjected to sequencing library construction using the Nextera XT DNA Library Preparation Kit (Illumina), and DNA sequencing was carried out using a MiSeq instrument (Illumina).

Sequence data were trimmed (base = 80 bp, error threshold = 0.05, n ambiguities = 2) using Trimmomatic 0.36 [43] prior to de novo assembly with the default option (automatic bubble size, minimum contig length =1000 bp). The trimmed reads were assembled into contigs using the SPAdes 3.7 assembler using k-mer size –k 21,33,55,77,99 [44] (similarity = 95%, length fraction = 75%); contigs <1000 bp were excluded. BLAST searches against these assemblies using previously published nucleomorph genes as queries resulted in the identification of putative nucleomorph-derived contigs using Genome Search Plotter [45] in all four newly sequenced species. These contigs were investigated more closely and confirmed to be of nucleomorph origin; their gene contents were similar to the previously published nucleomorph genomes of C. paramecium CCAP977/2a [12] and Ch. mesostigmatica [11]. For chromosome-level scaffolding, we carried out map**-based scaffolding in Geneious Prime 2020 [46] using reference genome C. paramecium CCAP977/2a [12]. Contigs were aligned to the reference genome and their order and arrangement inferred from the alignment.

Gene prediction, annotation, and comparative analyses

To aid in gene annotation, we created a database of protein-coding, rRNA, and tRNA genes from previously sequenced cryptophyte nucleomorph genomes. Preliminary annotation of protein-coding genes was performed using AGORA [47] and GeneMarkS [48]. The final annotation file was checked in Geneious Prime 2020 [46] using ORF Finder (https://www.ncbi.nlm.nih.gov/orffinder/) with the standard genetic code setting. Predicted open reading frames (ORFs) were checked manually with tBLASTn results with AGORA, and the corresponding ORFs (and predicted functional domains) were annotated. Hypothetical ORFs >50 amino acids in size were identified and annotated using the NCBI ORF Finder (standard genetic code). ORFs were searched against the non-redundant protein sequence (nr) database using BLASTp (https://blast.ncbi.nlm.nih.gov/Blast.cgi). ORFs with annotated homologs identified by BLASTp (e-value < 0.05, word size=6) only in nucleomorph genomes were designated “conserved nucleomorph ORFs” (nORFs). Hypothetical ORFs with no obvious similarity to ORFs in any other genome were designated strain-specific “nucleomorph ORFans” (nORFans). For consistency, functional categorization of genes/proteins followed procedures used previously for G. theta [9], H. andersenii [10], C. paramecium CCAP977/2a [12], and Ch. mesostigmatica [11]. The tRNA genes were identified using tRNAscan-SE version 1.21 [49] with the default settings using the “Eukaryotic” sequence source and “Universal” genetic code. To help identify rRNA gene sequences, a set of nucleomorph-encoded rRNA sequences from the public database was used as a query sequence to search our new genomic data using BLASTn. Physical maps were visualized with OrganellarGenomeDRAW 1.3.1 [50]. The previously published nucleomorph genome sequence of C. paramecium CCAP977/2a was downloaded from GenBank [12]. For structural and synteny comparisons, genomes were aligned using GeneCo [51] with default settings. In order to visualize high-level gene order conservation at the intra- or inter-species level, Circos plots were created with Circa (http://omgenomics.com/circa). For three-way inter-species comparisons, blocks of synteny were visualized in a pairwise fashion (i.e., gene order conservation was considered between two species at a time).

Molecular phylogenetics

Phylogenetic analysis was carried out on a 1423-nucleotide alignment of 174 cryptophycean nucleomorph SSU rRNA genes (Supporting Information Fig. S5). The alignment was produced using ClustalW in the program MacGDE2.6 [52, 53]. Bayesian analyses were performed with MrBayes 3.2.7 [54]; the best-fit model was selected by the Bayesian information criterion of jModelTest2 [55], which resulted in the GTR+I+G model being chosen, i.e., the general time-reversible model incorporating invariant sites and among-site rate variation approximated by a discrete gamma distribution. The phylogenetic tree was generated using a random starting tree, two simultaneous runs (nruns = 2) and four Metropolis-coupled Markov chain Monte Carlo (MC3) algorithms for 2 × 107 generations, with one tree retained every 1000 generations. The burn-in point was identified graphically by tracking the likelihood values using TRACER v. 1.6 (http://tree.bio.ed.ac.uk/software/tracer/).