Background

Mitochondrial genomes of Viridiplantae, outside of land plants, have been largely understudied. The disparity among mitochondrial genome data available for different classes of Viridiplantae provides a limited view of the diversity belonging to these groups. Being the closest relatives to land plants, studying the evolution of streptophyte algae has prompted the sequencing of multiple genomes from charophycean lineages [1–5]. Most representatives of mt genomes in the class Chlorophyceae (phylum Chlorophyta, sister group to Streptophyta) come from the order Volvocales, a morphologically diverse clade, ranging from the single celled Chlamydomonas to the complex multicellular Volvox. Genome size in Volvocales is thought to be correlated with organismal complexity [6], but overall the mt genomes in this order are highly derived, showing great reduction in gene number and considerable diversity in size and non-coding content.

Recently garnering attention, nine mt genomes were sequenced from the sister group of Volvocales, the order Sphaeropleales, previously characterized by only Acutodesmus obliquus [7, 8]. Through phylogenetic analyses and examination of gene rearrangement and intron content, variation was seen across the sampling of Sphaeropleales. Considered not as derived as in the Volvocales, the mt genome organization and gene content of Acutodesmus obliquus is derived compared to non-chlorophyceans, with a similar general pattern following in the nine other mt genomes surveyed from the order [7, 8]. In this way, A. obliquus has been noted as having an intermediate mt genome among the green algal lineages [8]. Volvocalean mt genomes can be highly compact or bloated with repeats, harbor a reduced number of protein coding genes with a broad range of intron densities and GC content, and assemble as both linear and circular, depicting sphaeroplealean mt genomes as comparatively conserved [7, 9, 10].

The known mt genomes of Sphaeropleales consistently share 13 protein coding genes, 4 rnl and 2 rns fragments, and 22–27 tRNAs. These genomes vary mostly in size and intron content, but it is appropriate to say that the level of mt genome variation is moderate across the order, allowing the genome of A. obliquus to generally characterize the order. Though they are sister groups, sphaeroplealean mt genomes are revealing to be substantially different from those in Volvocales in terms of content and variability.

Morphologically, the family Hydrodictyaceae stands out in the order Sphaeropleales due to the characteristic star-like colonies formed by its representative Pediastrum Meyen 1829, as well as the large, even macroscopic net-like colonies formed by the genus Hydrodictyon Roth 1797 (Fig. 1). In spite of the numerous morphological characters available for species delimitation, the systematics of Hydrodictyaceae has been frequently debated and revised over the last decade [11–14], with DNA sequence data revealing extensive cryptic diversity. Several phylogenetic relationships remain to be clarified in the family, particularly the non-monophyly of Pediastrum duplex (Fig. 1c) and its relationship to Hydrodictyon (Fig. 1a) [13] that represent a taxonomic and evolutionary conundrum. Genome-scale data will likely be necessary to fully understand the molecular and morphological evolution in Hydrodictyaceae.

Fig. 1
figure 1

Morphological diversity in the green algal family Hydrodictyaceae. a Hydrodictyon reticulatum, scale bar = 2 mm, b Pseudopediastrum boryanum, scale bar = 10 μm, c Pediastrum duplex, scale bar = 10 μm, d Stauridium tetras, scale bar = 10 μm

Full size image

To contribute to the sampling of mt genomes in Sphaeropleales and test the phylogenetic patterns of gene arrangements and genome architecture within a family, we offer four fully assembled hydrodictyacean mt genomes, as well as protein-coding gene sequences from the fragmentary mt assembly of Hydrodictyon reticulatum. As Fučíková et al. [7] only presented fragmentary genome data from Hydrodictyaceae (represented by Pediastrum duplex), our study is the first one to report the full mt genomes and is able to investigate the mt genome architecture in this family. We purport that the variability seen thus far across the order is present within a single family, as has been shown for Bracteacoccaceae and Neochloridaceae. These data show promise for resolving phylogenetic relationships among taxa within Hydrodictyaceae while also lending themselves to aid in the further understanding of mt genome evolution across the green algal phylogeny.

Results

Data collection resulted in 9.8 million reads for Stauridium tetras [GenBank:KR026341], 11.1 million reads for Pseudopediastrum boryanum [GenBank:KR026342], 8.7 million reads obtained for Pediastrum duplex AL0403MN [GenBank:KR026339], and 7.6 million reads for Pediastrum duplex PL0501b [GenBank:KR026340]. All four mitochondrial genomes assembled with no gaps, the mean coverage for each was as follows: P. duplex AL0403MN 381.3X, P. duplex PL0501b 160X, Ps. boryanum 86.4X, and S. tetras 15.6X. Each genome assembled as a circular molecule (Additional files 1, 2, 3 and 4) and contained 4 rnl gene fragments and 2 rns fragments, 24 tRNAs (except Pediastrum duplex PL0501b with 23 tRNAs), and 13 protein coding genes including nad1, nad2, nad3, nad4, nad4L, nad5, nad6, cob, cox1, cox2a, cox3, atp6, and atp9. The smallest mt genome belongs to S. tetras with 37,723 bp. The mt genome of Ps. boryanum has a length of 42,110 bp, and P. duplex AL0403MN and P. duplex PL0501b mt genomes are 46,416 and 53,560 bp, respectively.

A complete mt genome assembly of Hydrodictyon reticulatum was not possible due to contamination and consequently low coverage. Nine of the 13 protein coding genes expected to be present in all Sphaeroplealean mt genomes were sequenced in full and while complete sequences were not recovered for atp9, cox1, nad2, and nad4L, the portion of missing data was minimal (1.7 % missing for Hydrodictyon and 0.5 % total missing characters in the final trimmed alignment).

The position of nad4L through trnL-CUA comprises a conserved cluster of 19 genes in all four complete genomes (Fig. 2). The gene cluster trnD-GUC, nad3, cox3, rnl2, rns2 is also conserved in each genome. Nad1 has a completely different position in each genome. The rns and rnl gene fragments are clustered together in each genome, though not in identical orders. Moreover, the order of the fragments is scrambled (i.e., rnl1, rnl2, rnl3, and rnl4 are not arranged consecutively), but scrambled in the same manner across the genomes (Fig. 2).

Fig. 2
figure 2

Mauve alignment of four linearized Hydrodictyaceae genomes, beginning at the gene cob. Blocks of genes (represented by corresponding colors) are labeled in Stauridium tetras. Boxes under each genome map represent protein-coding genes (white boxes), rRNA genes (red boxes) and tRNA genes (small green boxes)

Full size image

Introns are present only in cox1, cob, and rnl4 (Fig. 3). Stauridium tetras, Hydrodictyon, and P. duplex UTEX LB1364 possess a homologous group II intron in cox1, while P. boryanum and P. duplex AL0403MN contain a group IB intron at a nearby position in cox1. Pediastrum duplex PL0501b has a group ID intron roughly in the middle of cox1. Introns are present in the cob gene in P. duplex AL0403MN, Ps. boryanum, and Hydrodictyon - the latter two are homologous group ID introns, while the first is group II and appears related to the cox1 group II introns. The rnl4 gene is uninterrupted in Ps. boryanum, S. tetras, P. duplex UTEX LB1364, and possibly also in Hydrodictyon (for which only incomplete rnl4 data are available). A group IA intron is inserted in the rnl4 gene of P. duplex AL0403MN and a group IB intron is present in P. duplex PL0501b, at a different position.

Fig. 3
figure 3

Mitochondrial intron types and distribution in three mitochondrial genes of the order Sphaeropleales

Full size image

The tRNA specifying threonine is absent from all genomes, and keeping with the deviant genetic code of Acutodesmus obliquus, leucine is coded for by UAG (normally a stop codon). The sphaeropleales-specific stop codon UCA is used infrequently in comparison to UAA, but terminates nad1 in all examined hydrodictyaceans. Stauridium tetras is an exception, as this taxon uses UCA in 9 of the 13 protein-coding genes and only uses UAA in atp6, cox1, cox2a, and nad6.

Phylogenetic analyses

Results of all phylogenetic analyses, including all five ML searches and the Bayesian consensus tree, agreed on the same, completely resolved topology (Fig. 4). All relationships among Sphaeropleales corresponded to the mitochondrial tree presented in Fučíková et al. [7], including Neochloris being the closest relative of Hydrodictyaceae and Chlorotetraedron being sister to Neochloris + Hydrodictyaceae. Relationships among the six included Hydrodictyaceae were well supported, with Stauridium being sister to the remaining representatives, Pseudopediastrum boryanum sister to Hydrodictyon + Pediastrum duplex. Pediastrum duplex was recovered as monophyletic with absolute support. Badger analysis of gene rearrangements within Sphaeropleales grouped the four representatives of Hydrodictyaceae as a group with 0.96 support (Additional file 5). Pseudopediastrum boryanum + S. tetras were resolved as sister to P. duplex AL043MN + P. duplex PL0501b.

Fig. 4
figure 4

Phylogenetic tree yielded by ML and Bayesian analyses of 13 mitochondrial protein-coding genes. Bayesian consensus tree is shown with support indicated by numbers at branches, representing Bayesian posterior probabilities (BPP) and ML bootstrap (BS) support values, respectively. Scale bar represents the estimat­ed number of substitutions per site. Bars on the right represent mt genome size and composition, as well as total intron number for each taxon

Full size image

Discussion

Mitochondrial genomes of Viridiplantae range from hyperinflated and hypervariable in land plants to compact and conserved in some lineages of green algae. From the information available for the class Chlorophyceae, Sphaeropleales appear to be closer to the conservative end of the spectrum, especially compared to their sister order Volvocales [6, 7]. Our study zooms in on the sphaeroplealean family Hydrodictyaceae and compares four fully sequenced and two partially sequenced mt genomes from taxa that span the phylogenetic breadth of the family. This study is the first to report a complete mt genome sequence for the family Hydrodictyaceae, and among the first ones to compare within-family mt genome diversity in green algae.

Genome size and structure

The large size difference among genomes of different strains is primarily due to expansion of intergenic regions (spacers), and the number of introns does not scale directly with genome size due to high variation in size of individual introns. This correlation appears to be the case across Sphaeropleales: for example, Neochloris’s 11 introns (highest intron number in Sphaeropleales) combined are of lesser length than the 9 introns of Kirchneriella (second-largest mt genome in Sphaeropleales; Fig. 4). The newly sequenced Pediastrum duplex PL0501b has the largest reported mt genome in Sphaeropleales but contains only two introns.

The deepest-diverging hydrodictyacean, Stauridium tetras, has a genome size very similar to those of the closest known relatives of Hydrodictyaceae: Neochloris and Chlorotetraedron (Fig. 4). It is therefore possible that the ancestral mt genome size for the Neochloridaceae + Hydrodictyaceae clade was in the vicinity of 38 kb. However, in contrast to S. tetras and Chlorotetraedron, Neochloris has a much higher intron content and lower spacer content, which makes such an interpretation of genome size evolution less straightforward as there may be different driving forces behind the proliferation of introns and contraction of intergenic spacers. In the P. duplex group the mt genomes appear to have expanded both due to intron acquisition and spacer expansion (especially in P. duplex PL0501b).

Within Chlorophyta, the cox2 gene shows varying degrees of fragmentation and migration. Algae belonging to Prasinophyceae, Ulvophyceae, and Trebouxiophyceae contain orthodox, intact mitochondrial cox2 genes [15]. Chlorophycean algae exhibit split cox2 genes, possessing either a mitochondrion-encoded cox2a and a cox2b fragment that has migrated to the nucleus (Scenedesmus-like) or having both cox2a and cox2b relocated to the nucleus following a reduced-derived pattern (Chlamydomonas-like) [15]. As expected, the four newly sequenced genomes presented appear to have a mitochondrial cox2a gene, emphasizing the intermediate evolutionary pattern of Sphaeropleales and building a strong case for distinguishing phylogenetic relationships.

Orientating the gene arrangements so that cob begins and nad6 ends the sequence, gene rearrangement is prominent in the regions flanked by the rRNA fragments and “end” of the genomes in both P. duplex, while in S. tetras and P. boryanum gene rearrangement is present only from the “beginning” to the end of the rRNA fragments (Fig. 2). Additional patterns of gene rearrangements are discussed below.

Intron diversity

Considerable diversity and abundance of introns in the cob, cox1, and rnl4 genes was previously described for sphaeroplealean mt genomes [7]. Compared to the diversity found across the order, Hydrodictyaceae have relatively intron-poor mt genomes and, in the strains examined here, no more than one intron per gene is found. Yet, all of the major intron types (groups IA, IB, ID, and II) found across Sphaeropleales are found in Hydrodictyaceae as well (Fig. 3). The group II intron found near the 3′ end of cox1 in Hydrodictyon, Pediastrum duplex UTEX LB1364, and Stauridium is unique among Sphaeropleales. Interestingly however, BLAST analyses show its relatedness to the cob gene of P. duplex AL0403MN (71–89 % identity in conserved intron regions including the intronic ORF - a putative reverse transcriptase gene), which appears homologous with an identically positioned cob intron in Chromochloris zofingiensis (67 % identity in the conserved intron domain V; Fig. 3). A notable 67 % identity in the conserved regions with a cox1 intron of the red alga Pyropia was also detected (however, the intron of Pyropia is inserted at a different cox1 position). The ancestry of this intron is unclear from the available data, but it can be hypothesized that the intron was either present ancestrally in the cob gene of all Sphaeropleales and lost many times (even within Hydrodictyaceae), or transferred horizontally into the cob gene of P. duplex AL0403MN. Either way, this intron may have been present in the ancestral mt genome of Hydrodictyaceae in the cox1 gene (where it likely arrived from the cob gene) and subsequently lost in some members of the family. Much denser taxon sampling will be required to accurately reconstruct the undoubtedly complex evolutionary history of these introns.

The group ID intron found in the cob gene of Hydrodictyon and Pseudopediastrum appears homologous to the same type introns present in several other sphaeroplealeans (especially Chlorotetraedron and Chromochloris - 68–69 % identity over 50 % or more of the intron’s length), and might therefore be ancestral to the order. In fact, its conserved 3′ end bears significant sequence similarity (78–80 % as determined by BLAST) to identically positioned cob introns of Volvox carteri and other green algae including Trebouxia aggregata (Trebouxiophyceae) and the streptophytes Chlorokybus atmophyticus, Mesostigma viride, and Chaetosphaeridium globosum, as well as several fungi (also similar position in cob). The original invasion of this intron into mt genomes might therefore reach deep into the green algal, or even eukaryote, phylogeny.

Phylogeny of Hydrodictyaceae

Previous studies concerning the systematics of Hydrodictyaceae split Pediastrum duplex into two clades, of which “group I” (in our study represented by UTEX LB1364 and AL0403MN) was more closely related to Hydrodictyon than to “group II” (here represented by PL0501b). These relationships were mainly supported by the nuclear 28S ribosomal large subunit gene [12, 13] and weakly indicated also by results based on the nuclear ITS-2 and chloroplast rbcL, psaA, psaB genes [13, 16], unpublished data. Contrarily, our study recovered P. duplex as a monophyletic species with strong statistical support with Hydrodictyon as its sister taxon (Fig. 4). Of course, the present taxon sampling is sparse compared to the studies of nuclear and chloroplast genes, which included dozens of hydrodictyacean isolates. However, it is worth noting that the 28S gene, which supports non-monophyly of P. duplex, only contained 94 parsimony-informative characters (less than 5 % of the data set) [13]. The rbcL gene is more variable (21.4 % or 253 parsimony-informative sites) [13] but does not provide any statistical support for the grouping of P. duplex group I and Hydrodictyon. In contrast, our mt data set contains 6473 parsimony-informative sites (60.6 %), and when reduced to reflect the phylogenetic breadth included in McManus and Lewis [13], i.e., Hydrodictyaceae with Neochloris as outgroup, 1755 parsimony-informative characters remain (16.4 %).

The pattern of gene rearrangement, along with other characteristics seen in the genomes, is reflected in the phylogeny (Figs. 2 and 4). However, without a much denser sampling of Hydrodictyaceae and their closest relatives, it is impossible to determine the ancestral mt genome structure in the family, with the exception of a handful of conserved blocks of genes. Stauridium tetras and Ps. boryanum, the two deepest diverging hydrodictyaceans sampled in this study, share the same gene order over a span of 24 genes (from nad4 to trnL-CUA). This block of genes might be ancestral to Hydrodictyaceae but is not found in closest relatives of the family (Chlorotetraedron and Neochloris, both of which have mt genomes very differently arranged and are therefore not included in Fig. 2 for simplicity). A cluster of 10 tRNA genes forms a consistent block with the cob gene, and this block is likely ancestral to Hydrodictyaceae. The majority of the tRNA gene cluster is also found in Chlorotetraedron and Neochloris (but not other Sphaeropleales), but it follows nad1 instead of cob. The nad3-rns2 block and the rnl4-rnl1 block are in close proximity to each other and on the same strand, with the exception of Ps. boryanum. The gene order in both P. duplex genomes is identical from cob to rnl1, and toward the “end” from nad4 to nad6. In P. duplex PL0501b, the rnl4-rnl1 block and the adjacent cox1 gene appear to have relocated as a single unit relative to AL0403MN. Additional mt genomes from across the family (in particular P. angulosum) are necessary to determine if gene order will aid with resolving phylogeny, and especially to reconstruct the relative timing of rearrangement events that shaped the extant genomes.

Conclusions

Presenting the first mt genomes of the family Hydrodictyaceae, and being one of the only family-level studies of mt genome diversity in the green algae, this study highlights the potential of phylogenetic informativeness and resolving power of fine scale, whole genome studies. Even though this study lacks the taxon sampling to credibly resolve the question of monophyly of P. duplex, our data nevertheless show great promise for addressing systematic questions in Hydrodictyaceae. Expanding the representation of mt genome data across the family will further aid in our understanding of mt genome evolution across the green algae.

Methods

Strains of Hydrodictyaceae were collected and isolated from various fresh bodies of water in the U.S.A. and Australia following procedures from McManus and Lewis [13], or from public culture collections (UTEX, ACOI) (Table 1). Locality information for the strains is included in McManus and Lewis [13]. Cultures were maintained at 20 °C under a 16:8 h light:dark (L:D) cycle on agar slants. The agar slants consisted of a 50:50 mixture of Bold’s basal medium (BBM) [17, 18] and soil water prepared following McManus and Lewis [13].

Table 1 Comparison of mitochondrial genomes of Hydrodictyaceae and their closest sphaeroplealean relatives
Full size table

Genomic DNA was extracted from living cells using CTAB extraction or the PowerPlant® Pro DNA Isolation Kit by MoBio Laboratories, Inc. (Carlsbad, CA). DNA was then submitted to Cold Spring Harbor Laboratories for TruSeq library preparation followed by sequencing on Illumina HiSeq2500 to produce 2×100 bp paired reads. Hydrodictyon reticulatum was sequenced on a MiSeq instrument at the University of Connecticut set for 250 bp long paired-end reads. Geneious v.6.1.5 (www.geneious.com) [19] was used to pair, trim, assemble, and annotate reads. Contigs of mitochondrial origin were sorted from the paired reads and served as a baseline for the initial map to reference. Assembly was achieved through a series of reference mapping iterations and de novo assembly of contigs until a single, final contig resulted. To verify, paired and trimmed reads were mapped to the final consensus sequence followed by a de novo assembly of the reference reads and reassembly of the contigs produced. Annotations, comprising protein coding genes, large and small ribosomal subunit genes, and tRNAs, were completed for each genome utilizing DOGMA (dogma.ccbb.utexas.edu/) [20], RNAWeasel (http://megasun.bch.umontreal.ca/cgi-bin/RNAweasel/RNAweaselInterface.pl) [21], tRNAscan-SE 1.21 [22], BLAST (http://blast.ncbi.nlm.nih.gov/) [23], Geneious, and previously annotated genomes. Intron type and homology was determined using RNAWeasel and blastn [24, 25]. Preliminary annotations for protein coding genes and large and small ribosomal subunit genes were done with DOGMA and imported into Geneious. After mapping DOGMA annotations to their corresponding genome, open reading frames (ORFs) correlating to Scenedesmus (Acutodesmus) obliquus were found using Geneious to verify the start and stop of each gene in conjunction with DOGMA. Annotations from published genomes were used to manually find and annotate rRNA gene fragments in our genomes. Annotations of tRNAs were completed with RNAWeasel and verified using tRNAscan-SE. Gene alignments were created for all protein coding sequences through a translational alignment in Geneious and rnl and rns fragments were aligned as well (Additional file 6). A Mauve alignment of all four complete hydrodictyaceae mt genomes was produced in which each genome was oriented to begin at cob (Fig. 2) [26]. For Hydrodictyon reticulatum, protein mt gene sequences were recovered by reference assembly to P. duplex (UTEX 1364) gene sequences in Geneious and subsequent reference assemblies to the resulting Hydrodictyon consensus sequences.

Phylogenetic analyses

Alignment of individual genes and removal of unalignable characters was performed as described in Fučíková et al. [7]. Combined Maximum Likelihood (ML) and Bayesian phylogenetic analyses of the 13 protein-coding mt genes were carried out using the same taxon set, with added hydrodictyacean representatives, and applying the same partitioning scheme and model of evolution (GTR+I+Γ) used in Fučíková et al. [7]. Phycas v.2 [27] was used for Bayesian inference, carrying out 100,000 cycles with polytomies allowed, sampling every 100 cycles. The first 200 trees of the run were discarded as burn-in and Bayesian marginal posterior probabilities (BPP) of splits were estimated from the post-burn-in sample of trees. RAxML [28] was used for a ML search for best tree with five independent searches, and a bootstrapping analysis was carried out for 1000 pseudoreplicates. Badger v.1.02 [29] was used to infer the phylogeny of Sphaeropleales from mt gene order. Badger was run for 5,000,000 generations with otherwise default settings. The data sets supporting the results of this article are included as Additional files 6 and 7.

Availability of supporting data

All supporting data are included as additional files.