Combined Genome and Transcriptome Analyses of the Ciliate Schmidingerella arcuata (Spirotrichea) Reveal Patterns of DNA Elimination, Scrambling, and Inversion.

Schmidingerella arcuata is an ecologically important tintinnid ciliate that has long served as a model species in plankton trophic ecology. We present a partial micronuclear genome and macronuclear transcriptome resource for S. arcuata, acquired using single-cell techniques, and we report on pilot analyses including functional annotation and genome architecture. Our analysis shows major fragmentation, elimination, and scrambling in the micronuclear genome of S. arcuata. This work introduces a new nonmodel genome resource for the study of ciliate ecology and genomic biology and provides a detailed functional counterpart to ecological research on S. arcuata.

Significance

Our understanding of genome organization in nonmodel ciliates is limited because 1) most species are uncultivable and 2) it requires the separate amplification of a ciliate’s two nuclei (the germline micronucleus and the somatic macronucleus), which is technically difficult. By using single-cell ‘omics, we were able to separately sequence both genomes of the ciliate Schmidingerella arcuata, which revealed patterns of extensive genome rearrangement and fragmentation, and also allowed us to analyze functional details of its somatic genome. This research contributes information on the genomic architecture and functional of an ecologically important ciliate and expands our understanding of ciliate genomics beyond model species.

Introduction

Ciliates are an ancient and diverse clade of microbial eukaryotes that inhabit nearly every environment on Earth. They have long served as models in the research of cell biology and recently have become an ideal system for the study of genome fragmentation and organization (Greider and Blackburn 1985; Lynn 2008; Parfrey et al. 2011). Although genome rearrangements have been discovered throughout the eukaryotes, the phenomenon appears to be especially elaborate within ciliates, which can exhibit the mass elimination, fragmentation, and scrambling of loci (Landweber et al. 2000; Prescott 2000; Chalker 2008; Bracht et al. 2013; Chen et al. 2014; Gao et al. 2015; Maurer-Alcalá, Knight, et al. 2018).

The complexity of ciliate genome architecture results from their nuclear dimorphism, a unique segregation of germline and somatic functions into separate nuclei (Lynn 2008). Chromosomal rearrangements can occur between the germline-limited micronucleus (MIC) and the transcriptionally active macronucleus (MAC). Following conjugation, the zygotic nucleus divides to form a new MIC and a new MAC. In creating the new MAC, a series of rearrangements and deletions can occur, including the extensive elimination of MIC segments called internally eliminated sequences (IESs) (Gratias and Bétermier 2001; Riley and Katz 2001; Katz et al. 2003; Fass et al. 2011). Regions that are not eliminated (i.e., present in both MIC and MAC) are called macronuclear-destined sequences (MDSs) and are cut and pieced together to form MAC loci (Swart et al. 2013) (fig. 1). In some ciliates, MDS regions are arranged out of order in the MIC, a complex pattern of genomic architecture called “scrambling.” Scrambled loci can additionally be “inverted” if they are transcribed on opposing strands of the MIC scaffolds (fig. 1).

Fig. 1

Genome architecture in Schmidingerella arcuata. (A) Patterns of nuclear architecture in the germline MIC and somatic MAC. The MIC contains the required MDSs for the generation of functional genes during development of the MAC. MDSs can be interrupted by IESs. Development of the MAC requires the precise excision of IESs and the correct rearrangement of MDS regions. MDS loci may be scrambled (e.g., MDS 1–4), inverted (e.g., MDS 4), or a combination of both. The organization of these MDSs is guided by pointer sequences (2–10 bp) that occur at MDS/IES boundaries. Green blocks capping ends indicate telomeres, which are added de novo to the ends of MAC chromosomes. (B) Exemplar micronuclear patterns of loci elimination, scrambling, and inversion identified in S. arcuata. (i) Consecutive MDS regions of varying size separated by IESs and guided by 3–5-bp pointers. (ii) MDSs separated by IESs of variable lengths; a mix of scrambled, nonscrambled, and partially inverted loci, with short pointers (2–3 bp). Pointer sequences are shown in white blocks; those appearing twice indicate their secondary location in the MIC (pointers occur twice in MIC and once in MAC). MDSs are numbered according to their somatic order in the MAC. Arrows at the end of each MDS indicate MDS directionality in the MIC. (C) Micronuclear architecture of a beta-tubulin gene in various ciliates. Schmidingerella arcuata separates the gene region into two MDSs, interrupted by a single IES and guided by an 8-bp pointer region. Different colors of MDS correspond to different classes, indicated at right. Accession numbers or gene identifiers: Oxytricha trifallax (PRJNA194431; OxyDB: Contig11167.0.g9), Stylonychia lemnae (X06874.1), Tetrahymena thermophila (L01416.1), Paramecium caudatum (AB070222.1), and Chilodonella uncinata (MH388464).

As most ciliates are not amenable to culture, research on ciliate genomics and nuclear architecture has mostly been limited to a few model species (e.g., Paramecium and Tetrahymena) (Hamilton et al. 2016). In addition, traditional sequencing methods are biased toward highly amplified MAC regions, which has made it challenging to isolate MIC regions. However, the process of multiple-displacement amplification used in single-cell genomics is biochemically biased for long-template DNA (2–70 kb), which allows for the selection of MIC chromosomes (Spits et al. 2006). Combined with single-cell transcriptomics, we are now are able to elucidate the patterns of genome rearrangement, elimination, and scrambling, all from a single cell (Maurer-Alcalá, Knight, et al. 2018).

Here, we use single-cell ‘omics (genomics and transcriptomics) to study the MIC genome and the transcriptome (a proxy for the gene-sized chromosomes of the macronuclear genome) of S. arcuata, a marine ciliate (class Spirotrichea, order Tintinnida). Long used as an ecological model in planktonic food web studies, S. arcuata is ubiquitous in coastal waters, where it periodically dominates the ciliate community (Agatha and Strüder-Kypke 2012; Dolan and Pierce 2013; Santoferrara et al. 2018). Schmidingerella arcuata is also one of the few marine ciliates amenable to culture and thus represents a ciliate that is both ecologically relevant and cultivable (Dolan 2012; Montagnes 2013; Echevarria et al. 2016; Jung et al. 2016; Cobb 2017; Gruber et al. 2019). Although tintinnids have a long history of taxonomic study (Müller 1779; Haeckel 1866), there exists no published data on their MIC or genomic architecture, and only limited transcriptome data exist for Schmidingerella (Keeling et al. 2014), the only tintinnid genus with transcriptome data. Here, we present a genome and transcriptome resource for S. arcuata, acquired using single-cell techniques, and we report on pilot analyses of its genome architecture and transcriptome. This represents a new resource for the study of ciliate genomic architecture and provides a detailed genomic counterpart to ecological research on this model microzooplankton.

Materials and Methods

Culturing

Schmidingerella arcuata was collected from the surface waters of northeastern Long Island Sound, CT (41.31°N, 72.06°W), using a 20-m-mesh plankton net. Single cells were isolated with drawn capillaries and moved to six-well culture plates with 0.2-m-filtered sample water. Clonal cultures of S. arcuata were fed saturating concentrations (∼3 × 103 cells/ml) of the dinoflagellate Heterocapsa triquetra and the prymnesiophyte Isochrysis galbana (strain TISO). Cultures were kept at 18 °C under a 12:12 h light:dark cycle. Morphology and 18S rDNA sequences (see Santoferrara et al. 2013) confirmed the taxonomic identification of S. arcuata (Agatha and Strüder-Kypke 2012).

Isolation of Single Cells

Individuals were transferred from growing cultures to autoclaved, 0.2-m-filtered seawater and starved for 12 h to ensure the clearance and digestion of prey. The cells were then picked and rinsed a minimum of five times in autoclaved, 0.2-m-filtered seawater using drawn capillaries under a stereo microscope. Each cell was then transferred into the appropriate buffer for transcriptome or genome sequencing and brought to volume with nuclease-free water (as specified in the kits detailed below).

Single-Cell Transcriptome and Genome Amplification

The SMART-Seq2 v4 Ultra Low input RNA kit (Cat: 634889; Takara, Mountain View, CA) was used for whole-transcriptome amplification (WTA) following manufacturer’s protocols, with the exception that we quartered the reaction volumes. For whole-genome amplification (WGA), the Repli-g single-cell kit (Cat: 150343; Qiagen, Hilden, Germany) was used following manufacturer’s protocols. The products (cDNA for WTA, gDNA for WGA) were quantified with the dsDNA Qubit assay (Invitrogen, Waltham, MA) and polymerase chain reaction-checked with eukaryotic 18S rDNA (Medlin et al. 1988) and genus-specific ITS (Costas et al. 2007) primers. Minimum bacterial contamination was confirmed by polymerase chain reaction with 16S rDNA primers (Lane 1991). Sequencing libraries were prepared with the Illumina Nextera XT kit (Cat: FC1311096; Illumina, San Diego, CA) then processed with Illumina HiSeq 2500 at Macrogen Sequencing (Geumcheon-gu, Seoul, South Korea).

Transcriptome and Genome Assembly

Raw reads from WTA and WGA sequencing were trimmed for quality and size (Q28 and minimum length of 200 and 1,500 bp, respectively) using BBMap (V38.39; Bushnell 2014) After trimming, two single-cell WTAs were assembled together using rnaSPAdes (V3.13.1; Bankevich et al. 2012), and seven singe-cell WGAs were assembled together using both SPAdes (V3.13.1) and MEGAHIT (V1.2.9; Li et al. 2015) .The MEGAHIT genome assembly was used for the final analysis because it yielded a higher mapping continuity (i.e., the amount of transcripts mapped to the WGA assembly per kilobase). Assemblies were processed through custom python scripts (http://github.com/maurerax/KatzLab/tree/HTS-Processing-PhyloGenPipeline) for the removal of rDNA and prokaryotic transcripts, and for the identification of orthologous gene families using USEARCH (V9.2; Edgar 2010) with OrthoMCLdatabases (V2.0.9; Fischer et al. 2011) (Maurer-Alcalá, Knight, et al. 2018). Additional steps included the prediction of open reading frames with AUGUSTUS (Hoff and Stanke 2019) using an Escherichia coli model to eliminate bacterial contaminants. Stop-codon usage was determined using a custom Python script, which quantified the frequency of in-frame occurrences of TAG/TGA/TAA when each codon was used as a termination site. The completeness of the MIC genome assembly was analyzed using Benchmarking Universal Single-Copy Ortholog (BUSCO; Waterhouse et al. 2018) (E-value <10−3, alveolate lineage database). OmicsBox (V5.2.5; Götz et al. 2008) was used with InterProScan (V5.42; Hunter et al. 2009) and BlastX (NCBI nonredundant database, E-value <10−4; V2.8.1; Altschul et al. 1990) for functional annotation and for the identification of gene ontology (GO) terms involved in Kyoto Encyclopedia of Genes and Genomes (Kanehisa and Goto 2000) pathways.

Genome Architecture Analysis

Custom Python scripts were used to identify and organize the genome architecture of S. arcuata (Maurer-Alcalá, Knight, et al. 2018). Putative MIC loci for MAC MDSs were identified via mapping of the MAC transcriptome to the MIC genome sequences using BLAST. For transcripts to be considered “MIC mapped,” at least 60% of their length was required to be mapped to the MIC (length threshold as suggested in Maurer-Alcalá, Knight, et al. [2018]). Loci were classified into categories of “unmapped,” “nonscrambled,” and “scrambled.” MDS–IES borders were required to be flanked by pairs of short (2–10 bp) tandem repeats called pointer regions to discriminate them from possible intron-exon boundaries (Bracht et al. 2013) (fig. 1). The GC content of MDS–IES boundaries was determined by evaluating the 40 bp located at both ends (i.e., the 5′ and 3′) of an MDS in the MIC.

Results and Discussion

Genome and Transcriptome Resources

Assemblies for the MIC and MAC of S. arcuata are about 49 and 6 Mb in size, respectively (table 1). Of the 11,673 transcripts, which are a proxy for the gene-sized MAC chromosomes of this species, roughly 15% (1,712) were mapped to the MIC. BUSCO analyses estimate that the MIC genome resource is about 19% complete (Complete: 18.8%, Fragmented: 7.6%, Missing: 73.6%, n: 171). This indicates the need for deeper sequencing, although BUSCO analyses have been found to underestimate highly fragmented genomes (López-Escardó et al. 2017) and thus may not be a reliable indicator of completeness in S. arcuata. About 80% of MAC transcripts have significant BlastX hits (NCBI nonredundant database, E-value <10−4), and of those, 74% have a confident assignment of GO terms. The majority of MAC transcripts related to cellular components correspond to membrane and organelle activity, whereas catalytic and binding activities are the primary molecular functions, and localization and biological regulation comprise the main biological terms (supplementary fig. S1, Supplementary Material online). About 15% of MAC transcripts were placed in Kyoto Encyclopedia of Genes and Genomes pathways. The three primary pathways identified are thiamine metabolism, purine metabolism, and aminoacyl-tRNA biosynthesis (supplementary fig. S1, Supplementary Material online). Annotation details regarding the MAC transcriptome can be found in supplementary data, Supplementary Material online, and full annotation files can be found at figshare at the link: https://doi.org/10.6084/m9.figshare.12686621.

Table 1

Summary Data on Micronuclear (MIC) and Macronuclear (MAC) Characteristics of Schmidingerella arcuata

Feature
Size of MIC assembly (Mb)	48.6
Size of MAC assembly (Mb)	6.3
Number of MAC transcripts	11,673
Number of MIC-mapped transcripts	1,718
Percentage of MIC covered by MAC	14.6
Number of scrambled transcripts	616
Percentage of MIC genome that contains scrambled transcripts	35.8
Average %GC content for all MIC-supported scaffolds	46.4
Average %GC content at MDS–IES boundaries	47.5
Average pointer length (bp)	3.7
Average %GC content of pointers	40.8
Average length of scrambled MDSs (bp)	361.7
Stop-codon usage	TGA/TAA

We assessed stop-codon usage and found that the codons TGA and TAA were rarely found in-frame, and their usage in S. arcuata matched homologs in the Oxytricha trifallax transcriptome (GenBank BioSample SAMN02953822). The combination of TGA/TAA as stop codons is not observed in the few published transcriptomes for this genus (Keeling et al. 2014; Heaphy 2018) or for other Spirotrichs, with TAA/TAG reported for Euplotes crassus and TGA for Oxytricha trifallax and Stylonychia lemnae (Kervestin et al. 2001; Lozupone et al. 2001; Swart et al. 2013; Heaphy 2018; Yan et al. 2019). However, ciliates have frequent stop-codon reassignments, and even context-dependent stop codons (Adachi and Cavalcanti 2009; Heaphy et al. 2016; Yan et al. 2019).

Patterns of Genome Architecture

Schmidingerella arcuata shows extensive genome fragmentation, including the unscrambling and inversion of loci during MAC formation (fig. 1). We considered MIC loci as scrambled if their associated transcripts mapped to MDSs that were out of consecutive order in the MIC, and if those MDS–IES boundaries contained pointer regions (Maurer-Alcalá, Knight, et al. 2018). Scrambled loci were determined to be inverted if nonconsecutive MDSs appeared on both strands of germline scaffolds (fig. 1). Of the MIC-mapped transcripts, roughly 36% were found to be scrambled. The average GC content at MDS–IES boundaries for S. arcuata (47.5%) was slightly higher than the overall GC for all germline-supported scaffolds (46.4%); this increase also occurs in other ciliate classes, although most report a sharper rise (10–14%) in %GC around these regions (Maurer-Alcalá, Knight, et al. 2018). Pointer sequence size was variable within and among MDS groups (fig. 1), with a range of 2–10 bp. We found no evidence for alternative processing (more than one MAC sequence resulting from a single MIC region; Katz and Kovner 2010).

Variations in Genome Architecture within and among Ciliate Classes

In general, the micronuclear arrangement of housekeeping genes in S. arcuata matched that of O. trifallax. However, we detected a major housekeeping gene with variable micronuclear organization among six ciliates with available data (fig. 1). In this example, the beta-tubulin gene (paralog 1) is separated into two similar-sized MDSs in the MIC of S. arcuata, interrupted by a single IES and connected by an 8 bp TC-iterative pointer sequence (fig. 1). In contrast, other ciliates of the class Spirotrichea (O. trifallax and S. lemnae) separate the paralog into three or four MDSs, with comparatively shorter IES regions. Model ciliates of the Oligohymenophorea class (Paramecium caudatum and Tetrahymena thermophila) contain this paralog as either three MDS regions of variable size or an uninterrupted sequence in the MIC (Dupuis 1992; Libusová and Dráber 2006). The Phyllopharyngean ciliate C. unicinata divides the MIC gene into three consecutive MDSs of variable size (75–600 bp each) with two 6–7-bp pointer sequences (Harper and Jahn 1989; Zufall and Katz 2007; Katz and Kovner 2010).

Significance of S. arcuata-Omics Resources

This work contributes a MIC genome and MAC transcriptome resource for the ecologically important ciliate S. arcuata. Single-cell omics allowed selective amplification the MIC and MAC, which revealed genomic scrambling, elimination, and inversion in S. arcuata. This study provides a nonmodel genome and transcriptome resource to a field represented mostly by model ciliates. The included annotation details are a valuable resource for future ecological research on S. arcuata and closely related ciliates, which are currently underrepresented in detailed genome-scale analyses. Additionally, research on nonmodel ciliates are beginning to reveal the significance of genome architecture in molecular evolution (Maurer-Alcalá, Ying, et al. 2018; Maurer-Alcalá and Nowacki 2019; Yan et al. 2019). Recent models indicate that slight differences in IESs and specific architectural patterns (e.g., alternative processing and scrambling) among intraspecific ciliate populations can cause rapid incompatibility, potentially leading to incipient speciation (Katz and Kovner 2010; Goldman and Landweber 2012; Gao et al. 2015; Yan et al. 2019). These slight errors in the rearrangement of loci could theoretically accumulate more frequently than (nonneutral) point mutations, which may help to explain the large disparity between the molecular and morphological diversity in ciliates (Gao et al. 2015).

Supplementary Material

Supplementary data are available at Genome Biology and Evolution online.

Click here for additional data file.