A novel kleptoplastidic symbiosis revealed in the marine centrohelid Meringosphaera with evidence of genetic integration

Summary

Plastid symbioses between heterotrophic hosts and algae are widespread and abundant in surface oceans. They are critically important both for extant ecological systems and for understanding the evolution of plastids. Kleptoplastidy, where the plastids of prey are temporarily retained and continuously re-acquired, provides opportunities to study the transitional states of plastid establishment. Here, we investigated the poorly studied marine centrohelid Meringosphaera and its previously unidentified symbionts using culture-independent methods from environmental samples. Investigations of the 18S rDNA from single-cell assembled genomes (SAGs) revealed uncharacterized genetic diversity within Meringosphaera that likely represents multiple species. We found that Meringosphaera harbors plastids of Dictyochophyceae origin (stramenopiles), for which we recovered six full plastid genomes and found evidence of two distinct subgroups that are congruent with host identity. Environmental monitoring by qPCR and catalyzed reporter deposition-fluorescence in situ hybridization (CARD-FISH) revealed seasonal dynamics of both host and plastid. In particular, we did not detect the plastids for 6 months of the year, which, combined with the lack of plastids in some SAGs, suggests that the plastids are temporary and the relationship is kleptoplastidic. Importantly, we found evidence of genetic integration of the kleptoplasts as we identified host-encoded plastid-associated genes, with evolutionary origins likely from the plastid source as well as from other alga sources. This is only the second case where host-encoded kleptoplast-targeted genes have been predicted in an ancestrally plastid-lacking group. Our results provide evidence for gene transfers and protein re-targeting as relatively early events in the evolution of plastid symbioses.

Graphical abstract.

Introduction

The evolution of plastids involved gains, losses, and replacements, spreading photosynthetic capabilities to most eukaryotic supergroups and creating a complicated pattern across the tree of life.¹ Eukaryotes first acquired photosynthesis in an ancestor of Archaeplastida through endosymbiosis with cyanobacteria, establishing the primary plastids.^2–4 Subsequent endosymbioses between eukaryotes have spread these plastids into new lineages on multiple independent occasions.^5,6 In addition to the acquisition of these permanent organelles, plastids can be acquired temporarily either by symbiotic interactions with microalgal whole cells (photosymbiosis) or by retaining plastids from prey species by kleptoplastidy, which enables the host to transiently acquire photosynthetic capabilities. Kleptoplastidy is increasingly recognized as a common interaction in aquatic ecosystems, occurring in a broad range of host species, including a few multicellular eukaryotes such as the saccoglossan sea slugs,^7,8 some marine flatworms,⁹ and a diverse list of protist hosts (including foraminiferans,¹⁰ ciliates,^11,12 and dinoflagellates^13,14). Kleptoplastidic interactions can be temporarily stable and thus offer unique insights into the establishment of plastids.¹⁵

Kleptoplastidy is a temporary association because the host cells cannot maintain stolen plastids indefinitely. In permanent plastids, a vast amount of protein import from host-encoded genes is required to compensate for their highly reduced plastid genomes.¹⁶ The imported proteins have mixed origins: some are the products of endosymbiotic gene transfer (EGT), others co-opted host proteins, and still others originate from horizontal gene transfers (HGTs).^17,18 In the absence of this suite of imported proteins, kleptoplasts eventually degrade, and so the host must re-acquire plastids.¹⁹

Although some host species do little to counter the degradation of their kleptoplasts, other species actively support their stolen organelles and, in some instances, dramatically extend their survival. For example, the Antarctic Ross Sea dinoflagellate (RSD) can maintain its haptophyte-derived kleptoplasts for at least 30 months without replacement,²⁰ probably by targeting host-encoded proteins to the kleptoplasts and thereby providing some of the necessary proteins.²¹ Interestingly, the nuclear genes that are targeted to the kleptoplasts often originated from HGT and not EGT, which aligns with the growing evidence of the importance of HGT for symbiotic integration.^6,22,23 To date, the presence of kleptoplast-targeted proteins in the host nucleus has only been documented in three cases: in RSD, in another dinoflagellate, Dinophysis acuminata, that has cryptophyte-derived kleptoplasts,²⁴ and in the euglenozoan Rapaza viridis that has green alga-derived kleptoplasts.²⁵ However, the survival of kleptoplasts can be extended with a variety of other strategies. For instance, sacoglossan sea slugs exhibit shielding strategies that protect their kleptoplasts from potentially damaging light intensity.^26–29 Similarly, some benthic foraminifera have been hypothesized to use behavioral strategies to protect their kleptoplasts from high light.³⁰ Alternatively, in karyoklepty, the nucleus of the prey is also retained, and it remains transcriptionally active and services the kleptoplasts by providing the necessary nuclear-encoded proteins. This was first documented in the ciliate Mesodinium rubrum (also called Myrionecta rubra),³¹ where the cryptophyte nuclei are retained for up to 30 days, and has since also been identified in the dinoflagellate Nusuttodinium aeruginosum.³² The diversity within host strategies used to control and extend kleptoplast survival remains an open question, as well as how many of these strategies involve host-encoded genes.

The import of host-encoded proteins into plastids is typically considered a benchmark of permanent and stable plastids. The prediction of protein import in kleptoplastidic systems indicates that a degree of genetic integration can already be achieved in temporarily retained plastids. It has been proposed that kleptoplastidic interactions controlled by host-encoded genes could represent a “tipping point” to stable genetically integrated plastids.^15,21 Furthermore, the acquisition of foreign genes (e.g., HGT) by a kleptoplastidic host could provide pre-adaptations to future plastids. In this way, kleptoplastidy fits within the “shopping bag” model of plastid evolution,³³ in which the serial uptake of symbionts provides necessary pre-adaptations required for permanent plastid establishment. Under this model, protein import establishes prior to permanent plastids and acts as an evolutionary ratchet that enables plastid fixation.^34–36

In this study, we investigated a little-known marine protist, Meringosphaera, and the autofluorescent “green bodies” consistently observed internally. Meringosphaera is a globally distributed genus with distinctive undulating silica spines,^37,38 which can reach dominance in planktonic assemblages.^39,40 Obiol et al.⁴¹ reported that Meringosphaera (referred to as Centrohelida-sp1) was “the most widespread and abundant ASV” (amplicon sequence variant) during the Malaspina 2010 Circum-navigation Expedition. Despite this, Meringosphaera remains remarkably understudied. First characterized in 1902,⁴² it has historically been treated as a chrysophyte-related alga due to the consistent presence of green bodies presumed to be chloroplasts.^38,43 Recently, the first Meringosphaera 18S rDNA sequence was obtained, which led to its re-classification as a centrohelid (Haptista: Centroplasthelida), in the environmental marine clade NC5.⁴⁴ Importantly, there are no permanent plastids known in centrohelids,⁴⁵ though there are two suggested cases of kleptoplastidic freshwater species⁴⁶ and endosymbiotic Chlorella have been found in the centrohelid Acanthocystis turfacea.⁴⁷ Centrohelids are typically not considered to have been ancestrally photosynthetic.⁴⁸ The origin of the green bodies in Meringosphaera is, therefore, unknown.

Since Meringosphaera is uncultivated, we used a suite of culture-independent approaches on surface water collected monthly from the North Sea (West coast of Sweden) to characterize the fluorescent bodies observed in the cells. Following manual isolation of Meringosphaera cells, we generated single-cell amplified genomes (SAGs) to study the genetic identity of the plastids and to search for evidence of genetic integration via host-encoded plastid-associated genes. Specific quantitative PCR (qPCR) assays were designed for environmental monitoring of Meringosphaera host and plastids over 14 months of sampling. Catalyzed reporter deposition-fluorescence in situ hybridization (CARD-FISH) was used with confocal microscopy to confirm the 18S rRNA of the host Meringosphaera and to visualize the presence/absence of plastids. Overall, the results suggest that Meringosphaera exhibits selective kleptoplastidy with evidence of host-encoded plastid-associated genes. The exact identity of the kleptoplasts depends on the season and the host lineage. Except for the euglenozoan Rapaza viridis, this is the only other known case of genetic integration in a kleptoplastidic host outside of dinoflagellates. This is particularly significant because centrohelids, unlike dinoflagellates but like Rapaza viridis, are not ancestrally plastid-bearing, and thus these host-encoded genes arose in a nuclear genome that had not previously co-evolved with a plastid.

Results

Origin and diversity of Meringosphaera plastid association

We used 15 manually isolated Meringosphaera cells, collected in October and November 2018, to perform individual genome amplification and assemble SAGs. Using the previously reported Meringosphaera 18S rDNA sequence as query (GenBank: MZ240752⁴⁴), we found by BLASTn a highly similar 18S rDNA sequence (95%–99% identity) in every SAG. Using two centrohelid 28S rDNA sequences as queries (GenBank: JQ245080 and AY752993), we further identified a 28S rDNA sequence in every SAG. When placed in a concatenated 18S and 28S rDNA phylogeny alongside representative centrohelid sequences (Figure 1A), the Meringosphaera sequences and a few environmental sequences formed a well-supported monophyletic clade (bootstrap percentage[BP] = 95%). The SAGs were split into two groups (“group 1” and “group 2,” BP = 100% and 92%, respectively), representing previously unknown genetic diversity.

Figure 1. Diversity and phylogeny of the Meringosphaera host and its plastid from the Swedish west coast.

(A) Maximum likelihood tree of selected representative concatenated 18S and 28S rDNA sequences showing Meringosphaera within the Centroplasthelida. The tree was reconstructed with a GTR model, with 4 gamma categories, and support values correspond to rapid bootstrap with 1,000 replicates. The two groups are designated as group 1 and group 2, these two groups corresponded to clustering the Meringosphaera 18S rDNA sequences by >97% identity (the separate 18S and 28S phylogenies are available in the Figshare repository D2.1). The Meringosphaera sequences written in bold were generated by this project, and all other previous sequences are unidentified environmental sequences. Haptophyte sequences were used as an outgroup.

(B) Maximum likelihood tree of selected cultured and representative 16S rDNA sequences showing the Meringosphaera plastids within the Dictyochophyceae plastids. See Figure S1 for the phylogenies of the additional plastid genes psbA and rbcL. The tree was reconstructed using a TIM3 + F + I + G4 model, and support values correspond to ultrafast bootstrap values from 1,000 replicates. The Meringosphaera plastid sequences written in bold were generated by this project. Pelagophyceae sequences were used as outgroup. In both trees, support values over 50% are shown on the trees, with values over 90% represented by a black circle on the branch. See STAR Methods for details regarding the tree formation.

To assess whether the green bodies commonly observed within Meringosphaera correspond to full endosymbiont cells, we searched for any additional 18S rDNA sequence in the SAGs using a diverse set of reference sequences (see STAR Methods). In only two SAGs did we identify a second 18S rDNA sequence (S5 and S7, both with 100% sequence similarity with the green algae Nannochloris sp.); however, given that it was rare, inconsistent across the samples, and mismatched to the plastid identity (see below), we considered it unlikely to correspond to the observed fluorescent bodies. In the remaining 13 SAGs, apart from the Meringosphaera 18S rDNA, no other 18S rDNA sequences were found. This suggests the absence of another eukaryotic nucleus (from, e.g., prey or photosymbionts), and thus that the fluorescent bodies likely correspond to plastids.

The identity of the plastids was determined by extracting 16S rDNA sequences from the SAGs. Despite that all picked cells appeared to contain green fluorescent bodies, we only found 16S rDNA plastid sequences in eight of the 15 SAGs. A 16S rDNA phylogeny showed that all eight plastid sequences are closely related to plastids from Dictyochophyceae and a few environmental sequences but were clearly distinct from the few described species (Figure 1B). The plastid phylogeny based on maximum likelihood showed a similar pattern to the host 18S rDNA phylogeny, with two distinct and well-supported clades (group 1 plastids BP = 97%, group 2 plastids = 100 %). In order to confirm the pattern given by the 16S rDNA plastid phylogeny, we also constructed phylogenies for two other plastid molecular markers: the large subunit of ribulose 1,5-bisphosphate carboxylase/oxygenase (rbcL) and the D1 protein of photosystem II (psbA). Both genes were only recovered in a subset of the SAGs (but fully overlapping with the eight SAGs with plastid 16S rDNA sequences): rbcL in 11 SAGs and psbA in nine SAGs (summarized in Table S1). Congruent with the 16S rDNA phylogeny, the rbcL and psbA sequences from the Meringosphaera plastids were most closely related to Dictyochophyceae and partitioned into two distinct groups (Figure S1), providing additional evidence for the plastid origin and that distinct plastids are found in the two major host groups. Interestingly, for all three of the plastid genes, the branching pattern is congruent with the 18S rDNA phylogenies, i.e., the group 1 hosts have one type of plastid, whereas the group 2 hosts have the other.

Seasonal dynamics of Meringosphaera and its plastid

To investigate the seasonal dynamics of Meringosphaera and its plastids, we surveyed over a year of samples and three locations with targeted qPCR assays that distinguish between group 1 and group 2 for both the host Meringosphaera 18S rDNA and the plastid-encoded rbcL genes. The assay was applied to whole water samples, meaning that the plastid rbcL detection could originate from plastids either inside Meringosphaera or inside free-living Dictyochophyceae. The qPCR assay was designed based on the SAG data from October to November 2018, and as such the oligonucleotides could be biased toward the groups found during that time of year. The 18S rDNA results (Figure 2A) show that Meringosphaera group 2 was present throughout the year (mean ~ 1.1 × 10⁶ gene copies L⁻¹), but gene abundance varied throughout the months (ANOVA, F_11,45 = 8.53, p < 0.001). The highest gene copies (~1.5 × 10⁷ gene copies L⁻¹) were detected in April, in what could be a spring bloom. Otherwise, in general, the gene abundances were higher (~1.3 × 10⁵ gene copies L⁻¹) in late summer after which they decreased through autumn down to very low frequencies in winter (range of 3–374 gene copies L⁻¹). By contrast, the Meringosphaera group 1 host was only detected in October and November of 2021. The presence of group 1 host in 2021 and not 2020 could have biological reasons but may also be explained by sampling strategy given that in 2020 only one of the three sites was sampled.

Figure 2. Seasonal dynamics of Meringosphaera and its plastids.

(A) Abundance of Meringosphaera group 1 and 2 and respective plastids detected in the monthly sampling from the Swedish west coast. The top panel shows the number of 18S rDNA gene copies L⁻¹ for the Meringosphaera group 1 and group 2 and the lower panel shows the respective number of rbcL gene copies L⁻¹ forthe plastid group 1 and group 2. The group identity of the hosts and plastids is indicated bythe colorofthe points.The abundance is reported as the Log of the gene copies L⁻¹. The samples were taken monthly between March 2021 and February 2022, in addition two extra samples were taken in October and November 2020, and these are indicated by triangle-shaped points. There is variation in the number of locations sampled, which is why the number of points varies between the months (see Figure S2 and STAR Methods for details). If the target was detected in some but not all of the technical replicates, the sample was marked as detected but not quantifiable and was plotted with a value of 1.1. Data are represented as the mean ± SEM. See also Figure S3 for the hydrographic conditions at the sampling stations.

(B–E) Images of Meringosphaera and its kleptoplasts collected in April 2021 from the Å17 sampling station and visualized by CARD-FISH, showing that the plastids are numerous and intact despite the lack of detection in the qPCR assay. The cells were imaged on a confocal microscope with the 405, 488, and 633 nm lasers. In all panels, the green signal is from the Alexa 488 dye bound to the Mer482 probe that was designed to specifically target Meringosphaera 18S rRNA, the blue signal is from the nucleic acid stain DAPI, and the red signal is from chlorophyll a autofluorescence. (B)–(E) are images taken from the same cell, (B)–(D) show each channel individually, and (E) shows the overlay of the three channels. Scale bars, 2 μm. Video S1 is composed of the z stack images taken from this same cell.

The seasonal dynamics of the plastids only partially corresponded to that of the Meringosphaera host (Figure 2A). Neither plastid groups were detected from January to June, and in July, August, and September only the group 2 plastid was detected. The identified plastids from the SAGs were, therefore, only detected in the second half of the year. In the first half of the year, the group 2 Meringosphaera was either aposymbiotic or switched to a different plastid that was not detected by our assays. Either scenario suggests that this is kleptoplastidy and not a permanent plastid. By contrast, the group 1 plastid displayed a similar temporal dynamic with the group 1 host as both were only detected between October and December. However, group 1 hosts and plastids showed spatial separation as they were detected in different sampling locations that differ in their hydrological conditions (Figures S2 and S3). The data suggest that the geographical distribution of the group 1 host and plastid is complicated, with potential migrations into our sampling locations for only 2–3 months of the year. Unfortunately, we do not have sufficient sampling data to form a complete picture of the changing distribution of group 1—future work will need to address this. Interestingly, qPCR data for 2 consecutive years in October and November (2020 and 2021) showed consistency in gene copy number and plastid detection (Figure 2A). The plastid identity at this time of year was also consistent with SAGs from earlier samples in October and November 2018. This level of consistency suggests that the fluctuations in plastid identity could be repeating annually, and therefore seasonal dynamics might play an important role in the Meringosphaera kleptoplastidy (although a longer time series is needed to confirm these preliminary observations). No pattern between basic environmental conditions and Meringosphaera host or plastid gene abundances was found (Figure S3).

Visualization of the Meringosphaera plastids

To test whether the group 2 Meringosphaera was aposymbiotic from January to June or had instead switched plastid, we used CARD-FISH and confocal microscopy to visualize individual Meringosphaera cells. We designed a specific CARD-FISH probe for identifying the Meringosphaera host, which in combination with chlorophyll autofluorescence enabled the visualization of potential photosynthetic plastids. We used samples from the spring bloom of April 2021, a time when our qPCR assay detected high copies of the group 2 Meringosphaera host (e.g., 10⁵–10⁷ gene copies L⁻¹) but both plastid groups were below detection.

The CARD-FISH probe hybridized only to appropriately sized cells with the expected morphology and cell diameters (4–9 μm diameter spherical cells) indicating a successful CARD-FISH procedure. We did not detect the characteristic undulating spines of Meringosphaera, but this was expected given the acidic steps in the CARD-FISH procedure (see STAR Methods). The CARD-FISH images (Figures 2B–2E) clearly show the presence of multiple and intact plastids within a positively hybridized Meringosphaera cell. The chlorophyll a autofluorescence and 18S rRNA probe occupied distinct regions within the Meringosphaera cell, revealing good intracellular specificity (Figures 2B and 2C). The plastids appear located to the edge of the cell, in a very similar arrangement to previous observations of Meringosphaera cells with plastids (e.g., Figures 1A and 1B in Zlatogursky et al.⁴⁴). The z stack images (Video S1) confirmed that the plastids were positioned within the cell and not located either above or below. DAPI only stained the Meringosphaera nucleus (Figure 2D), but no potential nuclei associated with the plastids, supporting the SAG analysis that did not find a second nucleus. The clear presence of plastids within some Meringosphaera cells in April 2021 but a lack of detection by our specific qPCR assay suggests that Meringosphaera might be able to switch plastids.

Variation in plastid integrity and gene repertoire

Plastid contigs were recovered from a total of 11 SAGs, with six mapping into circular genomes. The completeness of the plastid genomes varied for the other SAGs: five possessed incomplete plastid genomes, four of which were fragmented across multiple contigs, and four SAGs lacked plastid sequences (Figure 3; Table S1). The six complete plastid genomes, recovered from both Meringosphaera group 1 and 2, were annotated and compared with the closest available Dictyochophyceae with reference plastid genomes (all free-living). The SAG plastid genome sizes varied according to group identity (Figure 3; Table S1). The group 1 genomes were slightly smaller with a mean size of 83,326 bp, whereas the group 2 genomes had a mean of 89,487 bp, but the number of predicted open reading frames (ORFs) were similar between the two groups varying between 137 and 141. Both the genome size and predicted gene number are lower than those of the reference Dictyochophyceae plastid genomes, which have plastid genomes ranging from 108,152 to 140,025 bp with 144–159 predicted genes.⁴⁹

Figure 3. Variation in the completeness of the plastid genomes from group 1 and group 2 Meringosphaera SAGs.

The plastid genome maps are shown for genomes predicted to be complete and circular. For the incomplete genomes, the number of fragments and size of each fragment is written per SAG. The function of the predicted ORFs in the complete genomes are indicated by their colour as described in the legend.

The functional repertoire of the plastids found in Meringosphaera shows that they retain the capacity to perform photosynthesis (Figures 4 and S4). In particular, they retain the core components of both photosystem I and II (psa- and psb- genes), carbon fixation (rbcL and rbcS), cytochrome b6/f complex (pet- genes), ATP synthase (atp- genes), and chlorophyll biosynthesis (chlI). Compared with the Dictyochophyceae plastids, six genes are lacking in all of the Meringosphaera plastids (Figure 4A). These missing genes span a range of functions: ribosomal proteins (rpl22 and rpl4), protein translocase (secA), iron-sulfur cluster assembly (sufB), and two uncharacterized genes (ycf39 and ycf66). All of these genes have been lost from the plastid genome of other algae, for example, rpl22 has been lost within the Alveolates,⁵⁰ rpl4 and secA lost in Pelagophyceae,⁵¹ sufB lost in Pteridomonas⁵² and transferred to nucleus in green algae,⁵³ ycf39 lost in Eustigmatophytes and Ochromonas,⁵⁴ and ycf66 lost in some diatoms.⁵⁵ In addition, none of the plastids found in Meringosphaera have any of the three amino acid biosynthesis genes ilvB, ilvH, and serC, whereas the known Dictyochophyceae plastids have at least one copy. Furthermore, one gene, orf119, is present in all of the Meringosphaera plastids but not in Dictyochophyceae and encodes a 50s ribosomal L22 like-protein. It is potentially a derivative of the rpl22 gene that is missing in all the Meringosphaera plastids. The ten genes with a shared pattern across the Meringosphaera plastids compared with the free-living Dictyochophyceae (i.e., rpl22, rpl4, secA, sufB, ycf39, ycf66, ilvB, ilvG, serC, and orf119), indicate a level of convergence in the putative Meringosphaera kleptoplasts since plastid groups 1 and 2 are separated by free-living species.

Figure 4. Comparison of the gene repertoires of the Meringosphaera plastid genomes with free-living Dictyochophyceae.

(A) Presence/absence of annotated genes between the six complete SAG plastids and four free-living photosynthetic Dictyochophyceae. Only genes that are not present in all sampled genomes are shown. The full set of plastid genes is available in Figure S4. The background color of the squares highlights the identity of the plastids (pale green, group 1 SAGs; dark green, group 2 SAGs; and pale blue, representative Dictyochophyceae).

(B) Bar plot showing the predicted function of the annotated genes, with color denoting the functional group as described in the legend. The same genomes are compared as in (A), and the color circle below the genome name highlights the identity using the same color scheme as for (A).

The gene content within the two main plastid groups of Meringosphaera also showed patterns of gene specificity (Figures 4A and S4). Group 1 has five genes not found in group 2: acpP, which encodes a cofactor of fatty acid synthesis; the accessory subunit of PSI, psaE; the plastid-encoded member of the TIC20 family, ycf60; and two genes of unknown function (orf271 and ycf19). The presence of the plastid copy of the TIC20 transporter, ycf60, within group 1 is particularly interesting because this could be part of a protein import machinery into the plastid. TIC20 has been found to be capable of forming an independent channel in the chloroplast inner membrane that does not require other TIC proteins to function.⁵⁶ By contrast, group 2 had one specific gene, trnR a tRNA-arginine, that is involved in translation.

Host-encoded plastid-associated genes

To determine whether kleptoplastidy in Meringosphaera shows evidence of host genetic integration, we searched for host-encoded genes of plastid-associated pathways. The 15 individual SAGs were co-assembled based on a 99% 18S rDNA identity threshold to improve genome coverage and to help address the biases of the multiple displacement amplification (MDA) process (Table S1; the STAR Methods detail which SAGs were coassembled together). Both SAGs with and without plastids were co-assembled. This led to the assembly of three COSAGs (COSAG1, COSAG2a, and COSAG2b): COSAG1 belongs to group 1 of the 18S rDNA phylogeny, and COSAG2a and COSAG2b belong to group 2. As expected, the co-assemblies had higher completeness than the individual SAGs (Table S2). We searched for 31 well-characterized nuclear-encoded proteins that function in key plastid pathways even in non-photosynthetic species (e.g., isoprenoid synthesis, iron-sulfur cluster biosynthesis, and protein import).^21,57 In addition, we searched for 35 metabolic transporters that have been previously identified as kleptoplast-targeted.²⁵ For each protein candidate we (1) identified its genomic context to ensure that it was within a Meringosphaera contig and not a contaminant, (2) identified the phylogenetic origin of the candidate protein, and (3) predicted the subcellular localization by searching for a plastid-targeted signal (see STAR Methods for further details). For the metabolic transporters, only candidates with predicted plastid targeting were kept to avoid the inclusion of homologous copies functioning in other cellular compartments.

In total, we found 22 candidates corresponding to Meringosphaera host-encoded plastid-associated proteins (Figure 5; summarized in Table S3). These proteins were found across the three COSAGs and across a diverse range of functions. 14 of the 22 candidates were predicted to have a plastid-targeting signal with a clear N-terminal extension (Figure S5; Table S3). Four homologs to metabolic transporters were predicted, corresponding to three different transporters. Of particular interest is PLT4, a probable sugar transporter predicted in COSAG2a, which could function to translocate fixed carbon into the cytosol. Dinoflagellates and stramenopiles were the main phylogenetic origins of the plastid-targeted proteins (six proteins were predicted as dinoflagellate in origin and nine proteins as stramenopile). Since Dictyochophyceae are stramenopiles, these nine proteins could have originated from EGT, although we could not determine a more precise origin to a specific stramenopile group. It is unclear why six proteins have a dinoflagellate origin, but their genes could have arisen from HGT from other food sources or via a hypothetical second kleptoplast during plastid switching.

Figure 5. Predicted candidates for host-encoded plastid-associated proteins in the co-assemblies.

Each column shows the data for one of the co-assemblies, and each row represents one of the host-encoded plastid-associated proteins, which have been grouped by their function. First, there are the 31 known host-encoded plastid-associated proteins that we searched for, and at the bottom, the three metabolic transporters identified with predicted plastid targeting. The color of the larger box indicates how confident we are in the contig origin; blue shows a confident centrohelid identity, and gray represents a mixed identity that includes centrohelids among others. The small box indicates the predicted targeting signal: dark green indicates predicted plastid targeting, gray indicates predicted targeting to other cellular compartments, and a cross is shown when the N terminus was incomplete and targeting could not be predicted (see Table S3 and Figure S5). The central box shows the phylogeneticorigin of the candidate protein, with colors representing different groups. For the metabolic transporters, only those with predicted plastid targeting were kept. If a plastid copy of the gene was found, a green circle is shown.

Some of these plastid-associated proteins are known to be encoded in either the plastid or nuclear genome in various species, so these need to be considered alongside the plastid genome content. For example, the iron-sulfur cluster assembly genes sufB and sufC are typically encoded on the plastid genome, whereas the remaining genes of the pathway are nuclear-encoded.⁵⁸ The plastid genomes of the SAGs contained sufC but not sufB (Figures 4 and S4), but both COSAG1 and COSAG2b had a nuclear candidate sufB gene. Similarly, although the group 1 plastid genomes encoded the plastid TIC20-homolog (ycf60), the nuclear TIC20 candidate was only found in group 2 (COSAG2a). It remains to be confirmed if these patterns of complementarity hold with more complete data, but both of these examples suggest that host-encoded proteins can substitute missing plastid-encoded functions.

Discussion

Meringosphaera is a poorly studied marine genus that can reach high abundance and has been known for a century to harbor photosynthetic green bodies. Here we identified these commonly observed bodies and investigated the nature of their partnership. Our analyses based on single-cell genomic data and microscopy found no evidence of endosymbiont nuclei, which suggests that the green bodies are isolated plastids and not whole endosymbiotic cells. The analysis of the SAG data revealed two main groups of Meringosphaera harboring different Dictyochophyceae plastids. Our monthly environmental monitoring spanning over a year uncovered seasonal dynamics of both host and plastid. In particular, the group 2 plastid was not detected between January and June, which we hypothesize is due to plastid switching based on the observation by fluorescence of plastids in Meringosphaera cells collected during this time (Figure 2). The lack of plastid detection combined with the lack of plastid markers in four of the SAGs, suggest that the plastids in Meringosphaera are kleptoplasts stolen from Dictyochophyceae prey. Dictyochophyceae are photosynthetic stramenopiles with red algal-derived plastids.⁴⁹ There are only two previous reports of symbiotic Dictyochophyceae: (1) as kleptoplasts found in the dinoflagellate host Dinophysis mitra (also known as Phalacroma mitra),⁵⁹ and (2) in the dinoflagellate Podolampas bipes⁶⁰ where endosymbiotic cells are apparently vertically transmitted to daughter cells. Unfortunately, there are no plastid genomes from either of these examples, so we cannot compare Meringosphaera kleptoplast genomes to other dictyochophyte symbionts.

The presence of kleptoplasts in Meringosphaera is consistent with historical observations that have reported cells both with photosynthetic bodies^38,42,43 but also more rarely without.^61,62 Furthermore, these few available microscopic observations reported contrasting colors of the Meringosphaera photosynthetic bodies, including green^42,63 and golden,⁴³ which is consistent with the presence of a cryptic diversity of hosts that harbor different kinds of kleptoplasts or with plastid switching. Seasonal dynamics within kleptoplastidy similar to the type we report here have been documented before. For example, temporal changes were found in the identity of the green algal kleptoplasts within the sea slug Plakobranchus ocellatus.⁶⁴ The ciliate Mesodinium spp. was also found to switch between red and green-pigmented cryptophyte plastids throughout the year.⁶⁵ However, it has not yet been demonstrated whether plastid switching in kleptoplastidy can actively select the “optimum” plastid for a set of given conditions or whether it is always passive and responds to prey availability. The latter would be akin to the discovery that symbiont biogeography dictates the association in some marine photosymbioses.⁶⁶

The plastid genomes of Meringosphaera varied in both their recovered integrity and gene content. The variation in integrity could be an artifact of the MDA process, which is known to lead to uneven amplification and coverage that can prevent proper assembly.⁶⁷ Alternatively, the variation might reflect different stages of kleptoplastidy as the plastids eventually degrade without the required nuclear support.⁹ This could explain the four SAGs with no plastid sequences, which might have been the oldest cells containing highly degraded plastids with insufficient intact DNA for the MDA process to recover. Among the differences in gene content, the most striking was the presence of the plastid homolog of TIC20 within the group 1 plastids. Intriguingly, a host-encoded TIC20 gene was identified in COSAG2a. Combined, this indicates that two of the three co-assembly groups of Meringosphaera have a possible protein import channel that has been demonstrated to permit some protein import without other subunits.⁵⁶

The predicted host-encoded plastid-associated proteins in Meringosphaera are one of the few examples of putative genetic integration shown in kleptoplastidy.^21,24,25 The candidates for the host-encoded plastid associated-proteins were part of different pathways, but no pathway was completely recovered in our analysis of plastid pathways. This could be indicative of mosaic pathways, where enzymatic steps occur in different cellular compartments. This has, for example, been found previously in the heme synthesis pathway of some dinoflagellates and apicomplexans, where the first step utilized the mitochondrial version of the 5-aminolevulinic acid synthase (ALAS) enzyme but the later steps with the hemB-E enzymes occurred separately in the plastid.⁶⁸ It is possible that some of the pathways within Meringosphaera are similarly mosaic and that some stages occur in other cellular compartments. Alternatively, these pathways might be incomplete, which would contribute to the eventual degradation of the kleptoplasts, or be due to the partial nature of the SAGs.

We do not yet know whether the host-encoded plastid-associated proteins in Meringosphaera are successfully targeted to the kleptoplasts where they function. Nonetheless, finding host-encoded genes putatively associated with kleptoplasts is significant for two main reasons. First, it shows that low levels of EGT and possibly HGT may take place in a fluctuating kleptoplastidy where the kleptoplasts are sourced from different preys. By contrast, Rapaza viridis and the dinoflagellate hosts with kleptoplast-targeted proteins have remarkably stable kleptoplastidy (over 30 months without renewal in RSD), and only one specialized prey species is known.^21,24,25 Secondly, unlike in dinoflagellates but as in R. viridis, centrohelids are not ancestrally plastidbearing, and therefore, kleptoplast-associated gene transfers occurred in a naive genome with no pre-adaptations to plastid hosting. These observations suggest that low-level protein import can occur early-on as a mechanism to regulate plastid retention in kleptoplastidy. The finding of protein import into the kleptoplasts of Meringosphaera is congruent with the predictions of the shopping bag model of plastid origin, where serial uptake of foreign genes from food facilitated the eventual fixation of a plastid.³³ Furthermore, the predicted kleptoplast-targeted metabolic transporters are consistent with the targeting-ratchet model,³⁵ which hypothesizes that transporters set up an evolutionary ratchet for plastid fixation. Overall, this work supports the hypothesis that protein import is a relatively early event that helps to stabilize plastids,^34–36 and as such is a mechanism, rather than a consequence, of plastid establishment.

In conclusion, Meringosphaera offers an exciting opportunity to examine kleptoplastidy within centrohelids and so provides new insights, from a relatively under-examined eukaryotic group, into this important process for the evolution of photosymbiosis. Future work needs to identify the second putative kleptoplast in Meringosphaera, after which it could be used as a model to investigate plastid switching. Moreover, as a globally distributed and at times highly abundant species, the study of Meringosphaera is also potentially important for the marine ecosystems in which it plays a role. This study into the little-known centrohelid Meringosphaera demonstrates the insights gained into important evolutionary transitions by continuing to explore the broad range of eukaryotic diversity.

Star★Methods

Detailed methods are provided in the online version of this paper and include the following:

• KEY RESOURCES TABLE

• RESOURCE AVAILABILITY

  • ○Lead contact
  • ○Materials availability
  • ○Data and code availability

• EXPERIMENTAL MODEL AND SUBJECT DETAILS

  • ○Sampling
  • ○Culturing attempts
  • ○Environmental Dictyochophyceae

• METHOD DETAILS

  • ○Single cell isolations & MDA
  • ○Sequencing
  • ○SAG assembly
  • ○Plastid genome assembly & comparisons
  • ○Co-assemblies & host-encoded plastid associated genes
  • ○CARD-FISH
  • ○Environmental monitoring with qPCR

• QUANTIFICATION AND STATISTICAL ANALYSIS

Star★Methods

Key Resources Table.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Chemicals, peptides, and recombinant proteins
Paraformaldehyde	Serva	Catalog number: 31628
SeaKem LE agarose	Cambrex	Catalog number: 50004
Lysozyme	Sigma Aldrich	Catalog number:12650-88-3
ProLong Diamond Antifade mountant	Thermo Fisher Scientific (Invitrogen)	Catalog number: P36961
Proteinase K	Qiagen	Catalog number: 19133
TaqMan buffer	Thermo Fisher Scientific (Applied Biosystems)	Catalog number: 4370074
Formamide	VWR	Catalog number: 1.09684.1000
Critical commercial assays
REPLI-g UltraFast Mini kit	Qiagen	Catalog number: 150033
ExoProStar 1-Step Kit	GE Healthcare	Catalog number: US77702
DNeasy Plant kit	Qiagen	Catalog number: 69104
Deposited data
Meringosphaera 18S rDNA sequences	GenBank	Accession numbers: OQ075975 to OQ075989
Meringosphaera 28S rDNA sequences	GenBank	Accession numbers: OR195151 to OR195157 & OR196762 to OR196769
Meringosphaera plastid 16S rDNA sequences	GenBank	Accession numbers: OQ091774 to OQ091781
Meringosphaera plastid psbA sequences	GenBank	Accession numbers: OQ078560 to OQ078568
Meringosphaera plastid rbcL sequences	GenBank	Accession numbers: OQ078569 to OQ078579
Meringosphaera complete plastid genomes	GenBank	Accession numbers: OQ161668 to OQ161673
Raw reads	NCBI Sequence Read Archive	under BioProject PRJNA917255, accession numbers SAMN32532880 to SAMN32532894
Data files - all the plastid contigs (both complete and incomplete), the host-encoded plastid-associated protein candidates, single gene trees, the assembled reads of the SAGs and COSAGS, and the qPCR results.	Figshare	https://doi.org/10.6084/m9.figshare.c.6313464
Custom scripts	GitHub	https://github.com/ MeganSorensen/Meringosphaera_SAGs
Oligonucleotides
5’-CATATGCTTGTCTCAAAGATTAAGCCA-3’	Cavalier-Smith and von der Heyden69	Primer name: Thx25F
5’-CACACTTACWAGGAYTTCCTCGTTSAAGACG-3’	Cavalier-Smith and von der Heyden69	Primer name: Helio1979R
Software and algorithms
TrimGalore v0.6.1	NA	www.bioinformatics.babraham.ac.uk/projects/trim_galore
Bbnorm (bbmap v38.08)	Bushnell70	https://github.com/BioInfoTools/BBMap
SPAdes v3.13.1	Bankevich et al.71	https://cab.spbu.ru/files/release3.13.1/manual.html
blast 2.10.1 +	Sayers et al.72	https://blast.ncbi.nlm.nih.gov/Blast.cgi
MAFFT v7.407	Katoh and Standley73	https://mafft.cbrc.jp/alignment/software/
trimAl v1.4.1	Capella-Gutierrez et al.74	https://github.com/inab/trimal
IQ-TREE v1.6.5	Minh et al.75	http://www.iqtree.org/
ModelFinder	Kalyaanamoorthy et al.76	http://www.iqtree.org/ModelFinder/
RAxMLv. 8.2.12	Stamatakis77	https://cme.h-its.org/exelixis/web/software/raxml/
GetOrganelle v1.7.3.3	Jin et al.78	https://github.com/Kinggerm/GetOrganelle
MFannot	NA	https://megasun.bch.umontreal.ca/cgi-bin/mfannot/mfannotInterface.pl
OGDraw	Greiner et al.79	https://chlorobox.mpimp-golm.mpg.de/OGDraw.html
R v.4.2.0	R Core Team80	https://cran.r-project.org/
RStudio	RStudio Team81	https://posit.co/download/rstudio-desktop/
BUSCO v5.3.1	Simao et al.82	https://busco.ezlab.org/busco_userguide.html
Prodigal v2.6.3	Hyatt et al.83	https://github.com/hyattpd/Prodigal
SequenceServer 2.0.0	Priyam et al.84	https://sequenceserver.com/
BWAv0.7.8	Li and Durbin85	https://bio-bwa.sourceforge.net/
IGVv2.4.2	Thorvaldsdottir et al.86	https://software.broadinstitute.org/software/igv/home
DEEPLOC 2.0	Thumuluri et al.87	https://services.healthtech.dtu.dk/services/DeepLoc-2.0/
TargetP 2.0	Almagro Armenteros et al.88	https://services.healthtech.dtu.dk/services/TargetP-2.0/
Other
Alexa fluor 488	Thermo Fisher Scientific	Catalog number: A20000
47 mm diameter nominal pore size (5–15 μm) paper membrane filter	VWR	Catalog number: 516-0813
47mm-diameter 5μm polycarbonate hydrophilic membrane filters	Sigma Aldrich. (Millipore)	Catalog number: TMTP04700

Resource Availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Megan Sørensen (Megan.Sorensen@hhu.de).

Materials availability

This study did not generate new unique reagents.

Experimental Model And Subject Details

Meringosphaera cells were isolated from environmental samples by searching for characteristic Meringosphaera morphology: up to 13 undulating spines typically 16-25 μm long from a spherical body of 4-9 μm in cell diameter containing up to six green/yellow bodies and the cells are non-motile. The sampling strategy is detailed below. Once isolated, the cells were processed immediately and were not maintained in the lab.

Sampling

Samples were collected from surface waters at three stations off the West coast of Sweden: Anholt East (56°40’00” N, 12°07’00” E), Å17 (58°16’30”N, 10°30’48”E) and Släggo (58°15’30” N, 11°26’00” E) routinely sampled by the Swedish Meteorological and Hydrological Institute (SMHI) on the Research Vessel (R/V) Svea. On one occasion, October 2020, the Å17 sampling was replaced by sampling at a nearby station, Å16 (58°16’00” N, 10°43’30” E); which is 12 km away from Å17. Whole water samples were collected from the surface by bucket. The salinity is around 20 practical salinity units (PSU) at Anholt E, 31 PSU at Å17 and 24 PSU at Släggo. Surface water temperature at these locations typically ranges from ~0°C to 20°C over the year. Basic environmental data (temperature, salinity, dissolved nutrient concentrations) was taken from the SMHI website (https://sharkweb.smhi.se/hamta-data) corresponding to the dates and locations of our samplings (Figure S3). The samples were first stored on board the ship in the dark and then transported to the laboratory in an insulated, opaque container with icepacks; the total time between collection and arrival at the laboratory varied between 3 – 12 days.

Culturing attempts

We made several attempts to culture Meringosphaera but none of these were successful. These attempts included the inoculation of 1-10 Meringosphaera cells in 40 mm Petri dishes with 33 ppt artificial seawater or filter-sterilised water from the original habitat with or without additional nutrient media (soil extract, 0.025% cerophyl extract) at 4°C with or without light. Some dishes were separately cultured with Neobodo flagellates or the mixed protist community from original sample was added as potential food. The growth of Meringosphaera was never observed in any of these conditions and alive cells were not found in the original samples after two weeks.

Environmental Dictyochophyceae

The environmental diversity of Dictyochophyceae is an important factor for the Meringosphaera kleptoplastidy. We do not unfortunately have our own data of this, but we have performed searches for environmental Dictyochophyceae using the resources available:

First, regarding the Dictyochophyceae diversity in the environment at sampling sites, we have looked at the phytoplankton data taken by the SMHI monitoring program from the same stations we used (available at https://sharkweb.smhi.se/hamta-data). From this data, 4 Dictyochophyceae species have been identified by microscopy (Apedinella radians, Dictyocha fibula, Dictyocha speculum and Pseudopedinella pyriformis) and there were 3 additional genus-level identifications (Dictyochales, Pseudochattonella and Pseudopedinella). So there appears to be a range of Dictyochophyceae present, with 2 of the 4 known orders being identified. Unfortunately, the available data is abundance only and not sequence data.

Secondly, we have checked the Tara Oceans databases^89,90 using the 16S rDNA plastid sequences as queries. Here there were some hits with 90-98.5% similarity. The upper part of this range is higher than the similarity to the known Dictyochophyceae species. Given that none of Tara stations are close to our sampling locations, a lack of exact matches could be due to environmental variation, and the top matches could represent either the free-living prey or additional Meringosphaera plastids. In addition, there are three environmental sequences that cluster with the Meringosphaera plastids in the 16S rDNA phylogeny, but again we do not know if these represent the free-living prey or not.

Method Details

Single cell isolations & MDA

The samples for single cell isolation were collected in October and November 2018 from the West coast of Sweden (stations Å17 and Anholt E; Table S1 lists which SAGs were sampled from which stations). In the laboratory they were gravity filtered onto a 47 mm diameter nominal pore size (5–15 μm) paper membrane filter (VWR, Radnor, Pennsylvania, USA; Cat No. 516-0813) held in a Millipore filtration tower to avoid damaging the cells. The filters were washed in a 60 mm diameter plastic Petri dish with 10 ml of seawater. The dishes were scanned for characteristic Meringosphaera morphology (up to 13 undulating spines typically 16-25 μm long from a spherical body of 4-9 μm in cell diameter containing up to six green/yellow bodies and the cells are not motile) using a 40X objective of a Nikon Eclipse Ts2R inverted microscope, equipped with phase contrast. Single cells that appeared to contain green/yellow bodies were isolated using an Eppendorf TransferMan 4r micromanipulator and pulled glass pipettes. The cells were passed through droplets of minimal seawater to reduce contamination and frozen in 200μl PCR tubes with minimal seawater (e.g., <5 μl). Frozen single cells in PCR tubes were thawed and subjected to lysis and multiple displacement amplification (MDA) using the REPLI-g UltraFast Mini kit (Qiagen, Hilden, Germany) following the manufacturer’s instructions. The MDA samples were initially screened for the presence of Meringosphaera 18S rDNA with the centrohelid-specific primers Thx25F (5’-CATATGCTTGTCTCAAAGATTAAGCCA-3’) and Helio1979R (5’-CACACTTACWAGGAYTTCCTCGTTSAAGACG-3’) in a PCR reaction.⁶⁹ After gel electrophoresis, if a band was present the PCR products were purified with ExoProStar 1-Step Kit (GE Healthcare; US77702) and Sanger-sequenced directly at Macrogen Europe. From this screening, 15 MDA samples containing Meringosphaera 18S rDNA were selected for the next steps.

Sequencing

The library preparations and sequencing were performed by SciLifeLab National Genomics Infrastructure (NGI), Stockholm, Sweden. Sequencing libraries were prepared with the TruSeq PCR-free library preparation, targeting an insert size of 350bp. Sequencing was performed in two batches: Batch 1 (became SAGs S1-4) were multiplexed on 1 lane and sequenced on Illumina NextSeq500 with paired-end ‘Mid-output’ chemistry; batch 2 (became SAGs S5-15) were multiplexed on 2 lanes and sequenced on Illumina NovaSeq6000 with paired-end ’NovaSeqStandard’ workflow in ’SP’ mode flowcell.

SAG assembly

The 15 datasets were trimmed using TrimGalore v0.6.1 (https://www.bioinformatics.babraham.ac.uk/projects/trim_galore/) with default parameters. Then normalised with bbnorm (bbmap v38.08⁷⁰) with a minimum coverage value of 5 and a target value of 100 to help account for the biases introduced by MDA. The normalised reads were assembled into contigs with SPAdes (v3.13.1⁷¹) in careful mode (spades.py –careful -k auto). Basic parameters of the assemblies are listed in Table S2.

The contigs were made searchable as local databases with makeblastdb (blast 2.10.1+⁷²) and searched with blastn. For the specific Meringosphaera 18S rDNA search GenBank accession MZ240752 was used as the query. For the general 18S rDNA search, a custom dataset containing diverse 18S rDNA sequences was used as the query (Figshare dataset D1.1). In the general search there were additional green algae picoeukaryote 18S rDNA identified in 2 of the SAGs (S5 and S7), but due to the lack of consistency, both across the samples and with the plastid identity, we believe this is contamination. For S7, we believe the degree of contamination was sufficient to increase the BUSCO completeness score (see Table S2).

We chose to use the 16S rDNA, psbA & rbcL genes as the plastid markers because they maximised the number of the SAGs with a plastid copy and are commonly sampled genes, meaning that environmental references could be included. For these plastid markers a custom dataset was used for each containing diverse plastid sequences (16S rDNA Figshare datasets D1.2, psbA D1.3, rbcL D1.4). The hits from the blast searches were aligned with MAFFT v7.407 (mafft -auto -adjustdirectionaccurately -reorder)⁷³ and trimmed with trimAl v1.4.1⁷⁴ (trimal -gappyout -fasta). For the most part, phylogenetic trees were made with IQ-TREE v1.6.5⁷⁵ with ModelFinder⁷⁶ to determine the best-fit model (for the 18S rDNA tree the TN+F+I+G model was chosen, for the plastid 16S rDNA the TIM3+F+I+G4 model, for psbA the GTR+F+G4 model and for rbcL the GTR+F+I+G4 model). Support values are from 1000 ultrafast bootstrap replicates and SH-aLRT test (iqtree -m TEST -bb 1000 -alrt 1000). However, the phylogenetic trees of the concatenated 18S and 28S rDNA (Figure 1A) sequences and the separate 28S rDNA phylogeny (Figshare D2.1B) were reconstructed with RAxML v. 8.2.12⁷⁷ GTR models with 4 gamma categories and support values are from 1000 rapid bootstrap replicates.

The length of the Meringosphaera sequences used in the phylogenetic analyses are as follows: The Meringosphaera plastid 16S rDNA sequences (Figure 1B) were 1470-1480 bp in length. The Meringosphaera 18S rDNA sequences (Figshare D2.1A) were 1710-1720 bp in length. The complete Meringosphaera psbA sequences (Figure S1A) were 1059-1082 bp in length, but SAG S12 was incomplete and only 443 bp in length. The complete Meringosphaera rbcL sequences (Figure S1B) were 1458-1466 bp in length, but SAG S7 and S15 were incomplete and only 553 bp and 844 bp in length respectively. The Meringosphaera 28S rDNA sequences (Figshare D2.1B) were 2292 bp in length. The concatenated 18S and 28S rDNA sequences (for Figure 1A) were 3850 bp in length, and in order to get sufficient number of reference centrohelid 18S and 28S sequences from the same sample we used the long read OTU sequences generated by Jamy et al.⁹¹

Plastid genome assembly & comparisons

The plastid genomes were predicted and assembled with GetOrganelle v1.7.3.3⁷⁸ with the other_pt database (get_organelle_from_- reads.py -F other_pt -R 15 -k 21,45,65,85,105). In addition, GetOrganelle was performed with the seed option (-s) using the identified SAG S4 plastid genome to assist with plastid identification in the other SAGs. Six plastid genomes were assembled as circular by GetOrganelle and are predicted as ‘complete’ (S2, S3, S4, S6, S8, and S14). We did not verify them with PCR. Plastid contigs were annotated with MFannot (https://megasun.bch.umontreal.ca/cgi-bin/mfannot/mfannotInterface.pl) and were drawn with OG-Draw.⁷⁹ The MFannot predictions of absent plastid genes within Meringosphaera plastids were confirmed by manual verification using Blast. For the plastid genome comparisons, the annotated SAG plastid genomes were compared to the four published photosynthetic Dictyochophyceae plastid genomes available at the time (GenBank accession numbers: MK518352, MK518353, MK561359 & MK561360). The analysis was conducted and visualised with R v.4.2.0⁸⁰ in RStudio.⁸¹

Co-assemblies & host-encoded plastid associated genes

The MDA process is known to produce uneven coverage that can lead to errors in assembly, meaning that gene absences at the SAG level could be either biological or an artefact. To help address this, and to increase the coverage overall, we formed co-assemblies for the investigations of the host genes. First, the plastid sequences were removed from the SAGS, and the remaining reads were clustered based on >99% Meringosphaera 18S rDNA. Within these clusters the reads were co-assembled in the same way as described above for the single assemblies. This created three co-assemblies: COSAG1 (from SAG S2,3 & 6), COSAG2a (SAG S10,13), & COSAG2b (SAG S1,4,5,7,8,9,11,12,14,15). In this way, the co-assemblies included SAGs both with and without plastids. The completeness of the SAGs and COSAGS was assessed with BUSCO v5.3.1⁸² in genome mode with the eukaryote database of 255 markers (Table S2). Basic parameters of the co-assemblies are listed in Table S2.

Open reading frames (ORFs) and amino acid sequences were predicted with Prodigal v2.6.3⁸³ using default parameters. The amino acid sequences were made searchable as local blast databases (as above). These were searched for the 31 target proteins using reference databases across diverse taxa as the queries (databases from Schön et al.⁵⁷: Figshare Collection. https://doi.org/10.6084/m9.figshare.c.5388176.v3, list of proteins also based on Hehenberger et al.²¹). In addition, we also searched for 35 metabolic transporters found to be kleptoplast-targeted in the euglenozoan Rapaza viridis.²⁵The resulting homologs were aligned with the reference database and IQ trees were made according to the method described above. The trees were manually inspected to identify any homologs that grouped with known plastid-bearing species. The identity of the up- and downstream proteins surrounding these candidates were then assessed. The surrounding proteins were run through three different databases: Blastp tsa_nr mode (Transcriptome Shotgun Assembly proteins), Blastp nr mode (non-redundant protein sequences), and SequenceServer 2.0.0⁸⁴ with the EukProt V3 database⁹² selected taxonomic groups Centroplashelida, Haptophyta and Dictyochophyceae. The top 20 hits were taken from each of these searches and IQ trees were made of the alignments; from these trees the identity of the contigs was established. Only candidates whose up- and downstream proteins had possible centrohelid identity were kept. Coverage was checked manually across the candidate contigs to check for mis-assemblies, using BWA v0.7.8⁸⁵ and visualised with IGV v2.4.2.⁸⁶ Each candidate was manually checked for completeness, using alignments to a variety of homologs, and particular care was taken with the N and C termini. The completeness of each candidate is noted in Table S3, along with a general summary of each candidate. Only if the N-terminus was complete did we predict the subcellular localisation with DEEPLOC 2.0⁸⁷ and TargetP 2.0.⁸⁸ For the transporters, only candidates with predicted plastid-targeting were kept because often homologous transporters function in different cellular compartments. Candidate proteins that had been split by Prodigal due to the presence of introns were concatenated together manually and when this was necessary it is noted in Table S3.

In addition, we searched for the proteins encoded by genes identified as plastid-to-nucleus transferred within a Dictyochophyceae species.⁴⁹ Homologs of these nine proteins (acpP, ilvB, petF, psb28, rpl12, rpl32, syfB, ycf35 and ycf42) were not found in any of the Meringosphaera co-assemblies. However, some of these were still within the plastid genomes: in particular, acpP was found in the group 1 plastids, and rpl12 and rpl32 were found in all of the SAG plastids (see Figures 4A and S4).

Card-Fish

Water samples (0.5L) were filtered on 5μm 47mm-diameter polycarbonate hydrophilic membrane filters (Whatman) by gravity filtration. The filters were fixed in the dark at room temperature in 2ml of 4% paraformaldehyde for 1 hour. The filters were then rinsed three times with 0.2μm-filtered sea water and stored frozen at -20°C.

The CARD-FISH procedure was performed on these filters at a later date (a few months after collection). First, the filters were embedded in 0.2% low melting point agarose. Next, an initial permeabilization step of incubation in 10mg ml^-1 lysozyme solution for 1 hour at 35°C, followed by a MilliQ wash. The rest of the procedure followed the protocol provided by Piwosz et al.⁹³ Briefly, the filters were then moved to 0.01M HCL solution for a 20-minute incubation at room temperature, before being washed in PBS and DI water. Hybridisation occurred at 35°C for ~20 hours, for which the 50ng/μl probe solution was mixed with the appropriate hybridisation buffer (see Piwosz et al.⁹³) in a 1:99 dilution, producing a final probe concentration of 0.5ng/μl. The formamide percentage in the hybridisation buffer was optimised per probe, and the final percentage used is given in Table S4. The filters were then washed at 37°C for 30 minutes, and the concentration of NaCl in the wash buffer was determined by the formamide percentage of the hybridisation buffer. The filters were incubated in 0.01% PBS-Triton for 45 minutes at 37°C. They were then incubated at 37°C in the dark with the amplification buffer and fluorochrome-labelled tyramide solution for the CARD step. Alexa fluor 488 was used in this process. The filters were then incubated in 0.01% PBS-Triton for 15 minutes in the dark at 37°C. The final wash occurred in MilliQ, and lastly ethanol (100%). The filters were then air-dried. The filters were mounted onto slides with 1μg ml^-1 DAPI in ProLong™ Diamond Antifade mountant (Invitrogen™). The slides were dried for 24 hours in the dark at room temperature and stored frozen at -20°C until visualization with a confocal microscope.

The CARD-FISH probes used in this experiment are listed in Table S4. One of which was designed for this project, Mer482, and it targets both group 1 and 2 Meringosphaera 18S rRNA (though within the April 2021 CARD-FISH samples only group 2 hosts are present). Mer482 probe was designed with the help of oligoN-design v0.1.0 pipeline⁹⁴ and then was checked for specificity with the Silva 18S rRNA Arb database.⁹⁵ Probe function was predicted by MathFISH,^96,97and the predicted formamide concentration was used as the starting point.

Each sample filter was cut into 16 equal slices (of ~1.1cm²) in order that the controls could be applied to the same sample. In this way, every CARD-FISH experiment included both a negative and positive control. In the negative control the Mer482 specific HRP probe for Meringosphaera was absent but the Alexa 488 dye was added to check for non-specific binding. In the positive control a general eukaryote probe EUK1209R (EUK1195) was used instead of the Meringosphaera probe so to ensure that the CARD-FISH procedure worked optimally.

Images were taken on inverted confocal microscope Zeiss LSM 780 (Zeiss Berlin Germany) using the 405, 488 and 633nm lasers with the 63×/1.4 oil objective lens. The ZEN black 2011 software was used to control the system and process the images. The same gain and intensity per channel were used when comparing between all the sample conditions and across the controls.

Environmental monitoring with qPCR

Surface water samples (0.5-1L) were filtered onto a 47mm diameter 5μm filter (Whatman) by gravity filtration or onto a 25mm diameter 5μm filter (Whatman) held in 25-mm diameter Swinnex filter holder (Millipore Billerca MA USA) using a peristaltic pump (Cole-Parmer, Vernon Hills, IL USA). We aimed to filter 1L of water per replicate and to have 3 replicates per sample location when possible. There were some differences in which of the three locations were sampled per month, this information can be seen in Figures S2 that splits the qPCR data by sampling location. The filters were snap frozen in liquid nitrogen and stored at -80°C. DNA was extracted from the filters with the DNeasy Plant kit (Qiagen, Hilden, Germany) with the following modifications to the manufacturer’s protocol: Tubes containing the 400μl API buffer and the filter were subject to three freeze/thaw cycles (alternating from submerged in liquid nitrogen to 65°C). Glass beads (0.65g of a 50:50 mix of 2.8mm and 1.4mm Zirconium oxide beads (Precellys)) were added to the tubes, and then these were subjected to a 2-minute bead beat step (at 8500 rpm). 45μl Proteinase K was added and the samples were incubated for 1 hour at 55°C. The rest of the procedure followed the Qiagen protocol starting with the addition of RNase A. The final elution volume was 30μl. Extracted DNA was stored at -20°C.

qPCR was performed using the TaqMan chemistry (Applied Biosystems) and using a Step One Plus Real-Time PCR system (Applied Biosystems). The qPCR program was: 50°C for 2 min, 95°C for 10 min and 45 cycles of 95°C for 15 secs followed by the Tm temperature for 1 min (the Tm temperatures per probe are listed in Table S4). The reaction volume was 25μl per well, containing: 12.5μl 2X TaqMan buffer (Applied Biosystems), 8μl nuclease-free water, 0.4μmol L^-1 forward primer, 0.4μmol L^-1 reverse primer, 0.2μmol L^-1 probe & 2μl of template DNA). The target-specific TaqMAN probe was 5’ labelled with a fluorescent reporter FAM (6-carboxyfluoesceom) or VIC (2′-chloro-7′phenyl-1,4-dichloro-6-carboxy-fluorescein) (see Table S4 for details per probe) and 3’labelled with TAMRA (6-caroboxytetramethylrhomadine) as a quenching dye. On every plate, standards for an 8-point standard curve were included and were run in duplicates. The standard curve was made from a 10-fold dilution series ranging from 10⁸ to 10¹ gene copies per reaction using synthesized Gblocks (IDT) of the target region. Every plate also included 3 wells for the negative controls, where nuclease-free water was used instead of template DNA. Each sample was run with 2 or 3 technical replicates. If the target could not be detected in all of the technical replicates the sample was marked as detected but not quantifiable and were plotted with a value of 1.1. Gene copy numbers were calculated from the mean cycle threshold (Ct) value of the technical replicates, this value was then quantified into copy number using the standard curve.

The qPCR primers and probes in this study were designed using the Primer Express (Applied Biosystems) software, and their specificity was tested against the Silva database, NCBI, and with the alignments created from the sequencing results. Cross-reactivity between the primers/probes of the different groups were tested, e.g., the 18S group 1 was tested on group 2 standards, and no amplification occurred. The details of the primers and probes are given in Table S4.

Quantification And Statistical Analysis

The statistical analysis was conducted with R v.4.2.0⁸⁰ in RStudio.⁸¹ Within the qPCR results, the abundance of Meringosphaera group 2 was analysed by ANOVA with month as a factor (N = 57). The data was tested for normality by creating Q-Q and residual vs fitted values plots in R. When error bars are shown in a figure, the type of error used is noted in the legend.

Highlights.

• Meringosphaera protists engage in kleptoplastidic behavior

• The kleptoplasts are derived from dictyochophyte algae

• Seasonal dynamics of hosts and plastids differ, suggesting plastid switching

• Host nuclear genomes encode possible kleptoplast-targeted genes

Data and code availability

The identified genetic sequences have been deposited at GenBank and are publicly available as of the date of publication (Meringosphaera 18S rDNA sequences GenBank:OQ075975 to OQ075989, Meringosphaera 28S rDNA sequences GenBank: OR195151 to OR195157 and OR196762 to OR196769, Meringosphaera plastid 16S rDNA sequences GenBank: OQ091774 to OQ091781, plastid psbA sequences GenBank: OQ078560 to OQ078568, plastid rbcL sequences GenBank: OQ078569 to OQ078579, and the complete plastid genomes GenBank: OQ161668 to OQ161673). The raw reads data have been deposited at NCBI Sequence Read Archive: BioProject PRJNA917255, accession numbers: SAMN32532880 to SAMN32532894. All data files are available at Figshare https://doi.org/10.6084/m9.figshare.c.6313464, this includes all the plastid contigs (both complete and incomplete), the host-encoded plastid-associated protein candidates, single gene trees, the assembled reads of the SAGs and COSAGS, and the qPCR results.

All custom scripts used in this study are publicly available at GitHub: https://github.com/MeganSorensen/Meringosphaera_SAGs.

Any additional information required to reanalyse the data reported in this paper is available from the lead contact upon request.