Chlamydiae as symbionts of photosynthetic dinoflagellates

Abstract Chlamydiae are ubiquitous intracellular bacteria and infect a wide diversity of eukaryotes, including mammals. However, chlamydiae have never been reported to infect photosynthetic organisms. Here, we describe a novel chlamydial genus and species, Candidatus Algichlamydia australiensis, capable of infecting the photosynthetic dinoflagellate Cladocopium sp. (originally isolated from a scleractinian coral). Algichlamydia australiensis was confirmed to be intracellular by fluorescence in situ hybridization and confocal laser scanning microscopy and temporally stable at the population level by monitoring its relative abundance across four weeks of host growth. Using a combination of short- and long-read sequencing, we recovered a high-quality (completeness 91.73% and contamination 0.27%) metagenome-assembled genome of A. australiensis. Phylogenetic analyses show that this chlamydial taxon represents a new genus and species within the Simkaniaceae family. Algichlamydia australiensis possesses all the hallmark genes for chlamydiae–host interactions, including a complete type III secretion system. In addition, a type IV secretion system is encoded on a plasmid and has previously been observed for only three other chlamydial species. Twenty orthologous groups of genes are unique to A. australiensis, one of which is structurally similar to a protein known from Cyanobacteria and Archaeplastida involved in thylakoid biogenesis and maintenance, hinting at potential chlamydiae interactions with the chloroplasts of Cladocopium cells. Our study shows that chlamydiae infect dinoflagellate symbionts of cnidarians, the first photosynthetic organism reported to harbor chlamydiae, thereby expanding the breadth of chlamydial hosts and providing a new contribution to the discussion around the role of chlamydiae in the establishment of the primary plastid.


Introduction
The Chlamydiota (also known as chlamydiae) is a phylum of obligate intracellular bacteria infecting eukaryotes [1,2].Despite their diversity, all known chlamydiae have a remarkably conserved biology; they are dependent on eukaryotic host cells for growth and survival, alternate between infectious extracellular elementary bodies and intracellular replicative reticulate bodies, and manipulate host cells through a type III secretion system (T3SS) [1].Because of their intracellular lifestyle, all chlamydiae also possess reduced genomes (1-3 Mb) [3], and are not culturable ex hospite, making them notoriously difficult to study.Although chlamydiae are best known for infecting mammals, their host range is incredibly diverse and includes arthropods, amphibians, sponges, corals, and protists [2].Thus far, however, there has been no record of photosynthetic organisms harboring chlamydiae.Intriguingly, genomic data suggest that numerous horizontal gene transfer events between chlamydiae and Archaeplastida (i.e.red algae, glaucophytes, and plants) have occurred, which were speculated to be remnants of ancestral interactions [4].This led to the hypothesis that chlamydiae facilitated the establishment of ancestral Cyanobacteria as plastids [4,5], though phylogenetic evidence for this hypothesis remains controversial [6].
Chlamydiae sequences were recently detected in 16S rRNA gene amplicon sequencing of several Symbiodiniaceae laboratory cultures [7,10,15].In a Cladocopium sp.culture (SCF049.01-seeSupplementary text and Fig. S1 regarding its taxonomy), a single chlamydial amplicon sequence variant (ASV) made up >60% of the associated bacterial communities [7], but the nature of the association or whether Cladocopium is the true chlamydial host was not investigated.By combining amplicon sequencing, f luorescence microscopy, and genome sequencing and analysis, we prove that Cladocopium harbors chlamydial cells and provide a detailed description of an association between a chlamydial representative and a photosynthetic organism.

Symbiodiniaceae culture and maintenance
The Symbiodiniaceae culture SCF049.01 was used in this study (see Supplementary text for taxonomic considerations).It was initially isolated at the Australian Institute of Marine Science from the coral Pocillopora damicornis collected from Davies Reef (central Great Barrier Reef, Australia).Symbiodiniaceae cultures were maintained in 15 ml Daigo's IMK medium (1×), prepared with filtered red sea salt water (fRSSW, 34 ppt salinity) in sterile 50 ml polypropylene culture f lasks where media was changed fortnightly.These f lasks were kept in a 12 h light:12 h dark incubator (50-60 μmol photons m −2 s −1 of photosynthetically active radiation) at 26 • C.

Fluorescence in situ hybridization on Symbiodiniaceae cells
Fluorescence in situ hybridization (FISH) was performed on Symbiodiniaceae cells as previously described [7] using the 16S rRNAtargeting, chlamydiae-specific probe Chls523 along with a competitor probe [20].The antisense nonEUB probe [21] was also used as a negative control.Samples were observed on a Nikon Air Confocal Laser Scanning Microscope (CLSM).Additional details are available in the supplementary data.

Transmission electron microscopy
For Symbiodiniaceae cell preparation for transmission electron microscopy (TEM), high-pressure freezing, freeze substitution, epoxy resin infiltration, sectioning, and observation were performed as previously described [22], with adaptations detailed in the supplementary data.

Determination of chlamydial copy numbers by digital polymerase chain reaction (dPCR)
To determine chlamydial cell numbers, we performed digital dPCR on the supernatant and cellular fraction of six culture replicates.DNA was extracted from all samples using the DNeasy PowerSoil Pro Kit (Qiagen) according to the manufacturer's instructions.Digital PCR was performed using the QIAcuity EvaGreen PCR kit (Qiagen) and the chlamydiae-specific primer pair Chl40F and Chl523R targeting the 16S rRNA gene on a QIAcuity One digital PCR device (Qiagen).The determined 16S rRNA gene copy numbers per milliliter correspond to chlamydial cell numbers per milliliter.In the cellular fraction, the chlamydial cell numbers per milliliter were normalized to the Cladocopium sp.cell counts.Additional details are available in the supplementary data.
For each sampling day/time, cell density was obtained using a Countess II FL cell counter (Thermo Fisher scientific).Before each cell count, f lasks were vigorously shaken to detach cells from the base of the f lask and minimize any bias.Samples (100 μl) were taken from each of the three replicate f lasks.Of this, two 10 μl samples were processed to obtain cell density for each f lask.Cultures were then sampled as previously described [7] in order to separate "loosely associated," "closely associated," and "intracellular" bacteria from Cladocopium sp SCF049.01 cultures.Brief ly, six replicates of 100 000 cells for each f lask were filtered through a 5-μm strainer (pluriSelect, Germany), which retains Cladocopium cells (>5 μm), but not planktonic bacteria (<5 μm).Three replicates were washed with fRSSW to wash away bacteria that are not tightly attached to the cell surface.Filtrates represent the "loosely associated bacteria."Filters were detached from the strainers and represent the "closely associated bacteria," i.e. intracellular bacteria and those tightly attached to the surface.The other three replicates were washed with 6% sodium hypochlorite (v/v) to wash away all extracellular bacteria.Filters were detached from the strainers, and these represent the "intracellular bacteria."Six filters (three for the growth phase experiment, three for the time series experiment) that only received fRSSW and six filters (three for the growth phase experiment, three for the time series experiment) that only received sodium hypochlorite (no algae) were also sampled as negative controls.All samples were snap-frozen and kept at −20 • C until processing for 16S rRNA gene amplicon sequencing.
Additional samples were taken during the growth phase experiment for FISH and f low cytometry quantification of Cladocopium cells infected by chlamydiae, at Days 1, 8, 15, 22, and 29.Approximately 3 × 10 6 Cladocopium cells from each f lask were sampled for each time point and fixed in 80% ethanol as previously described [7].Because of the repeated centrifugations/washes during the FISH protocol, these samples are considered as the closely associated fraction.FISH and f low cytometry analyses were conducted as previously described [7], and additional details are available in the supplementary data.

DNA extractions for amplicon sequencing
DNA extractions were performed using a salting-out method with modifications as previously described [23].Extraction blanks (two to four) were included to account for potential contaminants introduced during the extraction process.

Amplicon sequencing data analyses
Symbiodiniaceae amplicon sequencing data (ITS2) were processed using the SymPortal analytical framework (symportal.org)[28].Sequence information was submitted to the SymPortal remote database and underwent quality control including the removal of artefact and non-Symbiodiniaceae sequences.Relative abundances of ITS2 types were exported and plotted on GraphPad Prism 9.
18S rRNA gene amplicon sequencing data were processed using QIIME2 version 2021.8 [29].Amplicon sequencing data were obtained as paired-end, demultiplexed files with primers and adapters attached.The cutadapt plugin [30] was used to remove primer and adapter sequences, with an error rate of 0.2.The quality of trimmed sequences was determined using the DADA2 plugin [31], which denoises, filters, dereplicates, detects chimeras, and merges paired-end reads.Reads with low quality (Q-score < 30) were removed.Taxonomy was assigned by training a naive Bayes classifier with the feature-classifier plugin [29], based on a 99% similarity to the 18S rRNA gene in the PR 2 database v4.14.1 to match the primer pair used [32].Metadata file, phylogenetic tree, and tables with amplicon sequence variant (ASV) taxonomic classifications and counts were exported, and relative abundances were plotted using GraphPad Prism 9.
16S rRNA gene amplicon sequencing data were processed using QIIME2 version 2021.8 [29].Amplicon sequencing data were obtained as paired-end, demultiplexed files with primers and adapters attached.The cutadapt plugin [30] was used to remove primer and adapter sequences, with an error rate of 0.2.The quality of trimmed sequences was determined using the DADA2 plugin [31], which denoises, filters, dereplicates, detects chimeras, and merges paired-end reads.Reads with low quality (Q-score < 30) were removed.Taxonomy was assigned by training a naive Bayes classifier with the feature-classifier plugin [29], based on a 99% similarity to the V5-V6 region of the 16S rRNA gene in the SILVA 138 database to match the 784F/1061R primer pair used [33].Mitochondria and chloroplast reads were filtered out.The metadata file, phylogenetic tree, and ASV tables were imported into Rstudio for analyses, using the phyloseq package [34].At this stage of the analysis, datasets from the growth phase experiment and from the time series experiment were separated and analysed independently.Rare ASVs (percentage abundance lower than 1 × 10 −5 ) were removed from the dataset.Samples with low read numbers were removed (four for the growth phase experiment: Day 5-Intracellular-Flask C, Day 15-Intracellular-Flask B, Day 22-Intracellular-Flask A, Day 26-Intracellular-Flask B; three for the time series experiment: 12:00-Loosely associated-Flask C, 0:00-Loosely associated-Flask A, 6:00-Loosely associated-Flask B).Sequencing statistics for all amplicon sequencing experiments can be found in Table S1.Contaminant ASVs, arising from kit reagents and sample manipulation, were identified manually based on their abundance in negative controls: any ASV that was five times more abundant in the mean abundance of filter blanks, extraction blanks, or no template PCRs compared to the mean of all Cladocopium sp SCF049.01 samples, and that represented at least 500 reads in all Cladocopium sp SCF049.01 samples, was considered a contaminant and removed from the dataset.Known contaminants (e.g.Cutibacterium) were also removed manually.For the growth phase experiments, 299 ASVs were identified as contaminants, accounting for 5.99% of Cladocopium sp SCF049.01 reads (Table S2A).For the time series experiment, 33 contaminants were identified, accounting for 9.88% of Cladocopium sp SCF049.01 reads (Table S2B).ASVs assigned to the chlamydiae were specifically targeted and plotted separately.

Statistical analysis
Chlamydial abundance and the proportion of infected Cladocopium cells in the growth phase and time series experiments were analyzed and plotted using GraphPad Prism 9.Each experiment was analyzed independently.Within each experiment, the three different fractions (closely associated, intracellular, loosely associated) were analyzed independently.For each fraction, the effect of sampling time (day in the growth phase experiment, hour in the time series experiment) on chlamydial abundance and the proportion of infected Cladocopium cells was analyzed by performing a nonparametric Kruskal-Wallis test.Statistical tests were considered significant at α = 0.05, unless otherwise stated.

Sample preparation and DNA extraction for long-read sequencing of Cladocopium microbial communities
For long-read sequencing, the supernatants of SCF049.01 cultures were sampled to minimize host DNA quantities.Two weeks after subculturing (Cladocopium density was around ∼1 × 10 6 cells/ml), 150 ml of supernatant were carefully removed, without disturbing the attached Cladocopium cells, to minimize Cladocopium contamination.The supernatant was subsequently filtered at 5 and 1.2 μm to remove Cladocopium cells and >1.2 μm-sized bacteria (chlamydiae are <1 μm).The filtered supernatant was centrifuged for 15 min at 3000 × g to pellet the bacteria.The supernatant was discarded and the bacterial pellet kept at −20 • C. The bacterial pellet was lysed (lysozyme 100 mg/ml) and DNA was extracted with GenFind V3, according to manufacturer's instructions for bacterial samples (Beckman Coulter).A ligation sequencing library was prepared (ONT SQK-NBD114-96) and the resultant library run on a MinION f low cell (FLO-MIN114) using a GridION device.Data were basecalled with Super-accurate basecalling in MinKNOW v23.07.05.

Sample preparation and DNA extraction for short-read sequencing of Cladocopium microbial communities
For short-read sequencing, we employed a more comprehensive strategy to eliminate Cladocopium cells and DNA and maximize bacterial DNA yields.First, the "closely-associated" communities of around 5 × 10 6 Cladocopium cells were sampled as described above.Samples were then bead-beaten for 20 min at 30 Hz with 100 mg of sterile beads (400-600 nm) to open up the Cladocopium cells and stained with SYBR Green as previously described [12,15].Bacteria were then separated from Cladocopium cells and debris by f luorescence-activated cell sorting on an Aria III/FACS DiVa 9 software (BD Biosciences, Franklin Lakes, NJ) equipped with a 70 μm nozzle and run at 70 psi, as previously described [15].Selection was based on size and high SYBR Green f luorescence.A total of 3.75 million events were obtained.
DNA was extracted using a HostZERO Microbial DNA Kit (Zymo) according to the manufacturer's instructions.The final elution volume was 40 μl.A volume of 20 μl of purified DNA was concentrated using a DNA Concentrator Kit (abcam) according to the manufacturer's instruction, with a final elution volume of 3 μl.The resulting DNA was amplified by multiple displacement amplification (MDA) using a REPLI-g Single Cell Kit (QIAGEN) following the manufacturer's instructions.Following MDA at 30 • C for 8 h, the DNA polymerase was inactivated, and amplified DNA was stored at −20 • C. The sample was sequenced across two lanes of a NovaSeq 6000 SP 2 × 150 bp f lowcell (Illumina, San Diego, CA) at the Ramaciotti Centre for Genomics (UNSW Sydney, Australia).
Two contigs from the final bin were predicted to be circular.The content of one of the two contigs was not sufficient to function as a plasmid entity, and most chlamydiae only have a single plasmid [44], so we attempted to improve the contiguity of the circular contigs and obtain a single plasmid.Short reads were mapped to the two circular contigs using bowtie2 v2.4.2 [41] and Samtools v1.11 [42], and long reads were mapped to the two circular contigs using minimap2 v2.26 [45] and Samtools v1.11 [42].A hybrid assembly was conducted with the mapped short and long reads and Unicycler v0.5.0 [46], yielding a single plasmid.

Phylogenetic analyses
For comparative and phylogenetic analyses, a dataset of highquality chlamydial genomes based on Dharamshi et al. [48] was used.The chlamydial dataset was complemented by additional genomes available on GenBank/ENA/DDBJ on 29 May 2023.In general, only genomes with a completeness >70%, contamination <5% (both determined with CheckM2 v1.0.2 [47]), and average nucleotide identities <95% (determined with FastANI v1.33 [34]) were considered for the final dataset, which included 170 chlamydial genomes and 89 genomes from Planctomycetota and Verrucomicrobiota serving as outgroup (Table S3).To obtain the phylogenetic affiliation of Cla049, a set of 15 concatenated conserved nonsupervised orthologous groups (NOGs) was used (Table S4).These 15 NOGs are known to retrieve the same topology for chlamydial phylogeny as the application of larger protein sets and is now widely used for chlamydial phylogenies [48].We remain careful with our conclusions regarding this analysis because of the relatively low number of genomes available.Proteins of the chlamydial and Planctomycetota-Verrucomicrobiota outgroup genomes belonging to the 15 NOGs were aligned with MAFFT v7.520 L-INS-i [49].The resulting single protein alignments were subsequently trimmed with BMGE v2.0 [50] and concatenated.One chlamydial MAG with low completeness was excluded from phylogenetic analysis (Table S3).Maximum likelihood phylogeny was inferred with IQ-TREE v2.2.5 [51] using ModelFinder Plus (m MFP -mset LG,LG+C20 -mfreq " "mrate G4,R4) [52] with 1000 ultrafast bootstrap replicates [53] and 1000 replicates of the SH-like approximate likelihood ratio test [54] under the LG + C20 + R4 model.The resulting phylogenetic tree was used as guide tree for maximum likelihood tree inference under the posterior site mean frequency (PMSF) model with 100 nonparametric bootstraps [55].Phylogenetic trees were rooted using the outgroup and visualized with iTOL v6.8.1 [56].

Analysis of orthologous groups of proteins
For further comparative analysis of the chlamydial genomes, all encoded protein sequences were clustered into orthologous groups (OGs) with OrthoFinder v2.5.5 [66] under default parameters.OGs and their respective eggNOG annotations were merged in RStudio v2023.06.2 and analyzed.OGs present in the bin obtained in this study, but not present in any other chlamydial genome, were selected (this included all genes not classified in any OGs).Because most sequences did not show any similarity to known proteins in public databases, the structures of representative sequences of these OGs were predicted with AlphaFold v2.3.2 [67] and visualized with UCSF ChimeraX v1.6.1 [68].The resulting best-ranked protein structure models were then searched with structure search against RCSB protein data bank (PDB) [69].Only hits with global pLDDT >70 were considered in the results.

Chlamydiae reside inside Cladocopium cells
To verify that the previously detected chlamydiae infect Cladocopium cells, rather than other protists potentially present in the culture, we conducted 18S rRNA gene amplicon sequencing.All reads from three replicate culture f lasks were assigned to Symbiodiniaceae (Fig. S3), strongly suggesting that there were no other eukaryotes and thus, no other chlamydial hosts, in the culture.Additionally, FISH and confocal laser scanning microscopy (CLSM) using a chlamydiae-specific probe clearly showed whole bacterial cells f luorescing inside Cladocopium cells, as well as on their cell wall (Fig. 1A-C).Chlamydiae are usually encased in large host-derived inclusions inside host cells [70], though they can sometimes reside in single-cell inclusions [71] or in the cytoplasm [72].Transmission electron microscopy confirmed the presence of chlamydial cells (∼400-600 nm in diameter) inside singlecell inclusions within Symbiodiniaceae (Fig. 1D and E and S4).This is similar to previously observed single-cell inclusions of Protochlamydia amoebophila in the amoeba Acanthamoeba castellanii [71], though chlamydial density in Cladocopium cells appears much lower.
Absolute quantification of chlamydiae through digital PCR (dPCR) revealed that there were an average of 1716 chlamydial cells/ml of culture supernatant (Fig. 2A) and that each Cladocopium cell was infected by an average 0.46 chlamydial cells (Fig. 2B) suggesting the presence of the common chlamydial developmental cycle with extracellular elementary bodies and intracellular replicative bodies.Because not all Cladocopium cells harbored chlamydiae (Fig. 1A-C), we measured the number of Cladocopium cells stained by FISH (both intracellularly and attached to the cell wall) by f low cytometry across a host growth cycle (once a week for 4 weeks, Fig. S2).This showed that an average of 30% of cells were infected by chlamydiae across four time points (Fig. 2C).Therefore, using our previous chlamydial absolute quantification, each infected Cladocopium cell (i.e.30% of the total number of cells) harbors an average of 1.52 chlamydial cells (Fig. 2B).

Cladocopium-chlamydiae infection is temporally stable
The temporal stability of the association was assessed at the population level by characterizing the bacterial communities throughout a host growth cycle (twice a week for 4 weeks) and across a single day (eight time points across 24 h, including four during the day and four at night) (Fig. S2 and Table S1).Bacterial communities were sampled in three fractions as previously described [7]: loosely associated (planktonic bacteria), closely associated (intracellular and tightly attached to the cell wall), and intracellular.In both experiments, a single chlamydial ASV made up the majority of the reads (Table S5).Chlamydial relative abundance was highest in the closely associated fraction, where it increased from 13% to 48% during host exponential phase and decreased once stationary phase was reached (Figs 2D and S5A and Table S5A).The same trend was observed in the loosely associated fraction, with a peak at 7.5% in relative abundance during host exponential phase.These findings suggest chlamydiae may replicate or be transmitted more easily when their host cells are dividing.The relative abundance of closely associated chlamydiae was higher during the day (33.1% across the four time points) than at night (26.7%) ( Fig. 2E and S5B and Table S5B).This may be due to the lack of host photosynthesis at night, resulting in lower amounts of adenosine triphosphate (ATP) available for chlamydial survival and replication.Chlamydial relative abundance in the intracellular fraction was very low (<5%) in both experiments, suggesting elementary bodies might be more abundant than replicative bodies in this culture.Alternatively, it might be the result of a technical bias from the sodium hypochlorite treatment during sample fractionation, which may affect bacterial DNA even inside Symbiodiniaceae cells and introduce biases in the measured relative abundances.Thus, the results from the intracellular fraction are to be treated carefully.

Cladocopium-associated chlamydiae belong to an undescribed Simkaniaceae genus
Using a combination of long-and short-read sequencing, we recovered a metagenome-assembled genome (MAG) of the Cladocopium-associated chlamydiae (Cla049; Table 1).The Cla049 MAG is comprised of 1 799 045 bp across seven contigs and estimated to be 91.73%complete.The MAG also includes one 87 402 bp plasmid (pAa).The last common ancestor of all chlamydiae likely possessed a plasmid, which was lost and/or integrated into the chromosome of some chlamydial lineages, but conserved in others [44,63].Cla049's plasmid possesses key genes present on other chlamydial plasmids, including pgp1 (a replicative DNA helicase) and pgp2 (a virulence protein) (Table S6).Other genes typically present on chlamydial plasmids, such as praA/pgp5 (a chromosome partitioning protein) and pgp6 (involved in host cell response mediation), are on noncircular contigs and therefore may have been integrated into the genome.Cla049 encodes 11 transposases, including two on its plasmid, hinting at a potential for gene f low between the plasmid and chromosome that may explain the presence of plasmid genes on noncircular contigs.
To assess the taxonomic placement of the Cla049 MAG, we constructed two maximum likelihood phylogenetic trees based on 15 conserved marker genes (Table S4) and 169 chlamydial genomes (Table S3), calculated with IQ-TREE using different models (Figs 3  and S6).Cla049 represents an undescribed, deep-branching genus within the Simkaniaceae family.Cla049 is most closely related to a group of three MAGs, including two isolated from activated sludge in Hong Kong (HK-STAS-VERR_A-4 and HK-STAS-VERR_A-5), and one from a marine microbial biofilm in Norway (OFTM343).Average AAI is ∼40-47% with all Simkaniaceae and Parasimkaniaceae genomes ( Table S7), supporting the placement of Cla049 into a novel genus and species, which we propose to name Candidatus Algichlamydia australiensis gen.nov.sp.nov.(A.australiensis Cla049 hereafter).The name has been registered through SeqCode [73].

Algichlamydia australiensis Cla049 has a reduced metabolic potential
The metabolic potential of A. australiensis is heavily reduced (Tables 2 and S8), lacking pathways for the synthesis of nucleotides, vitamins, and most amino acids (only genes for glutamate, aspartate, lysine, and alanine biosynthesis were found, as well as glycine-serine interconversion).The A. australiensis genome contains the genes necessary for glycolysis, the tricarboxylic acid cycle, pyruvate oxidation, the pentose phosphate pathway, and glycogenesis.Algichlamydia australiensis also possesses a complete shikimate pathway, as well as a complete menaquinone biosynthesis pathway, a cofactor that promotes growth in C. trachomatis [74].Finally, the genome was predicted to produce two secondary metabolites, most closely related to nostovalerolactone (Table S9), a putative transcriptional regulator [75].These  S2 for experimental design).Bacterial community profiling was performed in three fractions as previously described [7]: loosely associated bacteria (planktonic bacteria; blue), closely associated (intracellular and tightly attached to the cell wall; green) and intracellular bacteria (orange).Cladocopium cell counts across the growth phase experiments are also provided (dashed black line, D).Each point represents one of three replicate f lasks, with lines connecting the means.For each fraction, the effect of sampling day or time on chlamydial relative abundance is indicated under the plot with its corresponding color, based on a Kruskal-Wallis test.ns: P > .05;* : P ≤ .05;* * : P ≤ .01.Single replicate values are available in Table S5.metabolic abilities, or lack thereof, are similar to other chlamydiae [ 48,76] and suggest that A. australiensis acquires most metabolites and energy from its Cladocopium host.
Algichlamydia australiensis was predicted to encode five nucleotide transport proteins (NTTs), which were all most closely related to the four NTTs of S. negevensis, the Simkaniaceae-type species (Fig. S7).NTTs are critical for chlamydiae to import essential nutrients, including ATP, from their host cell [77,78].It is worth noting that plastid ATP/ADP antiporters in photosynthetic eukaryotes are derived from chlamydiae [79].In S. negevensis, SnNTT1 is an ATP/ADP antiporter, SnNTT2 is a guanine nucleotide/ATP/H + symporter, SnNTT3 acts as an RNA nucleotide antiporter, and no substrate was identified for SnNTT4 [78].However, SnNTT1's putative ortholog in A. australiensis Cla049 was split into three CDSs (EFCAOE_00685/00690/00695), with a STOP codon at the end of EFCAOE_00690 and a STOP codon and a frameshift at the end of EFCAOE_00695, and is thus presumably not functional, drastically diminishing the potential for ATP import and consequently energy parasitism.Nonetheless, A. australiensis Cla049 appears to possess two orthologs of SnNTT3 (Fig. S7; EFCAOE_00240 and EFCAOE_06465), which can import ATP in exchange for other nucleotides and may therefore compensate for the hypothetical nonfunctionality of NTT1.Experimental analyses are needed to determine the true substrates of A. australiensis Cla049's NTTs.Alternatively, because Cladocopium sp. is photosynthetic, it may have lower intracellular ATP and higher intracellular glucose than heterotrophic host cells, and A. australiensis Cla049 may rely more on glycolysis than ATP parasitism.S4) in 169 chlamydial genomes (Table S3).Confidence values based on 1000 ultrafast bootstrap replicates and 1000 replicates of the SH-like approximate likelihood ratio test are provided.Scale bar represents 1 nucleotide substitution per site.This tree was calculated using IQ-TREE 2 [51] with ModelFinder Plus under the LG + C20 + R4 model [52]; an additional tree calculated under the posterior mean site frequency (PMSF) model [55] using this tree as seed confirmed the phylogeny and is available in Fig. S6.MCF: metagenomic chlamydial family; CC-III: chlamydiae clade III.The chromosome and plasmid of Algichlamydia australiensis Cla049 encode a type III secretion system and type IV secretion system, respectively Algichlamydia australiensis Cla049 encodes most chlamydial hallmark virulence-associated genes that are present in other chlamydiae (Table S10) [76].This includes adhesins, T3SS effectors, Ser/Thr kinases (for host cell modulation), and developmental regulators.We also found 17 genes encoding putative eukaryotic-like repeat proteins (12 ankyrin repeat proteins, 1 WD40 repeat protein, and 4 tetratricopeptide repeat proteins), which are putative mediators of host-bacteria interactions and also abundant in other chlamydiae [63].Finally, A. australiensis Cla049 encodes a complete T3SS and T4SS (Fig. 4), the latter being encoded on the plasmid.Both secretion systems show high synteny with the corresponding coding regions of the chromosome and plasmid, respectively, of S. negevensis (Fig. 4).The chlamydial T3SS is highly conserved and allows for the translocation of effectors into eukaryotic host cells [80].The T4SS has only been reported in Simkaniaceae and Parachlamydiaceae, and its role remains unclear, though it may be involved in plasmid propagation [44,63,81].The chlamydial T4SS is plasmid-encoded in only three other known chlamydial species, S. negevensis, Protochlamydia naegleriophila, and Rubidus massiliensis, and T4SSencoding genes have been integrated in the chromosome of many Parachlamydiaceae and Simkaniaceae species [44,63].It was hypothesized that a T4SS of alphaproteobacterial origin was integrated into the plasmid of the Parachlamydiaceae ancestor and subsequently acquired by the plasmid of the Simkaniaceae ancestor [44].The presence of a plasmid-encoded T4SS in two distinct Simkaniaceae genera, Simkania and Algichlamydia, supports this hypothesis.

Unique genes in A. australiensis Cla049 may impact host-symbiont interactions
Because A. australiensis Cla049 is the first confirmed chlamydial symbiont of a photosynthetic organism, we asked whether it possessed any unique genes, not present in other known chlamydiae, which may mediate interactions with a photosynthetic host.We found 20 orthologous groups (OGs) that were specific to A. australiensis Cla049 ( Table S11), as well as 189 genes that were not classified into any OG (Table S12).Only one OG (OG0007644) was reliably annotated and was predicted to encode a lanthionine synthetase.Lanthionine is a major component of lantibiotics, a class of antimicrobial compounds [82].Some chlamydiae defend their hosts against pathogens; for exampl e, Parachlamydia acanthamoebae protects its amoeba host against viruses [83].Lantibiotics may protect Symbiodiniaceae from bacterial pathogens; alternatively, they may help reduce competition from other bacteria within Symbiodiniaceae cultures and facilitate chlamydial growth.Symbiodiniaceae Other OGs and unclassified genes were predicted to encode hypothetical proteins without any known motifs.Thus, we predicted the structure of one representative sequence per OG using AlphaFold and queried the RCSB Protein Data Bank to identify structurally similar proteins.Only 29 genes, including 4 out of the 20 OGs, resulted in reliable structural predictions (pLDDT >70; Fig. S8 and Tables S11 and S12).Among them, one gene (EFCAOE_07285) had two particularly relevant close hits: (i) a putative ankyrin repeat protein, which may be involved in host-chlamydiae interactions, and (ii) a vesicleinducing protein in plastids 1 (VIPP1), a protein involved in the biogenesis and integrity of thylakoid membranes of plants, algae, and cyanobacteria [84].In Arabidopsis thaliana, VIPP1 knockdown results in thylakoid swelling under high light [85], and VIPP1 overexpression improves postheat stress recovery [86].Therefore, EFCAOE_07285 may mediate interactions between A. australiensis Cla049 and Cladocopium's thylakoids and impact the photophysiology of Cladocopium cells.

Algichlamydia australiensis Cla049
Our previous study of Symbiodiniaceae-associated bacteria showed that ASVs at least 99% identical to A. australiensis Cla049 also infects cultures of Gerakladium sp.G3, Breviolum minutum, Durusdinium trenchii, and Fugacium sp.F5.1, though at much lower abundances than in Cladocopium sp.SCF049.01 (Table 3) [7].Because these Symbiodiniaceae cultures are themselves symbionts of cnidarians, we asked whether A. australiensis Cla049 infects in hospite Symbiodiniaceae (within cnidarians) or cnidarians themselves that are known to be chlamydial hosts [2,[87][88][89].Analysis of a recent dataset of 12 009 cnidarian 16S rRNA gene amplicon sequencing samples from 186 studies [90] showed that only five samples (each from a different study) harbored ASVs at least 99% identical to the 16S rRNA sequence of A. australiensis Cla049 (Table 3).The five samples were from four different genera of scleractinian corals, and the relative abundance of A. australiensis Cla049 varied between 0.01% and 0.31% (Table 3).This may be underestimated because of known biases from common universal primers against chlamydiae (e.g.515F/806R of the V4 region, or 1391R of the V8 region [91]), as well the low Table 3. Relative abundance of Algichlamydia australiensis Cla049 in Symbiodiniaceae (A) and cnidarian (B) 16S rRNA gene metabarcoding samples.The Symbiodiniaceae data were obtained from a previous study that analyzed bacterial community composition in 11 Symbiodiniaceae cultures [ 7].The cnidarian data were obtained from a recent dataset combining 186 cnidarian microbiome studies and 12 009 cnidarian samples [90].numbers of A. australiensis Cla049 in Cladocopium cells, which may be missed if the sequencing depth is too low.Nonetheless, this low abundance and prevalence also suggest that A. australiensis Cla049 is not a common symbiont of in hospite Symbiodiniaceae or cnidarians, and perhaps does not interact with cnidarian hosts at all. A. australiensis Cla049 may not be able to bypass the cnidarian immune system, therefore not having direct access to in hospite Symbiodiniaceae.Additionally, Symbiodiniaceae cell division is much slower in hospite compared to free-living Symbiodiniaceae [92].This may limit chlamydial growth and transmission opportunities among in hospite Symbiodiniaceae.

Conclusion
We provided a genomic characterization of a member of the chlamydiae that infects cultures of the dinof lagellate Cladocopium sp.SCF049.01.Although Symbiodiniaceae have been shown to feed on bacteria in some cases [98], it is unlikely to be the case here for the following reasons: (i) chlamydiae are known to escape host phagocytosis once internalized and avoid digestion [1,99], and (ii) high relative abundances of Simkaniaceae in this Cladocopium sp.culture were detected as far back as 2019 [7], in 2021 (sampling for the growth phase and time series experiments, as well as for the short-read metagenomic sequencing), in 2022 (sampling for the f low cytometry experiment), and in 2023 (sampling for the digital PCR and long-read metagenomic sequencing), confirming the long-term stability of this association.Further investigation is required to determine the potential inf luence of A. australiensis on their Cladocopium sp.host.Whether mutualistic (e.g. by protecting from other pathogens) or pathogenic, this infection would likely inf luence the survival and proliferation of free-living Symbiodiniaceae.By increasing or decreasing the available pool of free-living Symbiodiniaceae, this chlamydial infection may, in turn, affect the colonization of cnidarian hosts and the health of coral reefs-despite our findings that A. australiensis does not seem to be present in cnidarian holobionts.
As far as we know, a chlamydial infection in a photosynthetic organism has never been described before [2].Nonetheless, chlamydial 16S rRNA gene reads were recently detected in cultures of the green alga Ostreobium sp.[100,101], and chlamydial MAGs were retrieved from cultures of the green alga Amoebophrya sp. and the kelp Saccharina japonica [102].However, in all these examples, it was not established whether the photosynthetic organism or associated protists were the chlamydial host.Thus, additional studies are required to fully appreciate the potential breadth of photosynthetic hosts of chlamydiae, which would provide valuable insight into the evolutionary history of chlamydiae.This example of a chlamydia infecting a photosynthetic host cell also sheds new light on a hypothesis about a possible contribution of chlamydiae to the establishment of cyanobacterial symbionts during the evolution of the first photosynthetic eukaryotes [4][5][6].

Figure 2 .
Figure 2. Chlamydial infection of Cladocopium sp. is temporally stable.(A, B) Chlamydial abundance in culture supernatant (A) or cellular fraction (B), measured by digital PCR.For the cellular fraction (B), chlamydial gene copy data were normalized to either total Cladocopium cell numbers (i.e.every Cladocopium cell is infected) or 30% of Cladocopium cell numbers [i.e.only 30% of cells harbor chlamydiae; see (C)].Boxes represent first quartile to third quartile for six independent replicates (also shown individually), the middle lines represent the medians, and the whiskers represent the minimum and maximum values.(C) Proportion of Cladocopium sp. cells stained by FISH with the chlamydiae-specific probe Chls523, as analyzed by f low cytometry.Each point represents one of three replicate f lasks, with lines connecting the means.Gray-shaded areas represent dark time, while white areas represent light time.The effect of sampling day is indicated under the plot with its corresponding color, based on a Kruskal-Wallis test (ns: P > .05).(D, E) Relative abundance of chlamydial ASVs in the Cladocopium culture across 29 days (D, growth phase experiment) or 24 h (E, time series experiment, performed on Day 8 of the growth phase experiment-see black arrow in D), determined by 16S rRNA gene amplicon sequencing (see Fig.S2for experimental design).Bacterial community profiling was performed in three fractions as previously described[7]: loosely associated bacteria (planktonic bacteria; blue), closely associated (intracellular and tightly attached to the cell wall; green) and intracellular bacteria (orange).Cladocopium cell counts across the growth phase experiments are also provided (dashed black line, D).Each point represents one of three replicate f lasks, with lines connecting the means.For each fraction, the effect of sampling day or time on chlamydial relative abundance is indicated under the plot with its corresponding color, based on a Kruskal-Wallis test.ns: P > .05;* : P ≤ .05;* * : P ≤ .01.Single replicate values are available in TableS5.

Figure 3 .
Figure 3.The chlamydial symbiont Candidatus Algichlamydia australiensis (Cla049 MAG) belongs to an undescribed, deep-branching Simkaniaceae genus.Chlamydial maximum likelihood phylogeny based on 15 conserved gene markers (TableS4) in 169 chlamydial genomes (TableS3).Confidence values based on 1000 ultrafast bootstrap replicates and 1000 replicates of the SH-like approximate likelihood ratio test are provided.Scale bar represents 1 nucleotide substitution per site.This tree was calculated using IQ-TREE 2[51] with ModelFinder Plus under the LG + C20 + R4 model[52]; an additional tree calculated under the posterior mean site frequency (PMSF) model[55] using this tree as seed confirmed the phylogeny and is available in Fig.S6.MCF: metagenomic chlamydial family; CC-III: chlamydiae clade III.

Figure 4 .
Figure 4. Completeness, similarity, and synteny of the genomic regions encoding the T3SS (A, sct genes) and T4SS (B, tra genes) in Algichlamydia australiensis Cla049, compared with Simkania negevensis Z. Synteny is shown by lines joining the two genomes; protein identity is color-coded.

Table 2 .
List of complete and incomplete metabolic pathways in Algichlamydia australiensis Cla049.