The endosymbiotic theory that explains the origin of mitochondria and plastids in eukaryotic cells has been widely accepted for more than a century. However, the genetic mechanisms underlying the establishment of permanent endosymbiosis remain largely unclear. While all eukaryotic mitochondria are now believed to originated from one single endosymbiotic event, plastids underwent a far more complex endosymbiotic evolutionary history, leading to a wide range of disparate eukaryotic lineages affiliated with each other. Evolutionary footprints of establishing permanent endosymbiosis in most lineages were blurred and almost erased over time. Thankfully, transient endosymbiosis has been found to exist in several kleptoplastic ('stealing' plastids for acquired phototrophy) protists and the sacoglossan sea slugs, which provides evolutionarily divergent models for studying the intermediate state that may lead to permanent endosymbiosis. During the PhD studies, my major interest lies in exploring gene regulation and genetic adaptation in two kleptoplastic species – the red tide ciliate Mesodinium rubrum and the 'solar powered' sea slug Elysia chlorotica.

Mesodinium rubrum, a worldwide red tide ciliate, preys on cryptophytes in the genus Geminigera, Teleaulax, and Plagioselmis. It sequesters chloroplasts, nuclei, and other cell organelles from the cryptophyte prey in order to harvest light energy and produce nutrients. Previous studies in the M. rubrum system were mainly focused on the gene expression of the cryptophyte nuclei before and after sequestration but left the host genes behind because they had difficulty distinguishing gene products originating from the host and the endosymbionts in this system. Whole genome sequences from the host are ideal for picking out the host gene products, but ciliate genomes are challenging to assemble due to their complex organization. To bypass the dependence on whole genome sequences, we have developed an economical and time-efficient strategy to construct separate reference gene sets for the host and the endosymbiont simultaneously, with short-read sequencing data alone and comprehensive bioinformatics methods.

The separate gene sets allowed us to investigate transcriptional changes of M. rubrum and the sequestered prey nuclei simultaneously. M. rubrum coordinates biological processes of itself and the cryptophyte prey through actively transcribing transmembrane transporter genes and implements considerable control over the prey nuclei. Specifically, cryptophyte genes related to subcellular structure and cell motility were generally suppressed, while overall nutrition metabolism, biosynthesis, DNA replication as well as photosynthesis were enhanced after sequestration. These transcriptional changes of the prey nuclei are expected to benefit the host M. rubrum. However, this control is not yet well adjusted to environmental changes as the prey nuclei lose light-induced transcriptional responses and undesirably express light-harvesting genes in the darkness upon sequestration by M. rubrum. M. rubrum itself also shows very few adjustments in its own gene expression in subject to light conditions. Our results suggest M. rubrum can force only one expression pattern out of its acquired cryptophyte nucleus regardless of light conditions, which might be a common trend in transient endosymbiosis.

Horizontal gene transfer (HGT) is considered to be a common mechanism that supplements kleptoplastic protists with necessary genes to maintain the stolen plastids. However, whether the kleptoplastic sea slugs integrated alga-derived HGTs into their genomes has been debated for decades. We thus sequenced the genome of E. chlorotica, a 'solar-powered' sea slug that is able to sequester plastids from its algal prey and survive by photosynthesis for up to 12 months in the absence of food supply. To conquer the challenges imposed by the extremely high heterozygosity and repeat content of this genome, we designed a hybrid and hierarchical assembly strategy that made use of both Illumina short reads and the PacBio long reads. Through a set of strict evaluations, we demonstrated that this hybrid assembly presents a high level of per-base accuracy and overall completeness. We examined the controversial HGTs in the genome of E. chlorotica but found no evidence for alga-derived HGTs in the E. chlorotica genome, suggesting that alga-derived HGTs are likely not responsible for the long-term maintenance of stolen plastids in this animal.

In summary, our works provide practice examples in the study of endosymbiotic systems and have broad implications for the evolution of establishing endosymbiotic relationships.