Nuclear gene transformation in a dinoflagellate

The lack of a robust gene transformation tool that allows functional testing of the vast number of nuclear genes in dinoflagellates has greatly hampered our understanding of fundamental biology in this ecologically important and evolutionarily unique lineage. Here we report the development of a dinoflagellate expression vector, an electroporation protocol, and successful expression of introduced genes in the dinoflagellate Oxyrrhis marina. This protocol, involving the use of Lonza’s Nucleofector and a codon optimized antibiotic resistance gene, has been successfully used to produce consistent results in several independent experiments. It is anticipated that this protocol will be adaptable for other dinoflagellates and will allow characterization of many novel dinoflagellate genes.


INTRODUCTION
As widely distributed primary producers, essential coral endosymbionts, and the greatest contributors of harmful algal blooms and biotoxins in the ocean, dinoflagellates are a diverse group of unicellular protists with great ecological significance, evolutionary uniqueness, and numerous cytological and genomic peculiarities. Dinoflagellates have immense and permanently condensed genomes with many chromosomes [1][2][3] ; their genomes have a low protein-DNA ratio and histones are functionally replaced with dinoflagellate viral nuclear proteins 4,5 ; there are high numbers of repetitive non-coding regions and gene copies, in some species up to ~5,000 copies, organized in tandem arrays 6,7 ; only 5-30% of their genes are transcriptionally regulated [7][8][9] , and microRNAs seem to be the major gene regulating mechanism 10 ; and they have undergone extreme plastid evolution, transferring a massive quantity of plastid genes to the nucleus in most of the autotrophic species [11][12][13] . However, the molecular underpinnings of these unusual features remain elusive. In attempts to address the gap of knowledge, an increasing amount of effort has been made in the last decade to analyze dinoflagellate transcriptomes [14][15][16][17][18][19][20][21][22][23][24][25][26][27] and genomes 10,[28][29][30][31][32] . These experiments have provided not only extensive information on predicted genes and biological pathways, but also an even greater wealth of genes that have weak similarity to characterized proteins or no significant matches in databases. With the increasing volume of dinoflagellate transcriptomic and genomic data, the functional characterization of these novel genes has become a major bottleneck in translating system-level data into a mechanistic understanding of basic dinoflagellate biology, warranting the need for a dinoflagellate genetic transformation system. Gene transformation attempts have been reported for dinoflagellates by three separate groups. Ten and Miller (1998) 33 utilized silicon carbide whiskers, polyethylene glycol (PEG), and vigorous shaking to introduce foreign DNA into Amphidinium sp. and Symbiodinium microadriaticum with a success rate of ~1 ppm. Seventeen years later, Ortiz-Matamoros et al. used PEG, glass beads, shaking and, in some cases, co-incubation with Agrobacterium tumefaciens to transform foreign DNA into Fugacium kawagutii (formerly Symbiodinium kawagutii), S. microadriaticum, and an unclassified Symbiodiniaceae species 34,35 . Neither of these reports used codon optimized plasmids for dinoflagellate expression nor did they contain potential dinoflagellate promoters; moreover, both methods remain to be reproduced in other laboratories. In a recent study plasmids containing dinoflagellate minicircle DNA and an antibiotic resist gene were designed and introduced successfully through particle bombardment into the chloroplast genome of the dinoflagellate, Amphidinium carterae 36 . Here, we report a successful nuclear gene transformation method for the heterotrophic dinoflagellate, Oxyrrhis marina.
O. marina is a widespread and ecologically significant heterotrophic dinoflagellate. It is an established model species for both ecological and evolutionary research due its easy cultivable nature, extensive studies related to feeding behavior and nutrition, and its basal position in dinoflagellate phylogeny [37][38][39][40][41] . Although O. marina is an early branching dinoflagellate species, it still shares many of the peculiar biological characteristics described above and also retains more typical eukaryotic features that are lacking in later diverging dinoflagellate taxa; thus, it represents a good model for understanding dinoflagellate evolution [41][42][43] . In addition, O. marina has represented planktonic heterotrophs in experiments examining both how they feed and their nutritional value [44][45][46] . Through various studies as a prey species for copepods and rotifers, O. marina has been considered a trophic upgrade as they produce long-chain fatty acids, sterols, and essential amino acids that phytoplankton alone cannot 45,47,48 . Their nutritional value lead to the proposition of using O. marina as nutraceuticals for humans and agriculture 45 .
Although O. marina lacks a published genome, several transcriptomic studies are available 16,[49][50][51][52] , and the most exciting finding is O marina possess a potential proton pumping rhodopsin with homology to proteorhodopsin 50,51 . Proteorhodopsin is a retinal protein/carotenoid complex that utilizes sunlight to pump protons across a membrane, a non-photosynthetic form of light harvesting 53 . Dinoflagellate species across the phylogenic tree have been found to possess proteorhodopsin homologs, allowing the translational study of this proteins function in O marina to the other dinoflagellate species 16,17,50 . Therefore, having a genetic transformation system in place for O. marina will greatly excel our understanding of heterotrophic protist ecology, deepen our evolutionary understanding of dinoflagellates within their own branch and relative to other alveolates, allow exploration of the many predicted and novel dinoflagellate genes, and could tap into new industrial applications for O. marina, such as a food source or a potential alternative fuel.
Additionally, since O marina is a heterotrophic species, it is easier to detect the expression of introduced florescent proteins without interference from chlorophyll florescence, as in photoautotrophic species.
In this study, based on genomic and transcriptomic data from several dinoflagellates, we constructed a dinoflagellate expression system (named as DinoIII) that contains potential promoter and termination regions as well as important RNA elements. We incorporated a codon optimized rifampin resistance gene (DinoIII-arrO) and green fluorescent protein gene, gfp, (DinoIII-gfp) into DinoIII, and transformed this DNA as either PCR amplified fragments, excluding the plasmid O. marina using Lonza 4D-Nucleofector TM X system (Basel, Switzerland), a gene transformation system enabling transfer of genes directly into the cells' nucleus 54 . We have been able to repeat transformation for antibiotic resistance several times and verified the presence of both the antibiotic resistance gene and green fluorescent protein several months after transfection.

Construction of dinoflagellate backbone expression vector
Initially, the RNA complex sequence from dinoflagellate the Karenia brevis (GenBank accession # FJ434727) was inserted into the pMD TM 19-T plasmid vector (Takara, Kusatsu, Shiga Prefecture, Japan) and was used as the vector's skeleton for a series of modifications. After the addition of

Transformation using Lonza 4D-Nucleofector TM X Unit system
With Lonza's 4D-Nucleofector TM X Unit, specifically designed for hard-to-transfect cell lines, we went through an extensive cell optimization protocol for O. marina, and identified seven adequate pulse code settings. We used these seven pulse codes for follow-up experiments ( For other algae to be studied, full optimization tests with the other available solutions should be utilized when using Lonza's 4D-Nucleofector TM X Unit.

GFP Expression
The

Rifampin Resistance as a selection marker
To facilitate purification of transformed O. marina cells, we screened several selection markers and found rifampin as the most suitable for O. marina. Rifampin is an antibiotic used to treat tuberculosis, leprosy, and Legionnaire's disease and its resistance in bacteria is due to Rifampin ADP-ribosyltransferase activity 55 . We introduced our codon-optimized homolog through our DinoIII vector (DinoIII-arrO) and obtained expression of arrO, which was verified in several ways. First, the transformed cell culture survived and grew while the wildtype died completely in rifampin-containing medium. Second, after approximately one month we isolated RNA and DNA from both the transformed cells cultured in rifampin-containing medium and a wildtype culture grown in rifampin-free growth medium, and performed reverse-transcription PCR (RT-PCR). We detected the expression of the resistance gene arrO only in the experimental treatment and not in the wildtype (Fig. 3A). In addition, we sequenced the PCR product and confirmed that it was arrO.
Finally, the expression of arrO was detected from the cDNA synthesized using Oligo-dT as the primer (Fig. 3B), indicating the transcript of arrO was polyadenylated, a phenomenon best known for occurring mostly in eukaryotes mRNA.
Although the cells survived for more than one month, the population increased very slowly and did not seem healthy, probably due to low expression efficiency of the resistance gene. We we saw an increase in growth rate under antibiotic selection and verified its expression, as reported above, but this time three month after transfection (Fig. 3C). We still, when writing this manuscript, have the cell lines in culture.

DISCUSSION
In order to improve understanding of basic dinoflagellate biology, a gene transformation protocol is urgently needed to characterize the function of dinoflagellate genes, particularly the vast number of nuclear genes. A robust and reproducible protocol has been long-awaited. After testing multiple methods (including previously reported ones) and numerous conditions, we have found a passage and herein report a genome-targeted transformation method using a dinoflagellate gfp vector (DinoIII-gfp) and two dinoflagellate rifampin resistance vectors (DinoIII-arrO and DinoIII-arrO-N) that were developed based on dinoflagellate genomic and transcriptomic data.

Effectiveness of the promoter elements
Our efforts began with utilizing expression vectors from the previous reported dinoflagellate transformation 34 10,60 , which is present 65 base pairs upstream from the start codon in our intergenic region and is also present in our "promoter" region with the closest motif 133 base pairs upstream from the start codon. Whether or not these sequences are important can be evaluated in future studies using our method.

Effectiveness of DNA introduction method
O. marina is a naked dinoflagellate that had a difficult time withstanding the physical forces used in previously reported dinoflagellate transformation methods. Electroporation is a gentler method, allowing DNA to pass through temporary pores in an organism's membrane and has been utilized in many organisms, but requires the removal of seawater and replacement with electroporation buffers, often unable to maintain the osmolality of marine organisms 61  antibiotic concentrations that can be used to select transformed cell lines, and apply an antibiotic cocktail that will reduce potential microbial communities in the culture.
Ideally a selection marker and a reporter gene can be located on the same plasmid allowing for dual expression. We attempted to put both the arrO and gfp genes within one single DinoIII vector in multiple arrangements (with or without stop codon in between, fused or not fused) but, unfortunately, the simultaneous expression of both genes was difficult to obtain. Future studies using arrO and gfp with the rhodopsin intergenic region in between could potentially get over this hurdle.

Conclusion
Despite the proven challenges, we have developed a

Constructing Dinoflagellate Expression Vectors
To optimize the utilization of our dinoflagellate expression system several regions were amplified from dinoflagellate genomes and were incorporated to serve as the vector backbone.  From the F. kawagutii genome sequence data 10 , we located the highly expressed light harvesting complex (LHC) gene. Its upstream "promoter" region (672bp; Supplementary Table 2) and downstream "termination" region (812bp; Supplementary Table 3) were PCR-amplified using the following primer sets: SymkaLHC5FN1 and SymLHC3_5R for the "promoter" and SymLHC5_3F and SymkaLHC3R1 for the "termination" region. All PCRs were performed at 94C for 1 min, 25 cycles at 95C for 15s, 68C for 30s, and 72C for 1 min, and 1 cycle of 72C for 10 min. The sizes of the amplicons were checked by electrophoresis and DNA was purified by passing through a DNA column (Zymo).
SymLHC3_5R and SymLHC5_3F were designed to contain an overhang of either a portion of the "termination" region or the "promoter" region, respectfully, in order to link the two PCR products; thus, the two products were used in an equal molar ratio as template for the second PCR at 94C EcoRI and treated with alkaline phosphatase to avoid self-ligation. After 3.5 hours of digestion both products were purified by ethanol precipitation and ligated overnight in a 2:1 molar ratio (LHC product:vector) and transformed into competent E. coli cells. The colonies obtained were picked randomly, and plasmids were isolated and sequenced to identify the best clone harboring the correct DinoSL Complex-LHC region sequence, giving rise to the dinoflagellate expression vector backbone, DinoIII (5137bp; Fig. 1; Supplementary Table 4).
SymLHC3_5R and SymLHC5_3F primers were designed to also have an XbaI and BglII site in between the "Promoter" and "termination" regions so that a gene, either a reporter or an antibiotic resistant gene, could be inserted in the correct orientation. Accordingly, both a gfp gene and a rifampin resistance gene were incorporated into the expression vector, to yield DinoIII-gfp and DinoIII-arrO, respectively (Supplementary Table 5  well. New antibiotic solution was added every 3 weeks but the concentration was dropped down to 200 g/L to allow for greater cell growth and D. tertiolecta in 225 g/L rifampin medium was supplied whenever they were no longer detected in the medium.

Detecting the transformed gene and its expression
Total DNA was isolated from both wildtype (WT) and +arrO O. marina cultures using our CTAB method 65 and total RNA was isolated using the Trizol-Chloroform method in combination with Zymo Quick RNA Miniprep Kit (Irvine, CA, USA) 66 . As the transformed cultures grew slowly under antibiotic pressure, only 200-400 cells were available. These cells were divided into two for DNA and collected on TSTP Isopore 3m membrane filters (MilliporeSigma, Burlington, MA, USA). DNA and RNA were isolated as reported 16,66 .
If cell numbers were very low only OdT was used for cDNA synthesis to maximize cDNA production. Negative controls were included where no reverse transcriptase was added and was instead replaced with DEPC water.
PCR was performed using both the DNA and cDNA as templates with primer set arr2Q1Fa-arr2Q1Ra. Due to the low number of transformed cells used in DNA and RNA isolation, this PCR did not yield detectable amounts of products. The PCR products were then diluted 1000-and 10000-fold and used as template for a nested PCR with arr2Q1F-arr2Q1R as the primer (Table 2) for quantitative PCR. The products were run on a gel and sequenced directly to ensure the arrO gene was correctly amplified.