*Article* **Metagenomic Analysis of the Species Composition and Seasonal Distribution of Marine Dinoflagellate Communities in Four Korean Coastal Regions**

**Jinik Hwang <sup>1</sup> , Hee Woong Kang <sup>1</sup> , Seung Joo Moon <sup>2</sup> , Jun-Ho Hyung <sup>2</sup> , Eun Sun Lee <sup>2</sup> and Jaeyeon Park 2,\***


**Abstract:** Biomonitoring of dinoflagellate communities in marine ecosystems is essential for efficient water quality management and limiting ecosystem disturbances. Current identification and monitoring of toxic dinoflagellates, which cause harmful algal blooms, primarily involves light or scanning electron microscopy; however, these techniques are limited in their ability to monitor dinoflagellates and plankton, leaving an incomplete analysis. In this study, we analyzed the species composition and seasonal distribution of the dinoflagellate communities in four Korean coastal regions using 18S rRNA amplicon sequencing. The results showed significantly high diversity in the dinoflagellate communities in all regions and seasons. Furthermore, we found seasonally dominant species and causative species of harmful algal blooms (*Cochlodinium* sp., *Alexandrium* sp., *Dinophysis* sp., and *Gymnodinium* sp.). Moreover, dominant species were classified by region and season according to the difference in geographical and environmental parameters. The molecular analysis of the dinoflagellate community based on metagenomics revealed more diverse species compositions that could not be identified by microscopy and revealed potentially harmful or recently introduced dinoflagellate species. In conclusion, metagenomic analysis of dinoflagellate communities was more precise and obtained results faster than microscopic analysis, and could improve the existing monitoring techniques for community analysis.

**Keywords:** dinoflagellates; metagenomics; next-generation sequencing; monitoring

### **1. Introduction**

Marine dinoflagellates are ubiquitous and play diverse roles in marine ecosystems [1,2]. Some dinoflagellate species can grow out of control due to various environmental factors, such as excessive inorganic nutrients (nitrogen (N) and phosphorus (P)) introduced from the land, forming a bloom [3,4]. Blooms from dinoflagellates have detrimental effects on a variety of aquatic animals, including fish and aquatic mammals, and can even be harmful to humans through toxin production [5,6]. Therefore, continuous monitoring of dinoflagellate communities is essential, as they can affect the diversity of surrounding aquatic life and cause ecosystem disturbance.

To date, monitoring of dinoflagellates in the aquatic environment has generally involved morphological identification using light microscopy observations. Recent advances in microscopy, including scanning electron microscopy (SEM), have enabled more precise identification [7,8]. However, the morphological classification of plankton via microscopy is still challenging, as plankton are difficult to observe with SEM due to the lack of an outer shell in dinoflagellates or the extremely small size of plankton. Recently, many types of species identification technology to distinguish dinoflagellates and molecular technology targeting species-specific genes have been developed [9,10]. In particular, next-generation

**Citation:** Hwang, J.; Kang, H.W.; Moon, S.J.; Hyung, J.-H.; Lee, E.S.; Park, J. Metagenomic Analysis of the Species Composition and Seasonal Distribution of Marine Dinoflagellate Communities in Four Korean Coastal Regions. *Microorganisms* **2022**, *10*, 1459. https://doi.org/10.3390/ microorganisms10071459

Academic Editor: Simon K. Davy

Received: 29 April 2022 Accepted: 18 July 2022 Published: 19 July 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

sequencing (NGS) has greatly expanded our understanding of the diversity and function of dinoflagellates in the aquatic environment. This technique allows for rapid, high-resolution analysis of microbial and dinoflagellate communities [11,12]. In addition, it is possible to accurately identify nano- and pico-sized plankton, which are difficult to distinguish with a conventional microscope, facilitating the identification of various plankton that have been overlooked because they do not appear or are difficult to distinguish in local environmental conditions [13]. Although the QIIME or USEARCH pipeline has been widely used to analyze 16S rRNA gene sequencing reads from microbial communities [14–16], many metagenomics studies examining the profile of marine dinoflagellates have been carried out using the CLC Genomics Workbench [17–20]. In this study, we analyzed taxonomic profiling and seasonal distribution of the dinoflagellate communities in four Korean coastal regions based on the reading of 18S rRNA sequences using the CLC Workbench. To verify the results calculated using the CLC tool, those results were compared with abundance measured by direct counting of cells using microscopy.

Outbreaks of harmful dinoflagellates have traditionally occurred in tropical or temperate regions which have the potential for enhancing the growth rate of phytoplankton cells under the appropriate environmental conditions. Jeju Island, located along the southern coast of Korea, is a temperate region, and the occurrence of benthic dinoflagellates producing phytotoxins has been frequently reported in Jeju [21]. Understanding the spatial and seasonal dynamics of the toxic dinoflagellates in this region is essential, and many researchers have continuously monitored the cell abundance around Jeju Island using microscopic identification [22–24]. In this study, we investigated the spatial and seasonal variation of dinoflagellate communities in four different sites in Korean coastal waters, including Jeju Island, using NGS-based (18S rRNA amplicon) metagenomics. For precise bioinformatics analyses, we established a reference database of dinoflagellates and analyzed the precision of NGS compared to conventional microscopic observation. Thus, the reference data for the dinoflagellate community classified based on the NGS findings in this study will provide a better understanding of the occurrence of toxic dinoflagellates in Korea.

#### **2. Materials and Methods**

#### *2.1. Study Areas and Seawater Sample Collecting*

Seawater samples for metagenomic analysis were collected from four coastal waters (Gunsan, Pohang, Tongyeong, and Seongsan) in March, June, September, and December 2019. The four selected sampling sites have different geological and environmental characteristics, representing the eastern coast (Pohang), southern coast (Tongyeong), western coast (Gunsan), and Jeju island (Seongsan), and all four locations are near a port with considerable human activity (Figure 1a). To remove large zooplankton and foreign substances in the sample, surface seawater at each region was sieved using meshes with pore sizes of 80 µm. Four liters of seawater samples for metagenomics analysis were filtered through a polycarbonate filter membrane (0.8 µm Millipore; MilliporeSigma, Burlington, MA, USA) to obtain environmental DNA samples, then transferred to the laboratory on dry ice. For microscopic analysis, 500 mL of seawater samples was fixed with Lugol's solution, and phytoplankton cells were identified to at least the genus level using an optical microscope (Axioskop; Zeiss, Oberkochen, Germany). The dinoflagellate cells were counted directly using a Sedgwick-Rafter counting chamber by light microscopy (BX53; Olympus, Tokyo, Japan). Environmental data, such as water temperature, pH, dissolved oxygen, and conductivity, were measured at each location using a YSI 566 Multi Probe System (YSI Inc., Yellow Springs, OH, USA).


**Figure 1.** Location of sample sites and environmental indices at these sites (**a**) in four regions of Korean coastal waters (**b**). **Figure 1.** Location of sample sites and environmental indices at these sites (**a**) in four regions of Korean coastal waters (**b**).

#### *2.2. DNA Extraction, Library Preparation, and NGS 2.2. DNA Extraction, Library Preparation, and NGS*

DNA was extracted from the filtered membranes containing dinoflagellates and microbial cells using a DNeasy PowerSoil Kit (Qiagen, Hilden, Germany) following the manufacturer's instructions. The amount of double-stranded DNA and the purity in the extracted DNA samples was measured by PicoGreen (Promega, Madison, WI, USA) using VICTOR Nivo (PerkinElmer, Waltham, MA, USA). Per the Illumina 16S Metagenomic Sequencing Library protocols, the V3-V4 region of 18S ribosomal DNA (rDNA) gene in each sample was amplified by PCR using the following primers: 18S amplicon PCR forward primer, 5′–TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGCCAGCASCYGC GGTAATTCC-3′, reverse primer, 5′–GTCTCGTGGGCTCGGAGATGTGTATAAG -AGA-CAGACTTTCGTTCTTGATYRA-3′ [25]. A subsequent amplification step with limited-cycle reaction was performed to add multiplexing indices and Illumina sequencing adapters. The PCR products were pooled, cleaned, and normalized using the PicoGreen, and the size of libraries was measured using a TapeStation DNA screen tape D1000 (Agilent DNA was extracted from the filtered membranes containing dinoflagellates and microbial cells using a DNeasy PowerSoil Kit (Qiagen, Hilden, Germany) following the manufacturer's instructions. The amount of double-stranded DNA and the purity in the extracted DNA samples was measured by PicoGreen (Promega, Madison, WI, USA) using VICTOR Nivo (PerkinElmer, Waltham, MA, USA). Per the Illumina 16S Metagenomic Sequencing Library protocols, the V3-V4 region of 18S ribosomal DNA (rDNA) gene in each sample was amplified by PCR using the following primers: 18S amplicon PCR forward primer, 5 0–TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGCCAGCASCYGC GGTAATTCC-3 0 , reverse primer, 50–GTCTCGTGGGCTCGGAGATGTGTATAAG -AGACAGACTTTCG-TTCTTGATYRA-30 [25]. A subsequent amplification step with limited-cycle reaction was performed to add multiplexing indices and Illumina sequencing adapters. The PCR products were pooled, cleaned, and normalized using the PicoGreen, and the size of libraries was measured using a TapeStation DNA screen tape D1000 (Agilent Technologies, Santa

Clara, CA, USA). Sequence libraries in the sample were verified using the MiSeq™ platform (Illumina, San Diego, CA, USA).

#### *2.3. Customized Dinoflagellate Reference Databases for CLC Workflows*

For the DNA reference databases of dinoflagellates, a list of 1555 species of dinoflagellates named in a previous study [26] was prepared in the form of Excel data, and the reference database deposited in the NCBI was additionally downloaded. A total of approximately 5000 dinoflagellate reference databases were retrieved. The files were imported into CLC and customized for use as databases specified for analyzing dinoflagellate species. The analysis program used in this study was CLC Genomics Workbench 21.0.4 with CLC Microbial Genomics Module 21.0 (CLC Bio, Qiagen Company, Aarhus, Denmark) and was used for future species identification (Figure S1).

#### *2.4. Data Quality Control and Taxonomic Profiling*

Data quality control and taxonomic profiling were performed using the CLC Microbial Genomics Module (MGM). First, Reads were trimmed using the Trim Reads tool. The percentage of trimmed from approximately 300,000 reads per sample was 71% (*n* = 16). We trimmed the 50 and 30 terminal nucleotides of the reads, and discarded unqualified reads showing that the quality limit was less than 0.001 or ambiguous nucleotides were more than two. The average length of reads after trimming was between 217–234 bp. Samples with less than 100 reads (minimum percent from the median = 50.0) were removed. Second, the remaining qualified reads were used for operational taxonomic unit (OTU) clustering based on SILVA 18s v132 Database including 1555 dinoflagellates at a 97% sequence similarity. The detected chimeric sequences and singletons (Chimera crossover cost = 3, K-mer size = 6) were discarded. A phylogenetic tree using the neighbor-joining method with 100 replicates was constructed based on the aligned OTU sequences by the MUSCLE tool v3.8.425. The phylogeny was applied for alpha and beta diversity measures. The beta diversity was measured using the Euclidean distance, and principal coordinate analysis (PCoA) based on a Bray–Curtis dissimilarity matrix was performed to illustrate a hierarchical clustering heat map showing the correlation between the examined samples.

#### **3. Results**

#### *3.1. Environmental Characteristics of Sampling Sites*

The four selected sampling sites had different geological and environmental characteristics. All the regions showed four distinct seasons; however, there was a regional difference in water temperature. The month of March showed the lowest water temperature (6.4–14.3 ◦C) throughout the region, and September (20.1–26.3 ◦C) showed the highest water temperature. On average, the water temperature at Jeju Island (Seongsan) was higher than that of the land. The salinity did not show a significant difference by region (31.4–33.7‰), and the pH and dissolved oxygen amount also did not show significant regional changes (Figure 1b).

#### *3.2. Metagenome Comparisons*

A pipeline for metagenomics analysis of environmental DNA samples was developed to address the identification of dinoflagellates species. On average, over 300,000 reads were acquired from each region using the MiSeq™ platform (Illumina, San Diego, CA, USA), with a read length of 301 bp. After quality trimming and filtering of reads, 70.3% of the raw reads remained (Figure 2a), with an overall higher G+C content for reads obtained from the library.

March samples of Pohang, Tongyeong, and Seongsan were also similar. The June and September samples of Tongyeong, in which a single species bloomed and became dominant, showed no similarity with the other samples. Furthermore, low similarity was found

at Gunsan in June compared with the other samples (Figure 2b).

**Figure 2.** Comparison of metagenome libraries. Next-generation sequencing metadata including number of reads and trimmed reads (**a**), β-diversity (principal coordinate analysis (PCoA), dinoflagellate genotype composition (proportions) was measured by Bray–Curtis distances (**b**). **Figure 2.** Comparison of metagenome libraries. Next-generation sequencing metadata including number of reads and trimmed reads (**a**), β-diversity (principal coordinate analysis (PCoA), dinoflagellate genotype composition (proportions) was measured by Bray–Curtis distances (**b**).

*3.3. Metagenomic Analysis of the Dinoflagellate Species Composition*  To identify marine dinoflagellates, we used the CLC genomics workbench program (CLC Microbial Genomics Module) on the assembled read sequences, followed by BLAST searches on the NCBI database and the newly created database of 1555 dinoflagellate species. Following the metagenomic analysis, 64 species of dinoflagellate were found in all regions on average. The top 10 dinoflagellates were selected based on the analyzed reads (Table 1). The nucleotide sequence similarity of the dinoflagellate genes was expressed by region using PCoA to illustrate the overall regional similarity according to the season. The December samples for Gunsan, Tongyeong, and Seongsan showed similarities, and the March samples of Pohang, Tongyeong, and Seongsan were also similar. The June and September samples of Tongyeong, in which a single species bloomed and became dominant, showed no similarity with the other samples. Furthermore, low similarity was found at Gunsan in June compared with the other samples (Figure 2b).

#### *3.3. Metagenomic Analysis of the Dinoflagellate Species Composition*

To identify marine dinoflagellates, we used the CLC genomics workbench program (CLC Microbial Genomics Module) on the assembled read sequences, followed by BLAST searches on the NCBI database and the newly created database of 1555 dinoflagellate species. Following the metagenomic analysis, 64 species of dinoflagellate were found in all regions on average. The top 10 dinoflagellates were selected based on the analyzed reads (Table 1).

**Table 1.** Seasonal variations and distribution of dinoflagellates in four coastal waters (Gunsan, Pohang, Tongyeong, and Seongsan) by metagenomic analysis. Total dinoflagellate reads and unidentified reads (**a**), and proportion(%) of the 10 most common dinoflagellate species (**b**).



In Gunsan, the western coast, the highest number of reads detected by metagenomics data was seen in December (Table 1a). In March, two species (*Karlodinium veneficum* and *Gyrodinium* sp.) were dominant. When the ratio (%) of the top 10 species was calculated based on the total reads matched with dinoflagellate sequence, *Karlodinium veneficum* was the most dominant species, approximately 32%. Next, *Gyrodinium* sp. (24%) and *Gymnodinium* sp. (8%). In June, the composition of *Gonyaulax* sp. showed approximately 45%, followed by that of *Symbiodinium* sp. (15%) and *Karlodinium veneficum* (8%). Similar to

March, the dominant species in September was *Karlodinium* sp., which accounted for 24%. In December, *Gyrodinium* sp. (34%) and *Amphidiniella* sp. (25%) were dominant as well as *Ceratium* sp. which accounted for 11% (Table 1b, Figure 3a). *Microorganisms* **2022**, *10*, x FOR PEER REVIEW 8 of 15

> **Figure 3.** Proportion of top 3 most abundant species in each coastal seawater by metagenome analysis. Gunsan (**a**), Pohang (**b**), Tongyeong (**c**), Seongsan (**d**). **Figure 3.** Proportion of top 3 most abundant species in each coastal seawater by metagenome analysis. Gunsan (**a**), Pohang (**b**), Tongyeong (**c**), Seongsan (**d**).

> Figure 4 illustrates the most common species in the four coastal waters. In March, there were three common species at all sampling sites: *Cochlodinium* sp., *Gyrodinium* sp. *Gymnodinium* sp., and *Pelagodinium* sp. The most common species in June were *Akashiwo* sp., *Karlodinium* sp., *Peridinium* sp., *Pelagodinium* sp., and *Prorocentrum* sp. In September, *Akashiwo* sp., *Bysmatrum* sp., *Ceratium* sp., *Katodinium* sp., *Sinophysis* sp., and *Peridinium*  sp. were common. The common species in December were *Akashiwo* sp., *Alexandrium* sp., *Bysmatrum* sp., *Gyrodinium* sp., Hetrocapsa sp., *Peridiniopsis* sp., *Prorocentrum* sp., *Scrippsiella* sp., and *Symbiodinium* sp. In Pohang, the eastern coast, *Gyrodinium* sp. was dominant in June and December, *Katodinium* sp. was dominant in March, and *Karlodinium* sp. was dominant in September. (Table 1, Figure 3b). In Tongyeong, the southern coast, the appearance of *Gyrodinium* sp. was high in March and December, and was dominant at 24% and 62%, respectively. In particular, *Cochlodinium* sp. formed a red tide and dominated over 77%, and in June, the dominance of *Prorocentrum* sp. was more than 50%. The diversity was the highest in December, when 28 species of dinoflagellate reads were detected (Table 1, Figure 3c). In Seongsan, *Gyrodinium* sp. appeared at a high rate in all seasons, while *Bysmatrum arenicola* and *Karlodinium veneficum* dominated in June (56%) and September (19%), respectively. The sand-dwelling dinoflagellate *Bysmatrum arenicola* was dominant at Seongsan, except in March (Table 1, Figure 3d).

> Figure 4 illustrates the most common species in the four coastal waters. In March, there were three common species at all sampling sites: *Cochlodinium* sp., *Gyrodinium* sp. *Gymnodinium* sp., and *Pelagodinium* sp. The most common species in June were *Akashiwo*

sp., *Karlodinium* sp., *Peridinium* sp., *Pelagodinium* sp., and *Prorocentrum* sp. In September, *Akashiwo* sp., *Bysmatrum* sp., *Ceratium* sp., *Katodinium* sp., *Sinophysis* sp., and *Peridinium* sp. were common. The common species in December were *Akashiwo* sp., *Alexandrium* sp., *Bysmatrum* sp., *Gyrodinium* sp., Hetrocapsa sp., *Peridiniopsis* sp., *Prorocentrum* sp., *Scrippsiella* sp., and *Symbiodinium* sp. *Microorganisms* **2022**, *10*, x FOR PEER REVIEW 9 of 15

**Figure 4.** Seasonal common dinoflagellate species in 4 coastal waters by metagenome analysis. **Figure 4.** Seasonal common dinoflagellate species in 4 coastal waters by metagenome analysis. March (**a**), June (**b**), September (**c**), and December (**d**).

#### March (**a**), June (**b**), September (**c**), and December (**d**). *3.4. Comparison of Metagenomic Analysis and Microscopic Observation*

*3.4. Comparison of Metagenomic Analysis and Microscopic Observation*  When the abundance of dinoflagellates was analyzed by microscopic observation, the number of species composition was mostly lower than from metagenomic analysis (Table 2). Overall, the number of species in December was lower than in other seasons, as the biomass was considerably low and mainly dominated by diatoms. At Gunsan, the abundance of *Gyrodinium* sp. species was 0.8–2.9 cells mL−1 in March, September, and December, which showed similar patterns to the metagenomic analysis. In Pohang, the species composition in June was more diverse than in the other seasons, and two species of *Heterocapsa rotundata* (77.8 cells mL−1) and *Heterocapsa triquetra* (12.1 cells mL−1) were dominant. Similarly, the number of reads of *Heterocapsa triquetra* detected by the metagenomic analysis in the same sample were high. In Tongyeong, cell abundance of *Prorocentrum triestinum* (June) and *Cochlodinium polykrikoides* (September) was 341.1 and 2034 cells mL−1, respectively, which was similar to the metagenome result that the number of reads of *Prorocentrum* sp. and *Cochlodinium* sp. was 11,335, and 53,412, respectively. Small thecated When the abundance of dinoflagellates was analyzed by microscopic observation, the number of species composition was mostly lower than from metagenomic analysis (Table 2). Overall, the number of species in December was lower than in other seasons, as the biomass was considerably low and mainly dominated by diatoms. At Gunsan, the abundance of *Gyrodinium* sp. species was 0.8–2.9 cells mL−<sup>1</sup> in March, September, and December, which showed similar patterns to the metagenomic analysis. In Pohang, the species composition in June was more diverse than in the other seasons, and two species of *Heterocapsa rotundata* (77.8 cells mL−<sup>1</sup> ) and *Heterocapsa triquetra* (12.1 cells mL−<sup>1</sup> ) were dominant. Similarly, the number of reads of *Heterocapsa triquetra* detected by the metagenomic analysis in the same sample were high. In Tongyeong, cell abundance of *Prorocentrum triestinum* (June) and *Cochlodinium polykrikoides* (September) was 341.1 and 2034 cells mL−<sup>1</sup> , respectively, which was similar to the metagenome result that the number of reads of *Prorocentrum* sp. and *Cochlodinium* sp. was 11,335, and 53,412, respectively. Small thecated dinoflagellate species, such as *Azadinium* sp. and *Bysmatrum* sp., occurred in the Seongsan region, located at Jeju Island. Some small nano-planktonic dinoflagellates, which are difficult to identify by

dinoflagellate species, such as *Azadinium* sp. and *Bysmatrum* sp., occurred in the Seongsan region, located at Jeju Island. Some small nano-planktonic dinoflagellates, which are difficult to identify by microscopy, were easily found at Seongsan and Tongyeong using the

metagenomic analysis (Table 2).

microscopy, were easily found at Seongsan and Tongyeong using the metagenomic analysis (Table 2).

**Table 2.** Species composition and cell number of dinoflagellates analyzed by microscopic observation. Seasonal (March, June, September, December) species composition in four coastal regions (Gunsan, Pohang, Tongyeong, Seongsan).


\*Small thecated dinoflagellates: *Apicoporus*, *Azadinium*, *Crypthecodinium*, *Durinskia*, *Heterocapsa*, *Pfiesteria*. \*Small naked dinoflagellates: *Amphidiniopsis*, *Biecheleria*, *Karlodinium*, *Gymnodinium*, *Gyrodiniellum*, *Paragymnodinium*, *Pelagodinium*, *Symbidinium*.

Although not all species of dinoflagellates identified by microscopic observation were included in the metagenomic analysis, the appearance of dominant species was found to be quite similar (Table 2).

#### *3.5. Seasonal Distribution of Harmful Species Based on Metagenomic Analysis*

Four species of dinoflagellates (*Cochlodinium* sp., *Alexandrium* spp., *Dinophysis* spp., and *Gymnodinium* sp.) were selected as the causative species of red tide formation or toxin production in Korean waters (Figure 5a), and their seasonal distribution characteristics based on the number of reads through metagenomic analysis was confirmed by region. In Gunsan, the reads of *Gymnodinium* sp. were considerable in March, and *Dinophysis* spp. appeared in June and September. In Pohang, *Gymnodinium* sp. was relatively high in March, and *Cochlodinium* sp. was also detected at a high distribution in December. In Tongyeong, the abundance of *Cochlodinium* sp. was especially high in September, when

a red tide from this species was occurring. In Seongsan, the appearance of *Gymnodinium* sp. in March and September was revealed by microscopic observation (Figure 5b). Based on these findings, the seasonal distribution of red tide-causing species, which was not confirmed by microscopic observation, was confirmed using metagenomic analysis. in March and September was revealed by microscopic observation (Figure 5b). Based on these findings, the seasonal distribution of red tide-causing species, which was not confirmed by microscopic observation, was confirmed using metagenomic analysis.

*Microorganisms* **2022**, *10*, x FOR PEER REVIEW 11 of 15

**Figure 5.** Seasonal distribution of red-tide-causing species through metagenome analysis. Photo of red-tide-causing species (*Cochlodinium* sp., *Alexandrium* sp., *Dinophysis* sp., *Gymnodinium* sp.) taken under a light microscope (**a**), seasonal changes in red-tide-causing species (**b**). **Figure 5.** Seasonal distribution of red-tide-causing species through metagenome analysis. Photo of red-tide-causing species (*Cochlodinium* sp., *Alexandrium* sp., *Dinophysis* sp., *Gymnodinium* sp.) taken under a light microscope (**a**), seasonal changes in red-tide-causing species (**b**).

#### **4. Discussion 4. Discussion**

Approximately 300 dinoflagellate species are known to cause red tides and produce toxins worldwide, and these harmful events are increasing with changes in human activities and the environment [4]. Toxic dinoflagellate blooms frequently occur in the southern coastal waters of Korea, where many cage fish farms are located. As shown in Table 1, *Cochlodinium* sp. were dominant at Tongyeong in September according to NGS, which corresponds to the cell abundance counted by microscopic observation*.* In June, the NGS result that *Prorocentrum* sp. were mainly observed at Tongyeong was similar to the occurrence detected by microscopy analysis at this location. In addition, Karlodinium sp., which produces Karlotoxin and induces hemolytic and cytotoxic activity associated with fish mortality, appeared in our NGS results [27]. Approximately 300 dinoflagellate species are known to cause red tides and produce toxins worldwide, and these harmful events are increasing with changes in human activities and the environment [4]. Toxic dinoflagellate blooms frequently occur in the southern coastal waters of Korea, where many cage fish farms are located. As shown in Table 1, *Cochlodinium* sp. were dominant at Tongyeong in September according to NGS, which corresponds to the cell abundance counted by microscopic observation. In June, the NGS result that *Prorocentrum* sp. were mainly observed at Tongyeong was similar to the occurrence detected by microscopy analysis at this location. In addition, *Karlodinium* sp., which produces Karlotoxin and induces hemolytic and cytotoxic activity associated with fish mortality, appeared in our NGS results [27].

In a situation where the morphological analysis method is the dominant method for diagnosing harmful dinoflagellates off Korean coasts, diagnosis using molecular biology

In a situation where the morphological analysis method is the dominant method for diagnosing harmful dinoflagellates off Korean coasts, diagnosis using molecular biology is considered to be a more objective number, and the development of technology through this method can lead to the development of new monitoring techniques [28]. Moreover, if NGS technology has been developed and applied to the monitoring of marine organisms, it is possible to simultaneously analyze a large amount of mixed samples and save the effort and time of long-term monitoring and research analysis [29–31].

Monitoring of marine microalgae using NGS has been used by many researchers because of its various advantages [11]. Metagenomic analysis using NGS has revealed a significant number of phytoplankton taxa previously missed by microscopy in recent efforts to sequence marine microorganisms [32]. Our study also revealed a significant number of dinoflagellate communities that could not be distinguished microscopically. The genetic analysis method used in this study, especially high-throughput sequencing, has shown effectiveness in the study of phytoplankton diversity and ecology, and it is considered that it can potentially replace the microscopic identification and population quantification methods currently used.

Light microscopy, which has been used for morphological classification and population evaluation, requires an extensive amount of consideration. Underestimation of phytoplankton, including dinoflagellates, in microscopic samples results in cell loss of taxa during preservation, storage, and handling, preferentially after treatment of samples with fixing fluid. Further, when counting cells, a sedimentation chamber is commonly used, which means that smaller cells that do not sink sufficiently are less counted or missed [33]. Moreover, identification of small dinoflagellates using microscopy is not easy when their cell size is under 20 µm with similar morphologies when fixed with Lugol's solution [34]. We found that a significant number of dinoflagellate species were confirmed by metagenomic analysis compared to that by microscopic analysis. The small dinoflagellate cells which were classified as 'small naked dinoflagellate' were positively identified as species belonging to the genera *Amphidiniopsis*, *Biecheleria*, *Gymnodinium*, *Gyrodiniellum*, *Paragymnodinium*, *Pelagodinium*, and *Symbidinium*, while 'small thecated dinoflagellate' included *Apicoporus*, *Azadinium*, *Crypthecodinium*, *Durinskia*, *Heterocapsa*, and *Pfiesteria*. In particular, the sand-dwelling dinoflagellate *Bysmatrum arenicola*, which is easily confused with *Scrippsiella* [35] in microscopic analysis, was found in the metagenomic analysis in June at Seongsan (Figure 3d). This suggests that the metagenomic analysis was more extensive.

Although the NGS technique showed a high resolution for species identification compared to that with conventional microscopic analysis, further studies are required for development of an understanding of the spatial and seasonal dynamics of the dinoflagellate community using NGS-based metagenomics. Thus far, molecular markers based on ribosomal DNA have usually been used to identify the species, even among relatives [36]. However, this approach is limited by interspecific divergence, while it is difficult to distinguish intraspecific variation. As the reference database of dinoflagellates via the NGS method in this study was established based on 18S rDNA sequences, the relative proportions of some dinoflagellates in field samples could be misidentified in the presence of other dinoflagellates which were similar. Large subunit (LSU) rDNA sequences of *Prorocentrum* species containing *P. rhathymum*, *P. mexicanum*, and *P.* cf. *rhathymum*, which are toxic, were closer to the relatives, showing 0.1–0.9% dissimilarity, and small subunit rDNA (SSU) sequences of most of these are nearly identical [37]. Edvardsen et al. [38] reported that SSU rDNA sequences among *Dinophysis acuminata*, *Dinophysis acuta*, and *Dinophysis norvegica* show approximately 0.3% distance, and differences of LSU rDNA sequences among these species show 0.4–1.6% distance. Moreover, species whose sequences are not available in the GenBank are hardly detected despite their potential presence in the sample analyzed by the NGS technique because of the absence of deposited sequences. To distinguish intraspecific similarity of the above-mentioned species, establishment of a reference database via the NGS technique based on biomarkers such as cytochrome c oxidase I (COX1) and the

cytochrome b (COB) gene which allows for the unambiguous identification of the species should be developed.

Metagenomic analysis of marine biodiversity and abundance based on NGS will provide precise indicators for understanding biological patterns and characteristics of species in different habitats. Given the lack of molecular reference library databases, it is necessary to collect vast amounts of sequence information targeting biomarkers such as SSU, LSU, COX1, and COB genes. However, in this study, we established a reference database of dinoflagellates that occur in the coastal waters of Korea based on SSU rDNA sequences using the NGS technique and analyzed field samples in the presence of this NGS reference database library. We expect that the newly established reference database via the NGS will provide a better understanding of the seasonal dynamics of toxic dinoflagellates, as well as a complementary approach to conventional microscopic analysis for monitoring dinoflagellate community compositions.

#### **5. Conclusions**

This study integrated analyses of high-resolution dinoflagellate community composition and distribution in South Korea. Altogether, the results presented here reveal a complex dinoflagellate community pattern. The NGS-based (18S rRNA amplicon) metagenomics were able to detect dinoflagellates with low abundance, and allow continuous monitoring of the phytoplankton community in environmental samples even though numerous DNA samples were simultaneously collected compared to the conventional microscopic analysis. Our analysis suggested that NGS-based characterization of the 18S rRNA gene holds great promise as a tool for phytoplankton monitoring, as it allows for simultaneous regional cluster analysis monitoring in a high-throughput, reproducible, and cost-effective manner.

In today's world, which requires advances in environmental monitoring due to largescale blooming of toxic algae and international regulations regarding their toxic substances, this study provides a technique for the rapid evaluation of environmental samples for existing taxa of major dinoflagellates and potentially harmful/invasive species. In addition, the extension of the reference database presented in this study and addition of the species list can further expand the taxonomic scope so it can be applied to real-time monitoring of temporal dynamics and species diversity problems of harmful algal blooms in a wide range of waters.

**Supplementary Materials:** The following supporting information can be downloaded at: https://www. mdpi.com/article/10.3390/microorganisms10071459/s1, Figure S1: Workflow chart for metagenome analysis, Using CLC Genomics Workbench program. Supplementary Table S1.

**Author Contributions:** Conceptualization, J.H., H.W.K. and. J.P.; field investigation, S.J.M., J.-H.H., and E.S.L.; software, J.H.; validation and formal analysis, J.H. and J.-H.H.; data curation, J.P; writing original draft preparation, J.H.; writing—review and editing, J.H. and J.P. All authors have read and agreed to the published version of the manuscript.

**Funding:** This study was supported by the project 'Development of aquaculture technology for pomfrets as an endemic fish species on West Sea' (R2022005) of National Institute of Fisheries Science (NIFS), Incheon, South Korea, and supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT)(NRF-2021R1A2C1005943).

**Data Availability Statement:** Not applicable.

**Acknowledgments:** We thank E.J.K. for field sampling. We thank the reviewers for their comments.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


## *Article* **A Strategy for Gene Knockdown in Dinoflagellates**

**Miranda Judd \* and Allen R. Place**

Institute of Marine and Environmental Technologies, University of Maryland Center for Environmental Science, 701 E Pratt St., Baltimore, MD 21022, USA; place@umces.edu

**\*** Correspondence: mjudd@umces.edu; Tel.: +1-410-234-8828

**Abstract:** Dinoflagellates are unicellular protists that display unusual nuclear features such as large genomes, condensed chromosomes and multiple gene copies organized as tandem gene arrays. Genetic regulation is believed to be controlled at the translational rather than transcriptional level. An important player in this process is initiation factor eIF4E which binds the 7-methylguanosine cap structure (m7G) at the 50 -end of mRNA. Transcriptome analysis of eleven dinoflagellate species has established that each species encodes between eight to fifteen eIF4E family members. Determining the role of eIF4E family members in gene expression requires a method of knocking down their expression. In other eukaryotes this can be accomplished using translational blocking morpholinos that bind to complementary strands of RNA, therefore inhibiting the mRNA processing. Previously, unmodified morpholinos lacked the ability to pass through cell membranes, however peptide-based reagents have been used to deliver substances into the cytosol of cells by an endocytosis-mediated process without damaging the cell membrane. We have successfully delivered fluorescently-tagged morpholinos to the cytosol of *Amphidinium carterae* by using a specific cell penetrating peptide with the goal to target an eIF4e-1a sequence to inhibit translation. Specific eIF4e knockdown success (up to 42%) has been characterized via microscopy and western blot analysis.

**Keywords:** dinoflagellate; knockdown; morpholino; translation

**Citation:** Judd, M.; Place, A.R. A Strategy for Gene Knockdown in Dinoflagellates. *Microorganisms* **2022**, *10*, 1131. https://doi.org/10.3390/ microorganisms10061131

Academic Editors: Shauna Murray and Thomas Mock

Received: 1 April 2022 Accepted: 27 May 2022 Published: 31 May 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

### **1. Introduction**

Dinoflagellates are single-celled eukaryotes and members of the Alveolate lineage [1]. Dinoflagellates exhibit extremely diverse trophic strategies, including predation, photoautotrophy, mixotrophy, and intracellular parasitism [2,3]. Most cultured dinoflagellate species are photosynthetic, making them key marine primary producers. They are wellknown for bloom formation in coastal waters, making toxins that bioaccumulate in the food chain, producing bioluminescence, and as coral symbionts [4–6].

Climate-change has caused a warming of the Earth's oceans, benefitting the formation of harmful algal blooms [7]. Of the algal species that have been reported as producing marine harmful blooms, 75% are dinoflagellates [8–11]. Accumulation of dinoflagellates in coastal waters has begun to increase the presence of red tides, bringing with it fish mass mortality and marine toxin-derived disease in humans [12]. Increasing water temperatures provide optimum growth conditions for many dinoflagellates, allowing for increased toxic effects on their environment [13,14]. Globally, previous research has confirmed the mechanism and structure of some of these toxins [15–20].

A hallmark of these toxic blooms can be traced to the production of complex secondary metabolites. Some of these toxins are thought to assist in prey capture through the formation of a nonspecific pore upon complexation with prey's sterol membrane components [16,21]. Unfortunately, research into the biosynthesis of these dinoflagellate toxins is sorely lacking. This is in large part due to dinoflagellates having unusual cell biology [22]. Their genomes are larger than typical protists, with about 1.2–112 <sup>×</sup> <sup>10</sup><sup>9</sup> base pairs of DNA per haploid genome [10], whereas other protist genomes range in the millibases [23,24]. Dinoflagellate chromosomes are condensed into liquid crystalline states throughout the cell cycle and lack

nucleosomes, instead using histone-like proteins (HLPs) that are more similar to bacterial DNA binding proteins. Many dinoflagellate genes are organized in multiple copies as tandem repeats, some of which may be present in up to ~10<sup>5</sup> copies. Increasingly transcriptomic data has shown that dinoflagellates express numerous genes, yet about 50% have no match to known sequences [25]. The function of these sequences, as well as the effects of identified sequences, still need to be established through functional genomic studies.

Control of post-transcriptional regulation in dinoflagellates is currently enigmatic, with mRNA levels showing no correlation to protein production. Because dinoflagellates are believed to regulate at the translational level, rather than during transcription, translation factors are of great interest when understanding dinoflagellate metabolism. In most known eukaryotic translation systems, eIF4Es function as a rate-limiting step toward protein synthesis [26,27]. eIF4E is part of an extended gene family found exclusively in eukaryotes. This translation factor binds to the mRNA cap to recruit the ribosome for translation initiation. In most studied eukaryotic systems (excluding plants), the eIF4E-1 family member is expressed ubiquitously in all cell types from a single copy, such as in *Homo sapiens* or *Saccharomyces cerevisiae* [28–32]. Early studies speculated that eukaryotic systems contain a single gene that encodes eIF4E [33], but since then sequencing projects have revealed that many organisms contain multiple genes encoding proteins that have sequence similarity to the recognized eIF4E [31,34–36]. In the case of dinoflagellates, genes regularly appear in multiple copies, with eIF4E being no exception [37,38]. These gene copies commonly appear as slightly different variants with distinctive degrees of diversity. Prior transcriptome analysis of eleven dinoflagellate species has established that each species encodes between eight to fifteen eIF4E family members, a number surpassing that found in any other eukaryotes, including other alveolates [22].

Core dinoflagellate eIF4E translation factors are divided into 3 clades (1, 2, and 3), along with 3 subclades within each (a, b, c); with a total of 9 members. Our previous work has shown that these eIF4E family members display divergences at critical amino acids, suggesting the family members are functionally distinct [22,39]. Of these 3 major clades, eIF4E-1 stands out as the most duplicated, and with the lowest number of substitutions. Based on the expression levels of the subclades, our lab has theorized that eIF4E-1a is likely the primary translation initiation factor.

Although expression of subclade eIF4E-1a is highest of all the family members, dinoflagellates still generate a greater diversity and degree of eIF4E duplications than seen in other eukaryotes [22]. Understanding their various roles will bring us closer to understanding how dinoflagellates adapt to their environment, giving insight into harmful algal bloom formations, as well as their production of complex secondary metabolites and toxin biosynthesis. Generally, eIF4E family members are known to have different roles in metazoan gene expression [40]. Similarly, we predict that dinoflagellate eIF4Es will have distinct functions, allowing for an increased dependence on the translational control of gene expression [22]. Determining the role of the eIF4E family members requires a method of knocking down their expression. The unusual cell biology of dinoflagellates makes common gene knockout strategies impractical, limiting the amount of genetic research that can be applied. This is because gene knockouts require a "deletion" of all operable gene-copies, which can be difficult to obtain when many copies with slight variations exist [38]. In this case, gene knockdown strategies are the most feasible, as they target the functional transcripts produced by the gene copies, which in many cases remain less diverse [39,41].

In particular, we are pursuing the use of an antisense-based knockdown approach in order to study how a decrease in target gene expression effects dinoflagellate metabolism. In prior research, the introduction of antisense-oligomers to dinoflagellate cells has been hampered by their thick, cellulosic cell wall [42,43]. Other studies have bypassed this obstacle by preparing spheroplasts, cells with a completely or partially removed cellwall, beforehand [42–44]. Spheroplast production is done by incubating cells on plates in a polyethylene glycol (PEG) solution, which promotes fusion of the vesicles and cell

membrane, and ultimately a decrease in total cellulose. So far only two studies have been successful in achieving gene knockdown with this spheroplast procedure; targeting a condensin subunit and targeting a cellulose synthase [42–44]. Once introduced, the antisense-oligo was able to bind to cytoplasmic mRNA and knockdown expression of the target gene. Although gene expression could be quantified in this way, it appears that some physiological effects were hidden by the effects of PEG on the cell wall, which causes the cells to lose rigidity. Also, the need for cell plating, rather than cell culturing, immensely limits the species of dinoflagellates that can be studied since many will not grow outside of a liquid medium.

There has also been evidence of RNA interference (RNAi) machinery within dinoflagellates, a naturally occurring mechanism for gene silencing through various methods such as RNA degradation, transcriptional repression and translation inhibition [25,45,46]. One study observed the effects of RNAi silencing tool on the proton-pump rhodopsin and CO2 fixing enzyme Rubisco encoding genes in dinoflagellates by introducing small interfering RNAs (siRNAs) to dinoflagellate cultures via immersion. Results showed success in gene suppression within the two dinoflagellate species studied, *Prorocentrum donghaiense* and *Karlodinium veneficum* [25]. This decrease in gene expression was observed with a decrease on overall growth rate for both species as well, compared to the control green fluorescent protein (GFP) labelled siRNA. This knockdown method in dinoflagellates was initially challenging due to the large copy number of target DNA and permanently-condensed chromatin, but recent research has shown that there is a strong possibility that knockdown procedures can be successful.

Recently our team has begun to develop a system for introducing antisense morpholinos into dinoflagellate cells without the use of PEG to warp the cell wall, or the use of RNAi. Instead, we are using a novel delivery peptide that delivers substances via an endocytosis-mediated process that avoids damaging the plasma membrane of the cell [47]. Not only this, but the knockdown process can be done completely by immersion. The peptide and antisense morpholino are added directly to the culture to stimulate endocytosis and morpholino uptake. To test this system on dinoflagellates, we used the manufacturer recommended concentrations of delivery-peptide and antisense-morpholino on a dense culture of *Amphidinium carterae*, a known algal bloom species. The antisense morpholino is targeted to what we believed to be the main dinoflagellate translation factor, eIF4E-1a [22,27,40,48]. Our preliminary data has shown that unlike with PEG addition, the delivery peptide does not cause the cell population to drastically decrease.

#### **2. Materials and Methods**

#### *2.1. Cell Culturing*

*Amphidinium carterae (Hulbert)* strain CCMP1314 was grown in ESAW artificial marine media with a salinity of 32 ppt supplemented with f/2 nutrients without silicates at 25 ◦C [49]. The medium was buffered with 1mM HEPES (pH 8.0). Since bacterized cultures have shown to affect analyses of translation rate in *A. carterae*, the cultures were maintained axenically with an antibiotic solution of kanamycin (50 µg/mL), carbenicillin (100 µg/mL), and spectinomycin (50 µg/mL) [50]. The cultures were grown under 100 µmol/m<sup>2</sup> s − light. Delivery of the morpholinos (described below) requires constant swirling to keep the reagents in solution, therefore the cultures were also placed on an orbital shaker at 60 rpm [47]. The cultures were allowed to acclimate to the swirling for a week before knockdown reagents were added.

#### *2.2. Morpholino Customization and Delivery*

The sequence of the initiation factor eIF4E-1a was found by our lab previously [22]. Morpholino antisense oligonucleotides (MOs) are nucleic acid analogues in which DNA bases are bound to a non-charged backbone (morpholine rings linked by phosphorodiamidate bonds) [51]. For our purposes, a translation-blocking MO was created that covered the eIF4E-1a translational start site (50 -TCATTGAAGCTCAAACAAGCCATTG-30 ). Specificity

for the intended target sites was verified by BLAST analysis against the *Amphidinium carterae* transcriptome. MOs were purchased from GeneTools (Philomath, OR, USA) and modified with a red-emitting fluorescent 30 Lissamine addition and then used at a concentration of 1 µM and 10 µM. Standard control MOs with the Lissamine addition were ordered as well (5'-CCTCTTACCTCAGTTACAATTTATA-3').

MOs were delivered using Endo-Porter reagent (GeneTools, Philomath, OR, USA) at a concentration of 4 µM. Cultures of *A. carterae* seeded in 12-well plates were treated with either Endo-Porter, MO, or both. Three biological replicates of each treatment were performed.

#### *2.3. Cell Counts and Fluorescence Quantification*

Following the addition of 4 µM Endo-Porter and 1 µM or 10 µM MO, the viability of cultures of *A. carterae* were observed over a 96-h period by measuring their autofluorescence via flow cytometry. Measurements of the cultures within the first hour of treatment were labelled as Hour 0. Cell counts for each condition were determined on a BD C6 Accuri Flow Cytometer (BD Biosciences, San Jose, CA, USA), equipped with laser excitation at 488 and 640 nm and emission at 533/30, 585/40, and >670 nm. The FSC-A and fluorescence channels were used to select for live cells; from this selection, cells with Lissamine emission signals were detected (585 nm). Cells were grouped into low and high Lissamine fluorescence, with high fluorescence showing an intensity minimum of 10<sup>4</sup> relative fluorescence units (RFUs).

#### *2.4. Cell Imaging*

Images of A. carterae cells with and without MO treatment were taken on a STELLARIS confocal microscope (Leica Microsystems, South San Francisco, CA, USA), equipped with 405, 552 and 638 nm lasers, and PMT and HyD detectors collecting emission within 590–600 nm and 680–720 nm, respectively.

#### *2.5. Quantification of Protein Expression*

Protein expression was quantified by Western blotting. For the initial Western blot analyses done for the 1 µM concentration of morpholinos, the cell density of the cultures were quantified via flow-cytometry to create equal volume pellets containing ~75,000 cells for each condition 48 h post-treatment. A NuPAGE 4 to 12% Bis-Tris 1.5 mm Mini Protein Gels was used. Cell pellets were prepared for electrophoresis in 3× sample buffer (Blue Loading Buffer Pack, New England Biolabs, Ipswich, MA, USA), heated to 96 ◦C for 10 min and centrifuged for 2 min at 10,000× *g*; from the 15 µL total, 10 µL of each extract was electrophoresed at 165 V until the dye front reached the bottom of the gel. The gel was transferred to a membrane with the Trans Blot Turbo Transfer system. Protein loading and relative expression levels were verified by probing the same blot with anti-eIF4E-1a mouse monoclonal and HRP-conjugated anti-mouse IgG. The labeled bands, as well as band intensities, were detected with ImageLab software.

For the subsequent experiment testing the higher 10 µM concentration of morpholinos, equal quantities of whole-cell lysates containing 100,000 cells were prepared from each of the triplicate sample cultures at 48 and 96 h. Samples were once again split equally over two NuPAGE 4–12% Bis-Tris Gels and run for 50 min at 165 V. Both gels were transferred to a membrane with the Trans Blot Turbo Transfer system. To control for cell loading error, total protein was detected on one membrane using No-Stain Protein labelling Reagent (Invitrogen, Waltham, MA, USA). Summation of total protein in each lane was found using ImageJ. Protein loading and relative expression levels were again verified by probing the second blot with anti-eIF4E-1a mouse monoclonal and HRP-conjugated anti-mouse IgG. The labeled bands were detected with ImageLab software, as well as band intensities. The relative production of eIF4E-1a was analyzed by dividing the eIF4E-1a volume by the summation of total protein.

#### *2.6. Statistical Considerations 2.6. Statistical Considerations*

volume by the summation of total protein.

All conditions were performed in triplicate. Pairwise sample comparisons within timepoints were analyzed in R-Studio with *t*-tests using pooled standard deviations [52]. A *p*-value of <0.05 was considered statistically significant. All conditions were performed in triplicate. Pairwise sample comparisons within timepoints were analyzed in R-Studio with *t*-tests using pooled standard deviations [52]. A *p*-value of <0.05 was considered statistically significant.

intensities. The relative production of eIF4E-1a was analyzed by dividing the eIF4E-1a

#### **3. Results 3. Results**

#### *3.1. Cell Viability 3.1. Cell Viability*

Immediately post-treatment, the cultures displayed a suspension of growth, but recovery was observed at 24 h (Figure 1). The cultures containing the antisense morpholino also showed significant differences in growth compared to the control immediately after treatment addition and at both 24 and 48 h. Immediately post-treatment, the cultures displayed a suspension of growth, but recovery was observed at 24 h (Figure 1). The cultures containing the antisense morpholino also showed significant differences in growth compared to the control immediately after treatment addition and at both 24 and 48 h.

*Microorganisms* **2022**, *10*, x FOR PEER REVIEW 5 of 12

**Figure 1.** Flow cytometer population counts from *Amphidinium carterae* cultures (N = 3). Conditions included the control, Endo-Porter only, antisense morpholino only, and the Endo-Porter and antisense morpholino combined. Statistical analysis was done using *t*-tests with pooled standard deviations. \* Significantly different from 'Control' based on a *p*-value of < 0.05. **Figure 1.** Flow cytometer population counts from *Amphidinium carterae* cultures (N = 3). Conditions included the control, Endo-Porter only, antisense morpholino only, and the Endo-Porter and antisense morpholino combined. Statistical analysis was done using *t*-tests with pooled standard deviations. \* Significantly different from 'Control' based on a *p*-value of < 0.05.

#### *3.2. Uptake of the Morpholino 3.2. Uptake of the Morpholino*

The intensity of the Lissamine signal within cells was measured by flow cytometry (Figure 2). Peak Lissamine fluorescence was observed at 48 h, with about 13% of the population uptaking a high amount of fluorescently-tagged morpholinos [47], and waned after this timepoint. The mean Lissamine fluorescence per cell with the Endo-Porter delivery system plus the MO was over 50X greater than that of the control, and the median was over 1.5X greater, showing significant uptake of the MO in *A. carterae*, as well as a large positively-skewed distribution of uptake efficiency (Supplementary Table S1). Cells were also able to uptake morpholino without Endo Porter, but to a significantly lower degree, with no correlation to experimental duration. The intensity of the Lissamine signal within cells was measured by flow cytometry (Figure 2). Peak Lissamine fluorescence was observed at 48 h, with about 13% of the population uptaking a high amount of fluorescently-tagged morpholinos [47], and waned after this timepoint. The mean Lissamine fluorescence per cell with the Endo-Porter delivery system plus the MO was over 50X greater than that of the control, and the median was over 1.5X greater, showing significant uptake of the MO in *A. carterae*, as well as a large positively-skewed distribution of uptake efficiency (Supplementary Table S1). Cells were also able to uptake morpholino without Endo Porter, but to a significantly lower degree, with no correlation to experimental duration.

**Figure 2.** Percent of A.carterae populations with high Lissamine uptake after 48 h (cutoff 104 RFUs). Conditions included the control, Endo Porter only, antisense morpholino only, and the Endo Porter and antisense morpholino combined (N = 3). Statistical analysis was done using *t*-tests with pooled standard deviations. \* Significantly different from 'Control' based on a *p*-value of < 0.05. † Significant difference between "Antisense Morpholino Only" and "Endo Porter & Anti-sense Morpholino" **Figure 2.** Percent of A.carterae populations with high Lissamine uptake after 48 h (cutoff 104 RFUs). Conditions included the control, Endo Porter only, antisense morpholino only, and the Endo Porter and antisense morpholino combined (N = 3). Statistical analysis was done using *t*-tests with pooled standard deviations. \* Significantly different from 'Control' based on a *p*-value of < 0.05. † Significant difference between "Antisense Morpholino Only" and "Endo Porter & Anti-sense Morpholino" based on a *p*-value of < 0.05. **Figure 2.** Percent of A.carterae populations with high Lissamine uptake after 48 h (cutoff 104 RFUs). Conditions included the control, Endo Porter only, antisense morpholino only, and the Endo Porter and antisense morpholino combined (N = 3). Statistical analysis was done using *t*-tests with pooled standard deviations. \* Significantly different from 'Control' based on a *p*-value of < 0.05. † Significant difference between "Antisense Morpholino Only" and "Endo Porter & Anti-sense Morpholino" based on a *p*-value of < 0.05.

based on a *p*-value of < 0.05. Images of cells were also captured by confocal microscopy (Figure 3). Lissamine fluorescence was detected within the range of 590–600 nm (peak 593 nm), and autofluorescence was measured between 700–720 nm. Confocal images display a diffuse pseudo-blue Images of cells were also captured by confocal microscopy (Figure 3). Lissamine fluorescence was detected within the range of 590–600 nm (peak 593 nm), and autofluorescence was measured between 700–720 nm. Confocal images display a diffuse pseudo-blue coloring within the cytoplasm of the dinoflagellate cells, as well as in a large area near the nucleus, indicating both diffuse and localized morpholino presence (Figure 3). Images of cells were also captured by confocal microscopy (Figure 3). Lissamine fluorescence was detected within the range of 590–600 nm (peak 593 nm), and autofluorescence was measured between 700–720 nm. Confocal images display a diffuse pseudo-blue coloring within the cytoplasm of the dinoflagellate cells, as well as in a large area near the nucleus, indicating both diffuse and localized morpholino presence (Figure 3).

**Figure 3.** Confocal images of *Amphidinium carterae*. A control cell with no antisense morpholino introduced is on the far left. Red boxes encompass images of cells from a culture with Endo Porter and antisense-morpholino, fluorescently tagged with Lissamine (593 nm, pseudo-blue) at 24-, 48 and 96-h post-treatment. The grey areas are the auto-fluorescence emitted within 720–759 nm. **Figure 3.** Confocal images of *Amphidinium carterae*. A control cell with no antisense morpholino introduced is on the far left. Red boxes encompass images of cells from a culture with Endo Porter and antisense-morpholino, fluorescently tagged with Lissamine (593 nm, pseudo-blue) at 24-, 48- and 96-h post-treatment. The grey areas are the auto-fluorescence emitted within 720–759 nm.

From the initial Western Blot analysis, we found a statistically significant decrease in the expression of eIF4E-1a of about 30% compared to the control in cultures containing

*3.3. Initial Western Blot Analysis* 

and 96-h post-treatment. The grey areas are the auto-fluorescence emitted within 720–759 nm.

difference was not statistically significant.

*3.3. Initial Western Blot Analysis* 

difference was not statistically significant.

**Figure 3.** Confocal images of *Amphidinium carterae*. A control cell with no antisense morpholino in-

and antisense-morpholino, fluorescently tagged with Lissamine (593 nm, pseudo-blue) at 24-, 48-

the expression of eIF4E-1a of about 30% compared to the control in cultures containing both the 1 μM morpholino and Endo Porter (Figure 4). Cultures with Endo Porter showed a decrease of 11% in eIF4E-1a expression over cultures without Endo Porter, although the

From the initial Western Blot analysis, we found a statistically significant decrease in

#### *3.3. Initial Western Blot Analysis*

From the initial Western Blot analysis, we found a statistically significant decrease in the expression of eIF4E-1a of about 30% compared to the control in cultures containing both the 1 µM morpholino and Endo Porter (Figure 4). Cultures with Endo Porter showed a decrease of 11% in eIF4E-1a expression over cultures without Endo Porter, although the difference was not statistically significant. *Microorganisms* **2022**, *10*, x FOR PEER REVIEW 7 of 12

**Figure 4.** Western blot analyses for eIF4E-1a concentrations within control and treated cells at 48 h post-MO [1μM] addition (N = 3). Protein loading and relative expression levels was verified by probing with anti-eIF4E-1a mouse monoclonal and HRP-conjugated an-ti-mouse IgG. eIF4E-1a Protein area volumes were reduced by 30% after introduction of custom translation blocking morpholino and Endo Porter after 48 h. Statistical analy-sis was done using *t*-tests with pooled standard deviations. \* Significantly different from "Control" based on a *p*-value of < 0.05. No significant difference found between "Antisense Morpholino Only" and "Antisense Morpholino + Endo Porter" based on a *p*-value of < 0.05. **Figure 4.** Western blot analyses for eIF4E-1a concentrations within control and treated cells at 48 h post-MO [1 µM] addition (N = 3). Protein loading and relative expression levels was verified by probing with anti-eIF4E-1a mouse monoclonal and HRP-conjugated an-ti-mouse IgG. eIF4E-1a Protein area volumes were reduced by 30% after introduction of custom translation blocking morpholino and Endo Porter after 48 h. Statistical analy-sis was done using *t*-tests with pooled standard deviations. \* Significantly different from "Control" based on a *p*-value of < 0.05. No significant difference found between "Antisense Morpholino Only" and "Antisense Morpholino + Endo Porter" based on a *p*-value of < 0.05.

#### *3.4. Increase in MO Concentration 3.4. Increase in MO Concentration*

Once the concentration of morpholinos was increased from 1 μM to 10 μM, we found that with the custom eIF4E-1a target morpholino and Endo Porter there was a decrease in population eIF4E-1a expression at 48 h compared to the control of about 42% (Figure 5). This was an increase from the 30% found when using only 1 μM of MO, constituting a total 11.8% decrease with the increased MO concentration (Table 1). The cultures containing eIF4E-1a-target morpholino only also showed reduced target-protein production compared to the control of 21%. When compared to the culture containing only the standard non-target morpholino and Endo Porter, both of the cultures containing the eIF4E-1a-target morpholino with and without the Endo Porter appeared to exhibit a significant decrease of 47% and 29%, respectively (Figure 5). No significant difference was found be-Once the concentration of morpholinos was increased from 1 µM to 10 µM, we found that with the custom eIF4E-1a target morpholino and Endo Porter there was a decrease in population eIF4E-1a expression at 48 h compared to the control of about 42% (Figure 5). This was an increase from the 30% found when using only 1 µM of MO, constituting a total 11.8% decrease with the increased MO concentration (Table 1). The cultures containing eIF4E-1a-target morpholino only also showed reduced target-protein production compared to the control of 21%. When compared to the culture containing only the standard nontarget morpholino and Endo Porter, both of the cultures containing the eIF4E-1a-target morpholino with and without the Endo Porter appeared to exhibit a significant decrease

tween the cultures containing eIF4E-1a-target morpholino with or without the Endo Porter, although the cultures with the Endo Porter showed an average decrease of 25% com-

**Table 1.** Volume of eIF4E-1a produced compared to the control. Volumes of eIF4E-1a were quantified via western blot and compared to the density of a whole protein stain per sample (N = 3). Percent of eIF4E-1a for each sample were compared to the control sample to describe relative eIF4E-1a

production. Statistical analysis was done using *t*-tests with pooled standard deviations.

Antisense Morpholino Only 78.8% \* 78.3% 0.5%

*Microorganisms* **2022**, *10*, x FOR PEER REVIEW 8 of 12

**Morpholino Concentration 1 μM 10 μM** 

of 47% and 29%, respectively (Figure 5). No significant difference was found between the cultures containing eIF4E-1a-target morpholino with or without the Endo Porter, although the cultures with the Endo Porter showed an average decrease of 25% compared to the cultures without. Antisense Morpholino + Endo Porter 70.1% \* 58.3% \* 11.8% **\*** Significant difference when compared to the control based on a *p*-value of **<** 0.05.

 **Relative eIF4E-1a Production Percent Change** 

**Figure 5.** Western blot analyses for eIF4E-1a concentrations within control and treated cells at 48 post-MO [10 μM] addition (N = 3). Protein loading and relative expression levels was veri-fied by probing with anti-eIF4E-1a mouse monoclonal and HRP-conjugated anti-mouse IgG and compared to total protein volumes. Relative eIF4E-1a levels are lower in Am-phidinium population after 48 h of being subjected to custom translation-blocking morpholinos. Statistical analysis was done using *t*-tests with pooled standard deviations. \* Significantly different from 'Control' based on a *p*-value of < 0.05. † Significantly different from "Standard Morpholino + Endo Porter" based on a *p*-value of < 0.05. No significant difference found between "Antisense Morpholino Only" and **Figure 5.** Western blot analyses for eIF4E-1a concentrations within control and treated cells at 48 post-MO [10 µM] addition (N = 3). Protein loading and relative expression levels was veri-fied by probing with anti-eIF4E-1a mouse monoclonal and HRP-conjugated anti-mouse IgG and compared to total protein volumes. Relative eIF4E-1a levels are lower in Am-phidinium population after 48 h of being subjected to custom translation-blocking morpholinos. Statistical analysis was done using *t*-tests with pooled standard deviations. \* Significantly different from 'Control' based on a *p*-value of < 0.05. † Significantly different from "Standard Morpholino + Endo Porter" based on a *p*-value of < 0.05. No significant difference found between "Antisense Morpholino Only" and "Antisense Morpholino + Endo Porter" based on a *p*-value of < 0.05.

"Antisense Morpholino + Endo Porter" based on a *p*-value of < 0.05. Interestingly, we also found that expression levels became relatively similar after 96 h, showing the temporary effects of the morpholino (Supplemental Figure S1). **Table 1.** Volume of eIF4E-1a produced compared to the control. Volumes of eIF4E-1a were quantified via western blot and compared to the density of a whole protein stain per sample (N = 3). Percent of eIF4E-1a for each sample were compared to the control sample to describe relative eIF4E-1a production. Statistical analysis was done using *t*-tests with pooled standard deviations.


cantly decreasing viability. This method delivers substances into the cytosol of cells by an **\*** Significant difference when compared to the control based on a *p*-value of **<** 0.05.

The above data was collected using the manufacturer recommended concentrations

of reagents, but as we look further into these systems, we hope to optimize the outcome by adjusting protocol values. Currently, with 1 μM of MO added, we have been able to

endocytosis-mediated process that avoids damaging the plasma membrane of the cell [47,53]. With this new system, exploration into the key players of different metabolic path-Interestingly, we also found that expression levels became relatively similar after 96 h, showing the temporary effects of the morpholino (Supplemental Figure S1).

ways may be closer than expected for dinoflagellates.

#### **4. Discussion**

Here we have shown that novel delivery peptide technology has allowed for successful introduction of custom antisense morpholinos into *A. carterae* cells without significantly decreasing viability. This method delivers substances into the cytosol of cells by an endocytosis-mediated process that avoids damaging the plasma membrane of the cell [47,53]. With this new system, exploration into the key players of different metabolic pathways may be closer than expected for dinoflagellates.

The above data was collected using the manufacturer recommended concentrations of reagents, but as we look further into these systems, we hope to optimize the outcome by adjusting protocol values. Currently, with 1 µM of MO added, we have been able to successfully introduce a high concentration of the MO into approximately 13% of the population (Figure 2), with a significant decrease in eIF4E-1a protein production. Dose-dependent effects of the MOs have also been observed, with an increase in MO concentration from 1 µM to 10 µM resulting in a decrease in eIF4E-1a from 30% to 42%; an 11.83% decrease in total (Table 1). Optimistically we would like to increase this percentage to a level where protein production is functionally suspended, or be able to separate cells from the population based on their MO uptake.

Interestingly. western blot analyses for both the initial 1 µM and subsequent 10 µM MO experiment showed no statistically significant difference between the cultures with and without Endo-Porter, although the cultures with the Endo-Porter showed consistently lower average eIF4E-1a protein produced (Figures 4 and 5). This would indicate that even without a delivery peptide, *A. carterae* cells are able to uptake the MO. Reasons for MO uptake without a delivery peptide are still unclear. Over a million identified peptide sequences within various dinoflagellates are still of unknown function and origin, making their evolutionary history ambiguous [54]. Dinoflagellates are thought to have undergone multiple organellogensis events, where the genome of endosymbiotic algae becomes a plastid and/or genes from the endosymbiont are transferred to the nucleus [55]. The evolutionary nature of dinoflagellates to accept foreign genes appears to be high [56]. One theory for why the *A. carterae* cells took up the unaided MO could be the possibility that dinoflagellate systems are more open to horizontal gene transfer (HGT) than previously imagined [57]. Recent studies have shown that genomes within dinoflagellates may be more open to foreign contributions from both bacteria and eukaryotes compared to other organisms [57,58]. Further research needs to be conducted to follow how foreign genetic material is received by dinoflagellates, as well as how transcripts are processed.

A wide distribution of MO uptake may account for the discrepancy of measured eIF4E-1a protein production compared to the cellular MO uptake (Supplemental Table S1). Although we suggest about 13% of the cells have a high uptake after addition of 1 µM of MO, the eIF4E-1a production was decreased by about 30% according to the western results (Figure 4). We assume that there is a range of efficiency uptake within the population, so some cells may be producing little or no eIF4E-1a, while others may be producing their average amounts. Once again, changes to the concentrations used or the use of a cell-sorter may be necessary to observe higher gene knockdown efficiency.

MO localization does occur within the *A. carterae* cells, usually in a large area by the nucleus (Figure 3). In previous studies, Endo-Porter sometimes results in unsuccessful endosomal acidification, and therefore no release of the MO into the cytoplasm from their vesicle, but these localization points are usually known to appear as punctate fluorescence throughout the cell [47]. The singular, large area of localization may signify MO aggregation in the nucleolus or an RNA-granule [55,59–62]. Dinoflagellates are known to have very unusual nuclei, specially named the "dinokarya." Among other peculiar features are recently discovered "nuclear tunnels" which extend from the nuclear envelope of *Polykrikos kofoidii*, specifically during mitosis [59]. These nuclear envelope tunnels are also connected with a membranous structure throughout the nucleus known as the "nuclear net". The discovery of these structures adds a new level of complexity to the dinoflagellate nuclear membrane, and may allude to more complex processing for transcription and translation. The unusual nuclear envelope tunnels along with our MO localization could very well be connected. Since dinoflagellates are known to regulate gene expression at the translational level, this cellular organization of mRNA may be a crucial step for gene expression, which needs to be analyzed further.

In theory, this method of gene knockdown is progress towards expanding research into the biosynthetic pathways within dinoflagellate. Currently, this research is sorely lacking, largely due to their complex genomes and unusual cell biology. Translation regulation is now understood as a crucial step in gene expression, far beyond that of transcriptional control [22,48]. In addition, further understanding of the components of the translation machinery is required to understand the expression of specific genes, which make up the dinoflagellate "translational toolkit". As we work to optimize this procedure, other pathways could be targeted as well in other species of dinoflagellates.

One interesting area would be to target the transcript responsible for toxin production. The function of toxins produced by dinoflagellates has been theorized but is still unknown. Theories include allelopathy, prey-capture, and as a defense [18]. One way to discern the function of the toxins would be to knockdown their expression to see how this affects feeding and swimming behavior. Mixotrophic characteristics of dinoflagellate species known to produce toxins could be monitored through photosynthesis and respiration rates, as well as swimming behavior via digital holographic microscopy [63]. This data could produce more evidence to the intended functionality of dinoflagellate toxins.

#### **5. Conclusions**

This primary study provides proof-of-principle for the possibility to specifically downregulate gene expression in dinoflagellates using antisense morpholinos and a novel delivery system. Additional work is necessary to validate and optimize these findings, and to extend them to other biosynthetic pathways. Further work will also be required to investigate the effects of gene knockdown on the various eIF4E family members in order to possibly unveil a translation toolkit used by dinoflagellates to regulate gene expression. Studies involving toxin biosynthesis pathways would also benefit from successful knockdowns in order to identify functionality and key synthesis steps.

**Supplementary Materials:** The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/microorganisms10061131/s1, Figure S1: Western blot analyses for eIF4E-1a concentrations within control and treated cells at 96 h post-MO [10 µM] addition; Table S1: Lissamine uptake in *A. carterae* cells 48 h post treatment.

**Author Contributions:** Conceptualization, review and editing by M.J. and A.R.P.; Methodology by M.J. and A.R.P.; Supervision by A.R.P.; Formal analysis and writing by M.J. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Not applicable.

**Acknowledgments:** I would like to thank the Ernest Williams for his wealth of knowledge and advice, as well as helping me finalize this project. This is contribution #6181 from UMCES and #22-104 from IMET.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**

