Phaeocystis globosa is a cosmopolitan marine phytoplankton that plays important roles in carbon and sulfur biogeochemical cycles (Schoemann et al., 2005). P. globosa forms large blooms and is considered to be a harmful algal bloom species due to its negative economic and ecological impacts. Blooms of P. globosa in temperate regions, especially in the North Sea, have been observed regularly since the 1970s and are associated with nutrient enrichment from rivers (Cadee and Hegeman, 2002). Outbreaks of P. globosa blooms have been reported in tropical and subtropical regions frequently in the past two decades, including coastal waters of China (Qi et al., 2004), Viet Nam (Tang et al., 2004; Doan-Nhu et al., 2010; Smith et al., 2014) and the Arabian Sea (Madhupratap et al., 2000).

A unique feature of Phaeocystis is its polymorphic life cycle, which includes free-living solitary cells and colonies. A colony is balloon-like, with an envelope consisting of mucilaginous polysaccharides excreted by cells that are in turn embedded in this envelope (Baumann et al., 1994; Vanrijssel et al., 1997). The interior of the colony is filled with seawater. P. globosa colonies usually reach several millimeters in diameter; however, observations in Asian waters have shown that colonies can reach 1cm, with the maximum diameter being ca. 3cm (Qi et al., 2004). Individual cells (both those embedded in colonies and solitary forms) are only 3–12.4μm in diameter (Rousseau et al., 1994; Smith et al., 2014). Colony formation may provide a number of competitive advantages relative to free-living solitary cells. For example, the large size protects colonial cells from being grazed by small zooplankton due to size mismatch (Jakobsen and Tang, 2002). Additionally, the polysaccharide envelope may physically protect colonial cells against infection from bacteria and viruses (Hamm, 2000). Numerous studies have shown that colonies are the dominant forms in P. globosa blooms (Qi et al., 2004; Smith et al., 2014), supporting the perspective that Phaeocystis blooms are linked to the ecological advantages of colony formation. It remains unclear why the sizes of Asian Phaeocystis colonies are greater than those found in other areas of the ocean.

Phaeocystis colonies have special characteristics imparted by the envelope. The envelope is a thin yet mechanically stable skin (Hamm et al., 1999), forming a diffusive boundary layer that may limit nutrient and solute fluxes across the envelope (Ploug et al., 1999). The interior liquid develops into a unique microenvironment that is significantly different from ambient seawater. Smith et al. (2014) demonstrated that dissolved organic carbon (DOC) concentrations of intracolonial fluid were 25 times greater than ambient concentrations. Furthermore, the pH inside young healthy colonies can exceed those of ambient seawater (Lubbers et al., 1990), suggesting the flux of some ions is restricted, a finding that direct experiments verified (Smith et al., 2014).

The microenvironment provided by phytoplankton offers microscale niches for its associated heterotrophic bacteria (Seymour et al., 2017). For example, in the Trichodesmium holobiont, Trichodesmium colonies create a chemically heterogeneous microenvironment (Klawonn et al., 2020), presumably supporting diverse microbial composition and metabolisms, including phosphate and iron acquisition and denitrification pathways in millimeter-sized colonies (Gradoville et al., 2017), which in turn facilitate growth of Trichodesmium. The unique intracolonial microenvironment of P. globosa may harbor specific microbial consortia, which in turn facilitate the growth of P. globosa colonies. However, the microbial diversity and composition of the intracolonial microenvironment is unknown.

In the Amundsen Sea, it has been shown that microbial diversity within P. antarctica blooms was greatly reduced (Delmont et al., 2014), and the dominant microbes had metabolic features that were tied to the metabolism of Phaeocystis and sulfur cycling (Delmont et al., 2015). A number of bacteria in ambient seawater were clearly associated with Phaeocystis, and it was suggested that the bacteria and Phaeocystis established a mutualistic association (Delmont et al., 2014). In the northern Beibu Gulf of China, Xu et al., 2021 demonstrated that microbial communities from P. globosa bloom and non-bloom sites exhibited distinct community composition, and the bloom microbes were likely driven by high environmental selection. However, the microbial community enclosed by the colony envelope is still unknown, and hence the importance of the microenvironment in establishing the unique phytoplankton-bacteria association remains unclear.

We sampled the internal fluid of giant P. globosa colonies collected in Vietnamese coastal waters and analyzed its bacterial composition, sequenced 16S rRNA amplicons, and compared them to those found in ambient seawater to test if the intracolonial microenvironment of P. globosa provides a unique habitat for the development of a specific microbial assemblage. Our results suggest that the microenvironment is indeed different and leads to the development of a microbial consortia that is distinct from the ambient environment.

DNA sequencing of 16S rRNA gene fragments revealed that the microbial assemblage in the intracolonial fluid of P. globosa colonies is significantly different from that in ambient seawater. As the intracolonial fluid extracted from P. globosa colonies contained a mixture of bacteria and P. globosa cells, we used a specific 799F-1193R primer set to minimize contamination of chloroplast sequences (Chelius and Triplett, 2001; Bodenhausen et al., 2013), which has been used in several studies on bacteria associated with plants and macroalgae (Bulgarelli et al., 2012; Vieira et al., 2016; Serebryakova et al., 2018). This primer set displayed very low amplification of non-target DNA and retrieved high number of OTUs as well as exhibiting high Inverse Simpson diversity (Beckers et al., 2016). The use of 799F-1193R primer pair also revealed that Shannon diversity of bacteria from seawater was high and comparable to previous studies (Wang et al., 2019). Although PCR amplification bias introduced by primer selection was inevitable, the primer pair 799F-1193R was well suited for Phaeocystis-associated microbiome analysis.

Our results indicated that both the microbial richness and diversity of the surrounding water were significantly higher than those of the intracolonial fluid, suggesting that the microenvironment of colonies allowed the development of an independent consortia from that outside the colony. Moreover, the four dominant genera, which represented 89.6% of the intracolonial microbiota, were present in extremely low abundances in seawater. This is consistent with previous research that reported that the diatom Thalassiosira rotula-associated microbial assemblages have lowered diversity and are generally dominated by fewer bacterial species (Mönnich et al., 2020). Phytoplankton microenvironments may be much more common in the environment than previously thought and have considerable impacts on biogeochemical cycles.

Although numerous studies revealed the mutually beneficial relationship between phytoplankton and their associated bacteria, there is currently no unifying theory for the mechanism of bacterial assembly on marine phytoplankton. Two hypotheses have been proposed to explain the development of specific phytoplankton-associated bacterial consortia. One is the lottery hypothesis, which implies whoever “gets there first” ultimately dominates the consortia (Munday, 2004). Another hypothesis is “habitat filtering effect,” which emphasizes interactions and the selection from the host (Costello et al., 2012). The second theory has been supported by numerous direct experimental investigations, including the recent replicated and reproducible primary colonization experiment (Mönnich et al., 2020). Phaeocystis colonies likely form from solitary cells that multiply and remain together, ultimately forming colonies (Mathot et al., 2000), although it is also possible for colonies to fragment and create smaller colonies. Recently a colony formation model of P. globosa has been proposed, which suggested that even a single solitary cell with flagella can form the polysaccharide matrix (Zhang et al., 2020). Thus, the volume of seawater inside the matrix would initially be very small and contain only a limited number of bacteria, implying that either a few specific P. globosa-associated bacteria or stochastic bacteria in the ambient seawater would have been initially enclosed in the envelope. However, as colonies grew larger, the microenvironment becomes altered, being substantially different from ambient seawater [i.e., it has greater DOC concentrations (Smith et al., 2014) and elevated pH levels (Lubbers et al., 1990)], allowing for a selection of bacterial forms as the colony grows and continues to alter the intracolonial fluid.

Many observations confirm that phytoplankton-associated bacteria are often restricted to only a small number of groups, dominated by specific members of Bacteroidetes, Alphaproteobacteria, and Gammaproteobacteria (Buchan et al., 2014). The most enriched OTUs in the intracolonial fluid were identified as belonging to the genera Balneola and Labrezia, which are Bacteroidetes and Alphaproteobacteria, respectively. The most dominant OTU (Balneola) comprised 48.6% of the intracolonial fluid sequences. Balneola has been frequently observed to associate with marine phytoplankton, including Emiliania huxley and Cochlodinium polykrikoides (Taniguchi et al., 2011; Rosana et al., 2016). An arabinogalactan utilization gene cluster present in Balneola sp. genome indicated its ability to degrade polysaccharide, thus the heterotrophic bacteria could use polysaccharide synthesized by P. globosa as its carbon source. Moreover, Balneola has been implicated in the degradation of aromatic benzenoid compounds such as toluene (Li et al., 2012). Aromatic benzenoid compounds can be significant organic contaminants, but some benzenoid compounds serve as a biochemical defense when plants or microorganisms encounter stress (Misztal et al., 2015). No evidence is available for toluene production by P. globosa, although significant toluene fluxes to the atmosphere from Emiliania huxleyi blooms (a haptophyte like P. globosa), have been observed (Misztal et al., 2015). Isotopic labeling experiments have confirmed that plants release toluene and increase toluene production under stress conditions (Heiden et al., 1999; Misztal et al., 2015). Although the biological function of toluene in plants and phytoplankton remains unclear, we speculate that toluene accumulated in P. globosa colonies could provide carbon source to Balneola.

The second dominant OTU was classified as Labrenzia, which frequently forms close associations with a variety of marine autotrophs including diatoms, dinoflagellates, green and red algae (Weber and King, 2007). However, the close relationship between Labrenzia and P. globosa, and the biological function of enriched Labrenzia in P. globosa intracolonial fluid, have not been reported. It has long been recognized that P. globosa is an important DSMP producer and consumer (Keller et al., 1989; Stefels and Van Boekel, 1993). However, the role of the marine heterotrophic bacteria Labrenzia in DMSP producing and consuming has been demonstrated recently (Curson et al., 2017). In addition to serving as an osmolyte in many marine algae, including prymnesiophytes and dinoflagellates, DMSP and DMS also function as antioxidants under stressful conditions (e.g., high solar photon fluxes and nutrient limitation), which result in substantial increases in cellular DMSP and DMS concentrations (Sunda et al., 2002; Brisbin and Mitarai, 2019). Marine phytoplankton will encounter challenges caused by global change in the future, and as a result, cellular DMSP and DMS production may increase due to this stress. If this were to occur, release of these volatile sulfur compounds to the atmosphere will increase, enhancing atmospheric feedback loops that influence global climate and hydrological cycles. Research to date has not focused on the source of DMSP and DMS in P. globosa colonies, but future research should determine the dominant source of DMS/DMSP and consider the interaction between the two biotic components.

In the present study, dominant bacteria in the intracolonial fluid were completely different from the dominant bacteria observed by Xu et al. (2021) during P. globosa blooms, neither in free-living fraction (0.22–3μm) nor in particle-attached fraction (3–20μm). However, NS5 marine group, which belongs to Proteobacteria, was observed as one of the dominant bacteria in seawater free-living fractions at the bloom site from both studies. The results suggested the significance of microenvironment enclosed by colonial envelope, which established the distinct bacterial community. The interaction between heterotrophic bacteria and P. globosa solitary cells in the microenvironment might benefit the development of P. globosa colony.

Bioinformatic prediction by PICRUSt offers an effective way to understand the function of bacteria inside P. globosa colonies when metagenomic data are not available. However, methods, including NSTI scores, should be used to test the reliability of PICRUSt. Generally, accuracy of PICRUSt decreased with increasing NSTI score, but still produced reliable results for a dataset of soil samples with a mean NSTI score of 0.17 (Langille et al., 2013). In the present study, mean NSTI scores were relatively low (0.111) for intracolonial fluid samples, and relatively high for seawater samples (0.258). Langille et al. (2013) indicated that environments containing much unexplored diversity, such as the Guerrero Negro hypersaline microbial mats, showed a markedly lower accuracy with a mean NSTI score of 0.23. The relatively high NSTI scores obtained here indicate that the PICRUSt predictions must be treated with caution. The results, however, still provide some interesting insights into potential bacterial community functioning, that should be confirmed in future with more reliable and quantitative studies.

PICRUSt prediction revealed that the 15 abundant KEGG level 2 pathways were the same in P. globosa intracolonial fluid and ambient seawater, which demonstrated that both microbiomes have the same basic biological functions. However, significant functional differences were observed between two microbiomes by STAMP analysis. Surprisingly, the KEGG pathway of penicillin and cephalosporin biosynthesis (ko00311) was significantly enriched in intracolonial bacteria communities. Synthesis of antibacterial compounds is one mechanism by which one particular species of bacteria can out-compete another and increase the successful colonization associated with phytoplankton. The widespread marine roseobacter Phaeobacter inhibens produces the broad-spectrum antibiotic tropodithietic acid (TDA), which confers to it a competitive advantage (Majzoub et al., 2019). Thus, P. inhibens dominates the diatom T. rotula associated microbiota (Majzoub et al., 2019); it also colonizes the surface of Ulva australis (Rao et al., 2006). Our results suggest that elevated bacterial competition in P. globosa intracolonial fluid via the release of antibacterial compounds likely resulted in lower species diversity of the P. globosa-associated microbiota than that in the ambient seawater. Consistent with the function of dominant genus Balneola in the intracolonial fluid, the aromatic compound degradation pathways were also depicted by PICRUSt prediction. Although the source of the aromatic compounds in the intracolonial fluid is unknown, the predicted function suggests that the associated bacteria could use fluorobenzoate and toluene as a carbon and energy source.

Based on the dominated OTUs and the enriched predicted KEGG pathways, our study suggests that the P. globosa colonial envelope acts as a selective mechanism, and allows the development of a bacterial consortia and their specific functions. Furthermore, upregulated synthesis of antibacterial compounds also promoted the specificity of intracolonial bacteria. Enriched bacteria in the intracolonial fluid involved in DMSP and DMS production, which may affect DMSP and DMS synthesis by P. globosa, thus have a substantial influence on the sulfate cycle of the ocean. The results presented here highlight the close interaction between P. globosa and bacteria in the intracolonial fluid. However, the specific mechanisms are still unknown. For example, although both the toluene-degrader Balneola and the predicted function of toluene degradation have been suggested, the concentration of toluene inside the colonies is unknown. Thus, further research is needed to determine whether P. globosa can produce toluene, under what conditions this production is enhanced, and to investigate the interaction between Balneola and P. globosa. In addition, the contribution of Balneola and Labrezia in P. globosa intracolonial fluids should be confirmed by qPCR. While metabolic function analysis was only predicted by PICRUSt, its accuracy relies on the availability of completely sequenced genomes for the representative organisms (Wilkinson et al., 2018). Therefore, metagenomics and metatranscriptomics analysis are needed to verify the actual bacterial structure and function of P. globosa-associated communities to reveal the interaction between microbiota and P. globosa.

We assessed the bacterial composition and their predicted functions in P. globosa intracolonial fluids. The results indicated that bacterial richness and diversity were significantly lower in intracolonial fluid than that in ambient seawater due to the selection by unique intracolonial microenvironment. In addition, the close interaction between P. globosa solitary cells and heterotrophic bacteria might benefit the development of P. globosa and further impact the global biogeochemical cycles. In summary, the results suggests that the P. globosa colonial envelope acts as a selective mechanism and allows the development of a bacterial consortia and their specific functions.

The raw reads are available by the valid accession number: PRJNA720571.

ZZ, RM, WS, HD-N, and LN-N designed the experiment. ZZ, RM, WS, and HD-N collected sample in Viet Nam. 16S rDNA sequence analyses and bioinformatics analyses were conducted by ZZ and XJ. ZZ wrote the manuscript with contributions from RM and XJ. WS and HD-N revised the manuscript. All authors agree to be listed in this manuscript and approve the submitted version of the manuscript.

This work was supported by grants from the National Foundation for Science and Technology Development (NAFOSTED) of Viet Nam awarded to HD-N and LN-N (DFG 106-NN.06-2016.78).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.