Mucospheres produced by a mixotrophic protist impact ocean carbon cycling

Mixotrophic protists (unicellular eukaryotes) that engage in both phototrophy (photosynthesis) and phago-heterotrophy (engulfment of particles)—are predicted to contribute substantially to energy fluxes and marine biogeochemical cycles. However, their impact remains largely unquantified. Here we describe the sophisticated foraging strategy of a widespread mixotrophic dinoflagellate, involving the production of carbon-rich ‘mucospheres’ that attract, capture, and immobilise microbial prey facilitating their consumption. We provide a detailed characterisation of this previously undescribed behaviour and reveal that it represents an overlooked, yet quantitatively significant mechanism for oceanic carbon fluxes. Following feeding, the mucospheres laden with surplus prey are discarded and sink, contributing an estimated 0.17–1.24 mg m−2 d−1 of particulate organic carbon, or 0.02–0.15 Gt to the biological pump annually, which represents 0.1–0.7% of the estimated total export from the euphotic zone. These findings demonstrate how the complex foraging behaviour of a single species of mixotrophic protist can disproportionally contribute to the vertical flux of carbon in the ocean.

immobilised in mucospheres captured during experiments assessing prey consumption capabilities. Green circles show P. cf. balticum cells and red circle show prey species. Note images j1-j2, l1-l2, n1-n2, p1-p2 are the same cells but with either the P. cf. balticum cell or the prey cell in focus. Images were taken on an inverted fluorescence microscope (Nikon Eclipse Ti) with a black and white camera (Nikon DS-QiMc) under 200 or 400x magnification. Scale bars = 50 µm.
14 Supplementary Figure 11. A series of brightfield images through time of the same field of view taken at intervals over 24 hours (a1 = 0 hours; a2 = 6 hours; a3 = 7 hours; and a4 = 24 hours) using the INCell High Content Analyzer 2200 automated microscope system using a 10× objective demonstrating the negative buoyancy and propensity for sinking of a mucosphere laden with excess eukaryotic (Rhodomonas salina) and prokaryotic prey cells. The sinking mucosphere (arrow) is visible from 7 hours onward (a3-a4). Scale bar = 100 µm (c) Shows representative differential interference contrast images of each variant taken using live cells on an upright fluorescence microscope (Nikon Eclipse Ni, Japan) with a mono camera (Nikon DS-Qi2) demonstrating very little change in the cell proportions due to antibiotic treatment but clear variation in cell size. Figure 15. Growth curve of P. cf. balticum grown with an organic (sodium β-glycerophosphate) (bright coloured square symbols) and inorganic (sodium phosphate) (pale coloured triangle symbols) form of phosphorus. In the absence of bacteria P. cf. balticum was not able to grow with organic phosphorus (axenic).

Supplementary
Mean ± standard deviation (n=5 biological replicates).  There is a significant difference between sample medians

Supplementary Note 1. Detailed morphological and toxicological description of P. cf. balticum
The small (13-16 µm transdiameter) spheroid, laterally compressed dinoflagellate, is fast swimming, propelled by a longitudinal and transverse flagellum projecting from a single flagellar pore located anteriorly ( Supplementary Fig. 1a), typical of Prorocentroid desmokonts ( 2 , p 402). Golden-brown chloroplasts are located at the periphery of cells (Fig. 1c) protected by a theca consisting of two halves or valves. Scanning Electron Microscopy (SEM) revealed the valves are covered in short spines (150 nm high), irregularly scattered small pores (150 nm diameter) and a small number of larger pores (330 nm diameter) located near the apical area, from which longer spines (600 nm high) emanate, and valves are joined by a distinct, ornamented intercalary band ( Fig. 1d; Supplementary Fig. 1a-g). These morphological features and the presence of two wing-like apical projections bordering the periflagellar area ( Fig. 1d; Supplementary Fig. 1b and e) align with the light microscope description of Prorocentrum balticum (Lohmann) Loeblich III

(basionym Exuviella baltica Lohmann) by Lohmann 3 from the Baltic Sea, and the Scanning Electron
Micrograph illustration from near the type locality ascribed to the same species by Elbrächter in Hoppenrath, et al. 4 (Fig. 78 m-n).
Phylogenetic analysis of ribosomal DNA from the internal transcribed spacer region (ITS) located between the 18S and 5.8S genes, the 28S large subunit (LSU) and the 18S small subunit (SSU) showed the strains to be indistinguishable from each other, though distinct from all genetically represented species within the genus Prorocentrum ( Fig. 1a and Supplementary Fig. 2a-b). The strains grouped within a clade comprising the morphologically similar planktonic species P. minimum (=P. cordatum) and morphologically dissimilar species P. dentatum, P. donghaiense and P. shikokuense (=P. obtusidens) but were most closely related to taxonomically undefined strains (notation Prorocentrum sp. culture QUCCM SS1-13 from the Arabian Gulf and P. cf. balticum, culture CCMP 1787 from New Zealand, CCMP 1260 from Gulf of Mexico) ( Fig. 1a and Supplementary Fig. 2a-b). Unfortunately, the original description of P. balticum by Lohmann 3 lacks critical detail of thecal plate ornamentation and no archived type material exists, nor are genetic sequences available for strains isolated from the type locality. Without a direct comparison, we cannot unequivocally determine if our strains are undescribed or are indeed authentic P. balticum and have therefore referred to them herein as Prorocentrum cf. balticum.
It is worth noting that the morphologically similar and phylogenetic sister species Prorocentrum minimum, has been linked with Harmful Algal Bloom (HAB) events and shown to produce a water-soluble neurotoxin found to cause detrimental effects to scallops, oysters, and clams 5 , mortalities of fish, shellfish and zoobenthos ( 6 and references therein), and death to mice 7 . Negative human health effects have also been reported, linked with the production of tetrodotoxin 8 . Subsequent studies have questioned whether production of the toxin was associated with the microalgal cells themselves, or a symbiotic bacterium 9 . Additional toxins associated with negative human health effects are also produced by other species from the genus Prorocentrum, including okadaic acid 10-12 and dinophysistoxins 10,11,13 . When tested using Liquid Chromatography-Tandem Mass Spectrometry (LC-MS/MS), our P. cf. balticum strains and their associated bacterial microbiome did not produce okadaic acid, dinophysistoxin-1 and -2, or tetrodotoxin, thus this species is unlikely to be associated with human illness or detrimental ecosystem impacts.

Supplementary Note 2. Nutrient limitation induced initiation of sexual reproduction
Phototrophy is a successful strategy for P. cf. balticum when dissolved inorganic nutrients are sufficient, and phago-heterotrophy an effective behaviour when they are not, but there can be occasions when dissolved  15,16 ) in which the description of the nuclei donation process through a fertilisation tube remarkably resembles the peduncular phago-heterotrophic feeding mechanism described within this study.
Phagocytosis involves the interception of a prey cell and complete engulfment via an opening in the dinoflagellate sulcal region [18][19][20] or the apical horn 19 , and is mainly observed in, but not restricted to, unarmored dinoflagellates. In contrast, pallium and peduncular feeding are generally observed in armored dinoflagellates that are inhibited from ingesting large and complete prey through the narrow opening in the sulcal region 21 .
Pallium feeding involves the capture of a potential prey item using a tow filament, followed by the extrusion of a pseudopodial "feeding veil" that engulfs and digests the prey outside of the dinoflagellate cell [22][23][24] .
Similarly, peduncular feeding involves the capture of a potential prey item using a tow filament, extrusion of the peduncle, which penetrates the prey membrane and through which the cellular contents are extracted and deposited in food vacuoles for intracellular digestion within the dinoflagellate [24][25][26] . This process is also known as myzocytosis or "cellular vampirism" 27 and allows the consumption of prey items as large as 10 times the dinoflagellate's size 21 .
Without direct observation of feeding, it is difficult to confirm the presence of a peduncle. On occasion, peduncle-like tubular organelles can be detected in SEM images (e.g., 28,29 ) or inferred from the presence of intracellular microtubular baskets in Transmission Electron Microscope (TEM) images (e.g., 30,31 ) and the presence of these have been proposed to indicate peduncular feeding capabilities of dinoflagellates in general 26 . Such structures have been reported for other Prorocentrales (e.g., P. norrisianum and P. tropicalis in 28 ; P. lima in 32 ; P. micans in 31,33,34 , P. arenarium in 29 , including the morphological and phylogenetically similar species P. minimum (as P. mariae-lebouriae in 31,35 ). It has therefore been assumed that some Prorocentrum species feed via a peduncle, but this has never been observed until now. Curiously, the feeding mechanism for P. micans 19 and P. minimum has been reported as direct engulfment via separation of the intercalary band through the sutures between the two valves 19,36 . This contrasts with previous reports of microtubular structures in these species and the peduncular feeding we observed in P. cf. balticum.

Supplementary Note 4. Mixotroph classification
Further classification of mixotrophic protists differentiates between taxa with the innate ability to photosynthesise through vertical transmission of plastids, as opposed to those which acquire the ability through ingestion and domestication of phototrophic prey, known as constitutive and non-constitutive mixotrophs, respectively 37 . Constitutive mixotrophs can be further defined by their preference for phototrophy versus phago-heterotrophy 38,39 . Some protists are considered obligate phototrophs, acquiring carbon from photosynthesis, and supplementing nutrients through phago-heterotrophy (facultative phago-heterotroph) (defined as type II mixotrophs in 39 ). While other protists are obligate phago-heterotrophs, only photosynthesising (facultative phototrophs) when prey are limiting (defined as type III mixotrophs in 39 ).

Supplementary Note 5. Mucosphere composition
Mucus can be produced by microorganisms through a variety of mechanisms, but the most common method recognized for Prorocentrales, is secretion through pores in the thecal valves 2 . Equally, most pores in the thecal valves of Prorocentrales are associated with trichocysts, a type of extrusome that is involved in phagoheterotrophic feeding 21 . P. cf. balticum has two types of pores ( Fig. 1d; Supplementary Fig. 1e), including a small number of large pores (330 nm diameter) located near the apical area from which large spines emanate, a feature that is unique to the strains described in our study. Trichocysts and mucocysts were not observed but it is possible that this unique morphological characteristic in P. cf. balticum is associated with the release of the mucoid material that is used to construct its distinctive mucospheres. Effort was made to further characterise the chemical composition of the mucospheres by staining with a series of fluorescent stains including SYPRO-Red for proteins 40 , Calcofluor white for β-polysaccharides 41 , BODIPY for neutral lipids 42 and Acridine Orange for mucopolysaccharides 43 but all were negative (Supplementary Fig. 15c-f).