The characteristics of host lipid body biogenesis during coral-dinoflagellate endosymbiosis

Coral husbandry

Corals (Euphyllia glabrescens) were collected from a reef at the inlet of the Third Nuclear Power Plant in Nanwan Bay, Taiwan (21^∘57.376′N, 120^∘ 45.291′E) in 2012. Coral collection was approved by the Kenting National Park Management Office. Ten∼ twelve colonies were cultured over eight years in each 4-ton tanks with flow-through seawater (exchange rate of ∼80L/h) in the husbandry center of the National Museum of Marine Biology and Aquarium (NMMBA). Twenty-four colonies for experiments were maintained under an ambient diel cycle with natural sunlight, and temperature was maintained at 26.5 ±1 ^∘C using a microprocessor-controlled cooler (First FC-45, Aquatech, Kaohsiung, Taiwan). Light intensity and water temperature were continuously recorded using a HOBO Pendant^® logger placed near the colonies in each tank (UA-002; Onset, Pocasset, MA, USA).

Analysis of in situ LB distribution

Tentacle samples were collected at the following nine times during one diel cycle: 06:00 (sunrise), 9:00, 12:00 (noon), 15:00, 18:30 (sunset), 21:00, 00:00 (midnight), 03:00, and 05:00. Tentacles were collected from three distinct colonies (n = 3 tentacles/colony) at each time point. A single polyp was targeted only once to avoid sampling-induced stress. For accurate analysis of LBs, preserving the natural tissue architecture of the expanded tentacle was critical. Sampling the tentacle without using a capillary clip resulted in a sectioned gastroderm exhibiting multiple layers of cells containing endosymbionts (Fig. 1A). We circumvented this limitation by successfully preserving the tissue architecture of the naturally expanded tentacle with a clamping technique marked using solid arrows in Figs. 1B and 1C. To verify the effectiveness of this technique, architectural differences in the gastrodermal tissue of extended and contracted tentacles were assessed by staining LBs with osmium tetroxide, cryosectioning fixed specimens, and imaging sections by using differential interference contrast microscopy (Figs. 1A & 1D) as described by Chen et al. (2012). This technique allowed the gastroderm to consist of a single cell layer (Fig. 1D).

Ultrastructure analysis through Transmission Electron Microscope

Clamped tentacles were amputated using micro scissors and immediately fixed with 2.5% glutaraldehyde and 2% paraformaldehyde in 100 mM sodium phosphate containing 5% sucrose (pH 7.3) for 2.5 h at 4 ^∘C then washed with 100 mM sodium phosphate at 4 ^∘C. Specimens were post fixed in 1% osmium tetroxide in 50 mM sodium phosphate (pH 7.3) for 1 h at 4 ^∘C. Samples and data were then proceeded and analyzed according to previously methods by Chen et al. (2012). Individual LB sizes were determined by measuring their maximum diameter (μm), and LBs were categorized into three size classes (micro-LBs: <1 μm; mid-LBs: 1–3 μm; large-LBs: >3 μm) and two distribution categories (near the mesoglea or near the coelenteron) by using Metamorph 6.3 (Molecular Devices).

Transcriptomic analysis of ER genes related to LB lipogenesis

Diel expression pattern of BiP in host cells and purified LBs

Statistical analysis

Data were analyzed using SPSS (version 14.0; SPSS Inc., Armonk, NY, USA). One-way analyses of variance and Tukey pairwise comparisons were conducted to determine the effect of sampling time on LB density, size, location, and gene and protein expression, and results were considered statistically significant when p was <0.05. Values are presented as means ± SDs.

Ultrastructural and morphological characteristics of host LBs in the gastrodermis

Similar to the biogenesis of lipid droplets in other eukaryotic cells, that of LBs in coral–Symbiodiniaceae endosymbiosis has several unique morphological characteristics. We find multiple lines of evidence suggesting that lipid droplets are derived from the ER. First, coral host LBs were in close spatial proximity to mitochondria (Fig. 2), endoplasmic reticulum (ER, Fig. 2A), and Golgi apparatus (Fig. 2D and 2G), which exhibit high electron density upon transmission electron microscopy (TEM) examination. Our in situ morphology of coral host LBs revealed that they originate in the ER (Fig. 2A) and are surrounded by the ER (Fig. 2B). Figures 2B and 2C illustrates that micro-LBs and some mid-LBs accumulated in the coral host ER and then migrated and fused together. LBs formed a single globule with few inclusion bodies (Figs. 2C–2F) and had dense Golgi spots (Figs. 2D and 2G). When fusion occurred, numerous granules with high-electron density (Figs. 2E, 2F, 2H and 2I) appeared within the mid-LBs as they fused with micro-LBs, (Figs. 2E and 2H), and when mid-LB fused with large-LB (Figs. 2F and 2I). Finally, large-LBs continued to integrate into larger LBs.

Diel pattern of LB size and distribution

To assess the spatiotemporal dynamics of coral LBs during the diel cycle, gastrodermal LBs were analyzed using TEM for three sizes (i.e., micro-LBs: <1 μm; mid-LBs: 1–3 μm; large-LBs: >3 μm) and two distributions near the mesoglea or coelenteron within the gastroderm (Fig. 3A). Total LBs started to increase at sunrise and gradually increased during the day, reaching maximum density at sunset (Fig. 3B).

Micro-LBs comprised the largest portion of LBs throughout the diel cycle (Fig. 3C; Table S1). A significantly larger portion of micro-LBs were located near the mesoglea (approximately 40–80%) and they increased moderately throughout the diel cycle (Fig. 3C; Table S1; light period: 37.2 ± 2.9% to 44.3 ± 3.8%; dark period: 40.1 ± 3.4% to 53.3 ± 2.8%). The percentage of micro-LBs near the coelenteron gradually decreased from 22.3 ± 1.7% at sunrise to 10.0 ± 3.1% at sunset (Fig. 3C; Table S1) and increased overnight to 13.8 ± 1.6% at 21:00 and 20.3 ± 4.2% at 05:00. The number of mid- and large-LBs near the mesoglea remained relatively low (0.7 ± 0.3% and 13.5 ± 4.4%, respectively; Fig. 3C; Table S1) and nonsignificantly different during the diel cycle (mid-: p = 0.45; large-: p = 0.11). However, LB populations close to the coelenteron exhibited significant diel fluctuations (mid- and large-: p  <  0.05). The percentage of these larger LBs increased during the light period and returned to baseline levels overnight (Fig. 3C). The peak in relative abundance for mid- and large-LBs near the coelenteron occurred at 18:30 (23.3 ± 1.1%) and 21:00 (19.3 ± 1.7%), respectively.

Transcriptomic analysis of UPR activation/ER stress genes related to lipogenesis

In response to ER stress, a signal transduction pathway known as UPR is activated, and it may affect the metabolic process of de novo lipid biosynthesis. We analyzed the transcriptome of the coral Euphyllia glabrescens and focused on unigenes and pathways related to LB metabolism. LB lipogenesis occurs primarily in host cells, where metabolic signals regulate the expression of key enzymes in lipogenic pathways (Fig. 4A, modified from previous models; (Morris et al., 1997; Zheng, Zhang & Zhang, 2010; Basseri & Austin, 2012)). Specifically, as a monitor of UPR activation/ER stress, the ER chaperone and signaling regulator GRP78/BiP exhibited a fourfold increase in expression at noon. Nearly simultaneously, the two sensor proteins protein kinase R-like ER kinase (PERK) and inositol-requiring enzyme 1 (IRE1), which induce various pathways and drive lipid biosynthesis, exhibited near a two-fold increase in expression. IRE1 expression started to increase at sunrise and could reach a two-fold change in transcription factor X-box binding protein 1 (XBP1) expression by noon. The eukaryotic translation initiation factor 2 subunit 1 (eIF2 α) expression started to increase twofold change at sunrise. Transcription factor sterol regulatory element-binding protein 1 (SREBP1) activated by PERK also had increased expression at noon. The activating transcription factor 6 (ATF6) sensor protein was activated after midnight, whereas membrane-bound transcription factor site-1 protease (S1P) exhibited increased expression at sunset. Figure 4B presents a box plot of the distributions and outliers of gene expression for UPR activation/ER stress genes related to LB lipogenesis. Genes including BiP, PERK, eIF2α, SREBP1, IRE1, XBP1, ATF6, and S1P were upregulated starting at sunrise and downregulated from sunset to midnight.

Diel fluctuations of ER marker BiP locations

Figure 5A illustrates the successful fluorescence staining of purified LBs with lipid marker BODIPY 493/503 and ER-tracker Blue-White dye. The LBs contained neutral lipids and were enclosed in an ER membrane. The aforementioned in situ morphological findings also supported this finding (Fig. 2). BiP was selected to be an ER-specific marker protein to assess the functional association of coral gastrodermal LBs with the host ER. BiP expression was detected only in the coral host and purified LBs but not in endosymbionts (Fig. 5B). Furthermore, the 78-kDa BiP doublet was present in the purified LBs, indicating the close association of BiP with LBs. The two bands of BiPs in the host and purified LBs (Fig. 5B) were removed and analyzed using mass spectrometry (Fig. S1). The upper band in both LBs and the host was identified as a 78-kDa glucose-regulated protein (GRP78/BiP), and the lower band in the host fraction was identified as a 73-kDa heat shock protein (HSP70).

BiP expression during the diel cycle was assessed through Western blotting (Fig. 5C) and then quantified (Fig. 5D). To understand the relative ratios of BiP involved in LB lipogenesis of the total host fraction, the diel fluctuations in the relative concentration of BiP associated with purified LBs were assessed and compared with those of the remaining host fraction (Fig. 5D). The relative expression of BiP associated with LBs increased significantly at 15:00 and decreased at 18:00 (Fig. 5D). In addition, BiP abundance in the host fraction was significantly higher at night and peaked at 21:00. Thus, BiP exhibited distinct temporal fluctuation patterns in the LB and host fractions.

Cytological characteristics of LB formation reveal LB origin

We successfully preserved the gastroderm consisting of a single layer of cells (Fig. 1). LBs move to self-assemble and fuse with each other (Boström et al., 2005). This is consistent with our observations that smaller LBs fuse to form larger LBs, resulting in irregularly shaped large LBs (Figs. 2B and 2C). The fusion of several smaller lipid droplets to larger lipid droplets also contributes to lipid droplet growth (Guo et al., 2008; Cheng, Fujita & Ohsaki, 2009). The LBs were in close spatial proximity to mitochondria, ER, and Golgi apparatus. Originating from the ER, lipid droplets can associate with most other cellular organelles through membrane contact sites (Jacquier et al., 2011; Peng et al., 2011; Olzmann & Carvalho, 2019). LB mobility was also attributed to the dynamic interactions of LBs with other organelles, including mitochondria (Figs. 2A, 2D, 2E and 2F), ER (Fig. 2A, Fig. 2B and Fig. 2C), the Golgi apparatus (Fig. 2D and 2G), peroxisomes, endosomes, and lysosomes (Herms et al., 2015; Gao & Goodman, 2015; Crossland, Barnes & Borowitzka, 1980; Hayes & Goreau, 1977; Vandermeulen, 1974). Therefore, LBs may play a role in the intracellular trafficking of lipids as well as proteins and other molecules between organelles. Moreover, LBs may migrate within host gastrodermal cells toward intracellular sites of lipid utilization and catabolism. Importantly, LBs biogenesis and degradation, as well as their interactions with other organelles, are tightly coupled to cellular metabolism and facilitate the coordination and communication between different organelles and act as vital hubs of cellular metabolism (Olzmann & Carvalho, 2019). In many cells, lipid droplets undergo active motion, typically along microtubules. This motion has been proposed to aid growth and breakdown of droplets, to allow net transfer of nutrients from sites of synthesis to sites of need and to deliver proteins and lipophilic signals (Pol et al., 2004; Welte, 2009). Thus, nascent LB globule biogenesis likely occurs near the mesoglea. As LBs grow and mature, they migrate across gastrodermal cells and fuse to form large LBs adjacent to the coelenteron.

The time-dependent LB relocation may occur inside a single gastrodermal cell during their biogenesis which ultimately manifests as a redistribution of LBs across the gastrodermal tissue layer. LBs move along microtubules and relocate within the cytoplasm (Welte, 2009). A proteomic investigation indicated that cytoskeletal proteins and proteins involved in intracellular trafficking were associated with coral LBs (Peng et al., 2011). The mechanism by which LB populations migrate across the gastroderm to cause this redistribution remains unknown. By correlating LB size and location throughout the diel cycle in naturally expanded tentacle specimens, a study validated the proposed redistribution of LBs from the mesoglea to the coelenteron (Chen et al., 2012). Our data further demonstrated that micro- and mid-sized LBs fuse to form large LBs near the coelenteron. In particular, micro-LBs were the most abundant near the mesoglea, whereas mid- and large-sized LBs increased in abundance near the coelenteron during the light period (Fig. 3). This finding was consistent with a previous study suggesting that LBs migrate across the gastroderm during the diel cycle (Chen et al., 2012).

Transcriptional variation of BiP is correlated with LB lipogenesis

Ultrastructural morphology studies have revealed that lipid droplets are closely associated with the ER membrane (Fig. 2), and proteomic studies of lipid droplets isolated from various cell lines have revealed numerous ER proteins (Chen et al., 2012; Peng et al., 2011; Liu et al., 2004; Brasaemle et al., 2004; Wan et al., 2007). The ER is the main site of the synthesis of lipids that constitute most of the lipid components of all biological membranes, and the ER is highly sensitive to changes in intracellular homeostasis and extracellular stimuli. During the perturbation of ER homeostasis, referred to as ER stress, UPR is activated when misfolded proteins accumulate, when reactive oxygen species (ROS) are produced, or when the accumulation of FAs resulting from lipolysis alter ER membrane lipids (Volmer & Ron, 2015; Walter & Ron, 2011; Chitraju et al., 2017). We demonstrated that the increase in BiP gene expression from sunrise to noon (Fig. 4) may reflect a localized increase in BiP in daytime within the ER microdomain involved in LB formation (Fig. 5). BiP is the primary regulator of the UPR and is upregulated in response to ER stress (Bertolotti et al., 2000; Morris et al., 1997). Furthermore, PERK and IRE1 cascade facilitates LB formation and possibly lipogenesis, and the ATF6 branch enhances phospholipid biosynthesis (Zheng, Zhang & Zhang, 2010; Basseri & Austin, 2012; Zha & Zhou, 2012). Our transcriptiomic variation results (Fig. 4B) were also consistent with those of previous studies, which indicated a general increase in total lipids, FA concentration, and density in LBs during the light period followed by a decrease at night (Chen et al., 2012; Chen et al., 2017).

In addition to its core function in lipid formation, the diel cycle may provide a mechanism for managing daily oxidative stress induced by photosynthesizing Symbiodinaceae in coral host tissue (Levy et al., 2006). In addition to ROS production in the ER resulting from oxidative protein folding, this could overwhelm ER homeostasis and induce daily ER stress response (Malhotra & Kaufman, 2007; Levy et al., 2011; Oakley et al., 2017). LB formation may play a role in alleviating ER stress by sequestering unfolded proteins and transiently storing them for degradation (Ploegh, 2007). This has been supported by the observations of increased LB biogenesis induced by ER stress (Fei et al., 2009; Lee et al., 2012), but this functional link remains uncertain (To et al., 2017).

Translational fluctuations of BiP revealed LB origins

After analyzing diel transcriptional variations and the transcriptomic pathway, we selected the ER marker BiP associated with LBs to elucidate the connection among LB formation, the ER, and the diel cycle. We observed an increase in the relative concentration of BiP associated with LBs in midafternoon (Fig. 5). However, BiP expression in the host fraction continued to increase at night (21:00). The increase in the apparent relative abundance of BiP may be due to an increase in either ER network volume or localized BiP concentration in response to upregulated BiP expression. Our current data cannot be used to distinguish one cause from the other. Nevertheless, the consistent association of BiP with LBs throughout the diel cycle and the diel patterns in BiP concentrations between the LB and host fractions demonstrated that LBs are closely associated with a specialized domain of the host ER. The peak in LB-associated BiP was associated with an increase in the total volume of LBs (Fig. 5), thereby supporting the involvement of ER expansion in LB biosynthesis. Moreover, the nocturnal increase in host ER BiP but not in LB formation may indicate ER expansion at night. The ER is a dynamic organelle, and its tubular network can expand to meet the demands of ER-based processes, including protein folding, modification, and secretion; lipid metabolism; and calcium homeostasis (Federovitch, Ron & Hampton, 2005).

A persisting question in LB research is whether LBs ever truly detach from the ER (Mishra et al., 2016). This is particularly relevant considering the substantial migration and fusion of LBs in the coral gastroderm. After the peak in LB-associated BiP occurred in the midafternoon, the relative BiP concentration decreased at sunset (Fig. 5D). However, total LB volume reaches its highest concentration at sunset (Chen et al., 2012; Chen et al., 2017; Peng et al., 2011). Therefore, larger LBs that dominate at sunset likely detach from the ER or are at least less closely associated with the ER. The continuity of the ER membrane with the LB surface may be maintained permanently with stalk-like projections of the ER membrane that allow for the continued lipid and protein trafficking between the organelles (Jacquier et al., 2011; Zehmer et al., 2009). Moreover, the dynamic fluctuation of BiP expression between the coral host and LBs rules out the possibility that BiP within the LBs was merely an impurity collected in the isolation process. Rather, our findings suggested that coral gastrodermal LBs are functionally connected to a specialized domain of the host ER.

Dynamic schematic model of LB formation

Coral gastrodermal LBs are highly dynamic organelles, and LB formation is highly associated with coral host ER. Imaging analyses of ultrastructural morphological characteristics and the Western blotting of BiP–ER marker protein distribution supported the model of LB formation (Fig. 6). We developed this model of ER involvement in LB biogenesis to illustrate the distribution of BiP expression between a coral host and LBs (Fig. 6A), resulting in dynamic fluctuations in BiP expression (Fig. 5). Furthermore, we proposed a six-stage process of LB formation and maturation (Fig. 6B). First, micro-LBs bud from the ER surrounded by mitochondria (Fig. 6B [i]). Subsequently, more micro-LBs gather and are surrounded by ER sheaths (Fig. 6B [ii]). In the third stage, micro-LBs fuse into larger LBs with different lipid compositions (Fig. 6B [iii]). This might result in different colors in TEM image analysis. During the diel cycle, LB growth and fusion result in mid-LBs. At the same time, a large population of micro-LBs is maintained (Fig. 6B [iv]); this occurs with dense Golgi spots in particular. Mid-LBs continue to fuse to form large-sized LBs (>6 μm). When fusion occurs, numerous “dots” form on vesicles in between the LBs, as illustrated in Fig. 6 [v]. Finally, the large-LBs fuse to form mature LBs, as presented in Fig. 6 [vi].

LB size must be strictly regulated to balance lipid storage and mobilization in response to energetic demand (Yang et al., 2012). In response to excess lipid molecules, a cell can increase either the number of LBs or the size of existing LBs. Changes to LB size alter the LB volume-to-surface-area ratio, which affects the rate of LB catabolism. The formation of large LBs as observed in the coral gastroderm (Fig. 3C) is the most efficient form of lipid storage (Yang et al., 2012). Moreover, micro-LBs potentially function to meet immediate energetic demands because of the rapid availability of lipids for catabolism and oxidation (Ariotti et al., 2012). Therefore, heterogeneity in gastrodermal LBs may reflect diversity in LB functions (Zhang et al., 2016). The overnight disappearance of large LBs was hypothesized to occur because of nocturnal lipid catabolism and utilization (Chen et al., 2012), however, the underlying mechanism is not clear.