Introduction

Coral reefs are important tropical ecosystems that are both economically (Spalding et al. 2017) and culturally (Cvitanovic et al. 2013) valuable. Their persistence in nutrient-poor tropical waters (Hoegh-Guldberg 1999) is made possible by an intracellular symbiosis formed between the coral animal and dinoflagellates in the family Symbiodiniaceae (formerly genus Symbiodinium) (LaJeunesse et al. 2018). These symbionts transfer metabolites (Sogin et al. 2017) and photosynthetic products (Hoegh-Guldberg et al. 2007) that supplement a coral’s energy requirements for vital life processes such as growth (Little et al. 2004), calcification (Colombo-Pallotta et al. 2010), and reproduction (Edmunds and Spencer Davies 1986). Warming trends over the past several decades have threatened this symbiosis in many ways. The most well-known threat is coral bleaching, which results from the loss of the symbiont itself (Weis 2008) or from bleaching of the symbiont’s photopigments (Hoegh-Guldberg 1999). It is understood as a breakdown in symbiosis (Baker 2003), particularly when bleaching results from the loss of the symbiont cell.

Bleaching is suspected to result through processes such as: exocytosis of the symbiont cell, apoptosis or necrosis of the host cell containing the symbiont, pinching off the portion of the host cell occupied by the symbiont, or detachment of the entire host cell (Gates et al. 1992). Coral bleaching is generally believed to be triggered by oxidative stress that can be either host- or symbiont-derived (Oakley and Davy 2018). However, the role of symbiont cell surface proteins and how they change in response to stimuli is worth consideration, as this is a putative site of signal and nutrient exchange and may, therefore, represent an important point of interaction between partners.

Cell surface studies are gaining in importance and it is becoming clear that the extracellular surface of all organisms is actively maintained. In plants, for example, the cell wall is constantly remodeled to accommodate osmotic pressure (Deniaud-Bouët et al. 2017). Additionally, new and emerging roles for processes like extracellular redox in maintaining the intracellular redox state (Jones et al. 2015) are coming to light. In the coral-Symbiodiniaceae symbiosis, cell surface elements such as lectins and glycans have been historically viewed as crucial for successful symbiont infection of host tissues (Logan et al. 2010; Davy et al. 2012; Jimbo et al. 2010; Koike et al. 2004), although recent evidence suggests that these may be more important for post-phagocytic processes (Parkinson et al. 2018). Regardless, despite the importance of this subcellular locale, the symbiont cell-surface protein composition remains unknown. Furthermore, it is also unknown if this composition is dynamic in nature, and if so, what role that may play during thermally induced symbiosis breakdown.

To investigate this, the present study sought to determine the effect of elevated temperature on protein composition at the symbiont cell surface in vitro. Proteins were isolated from the symbiont cell surface using a membrane-impermeable biotin probe (Li et al. 2013; Pelz et al. 2018) on intact symbiont cells with three hypotheses in mind: (1) that the cell-surface protein abundance will change under elevated temperature; (2) that the symbiont cell surface will exhibit stress-mitigating mechanisms under elevated temperature; and (3) that changes occurring at the symbiont cell surface have the potential to elicit immune responses from a host. We found evidence to support all three of these hypotheses and document the first cell surface protein composition of a Symbiodiniaceae species. This work highlights host–symbiont interaction mechanisms that may be important during symbiosis breakdown.

Methods

Study design

Breviolum psygmophilum (formerly Symbiodinium psygmophilum, clade B (LaJeunesse et al. 2018); isolated from Oculina diffusa, Western Atlantic, Bermuda) was obtained from T. LaJeunesse (Pennsylvania State University) and grown in the Mydlarz lab. Three replicate cultures for the control treatment and three replicate cultures for the heat treatment were grown up to a target density of 500,000 cells/ml. Cultures were grown at 26 °C under a 12-h light/dark cycle at an irradiance of approximately 55 \(\upmu\)mol quanta m−2 s−1 (measured with an LI-COR model LI-250 light meter, LI-COR Environmental) in ASP-8A medium (Chang et al. 1983; McGinty et al. 2012). Light conditions remained the same throughout the experiment. Once cultures reached target density, each replicate culture was separated into three 5 ml subsample volumes (referred to as replicate subsamples henceforth) that were collected and processed at 0-h, 12-h, and 24-h time points after exposure to treatment. 0-h subsamples were collected and processed immediately and were not exposed to either treatment condition.

Replicate subsamples were placed in individual water baths (e.g., 12-h and 24-h replicate subsamples for replicate 1 were in an individual water bath, while 12-h and 24-h replicate subsamples for replicate 2 were placed in a separate water bath) (supplementary figure S1). Water baths were fitted with heaters set to target temperatures (26 °C ± 1 °C for control treatment and 32 °C ± 1 °C for heat treatment to simulate bleaching-inducing conditions). Air pumps were used to circulate water for an even temperature distribution. All water baths were initially set to control treatment temperatures. Once cultures were placed in their respective water baths, heat treatment water baths were subjected to a 2-h temperature ramp. Control treatment water baths remained at a constant temperature during this time. Time of exposure began once the temperature ramp was completed.

Experimental sample processing

At time of collection, replicate subsamples were removed from treatment and pelleted in a benchtop centrifuge at 959×g for 10 min. At this speed, it was microscopically confirmed that Symbiodiniaceae cells do not lyse. Intact cells were washed thoroughly using sterile 1× PBS at room temperature. All washes and incubations were conducted using 1× PBS when specified. Although 1× PBS may induce hypoosmotic stress, differences in protein expression and abundance should still result from temperature treatments as each sample was handled identically. Additionally, the exposure time during washing and incubation was relatively minimal.

Replicate subsamples were then incubated for 30 min at room temperature with a membrane-impermeable, cleavable biotin probe (Sulfo-NHS-SS-Biotin, G-Biosciences) (supplementary figure S2). This probe is used commonly in cell-surface studies (e.g., Elschenbroich et al. 2010; Li et al. 2013; Pelz et al. 2018) due to its hydrophobic nature. In this system, the biotin probe is conjugated to a sulfonated NHS ester by a cleavable disulfide linker (Elschenbroich et al. 2010). The sulfo-group transfers hydrophobicity to the probe, and the cleavable property negates the need to chemically unbind the biotin motif from avidin when rescuing isolated proteins (Elschenbroich et al. 2010). This was ideal for a cell-surface study in Symbiodiniaceae, because cell walls are negatively charged (Shomer et al. 2003) and, as such, the negatively charged biotin probe used would not be able to cross it. The probe will, therefore, primarily detect proteins on the external face of the symbiont cell wall.

To quench the biotinylation reaction after incubation with the biotin probe, 25 mM tris buffer (Trizol, Sigma-Aldrich) was added. After 5 min at room temperature, time point subsamples were spun down in a benchtop centrifuge at 15,330 × g for 5 min, washed thoroughly with sterile PBS, and resuspended in sterile PBS after the final wash (Howes et al. 2010; Jo et al. 2010; Suzuki et al. 2010). Breviolum psygmophilum cells were then lysed via bead beating with glass beads for 1 min.

Isolation of biotin-labeled cell surface proteins was conducted by avidin affinity purification using spin columns and monomeric avidin (G-Biosciences). Following established protocols (Lee et al. 2009; Shimus et al. 1985), replicate subsamples were incubated in spin columns with avidin for 30 min at room temperature. Avidin-bound proteins were freed using 50 mM DTT to cleave the probe’s disulfide bond via reduction. Freed proteins were rescued via spinning out into a collection tube. The resulting protein isolates were considered cell-surface protein enriched.

Protein isolates were then precipitated using a modified chloroform and methanol protocol (Ferro et al. 2000), wherein the reagents were added to the protein samples in a 3:4:1 v/v/v ratio (protein/methanol/chloroform) and centrifuged in a benchtop centrifuge at 15,330 × g at 4 °C. DTT was removed by resuspending the precipitated protein pellet in 10% SDS and undergoing a second round of precipitation. Once DTT was removed, pellets were resuspended in 1% SDS and protein quantification of replicate subsamples was performed using a BCA assay (G-Biosciences) in a microplate spectrophotometer (BioTek).

Nanospray-LC–MS/MS

4 ng of protein per replicate subsample were tryptically digested in solution following established protocols (Chakrabarty et al. 2016). In summary, protein isolates were incubated with 10 mM DTT at 56 °C for 45 min under continuous agitation. Protein isolates were then incubated with 10 mM iodoacetamide in the dark at room temperature for 30 min. 50 mM ammonium bicarbonate was then added followed by trypsin in a 1:50 w/w ratio (trypsin/protein) and incubated overnight at 37 °C under continuous agitation. After tryptic digestion, 0.1% formic acid was added to neutralize the pH and protein isolates were then dehydrated in a speedvac (Vacufuge plus, Eppendorf). Once dehydrated, protein isolates were reconstituted in 0.1% formic and introduced to a Velos Pro Dual-Pressure Linear Ion Trap Mass Spectrometer (ThermoFisher Scientifc). Nanospray-LC–MS/MS was carried out using a data-dependent protocol. Protein fragmentation was achieved by collision-induced dissociation (CID).

Isolated protein sequences were identified from mass spectra using Proteome Discoverer software (ver. 2.0, ThermoFisher Scientifc). Using the Sequest HT algorithm within the software, spectra were matched against a translated Breviolum psygmophilum transcriptome publicly available from Reef Genomics databases (Liew et al. 2016; Parkinson et al. 2016). Sequest HT criteria were as follows: the proteolytic enzyme was indicated as trypsin; two missed cleavages were allowed; precursor mass range of 350–5000 Da; fragment mass tolerance of ± 2.5 and 0.6 Da; peptide charges excluded + 1 (Kamal et al. 2018).

Data set building

A decoy search strategy was employed in Proteome Discoverer software using a 5% False Discovery Rate (FDR) (Wilhelm et al. 2014). Breviolum psygmophilum proteins in replicate subsamples were considered identified with high confidence at \(\le\) 5% FDR if they met either of the following criteria: (a) ≥ 2 peptides were detected in ≥ 2 replicates; or (b) ≥ 1 peptide was detected in all three replicates (Kamal et al. 2018). Using these criteria, a data set of 147 proteins was compiled (supplementary table S1). Proteome Discoverer utilizes the label-free method of spectral counting to quantify protein expression (peptide spectral matches; i.e., PSMs). PSMs of confidently identified proteins were normalized as % total PSMs per replicate subsample (Kamal et al. 2018).

To annotate the 147 proteins identified within the B. psygmophilum transcriptome, their sequences were BLASTed against the Uniprot KB Swiss-Prot database. An e value ≥ e−5 was considered a confident annotation (Mayfield et al. 2018). If a B. psygmophilum sequence could not meet the criteria for confident annotation, it was BLASTed against the entire Uniprot KB database (i.e., Swiss-Prot and TrEMBL databases).

Once annotated, GO terms (Gene Ontology) and literature searches were utilized to categorize the proteins into functional groups based on their roles when expressed at the cell surface or secreted into the extracellular space. In instances where GO terms agreed with the known extracellular role, the GO term was used to group proteins of the same function. In all other instances, proteins were grouped according to known functions documented in the literature (supplementary table S2). Literature searches were conducted by providing the search term “extracellular”, “secreted”, or “cell surface” before the protein name, and only manuscripts found with these searches were used to determine if proteins possessed a cell surface or extracellular presence. Additionally, if these searches yielded proteins known to interact with a host immune system when present at the cell surface or in the extracellular space, then it was classified as either immune-activating, -regulating, or -suppressing (e.g., cell surface heat shock protein 70 promotes phagocytosis and is, therefore, classified under immune activation). The regulatory category encompassed proteins with known regulatory roles or whose extracellular effects on a host immune system were conflicting.

These methods found that 67 of the 147 proteins identified have a documented extracellular function in the literature and 12 proteins have an extracellular presence but an unknown function. Statistical analyses were carried out on these proteins (79 in total) to determine how treatment and length of treatment affected the composition of proteins and protein function at the cell surface. 47 proteins were known chloroplast constituents and were, therefore, considered contamination. Chloroplast contamination is not uncommon in cell wall/cell surface studies in dinoflagellates (Li et al. 2012; Wang et al. 2004a). Chloroplast contamination is likely due to the extreme peripheral position of the large dinoflagellate chloroplasts (Lee et al. 2014). Chloroplast constituents did not change with heat (supplementary figure S3) and were not considered further.

Validation of NodG homolog and Nod homolog searches among Symbiodiniaceae

A nod factor G (nodG) homolog was identified (comp8899_c0_seq1.p1) in Breviolum psygmophilum and classified under signal transduction (Table 1). Because of its putative roles in symbiosis, a special attention was paid to the validation of its presence within the B. psygmophilum transcriptome. The presence of nod factors in other Symbiodiniaceae was also investigated for this reason. A Pfam protein domain search was compared between the B. psygmophilum nodG and the reviewed uniprot nodG sequence which it was matched to via BLAST (nodG, uniprot ID P72332 from from Rhizobium sp. strain N33). Additionally, an EMBL-EBI pairwise sequence alignment using the EMBOSS Water algorithm was conducted with the following criteria: EBLOSUM62 was used at the matrix, gap penalty was set to 10, and extend penalty was set to 0.5 (Madeira et al. 2019).

Table 1 Protein functional groups
Full size table

Investigations into nod factor presence in other Symbiodiniaceae were carried out on six species whose genomes or transcriptomes are publicly available on Reef Genomics databases: Breviolum aenigmaticum, B. minutum, B. pseudominutum, B. psygmophilum, Cladocopium (species unknown), and Fugacium kawagutii. Symbiodiniaceae sequences were BLASTed against a databased composed of the 148 nod factor sequences available through the Uniprot KB database. An e value ≥ e−5 cut-off was imposed. The top ten strongest BLAST hits plus the sequence of the nodG homolog identified in this study were then phylogenetically compared using the Clustal Omega algorithm and bacterial nodI as an outgroup (unprot ID Q39GT7, Burkholderia lata) (Madeira et al. 2019).

Finally, the same six Symbiodiniaceae species were queried for the presence of nod factors A, B, C, and D (NodABCD). NodABCD is necessary for the synthesis of the lipochitooligosaccharide backbone of all nod factors (Roche et al. 1996), which are later modified by other nod factors for host specificity (Wang et al. 2018). Symbiodiniaceae genomes/transcriptomes were BLASTed against a database comprised of all available NodABCD sequences available through the Uniprot KB database. An e value ≥ e−5 cut-off was again imposed.

Statistical analysis

All statistical analyses were conducted using R statistical software (R Development Core Team 2015). Identified proteins were divided into groups based on protein function (i.e., functional groups). Bray–Curtis distances were utilized by similarity percentages analysis (i.e., SIMPER analysis) to calculate the strongest drivers of differences observed between control and heat-treated samples (Clarke 1993; Warton et al. 2012). From SIMPER analyses, the most influential functional groups and/or individual proteins within a functional group were determined. SIMPER was carried out using the ‘simper’ function in the R package ‘vegan’ (Oksanen et al. 2018). PCA was conducted on the cumulative protein abundance for functional groups of interest using the ‘ggbiplot’ function in the R package ‘ggbiplot’ (Vu 2011).

To address the possible correlation in protein expression within resampled experimental units, repeated measures MANOVA was conducted using the ‘RM’ function in the R package ‘MANOVA.RM’ (Friedrich et al. 2018). Non-parametric t tests were then conducted on the cumulative protein abundance for influential functional groups (e.g., cumulative abundance of proteins with immune modulatory functions in control vs. heat-treated samples). Within functional groups, non-parametric t tests were also carried out on the abundance of individual proteins that were determined to be influential by SIMPER (e.g., abundance of the protein V-type H+-ATPase in control vs. heat-treated samples).

To quantify the biological significance of the differences observed in protein abundance observed, effect size was calculated using Cohen’s d estimation (Cohen 1992a, b; Rice and Harris 2005). Effect size is defined as the discrepancy between the null hypothesis and the alternate hypothesis (Cohen 1992a). The small sample size (n = 3 per treatment) in combination with the variability observed between replicates can potentially underinflate statistical significance at α = 0.05. This can, therefore, obscure findings of biological importance. Effect size is thus reported in addition to p values to provide more transparent and accurate statistical interpretation (Greenland et al. 2016; Wasserstein and Lazar 2016). Cohen’s d was calculated using the ‘cohen. d’ function in the R package ‘EffSize’ (Torchiano 2018). A small-effect size is a Cohen’s d ~ 0.2, a medium-effect size is a Cohen’s d ~ 0.5, and a large-effect size is a Cohen’s d ~ 0.8 (values noticeably lower than 0.2 are considered negligible, while values noticeably greater than 0.8 are considered very large-effect sizes) (Rice and Harris 2005; Torchiano 2018).

Results

Constitutive cell-surface protein composition of Breviolum psygmophilum

Control samples and heat samples at 0 h of exposure to treatments are pooled and considered the constitutive state of the Breviolum psygmophilum cell surface. A total of 147 proteins were identified at \(\le\) 5% FDR (supplementary table S1). 79 identified proteins are known to be either secreted or actively released into the extracellular space, or expressed at the cell surface in various prokaryotic and eukaryotic species (supplementary table S2). These 79 proteins were used in statistical analyses for this study. 12 proteins had either no literature documentation of cell surface presence or the literature concerning the protein was conflicting. These proteins were also considered contamination. Three proteins could not be identified by BLAST.

The 79 proteins known to occur at the cell surface or in the extracellular space encompassed nine functional groups: protein folding, cell structure, adhesion, CO2 uptake, extracellular ATP synthase, extracellular redox, signal transduction, ion homeostasis, and an unknown category representing proteins whose function is unknown when expressed in the extracellular space (they will not be addressed further as a result; Table 1). Adhesion proteins represent the most abundant functional group at the cell surface of B. psygmophilum, while proteins representing the ion homeostasis functional group were least abundant (Fig. 1).

Fig. 1
figure 1

Stacked plot depicting relative abundance of functional groups. Percentages reported are based on the average abundance of each functional group between treatment replicates. Adhesion: n = 14 proteins; cell structure: n = 9; CO2 uptake: n = 2; extracellular ATP: n = 7; extracellular redox: n = 11; signal transduction: 9; ion homeostasis: n = 2; protein folding: n = 14; unknown = 12. Y-axis represents percent normalized peptide spectral matches (PSMs)

Full size image

Nodulation factors present in Breviolum psygmophilum and other Symbiodiniaceae species

The B. psygmophilum nodG homolog identified is, indeed, a putative nodulation factor. Pairwise sequence alignment between B. psygmophilum nodG homolog and Rhizobium sp. strain N33 nodG (uniprot ID P72332) achieved a high sequence alignment: 48.4% identity match, 66.5% similarity, and 3.6% gaps (Fig. 2a). The overall alignment score was 547. Pfam searches between the Breviolum psygmophilum nodG homolog and the Rhizobium sp. strain N33 nodG also displayed identical protein domain structure (Fig. 2b).

Fig. 2
figure 2

Validation of nodG in Breviolum psygmophilum a sequence alignment between B. psygmophilum nodG homolog and Rhizobium sp. nodG (uniprot ID P72332); b Pfam protein domains present in B. psygmophilum nodG homolog and Rhizobium sp. nodG; c Phylogenetic analysis of top ten strongest BLAST hits for nod factors across Symbiodiniaceae species. (*) denotes B. psygmophilum nodG identified in the current study

Full size image

Potential nodulation factors are ubiquitous in Symbiodiniaceae, with a total of 6,557 matches identified across the six Symbiodiniaceae species investigated (supplementary file 1). Potential NodABCD homologs were found in all species investigated as well (supplementary file 2). All six species were represented in the top ten strongest hits found via BLAST. The Nod sequences investigated grouped according to Symbiodiniaceae species (Fig. 2c).

Response of Breviolum psygmophilum cell-surface proteins to heat

Proteins at the cell surface were responsive to temperature stress, particularly when time is taken into account (Table 2). One protein was identified as uniquely present in the control samples (comp36516_c0_seq1.p1; ATP synthase subunit) and one protein was identified as uniquely present in the heat-treated samples (comp35699_c0_seq1.p1; calreticulin). For the shared proteins (i.e., proteins found in both control and heat-treated samples), total protein abundance does not differ between treatments until after 24 h of exposure to heat treatment (p = 0.400, Cohen’s d = − 0.440, Fig. 3). Differences are primarily seen in the abundance between the proteins uniquely expressed in either the control or heat-treated samples: after 24 h of exposure to heat treatment, abundance of the heat-treatment-unique protein was greater than that of the control treatment (p = 0.176, Cohen’s d = − 0.894, Fig. 3). The control-treatment-unique protein was not present after 24 h.

Table 2 Statistical comparison of treatment on all response variables
Full size table
Fig. 3
figure 3

Total abundance of proteins found in each treatment. Left: total abundance of proteins for sequences found only in either control (n = 1) or heat-treated (n = 1) samples. Right: total abundance of proteins for sequences found in both control or heat-treated samples (n = 77). Blue lines with circles represent control treatment; red lines with triangles represent heat treatment. At 24 h of exposure, shared proteins do not differ in abundance (p = 0.400, Cohen’s d = − 0.440), but unique proteins do (p = 0.176, Cohen’s d = − 0.894). Y-axis represents normalized peptide spectral matches (PSMs)

Full size image

Total abundance of proteins within functional groups showed differences through time and by treatment (Fig. 4). The functional groups driving the differences observed between control and heat-treated samples were, in order of most influential: protein folding, cell structure, adhesion, CO2 uptake, and extracellular ATP synthase (Table 2). When the most influential functional groups were used to characterize the protein abundance data, only those samples that belong to the 24-h heat treatment were distinct (Fig. 5). This demonstrates that the exposure to 1X PBS was not long enough to substantially affect protein expression. Only those differences at the 24-h time point were considered in further analyses.

Fig. 4
figure 4

Total abundance of proteins found in each treatment for each functional group through time. Blue hues represent control treatments; red hues represent heat treatments. Adhesion: n = 14 proteins; cell structure: n = 9; CO2 uptake: n = 2; extracellular ATP: n = 7; extracellular redox: n = 11; signal transduction: 9; ion homeostasis: n = 2; protein folding: n = 14; unknown = 12. Y-axis represents normalized peptide spectral matches (PSMs)

Full size image
Fig. 5
figure 5

PCA plot comparing treatments to functional groups

Full size image

Protein responses in the most influential functional groups to heat

The protein-folding functional group was represented by 14 unique proteins that represented five protein types (Table 1). The top two most influential proteins in the protein-folding group were heat shock protein (HSP) 70 and HSP 90 (Table 3, SIMPER cumsum ~ 80%). After 24 h of exposure to heat treatment, both proteins were in greater abundance when compared to control samples (p = 0.400 and Cohen’s d = − 1.174, p = 0.100 and Cohen’s d = − 2.870; respectively; Table 3, Fig. 6).

Table 3 Statistical comparison of two most influential proteins within each functional group
Full size table
Fig. 6
figure 6

Top two most influential proteins for each functional group according to simper. Purple: protein folding; orange: cell structure; red: Adhesion; olive green: CO2 uptake; green: extracellular ATP; teal: extracellular redox; light blue: signal transduction; blue: ion homeostasis. (*) represents those differences between treatment that are stastically significant; (a) represents those differences with large-effect sizes. Effect size was calculated using Cohen’s d estimation. Y-axis represents normalized peptide spectral matches (PSMs)

Full size image

Cell structure was represented by nine unique proteins that represented four protein types (Table 1). Tubulin and a major outer membrane lipoprotein were the two most influential proteins (Table 3, SIMPER cumsum ~ 83%). After 24 h of exposure to heat treatment, tubulin abundance decreased (p = 0.400 and Cohen’s d = 0.981, Table 3, Fig. 6), while the abundance of the major outer membrane lipoprotein increased (p = 0.100 and Cohen’s d = − 1.243, Table 3, Fig. 6).

Adhesive proteins were represented by 14 unique proteins that encompassed six protein types (Table 1) and were most influenced by enolase and triosephosphate isomerase (TPI) (Table 3, SIMPER cumsum ~ 55%). After 24 h of exposure to heat treatment, neither protein differed in abundance compared to control treatments (p = 0.633, Cohen’s d = − 0.441 and p = 0.960, Cohen’s d = 0.043; respectively; Table 3, Fig. 6).

CO2 uptake was represented by two unique proteins that were both identified as carbonic anhydrase by BLAST (Table 1). Extracellular ATP synthase was similarly represented by four unique proteins that were all identified as ATP synthase subunits by BLAST (Table 1). After 24 h of exposure to heat treatment, carbonic anhydrase did not differ from control samples in abundance (p = 0.931, Cohen’s d = 0.078, Table 3, Fig. 6); however, variation between samples was much higher for heat-treated samples vs. control samples. ATP synthase decreased in abundance in heat-treated samples compared to control samples (p = 0.378, Cohen’s d = 0.814, Table 3, Fig. 6).

Responses of extracellular redox, signal transduction, and ion homeostasis proteins to heat

Extracellular redox was represented by 11 unique proteins that represented seven protein types (Table 1). Fumarate reductase and cytochrome c were the most influential proteins within the extracellular redox functional group (Table 3, SIMPER cumsum ~ 64%). Fumarate reductase was found in greater abundance in heat-treated samples compared to control samples, while cytochrome c did not differ in abundance, but was greater in between-sample variation (p = 0.700, Cohen’s d = − 0.954 and p = 1.000, Cohen’s d = 0.214; respectively; Table 3, Fig. 6).

The signal transduction group was represented by nine unique proteins that represented eight protein types (Table 1). The 14-3-3-like protein and calreticulin were the most influential proteins within this functional group (Table 3, SIMPER cumsum ~ 40%). 14-3-3 increased in abundance after 24 h of exposure to heat treatment, while calreticulin was only found in heat-treated samples (p = 0.200, Cohen’s d = − 1.459 and p = 0.197, Cohen’s d = − 1.564; respectively; Table 3, Fig. 6). Ion homeostasis was represented by a K+-stimulated sodium pump and a V-type proton ATPase (Table 1). Only the V-type proton ATPase was found in greater abundance after 24 h of exposure to heat treatment (p = 0.391, Cohen’s d = − 0.865, Table 3, Fig. 6).

Immune modulatory proteins present at the Breviolum psygmophilum cell surface and their response to heat

Proteins known to modulate a host immune system via immune activation, suppression, or regulation were identified in cell-surface protein isolates (supplementary table S3). Proteins were assigned to each category based on information available within the literature (supplementary table S3). The regulatory category encompassed proteins with known regulatory roles or whose extracellular effects on the host immune system were conflicting.

Immune modulatory proteins were affected by both treatment and time (RM MANOVA, Table 4). Differences only at the 24-h time point are addressed: cumulative abundance of all immune-activating proteins increased in response to heat (p = 0.1, Cohen’s d = − 2.131, Table 4, Fig. 7), while the cumulative abundance of all immune-suppressing proteins decreased (p = 0.505, Cohen’s d = 0.816, Table 4, Fig. 7); immune regulatory proteins were not responsive to heat (p = 0.4, Cohen’s d = 0.585, Table 4, Fig. 7).

Table 4 Statistical comparison of treatment on immune modulating proteins
Full size table
Fig. 7
figure 7

Response of immune modulation proteins after 24 h exposure to heat. Blue boxes represent control treatments; red boxes represent heat treatments; (*) represents those differences between treatment that are stastically significant; (a) represents those differences with large-effect sizes. Activation: n = 11; regulation: n = 3; suppression: n = 1. Effect size was calculated using Cohen’s d estimation. Y-axis represents normalized peptide spectral matches (PSMs)

Full size image

Discussion

The nature of the coral-Symbiodiniaceae symbiosis implies an important role for the symbiont surface as a possible point of interaction between the coral cell and the intracellular symbiont. As such, proteomic investigation at this locale is important in understanding potential mechanisms in partner dynamics. In this study, the Breviolum psygmophilum transcriptome developed by Parkinson et al. (2016) was used as a database to inform protein identification from cultured B. psygmophilum, and only proteins encoded within the B. psygmophilum genome were identified as a result. Our investigation reveals elements of host-specific interaction mechanisms and shows cell-surface proteins are responsive to heat stress. We also show that stress-mitigating mechanisms have the potential to influence a host immune system.

The biotin probe utilized within this study ensures that a cell-surface-enriched protein fraction was analyzed due to the probe’s hydrophobic nature (Elschenbroich et al. 2010). Identified proteins were further validated for cell surface presence or absence and classified into functional groups based on rigorous, non-biased literature searches. While we did observe the presence of proteins that are not exclusively extracellular, it is coming to light that many proteins can have multiple subcellular locations (e.g., Jeffery 2015; Gancedo et al. 2016), and that many proteins can be destined for the cell surface and extracellular space despite lacking a proper signal peptide (Robinson et al. 2016). Exosomes and microvesicular bodies (MVBs) are gaining momentum as a mechanism for the transport of intracellular proteins to the cell surface, and may represent a possible explanation for the presence of these proteins in the cell surface-enriched protein fraction of B. psygmophilum. Indeed, MVBs are acknowledged as ubiquitous across the tree of life and have been observed traversing the cell walls of plants, fungi, and bacteria (Rodrigues et al. 2015; Samuel et al. 2015). Alternatively, the presence of these proteins may also be explained by cell lysis in culture, and adherence of the proteins to the cell surface of intact cells. However, the thorough washings conducted during sample processing minimize this as a possibility. Furthermore, the lack of canonical intracellular markers that do not have a documented cell surface presence [e.g., CYP450, ER (Scotcher et al. 2017); histone, nuclear (Yang et al. 2004); G6Pase, ER (Scotcher et al. 2017); rab proteins, endosomal (Smith et al. 2016); ras proteins, endosomal (Arike and Peil 2014); RPL11, ribosomal (Brodersen and Nissen 2005); HSP60, mitochondrial (Bonifati et al. 2003)] indicate that intracellular contamination was minimal.

Despite the probe’s hydrophobic properties, a large number of chloroplast constituents were isolated. Chloroplast contamination is not uncommon in cell wall/cell surface studies in plants and algae (e.g., Li et al. 2012; Wang et al. 2004a; Calderon-Rodrigues et al. 2014). Some authors have explained their presence as a result of mistargeting by the protein’s signal peptide (Slabas et al. 2004). Indeed, proteins destined for the mitochondria can be mistargeted to the cytosol (Wrobel et al. 2015), providing precedence for such a phenomenon. Others have hypothesized a role for the underlying biology of free-living dinoflagellates, whereby individual cells are linked together via an attachment pore are broken apart by a mechanical disturbance that leads to chloroplastic leakage (Li et al. 2012). An alternative possibility for their presence in the current study may be related to the extreme peripheral position of the chloroplast in B. psygmophilum (Lee et al. 2014), which may result in co-isolation with proteins specifically bound by the cell surface probe. We currently cannot explain definitively why this was observed. Because chloroplast constituents did not change with heat, they were considered as random contamination that does not represent the intracellular state and were not addressed further.

Constitutive cell-surface proteins of Breviolum psygmophilum carry out essential functions

The extracellular matrix and cell membrane carry out important functions. They are often viewed as a first line of defense against assaults on cellular integrity (Deniaud-Bouët et al. 2017) and are also responsible for waste exchange and nutrient uptake (Hahn and Mendgen 2001). In addition, they are important for modulating osmotic pressure (Deniaud-Bouët et al. 2017) and sensing cues from the extracellular environment that lead to cell growth or differentiation (Deniaud-Bouët et al. 2017). Recent advances in cell-surface research are highlighting this dynamic nature across the tree of life (Shi et al. 2016; Lemmon et al. 2016). It is now clear that cells actively maintain the cell surface and the extracellular space directly adjacent to the cell surface to preserve homeostasis.

At the constitutive state, the adhesive functional group had the highest abundance at the Breviolum psygmophilum cell surface and was primarily represented by glycolytic proteins. When expressed at the cell surface, many of these proteins bind to laminin (Amblee and Jeffery 2015), a major constituent of animal extracellular matrices, including those of Cnidaria (Sarras and Deutzmann 2001), and plasminogen (Gancedo et al. 2016), the inactive form of the serine protease plasmin (Aisina and Mukhametova 2014). Of interest is that plasmin is important for the degradation of animal cell extracellular matrices (Chana-Muñoz et al. 2019), and plasminogen-binding proteins have been utilized by a number of pathogens to promote host invasion (Ayon-Nunez et al. 2018). It is, however, unclear if invertebrates possess plasmins (Chana-Muñoz et al. 2019; Chao et al. 2012). The presence of proteins in B. psygmophilum that can potentially bind plasminogen may, therefore, simply be relevant for adhesion alone rather than host infection.

The abundance of the adhesive functional group was followed closely by the cell structure and CO2 uptake proteins. CO2 uptake was represented by two proteins identified as carbonic anhydrase (CA). In the unicellular green alga Dunaliella tertiolecta, cell-surface CA assists in the uptake of CO2 from the surrounding water (Aizawa and Miyachi 1984). The same role is carried out by cell-surface CA in various other phytoplankton (Mustaffa et al. 2017), including Symbiodiniaceae (Yellowlees et al. 1993; Karim et al. 2011). Cell-surface CA is responsible for dehydrating HCO3− to CO2, which is then diffused into the cell (Aizawa & Miyachi 1984), although it appears to be the case that symbionts in hospite do not readily utilize HCO3− (Yellowlees et al. 1993), and may instead rely on the cnidarian host to concentrate extracellular HCO3− to provide a sufficient level of CO2 for photosynthesis (Allemand et al. 1998; Furla et al. 2000). This suggests that cell-surface CA may be more relevant in the free-living state. In either case, the identification of cell-surface CA at the cell surface of B. psygmophilum corroborates a growing body of literature demonstrating the ubiquity of proteins with pleiotropic (Orjalo et al. 2009; Ebnet 2017) and moonlighting (Jeffery 2015; Gancedo et al. 2016) properties. This is with particular regard to extracellular protein function (Wang and Jeffery 2016).

Other emerging roles for the cell surface are extracellular ATP synthesis (Federica and Antonio 2018) and extracellular redox processes (Banerjee 2012). Extracellular ATP synthase (i.e., eATP synthase) has only recently been accepted as a truly functional complex when expressed at the cell surface (Federica and Antonio 2018) and, as such, is poorly understood. They do, however, have ion regulating properties that are believed to result from the movement of hydrogen ions into and out of the cell during the synthesis and hydrolysis of ATP (Federica and Antonio 2018). There may be as of yet unknown functions at the cell surface of B. psygmophilum that rely on eATP synthase or eATP. Extracellular redox, on the other hand, is known to be important in maintaining the intracellular redox environment (Banerjee 2012). Interestingly, extracellular redox modulation has roles in inflammatory processes (Carta et al. 2009) and may, therefore, be an important aspect governing symbiosis dynamics between corals and B. psygmophilum.

Nod factors are ubiquitous across Symbiodiniaceae species

Nod factors are secreted lipochitooligosaccharide molecules that are primarily characterized in the plant endosymbiotic bacteria, Rhizobium. In the plant-rhizobia model, flavonoids are secreted by the host plant to attract bacteria (Hassan and Mathesius 2012), and the detection of these flavonoids by the bacteria in turn produces Nod factors that are then secreted by the bacteria (Oldroyd and Downie 2004). A complete Nod factor is synthesized by the “common” Nod factors (Nods A,B,C,D), which are responsible for synthesizing the chitin backbone of the molecule (Roche et al. 1996), and the “specific” Nod factors (all other nod factors), which modify the chitin backbone for host specificity (Wang et al. 2018). Upon secretion, Nod factors are perceived by plant LysM proteins, which consist of an extracellular LysM motif and an intracellular protein kinase domain with autophosphorylation properties (Madsen et al. 2011), and perception of a compatible bacteria–Nod factor combination by the plant tissues then causes the iconic root hair deformation and nodulation characteristic of a successful symbiosis (Oldroyd and Downie 2004). Importantly, incompatible Nod factors will prevent a symbiosis from forming (Oldroyd and Downie 2004).

The presence of the nodG homolog in Breviolum psygmophilum corroborates the previous reports of Nod factors present in Symbiodiniaceae (e.g., Lin et al. 2015; Weston et al. 2012). The nodG homolog identified at the B. psygmophilum cell surface shows high homology to the reviewed Rhizobium sp. nodG. The occurrence of other Nod factors across Symbiodiniaceae was thus further explored, as well as the presence of the common Nod factors due to their necessity for the synthesis of the Nod factor backbone. Each of the six Symbiodiniaceae species investigated possessed sequences homologous to Nods A, B, C, and D within their respective genomes/transcriptomes, and the subset of sequences that were phylogenetically investigated show Nod factors grouping by species.

Lectin and glycan interactions between Symbiodiniaceae and the coral host have historically been the favored model for partner specificity (Logan et al. 2010; Wood-Charlson et al. 2006). However, recent evidence suggests that other aspects may take part in this process, as manipulation of Symbiodiniaceae glycans does not appear to alter host infection rates (Parkinson et al. 2018). While at present, it is unknown if Nod factors in Symbiodiniaceae carry out the same functions as in Rhizobium, nor if the coral host has the machinery necessary to detect and/or respond to them, the presence of Nod factor-like proteins at the B. psygmophilum cell surface opens new lines of inquiry when considering alternative elements for partner selection.

The cell-surface of Breviolum psygmophilum is responsive to heat stress over time

Heat affects the Breviolum psygmophilum cell surface primarily after 24 h of exposure to heat treatment. As such, comparisons were made at the 24-h time point. Protein folding was the most influential functional group driving the differences between control and heat-treated samples. This resulted from an increase in heat shock protein (HSP) 70 and HSP 90 at the B. psygmophilum cell surface. HSPs are commonly upregulated in response to stress (Wiersma et al. 2015) and function to protect the existing proteins from denaturing (Hasanuzzaman et al. 2013; Wang et al. 2004b). Here, we show that cell-surface HSPs are a key response to heat stress in B. psygmophilum. HSP action may be facilitated by eATP synthase, as both HSP 70 and HSP 90 require ATP to bind target proteins and carry out chaperone functions (Hasanuzzaman et al. 2013; Wang et al. 2004b). It may be of biological importance that eATP synthase decreases in response to heat, while HSP proteins increase. Uncoupling of the two may reflect dysfunction brought on by heat stress. Regardless, the observation of increased HSPs demonstrates that stress experienced by B. psygmophilum under elevated temperatures manifests at the cell surface and has implications for an intracellular symbiosis in hospite.

Tubulin also decreased after 24 h of exposure to heat. Dynamic tubulin modulation is important for cell wall remodeling (Chan et al. 2010; Ochs et al. 2014), and decreases observed in response to heat are somewhat paradoxical within the context of HSP increases. This is likely because remodeling to accommodate responses such as protein translocation and insertion took place prior to 24 h. Cell wall remodeling in B. psygmophilum may, therefore, occur during early responses to stress. Conversely, the abundance of a major outer membrane lipoprotein increases in response to heat after 24 h of exposure. In the dinoflagellate cell wall, these proteins are important for protein binding, lipid anchoring, and calcium binding (Wang et al. 2011), and may have similar roles in the B. psygmophilum cell wall.

Increases in extracellular redox proteins demonstrate an increased need by B. psygmophilum to maintain their intracellular redox environment, and may result from an increased energy demand that can potentially fatigue redox gradients across the mitochondrial membrane (Banerjee 2012). Related is the increase in the V-type H+-ATPase, a protein responsible for the transport of protons into and out of the cell (Miles et al. 2017). Responses by both the extracellular redox and ion homeostasis functional groups demonstrate the importance of the cell surface in maintaining the intracellular environment. Importantly, H+-ATPase was thought to only be expressed by Symbiodiniaceae when in a symbiotic state (Bertucci et al. 2010; Miles et al. 2017). However, this was supported by gene expression alone. Using proteogenomic methods, we show that this protein is indeed found in a Symbiodiniaceae species outside of a symbiosis and that it is responsive to heat stress.

Proteins known to stimulate host immune responses are present at the cell surface of Breviolum psygmophilum and increase with heat

Proteins known to modulate a host immune system were present at the Breviolum psygmophilum cell surface. Within the 15 proteins detected, three categories could be identified based on literature searches: immune activation (i.e., eliciting an immune response from a host upon detection); immune regulation (i.e., roles in immune activation and resolution); and immune suppression (i.e., preventing or hindering a host immune response). The majority of these proteins (11/15) are known to activate a host immune system. Three proteins, ubiquitin (Majetschak 2011), ATP synthase (Chivasa et al. 2009), and peptidyl-prolyl cis–trans isomerase (Ünal and Steinert 2014), have roles in immune regulation. One protein, nicotinamide phosphoribosyl transferase, is known to suppress a host immune system upon extracellular detection (Audrito et al. 2015).

Although the effect of specific proteins on the coral immune system was not itself investigated, it is a worthwhile consideration when addressing cell-surface proteins. For example, cell-surface heat shock protein 70 can promote phagocytosis (Wang et al. 2006), and inflammatory cytokine production through interaction with TLR2 and TLR4 (Asea et al. 2002). Furthermore, the existence and persistence of an intracellular symbiont ultimately involves the immune system as it inherently implies that the host is not clearing a foreign body.

In Symbiodiniaceae, persistence within the host is generally attributed to host immune suppression. This is evidenced by phenomena such as corals displaying decreased disease susceptibility when bleached, or in other words, when corals have a lower symbiont load (Merselis et al. 2018) and the upregulation of immune-suppressing TGF-\(\beta\) in the coral host during the onset of symbiosis (Berthelier et al. 2017). It may, therefore, be important that immune-activating proteins increase, while regulating proteins decrease and immune-suppressing proteins virtually disappear after 24 h of exposure to heat stress. Should such a pattern persist when in a symbiotic state, it would support the hypothesis that thermally induced bleaching results from a host immune response against Symbiodiniaceae. It would also support the hypothesis that the immune response results, in part, from symbiont dysfunction.

Breviolum psygmophilum present an “eat me” signal after experiencing heat stress

Calreticulin was present at the Breviolum psygmophilum cell surface after 24 h of exposure to heat stress. Calreticulin typically provides chaperone-like functions in the endoplasmic reticulum (Wang et al. 2004b); however, it is known to accumulate at the cell surface during stress events (Park and Kim 2017). In apoptotic cells, this accumulation can promote cell clearance by serving as an “eat me” signal to phagocytic cells (Park and Kim 2017). Because the symbiosome is established as an arrested phagosome (Mohamed et al. 2016), one possibility is that cell-surface calreticulin “re-activates” the fusion of the symbiosome to the previously inhibited lysosome. Calreticulin could, therefore, serve as a signal for dysfunction to the coral host and induce symbiophagic processes.

Conclusions

The coral-Symbiodiniaceae symbiosis is responsible for the persistence of coral reefs in tropical waters. Rising global temperatures are a primary threat to this symbiosis. This work joins an emerging body of research highlighting the importance of cell-surface modulation. Here, we present the first formal investigation into the response of cell-surface proteins to elevated temperatures in a Symbiodiniaceae species and show that this locale is dynamically modified in response to heat. These data demonstrate that stress experienced within the cell is manifested at the cell surface, and that these proteins have the potential to influence host responses during temperature stress. As coral bleaching (i.e., symbiosis breakdown) continues to decimate reefs, continuing investigation into responsible mechanisms is of vital importance for informing conservation and management practices.