As Earth’s oceans continue to warm and acidify  , scientists are racing to understand the physiological implications of such climate change impacts on marine organisms  ; there has been a particular focus on taxa known to be environmentally-sensitive, such as reef-building corals  . Although it is true that many scleractinian-dinoflagellate (genus Symbiodinium) endosymbioses readily disintegrate (i.e., “bleach”) upon prolonged exposure to unfavorable environmental conditions  , certain species/populations have proven to be markedly plastic and readily acclimatize/acclimate to an array of different environmental conditions in situ   , as well as in the laboratory     .
Southern Taiwan has served as an exemplary natural laboratory for understanding how environmental heterogeneity influences coral biology, as there are well-developed coral reefs experiencing very different oceanographic conditions in near vicinity of each other  . For instance, corals of Houwan (HWN), a reef within the Taiwan Strait (Figure 1), experience a relatively stable environment with respect to seawater temperature  , which rarely
Figure 1. Map of Southern Taiwan featuring the two study sites: Houbihu (upwelling site within Nanwan Bay) and Houwan (non-upwelling, Taiwan Strait control site). Please see Table 1 for a summary of the oceanographic differences between these two sites. Both full-scale (upper inset) and “macro” (lower inset) images of the model coral for this study, Seriatopora hystrix (which is abundant at both sites), have been presented, and white scale bars represent ~10 cm and ~5 mm, respectively. NMMBA = National Museum of Marine Biology and Aquarium.
fluctuates more than a 1 - 2˚C within a single day. In contrast, the thermal environment of nearby (~15 km) Houbihu (HBH), which is just around the Maobitou Cape within Nanwan Bay (Taiwan’s southernmost embayment; Figure 1), differs dramatically  ; Nanwan Bay experiences spring tide-induced upwelling events in the boreal summer, during which temperature can change by up to 9˚C in just a few hours.
To gain greater insight into the molecular biology underlying the ability of corals to thrive in these upwelling environments, biopsies of the common, Indo-Pacific scleractinian Seriatopora hystrix Dana 1846 were taken from the same colonies from which laboratory-based experiments were previously conducted      immediately upon removal of the colonies from the ocean at each of the two aforementioned study sites in Southern Taiwan, HBH (the upwelling site) and HWN (the non-upwelling site), during a stable-temperature period (i.e., between upwelling events in the case of HBH). It was hypothesized that a two-dimensional (2D) gel electrophoresis approach followed by sequencing of proteins uniquely synthesized by corals of one site and not the other (i.e., uniquely synthesized proteins [USPs]) with mass spectrometry (MS) could aid in elucidating the proteomic basis of survival in a highly variable-temperature environment. Indeed, proteomics-based approaches have aided in our understanding of both the fundamental cell biology of anthozoan-dinoflagellate endosymbioses (e.g.,    ), as well as their responses to high temperatures  , and it was hypothesized that a number of proteins would be differentially concentrated (i.e., differentially concentrated proteins [DCPs]) between corals of the two study sites; such USPs/DCPs might be linked to cellular processes involved in combatting the major abiotic challenges facing the corals of each habitat.
2. Materials and Methods
2.1. Study Sites
Both the upwelling site HBH (21˚56'18.01"N, 120˚44'45.54"E) and the non-upwelling site HWN (22˚01'23.30"N, 120˚41'18.29"E) have been well characterized with respect to their oceanography  and coral reef ecology (e.g.,   ). Although the difference in temperature variability between study sites was the sole focus of prior environmental physiology works on seriatoporid corals from HBH and HWN  , other oceanographic variables were shown by Liu et al.  to differ between them, and these additional seawater quality parameters could likewise influence the physiology of the resident corals. We therefore undertook a meta-analysis of seawater quality at each site by pooling our own published data (    ) with those of Liu et al.  . We now briefly review how these data were collected.
2.2. Oceanographic and Ecological Data Collection
The temperature regimes of both the upwelling site HBH and the non-upwelling, Taiwan Strait, control site HWN (Figure 1) were characterized in detail in 2010 at 7 - 8 m depth (the depth of coral collection  ); specifically, temperature was measured at hourly intervals with HOBO® Pendant data loggers (Onset) for one year. Although the mean annual temperature did not differ between the two sites, the variability did (Table 1); the mean monthly temperature range was 2-fold higher at HBH than HWN due to spring-tide upwelling events that occur during the boreal summer at the former site only  .
As mentioned above, prior works on seriatoporid corals from HBH and HWN have focused mainly on these temperature differences as being responsible for the physiological heterogeneity documented across the sampled coral colonies (e.g.,      ). However, we considered additional seawater quality parameters herein as being potential drivers of physiological variation between corals of the two study sites by incorporating the data from Liu et al.  , in which stations 2 and 7 - 8 correspond to HWN and HBH, respectively. Specifically, Liu et al.  measured salinity, pH, dissolved oxygen (DO) content (%), biochemical oxygen demand (BOD5; mg/L), suspended solid levels (mg/L), turbidity (NTU), and concentrations of nitrite (μg/L), nitrate (μg/L), ammonia (μg/L), phosphate (μg/L), silicate (μg/L), and chlorophyll a (chl-a; μg/L) at monthly intervals between 2001 and 2008, and their respective averages across this timespan have been included in Table 1. Photosynthetically active radiation (PAR) was instead measured at the depth of coral collection (7 - 8 m) at hourly intervals by Mayfield et al.  , with the mean hourly daytime PAR calculated across several days in May 2010 (the month of coral collection) presented in Table 1. Finally, coral cover was estimated at each site as in Tkachenco et al.  , with modifications made by Liu et al.  , and the mean average live coral cover (ALCC) values of the latter work have been included in Table 1 for both study sites. Data were compared statistically across sites as described below.
2.3. Coral Sampling and Incorporation of Data from Past Studies
Sampling of six S. hystrix colonies (see insets of Figure 1) was undertaken at each site in May 2010 (a month in which upwelling events were frequent  ) during a period of stable (i.e., non-upwelling) temperatures (~26˚C) as in Mayfield et al.  . Only visibly healthy corals were sampled, and there were no evident signs of stress (e.g., excessive release of mucus, bacterial infection, bleaching, recent tissue loss, etc.) in any of the 12 colonies. Once at the surface, three small (~50 mg) biopsies taken from each colony were submerged in ~2 ml of TRIzol™ (Life Technologies), and the 36 biopsies were later homogenized with a mortar and pestle at the lab as in Mayfield et al.  . Additional tissue biopsies were taken immediately upon removal of the colonies from the ocean in order to determined in situ chl-a content and Symbiodinium density (described in detail in Mayfield et al.  ). The colonies from which these “in situ biopsies” were taken were transported to the laboratory and fragmented into nubbins for use in a laboratory-based reciprocal transplant  . Physiological- and molecular-scale
Table 1. Oceanographic differences between the two study sites and physiological variation between their resident seriatoporid corals. Please note that all coral physiological and molecular response variables were assessed only after four weeks of husbandry except for the following: chlorophyll a (chl-a; in situ and post-husbandry data obtained), maximum quantum yield of photosystem II (Fv/Fm; in situ and post-husbandry data obtained), Symbiodinium (Sym) density (in situ and post-husbandry data obtained), and two-dimensional (2D) gel electrophoresis + mass spectrometry (MS)-derived protein concentrations (in situ [herein] and post-husbandry data  obtained); in the case of the former three parameters only, their in situ values are instead included in the table. All values in the “HBH” (Houbihu) and “HWN” (Houwan) columns represent means (±std. dev.) unless noted otherwise. In the “Conclusion” column, “<” and/or “>” denote statistically significant differences between sites (P < 0.05). ALCC = average live coral cover. exp. = experiment. GCP = genome copy proportion (a molecular proxy for Sym density). PAR = photosynthetically active radiation. SE = standard error. For full gene names, please see the respective references.
aunequal variance; blog-transformed data; csquare root-transformed data.
data from the same coral colonies as those analyzed herein (      ) have been summarized in Table 1 in a similar manner as for the previously acquired seawater quality data in order to more thoroughly assess environmental differences in the physiology of S. hystrix in Southern Taiwan. This meta analysis was also carried out to better understand the effect of experimental husbandry on coral physiology (sensu Mayfield et al.  ), since Symbiodinium densities and chl-a concentrations, as well as the maximum dark-adapted yield of photosystem II (Fv/Fm), were documented both in situ and after four weeks of aquarium husbandry for corals of both study sites.
2.4. Coral Protein Extraction, 2D Gel Electrophoresis, and MS
Proteins from one of the three technical replicates from each of three randomly chosen colonies from each of the two sites were purified as in Mayfield et al.  . These six samples represent those biopsies preserved immediately upon removal from the source colonies from the ocean and were therefore meant to be representative of “in situ protein concentrations.” Proteins (n = 6) were extracted as in Mayfield et al.  , precipitated in acetone, washed, dried, dissolved in ~150 μL of rehydration buffer (9.5 M urea, 2% CHAPS, 0.5% carrier ampholytes, and 65 mM dithiothreitol), and 20 μL of solubilized protein were quantified with the 2D Quant kit (Amersham Biosciences). The transcriptomes of all 12 colonies, including the 6 whose proteomes were analyzed herein, were sequenced previously  , albeit after three weeks of acclimation and one week of experimentation (stable vs. variable temperature regimes  ).
Proteins (~130 μg) were electrophoresed across 2D as in Mayfield et al.  ; however, as this book chapter is not freely available except on ABM’s personal website (http://coralreefdiagnostics.com/), certain key details, which can also be found in Mayfield et al.  , been reiterated below. Briefly, upon electrophoresing the protein samples (n = 3 for each of two sites of origin: HBH vs. HWN) across 2D (isoelectric focusing for isoelectric point [pI] determination and SDS-PAGE for molecular weight [in kDa] determination) as described in Mayfield et al.  , gels were fixed and stained with SYPRO® Ruby (Life Technologies). They were then imaged by a Typhoon Trio™ scanner (GE Healthcare), and ImageQuant TL software (GE Healthcare) was used to identify USPs, or, when USPs were not uncovered, DCPs, between each pair of HBH and HWN gels. Two gels were run simultaneously on each of three days: one HWN sample and one HBH sample. DCP/USP pIs and molecular weights were recorded from the 1st and 2nd HBH vs. HWN gel pairs. When these same protein spots were also differentially concentrated between gels, or uniquely synthesized by corals of one of the two sites, in the third pair of biological replicates, they were removed from the gel in which their concentration was higher (referred to herein as the “representative” gel), digested with trypsin, and prepared for MS as in Mayfield et al.   .
In general, only protein spots that were uniquely synthesized by coral samples of one site and not the other in all three pairs of gels were removed from the final, representative gel and sequenced, though two protein spots that were instead more concentrated by HBH samples in all three gel pairs (i.e., DCPs, rather than USPs) were removed from the final HBH gel and analyzed by MS. In total three HBH > HWN and six HWN > HBH spots were in-gel digested with trypsin and purified as in Mayfield et al.   . Then, 2 μL of purified, trypsin-digested peptides were injected into a nano-liquid chromatography system and detected by an LTQ Orbitrap “Discovery Hybrid Fourier Transform” mass spectrometer (Thermo-Fisher) at a resolution of 30,000 coupled with a nanospray source that was executed in positive ion mode. The nano-UPLC system (“nanoACQUITY”), desalting (Symmetry C18; 5 μm × 180 μm × 20 mm), and analytical (BEH C18; 1.7 μm × 75 μm × 150 mm) columns were all purchased from Waters. The peptide eluate from the column was directed to the nanospray source, and the MS was operated in data-dependent mode.
2.5. Data Analysis-I: Oceanographic, Ecological, and Target Response Variable Data
When oceanographic (e.g., temperature), ecological (ALCC), and molecular physiological (i.e., the 23 response variables analyzed previously in the sampled colonies; Table 1) data were normally distributed and of homogeneous variance across the two study sites, their means were compared across sites with student’s t-tests. Wilcoxon tests were used when log- or square root-transformations did not lead to normally distributed (Shapiro-Wilk test, P < 0.05) or homogeneously variable (Levene’s test, P < 0.05) datasets between sites. In certain cases, both in situ data (Symbiodinium density, Fv/Fm, and chl-a content) and data following four weeks of experimental husbandry (described in Mayfield et al.  ) were available. In these cases, repeated-measures ANOVAs were instead used to determine whether husbandry had a differential effect on corals of the two study sites. All univariate statistical analyses were carried out with JMP® (ver. 12.0.1). Finally, a multivariate approach aimed at modeling site differences in coral physiology (sensu Mayfield  ) was carried out with PRIMER (ver. 6) with all 23 response variables (standardized prior to analysis), testing site of origin as the fixed factor. PERMANOVA is based on similarity (Bray-Curtis) between samples and so does not require that data are normally distributed and of homogenous variance (as does MANOVA, which is sensitive to significant deviations from normality and cannot be used in cases such as this study in which there are more response variables [n = 23] than samples [n = 12]).
2.6. Data Analysis-II: MS
.MGF data files from the mass spectrometer (n = 6) were directly uploaded into the MS-SCAN program featured on the S. hystrix-Symbiodinium transcriptome server (http://symbiont.iis.sinica.edu.tw/s_hystrix/static/html/#mscan), and all default conditions of the MS-GF+ script  upon which MS-SCAN is based were used (discussed on the open access website housing the script: https://omics.pnl.gov/software/ms-gf); this included up to two missed cleavages allowed. However, as 1) Chiva et al.  found that sequence datasets featuring such missed cleavages are not inherently biased with respect to quantification, and 2) Mayfield et al.  found the S. hystrix-Symbiodinium proteome to be lysine- and arginine-rich, we generally included sequences containing over two missed cleavages provided that either 15 amino acids (AA) were sequenced at minimum or, alternatively, two peptides mapped to the same reference protein whose collective length was ≥15 AA. A decoy database (sensu  ) was not queried to calculate a false discovery rate since there is no fully sequenced proteome for a coral or Symbiodinium. Instead, the aforementioned peptide length criteria were increased above the default minimum of 6 AA of MS-GF+ to ensure that only peptides that could be identified with confidence were considered in the analysis. This undoubtedly reduced the total number of DCPs identified. Additional details of MS-GF+, which, unlike Mascot (Matrix Sciences), is free, can be found in Kim and Pevzner  .
When peptides could be identified with confidence (≥15 AA mapping to a translated contig in the S. hystrix-Symbiodinium transcriptome) with MS-SCAN, they were assigned a compartment of origin (host, Symbiodinium, intermediate [either host coral or Symbiodinium], or unknown), as well as a protein identity and functional category (e.g., metabolism) when the top hit contig (mRNA) hypothetically encoding the sequenced peptide aligned significantly (e < 10−5) to a functionally characterized protein in the NCBI nr database. The full mRNA sequence of the top hit contig derived from MS-SCAN analysis of the spectral data, rather than the peptide sequence itself, was used as the query (BLASTx) of the NCBI database; this is because trypsin-digested peptides are generally short (mean length = 22 ± 9 [std. dev.] AA herein; see Table S1 located at the end of the manuscript.), and such short sequences may not align significantly to any homolog on a public sequence repository like NCBI. However, when >30 contiguous AA were sequenced from a single protein, the peptide sequence itself was additionally BLASTed (BLASTp) against the NCBI database to attempt to corroborate the BLASTx analysis of the respective mRNA.
The compartmental breakdowns of the differentially concentrated proteomes were compared to the S. hystrix: Symbiodinium mRNA ratio of 1.8  with z-tests. This approach aimed to determine whether one member of the endosymbiosis contributed relatively more USPs/DCPs than the other for 1) all DCPs, 2) the HWN > HBH DCPs, and 3) the HBH > HWN DCPs. Two-sample proportion tests were used to determine whether 1) certain functional categories differed in proportional abundance between the HWN > HBH and HBH > HWN proteomes and 2) certain functional categories were significantly over-represented relative to the stable vs. variable differentially concentrated proteome of Mayfield et al.  . For all such statistical analyses, an alpha level of 0.05 was established a priori.
As the transcriptomes of nubbins generated from the colonies analyzed herein were sequenced previously (6 nubbins from each site of origin sequenced after one month of experimental husbandry  ), the expression data from all mRNAs encoding the 53 DCPs uncovered herein were acquired from the S. hystrix-Symbiodinium transcriptome server (http://symbiont.iis.sinica.edu.tw/s_hystrix/static/html/#stat), and 2-way ANOVAs were performed to determine the effects of site of origin (HWN vs. HBH), temperature treatment (stable vs. variable; see Mayfield et al.   for details.), and their interaction. When a site of origin difference was statistically significant (P < 0.05) and matched that observed at the protein level documented herein, congruency between mRNA expression and protein concentration was said to have occurred. For instance, if the 2-way ANOVA revealed that a gene was expressed at 2-fold-higher levels in corals of HBH, and the protein spot was more highly concentrated (or uniquely identified) in the final, representative HBH protein gel, molecular congruency was deemed to have been verified. Two-sample proportion tests were used to determine whether congruency differed significantly across 1) compartments (host coral vs. Symbiodinium), 2) site of origin (HBH vs. HWN), and 3) experiments (this study compared to a variable temperature study performed with these same samples  ).
3.1. Seawater Quality
Although most seawater quality parameters were similar between HWN and HBH, several differed significantly upon pooling data over an eight-year period (Table 1); notably, silicate concentration, suspended solid levels, and turbidity were all significantly higher at HBH (P < 0.05), which abuts a marina, than HWN, which is adjacent to Taiwan’s National Museum of Marine Biology and Aquarium (NMMBA; Figure 1; where all laboratory work was undertaken). As reported previously  , the mean monthly temperature range at HBH of ~6˚C was approximately double that of HWN (~3˚C; Table 1) in the year of coral sampling (2010).
3.2. Overview of Coral Physiology Results
In addition to the 2D + MS results generated herein, we aimed to first provide a brief overview of previous findings obtained from these same coral colonies. We first summarize the influence of a one-month experimental husbandry on coral nubbins from the two study sites that were fragmented from the same source colonies from which biopsies were taken herein for 2D + MS analysis (“Effect of experimental husbandry”); these findings are described in detail in Mayfield et al.  . Then, we briefly summarize the findings of two prior works (Mayfield et al.   ) in which 23 molecular-physiological response variables were measured in coral nubbins generated from the same source colonies as those analyzed herein, albeit exposed to stable or variable temperatures for one week (following three weeks of acclimation; “Coral molecular physiology differences between upwelling (HBH) and non-upwelling (HWN) study sites”). Upon presenting the 2D + MS data produced herein (“2D + MS”), we then compare the results obtained to those of another study with these same samples  that instead sought to model temperature-related, rather than site of origin-associated, differences in protein concentrations (“Comparison with a stable vs. variable temperature regime study carried out with S. hystrix”). It should be mentioned that all 12 colonies were of the identical genotype (determined by analysis of microsatellites  ) and hosted Symbiodinium of clade C exclusively  .
3.3. Effect of Experimental Husbandry
As discussed in more detail in Mayfield et al.  , some coral response variables did change in response to four weeks of experimental husbandry (three weeks of 26˚C-acclimation and one week of experimentation [stable vs. variable temperature regime]; see Mayfield et al.  for details.), and husbandry differentially affected corals of the two sites; Symbiodinium density increased more after four weeks of husbandry for corals of HWN than for those of HBH, whereas chl-a concentration and Fv/Fm increased more over this period for corals of HBH (Table 1).
3.4. Coral Molecular Physiology Differences between Upwelling (HBH) and Non-Upwelling (HWN) Study Sites
Of the 23 response variables measured in corals of the two study sites (Table 1; see Mayfield et al.   for details.), only the protein/DNA ratio and Symbiodinium photosystem I (subunit III; psI) mRNA expression differed significantly between corals of the upwelling and non-upwelling sites (Table 1). Regarding the former, S. hystrix colonies of HWN were characterized by 1.4-fold higher protein/DNA ratios than colonies of the same genotype and Symbiodinium assemblages of HBH. It should be mentioned that, despite this difference, equal quantities of protein were loaded into all 2D gels. In contrast, psI mRNA expression was 1.5-fold higher in Symbiodinium (clade C  ) populations within corals of HBH. Although PERMANOVA did not detect a multivariate effect of site of origin across the 23 response variables (Table 1), it is possible that a hypothesis-neutral, proteomics-based approach could nevertheless uncover site-related differences in coral proteo-biology; such was indeed the case, and these findings are discussed in detail below.
3.5. 2D + MS
Of the 6 and 3 protein spots concentrated at higher levels by samples of HWN (Figure 2(a)) and HBH (Figure 2(b)), respectively, 6 and 1, respectively, were uniquely synthesized by corals of one site of origin and not the other; two DCPs were additionally isolated from the representative HBH gel (Figure 2(b)). From the 6 HWN > HBH and 3 HBH > HWN protein spots, 38 (Table 2) and 15 (Table 3) proteins, respectively, were identified with MS-SCAN using the S. hystrix-Symbiodinium transcriptome as a reference database (Figure 3). Regarding the compartmental breakdown of all 53 proteins identified (Figure 3(a)), 55 and 38% were of host coral and Symbiodinium origin, respectively. This 1.4:1 ratio did not differ significantly from the 1.8:1 mRNA ratio of this coral  (z-test, P > 0.05), nor did it differ from the 1.6:1 host/Symbiodinium DCP ratio of another proteomic study of S. hystrix  (2-sample proportion test, P > 0.05).
With respect to the functional breakdown of all 49 proteins that could be assigned a compartment of origin (Figure 3(d)), nearly half aligned to proteins that had not been characterized. Of the remaining 28 proteins (57%) that did
Figure 2. Representative two-dimensional (2D) protein gels from Seriatopora hystrix colonies sampled from Houwan (HWN; (a)) and Houbihu (HBH; (b)). The x- and y-axes represent the isoelectric point (pI; 4 - 7) and molecular weight (6 - 200 kDa), respectively. Protein spots encircled and given a code (e.g., “HW1”) were removed from the gel, digested, purified, and sequenced with mass spectrometry. Matched spots in the opposing gel (i.e., same molecular weight and pI) have been encircled to highlight their differential concentrations between sites.
align to characterized proteins, the dominant functional categories were cytoskeleton, stress response, transport, and transcription. Of the 14 HBH > HWN DCPs (Figure 3(e)), cytoskeleton and the stress response were the most represented categories, and proteins involved in the stress response were significantly more likely to be documented at higher levels by corals of HBH (22% of the differentially expressed proteome) than conspecifics of HWN (Figure 3(f); 3%; 2-sample proportion test, P < 0.05). For the 35 HWN > HBH DCPs (Figure 3(f)), proteins involved in transcription and transport were instead of higher proportional abundance.
3.6. Host Coral and Symbiodinium Differentially Concentrated Proteomes
When looking only at the 29 host coral DCPs (Figure 3(g)), proteins involved in the stress response, cytoskeleton, and transcription were most abundant. As when the host and Symbiodinium data were pooled (Figure 3(d)-(f); discussed above), the former process was proportionally more abundant in the HBH > HWN proteome (Figure 3(h)); 22% of the 9 HBH > HWN and 0% of the 20 HWN > HWN host DCPs were involved in the stress response (Figure 3(i); 2-sample proportion test, P < 0.05). Instead, transport and transcription were
Figure 3. Pie graphs depicting compartmental and functional breakdown of proteins whose concentrations differed in Seriatopora hystrix colonies from two study sites: Houbihu (HBH; upwelling site) and Houwan (HWN; non-upwelling site). The compartmental breakdown of all uniquely synthesized (USPs) and differentially concentrated proteins (DCPs) has been shown for all DCPs + USPs (a), as well as for those HBH > HWN ((b); see Table 3.) and HWN > HBH ((c); see Table 2.) DCPs. Similarly, the host + Symbiodinium (Sym) DCPs (d)-(f), host DCPs only (g)-(i), and Sym DCPs only (j)-(l) have been presented across all DCPs ((d), (g), and (j), respectively), the HBH > HWN DCPs only ((e), (h), and (k), respectively), and the HWN > HBH DCPs only ((f), (i), and (l), respectively). When a functional category was represented at a significantly different proportion between the HBH > HWN and HWN > HBH proteomes, a bar has been placed over the category name in the proteome in which the value was higher. When the proportions of the host + Sym (d) and Sym functional categories (j) were significantly higher than in their respective proportions in the 117-protein host + Sym and 42-protein Sym-only stable vs. variable temperature proteomes of Mayfield et al.  , respectively, the functional category name has been underlined. No functional categories differed in proportional abundance between the 29 host coral DCPs uncovered herein and the 75 found to be differentially concentrated between stable and variable temperature treatments in Mayfield et al.  .
the most numerically dominant functional categories in the HWN > HBH differentially concentrated host coral proteome (Figure 3(i)).
Table 2. The 38 proteins whose concentrations were higher in corals of the non-upwelling control site: Houwan (HWN). Spots 1, 2, 3, 4, 5, and 6 in the left-most column correspond to spots “HW1,” “HW2,” “HW3,” “HW4,” “HW5,” and “HW6,” respectively, in Figure 2(a); all were found only in the three HWN gels (and therefore in none of the three Houbihu [HBH] gels). The average coverage of 15 ± 13% (std. dev.) did not differ significantly from that of the 15 differentially concentrated proteins found within the 3 HBH > HWN spots (14 ± 9% [std. dev.]; Table 3; student’s t-test, P > 0.05). Contigs denoted by asterisks were associated with congruency between mRNA expression and protein concentration. The host/Symbiodinium (Sym) ratio of 20/15 (1.3:1) did not differ significantly from the HBH > HWN ratio of 9/5 (1.8:1; Table 3; 2-sample proportion test, P > 0.05), nor did it differ from the Seriatopora hystrix/Sym mRNA ratio of 1.8:1  (z-test, P > 0.05). For the peptide sequences, please see Table S1.
aMaintained at lower concentrations in corals exposed to variable temperature regimes  ; bClosely related to a protein documented at lower concentrations in corals exposed to variable temperature regimes  .
Table 3. The 15 proteins whose concentrations were higher in corals of the upwelling site: Houbihu (HBH). Spots 1, 2, and 3 in the left-most column correspond to spots “HB1,” “HB2,” and “HB3,” respectively, in Figure 2(b). Only the latter spot was unique to the HBH gels; spots HB1 and HB2 were concentrated at higher levels in the HBH gels than the HWN gels and so were differentially concentrated proteins (DCPs) rather than uniquely synthesized proteins (USPs). Contigs denoted by asterisks were associated with congruency between mRNA expression and protein concentration. The host/Symbiodinium (Sym) DCP + USP ratio of 9/5 (1.8:1) was the same as the Seriatopora hystrix/Sym mRNA ratio  (1.8:1; z-test, P > 0.05). For the peptide sequences, please see Table S1.
aMaintained at lower concentrations in corals exposed to variable temperature regimes  .
In general, the same functional categories differed in proportional abundance between sites of origin for Symbiodinium (Figure 3(j)): stress response, transcription, and transport. However, the Symbiodinium differentially concentrated proteome also featured proteins involved in metabolism. When compared to another study performed with Taiwanese S. hystrix samples   , the Symbiodinium proteome sequenced herein featured significantly higher proportions of proteins involved in the stress response, transport, and transcription (2-sample proportion tests, P < 0.05). In the stable vs. variable temperature experiment, on the other hand, proteins associated with lipid bodies and mRNA processing were more likely to be maintained at different levels by Symbiodinium populations (2-sample proportion tests, P < 0.05).
Only five Symbiodinium proteins were identified across the three HBH > HWN protein spots, and only one could be identified with confidence (Figure 3(k)): a stress-associated HSPB1-associated protein 1-like protein was concentrated at higher levels in Symbiodinium from corals of HBH (Table 3). In contrast, the majority of the 15 HWN > HBH Symbiodinium DCPs could be identified with confidence (Figure 3(l)), and these proteins were involved in metabolism, transport, and transcription. No functional category differed in proportional abundance between host and Symbiodinium for the HBH > HWN or HWN > HBH differentially expressed proteomes.
3.7. Congruency between mRNA Expression and Protein Concentration
The congruency between mRNA and protein expression did not differ between the two compartments of the S. hystrix-Symbiodinium endosymbiosis (2-sample proportion tests, P > 0.05). Specifically, 2 of the 29 host coral molecules (7%) and 1 of the 20 Symbiodinium molecules (5%) demonstrated congruency between mRNA expression and protein concentration. These three molecules included 1) a host O-aminophenol oxidase involved in the oxidative stress response (the lone HBH > HWN molecule demonstrating congruency between mRNA and protein expression; 7%; Table 3), 2) one host coral molecule of unknown function (HWN > HBH; Table 2), and 3) one Symbiodinium molecule of unknown function (HWN > HBH; Table 2). In addition, one HWN > HBH USP of unknown cellular origin (Table 2) also demonstrated congruency between mRNA expression and protein concentration. The overall mRNA vs. protein congruency of 7.5% (4/53) was significantly higher than that of another study performed with Taiwanese S. hystrix samples (2-sample proportion test, P < 0.05); Mayfield et al.  found only 2 molecules out of 167 total (<2%) that demonstrated congruency between mRNA expression and protein concentration. A more detailed comparison with this only other proteomic assessment of S. hystrix  can be found below.
3.8. Comparison with a Stable vs. Variable Temperature Regime Study Carried out with S. hystrix
Besides a host coral beta-gamma crystallin protein, only three additional proteins were found to be differentially concentrated between sites of origin herein and between stable and variable temperature treatments in S. hystrix nubbins made from these same colonies by Mayfield et al.  . All three proteins were of host coral origin, though only two could be identified with confidence: a serine-arginine repetitive matrix protein (RMP) involved in mRNA processing and a translation factor known as eukaryotic translation initiation factor 3 subunit A-like. The serine-arginine RMP was one of only four DCPs uncovered herein involved in mRNA processing and editing (7.5%), the most temperature-sensitive cellular process documented previously in S. hystrix  (16%). In contrast to the variable temperature study, proteins whose concentrations differed between the two study sites herein were more likely to be involved in the stress response, transcription, transport, and the cytoskeleton.
4.1. The Environmental Physiology of S. hystrix in Southern Taiwan
Cell and molecular biology-driven approaches have aided in developing our understanding of both the fundamental biology of anthozoan-dinoflagellate endosymbioses     , as well as their environmental physiology    . Herein we utilized a differential proteomics approach to uncover proteins whose concentrations differed between corals of an upwelling site (HBH) and a non-upwelling site (HWN). In addition to the difference in the mean monthly temperature range between these two sites uncovered in prior works (e.g.,  ), a meta-analysis of previously published data conducted herein also revealed significant differences in silicate concentrations, suspended solid levels, and turbidity between them; specifically, these parameters were all significantly higher at HBH. As the PAR levels reaching the sampled corals were identical at each site, the elevated turbidity and suspended solid levels at HBH did not evidently result in a decrease in PAR reaching the sampled S. hystrix colonies. However, it is possible that such suspended solids affected other wavelengths of light that were not measured; for instance, high levels of suspended particulate matter lead to the attenuation of ultraviolet radiation (UVR) in wastewater  .
If diminished UVR levels were reaching the HBH S. hystrix colonies, then they could be hypothesized to be at lower risk of high temperature + UVR-induced bleaching  . In fact, relatively more proteins in the HBH>HWN differentially expressed host + Symbiodinium proteome (22%) were involved in the stress response compared to the 35-protein HWN > HBH proteome (3%). These three, presumably stress-targeted HBH > HWN proteins included two host coral proteins, beta-gamma crystallin and O-aminophenol oxidase, as well as the Symbiodinium protein known as HSPB1-associated protein 1-like. Beta-gamma crystallin, which has only ever been hypothesized to be involved in the stress response (as well as calcium binding  ), was actually found to be down-regulated in S. hystrix specimens exposed to a variable temperature regime for one week  and represents one of only four proteins found to differ in concentration between sites of origin herein and across temperature treatments in Mayfield et al.  ; as such, and in reiterating the recommendation of Meyer et al.  , we advocate that the role of beta-gamma crystallin in coral thermal adaptation be more thoroughly characterized in future works.
Given the fact that mean monthly temperature range, silicate concentrations, and suspended solid levels and turbidity were all higher at HBH (the upwelling site), it is tempting to speculate that the relatively higher number of stress response-associated proteins within the HBH > HWN proteome is indicative of sub-lethal levels of cellular stress in corals of HBH. However, the physiological performance of these corals did not differ from those of HWN (   and Table 1). This may mean, in contrast, that such constitutively elevated stress protein levels instead represent “front-loading”   , whereby stress proteins are maintained at high intracellular concentrations such that the molecular machinery requisite for dealing with large shifts in the abiotic environment (e.g., an upwelling event in the case of corals of HBH) is engaged at any given time. This strategy is relatively uncommon in nature given the high energetic expense of being constitutively “stressed” at all times  , though it does characterize the cellular biology of some intertidal invertebrates  . By exposing corals from both study sites to variable temperature regimes and elevated suspended solid levels in the laboratory, it may ultimately be determined whether the protein expression signatures of corals of HBH documented herein are resultant of cellular stress due to, for instance, direct impacts of sediments with coral tissues (sensu   ), or, alternatively, represent a protective response to counter future environmental change. If healthy corals indeed constitutively maintain high cellular concentrations of stress-associated proteins, as has also been documented in the most-remote reaches of French Polynesia  , then this may complicate the interpretation of data derived from molecular biomarker panels aimed at assessing coral health; in other words, are high concentrations of molecular chaperones and other stress proteins indicative of healthy corals or stressed ones?
4.2. The Role of Osmoregulation in Coral Acclima(tiza)tion to Environmental Change
Osmoregulation has been hypothesized to be the crux of the coral stress and bleaching response  . This theory stems from the fact that high temperature- and light-induced photoinhibition  would presumably result in a reduction in osmolyte flux from Symbiodinium to host. This would lead to a drop in osmotic pressure in the endosymbiotic gastrodermal cell  , which would manifest in changes in cytoskeletal architecture  . This may explain why a large number of host cytoskeleton-associated proteins were differentially concentrated between corals of the two sites. Specifically, actin and gelsolin (which is involved in actin assembly/disassembly) were synthesized at higher levels by corals of HBH. The fact that these corals are more likely to undergo osmotic pressure fluctuations as a result of upwelling-induced temperature changes may explain why they constitutively synthesize larger quantities of gelsolin, in particular, as this protein acts in cytoskeleton rebuilding. Not surprisingly then, genes encoding proteins involved in the cytoskeleton have found to be differentially expressed in corals exposed to elevated temperatures  . Given these findings, as well as the fact that osmoregulation is the most energetically expensive task a cell undertakes, the role of osmoregulation in the coral thermal stress response should be more thoroughly characterized in future works. It should be noted here that, due to the rigid cell walls of Symbiodinium, the osmotic stress associated with temperature + UVR stress-derived photoinhibition mentioned above is not hypothesized to dramatically affect Symbiodinium cell volume  ; therefore, it is unsurprising that no Symbiodinium cytoskeleton proteins were uncovered herein.
4.3. Congruency between mRNA Expression and Protein Concentration
Unfortunately, it has become commonplace in the coral biology field to make conjectures about protein behavior based on mRNA expression data alone (sensu   ), despite the fact that mRNA vs. protein congruency has been found to be as low as 0% in Symbiodinium populations within S. hystrix nubbins of another study  . Likewise, the degree of congruency between mRNA expression and protein concentration was markedly low herein; only 4 of the 53 DCPs uncovered were associated with an mRNA whose expression also differed significantly between sites of origin (7.5%). One such molecule, the host coral O-aminophenol oxidase, was discussed above in the context of the coral stress/environmental acclimation response. Given the low congruency between mRNA expression and protein concentration in both this study (7.5%) and others (2% in Mayfield et al.  and 10.5% in Mayfield et al.  ), we advise that those researchers looking to model the response of anthozoan-dinoflagellate endosymbioses to environmental change instead exploit proteomics-based approaches in their experiments; unlike the mRNAs that encode them, proteins actually enact physiological changes in cells and are likely to play key roles in thermal acclimation in both coral hosts  and their in hospite Symbiodinium populations.
The authors would like to thank Dr. Yu-Bin Wang for developing the MS-SCAN software and implementing it on the S. hystrix-Symbiodinium transcriptome server, as well as Drs. Peter Edmunds and Tung-Yung Fan for fruitful discussions on the eco-physiology of corals of Southern Taiwan. This work was funded by the United States National Science Foundation (postdoctoral research fellowship to ABM; OISE-0852960) and the Khaled bin Sultan Living Oceans Foundation (postdoctoral research fellowship to ABM).
Supplemental Data-Peptide Sequences
Table S1. “AA” = amino acids. “Sym” = Symbiodinium. * = two peptide sequences overlapped. ** = three peptide sequences overlapped.
acompartment of origin could not be verified.