SUMMARY
Ocean acidification, or the increased uptake of CO2 by the ocean due to elevated atmospheric CO2 concentrations, may variably impact marine early life history stages, as they may be especially susceptible to changes in ocean chemistry. Investigating the regulatory mechanisms of early development in an environmental context, or ecological development, will contribute to increased understanding of potential organismal responses to such rapid, large-scale environmental changes. We examined transcript-level responses to elevated seawater CO2 during gastrulation and the initiation of spiculogenesis, two crucial developmental processes in the purple sea urchin, Strongylocentrotus purpuratus. Embryos were reared at the current, accepted oceanic CO2 concentration of 380 microatmospheres (μatm), and at the elevated levels of 1000 and 1350 μatm, simulating predictions for oceans and upwelling regions, respectively. The seven genes of interest comprised a subset of pathways in the primary mesenchyme cell gene regulatory network (PMC GRN) shown to be necessary for the regulation and execution of gastrulation and spiculogenesis. Of the seven genes, qPCR analysis indicated that elevated CO2 concentrations only had a significant but subtle effect on two genes, one important for early embryo patterning, Wnt8, and the other an integral component in spiculogenesis and biomineralization, SM30b. Protein levels of another spicule matrix component, SM50, demonstrated significant variable responses to elevated CO2. These data link the regulation of crucial early developmental processes with the environment that these embryos would be developing within, situating the study of organismal responses to ocean acidification in a developmental context.
INTRODUCTION
Ecological development, or ‘eco-devo’, is the study of development in an ecological context, moving the study of development out of the laboratory and into the ‘real world’ environment (Gilbert, 2001; Sultan, 2007). Additionally, the aims of eco-devo are to incorporate ‘both abiotic and biotic factors into studies of gene expression and regulatory pathways... [and study] the expression of traits important to function, fitness and ecological interactions in [the] environment’ (Sultan, 2007). Understanding how regulatory pathways operate under differential environmental conditions has been suggested as the necessary next step for developmental and physiological studies, as environmental factors play a crucial role in the execution of development (Gilbert, 2005; Sultan, 2007; Gilbert and Epel, 2009).
The need for an ecological perspective on development has become increasingly clear in light of anthropogenically driven ocean warming and ocean acidification (Kurihara, 2008; Doney et al., 2009; Dupont et al., 2010). Very little is known about the capacity for phenotypic plasticity in organisms during their early life history stages and whether they can successfully respond to rapidly changing environmental conditions. The anthropogenically driven rise in atmospheric CO2, which has steadily increased since the industrial revolution, has led to an increase in oceanic CO2 (Forster et al., 2007; IPCC, 2007; Feely et al., 2008) and to ocean acidification (Doney et al., 2009; Feely et al., 2009). The continued rise in oceanic CO2 could present challenges for many marine organisms, as the increasingly acidic water is potentially corrosive to those organisms with calcified hard parts (Feely et al., 2008) and can interfere with the extracellular and intracellular acid–base balance necessary for proper cellular functions (Fabry, 2008).
Given that early developmental processes are especially sensitive to environmental perturbation (Hamdoun and Epel, 2007; Epel, 2003), successful development relies heavily on physiological tolerance of the environment and on whether embryos and larvae are able to respond to changing abiotic factors. There are a growing number of studies that have investigated how marine invertebrate embryos and larvae may handle changes in carbonate chemistry as a consequence of ocean acidification (Kurihara, 2008; Dupont and Thorndyke, 2009; Byrne, 2011). Many of these studies have focused on mortality and phenotypic characteristics such as growth rate and structural and morphological changes in larvae (see Kurihara, 2008). Specifically, the calcium carbonate spicules and skeletons of larval echinoids are directly affected by changes in oceanic CO2 concentration (Kurihara and Shirayama, 2004; Kurihara, 2008; Dupont et al., 2008; Byrne et al., 2009; O'Donnell et al., 2010). Few studies, however, have investigated the molecular-level responses and signaling pathways that regulate phenotypic expression and responses relative to large-scale environmental changes like ocean acidification (see Sultan, 2007; Todgham and Hofmann, 2009; O'Donnell et al., 2010). In this study, we sought to make the connection between these larger scale environmental phenomena and molecular-level responses. We measured expression changes in two regulatory pathways in a key gene regulatory network – the primary mesenchyme cell gene regulatory network (PMC GRN) – in Strongylocentrotus purpuratus sea urchin embryos as they develop through gastrulation at elevated CO2 levels. Notably, we chose CO2 levels that are consistent with future emission scenarios predicted by the IPCC (Meehl et al., 2007) and, additionally, reflect occasional high PCO2 exposures for S. purpuratus embryos in episodic upwelling zones on the coast of the northeastern Pacific (Feely et al., 2009; Hauri et al., 2009).
Of the numerous processes in early development, gastrulation is an extremely critical one that, if not executed properly, has dire and typically fatal results for the embryo (Wolpert, 1992; Leptin, 2005). In sea urchin embryos, the gastrulation process overlaps another key process, spiculogenesis. A group of spiculogenesis-specific cells, the primary mesenchyme cells (PMCs), ingress and migrate during gastrulation, and initiate the events leading to the formation of the spicules (Wilt, 2005). Not surprisingly, many of the regulatory pathways that regulate the initiation of spiculogenesis are also crucial for the regulation of cellular movements in gastrulation (Oliveri et al., 2008; Ettensohn, 2009). These shared pathways make up part of the PMC GRN, which regulates the process of biomineralization (the process by which organisms take up minerals in order to form their hard parts). The PMC GRN is one of the best understood gene regulatory networks in sea urchins (Oliveri et al., 2008; Ettensohn, 2009), providing a detailed model of the cascade of gene regulatory events required of the embryo to carry out gastrulation and spiculogenesis. While many studies have utilized the PMC GRN as a model to garner an increased fundamental and mechanistic understanding of these two crucial aspects of the urchin developmental program, no studies to date have sought to determine whether and how environmental factors, such as increased CO2, can affect the transcriptional regulation of these processes by the PMC GRN.
In this study, we investigated how a subset of regulatory pathways for two developmental processes in S. purpuratus embryos were affected by elevated CO2 concentrations in comparison with embryos that develop under control conditions. Aside from being a model organism with a thoroughly characterized PMC GRN, S. purpuratus was chosen as the study organism because it is a keystone echinoderm species that inhabits a large biogeographic range along the North American west coast. Additionally, S. purpuratus larvae, like other echinoplutei, develop a calcium carbonate (CaCO3) skeleton that is susceptible to increases in oceanic CO2 (Kurihara and Shirayama, 2004; O'Donnell et al., 2009; Dupont et al., 2010). We have investigated a subset of key pathways, which either regulate or are regulated by the gene Alx1, in the PMC GRN that are crucial for the developmental processes of gastrulation and spiculogenesis.
An overview of the targeted PMC GRN genes
Alx1 is an upstream PMC GRN regulatory gene that is essential for PMC ingression and their movement within the blastocoel (Ettensohn et al., 2007). The genes chosen for this study were in pathways that either regulate or are regulated by Alx1. These genes (shown in Fig. 1) were Wnt8, Pmar1, Alx1, VegFR, SM30b, Msp130 and SM50. Alx1 itself is regulated by Pmar1, a zygotically expressed transcriptional repressor responsible for activating micromere specification (micromeres that are then specified to become the PMCs) (Oliveri et al., 2008). Wnt8 is a zygotically expressed regulatory gene that regulates Pmar1 through the activation of β-catenin (Oliveri et al., 2008). As one of the earliest genes to activate the PMC GRN, Wnt8 is associated with cell fate determination through canonical signaling pathways and is important for the morphogenetic movement of the PMCs through non-canonical signaling (Leptin, 2005).
Alx1 also acts as a regulatory input to VegFR. VegFR, which is expressed solely by the PMCs, encodes a receptor of the ectoderm-derived ligand VegF (vascular endothelial growth factor) (Ettensohn, 2009). VegF originates from the two ventrolateral positions in the ectoderm where the triradiate spicule rudiments are later deposited (Okazaki and Inoue, 1976), and this ectoderm–PMC interaction between VegF and VegFR is necessary for the correct migration and orientation of the PMCs, and proper spicule formation (Duloquin et al., 2007). Additionally, VegF/VegFR signaling is thought to be the regulatory input to one of the terminal differentiation genes of the PMC GRN, SM30b (Duloquin et al., 2007; Oliveri et al., 2008). SM30b is a biomineralization gene important for spicule rod growth and elongation (Kitajima and Urakami, 2000; Urry et al., 2000). It encodes the spicule matrix protein SM30, one of over four dozen identified spicule matrix proteins that make up the occluded matrix of the spicule (Wilt, 2005; Livingston et al., 2006).
Msp130, also regulated by Alx1, is a terminal differentiation gene that encodes a cell surface glycoprotein, MSP130, which completely surrounds the PMCs and the spicules (Kiyomoto et al., 2007; Wilt et al., 2008a). Msp130 has been implicated in calcium deposition (Carson et al., 1985). Lastly, Alx1 acts as a regulatory input for SM50, a gene that encodes another spicule matrix protein, SM50. SM50, like SM30, occurs within the occluded matrix of the spicule (Wilt, 2005; Livingston et al., 2006); it is important for spicule elongation (Kitajima and Urakami, 2000; Urry et al., 2000) and is necessary for deposition of the calcite spicule rudiment (Wilt et al., 2008a).
We have evaluated the expression of this subset of PMC GRN genes in response to varying levels of CO2 during early development; specifically, over the course of gastrulation. We also investigated whether expression of a protein encoded by one of the biomineralization genes (SM50) varied as a result of differential CO2 levels. Investigating these pathways will not only provide information on how upstream PMC GRN events affect biomineralization and spicule formation but also help us to elucidate how the developmental programs of spiculogenesis and gastrulation respond to increased CO2 and ocean acidification.
MATERIALS AND METHODS
Strongylocentrotus purpuratus collection and embryo culturing
Adult S. purpuratus (Stimpson 1857) were collected in the Santa Barbara Channel during the middle of the spawning season (mid-February 2010) and held in flow-through seawater tanks at 14–16°C. Individuals were spawned by coelomic injection of 0.5 mol l–1 KCl within 2–3 weeks of collection. Eggs from a single female were then fertilized by a single male, following a protocol outlined elsewhere (Foltz et al., 2004), in ambient CO2 seawater to ensure the highest fertilization success as low pH levels can compromise this process (Havenhand et al., 2008; Parker et al., 2009) (but see Byrne et al., 2009; Byrne et al., 2010). Only batches of eggs with a fertilization success of >90% were used. A total of four mating pairs were spawned and eggs fertilized.
Embryos from each mating pair were divided approximately equally and raised in larval culture buckets in the CO2 delivery system (see Fangue et al., 2010) at a density of ∼100 embryos ml–1. Because of the physical constraints of the CO2 delivery system, only 12 culture buckets could be run at a time. To maximize the number of mating pairs and replicates within the 12 bucket system, embryos from two mating pairs were raised simultaneously in two sets of six larval culture buckets. Of the six larval buckets used for each mating pair, two buckets had a CO2 concentration of 380 microatmospheres (μatm), which served as the controls, two treatment buckets had a CO2 concentration of ∼1000 μatm, and the remaining two treatment buckets had a concentration of ∼1350μatm Therefore, each treatment had two replicates per mating pair (totaling six culture buckets). Each set of 12 buckets was referred to as a ‘trial’ for statistical analyses. Four mating pairs were used; therefore two trials were run (i.e. a total of 24 buckets run at 12 buckets at a time). The temperature of all larval culture buckets was maintained at ∼15°C.
Seawater chemistry
During the embryo culturing described above, seawater chemistry was performed using a Mettler-Toledo T50 (Columbus, OH, USA) alkalinity titrator and a spectrophotometric pH assay as described previously (Fangue et al., 2010). In preparation for the experiments, pH and alkalinity measurements were taken every few hours until the seawater reached the desired CO2 concentration, and once the desired conditions were achieved (see Table 1 for summary data for seawater chemistry), spawning was performed and embryos were introduced into the culture buckets. During embryo culturing, pH was measured at all embryo sampling time points while alkalinity was measured once during the 19 h sampling period. Following the protocol outline by Fangue (Fangue et al., 2010) (see also DOE, 1994), in brief, the seawater from each culture was filtered of embryos and siphoned into separate, pre-warmed (25°C) sampling cuvettes. The increase from 15 to 25°C was corrected for later using a carbonate calculator (Lewis and Wallace, 1998) described below. Each cuvette was filled and capped with no head space to minimize CO2 exchange with the atmosphere. Each sample was measured before and after addition of the indicator dye m-Cresol Purple. The absorbance wavelengths measured corresponded to one non-absorbing and two absorption maxima for the m-Cresol Purple indicator. Final pH values were determined using the absorbance values and the pH determination calculations in the modified standard operating procedures (see DOE, 1994). The accuracy of these results was ensured by periodic measurement of a certified reference material (CRM) for pH (Ocean Carbon Dioxide Quality Control, Scripps Institute of Oceanography) (Fangue et al., 2010). Total alkalinity measurements were made as previously described (Fangue et al., 2010). Briefly, seawater was siphoned and filtered, ensuring no air bubbles or embryos, into 125 ml borosilicate glass-stoppered bottles (1500-125, Corning, Corning, NY, USA) bottles, allowed to overflow, capped and brought to 25°C for immediate measurement via titration. The total alkalinity (μmol kg–1), along with the corresponding pH measurement, were then entered into the CO2SYS carbonate calculator (Lewis and Wallace, 1998) with the pre-determined dissociation constant for carbonic acid (Mehrbach et al., 1973) to determine the concentrations of dissolved inorganic carbon (DIC) and PCO2, the remaining parameters needed to fully characterize the carbonate system in seawater and fully characterize the amount of CO2 in the seawater sample (Fangue et al., 2010).
Embryo sampling
Embryo samples were collected by reverse filtration at three time points during the experiment: (1) directly after hatching and PMC ingression (21 h post-fertilization, h.p.f.), (2) at primary invagination (30 h.p.f.) and (3) at full gastrula (36 h.p.f.). Each of the samples (which contained ∼40,000 whole embryos per sample) were homogenized by vortexing in 500 μl of TRIzol Reagent (Invitrogen, Carlsbad, CA, USA) and stored at –80°C until RNA extraction. Total RNA was extracted using the TRIzol manufacturer's protocol with the following modifications. Once homogenized, samples were thawed, passed through a 25 gauge needle 3–5 times and washed with 100 μl chloroform, and the RNA was precipitated with 125 μl isopropanol + 125 μl precipitation salt solution (1.2 mol l–1 NaCl, 0.8 mol l–1 disodium citrate in DEPC-treated RNase-free water). The RNA was re-suspended in nuclease-free water and RNA concentrations were measured and purity was checked using a Nanodrop ND-1000 full-spectrum UV/Vis spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA).
Gene expression analysis using qPCR
Following extraction, 100 ng total RNA was reverse transcribed to synthesize cDNA using oligo dT and the ImProm-II Reverse Transcription System (Promega, Madison, WI, USA) following the manufacturer's protocol. The resulting cDNA was amplified via quantitative real-time PCR (qPCR) using primers for the PMC GRN-specific genes: Wnt8, Pmar1, Alx1, Msp130, SM30b, SM50 and VegFR (Table 2). Gene-specific primers for Wnt8, Pmar1, Alx1, Msp130, SM30b, SM50 and ef1-α were designed using Primer Express software (version 2.0.0; Applied Biosystems, Foster City, CA, USA) with the published S. purpuratus sequences for ef1-α (GenBank accession no. NM_001123497.1), Wnt8 (SPU_020371), Pmar1 (SPU_01422), Alx1 (SPU_214644), Msp130 (SPU_013821), SM30b (SPU_000826) and SM50 (SPU_018811). Confirmed primer sequences for VegFR were obtained from SpBase (http://www.spbase.org/SpBase/resources/methods/q-pcr.php).
Briefly, qPCR was carried out in 20 μl SYBR Green Supermix (Bio-Rad, Hercules, CA, USA) reactions containing 2 μl of template cDNA and 1 μl of each primer. The qPCR reactions were run with the following protocol: 1 cycle at 95°C for 3 min, 40 cycles at 58°C for 10 s and 95°C for 1 min, 1 cycle at 55°C for 1 min. Melt curve analyses were also conducted to ensure only a single product for each primer set was amplified. Relative mRNA levels were quantified as follows. Wnt8, Pmar1, Alx1, Msp130, SM30b, SM50, VegFR and ef1-α cycle threshold (Ct) values were first normalized to a single S. purpuratus embryo standard control sample run on every qPCR plate. Four, 10-fold serial dilutions of this standard were used to develop a standard curve on each plate and Ct results from all samples were expressed relative to these curves. The Ct values were then normalized to the corresponding ef1-α values. The ef1-α gene was used as an internal control as its expression does not significantly differ in response to elevated CO2. The resulting normalized values are the reported relative concentrations for Wnt8, Pmar1, Alx1, Msp130, SM30b, SM50 and VegFR.
Quantification of spicule matrix protein SM50
To determine whether elevated CO2 differentially affected the translation of transcripts for the terminal differentiation genes, protein analysis was carried out on SM50. Approximately 1500 embryos were collected from each culture at each sampling time point for protein extraction and immunoblot analysis of SM50 protein. Embryos were collected by microcentrifugation, all seawater was removed, and the samples were immediately snap-frozen and stored at –80°C.
Whole-embryo samples were homogenized by sonication in a lysis buffer containing 1.4× protease inhibitor cocktail (Roche, Pleasanton, CA, USA), 1 μg ml–1 leupeptin, 0.1% NP-40 substitute, 100 mmol l–1 DTT and 100 mmol l–1 PMSF. Samples were quantified using the BCA assay (Pierce Biotechnologies, Rockford, IL, USA), and 10 μg of homogenate from each sample was dissolved in sample loading dye and heated at 100°C for 3 min. Proteins were resolved on a 4% stacking/12% resolving discontinuous SDS-page gel and electrophoretically transferred to Hybond ECL 0.45 μm nitrocellulose membrane (GE Healthcare, Piscataway, NJ, USA). Blots were washed with phosphate-buffered saline (PBS) and blocked overnight in 5% non-fat dry milk in PBS with 0.1% Tween-20 at 4°C. Blots were then incubated in primary antibody (polyclonal antisera prepared in rabbits against SM50 at a 1:500 dilution in blocking buffer, courtesy of Dr Fred Wilt) (Killian and Wilt, 1996) for 1 h at room temperature, washed in PBS with 0.1% Tween-20, and then incubated in secondary antibody [goat anti-rabbit horseradish peroxidase conjugate (BioRad) at 1:5000 dilution in blocking buffer] for 1.5 h at room temperature. Blots were developed in chemiluminescent SuperSignal West Dura Extended Duration Substrate (Pierce Biotechnologies) and imaged using Quantity One software and VersaDoc imaging system (BioRad). Blots were quantified and analyzed with ImageJ (http://lukemiller.org/index.php/2010/11/analyzing-gels-and-western-blots-with-image-j/). Sample intensities were quantified relative to intensities of a standard protein extract (S. purpuratus prism stage embryos); 10 μg of standard were loaded onto each gel to ensure relative intensities of SM50 on each gel could be directly compared.
Statistical analyses
Statistical analyses were performed using JMP 8.0 (SAS). For each gene, MANOVA analyses were conducted to determine the significance of CO2 on the expression of each gene, with CO2, trial and mating pair nested within trial as factors. A MANOVA was also used to test for differences in SM50 protein levels with CO2, trial and mating pair nested within trial as factors. Stages were analyzed separately using one-way ANOVA and post hoc Tukey–Kramer HSD was used to determine differences between CO2 concentrations.
RESULTS
Overview of expression profiles at control CO2 concentrations
For control conditions, embryos were raised under 380 μatm CO2, an accepted level of ambient CO2 that yields a control pH of ∼8.1, and transcript levels for the seven candidate genes were evaluated using qPCR at three time points between hatched blastula and gastrula stages (Fig. 2). All transcripts were present at detectable levels at all three time points and generally matched patterns of expression observed in other studies (see below for details of individual genes). Of the genes evaluated in this study, Wnt8 exhibited the highest level of expression between hatched blastula and gastrula stage, consistent with the requirement for Wnt8 signaling in multiple regulatory pathways in the embryo (e.g. endoderm specification and non-skeletogenic mesoderm specification as well as PMC specification) (Smith and Davidson, 2008).
At hatched blastula stage, Wnt8 exhibited the highest level of expression. Expression of Pmar1 and Alx1 were the next highest, and were similar in expression levels to each other. VegFR and Msp130 exhibited a lower level of expression, followed by SM50 and SM30b, exhibiting the lowest transcript levels (Fig. 2). At early gastrula stage, Wnt8 expression decreased significantly, but was still greater than that of all other genes. Pmar1 and Alx1 transcript levels also decreased and were similar to those of VegFR and Msp130. VegFR, Msp130, SM50 and SM30b displayed increases in transcript level. At full gastrula stage, Wnt8 transcript levels further decreased but, again, were still higher than those of the other genes. Expression of Pmar1 and Alx1 also exhibited a further decrease in expression. VegFR, Msp130 and SM50 all displayed decreases and had similar transcript levels. In contrast, SM30b transcript levels increased significantly by this developmental stage, an expression pattern consistent with previous findings (Guss and Ettensohn, 1997; Wei et al., 2006).
Patterns of gene expression in embryos developing under high CO2 conditions
Wnt8
As a general pattern, Wnt8 expression declined as development advanced. At the control CO2 concentration of 380 μatm, the expression profile was as follows: Wnt8 expression was highest at hatched blastula stage, decreased at early gastrula (P=0.0449) and further decreased at gastrula (P<0.0080; Fig. 3). Expression profiles were similar at 1000 and 1350 μatm CO2. MANOVA results indicate that Wnt8 expression varied significantly as a function of CO2 concentration across development (P=0.0042). Expression also varied with trial (P<0.0001). There was also a significant interaction between CO2 and trial (P=0.0136); however, this effect simply highlights the variation between mating pairs, as mating pair was nested within trial. A separate ANOVA and post hoc analysis revealed that there was a significant increase in expression from 380 to 1350 μatm at full gastrula stage (P=0.0217).
Pmar1
Under control conditions of 380 μatm, Pmar1 expression was highest at hatched blastula stage, decreased slightly at early gastrula and further decreased at full gastrula. When embryos developed under high CO2, there were different outcomes for the two CO2 treatments: the pattern of expression was similar to that of the control for the 1350 μatm CO2 treatment, whereas at 1000 μatm there was some variation (Fig. 3). Specifically, there was a decrease in hatched blastula expression, an increase at early gastrula stage and a decrease at full gastrula stage. However, these shifts in expression observed in the mid-CO2 level treatment were not significant (all P>0.05; Fig. 3). Thus, expression of Pmar1 during early development was not affected by elevated CO2 concentrations (MANOVA, P=0.2342). There was also no variation due to trial, nor was there a significant interaction between CO2 and trial (P=0.2867). Furthermore, for embryos at all three CO2 concentrations, expression at full gastrula stage was significantly lower than at hatched blastula stage (P=0.0012).
Alx1
At the control CO2 concentration of 380 μatm, transcript for Alx1 showed an overall reduction as development proceeded: Alx1 gene expression was highest at the hatched blastula stage, decreased at the early gastrula stage, and further decreased at the full gastrula stage. The expression profiles at 1000 and 1350 μatm were nearly identical to the expression profile at 380 μatm (Fig. 3). There was no significant variation in Alx1 expression as a result of CO2 (MANOVA, P=0.2126), but there was a significant effect of trial (MANOVA, P<0.0001). For all CO2 concentrations, expression at hatched blastula stage was significantly greater than that of early gastrulae (P=0.0016) and full gastrulae (P<0.0001). At hatched blastula stage, Alx1 expression was highest at 1350 μatm; however, this was non-significant (P=0.0700).
VegFR
Under control CO2 conditions, VegFR transcript levels were highest at early gastrula stage and lowest at full gastrula stage. The general expression pattern was the same at 1000 and 1350 μatm (Fig. 4). Overall, there was no effect of CO2 concentration on VegFR expression (MANOVA, P=0.2562), nor was there any variation attributable to trial (MANOVA, P=0.5245) or a significant interaction between CO2 concentration and trial (MANOVA, P=0.4192). At each CO2 concentration, expression at hatched blastula stage was significantly lower than that at early gastrula (P=0.0014) and significantly higher than that at full gastrula (P=0.0089).
SM30b
At 380 μatm CO2, SM30b transcript levels were very low at hatched blastula stage, increased significantly at early gastrula, and further increased at full gastrula stage, where expression was greatest. The same trend was true for embryos cultured at 1000 and 1350 μatm CO2 (Fig. 4). SM30b expression, however, did vary as a function of CO2 concentration through development (MANOVA, P=0.0163), as indicated by increased expression at 1000 μatm in full gastrula stage. There was also variation as a result of trial (MANOVA, P=0.0181). The effects of CO2 concentration and trial were independent of each other (MANOVA, P=0.5557). At each CO2 concentration, the increase in expression from hatched blastula stage to early gastrula (P<0.0001) and full gastrula stage (P<0.0001) was highly significant.
Msp130
At the control CO2 concentration of 380 μatm, Msp130 expression was highest at early gastrula stage and lowest at hatched blastula stage at each CO2 concentration (Fig. 4). Msp130 expression did not vary as a result of CO2 concentration (MANOVA, P=0.3419). However, expression did vary as a result of trial (MANOVA, P=0.0004), which was independent of CO2 (MANOVA, P=0.4040). The expression at early gastrula stage was significantly higher than hatched blastula expression (P<0.0001), and the increase in expression at full gastrula over hatched blastula stage was also significant (P=0.0117).
SM50 gene expression and protein levels
Under control conditions, SM50 transcript levels were lowest at hatched blastula stage, increased at early gastrula, where expression was highest, and decreased at full gastrula stage at 380, 1000 and 1350 μatm CO2 (Fig. 4). Expression throughout development did not vary as a function of CO2 concentration (MANOVA, P=0.7506) but did vary with trial (MANOVA, P=0.0012). There was no interaction between CO2 concentration and trial (MANOVA, P=0.9127). At all CO2 concentrations, SM50 gene expression at hatched blastula stage was significantly lower than at early gastrula stage (P<0.0001) and full gastrula stage (P<0.0001).
Conversely, while there was no significant effect of CO2 on SM50 transcript levels, there was a significant effect on SM50 protein levels (MANOVA, P<0.0001; Fig. 5). SM50 protein level from embryos at each CO2 concentration followed the same expression pattern as SM50 transcript level: expression was lowest at hatched blastula stage, increased at early gastrula, where expression was greatest (P<0.0001 relative to hatched blastula), and decreased slightly at full gastrula stage but was still significantly greater than expression at hatched blastula (P<0.0001). SM50 protein expression did vary within trials (MANOVA, P<0.0001). There was also a significant interaction between CO2 and trial (MANOVA, P=0.0119). However, the interaction was caused by expression in embryos at 1000 μatm displaying a slightly different expression pattern from that in embryos at 380 and 1350 μatm – early gastrula stage embryos at 1000 μatm displayed greater expression than embryos at both 380 and 1350 μatm. One-way ANOVA (with Bonferroni-corrected α=0.0167) results at each stage show that embryos at hatched blastula stage at 1000 μatm (P=0.0006) and 1350 μatm (P=0.0044) displayed significantly lower SM50 protein levels than those at 380 μatm. At early gastrula stage, embryos at 1000 μatm exhibited the greatest expression level, which was significantly greater than that at 1350 μatm (P=0.0002) but not that at 380 μatm (P=0.1867). Protein expression at 380 μatm was greater than that at 1350 μatm, but this was not significant (P=0.0493). Finally, there were no significant differences between CO2 treatments at full gastrula stage (P=0.0261).
DISCUSSION
The molecular exploration of embryonic gene regulatory networks is extensive, but placing the study of regulatory pathways within these networks in an ecological or environmental context is relatively unexplored. In this study, a subset of key pathways in the PMC GRN necessary for gastrulation and the initiation of spiculogenesis (see Fig. 1) were investigated in embryos reared at relevant, elevated CO2 levels through gastrulation. With regard to the effects of ocean acidification conditions on development, the results of this study revealed only subtle changes in SM30b and Wnt8 expression at the high CO2 concentrations of 1000 and 1350 μatm, and variable expression in a biomineralization-specific protein, SM50. Additionally, Alx1, Pmar1, VegFR, Msp130 and SM50 were unaffected in terms of transcript level and timing of expression (for a summary, see Table 3).
Embryo and micromere specification genes
The CO2 concentration of 1350 μatm had a significant effect on Wnt8 expression at full gastrula stage: transcript levels were 20–28% greater than control levels at this stage. An increase in Wnt8 transcript level of this magnitude may not adversely affect the embryo, as no morphological abnormalities were observed in these embryos (not shown). Significant overexpression would likely have caused abnormalities such as the multiple invagination sites around the embryo found in a previous study (Wikramanayake et al., 2004). Given the current data and lack of morphological abnormalities, it may be that 1350 μatm, while causing some change in Wnt8expression at full gastrula stage, may not interfere with the initiation and regulation of embryo patterning, endoderm specification and, by extension, gastrulation.
Another study (Todgham and Hofmann, 2009) that explored the transcriptomic response to ocean acidification in S. purpuratus at prism stage (∼40 h.p.f.) found that Wnt8 expression was decreased at the 540 μatm CO2 level and exhibited no change at ∼1020 μatm. Differences in embryo responses between that study and the current study at the CO2 concentration of ∼1000 μatm are likely a result of the use of a microarray approach by Todgham and Hofmann, compared with the present qPCR-based study. Additionally, differences may be attributed to developmental stage, as in the latter, early prism stage Wnt8 expression levels are already significantly lower than those of earlier stages under normal conditions (Wikramanayake et al., 2004).
The present data indicate that elevated CO2 did not alter Alx1 expression. Alx1 is necessary for specification and morphogenetic movement of the PMCs, while also activating the genes necessary for the epithelial mesenchymal transitions of gastrulation (Ettensohn et al., 2003; Wu and McClay, 2007). Alx1 expression was also found to activate expression of downstream genes in the PMC GRN that are in turn essential for activation of biomineralization genes (such as Msp130 and SM50) (Ettensohn et al., 2003). Therefore, the current data suggest that PMC ingression and the subsequent initiation of the morphogenetic movements of gastrulation would be unaffected by elevated CO2. Morphological studies of early developmental stages exposed to elevated CO2 (Hammond, 2010) indeed show this; gastrulation proceeds successfully although with a possible modest delay in timing. Given that Pmar1 expression also did not change in response to elevated CO2 levels, a change in Alx1 expression would not be anticipated. It is important to note, however, that recent data suggest that Alx1 expression is not solely regulated by Pmar1. Sharma and Ettensohn, using pharmacological inhibitors of MEK and WMISH with LvEts1, demonstrated that while Alx1 is activated by Pmar1, its expression is maintained by MAPK signaling and expression of ets1, another transcriptional regulator in the PMC GRN (Sharma and Ettensohn, 2010); thus, ets1 expression under elevated CO2 conditions should also be considered. The present data contrast with those of Todgham and Hofmann, who observed a decrease in Alx1 expression at 1020 μatm in prism (∼40 h.p.f.) stage S. purpuratus (Todgham and Hofmann, 2009). As noted above, these differences may be due to the different developmental stages evaluated and different quantification methods used in each of the studies.
Skeletogenesis-specific genes and proteins
The SM30b expression pattern observed here is similar to previous findings (e.g. Oliveri et al., 2008; Killian et al., 2010) where expression in the PMCs begins at ingression and continues steadily after this (Guss and Ettensohn, 1997; Livingston et al., 2006). Previous work (Duloquin et al., 2007) has demonstrated that, unlike the other terminal biomineralization genes in the PMC GRN that are regulated predominately by the Alx1 signaling cascade, SM30 expression is largely activated by the ectodermal signaling of VegF/VegFR. As CO2 did not perturb VegFR expression, the CO2 induced changes in SM30b expression at full gastrula stage must have occurred by an undetermined mechanism independent of VegFR. Moreover, because spiculogenesis and spicule rudiment formation have been shown to occur normally in embryos raised at 1000μatm (Hammond, 2010), such an increase in SM30b expression at 1000 μatm suggests that at this CO2 concentration the embryo may upregulate SM30b as a way of compensating for the heightened CO2 conditions and to ensure spiculogenesis continues. In contrast, O'Donnell and colleagues observed downregulation of SM30-like expression in early Lytechinus pictus echinoplutei in response to a CO2 concentration of 970 μatm (O'Donnell et al., 2010). However, these findings were in a much later developmental stage, possibly explaining the differences in expression in response to CO2 (i.e. expression at the initiation of spiculogenesis vs expression after significant completion of the larval skeleton). Additionally, variation in response could be related to species-specific differences. Further studies should elucidate the variable patterns of expression in S. purpuratus.
The pattern of expression of Msp130 exhibited in this study is similar to previous findings (Guss and Ettensohn, 1997) with a peak in Msp130 expression at early gastrula stage. Msp130 encodes a PMC-specific cell-surface glycoprotein (Illies et al., 2002). While the exact function of Msp130 is unclear, it has been suggested to play a role in Ca2+ sequestration and/or deposition during spicule formation as the N-linked oligosaccharide chain on the Msp130 protein binds Ca2+ (Carson et al., 1985; Farach-Carson et al., 1989). As elevated CO2 did not affect the expression of Msp130, if Msp130 protein is indeed necessary for Ca2+ sequestration and deposition, then CO2 levels up 1350 μatm may not interfere with Ca2+ uptake and transport.
SM50 gene expression and and protein levels
SM50 gene expression typically begins between hatched blastula stage and PMC ingression (Guss and Ettensohn, 1997) and follows the pattern observed in the current study. SM50 gene expression is necessary for spicule elongation. Blockage of the gene and protein by injection of morpholino antisense oligonucleotides has been shown to prevent biomineralization (Peled-Kamar et al., 2002; Wilt et al., 2008b). Elevated CO2 concentrations up to 1350 μatm appeared to have no effect on SM50 expression (Fig. 4). Given that one of the confirmed regulatory inputs into SM50, Alx1 (Oliveri et al., 2008), was not affected by 1350 μatm CO2, the lack of a shift in SM50 transcript levels is not surprising. As SM50 appears to be necessary for the deposition of the spicule rudiment (Wilt et al., 2008b), uninterrupted SM50 expression would ensure minimal interference with calcite crystal deposition in high CO2 seawater, a notion which is supported by findings that elevated CO2 up to 1000 μatm did not negatively affect the timing of spicule deposition in S. purpuratus gastrulae (Hammond, 2010). It is important to note that there is evidence that the deposition of the calcite crystal and spicule elongation are two different processes; the deposited rudiment is pure calcite crystal whereas spicule elongation involves the conversion of deposited amorphous calcium carbonate to calcite over time in the elongating spicule (Beniash et al., 1997; Ingersoll and Wilt, 1998; Peled-Kamar et al., 2002; Wilt et al., 2008b).
The difference in expression between SM50 transcript levels and SM50 protein levels suggests that elevated CO2 could have an effect on translation rates or protein stability. The significant increase in levels of SM50 protein in early gastrula stage embryos at 1000 μatm vs control CO2 levels suggests that the embryos respond to CO2 at a post-transcriptional level, though whether this is specific to SM50 alone or is a general effect on the proteome is not known. Interestingly, however, the significant decrease in SM50 protein levels at high CO2 levels (1350 μatm) suggests that the embryo may not able to compensate for such a high CO2 concentration. Such a depression may result in negative interference with spicule deposition or elongation; however, further studies into spiculogenesis at elevated CO2 are needed to determine whether CO2 concentrations at this level have an effect on spicule deposition or elongation.
CONCLUSION
The results of this study demonstrate that many of the genes in a subset of PMC GRN genes were unaffected by elevated levels of CO2, and for those that were, the increases, while significant, were only subtle at different life history stages (see summary in Table 4). Interestingly, one of these genes with slightly increased expression was Wnt8, which may reflect the ability of the embryo to compensate for increased CO2 by upregulating this gene, which is necessary for the initiation of the PMC GRN. The slightly increased expression of Wnt8 has positive implications for the process of gastrulation; embryo patterning and endomesodermal specification are most likely unaffected. The second of the affected genes was the biomineralization-specific SM30b. Upregulation of this gene suggests the continued ability to initiate and carry out spiculogenesis. Furthermore, while expression of another biomineralization gene, SM50, did not vary with elevated CO2, the protein it encodes, SM50, displayed an increase in expression at 1000 μatm, but a decrease in expression at 1350 μatm, suggesting the embryo may initially be able to compensate for the increased CO2 by increasing the translation or stability of SM50, but as CO2 increases, this ability may be compromised, with potential negative effects on spicule deposition and elongation.
While further studies are needed to clarify the variable responses to elevated CO2, including studies comparing transcription and translation, it appears that early developmental and skeletogenic gene expression in S. purpuratus embryos was little affected, suggesting that these embryos may be somewhat resilient at the molecular level with regard to elevated CO2 concentrations. Given recent work suggesting that coastal organisms in the eastern Pacific may be exposed to acidic, high PCO2 seawater as a result of seasonal, episodic upwelling more regularly than once believed (Feely et al., 2009; Hauri et al., 2009), the seeming robustness is not altogether surprising. The seasonal, potentially long-term exposures experienced by S. purpuratus, a keystone benthic species along the North American west coast, may have allowed for some degree of adaptation to elevated CO2 concentrations.
The present study is one of the first to investigate embryonic responses to elevated CO2 in PMC GRN-specific genes, especially those upstream of the terminal biomineralization genes. Moreover, it attempts to connect the expression of genes involved in crucial developmental processes and gene batteries to the changing environmental conditions caused by anthropogenically driven CO2-induced ocean acidification. The ecological development approach to understanding the possible consequences for early life history stages will increase our understanding of key developmental processes in context while enabling a greater ability to forecast how marine organisms will or will not respond to their rapidly changing environments.
FOOTNOTES
FUNDING
This study was supported in part by a National Science Foundation (NSF) GRFP Fellowship to L.M.H., an NSF grant [OCE 0425107] to G.E.H., and funds from the University of California in support of a multi-campus research program, Ocean Acidification: A Training and Research Consortium (http://oceanacidification.msi.ucsb.edu/) to G.E.H.
Acknowledgements
We would like to thank Dr Andrew Dickson (Scripps Institution of Oceanography) for his support in developing protocols for our ‘in-house’ seawater analysis, Anna MacPherson for assistance with sample collection and water chemistry, and Dr Pauline Yu for guidance with protein analysis. Additionally, we would like to thank Dr Steven Gaines for statistical advice and Dr Kathleen Foltz for insightful comments.