The evolution and future of carbonate precipitation in marine invertebrates: Witnessing extinction or documenting resilience in the Anthropocene?

The cumulative and continuing emissions of anthropogenic emissions of CO₂ since the beginning of the industrial revolution are projected to decrease the pH of the surface oceans an additional 0.2 to 0.5 pH units by the year 2100 (IPCC, 2013), leading to a nearly 50% drop in carbonate ion concentration (Feely et al., 2009). It has also been suggested that the average pH decrease and overall diel pH variability could be even more pronounced in specific areas, such as coastal seas (Jury et al., 2013; Melzner et al., 2013). The change in carbonate chemistry potentially could result in a decreased ability for marine organisms to precipitate CaCO₃ (Doney et al., 2009), and has been experimentally correlated with reduced calcification rates as well as changes in ecosystem structure (Gazeau et al., 2007; Anthony et al., 2008; Cohen et al., 2009; Sheppard Brennand et al., 2010). However, the effects of ocean acidification on calcifying marine invertebrates are not consistent and suggest a degree of physiological acclimation and potential genetic adaptation (Fine and Tchernov, 2007; Ries et al., 2009; Kelly et al., 2013; Shamberger et al., 2014). One suggested mechanism for the retention of biological calcification is low-pCO₂ refugia (Tittensor et al., 2010; Manzello et al., 2012). While processes responsible for carbonate precipitation in marine invertebrates are not completely understood, we suggest that a specific suite of proteins allows biomineralization to persist in these organisms even under low pH conditions.

Although biomineralization is a common phenomenon across the tree of life, it is difficult to reduce the process to simple, general principles (Lowenstam and Weiner, 1989; Dove et al., 2003; Müller, 2011). By far, the most common biominerals produced in the contemporary ocean are calcium carbonates, especially aragonite and calcite. Although the two ingredients required to produce these mineral forms, calcium and bicarbonate ions, are present in seawater in excess of 10 and 2 mM, respectively, CaCO₃ does not spontaneously precipitate in the contemporary ocean. The precipitation reaction requires catalysis by living organisms to overcome kinetic barriers.

The carbonate structures formed by metazoans are often well preserved in the fossil record. Their elemental and isotopic compositions have been extremely helpful in reconstructing the thermal and chemical history of the ocean. Carbonates themselves have been examined at different stages of the biomineralization process in marine invertebrates. For more than 50 years, it has been generally understood that the precipitation and organization of biominerals is modulated by specific biomolecules, especially proteins (Müller, 2011). However, how the proteins function to facilitate the precipitation of minerals in marine invertebrates is poorly understood.

Over the past decade there has been a proliferation in the number of genomes and transcriptomes of mineralizing invertebrates (Jackson et al., 2006; Sea Urchin Genome Sequencing Consortium et al., 2006; Joubert et al., 2010; Shinzato et al., 2011; Traylor-Knowles et al., 2011; Takeuchi et al., 2012). These data sets have allowed numerous studies of the proteomes of sea urchin teeth and spicules, mollusk shells, and most recently coral skeletons (e.g., Mann et al., 2008b, a; Mann et al., 2010b; Marie et al., 2010; Marie et al., 2011a; Marie et al., 2011b; Zhang et al., 2012; Drake et al., 2013; Marin et al., 2013; Ramos-Silva et al., 2013). We are poised to compare the evolutionary history of biomineralizers in the fossil record (Knoll, 2003) with this developing biochemical and molecular biological understanding of how the minerals are formed by the animals.

Here we review over 100 studies in a search for similarities between marine invertebrate carbonate organic matrix (COM) proteins that have been identified through genomics, transcriptomics, and proteomics. We focus on stony corals (Cnidaria), bivalve and gastropod mollusks (Mollusca), and sea urchins (Echinodermata), thereby covering representatives from three widely divergent metazoan phyla. We first outline the physiology and known biomineralization processes in each phylum, then detail the COM protein groups conserved, if not in sequence, then in function. Finally, we discuss the affects of impending ocean acidification on both the function of some of these COM proteins as well as the continuation of these organisms.

Despite their shared use of CaCO₃, evolutionary evidence suggests that biomineralization evolved independently several times across the tree of life (Knoll, 2003; Murdock and Donoghue, 2011). Over the ∼600 million years during which Cnidaria, Mollusca, and Echinodermata diverged from shared ancestors (Erwin et al., 2011), biomineralization was selected to support several functions. The first is as a substrate for growth. In modern stony corals, precipitation of aragonite allows for growth of the thin film of animal on a hard, continuous substrate. The process begins when larvae settle and colonies adhere to a benthic matrix. In contrast, bivalve and gastropod mollusks precipitate aragonite and/or calcite shells as a protective wall surrounding soft tissues. These structures are very effective in defending against predators. In sea urchins, calcitic spines are also a deterrent to predators, but being relatively light, they do not interfere with motility of the animal.

Stony corals in the phylum Cnidaria are the earliest known metazoans to precipitate calcium carbonate (Knoll, 2003). Cnidaria also contain Order Alcyonacea, or soft corals, which, unlike stony corals, produce small internal sclerites, or ∼ mm-sized calcite plates. In Paleozoic time, stony coral structures appear to have been formed primarily from the precipitation of calcite, whereas over the past ∼ 250 Ma, with the appearance of Scleractinian corals, these organisms have evolved a de novo aragonite precipitation process (Porter, 2010). This period also broadly corresponds to their transition as host animals to symbiotic photosynthetic dinoflagellates of the genus Symbiodinium (Stanley and Schootbrugge, 2009; Tchernov et al., 2012). While some corals that lack symbionts can produce CaCO₃ skeletons, the builders of the relatively rapidly accreting massive reefs observable from space are hermatypic, or symbiont-containing.

Up to 90% of the carbon requirement of the coral host is translocated in the form of glycerol and/or glucose from Symbiodinium spp. (Falkowski et al., 1984; Burriesci et al., 2012); additionally, corals can obtain nitrogen and phosphorus by feeding on dissolved and particulate matter from the overlying water supply (reviewed by Houlbrèque and Ferrier-Pagès, 2009). Some of the energy in the translocated carbon is suggested to support biomineralization, a phenomenon termed ‘light enhanced calcification’ (Kawaguti and Sakumoto, 1948; Goreau, 1959; Rinkevich and Loya, 1984)– although the mechanism(s) remain unclear (review by Tambutté et al., 2011).

Stony corals precipitate an extracellular skeleton in the form of aragonite under their calicoblastic ectodermal layer (Figures 1A & B). Recently, there have been suggestions of an amorphous phase through atomic force microscopy and synchrotron-based XANES (Cuif et al., 2008; Falini et al., 2013). Planulae begin to precipitate aragonite only upon settling. Proteins related to adhesion have been localized toward the aboral end immediately prior to settlement (Hayward et al., 2011). Growth rate depends strongly on environmental factors, including irradiance and temperature, with some corals showing gross precipitation as high as 10 kg m^-2 yr^-1 (Milliman, 1993). Several N-terminal sequences have been obtained from proteins associated with soft coral calcitic sclerites, but the vast majority of research on COM proteins from corals has focused on the tropical symbiotic stony corals.

Mollusks, the oldest extant biomineralizing bilaterians (reviewed by Knoll, 2003), encompass a far more advanced and varied body plan than stony corals. Generally, the interior of the body with defined digestive organs is protected from the exterior environment by one or two CaCO₃ shells (although in some classes this protection is much reduced or absent; Marin et al., 2012). The shell is generally covered with a thin organic layer, the periostracum, which not only separates the shell from the overlying water, but also plays a role in the biomineralization process. While some gastropod and bivalve mollusks contain endosymbiotic Symbiodinium spp., the majority are filter feeders, particularly when sessile, or scrape off algae from surfaces when motile (Klumpp et al., 1992).

In larval gastropods and bivalves, a protein-rich layer, the periostracum, is secreted from a gland in the shell prior to onset of biomineralization (Figure 1). Ancestral mollusks have been proposed as solely aragonite producers (Porter, 2010), but in modern species the initial mineral is often amorphous calcium carbonate (ACC), though can be calcite or aragonite (Weiss et al., 2002). In adults, the periostracum is secreted by cells in the mantle folds and is maintained throughout the life of the mollusk, thus providing a barrier between the mineral CaCO₃ and the environment (Marin et al., 2012; Figure 1C & D). Additionally, the periostracum serves as a preliminary scaffold on which the organism lays down CaCO₃ (Checa, 2000). Together the periostracum and mantle provide a distinct area, the extrapallial space, in which calcification occurs in the presence of COM composed of chitin and proteins (Nudelman et al., 2006; Marin et al., 2012; Figure 1D). The COM proteins, polysaccharides, and lipids secreted by the mantle guide nucleation, elongation, and finally termination of aragonite and/or calcite crystals in well-defined locations and morphologies (reviewed by Marin and Luquet, 2004). Additionally, specific COM proteins, such as Asprich, stabilize the ACC (Politi et al., 2007) and influence the mechanical properties of the biomineral (reviewed by Weiner and Addadi, 2011). Vesicles containing mineral precursors are also exported from specific cells within the animal and aid in facilitating biomineralization in some mollusks (reviewed by Addadi et al., 2006).

Although echinoderms display radial symmetry as adults, they evolved from a bilaterian ancestor shared with mollusks and display bilateral symmetry as embryos (Hyman, 1955). Echinoderms have a defined digestive tract consisting of mouth, pharynx, esophagus, stomach, intestine, and anus. While some echinoderms are carnivorous, sea urchins – currently the only echinoderm with a sequenced genome (Sea Urchin Genome Sequencing Consortium et al., 2006) – are predominately herbivores. They use calcified spicules for motility, protection, and to preserve body shape, and five polycrystalline magnesium calcite teeth for foraging (Ma et al., 2008).

Larval planktonic sea urchins secrete CaCO₃ spicules and juveniles produce a calcified test and spines, while adults construct these features plus magnesium-enriched calcite teeth, pedicellariae, and ossicles in their tube feet, all in close proximity to mineralizing cells (Killian and Wilt, 2008). Larval sea urchins begin to precipitate CaCO₃, guided by COM from primary mesenchyme cells, prior to settling (Figure 1E & F). As is often the case with bivalve and gastropod mollusks, it appears that this CaCO₃ precipitation begins in the form of amorphous calcium carbonate (ACC) but within days changes to calcite (Politi et al., 2008). Conversion of ACC to calcite is also known in adult urchins (Politi et al., 2004; Killian et al., 2009), and has been suggested as a general tactic in biomineralization, to have mineral precursors ready for crystallization (reviewed by Gilbert and Wilt, 2011).

The concept of biologically controlled mineralization is widely accepted (Mann, 2001). In stony corals, the “biomineralome” contains an organic component that is 0.1 to 5% by weight with respect to that of the total biomineral and contains a high preponderance of acidic residues in the skeletal proteins (Young, 1971; Cuif et al., 2004; Mass et al., 2012). The proteome of coral COM contains an assemblage of adhesion and structural proteins as well as highly acidic proteins (Drake et al., 2013; Ramos-Silva et al., 2013). These highly acidic proteins exhibit pKas of < 5 and a high relative composition (≥ 28%) of the acidic amino acids, aspartic (Asp) and/or glutamic (Glu) acid. Four coral-acid rich proteins (CARPs), one of which was found in both coral skeleton proteomic studies (Drake et al., 2013; Ramos-Silva et al., 2013), were individually cloned and purified. All four proteins precipitate calcium carbonate in vitro in artificial seawater at ambient carbonate concentrations at pH 8.2 and 7.6 (Mass et al., 2013). These results have led to the proposal of a nucleation mechanism in corals in which clusters of Asp and/or Glu residues coordinately bind Ca²⁺ ions, potentially leading to a Lewis acid displacement of protons in bicarbonate anions (Mass et al., 2013). The resulting reaction alters the mineralizing environment below the calicoblastic ectoderm such that the pKa of the local environment on the protein surface decreases and carbonate crystals are precipitated on the protein matrix, even at relatively low ambient pH (Mass et al., 2013).

As with stony corals, the organic component of the bivalve and gastropod mollusk shell minerals generally comprises between 0.01 to 5% by weight, with many of the (glyco)proteins containing a high proportion of Asp residues and/or post-translational decorations of sulfate groups (Lowenstam and Weiner, 1989). As in corals, these proteins are highly acidic. Initially much of the focus on biomineralizing proteins in mollusks focused on these highly acidic proteins and their possible template-specific roles in nucleating precipitation, with distances between acidic residues in β-sheet conformation corresponding to distances between Ca²⁺ ions in both calcite and aragonite (Weiner and Hood, 1975). More recently, the role of highly acidic proteins in mollusk and echinoderm mineralization has grown to encompass the inclusion of these proteins in a hypothesized organic gel, the potential stabilization of a transitional amorphous calcium carbonate step, and the binding and inhibition of growing crystal growth faces (reviewed by Marin, 2007). Additionally, a number of proteomic comparisons have shown that COM proteins include many of unknown function as well as those specific to aragonite versus calcite (e.g., (Jackson et al., 2006; Marie et al., 2010; Marie et al., 2012; Marin et al., 2013)). Of the proteins with known function, many are involved in maintenance of the extrapallial environment and in reworking of calcium carbonate.

Analyses of the proteins in COM from spicules, tests, and teeth of sea urchins have yielded a non-redundant list of over 200 individual molecules (Mann et al., 2008b, a; Mann et al., 2010b). While some of these proteins appear to be specific to one type of mineral or structure (spines versus teeth, etc.), many overlaps are observed (reviewed by Gilbert and Wilt, 2011). Some of these proteins have been localized to larval primary mesenchyme cells, but other cells, potentially secondary mesenchyme cells, are more likely responsible for biomineralization in adult animals (Killian and Wilt, 2008). Although highly acidic proteins are also found in sea urchin biominerals (George et al., 1991; Illies et al., 2002; Mann et al., 2008a), non-acidic proteins that control either the mineralizing environment or activity of other proteins are also appreciated as essential (example is SM30 family Wilt et al., 2013).

The term ‘acidic’ protein has a variety of definitions, but ultimately describes a protein that possesses a net negative charge on some or all of the primary sequence or through post-translational modification. In an attempt at clarity, we compiled a list of the ‘acidic’ proteins and their gene accession numbers in Table 1. It is clear from a phylogenetic analysis that these proteins arose independently several times in the evolution of animals through convergent evolution, with gene fusion potentially conferring additional roles to these peptides (Miyamoto et al., 1996; Mass et al., 2013). These proteins appear to be essential for biomineralization. We conducted a protein sequence similarity analysis, using > 1500 non-redundant coral, mollusk, and sea urchin mineral-derived proteins reported in the literature. Through these analyses, we found only one case of cross-phyla sequence similarity among any of the acidic biomineralization proteins for which sequences are available. Prismalins, shematrins, and some shell matrix proteins from mollusks show 30 to 40% similarity with several collagens from echinoderms and corals (Figure 2, SI Table 1); this is likely due to the high glycine content of shematrins (Yano et al., 2006) similar to the primary structure of collagens.

Within corals, several acidic proteins show sequence similarity across Orders Scleractinia and Alcyonacea. These include stony coral CARPs 4 and 5 (Drake et al., 2013), SAARPs 1 and 2, SOMP, and Amil-SAP2 (Moya et al., 2012; Ramos-Silva et al., 2013) and the soft coral protein ECMP-67 (Rahman et al., 2011b; SI Table 2). Although not yet analyzed in Acropora digitifera and Pocillopora damicornis skeleton, CARP4 homologues are found in transcriptomes of these species as well (Shinzato et al., 2011; Vidal-Dupiol et al., 2013). The CARP4 sub-family has been shown to be both highly conserved across and limited to stony corals (Drake et al., 2013), and now the N-terminal sequence from the soft coral, Lobophytum crassum, may be assigned to this family based on sequence alignment (Figure 3).

Stony corals precipitate aragonite whereas soft corals form calcite, suggesting that the CARP4 sub-family arose in Anthozoa prior to divergence between Scleractinia and Alcyonacea. However, because Actiniarians, or sea anemones, which lack the CARP4 subfamily, diverged from Scleractinia after Alcyonacea (Berntson et al., 1999), they may have lost the gene. It has been proposed that ECMP-67 encourages the precipitation of only calcite (Rahman et al., 2011b), despite its similarity to the aragonite-associated CARP4 and SAARP1 (Figure 3). Interestingly, several mollusk proteins have also been suggested to function in both aragonite and calcite precipitation (MS17; Feng et al., 2009) or in polymorph control (Aspein; Takeuchi et al., 2008). Additionally, Aspein, Asprich, and Pif, which all are components of molluscan shells, share 30% sequence similarity (SI Table 2; Tsukamoto et al., 2004; Gotliv et al., 2005; Suzuki et al., 2009). Thus although the acidic proteins are independently derived, this sequence comparison suggests a common mechanism of CaCO₃ precipitation by acidic proteins in the three marine invertebrate phyla.

Some of the biomineralizing proteins appear to have dual functions. For example, perlwapin, prismalin, molluscan shell prism nacre protein, shematrins, and several collagens, are all implicated in organization and structural support of the bioinorganic matrix. These are called ‘other’ gene ontology (GO) terms in Figure 2 due to equal distribution of collagen, membrane, and extracellular GO assignments. These proteins may also serve a dual role as terminators of biomineralization and are not the only COM proteins, both acidic and non-acidic alike, shown to have crystallization inhibition activities (e.g., Ma et al., 2007; Politi et al., 2007; Feng et al., 2009). A dual nature of a biomineralizing protein is also observed for the mollusk protein, nacrein, which exhibits both internal acidic regions and carbonic anhydrase activity (Miyamoto et al., 1996). In sea urchins, SpP16 is an acidic protein with a transmembrane region, suggesting that it is anchored to the primary mesenchyme cells to which it has been localized; while its precise mechanism in calcitic spicule formation remains to be determined, it shows some similarities to other calcium binding proteins to regulate calcium transport, again suggesting dual function (Illies et al., 2002).

Carbonic anhydrases, which interconvert CO₂ and HCO₃^-, are integral within carbonate-based biominerals (Bertucci et al., 2013). Their use in biomineralization has been suggested to date back to the earliest calcifying metazoans (Jackson et al., 2007). Gene duplication events and further mutations since the Paleozoic have resulted in a wide variety of the enzyme, many of which are localized within the bioinorganic matrix. Our proteomic analysis revealed representatives of divergent carbonic anhydrases in the biomineral matrix in each of the three invertebrate phyla (Table 2), which reinforces their functional necessity in the evolution of the precipitation of extracellular calcium carbonate in each of the three metazoan phyla. Indeed, multiple carbonic anhydrases are found in COM of each phylum, suggesting a level of redundancy for the biomineralization process.

Just as bicarbonate is required for the biomineralization mechanism, so is Ca²⁺. In animals, Ca²⁺ is transported into and out of cells by plasma membrane-bound Ca-dependent ATPases. These enzymes are widely distributed in mollusks, sea urchins, and corals (Zoccola et al., 2004; Wang et al., 2008; Todgham and Hofmann, 2009). However, of the more than 100 reports of COM, only one identified a Ca-ATPase in a ‘biomineralome’ (Joubert et al., 2010). An additional mechanism for Ca²⁺ transport in corals is the translocation of calcium from the bulk water to the site of mineralization (Gagnon et al., 2012) by a so-called “paracellular” pathway (Tambutté et al., 2012). However, without a chelating or trapping molecule, this mechanism would not concentrate calcium ions. Finally, proton pumps not associated with mitochondrial ATPases, an additional method to modify the saturation state of the mineralizing environment (Corstjens et al., 2001), were not observed in COM by the authors of any of the studies used in our bioinformatics analysis. Additionally, none of the 295 non-redundant proteins conserved across phyla returned molecular function GO terms associated with non-mitochondrial proton pumps.

A number of non-acidic COM proteins are either conserved by sequence (Figure 2) or function (Table 2). For heuristic purposes, we have aggregated these proteins by function into major groups: (i) proteins that participate in adhesion of cells to their substrates and/or provide a scaffold for organizing the biomineral, and (ii) proteins that signal and regulate biomineralization. Our list is not exhaustive (see SI Table 2). However, it is useful in providing a framework for comparing similarities and differences in COM proteins across the three phyla.

Actins and beta-tubulin, cytoskeletal proteins that work as polymerization molecular motors, are present in COM of stony corals, mollusks, and sea urchins (Figure 2). Although they could be contaminants from the animal component (i.e., not truly part of the COM; Mann et al., 2008b), physiological and genomic data suggest they may be part of the biomineralization “toolkits” for these organisms. Inhibition of these proteins significantly reduces organic matrix synthesis and calcification in corals (Allemand et al., 1998). While similar studies have not been conducted on sea urchin biomineralization, actin is crucial to the primary mesenchyme cell ingression stage just prior to onset of embryonic biomineralization, with actin inhibition resulting in loss of ingression (Wu et al., 2007). Secondly, biomineralization in paralogue organisms such as diatoms, which precipitate amorphous silicate, involves actins and tubulins (Hildebrand et al., 2008). In contrast, more likely contaminant proteins, such as nuclear and mitochondrial proteins, are noted in SI Table 1.

A number of proteins that adhere to or interact with actin and tubulin are found in COM across marine invertebrate phyla, but are not conserved by amino acid sequence, and therefore are assumed to be independently evolved (Table 2). For example, laminins, a family of glycoproteins, along with fibronectins and integrins, serve as structural support in mollusks, echinoderms, and corals, respectively (Table 2). Peroxidasins and thrombospondins (in corals and sea urchins) may also serve as extracellular structural and adhesive proteins (Table 2; Péterfi et al., 2009).

Finally, collagens, which provide a flexible framework and confer structural support for biominerals, appear to be widely distributed in corals and sea urchins (Figure 2, Table 2). It is unclear whether collagens are actually found in mollusks (Blank et al., 2003; Xuan Ri et al., 2007). Alternatively, chitin, a potential structural replacement for collagen is found in mollusks (reviewed by Ehrlich, 2010). In this case, chitin, like collagen, can provide a highly structured scaffold on which COM proteins can bind (reviewed by Falini and Fermani, 2004). It has also been suggested to have a more direct role in polymorph control (Falini et al., 1996). Chitin synthase (Joubert et al., 2010) and a chitin-binding protein (Mann et al., 2012) have been detected in mollusk shell proteomes; they do not show cross-phyla sequence similarities in our analysis (SI Table 2).

Low-density lipoprotein (LDL) receptors are ancient proteins that aid in signal transduction across cell membranes and are responsible for cell specification and axis patterning in cnidarians and sea urchins (Wikramanayake et al., 2004; Lee et al., 2006; Lee et al., 2007). Notch proteins, which contain an epidermal growth factor domain (EGF), are calcium-binding transmembrane signal transduction pathway proteins that interact with LDL receptors (Baron, 2003). In sea urchins, they direct secondary mesenchyme cell development and spiculogenesis (Suyemitsu et al., 1990; McClay et al., 2000). Several of these proteins, with ‘membrane’ GO terms assigned, are reported in sea urchin and coral COM (Figure 2).

Bone morphogenic proteins (BMPs) are responsible for signal gene regulation in apatite producing cells (reviewed by Canalis et al., 2003) and up-regulate Notch signaling genes in osteoblast precursor cells (de Jong et al., 2004). BMPs are reported in COM from all three phyla (Table 2) but, as ‘extracellular’ GO terms, are only conserved between mollusks and sea urchins (Figure 2). As signal transduction pathway proteins, BMPs from corals and mollusks induce mesenchyme cell differentiation and osteogenesis (Zoccola et al., 2009; Takami et al., 2013) and are involved in sea urchin cell specification and axis patterning (Angerer et al., 2000). They have also been immunolocalized to calicoblastic ectodermal cells in corals (Zoccola et al., 2009) and a novel protein that shows structural (from structure prediction) but not sequence similarity to the BMP inhibitor Noggin was recently observed in coral COM (Drake et al., 2013).

In addition to cell signaling, reworking of individual COM proteins is important to the function of the complex. Matrix metalloproteinases (MMPs) are specific extracellular matrix degradation enzymes that require cations for their catalytic activity. COM MMPs, as extracellular proteins, show sequence similarity between corals and sea urchins (Figure 2). MMPs affect sea urchin spicule elongation but not nucleation (Ingersoll and Wilt, 1998) and regulate extracellular matrix in humans (reviewed by Woessner, 1991; Birkedal-Hansen et al., 1993). They are found in mollusk hemolymph and an MMP inhibitor has been extracted from mussel pearls (Mannello et al., 2001; Jian-Ping et al., 2010), suggesting that they may be present in mollusk COM as well.

Ubiquitin is a protein that attaches to and signals other proteins for degradation or relocation, or inhibits protein binding or activity. It is highly conserved in COM proteins across all three phyla and is generally assigned a ‘membrane’ cellular component GO term (Figure 2). Ubiquitylated proteins are reported from the prismatic layer of mollusks; removal of ubiquitin decreased the ability of these proteins to inhibit calcium carbonate mineralization (Fang et al., 2012). Hazelaar et al. (2003) proposed that, in diatoms, ubiquitin may signal templating proteins for degradation as the cell walls develop pores; a similar role in invertebrate calcium mineralization has yet to be examined.

Morphological, phylogenetic, and now proteomic evidence indicates that marine CaCO₃ precipitation is a convergent evolutionary process. Coral, mollusks, and sea urchins have all evolved their own suites of acidic proteins to catalyze the nucleation of CaCO₃ (Illies et al., 2002; Gotliv et al., 2005; Drake et al., 2013) and co-opted a number of adhesion, structural, signaling, and regulation proteins to provide structure and environmental stability to the calcifying regions (Marin and Luquet, 2004; Zhang and Zhang, 2006; Allemand et al., 2011). This convergence suggests that there are general roles for biomolecules in the biomineralization process and that, although biomineralization toolkits may not be conserved across phyla, the requirements to have these biological roles fulfilled is. Further, the precise macroscopic morphologies of skeleton, shell, and spicule/teeth, etc. may then be driven by those proteins or other biomolecules that are functionally individual not only to each phylum, but also often to specific minerals. Examples of these functionally individual proteins, which in some cases have been reviewed elsewhere, are lysine-rich matrix proteins and amorphous calcium carbonate binding proteins in mollusks (Zhang et al., 2006; Ma et al., 2007), galaxins in corals (Fukuda et al., 2003), and the mesenchyme specific cell surface glycoproteins in sea urchins (Leaf et al., 1987; Killian and Wilt, 2008).

Estimates of CO₂ emissions and the projected effects on upper ocean ecosystems over the next 100 years are dire, especially for marine calcifiers that will face the combined issues of warming and ocean acidification (Raven, 2005; Feely et al., 2009; Ries et al., 2009; Hoegh-Guldberg and Bruno, 2010; Pandolfi et al., 2011). Models of the average calcite saturation state (Ω_calcite) in the surface ocean projects a decrease from 5 to 2 in the next 100 years (Doney et al., 2009; Norris et al., 2013). Studies of biomineralizing marine invertebrates indicate a variety of potential responses to ocean acidification, from potential decreased calcification rates (Ries et al., 2009; Holcomb et al., 2010; Kroeker et al., 2010) to ecosystem shifts (Anthony et al., 2008; Crook et al., 2012). Additionally, each of the marine invertebrate mineralizers discussed here have representatives that naturally persist in waters with low calcium carbonate saturation (i.e., > 1000 m depth or in regions of upwelling, high respiration rates, or rainwater runoff; e.g., (Anagnostou et al., 2011; Shamberger et al., 2014)). However, in addition to examining the responses of current marine biomineralizers, it is useful to consider both the micro- and macroscopic effects of increased atmospheric CO₂, particularly as it pertains to ocean acidification.

As detailed above, several highly acidic coral proteins function at ambient and low pH (Mass et al., 2013). This capacity suggests that while crystal growth rate and morphology can be dependent of the saturation state of the medium (Holcomb et al., 2009), the pKas of these and likely other highly acidic proteins are three orders of magnitude lower than seawater. Thus, in spite of the fact that pH may decline by up to 0.5 pH units in the coming century, the reactivity of these proteins in the precipitation of carbonates will remain virtually unchanged. This built-in molecular resilience will likely allow the nucleation reaction to continue. Although this resilience does not address the physiological response of marine biomineralizers to a surrounding medium whose pH is stressful to processes beyond the nucleation of minerals (Lannig et al., 2010; Stumpp et al., 2011; Moya et al., 2012), these organisms are not static and there is evidence of genetic and transcriptional adaptation to low pH conditions (Moya et al., 2012; Hüning et al., 2013; Kelly et al., 2013).

Given the inevitable acidification of the contemporary ocean in the coming decades it is potentially also instructive to examine the geological record of analogues. One possible analogue is the Paleocene Eocene Thermal Maximum (PETM; Norris et al., 2013). This period, 55.5 to 55.9 million years ago, was characterized by a 4 to 5° C increase in sea surface temperatures in the tropics (Zachos et al., 2003) and nearly 1000 ppm increase in CO₂ in the atmosphere (Beerling and Royer, 2011). While the PETM corresponds with a dramatic reduction in CaCO₃ in the geologic record (Zachos et al., 2005) and a decrease in the coverage of coral reefs (reviewed by Norris et al., 2013), all marine invertebrate taxa discussed here survived through this extraordinary period. Indeed, some experienced rapid radiation and diversification in the Eocene (e.g., Stanley, 2003).

The survival of the major carbonate-precipitating marine invertebrates in the face of ocean acidification is likely because (i) the site of biomineralization is removed from the surrounding medium, (ii) the proteins responsible for the precipitation of carbonates function at low pH, and (iii) genetic adaptation allows the organisms to persist, albeit perhaps not optimally, in their new environments. Additionally, recent research suggests that these organisms are able to modify their mineralizing environment in response to pH stress (Yuen et al., 2006; Venn et al., 2013). This ability suggests that the biomineralization processes will almost certainly persist throughout the Anthropocene. It appears that these organisms are far more resilient than is often acknowledged. However, from a molecular biology perspective, significant research remains toward understanding 1) the expression rates of known COM proteins under variable environments, as well as the biochemical cascades responsible for any changes; 2) the interactions of the COM proteins with each other to regulate the initiation, elongation, and termination steps of mineralization; and 3) how signals of COM protein evolution, which also guided biomineralization in the past, may be retained in fossil minerals.

All data were taken from publically available datasets. Original sources and accession numbers of all sequences are noted in Table S2.

© 2014 Drake, Mass and Falkowski. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

We thank Athena Fu for collecting sequence data, Mary Katherine Battles for drawings, and Morgan Schaller and Ehud Zelzion for constructive discussions.