HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Data Supplement Alert me when this article is cited Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Reitzel, A. M. Articles by Finnerty, J. R. Search for Related Content PubMed PubMed Citation Articles by Reitzel, A. M. Articles by Finnerty, J. R. Related Collections Evolution Genomics Cnidarians

Genomic Survey of Candidate Stress-Response Genes in the Estuarine Anemone Nematostella vectensis

Adam M. Reitzel^1,*, James C. Sullivan^2,*, Nikki Traylor-knowles^2,* and John R. Finnerty^2,

¹ Biology Department, Woods Hole Oceanographic Institution, Woods Hole, Massachusetts 02543
² Department of Biology, Boston University, 5 Cummington Street, Boston, Massachusetts 02215

To whom correspondence should be addressed. E-mail: jrf3{at}bu.edu

	Abstract

TOP Abstract Introduction Materials and Methods Results Chemical stress response Innate immunity and biological... Wound healing Comparison of protein counts... General Discussion and... Early evolution of animal... Evolution of stress tolerance Microevolution of stress... Nematostella vectensis as a... Literature Cited

 
Salt marshes are challenging habitats due to natural variability in key environmental parameters including temperature, salinity, ultraviolet light, oxygen, sulfides, and reactive oxygen species. Compounding this natural variation, salt marshes are often heavily impacted by anthropogenic insults including eutrophication, toxic contamination, and coastal development that alter tidal and freshwater inputs. Commensurate with this environmental variability, estuarine animals generally exhibit broader physiological tolerances than freshwater, marine, or terrestrial species. One factor that determines an organism's physiological tolerance is its ability to upregulate "stress-response genes" in reaction to particular stressors. Comparative studies on diverse organisms have identified a number of evolutionarily conserved genes involved in responding to abiotic and biotic stressors. We used homology-based scans to survey the sequenced genome of Nematostella vectensis, the starlet sea anemone, an estuarine specialist, to identify genes involved in the response to three kinds of insult—physiochemical insults, pathogens, and injury. Many components of the stress-response networks identified in triploblastic animals have clear orthologs in the sea anemone, meaning that they must predate the cnidarian-triploblast split (e.g., xenobiotic receptors, biotransformative genes, ATP-dependent transporters, and genes involved in responding to reactive oxygen species, toxic metals, osmotic shock, thermal stress, pathogen exposure, and wounding). However, in some instances, stress-response genes known from triploblasts appear to be absent from the Nematostella genome (e.g., many metal-complexing genes). This is the first comprehensive examination of the genomic stress-response repertoire of an estuarine animal and a member of the phylum Cnidaria. The molecular markers of stress response identified in Nematostella may prove useful in monitoring estuary health and evaluating coastal conservation efforts. These data may also inform conservation efforts on other cnidarians, such as the reef-building corals.

Abbreviations: ECM, extracellular matrix • EST, expressed sequence tag • ROS, reactive oxygen species

	Introduction

TOP Abstract Introduction Materials and Methods Results Chemical stress response Innate immunity and biological... Wound healing Comparison of protein counts... General Discussion and... Early evolution of animal... Evolution of stress tolerance Microevolution of stress... Nematostella vectensis as a... Literature Cited

 
To survive and reproduce, organisms must deploy an array of molecular, physiological, and behavioral responses to avoid or counteract detrimental environmental conditions. These responses to environmental stress are often complex, involving many genes and integrating events at cellular and organismal levels. Except in cases where it is possible to simply flee stressful environmental conditions, an organism's response to most environmental stressors—both natural and anthropogenic stressors—will typically involve changes in baseline gene expression that alter physiology, development, or behavior. Therefore, to study organismal stress responses, we must integrate environmental data, gene expression data, and organismal data (Ankley et al., 2006; Lee and Mitchell-Olds, 2006). The ongoing proliferation of genomic data is enabling comparative stress-response studies on a phylogenetically wider sampling of taxa occupying a broader range of habitat types (Lettieri, 2006). It is now possible to compare evolved differences between species occupying diverse environments at both shallow and deep evolutionary divergences.

Nematostella vectensis as an estuarine sentinel species
Nematostella vectensis Stephenson, the starlet sea anemone, is one of the relatively small number of macroscopic animals that are year-round residents in estuarine environments. It is a small infaunal anemone (typically <1 cm) inhabiting salt marshes, saline lagoons, and other sheltered brackish environments. The species’ range includes the eastern Pacific, western Atlantic, northern English Channel, and western North Sea (Hand and Uhlinger, 1994).

In recent years, Nematostella has emerged as a model system for developmental (Finnerty et al., 2004; Matus et al., 2006) and genomic studies (Sullivan et al., 2006; Putnam et al., 2007), and it has proven particularly informative for reconstructing the functional evolution of developmental regulatory genes whose origins can be traced to the cnidarian-triploblast common ancestor. It is useful for such studies because (1) it is an outgroup to the superphylum Triploblastica (=Bilateria; Fig. 1), (2) it is easy to culture in the laboratory through its entire sexual and asexual life history, and (3) its genome appears to have evolved in a relatively conservative fashion. A general finding of comparative genomic studies involving Nematostella is that the sea anemone exhibits remarkable genomic complexity and a surprising degree of genomic conservation relative to vertebrates, despite its extreme evolutionary distance from vertebrates (Technau et al., 2005; Sullivan et al., 2006; Putnam et al., 2007; Sullivan and Finnerty, 2007).

View larger version (14K): [in this window] [in a new window]   Figure 1. Metazoan phylogeny. The Cnidaria are the likely sister to the Bilateria (= Triploblastica). Dates for labeled ancestors are compiled from a number of sources (Aris-Brosou and Yang, 2002; Peterson et al., 2004; Peterson and Butterfield, 2005).

A first step in understanding how organisms sense and respond to stressors from the environment is to identify the genetic components of stress-response pathways. In this study, we searched the Nematostella genome to identify homologs of proteins known to be involved in sensing stress, counteracting particular stressors, and repairing damaged tissue. We divided the stress-response genes into three broad functional categories: (1) those that mediate chemical stresses, (2) those that counteract pathogens, (3) and those that underlie wound healing.

	Materials and Methods

TOP Abstract Introduction Materials and Methods Results Chemical stress response Innate immunity and biological... Wound healing Comparison of protein counts... General Discussion and... Early evolution of animal... Evolution of stress tolerance Microevolution of stress... Nematostella vectensis as a... Literature Cited

 
Identifying proteins implicated in stress response in other taxa
We compiled a list of proteins known to be involved in (1) chemical stress response, (2) innate immunity, and (3) wound healing. To identify genes involved in chemical stress response, we utilized a collection of "defensome" motifs compiled by Goldstone et al. (2006), with the following additions: zinc transporter (Pfam identifier PF02535); Ctr copper transporter family (PF04145); Cation efflux family (PF01545). To identify candidate genes involved in pathogen defense, we consulted the Innate Immunity Phylogenomic Explorer, ver. 2.0 (Krishnamurthy et al., 2006). We first eliminated those protein families whose functional connection to innate immunity is relatively tenuous—i.e., proteins annotated as "related to" innate immunity or that might "putatively" play a role in innate immunity. This left 128 protein families to which we could ascribe a role in innate immunity with higher confidence. To identify candidate genes involved in wound healing, we culled the recent literature on wound healing and regeneration (Kiritsy et al., 1993; Galko and Krasnow, 2004; Alvarado and Tsonis, 2006; Huxley-Jones et al., 2007).

Identifying homologous proteins in Nematostella
After assembling a list of stress-response query genes from other taxa (Fig. 2a), we sought to identify putative homologs in Nematostella. Our principal search strategy utilized conserved protein domains cataloged in the Pfam database, release 17 (Finn et al., 2006). For innate immune genes and chemical defense genes, lists of conserved Pfam domains had already been compiled (Goldstone et al., 2006; Krishnamurthy et al., 2006). For wound-healing genes, we identified conserved domains in the query proteins using the conserved domain search function at Pfam (Fig. 2b). We then identified all Nematostella proteins that contained one of these conserved Pfam domains (Finn et al., 2006) by searching a database of predicted proteins at StellaBase 1.0 (Sullivan et al., 2008) using a Hidden Markoff Model search algorithm (Fig. 2c; Durbin et al., 1998). Nematostella proteins were scored as possessing a particular Pfam domain if the match to a query sequence received an Expect value 1e–6. These proteins predicted by using StellaBase were then cross-referenced with predicted Nematostella proteins at NCBI using BLASTp (Fig. 2d). Since this search identifies all proteins matching a particular domain, even those that are not implicated in stress response, the proteins identified in our search were compared against the human Ref Seq database (downloaded 1 Oct. 2007, containing 24306 proteins) using BLASTp to determine whether a known stress-response gene is the best match (Fig. 2e–f).

View larger version (25K): [in this window] [in a new window]   Figure 2. Workflow for identifying stress-response proteins in Nematostella, based on Pfam domain searches and BLAST similarity.

Comparison of protein motifs counts across seven species
We used the Genome Comparison tool in StellaBase to compare the occurrence of each conserved Pfam motif in the sequenced genomes of seven different taxa. The taxa compared were Escherichia coli, Arabidopsis thaliana, Saccharomyces cerevisiae, Nematostella vectensis, Caenorhabditis elegans, Drosophila melanogaster, and Homo sapiens. Using the Pfam database (release 17), we identified Pfam domains in each genome by using a hidden Markov model search algorithm on the NCBI dataset for each of these species. For these searches, we queried all Pfam motifs for which we recovered at least one matching Nematostella protein. When we tallied the results across species, we included all proteins that matched a Pfam motif at an Expect value cut-off of 1e –6 (Suppl. Table 1; http://www.biolbull.org/supplemental/). To facilitate comparisons among taxa, we calculated the ratio of the number of each Pfam motif found in each taxon divided by the number found in human (Fig. 3).

View larger version (17K): [in this window] [in a new window]   Figure 3. Comparison of the abundance of Pfam motifs associated with responses to chemical stressors, wound, and pathogens in seven model systems. Bars depict the ratio of motifs recovered from each model system relative to the number recovered from human.

	Results

TOP Abstract Introduction Materials and Methods Results Chemical stress response Innate immunity and biological... Wound healing Comparison of protein counts... General Discussion and... Early evolution of animal... Evolution of stress tolerance Microevolution of stress... Nematostella vectensis as a... Literature Cited

	Chemical stress response

TOP Abstract Introduction Materials and Methods Results Chemical stress response Innate immunity and biological... Wound healing Comparison of protein counts... General Discussion and... Early evolution of animal... Evolution of stress tolerance Microevolution of stress... Nematostella vectensis as a... Literature Cited

 
The chemical stressors that impact biodiversity in aquatic ecosystems include a diverse array of naturally and anthropogenically produced compounds (Goldstone et al., 2006). Naturally produced chemical compounds, generated by organisms or by solar radiation, include reactive oxygen species, phytotoxins, and microbial metabolites. Anthropogenic chemical insults include toxic pollutants such as heavy metals, polychlorinated biphenyls (PCBs), pesticides, herbicides, pharmaceuticals, polycyclic aromatic hydrocarbons, flame retardants (polybrominated diphenyl ethers), and organochlorine compounds (DDT, endosulfan). Estuaries are particularly susceptible to anthropogenic chemical contaminants since they are often located in the vicinity of dense coastal population centers, and their freshwater inputs tend to concentrate runoff and point sources of pollution from upstream sources. At the same time, chemical stressors generated by organisms or by solar radiation can often achieve extremely high levels in shallow estuarine pools that are only infrequently flushed by the tides.

The combination of natural and anthropogenic chemical stressors can have considerable consequences for the biodiversity in estuaries (Pennings and Bertness, 2001; Weis et al., 2004, 2005). Two biotic metrics that reveal much about the health of estuarine ecosystems are (1) the community composition—in particular, the presence and abundance of key indicator species, and (2) the physiological status of resident organisms. Increasingly, molecular and genomic methods are being employed to assay both the community composition of estuaries, particularly the microbial communities, and the physiological status of resident organisms. For example, gene expression assays have recently been employed to characterize the chemical stress response of Fundulus heteroclitus, the mummichog, an estuarine fish (Meyer et al., 2005; Schulte, 2007). The mummichog is a highly mobile animal, living in the surface waters of estuarine pools and creeks. It can migrate extensively throughout an estuary to avoid locally unfavorable conditions. However, many animals commonly found in estuaries are sessile, benthic species with limited ability to escape locally stressful conditions. Due to their lack of mobility and their habit of living in or on sediments that tend to accumulate toxicants, the gene expression profiles of such species (e.g., Nematostella) are likely to more accurately reflect the recent history of chemical stressors at a particular geographic locale.

Nematostella is an ideal estuarine sentinel species for genomic studies on chemical stress. As a sessile benthic animal and a basal metazoan, it can provide results to complement the ongoing work on Fundulus, a motile vertebrate. It is an easily collected and common inhabitant of estuaries all along the east coast of North America from Nova Scotia to Georgia and along the Gulf coast. Introduced populations can also be found in England and on the Pacific coast of North America, from central California to British Columbia (Reitzel et al., 2008). It occupies both pristine and highly contaminated habitats. For example, Nematostella has been collected in the Hackensack Meadowlands, a restored marsh located along the Hackensack River in northeastern New Jersey. This well-studied site is known to be contaminated with heavy metals (Weis et al., 2004, 2005; Barrett and McBrien, 2007), chlorinated hydrocarbons (Bopp et al., 1998), and pesticides (Barrett and McBrien, 2007).

As Nematostella's apparently low dispersal ability probably renders it unable to escape toxic contamination in the sediment through migration, it is likely that some populations have evolved greater tolerance to some chemical contaminants. Population genetic surveys conducted throughout the animal's extensive range have identified sharp genetic breaks between neighboring estuaries, and even between adjacent pools within single estuaries (Darling et al., 2004; Reitzel et al., 2008). Although the swimming larva represents a potential dispersal vehicle, these genetic studies suggest that the effective dispersal potential for this animal may be quite low. We have observed that larvae in culture exhibit positive geotaxis, a behavioral tendency that would contribute to their apparently limited dispersal (Reitzel, Darling, Sullivan, and Finnerty, unpubl. data).

Goldstone et al. (2006) characterized the chemical "defensome" of the sea urchin. Their classification, which we follow here, included (1) stress-activated receptors, signal transduction proteins, and transcription factors; (2) efflux pumps; (3) oxidizing enzymes; (4) reducing and conjugating enzymes; (5) antioxidant proteins; (6) metal detoxicants; and (7) heat-shock proteins. The first stage in chemical defense is environmental sensing, which involves two principal groups of genes, the PAS (Per-ARNT-SIM) family of transcription factors that respond to oxygen and small molecules, and the nuclear receptor superfamily that bind a variety of ligands, some involved in stress. Both these classes of genes regulate transcription of a variety of effector genes including cytochrome p450s (CYPs), conjugating enzymes, and transporters. Some chemical compounds that breech the cell membrane are removed by efflux proteins such as the ATP Binding Cassette (ABC) and other ion transporters such as the organic anion and cation transporters (OAT and OCT, respectively). Other chemical stressors are biotransformed to inactivate and eliminate them. Biotransformation involves two phases: oxidation, then reduction or conjugation. Oxidation is typically carried out by two families of genes, the flavoprotein monooxygenases (FMO) and the CYPs. After oxidation, compounds are reduced or conjugated by a large suite of gene families including glutathione-S-transferases (GSTs), sulfotransferases (SULTs), and aldo-keto reductases (AKRs). The final group of genes in chemical defense summarized by Goldstone et al. (2006) is antioxidant defenses. Reactive oxygen species (ROS), including superoxide, H₂O₂, and hydroxyl radical, are a product of both metabolism and exogenous processes (pollutants, ultraviolet radiation, hypoxia). Regardless of their source, ROS affect signal transduction cascades, transcription factors, DNA integrity, and lipids, to name a few, resulting in effects on differentiation, apoptosis, stress-responsive genes, and aging (Adler et al., 1999). A variety of antioxidant proteins including superoxide dismutase (SOD), catalases (CATs), and peroxidases have been identified in diverse metazoans that regulate ROS in the cell.

Stress-activated receptors, signal transduction pathways, and transcription factors

Nematostella possesses a number of receptor and transduction genes involved in chemical defense in sea urchin and human (Table 1). For example, a recent phylogenetic study revealed that most of the gene families in the basic helix-loop-helix (bHLH) superfamily had evolved prior to the cnidarian-triploblast divergence. Nematostella was found to possess 68 genes representing 29–32 families, including two proteins that also possess a second domain, the Per-Arnt-Sim, or PAS, domain: (1) the hypoxia inducible factor 1 (HIF-1) and (2) the aryl hydrocarbon receptor nuclear translocator (ARNT; Simionato et al., 2007). In our search of the Nematostella genome, we identified three PAS proteins—ARNT, HIF-1, and the aryl-hydrocarbon receptor (AHR). AHR was not identified in the earlier study, and given its presence in Hydra, its apparent absence in Nematostella was ascribed to a lineage-specific loss in the sea anemone (Simionato et al., 2007). The putative AHR homolog we identified (SB_ 56923; gi|156394392), whose expression has been confirmed through an expressed sequence tag (EST) (JGI_CAGN20098.fwd), was not represented in the bHLH proteins culled from Nematostella in this previous study. However, a single "AHR related" gene, presumably the same protein, was identified in the publication of the Nematostella genome (Putnam et al., 2007). Both AHR and HIF-1 form complexes with ARNT and then regulate transcription of downstream targets through recognition of xenobiotic-responsive elements (XREs) or hypoxia-responsive elements (HREs), respectively. Further bioinformatic searches for these elements in upstream regions of potential effector genes would provide a fruitful avenue of research in constructing gene networks.

Our search recovered likely homologs to the bZIP transcription factors NF-E2 and Maf, which heterodimerize and activate gene expression in response to oxidative and xenobiotic stress. Both were recently identified in a number of cnidarians including Nematostella (Amoutzias et al., 2007). We also identified a KEAP1, an oxidative stress and electrophile sensor protein that binds NF-E2 in the cytoplasm, preventing NF-E2 translocation to the nucleus when the cell is not under stress (Jaiswal, 2004).

Predicted Nematostella proteins also exhibit strong matches to the metal-responsive transcription factor (MTF1) and the heat-shock factor HSF1. Because each of these transcription factors regulates gene expression by binding to well-characterized recognition motifs in target genes—metal-responsive elements for MTF: TGCRCNC (Saydam et al., 2001; Zhang et al., 2001); heat-shock elements for HSF: repeats of AGAAN and its complement recognized by trimers of HSF1 (Orosz et al., 1996)—it will be possible to screen the Nematostella genome for candidate target genes.

Nematostella, like other cnidarians surveyed (Grasso et al., 2001; Thornton, 2003; Bertrand et al., 2004), lacks many of the nuclear receptors traditionally studied in organismal stress (e.g., estrogen receptors, NR1 family). However, the receptors most relevant to environmental perturbation, hepatocyte nuclear factor 4 (Hnf4) and the retinoid X receptor (RXR), both members of the ancestral nuclear receptor subfamily 2, appear to be present in Nematostella (Reitzel and Tarrant, unpubl. data) and other cnidarians (e.g., Grasso et al., 2001). The functional role of these genes awaits further study, but some evidence suggests that cnidarians may be susceptible to a condition resembling endocrine disruption (Tarrant, 2005). Furthermore, understanding the interaction of nuclear receptors and other genes involved in endocrine-like function in Nematostella may reveal how the endocrine system evolved (Tarrant, 2007).

Efflux pumps
ATP-binding cassette (ABC) superfamily proteins are efflux transporters that pump compounds across cellular membranes against their concentration gradient (Dean and Annilo, 2005). ABC transporters are grouped into eight subfamilies (A–H) of which B, C, and G are known to expel toxic substances (Goldstone et al., 2006). Our search recovered 42 candidate ABC transporters. The ABC transporter complement from Nematostella is similar to that of human (n = 48), greater than Ciona (n = 31), and less than urchin (n = 65). Phylogenetic studies of these genes are ongoing and will allow a comparison of the ABC gene distribution in each subfamily.

Nematostella also has strong matches to other toxicant efflux proteins involved in removal of herbicides and toxic metals such as mercury and cadmium. We observed significant matches to the organic anion transporter polypeptide (OATP) family (n = 4, SB_12293, 7514, 12005, and 47307), the organic cation transporter (OCT) family (n = 1, SB_18740), and zinc transporters (n = 4, SB_34021, 4313, 28001, and 34688).

Oxidative biotransformation
Nematostella has putative homologs for all but one of the gene families involved in oxidative biotransformation (Table 2). The Nematostella genome encodes 48 predicted cytochrome p450s (Table 2, Supp. Table 1; http://www.biolbull.org/supplemental/), a gene family involved in oxidation of xenobiotic compounds including polycyclic aromatic hydrocarbons. Two other enzymes involved in oxidation of exogenous compounds—flavin-containing monooxygenase (FMO) and aldehyde dehydrogenase (ALDH)—are also present in Nematostella's genome. We did not identify a clear homolog for the prostaglandin-endoperoxide synthase (PTGS/COX) in Nematostella. This result is surprising given that PTGS genes have been identified in corals (Varvas et al., 1999; Järving et al., 2004). Although the PTGS protein of the coral Gersemia fruticosa is 50% identical to the human version, the best Nematostella match is only 23% identical to human PTGS and exhibits equivalent similarity to human peroxidases. Overall, for those gene families known to be involved in oxidative biotransformation, Nematostella appears to have fewer genes than human or urchin (Table 2).

Antioxidants and metal complexing
The Nematostella genome appears to encode a number of proteins involved in responding to ROS, including superoxide dismutase (SOD), various peroxidases, and the glutathione pathway (Table 3). Previous research with cnidarians, primarily anthozoans, has characterized the expression of SODs and catalase (CAT) in response to environmental stress (Richier et al., 2003, 2005; Yakovleva et al., 2004; Dash et al., 2006, 2007; Merle et al., 2007).

Heat-shock proteins
HSF, identified above, is a transcription factor that regulates expression of the heat-shock proteins (HSPs), which are involved in responses to temperature fluctuations and a wide variety of other stressors. HSPs are categorized into families on the basis of the molecular weight of the protein. Nematostella has multiple representatives in the hsp20 (7), hsp70 (6), and hsp90 (3) families. HSPs from various cnidarians have been studied in a variety of stress responses including temperature (Hayes and King, 1995; Kingsley et al., 2003; Schroth et al., 2005), coral bleaching (Downs et al., 2002), and aggression (Rossi and Snyder, 2001).

	Innate immunity and biological stress

TOP Abstract Introduction Materials and Methods Results Chemical stress response Innate immunity and biological... Wound healing Comparison of protein counts... General Discussion and... Early evolution of animal... Evolution of stress tolerance Microevolution of stress... Nematostella vectensis as a... Literature Cited

 
Broadly, the immune system of vertebrates with jaws can be divided into adaptive and innate immunity. Through the adaptive immune system, jawed vertebrates are capable of mounting a "learned response" against infectious agents via the generation of pathogen-specific antibodies. This complex, acquired response relies on B-cells (humoral system) and T-cells (cell-mediated systems), other lymphocytes, and an ability to model an extensive number of unique antibodies via light-chain variability of immunoglobulin proteins.

However, even in those organisms that possess an adaptive immune response, an important component of organismal health is the innate immune system. The innate immune system comprises a number of disparate host-defense mechanisms that interfere with the efficacy of infection by pathogens. The first line of defense, and an integral part of the innate immune system, is a physical barrier that impedes the entrance of infectious agents. Epithelium, mucous membranes, waxy cuticles, shells, and chitinous skeletons perform such a function in phylogenetically diverse organisms. Once an infectious agent gains access, vertebrate hosts may deploy numerous components of the innate immune system, including an inflammatory response, activation of the complement system, deployment of nonspecific macrophages and other leukocytes, production of free radicals and peroxide, and the production of antimicrobial peptides. Although such antimicrobial peptides lack the degree of specificity exhibited by immunoglobulins, they bind cell structural elements or macromolecules that are specific to major clades of infectious organisms.

A comparison of nonvertebrate deuterostomes (e.g., sea urchin) and basal vertebrates (lamprey, hagfish) reveals that the adaptive immune system is a vertebrate invention (Sodergren et al., 2006). In contrast, the evolutionary origin of the innate immune system remains obscure. Its presence in both deuterostomes (e.g., human and sea urchin) and protostomes (e.g., Drosophila sp. and Limulus polyphemus) indicates that this system originated prior to the radiation of triploblasts. We found that 63 of the 128 Pfam motifs specifically implicated in innate immune function appear to have homologs in the genome of Nematostella. These 63 Pfam motifs identify 1039 unique predicted proteins housed in StellaBase that may be involved in innate immune function (Table 4).

Five Toll-like domains (Pfam motif: TIR) have been identified in the Nematostella genome (Miller et al., 2007; Sullivan et al., 2007). Four of these domains are present in transmembrane proteins, suggesting a role in extracellular signaling. Three of these four transmembrane proteins are associated with two or more immunoglobulin motifs, suggesting a structure reminiscent of vertebrate interleukin receptors. The remaining transmembrane-linked TIR motif is associated with numerous leucine-rich motifs, which, when combined with phylogenetic analyses, suggests a structure conserved with true Toll- and Toll-like receptors in fly and vertebrates. The final TIR domain is not associated with a transmembrane motif, but rather with a DEATH domain, indicating that this protein may act as the first step in the intracellular signaling cascade (e.g., it may be a MyD88 homolog). Like Nematostella, Hydra magnipapillata was also found to possess predicted proteins combining Toll-like receptor domains in association with transmembrane motifs (Zheng et al., 2005).

Other key elements of the innate immune system have also been identified in Nematostella, including several members of the Toll and interleukin signaling cascades such as IRAK, TRAF, ECSIT, IKK, MKK, JNK, IKB, p38, and the transcription factors NF-B and AP-1 (Table 4; Miller et al., 2007; Sullivan et al., 2007). Additionally, the transcription Interferon-regulatory factor, also inducible by the Toll-like signaling cascade, may have a homolog in this basal animal (SB 44157, Genbank Accession gi|156396757|ref|XP_001637559.1; BLASTp E-value versus human interferon regulatory factor 2 [gi|153082752|ref|NP_002 190.2] = 5e^–40). Interestingly, although Nematostella was found to possess a complex repertoire of innate immune genes, a relatively reduced complement of innate immune genes was identified in the freshwater hydrozoan Hydra magnipapillata (Miller et al., 2007).

In cnidarians, it has been suggested that Toll-like receptors might be performing an ancestral function unrelated to innate immunity (Kanzok et al., 2004; Zheng et al., 2005). Kanzok et al. (2004) argued that animals lacking a coelom, including all diploblastic animals such as cnidarians, would not benefit from production of anti-microbial peptides—a hallmark of Toll-receptor-mediated innate immunity—because diploblasts lack an internal body compartment (namely, the coelom) into which these proteins could be secreted. For a time, this hypothesis was bolstered by an inability to recover an NF-B or other Rel-homology domain containing transcription factor from any basal animal.

Now that many of the components of innate immunity have been identified in cnidarians, we can begin to assess whether they are performing a role homologous to the role in triploblastic animals. If functional Toll- and Interleukin-signaling cascades are indeed present in Nematostella, downstream targets of the cascade may be conserved between cnidarians and bilaterians (e.g., canonical targets of NF-B). Homologs of known targets of NF-B in vertebrates have been identified in Nematostella through sequence homology (Table 4; Sullivan et al., 2007). Downstream components present in Nematostella include numerous apoptosis regulators (i.e., caspases [Eckhart et al., 2007], p53, and p63), domains associated with antimicrobial peptides, antioxidant enzymes, and a putative homolog of a complement component protein (Table 4). We are currently searching the 5' regions of these putative targets to identify loci that are enriched for B sites, and therefore potentially regulated by NF-B, perhaps via the Toll-signaling cascade.

C3-related complement system proteins
Among cnidarians, a homolog for the complement system protein C3 was first identified in the gorgonian Swiftia exserta (Dishaw et al., 2005). In Nematostella, a systematic search of the genome for components of the complement pathway identified a C3 protein in addition to integrins, suggesting that a simplified complement system may be present (Nonaka and Kimura, 2006). Such a system, however, would have to be extremely simplified relative to the complement system of vertebrates, whose complexity required a number of gene and genome duplications over the course of vertebrate evolution. Interestingly, Hydra magnipapillata appears to lack a C3 protein (Miller et al., 2007), although evolutionarily related A2M-domain–containing proteins are expressed in the endoderm of Hydra with an expression pattern similar to that of a C3 protein of the coral Acropora millepora.

	Wound healing

TOP Abstract Introduction Materials and Methods Results Chemical stress response Innate immunity and biological... Wound healing Comparison of protein counts... General Discussion and... Early evolution of animal... Evolution of stress tolerance Microevolution of stress... Nematostella vectensis as a... Literature Cited

 
Along with chemical insults and infection by pathogens, physical wounding is a major source of organismal stress. Indeed, different categories of stressors interact to threaten the well-being of organisms: for example, pathogens may be introduced upon wounding, and the inability to recover from wounds may be the proximate cause of death in animals that are already stressed by chemical insults or pathogens. All animal phyla exhibit some ability to recover from wounds. In those animals where it has been studied in detail, wound healing is a complex and dynamic process that requires the interaction of many factors including cytokines, intracellular signaling pathways, transcription factors, and extracellular matrix (ECM) proteins. The extent to which the wound-healing process of "basal animals" might be homologous to the wound healing of triploblastic animals is not yet known.

We expect that some genes in the wound-healing repertoire of cnidarians may be unique to this phylum and thus identifiable only through empirical studies on cnidarians. However, as the "epithelial level of organization" exhibited by Nematostella is an extremely ancient invention that dates to the common ancestor of the Eumetazoa (if not earlier), it is likely that epithelial repair mechanisms are similarly ancient, and core elements of these pathways may be deeply conserved across animals. If so, searching for homologs of known wound-healing genes should be useful to identify candidate wound-healing genes in Nematostella.

Wound healing in vertebrates
Traditionally in vertebrate models, the process of wound healing has been broken down into three distinct but overlapping phases (Singer and Clark, 1999): (1) inflammation, (2) proliferation, and (3) matrix rebuilding and remodeling. During the inflammatory stage, the coagulation pathway is activated and platelets appear at the wound site (Kiritsy et al., 1993). Under the influence of cytokines such as platelet-derived growth factor (PDGF) and transforming growth factor beta (TGF-beta), these platelets form a plug at the site of injury (Franchini and Ottaviani, 2000). This phase is characterized by the activation of neutrophils, macrophages, and mast cells, which help to remove debris and bacteria from the wound site. Simultaneously, cytokines stimulate circulating cells to express integrins, and the integrins enable these cells to interact with the ECM. The proliferative phase is marked by the presence of fibroblasts, which produce a number of ECM components including collagen, proteoglycans, and fibronectin. Keratinocytes proliferate and migrate out to the edge of the wound site, creating a natural defense barrier. Additional cytokines are produced from fibroblasts, keratinocytes, and macrophages, and these intercellular signaling factors help to coordinate this complicated series of events. Lastly, the ECM is remodeled by the modification of collagen fibers, which results in the contraction of the wound. Collagen reorganization appears to be a very ancient component of metazoan wound healing shared among protostomes and deuterostomes (Tettamanti et al., 2005).

Wound healing in cnidarians
In cnidarians, wound healing is often followed by regeneration, and these two processes tend to be conflated in the cnidarian literature (Henry and Hart, 2005). However, two lines of evidence reveal that wound healing and regeneration are evolutionarily and mechanistically separable processes. First, there are many animal species whose regenerative abilities are limited but whose wound-healing abilities are quite well developed—for example, if the tail of a mouse is surgically removed, the wound will heal, but the tail will not regenerate. Second, even in animals with extensive regenerative ability, such as cnidarians, wounds that do not result in the loss of a body region or structure do not necessarily trigger regeneration. In Nematostella, for example, complete bisection through the body column always results in the regeneration of a missing oral crown or a missing physa (Reitzel et al., 2007). However, incomplete bisection of the body column only rarely triggers the development of an ectopic oral crown or physa. Far more often, it triggers wound healing without regeneration. Here, we focus on wound healing and not regeneration, because the former is clearly a response to acute stress.

In cnidarians, the cellular events underlying wound healing have not been characterized at the same level of detail as in vertebrates, but at least one potential parallel has been noted. In vertebrates, the inflammatory stage is characterized by the influx of specialized cells to the wound site, including phagocytes (Gillitzer and Goebeler, 2001). Similarly, in some cnidarians, including gorgonians, scleractinian corals, and sea anemones, amoebocytes are recruited from the mesoglea to the wound site, and these cells engage in phagocytosis (Hutton and Smith, 1996; Meszaros and Bigger, 1999; Olano and Bigger, 2000; Henry and Hart, 2005; Mydlarz et al., 2006). These amoebocytes that migrate to the wound site also produce ROS (Hutton and Smith, 1996), as occurs in the circulating amoebocytes of other invertebrates (Nakamura et al., 1985; Coteur et al., 2002), suggesting that these cells are playing a role in an inflammatory response.

Identifying homologs of known wound-healing genes in Nematostella
In a first step toward determining whether the wound healing process in cnidarians is homologous to that of vertebrates, we can determine whether the same genes are involved in coordinating and regulating the process. Although a complex network of genes underlies the process of wound healing in vertebrates, none of the genes so far identified appears to be uniquely involved in wound healing. Thus, it is not possible to strictly categorize any gene as a "wound healing gene." However, we can determine which genes from the entire complement of genes active during wound healing in vertebrates are present in Nematostella, and subsequent investigations can determine whether these putative homologs are deployed in a homologous fashion during wound healing in the anemone.

The genes involved in vertebrate wound healing can be broken down into the following broad functional categories: (1) intercellular signaling molecules; (2) cell surface receptors and intracellular signaling molecules, (3) transcription factors, and (4) ECM components. On the basis of a survey of the literature, we identified 31 Pfam domains in proteins associated with wound healing, including signaling molecules, cell surface receptors, intracellular signaling factors, transcription factors, and ECM components. Every one of these Pfam domains scored a significant match to at least one predicted protein in StellaBase (Table 5, Suppl. Table 1; http://www.biolbull.org/supplemental/). Overall, the 31 domains scored significant matches to 485 different anemone proteins.

There are two types of TGF-beta receptors: type I and type II. In the absence of a ligand, these receptors exist as homodimers on the surface of the cell, but they typically bind to ligands as heterotetramers (Derynck and Zhang, 2003). Activation of this signaling cascade by the binding of a TGF-beta ligand to a receptor complex can result in the nuclear translocation of SMAD proteins, which can impact cell proliferation and migration. Also, the TGF-beta signaling pathway can be Smad-independent and result in the activation of MAPK pathways, some of which have a role in the Smad activation (Derynck and Zhang, 2003). In Nematostella, we identified a single TGF-beta receptor that best matches human TGB-beta receptor I.

Integrins
Integrins are a superfamily of integral membrane proteins that function in cell migration, cell-cell adhesion, cell-ECM adhesion, and signal transduction (Takada et al., 2007). In the cell, integrins are found as heterodimers, consisting of different alpha and beta subunits. The extracellular domains of integrins interact with a variety of ligands depending on the combination of receptor units. The cytoplasmic domain can interact with signaling proteins, focal adhesions, and cytoskeletal proteins, and it is through this interaction that integrin-initiated intracellular signaling occurs (Takada et al., 2007). It has been proposed that integrins can play a role in the nuclear translocation of the transcription factor NF-B, as well as in activation of the Jun-N-terminal Protein Kinase (JNK) (Nikolopoulos et al., 2005). The JNK pathway is activated during the proliferative phase of wound healing in vertebrates and during epidermal wound repair in Drosophila (Rämet et al., 2002; Galko and Krasnow, 2004).

Both alpha and beta subunits of integrins were previously cloned in the jellyfish Podocoryne carnea (Reber-Muller et al., 2001). The beta subunit has also been cloned from sponge and coral (Brower et al., 1997). A search of the Nematostella genome using the "integrin, beta chain" and "integrin beta tail domain" identifies two integrin beta loci. A tBLASTn search using human integrin alpha proteins identifies two integrin alpha loci in Nematostella.

The transcription factor, grainyhead
Grainyhead plays a role in maintaining the tension of the cuticle in Drosophila, and similarly, it is implicated in the epithelial integrity of mammals, where a mutation in this gene causes defects in wound healing and epithelial barrier formation (Uv et al., 1994; Kudryavtseva et al., 2003; Mace et al., 2005; Ting et al., 2005). Grainyhead can be recognized by the possession of a highly conserved CP2 domain, and this transcription factor has been identified in a wide range of triploblastic animals (Moussian and Uv, 2005). A search of the Nematostella domain using the "CP2" domain identified a single significant match. Reciprocal tBLASTn searches between the human genome and the sea anemone genome identifies this predicted protein as the Nematostella ortholog of human grainyhead.

Smads
Smads are transcription factors that are downstream of TGF-beta signaling. They are implicated in cell proliferation and migration. In general, all smads share sequence similarity in the amino-terminal and carboxy-terminal domains, MH1 and MH2 respectively (Xu, 2006). Searching the Nematostella genome for the MH1 and MH2 domains identifies four significant matches to MH1 and one to MH2. According to a BLASTp search, four of the five MH1-containing proteins and the single MH2-containing protein are matches to human Smads. The remaining MH1-containing protein matches a human BMP.

Extracellular matrix proteins important in wound healing
During the proliferative stage of wound healing, an epithelium begins to form over the wound site, and the ECM is deposited. The ECM is a complex netting of glycoproteins, collagens, and proteoglycans. It is important for cellular remodeling during wound healing, and it is implicated in cell attachment and growth, cellular differentiation, and structural support of tissues (Huxley-Jones et al., 2007). Many ECM proteins exhibit strong evolutionary conservation across the animal kingdom. This high degree of conservation is thought to show the importance of collagen-based extracellular matrices in animals (Tettamanti et al., 2005; Aouacheria et al., 2006; Huxley-Jones et al., 2007).

Collagens
The "collagen triple helix repeat" motif was a significant match for 19 predicted proteins in the Nematostella genome. When a BLAST search was performed against the human genome, these candidate Nematostella collagens exhibited best matches to multiple human collagen genes, including alpha 1 collagens (types V, XIII, XV, XXV), an alpha 2 collagen (type V), and an alpha 5 collagen (type IV). The BLAST searches also identified matches to a human contactin 4 (also known as BIG-2) and contactin 1 associated protein. The contactins are neuronal cell adhesion molecules that belong to the immunoglobulin superfamily (Yoshihara et al., 1995).

Fibronectins
Searches using the "fibronectin type II domain" identified nine matches in the Nematostella genome, though none of these identified a fibronectin as a best match in the human genome. For example, three Nematostella hits to "fibronectin type II" were best matches to human matrix metalloproteinases (MMP-2 and MMP-9), enzymes involved in the breakdown of extracellular matrix proteins. MMP-2 and MMP-9 are unique among metalloproteinases in harboring three repeats homologous to fibronectin II (Steffensen et al., 2001). The six additional Nematostella loci that exhibit significant similarity to the "fibronectin type II" domain are best matches to human notch 2 preprotein, relaxin, vitrin, neuronal pentaxin, neuropilin, and semaphorin.

Searches using the "fibronectin type III domain" identified 101 significant matches in the Nematostella genome. Two of the genes identified are best matches to the same human fibronectin type III domain containing 3B gene, but none of the best matches are to human fibronectins. A large fraction of the Nematostella genes that were identified in the search for fibronectin type III domains are best matches to human protein tyrosine phosphatase receptors.

In an attempt to identify a bona fide fibronectin protein in Nematostella, we used tBLASTn to query the genome with the human fibronectin 1. When the best hit in the Nematostella genome was used to search the human genome, the best human match proved to be a sidekick protein (SDK2), not fibronectin 1 (Table 6).

Small leucine-rich-repeat protein family
The small leucine rich repeat proteins (SLRPs) are a family of proteoglycans that are known to play important roles in organizing structural elements in the ECM, thereby impacting cell adhesion, wound healing, and the ability to withstand mechanical stress (Matsushima et al., 2000). The proteins are grouped into four classes based on the number and type of repeats and super-repeats they possess. Class I (e.g., decorin and biglycan) and class II (e.g., fibromodulin, lumican, keratocan, and PRELP) have 12 LRRs. Class III SLRPs (e.g., opticin and osteoglycin) have 7 LRRs. Class IV consists of chondroadherin. In vertebrates, specific LRR proteins are involved in distinct extracellular matrix types. For example, biglycan and decorin (class I) act in synergy to promote skin and bone integrity in mice (Corsi et al., 2002), while the class II protein lumican plays a role in wound healing in the cornea (Kao et al., 2006).

Leucine-rich repeats are found in a wide range of proteins, including bacterial virulence factors, cell adhesion molecules, enzymes, hormone receptors, and tyrosine kinase receptors, in addition to extracellular matrix-binding glycoproteins. A search of the Nematostella genome using the "leucine-rich-repeat" domain identified 54 matching genes, but none of these genes appears to be orthologous to any of the vertebrate SLRPs. When the human genome was queried using these 54 anemone sequences, the top BLAST hits included a number of predicted "leucine rich repeat containing" proteins but no members of the SLRP family. Using tBLASTn, we queried the Nematostella genome with human SLRP proteins (decorin, biglycan, fibromodulin, lumican, keratocan, PRELP, opticin, osteoglycin, and chondroadherin). We then queried the human genome with the top BLAST hits from Nematostella, and not one of these reciprocal BLAST searches identified the human SLRP protein originally used to query Nematostella as the top hit. In all cases, other human LRR-containing proteins were identified (Table 6). These same SLRPs also seem to be lacking in protostome animals (Table 7), suggesting that they are deuterostome or vertebrate inventions (Huxley-Jones et al., 2007). The complexity of this gene family in vertebrates may be correlated with the diversity of vertebrate ECM types (cartilage, bone, tendon, cornea, tooth, etc.)

To determine whether the ECM proteins apparently missing from Nematostella might be exclusive to deuterostomes or vertebrates, we used tBLASTn to search for them in protostome genomes (Table 7). We then used tBLASTn to compare the top protostome hits back against the human genome to determine whether the best match would correspond to the original human query sequence. These reciprocal tBLASTn searches produced clear evidence for protostome syndecan and tenascin, suggesting that these ECM proteins may have evolved on the triploblastic stem lineage. However, we failed to identify asporin, fibronectin, matrilin, osteomodulin, podocan, reelin, or vitronectin in any protostome animals, suggesting that these ECM proteins may have originated within the Deuterostomia.

	Comparison of protein counts among seven diverse species

TOP Abstract Introduction Materials and Methods Results Chemical stress response Innate immunity and biological... Wound healing Comparison of protein counts... General Discussion and... Early evolution of animal... Evolution of stress tolerance Microevolution of stress... Nematostella vectensis as a... Literature Cited

 
We compared the number of predicted proteins bearing Pfam motifs associated with chemical stress, innate immunity, or wound healing in seven diverse taxa (Fig. 3, Suppl. Table 1; http://www.biolbull.org/supplemental/). As expected, E. coli encodes the fewest proteins in all three categories. For motifs associated with chemical stress response, Saccharomyces cerevisiae, Nematostella, and Caenorhabditis elegans encode fewer proteins than human, while Drosophila melanogaster and Arabidopsis thaliana encode more. A. thaliana outnumbered human for proteins containing motifs associated with chemical stress by more than 3 to 1 due to extremely abundant representation of some Pfam motifs (e.g., p450s), despite lacking representatives for four of the assessed motifs.

For motifs associated with wound repair, the Nematostella genome encodes more proteins than any other taxon except human (50% as many as human), including representatives of every motif queried. However, although Nematostella harbors a large number of proteins containing motifs associated with wound healing, it actually lacks homologs for many of the vertebrate proteins specifically implicated in wound healing, including fibronectin, many ECM proteins, and SLRPs. However, all other taxa lack representatives of two or more of the Pfam motifs associated with wound-healing proteins. E. coli and S. cerevisiae contained the fewest matches for the wound-repair motifs, which is unsurprising given that they are single-celled organisms.

For pathogen response, the overall number of proteins is larger in A. thaliana and C. elegans than in humans. With respect to pathogen defense, Nematostella exhibits fewer hits than other metazoans, but the anemone is the only taxon to have at least one protein representing each of the Pfam motifs associated with pathogen defense. For example, human contained no matches for the PF05183 (RNA-dependent RNA polymerase), but Nematostella had four hits.

	General Discussion and Conclusions

TOP Abstract Introduction Materials and Methods Results Chemical stress response Innate immunity and biological... Wound healing Comparison of protein counts... General Discussion and... Early evolution of animal... Evolution of stress tolerance Microevolution of stress... Nematostella vectensis as a... Literature Cited

 
Characterizing the molecular stress-response repertoire of Nematostella can be informative at four levels: (1) comparisons to triploblastic animals and non-animal models will help us to reconstruct the early evolution of stress-response pathways; (2) comparisons to species with narrower environmental tolerances, especially closely related marine anemones, will help us to understand the genomic basis for Nematostella's hardiness and ability to survive in estuarine habitats; (3) comparisons among genetically and phenotypically distinct populations of the starlet sea anemone can reveal the microevolution of stress-response pathways; and (4) identifying molecular markers of stress will augment our ability to utilize Nematostella and other Cnidaria as environmental sentinels.

	Early evolution of animal stress-response pathways

TOP Abstract Introduction Materials and Methods Results Chemical stress response Innate immunity and biological... Wound healing Comparison of protein counts... General Discussion and... Early evolution of animal... Evolution of stress toler