Cnidaria have traditionally been viewed as the most basal animals with complex, organ-like multicellular structures dedicated to sensory perception. However, sponges also have a surprising range of the genes required for sensory and neural functions in Bilateria. Here, we: (1) discuss “sense organ” regulatory genes, including; sine oculis, Brain 3, and eyes absent, that are expressed in cnidarian sense organs; (2) assess the sensory features of the planula, polyp, and medusa life-history stages of Cnidaria; and (3) discuss physiological and molecular data that suggest sensory and “neural” processes in sponges. We then develop arguments explaining the shared aspects of developmental regulation across sense organs and between sense organs and other structures. We focus on explanations involving divergent evolution from a common ancestral condition. In Bilateria, distinct sense-organ types share components of developmental-gene regulation. These regulators are also present in basal metazoans, suggesting evolution of multiple bilaterian organs from fewer antecedent sensory structures in a metazoan ancestor. More broadly, we hypothesize that developmental genetic similarities between sense organs and appendages may reflect descent from closely associated structures, or a composite organ, in the common ancestor of Cnidaria and Bilateria, and we argue that such similarities between bilaterian sense organs and kidneys may derive from a multifunctional aggregations of choanocyte-like cells in a metazoan ancestor. We hope these speculative arguments presented here will stimulate further discussion of these and related questions.
The word “animal” implies muscle-driven motility coordinated by neural integration of sensory stimuli produced in specialized multicellular sensory structures. Consequently, a number of sets of questions spring to mind when considering evolution of metazoan sensation: where on the tree of animal life did the first sense organs evolve?
Do sense organs share a common evolutionary origin with other structures or organs? What type of sense organ evolved first and how are different classes of sense organs related to one another? Are bilaterian sense organs related to the sensory features in the more basal radiate taxa? Does the placement of Scyphozoa and Hydrozoa together in a medusozoan group support a derived condition for cnidarian sense organs? How does evidence suggesting common origin of bilaterian and cnidarian sense organs relate to the presence of bilaterian-like dorso-ventral axial organization in Cnidaria? In addition, for clarity in addressing these questions a definition of “sense organ” specific to the purposes of this discussion must be developed.
Not all of the preceding questions can be definitively answered at this time. However, developmental gene-expression studies, genome sequencing and expressed-sequence-tag studies, are shedding light on some of these issues. Interestingly, the initial answers to these questions are not always consistent with a priori expectations. For example, one might expect that evolution of genes thought to be explicitly involved in development of sense organs would coincide with the evolution of the radiates, as Cnidaria and Ctenophora are the most basally branching lineages with specialized sense “organs.” This expectation is not met; regulatory genes involved in sense-organ development in “higher” Metazoa are present more basally in sponges, as are genes considered essential for synaptic function. Although not explicitly muscular or neural, sponges exhibit coordinated contraction as well as coordinated cessation of pumping. Thus, a view of sponges as more active is replacing an older perception that held sponges to be virtually “inanimate.”
In this work, we touch on the features that distinguish sense organs. We then consider the questions listed earlier in the context of the basal branches of the metazoan tree, focusing on the cnidarian and sponge branches. In cnidarians, we address the relationship between cnidarian and bilaterian sensory structures, as well as shared aspects of sense organs and appendages. In the sponges, we discuss the possible evolutionary antecedents of sense organs. Lastly, we consider how different reconstructions of the metazoan tree influence these interpretations. The speculative hypotheses presented here emphasize differential persistence and modification of an ancestral condition, rather than invoking wholesale “cooptation” of genes, as an explanation for conflicting patterns of gene expression and morphology observed across the metazoan tree. In each instance considered, many other hypotheses could be advanced and we encourage others to generate specific competing hypotheses.
What do sense organs have in common?
Cells generally have an ability to assay aspects of their surroundings. However, multicellular organisms have the challenge of differential exposure of cells to external and internal environments and the opportunity to have cells with specialized sensory functions. Sensory structures that form part of the epidermis are found in all animal phyla from cnidarians onward. In cnidarians and some basal bilaterian groups (e.g., acoels, platyhelminths, and nemertines), sensory structures consist of “naked” sensory neurons whose dendrite is formed by a modified cilium (Chia and Koss 1979; Wright 1992). Cell bodies of sensory neurons are often sunken beneath the level of the epidermis or can even reside within the central nervous system. From these “naked” sensory neurons, one distinguishes sensilla and sensory organs. Sensilla constitute individual sensory neurons, or small arrays of sensory neurons, integrated with specialized nonneuronal cells that typically function in particular sensory modalities—light reception, mechanoreception (auditory/inertial/touch/stretch/vibration), and chemoreception (taste/smell). Finally, sense organs are large assemblies of sensory neurons and nonneuronal cells that form macroscopic structures. In some cases, such as the compound eyes and auditory organs of arthropods, arrays of contiguous sensilla are integrated into large sensory organs. In this view, “sense organs” already exist in cnidarians, in the form of eyes and statocysts despite the lack of mesoderm often invoked as a required condition for organ systems. Highly developed sensory organs are more widespread and exist for all sensory modalities in bilaterians. In many instances, sensory organs and sensilla coexist with naked sensory neurons in the same animal.
The sensory neurons of a sensory organ or sensillum usually bear cilia and/or microvillar structures on their apical surfaces and these surfaces are often modified into complex membrane features (Fain 2003). Photoreception and chemoreception involve seven-pass transmembrane G protein-coupled receptors (GPCRs), and membrane-bound ion channels transduce mechanical stimuli (other sensory-cell types can detect ionic concentrations or electrical fields). Such sensory neurons then communicate by electrical potential either through axons that are components of the sensory cells themselves (the typical invertebrate condition), or via synapses on the cell bodies to adjacent neural cells (a frequent vertebrate condition, as in the hair cells of the inner ear).
It is important to note that not all GPCRs are involved in sense organs or sensory perception. Multiple independent classes of these receptors are involved in synaptic, hormonal, and developmental signaling internal to the organism (e.g., http://www.sdbonline.org/fly/aignfam/gpcr.htm), and the proliferation of multiple classes of GPCRs appears to be a critical distinctive feature of animals relative to other eukaryotes (http://drnelson.utmem.edu/MHEL.7TM.html). Thus, sense organs are distinct in the particular application of GPCRs to external chemical and photoreception.
There are shared aspects of developmental gene expression in sense organs across the bilaterian tree and across classes of sensory structures in a single animal. Bilaterian data are discussed first; we then explore how these bilaterian-based inferences play out when compared to the limited cnidarians and sponge information. As noted earlier, our primary objective is to treat the range of multicellular sensory structures rather than naked sensory cells or simple sensilli.
Common aspects of sense-organ developmental gene regulation
Different proneural genes are required for the production of different types of sensilla and sense organs in Drosophila. Isolated mechanosensory sensilla require the expression of achaete–scute complex genes. Whereas, atonal, a distinct basic helix-loop-helix gene, is required for the development of sense organs that consist of closely stacked sensory units, such as chordotonal organs found in stretch receptors, auditory organs, or the ommatidia of the insect compound eye (Jarman and Ahmed 1998). Atonal, or its multiple vertebrate homologues, are expressed in, and function in, the development of all or virtually all sense organs in Drosophila and vertebrates. This includes eyes and chordotonal organs in Drosophila and the placodally derived eye, ear, and nose of vertebrates.
In addition to atonal, a number of other genes, initially identified by the loss of eyes in Drosophila mutants, function in the regulatory cascades governing the development of multiple classes of sense organ. These include eyes absent and dachshund, as well as members of the Six gene-family—a distinctive group of homeodomain-containing genes that includes sine oculis and optix. In addition, genes such as Brain3 are required for specifying aspects of sensory-cell and sensory-nerve-cell differentiation in multiple classes of sense organs (auditory, olfactory, and visual). Mouse Brain3 mutants are deaf and blind, and lack balance due to the absence of hair cells in the semicircular canals (Pan et al. 2005). Thus, a substantial list including upstream regulatory genes and downstream genes with sensory-cell-type specificity is a common feature of a wide range of sensory organs [Schlosser (2006) provides a summary of shared regulatory-gene control across vertebrate sensory structures].
Over-expression studies illuminate some of the commonality and combinatorial function of these genes. Famously, expression of the vertebrate homologue of eyeless (PAX6) successfully rescues eyes in eyeless mutants of Drosophila (e.g. Gehring and Ikeo 1999). However, over-expression experiments (that deliver the gene product throughout the organism) convert chordotonal organs to eyes (Halder et al. 1995). This conversion illustrates the shared developmental genetic regulation present in multiple classes of sense organ, as well as the role that Pax genes, such as eyeless, play in determining a subset of sense organs that includes eyes (Schlosser 2006).
Sharing of developmental regulatory genes across systems
The preceding section presented a coherent picture of the regulation of sense-organ development across divergent Bilateria, but, alas, additional complexities intrude on this seeming paradise of rational hierarchical organization. Developmental genes often serve multiple functions in development, thus hypotheses regarding common ancestry of function with distantly related organisms are not necessarily straightforward. They require attention to other lines of evidence that may suggest which facets of expression are likely to reflect shared ancestry. Many of the genes involved in the development of sensory organs are also involved in the development of structures that are not, or might not, typically be considered sense organs.
Overlap of expression of sense-organ regulatory genes in muscles and nerves is perhaps to be expected, given the functional and synaptic connections between these systems. In addition, gene duplication appears to have generated multiple players with separate functions in sensory cells, nerves, and muscles. There are many examples of this in groups of genes that evolved basal to the radiation of bilaterians; the three classes of Six genes (sine oculis, optix, and myotonix in Drosophila) are primarily involved in the development of sense organs in the first two instances and muscles in the later. In the NK2 homeodomain genes, tinman and bagpipe are involved in the differentiation of cardiac and smooth muscles, but vnd functions in the development of the medial nervous system (discussed in Jacobs et al. 1998; Holland et al. 2003). Invertebrate sensory cells also have neuronal processes (Fig. 1); vertebrate sensory cell and neuron are separate cells. Vertebrates also have multiple copies of many genes including the Brain3 gene. Separate copies of Brain3 in vertebrates appear to have distinct functions, seemingly coincident with the division of neural and sensory cell types in the vertebrate nervous system, relative to the single neurosensory cell that performs this combined function in most invertebrate sensory neurons. The above examples of gene family function and gene duplication suggest some of the typical and more prosaic ways in which genes involved in sense-organ developmental regulation appear to “coopt” new functions in their evolutionary history.
Instances of sense-organ developmental circuitry that go beyond these typical cooptive categories are potentially intriguing and challenging for evolutionary interpretation. They may provide evidence for relationships between classes of organs not usually considered in common. For example, a number of genes generally thought to be sense-organ-specific, such as the Six genes, as well as eyes absent, and dachshund homologues (Schlosser 2006) are also expressed in the pituitary, as is the POU gene PIT1, the most closely related POU gene to Brain 3. This is surprising on its face, but proves consistent with the evolution of the adenohypophyseal component of the pituitary, from an external chemosensory to an internal endocrine organ in the chordate lineage (Gorbman 1995; Jacobs and Gates 2003). Thus, the presence of the gene Pit1 in more basal taxa, including cnidarians and sponges (Jacobs and Gates 2001), is consistent with an evolutionarily antecedent to the vertebrate pituitary, perhaps involved in external reproductive communication. Other structures derived from cephalic placodes in vertebrates share aspects of regulation with formal sense organs (Schlosser 2006) and likely have a common evolutionary origin with sensory structures.
A still less-expected set of commonalities is evident between vertebrate sense organs, such as the ear and kidneys. Both sense organs and kidneys express the same suite of regulators in development, and there are a number of diseases that effect the ear and kidney, in particular leading to the biomedical term otic–renal complex (see Izzedine et al. 2004 for review). Common attributes of distinctly different organs are often dismissed as cooptation, but this is too easy. How such cooptation occurs is critical to understanding evolution. We argue that, whether considered cooptive or not, they likely reflect some aspects of shared ancestry, and that such common origin may be supported by examination of the basal lineages. In this unexpected case—the commonality of kidneys and sense organs—we argue below that this could reflect cellular organization in sponges, in which groupings of choanocytes may serve multiple functions and subsequently evolved into the sense organs and kidneys in bilaterians. We advance this particular argument based on the initially surprising commonality of sensory regulation and disease in distinctly different organs. However, this does not limit the possibility that many other systems in higher Metazoa may also have common origins; given the small set of differentiated cell and tissue types found in sponges, this may necessarily be the case. Given similar logic, the expression of atonal homologues associated with the neuroendocrine cells of the gut (Yang et al. 2001; Bjerknes and Cheng 2006) also suggests derivation from the choanocyte cell component.
The association of developmental regulatory genes with appendages and sensory organs is evident from regulators, such as dachshund, which are required for sense-organ development and proper limb development. Vertebrate limbs are novel-derived feature of gnathostome vertebrates; consequently, the pharyngeal arches are the vertebrate structure most directly related to invertebrate appendages (Shubin et al. 1997; Depew et al. 1999). Thus, the hearing organs of flies found in the joints of appendages, the chordotonal organs (e.g., Johnston's organ; Todi et al. 2004), and the inner ear, derived from the pharyngeal arches, are both appendage-derived sensory structures. Moreover, common developmental gene expression and motor proteins, such as prestin and myosinVIIa, function in the Drosophila and vertebrate “ears,” arguing for evolutionary continuity through a shared ancestral auditory/inertial or comparable mechano-sensory structure (Boekhoff-Falk 2005; Fritzsch et al. 2006) borne in this “appendage” context.
Fringe and associated regulators provide another interesting example of commonality of regulation of sense organs and appendages; they function along the equator (akin to a dorso-ventral compartment boundary) of the Drosophila eye, and also in the evolutionarily secondary Drosophila wing, where they are responsible for defining the wing margin that itself bears a row of sensory bristles. Although beyond the scope of the review, a number of other genes function similarly in the development of sensory and appendage imaginal discs in Drosophila, further supporting the commonality of sensory structures and appendages.
The presence of eyes on all the parapodia in some species of polychaete (Verger-Bocquet 1981; Purschke 2005) documents evolutionary conversion of limbs to sense-organ-bearing structures. They are evolutionary “phenocopies,” producing phenotypes comparable to those engendered by eyeless overexpression that convert limb-borne chordotonal organs to eyes, as was discussed earlier. The presence of eyes on the terminal tube feet (appendages) near the ends of the “arms” (axial structures) of sea stars (Mooi et al. 2005; Jacobs et al. 2005) provides an instance in yet another bilaterian phylum, where sense organs and are associated with “appendages.” We argue subsequently that this shared aspect of the development of sense organs and appendages evolved basal to the Bilateria “senu stricto” as it is also found in the Cnidaria.
Homology of medusan and bilaterian sensory structures
Cnidarian sense organs are thought to be exclusive to the medusa, a point we dispute subsequently. Nevertheless, the sense organs of the medusa are highly developed and distributed across Scyphozoa, Hydrozoa, and Cubazoa. In those hydrozoans with a medusa stage, many have eyes associated with the tentacle base. The relative position of the eye and tentacle appears to be evolutionarily plastic; the necto-benthic Polyorchis penicillatus feeds on the bottom and its eyes are on the oral side, presumably aiding in prey identification on the bottom, whereas the nektonic P. monteryensis (Gladfelter 1972) has eyes on the aboral side of the tentacle, presumably aiding in identification of prey in the water column. Nevertheless, the hydrozoan eye appears to be closely associated with the base of the tentacle.
The rhopalium, the sense organ bearing structure of Scyphozoa, as well as the Cubozoa (a modified group within the scyphozoans), contains the statocyst and eyes. It is borne on the margin of the bell in the medusa. The rhopalia of cubozoan medusae contain eyes with lenses, the most dramatic of cnidarian sense organs. These facilitate swimming in these very active medusae with extremely toxic nematocysts. Other cnidarian eyes are simpler. These eyes tend to be simple eyespots or pinhole camera eyes that lack true lenses (see Martin 2002; Piatagorsky and Kozmik 2004 for review). In the scyphozoan, Aurelia, the statocyst is effectively a “rock on a stalk,” with a dense array of mechano-sensory cells that serve as a “touch plate” at the base of the stalk, where it can contact the overlying epithelium of the rhopalium (Spangenberg et al. 1996; Arai 1997). In Scyphozoa, there are typically eight rhopalia that alternate with eight tentacles around the bell margin. Cubozoa have four rhopalia that similarly alternate with tentacles. Although there are exceptions to this alternating tentacle/rhopalia pattern (Russell 1970), they appear to be derived. Thus, appendages in the form of tentacles and the sense organ bearing rhopalia occupy a similar position/field that appears to assume alternative fates in development. This is consistent with the arguments relating appendages and sense organs in Bilateria developed earlier and relates to our discussion of tentacles considered as appendages as well as sense organs in cnidarians discussed subsequently.
Several studies document expression of regulatory genes in Cnidaria that typically function in the development of bilaterian sense organs. These studies document a common aspect of gene expression albeit with significant variation. In Cubozoa, a paired-class gene has been identified that is expressed in sense-organ development (Kozmik et al 2003). Interestingly, this PaxB gene does not appear to be a simple homologue of eyeless/Pax6 as it contains an eyeless/Pax6 type homeodomain combined with a paired domain typical of PAX 2/5/8—a regulatory gene more closely associated with ear development that is also expressed in statocysts in mollusks (O’Brien and Degnan 2003). Statocysts are ear-like in their inertial function and are localized with the eye in the cnidarian rhopalium. Given that cubozoan statocyst expresses PAXB along with the eye, a PaxB-type gene appears to have undergone duplication and modification in the evolution of the bilaterian condition such that eyes and ears are differentially regulated by separate PAX6 and PAX 2/5/8genes. This evolution in the ancestry of eyeless/Pax6 contrasts with a number of other sense-organ regulatory genes such as sine oculis (Bebeneck et al. 2004), Brain3 (Jacobs and Gates 2001), and eyes absent (Nakanishi et al. manuscript in preparation), all of which appear to be extremely similar in their functional domains to specific bilaterian homologues. Thus, eyeless/PAX6 may have evolved more recently into its role in the eye developmental cascade than a number of other genes critical to the developmental regulatory cascade in the eye many of which also function in other sense organs.
In the scyphozoan Aurelia, a homologue of sine oculis is expressed in the rhopalia (Bebeneck et al. 2004), as is the case for Brain3 (Jacobs and Gates 2001) and eyes absent (Nakanishi et al. manuscript in preparation). Six-class genes are also expressed in the development of the eyes in the hydrozoan Cladonima (Stierwald et al. 2004). These sorts of data, taken together, provide a substantial argument for a shared ancestry between bilaterian and cnidarian sense organs generally. Shared ancestry of specialized classes of sensory organs, such as eyes, also appears likely. However, given that many conserved regulators usually function in multiple classes of sense organs, such as the eye and the statocyst/ear, their expression, by itself, has not yet provided unambiguous support for shared ancestry of particular bilaterian and cnidarian sense organ types.
In opposition to the above argument is the perception that cnidarian sense organs are exclusive to the medusa, and that the medusan phase is derived given the basal placement of the Anthozoa that lack such a stage in their life cycle (Bridge et al. 1992; Collins et al. 2006). However, a variety of arguments limit the strength of support for completely de novo evolution of cnidarian sense organs. Neither the polyp nor the medusa are present in outgroups, consequently the power of tree reconstruction to resolve the presence or absence of medusa or polyp is minimal (Jacobs and Gates 2003). This, combined with the frequency of loss of the medusa phase in hydrozoan lineages, limits confidence in the inferred absence of a medusa in the common ancestor. In addition, features that may merit consideration as sense organs are present in planula and polyps (discussed subsequently). Accordingly, the emphasis on the medusan phase of the life history may be unwarranted. In particular, statocysts are found in some unusual hydrozoan polyps (Campbell 1972) and ocelli associate with the tentacle bases in some stauromedusan (Scyphozoa) polyps (Blumer et al. 1995). The view that sensory organs are shared ancestral features of Bilateria and Cnidaria finds further support in recent arguments that cnidarians also share attributes of bilaterian axial development (Finnerty et al. 2004; Matus et al. 2006). In the following paragraphs, we review the distribution of potential sensory structures in Cnidaria, reconsider the commonalities shared by appendages and sensory structures, and then touch on the implications of bilaterian/cnidarian origins.
The cnidocytes of Cnidaria are innnervated (Anderson et al. 2004) and have triggers that respond to sensory stimuli. In some instances, they synaptically connect with adjacent sensory cells (Westfall 2004). Thus, cnidocytes are, at once, a potential source of sensory stimulation and, presumably, modulate their firing in response to neuronal stimuli (Anderson et al. 2004). Having acknowledged this complexity, we set it aside and limit the discussion to the integration of more traditional sensory cells into what may be considered sense organs.
Sensory structures in planula and polyp
In the planula larvae of Cnidaria, FMRF-positive sensory cells are found in a belt running around the locomotory “forward” end (aboral after polyp formation) of the planula ectoderm (Martin 1992, 2002). The axons of these cells extend “forward” along the basement membrane of the ectoderm and are ramified forming what appears to be a small neuropile at the aboral pole of the planula (Fig. 1). This feature varies among taxa; in hydrozoans, such as Hydractinia, the array of sensory cells appears closer to the aboral end of the elongate planula. There is also ontogenetic variation in which the sensory cells move closer to the aboral end prior to settlement (Nakanishi et al. manuscript in preparation). Strictly speaking, the sensory neurons of the cnidarian planula correspond to the “naked” sensory neurons discussed previously; however, one might consider dense arrays of such chemoreceptive and/or mechanoreceptive neurons as “precursors” of sense organs (see FMRF staining in Fig. 1A). Expression data for atonal in hydrozoan planulae (Seipel et al. 2004) also suggest that this integrated array of sensory cells could merit “sense organ” status.
Oral structures, the hypostome and manubrium of the polyp and medusa, respectively, may rise to the status of sense organs. In Aurelia ephyrae (early medusa), sensory cells are present in rows on the edges of both the ectoderm and endoderm of the manubrium (Fig. 1). POU genes such as Brain3 (unpublished data) are expressed in the manubriium of Aurelia, as is a homologue of sine oculis (Bebenek et al. 2004). Similar sine oculis expression in the manubrium is evident in the hydrozoan Podycoryne, but this may not be the case in Cladonema where a related Six gene myotonix/Six4,5 is expressed in the manubrium (Steirwald et al. 2004). In Podycoryne, limited expression of atonal is evident in the manubrium (Seipel et al. 2004), and PaxB is expressed in the manubrium and hypostome (Groger et al. 2000).
Tentacles as sense organs and appendages?
Cnidarian tentacles are variable; ectoderm and endoderm layers and a central lumen connected to the gastrovascular cavity are typical of anthozoan tentacles. In contrast, polyp tentacles of scyphozoans and some hydrozoans lack a lumen; a single row of large vacuolated endodermal cells is present at the core of a slender tentacle. A variety of tentacle morphologies are also present in medusae. We discuss whether tentacles are (1) sense organs, (2) sense organ bearing structures, and (3) whether tentacles and rhopalia (that bear sense organs in scyphozoans) are alternative developmental outcomes of an initially common developmental field or program.
Ultrastructural studies as well as markers such as FMRF that typically recognize sensory cells and neurons (Fig. 1) document arrays of sensory cells, in tentacles that are substantially denser than those found in the body wall of the polyp or in the medusan bell (Fig. 1). Optix homologues are also expressed in certain presumed sensory neurons or cnidocytes in tentacles of Podocornyne (Stierwald 2004). Sensory cells form concentrations at the base of the tentacle or, in some instances, at the tips of the tentacles (Holtman and Thurm 2001); these concentrations merit consideration as sense “organs.”
Sense-organ-related genes are preferentially expressed near the bases of hydrozoan tentacles; sine oculis and PAXB are expressed here in Podocoryne, a hydrozoan medusa that lacks eyes (Groger et al. 2000; Steirwald et al. 2004). Sensory gene expression associated with tentacle bases is not exclusive to medusae. In the anemone Nematostella, PaxB homologues are expressed adjacent to the tentacles (Matus et al. 2007). In addition, the base of the tentacle is the locus of ocelli in some unusual polyps as discussed earlier (Blumer et al. 1995). Thus, a developmental field specialized for the formation of sensory organs appears to be associated with the bases of cnidarian tentacles, but tentacle terminal concentrations of sensory cells also occur, as is the case in the ployp tentacles of the hydrozoan Coryne (Holtmann and Thurm 2001).
In Hydra, an aristaless homologue is expressed at the base of tentacles (Smith et al. 2000), comparable to the proximal component of expression seen in arthropod limbs (Campbell et al. 1993). Transforming growth factor (TGF)-β expression always precedes tentacle formation in tentacle induction experiments (Reinhardt et al. 2004) and continues to be expressed at the tentacle base. Both decapentplegic and aristaless are involved in the localization and outgrowth of the appendages in flies (Campbell et al. 1993; Crickmore and Mann 2007). Thus, there are also common aspects of bilaterian appendage and cnidarian tentacle development.
As noted earlier, in typical Scyphozoa, rhopalia alternate with tentacles in a comparable bell-margin position; in Hydrozoa, sense organs associate with the tentacle bases. Overall, there is support for a common appendage/sense-organ field in Cnidaria comparable to that evident in Bilateria as discussed earlier. This appendage/sense-organ field appears to be a shared-derived feature of bilaterian and cnidarian body plans, which along with the recently demonstrated common aspects of dorso-ventral-axis formation (Finnerty et al. 2004; Matus et al. 2006) should aid in understanding the common aspects of divergent bilaterian and cnidarian form.
Sensory attributes of sponges
Sponges are thought to constitute the most basal branch, or branches, of the animal tree and a progressivist views of evolution have has long treated them as primitively simple (Jacobs and Gates 2003). Yet, there is increasing evidence that sponges are not as simple as often anticipated. Some sponge lineages exhibit (1) coordinated motor response to sensory stimuli and others posses an electrical-conduction mechanism; (2) sponges have genes encoding proteins that function in a range of bilaterian developmental processes; and (3) sponges have many of the genes employed in the development of sense organs. The presence of genes known to function in eumetazoan sense-organ development in a group lacking formal sense organs presents interpretive challenges. Certain sets of larval cells, or the grouping of choanocytes into functional arrays in, represent possible sponge structures potentially related to eumetazoan sense organs. We discuss these briefly and explore the possibility that multiple organs including kidneys and sense organs may share ancestry with ensembles of choanocytes.
Sponges exhibit contractile behaviors (reviewed by Leys and Meech 2006; Elliot and Leys 2003). In the small, freshwater sponge Ephydatia, an inhalent expansion phase precedes a coordinated contraction that forces water out of the osculum. This contractile activity generates high-velocity flow in the finer channel systems that then propagate toward the osculum. Effectively, this seems to be a “coughing” mechanism that eliminates unwanted material, chemicals, or organisms from the vasculature. Sponges are known to have specialized contractile cells, termed myocytes that have been compared to smooth-muscle cells; however, other epithelial cell types (pinacocytes and actinocytes) contribute to contractile behavior (reviewed by Leys and Meech 2006).
In hexactinellids, “action potentials” that appear to involve calcium propagate along the continuous membranes of the syncytium that constitutes the inner and outer surface of these sponges (Leys and Mackie 1997; Leys et al. 1999). This propagation of signals along the syncytium permits rapid coordinated choanocyte response in hexactinellids. In other classes of sponges, propagation of information appears to involve calcium dependent cell/cell communication (Leys and Meech 2006) discussed further below.
As mentioned earlier, recent work by Sakarya et al. (2007) documents the presence of “postsynaptic” proteins and argues that these proteins are organized into a postsynaptic density comparable to that found in eumetazoan synapses. This suggests surprising functionality given the absence of formal synapses in sponges. An EST study of the demosponge Oscarella (Nichols unpublished data; see Nichols et al. 2006 for methods, accession numbers follow name below) provides additional support for the presence of molecular components that are required for vesicle-related signaling function. These include: (1) synaptogamin (EC375291.1), involved in calcium-dependent vesicle fusion and required for many aspects of eukaryotic vesicle trafficking, including neurotransmitter release; (2) additional SNARE-complex components similarly involved in vesicle transport—syntaxin (EC370432.1, EC370199.1, EC370795.1, EC750565.1), N-ethylmaleimide-sensitive fusion protein attachment protein-α (EC374655.1, EC375028.1), and N-ethylmaleimide-sensitive factor (EC376726.1, EC369036.1); (3) neurocalcin (EC374277.1, EC373904.1, EC373904.1, EC371466.1, EC375078.1, EC374707.1), a neural-specific agent that modulates calcium-dependent interactions with actin, tubulin, and clathrin; (4) as well as genes typically involved in axon guidance such as slit (EC376833.1). Of these Oscarella sequences, inferred functions (1–3) involved in vesicle trafficking are essential for synaptic function; however, they also have other functions in eukaryotes. On the other hand, axon guidance (4) would appear more specific to metazoan cell fate and neural function. These recent observations in sponges suggest the high activity of equipment involved in vesicle transport, and the presence of some synaptic and developmental signaling components typically associated with bilaterian neural systems.
Given that sponges lack formal synapses, it is worth noting that nonsynaptic communication between cells via calcium waves can occur through a variety of mechanisms. One such class of mechanism involves gap junctions or gap junction components, which have yet to be documented in sponges and are presumed absent. Others involve the vesicular release of molecules such as ATP that can operate through receptors associated with calcium channels or through specific classes of GPCRs (see North annd Verkhratsky 2006 for review of purinergic communication). Such receptors are known to permit nonsynaptic intercellular communications in nerves and nonneuron components such as between glial cells. Mechanisms of this sort, involving nonsynaptic vesicular release of signaling molecules and a “calcium wave” propagation, seem broadly consistent with available information on communication in cellular sponges reviewed in detail by Leys and Meech (2006).
The ring-cells around the posterior pole (relative to direction of motion) of the parenchyma larva of the demosponge Amphimedon has been shown to be photosensitive and to respond to blue light (Leys et al. 2002; see Maldonado et al. 2003 for observations on other demosponge larvae). These cells effectively steer the sponge, using long cilia providing for a phototactic response. Sakarya et al. (2007) document that flask cells of larval sponges express proteins involved in postsynaptic organization in Bilateria, and speculate that these cells are sensory. These larval sensory attributes are of interest as larvae provide a likely evolutionary link with the radiate and bilaterian groups (Maldonado 2004).
Groups of choanocyte cells in adult sponges also bear some similarity to eumetazoan sensory structures as: (1) choanocytes are crudely similar in morphology to sensory cells, particularly mechanosensory cells; (2) the deployment of sponge choanocytes in chambers is similar to the array of sensory cells in sense organs; and (3) choanocytes are a likely source of stimuli that produce the contractions and electrical communications as noted above. Choanocytes of sponges and choanoflagellates present a cilium/flagellum surrounded by a microvillar ring on the apex of the cell, which bears at least superficial similarity to the typical organization of many sensory cells, such as those of the ear (Fritzsch et al. 2006; Fain 2003). Clearly, chemical signals in the water can induce contractile responses in demosponges (Nickel 2004; Ellwanger et al. 2007; Leys and Meech 2006). In addition, it appears likely that mechano and chemosensory responses to particles would be necessary for the feeding function of the choanocyte and that communication between adjacent choanocytes in the choanosome structure would also be essential to feeding. Feeding behavior appears coordinated across sponges, rather than just within choanosomes as different types of particles are preferred under different circumstance (Yahel et al. 2006).
The molecular complexity of sponges exceeds that expected based on their presumed “primitive” nature. Nichols et al. (2006) reported a range of extracellular matrix proteins as well as components of the major intercellular signaling pathways operative in metazoan development from their EST study of the demosponge Oscarella. Larroux et al. (2006) reported a diverse array of homeodomains and other DNA-binding regulatory genes from the demosponge Amphimedon queenslandica (formerly Reneira). Thus, sponges possess a significant subset of the equipment used to differentiate cells and tissues in Bilateria and Cnidaria [see Ryan et al. (2006) for a recent survey of cnidarian homeodomans from the Nematostella genome and Simionato et al. (2007) for survey of bHLH regulators across Metazoa, including cnidarians and demosponge genomic data]. Turning to sense organ-associated regulators, sine oculis homologues are present in all classes of sponges (Bebeneck et al. 2004), as are homologues of Brain3 (Jacobs and Gates 2001, 2003, and unpublished data). Similarly, relatives of atonal are present in demosponges (Simionato et al. 2007). Thus, sponges appear to have the regulatory gene cascades associated with sense-organ development in Eumetazoa.
As noted earlier, vertebrate sensory organs have a surprising amount in common with the kidney; for example, ear and kidney both express Pax6, eyes absent, and sine oculis in development and numerous genetic defects affect both structures (Izzedine et al. 2004). Consideration of sense organs, and organs that eliminate nitrogenous waste, both as evolutionary derivatives or relatives of a choanocyte chambers, may help explain these commonalities. The fluid motion engendered by choanocyte chambers renders these structures the central agency in nitrogenous waste excretion, in addition to their other functions (Laugenbruch and Weissenfels 1987); vacuoles involved in the excretion of solids following phagocytic feeding presumably represent a separate aspect of waste disposal (Willenz and Van De Vwer 1986). In a number of bilaterian invertebrates, nitrogen excreting protonephridia consist of specialized ciliates flame cells that generate the pressure differential critical for initial filtration much as choancytes do. These appear intermediate between choanocytes and matanephridia that rely on blood pressure for filtration (Bartolomaeus and Quast 2005). Thus, we draw attention to the potential evolutionary continuity of function and structure between associations of choanocytes and protonephridia, and ultimately metanephridia. These are of interest in the context of the potential for explaining the common features of sense organs and kidneys (Izzedine et al. 2004). Such explanations are necessarily speculative, but will soon be subject to more detailed test with an increasing knowledge of gene expression and function in sponges. It should also be noted that this argument does not negate the possibility that a number of other structures such as the neuroendocrine structure of the gut epithelium, as mentioned above, might also derive from or share ancestry with the choanosome.
Tree topology is critical to evolutionary interpretation of the events surrounding the evolution of sensory systems in the basal Metazoa. There is increasing agreement on the relationships between bilaterian phyla and the placement of Cnidaria as sister to the Bilateria (Halanych 2004), as well as the relationships between the classes of Cnidaria as discussed earlier. Recent work suggests (Borchiellini et al. 2001; Medina et al. 2001) that Eumetazoa derive from a paraphyletic sponge group. These analyses tend to place the Eumetazoa as sister to the calcareous sponges. Sponge paraphyly implies that the ancestral eumetaozan was sponge-like (Eerkes-Medrano and Leys 2006) with choanocytes and other broadly distributed attributes of sponges, lending credence to arguments that choanosome development may have contributed to the evolution of sensory structures as argued above.
The enigmatic Placozoa are also of interest as they may provide information on the nature of the stem of the metazoan tree potentially permitting interpretation of Vendian (late Precambrian) fossils (Conway-Morris 2003). The large size of the placozoan mitochondrial genome is comparable to those found in protists. Animal mitochondria are smaller, suggesting that Placozoa may be the most basal branch of the Metazoa, but placozoan mitochondrial sequence data yield tree topologies that place all basal Metazoa including Placazoa in a sister clade to Bilateria (Signorovitch et al. 2007). Ribosomal genes place Placozoa in a variety of basal postions (Borchiellini et al. 2001; Halanych 2004), but are consistent with the basal placement and/or paraphyly of sponges discussed above. Interestingly, there is evidence for PAX-like genes in the presumptively basal Placozoa (Hadrys et al. 2005). This is broadly consistent with evolution of many of the major classes of regulatory proteins that function in metazoan development prior to the evolution of the metazoan radiation (Derelle et al. 2007 provides a recent analysis of homeobox gene families in this context).
We have argued that many aspects of sense organ evolution preceded the evolution of formal organs in the triploblastic Bilateria. Clearly, Cnidaria have well-developed neural and sensory features some of which may merit treatment as “organs”; however, even sponges appear to have precursory elements of sensory organization. In addition, sense-organs share attributes with endocrine structures, appendages, and kidneys. We argue that these similarities are a product of derivation from common ancestral structures. In a more general sense as one compares structures in divergent ancient lineages such as the basal lineages of the Metazoa, we feel that similarities that are the product of shared ancestry are likely to be manifest in surprising and subtle ways. Thus, neither inferences of similarity as indicative of strict homology nor dismissal of similarity as products of convergence or cooptation should meet with facile acceptance.
We thank the symposium organizers for their efforts, Sally Leys, Chris Winchell, the Martindale and Oakley labs as well as the NESCent Catalysis Group on “Origins and Evolution of Chemoreception” for related discussions, anonymous reviewers for their helpful critique, and NASA and the NASA Astrobiology Institute for their support for research in the Jacobs lab.