Nervous System

1.1 Traditional scenarios for bilaterian nervous system evolution

Markedly narrative, traditional evolutionary scenarios aimed at delivering plausible explanations for the origin and diversification of animal morphology (e.g. the diversity of neural arrangements), while at the same time were used to define phylogenetic relationships among animal lineages. This inevitably led to circular reasoning (e.g. an animal lineage was placed in the phylogeny based on their morphology, and at the same time their given phylogenetic position was used to justify their morphology and evolution of a specific character), and endless debates between the advocates (see i.e. Dougherty, 1963). In most of the cases, these scenarios focused on defining linear transformations of adult body plans, with developmental biology only playing a minor role. When embryos were considered, the argumentation was strongly influenced by Haeckel’s ‘biogenetic law’ (Haeckel, 1866), namely that ontogeny (i.e. development) is a fast recapitulation of phylogeny. Embryos were not yet seen adaptive and evolvable, but as mirrors of ancient adult forms that existed once in the past, which were figuratively named after their corresponding developmental stage (e.g. ‘Blastaea’ and ‘Gastraea’). As we illustrate below with Reisinger’s ‘orthogon’-scenario (Reisinger, 1925), Sedgewick’s and Balfour’s ‘oral nerve ring’ scenario (Balfour, 1883; Sedgwick, 1884) and the ‘nemertean’ and ‘annelid’ scenarios by Hubrecht (Hubrecht, 1883, 1887) and Dohrn (1875) respectively, traditional scenarios proposed alternative evolutionary transformations to explain the emergence of condensed neural cords from a diffuse ectodermal nerve net composed out of basiepidermal interconnected neurons (Hejnol and Rentzsch, 2015).

1.1.1 The ‘orthogon’ scenario

An orthogonal nervous system is defined by multiple pairs of longitudinal cords distributed along the dorsoventral axis of the animal and connected with transverse commissures (Richter et al., 2010). Such nervous system architecture is mainly present in Platyhelminthes, but also in some representatives of other animal lineages (e.g. annelids). The putative position of Platyhelminthes as one of the earliest branches of Bilateria led Reisinger to propose that the condensation of a nerve-net like nervous system, as found in Cnidaria, into an orthogonal CNS, as found in Platyhelminthes, could explain the subsequent evolution of the different neuronatomies of bilaterian animals (Fig. 1A) (Hanström, 1928; Reisinger, 1925, 1971; Steinböck, 1966). In this way, the loss of the dorsal longitudinal cords of the ancestral orthogon would have originated the ventral cords found in many protostomian lineages, while the loss of the ventral cords of the orthogon would explain the dorsal location of the nerve cord in chordates (Fig. 1A). However, current molecular phylogenies do not place Platyhelminthes as the sister group of Bilateria (Dunn et al., 2008; Kocot et al., 2016; Laumer et al., 2015; Struck et al., 2014), but well-nested within Spiralia (see “The nervous system in the animal phylogeny” section below; Fig. 2), a major bilaterian clade that comprises animal groups with diverse neuroanatomies. Apart from Platyhelminthes, only a few other lineages, such as some annelids and mollusks (solenogastres, polyplacophorans and monoplacophorans) (Bullock and Horridge, 1965; Schmidt-Rhaesa, 2007; Schmidt-Rhaesa et al., 2016), exhibit a more or less evident orthogonal-like CNS (Fig. 2). It is thus unlikely that the platyhelminth orthogonal CNS arrangement represents the ancestral neuroanatomy of Spiralia, because if it does, it was then lost or heavily modified in most other spiralian taxa. Moreover, clear orthogonal CNS are absent or at least unclear in other bilaterian taxa (Bullock and Horridge, 1965; Schmidt-Rhaesa, 2007; Schmidt-Rhaesa et al., 2016), and the ancestral condition for Xenacoelomorpha, the sister taxa to all remaining bilaterians (see “The nervous system in the animal phylogeny” section below) is most likely a diffuse nerve net without any neural condensations (Hejnol and Pang, 2016). Therefore, the presence of an orthogonal arrangement of the neural condensations in the last common bilaterian ancestor is not supported.

Fig. 1. Traditional scenarios for the evolution of a centralized nervous system. (A) In the orthogon scenario, the ancestral bilaterian was a flatworm-like animal with an orthogonal arrangement of the nervous system. Dorsal and ventral nerve cords evolved by the subsequent loss of ventral and dorsal nerves of the orthogon, respectively. (B) The ‘oral nerve ring’ scenario explains the evolution of ventral paired nerve cords by the elongation and fusion in the middle of the oral nerve ring condensation of extant sea anemones. (C) In the nemertean scenario, the evolution of the chordate CNS occurs from a nemertean-like ancestor, through the movement of the ventrolateral nerve cords to the dorsal side. (D) In the annelid scenario, the chordate CNS evolves from an annelid-like ancestor that flips over its dorsoventral axis. Drawings are not to scale, and the CNS is in blue. See main text for references. An, anus; mo, mouth; vnc, ventral nerve cords.

Fig. 2. The diversity of neural anatomies in Metazoa. Distribution of neuroanatomical characters in representative metazoan lineages under the current phylogenetic relationships supported by molecular data (Dunn et al., 2014). Neuroanatomy is based on (Schmidt-Rhaesa, 2007).

1.2 The nervous system in the animal phylogeny

As described above, most scenarios for the evolution of the nervous system were strongly influenced by pre-cladistic considerations of animal interrelationships. However, our view of the animal tree of life has changed profoundly in the last two decades, since the advent of molecular phylogenies (Dunn et al., 2014). Bilaterian lineages –those with anteroposterior, dorsoventral, and left-right axes– have been rearranged into three major monophyletic groups, Deuterostomia, Ecdysozoa, and Spiralia (Fig. 2), with Deuterostomia (e.g. vertebrates) being the sister taxa to Ecdysozoa (e.g. arthropods) and Spiralia (e.g. annelids) (Aguinaldo et al., 1997; Laumer et al., 2015). Although the internal phylogeny of these three major clades still shows some uncertainties (Dunn et al., 2014; Giribet, 2016), there is robust evidence placing Xenacoelomorpha (e.g. acoel worms), which was previously placed within Platyhelminthes, as the sister group to all remaining bilaterians (Cannon et al., 2016; Ruiz-Trillo et al., 1999). The former Coelenterata (Cnidaria ​+ ​Ctenophora) has little support (Ryan et al., 2013; Simion et al., 2017; Whelan et al., 2015), and Cnidaria now stands alone as the sister group to Bilateria. The position of Ctenophora is still controversial (Shen et al., 2017; Simion et al., 2017), but most recent analyses place this group as the sister lineage to all remaining animals.

The new animal phylogeny generates uncertainty around central questions on nervous system evolution that were previously thought settled (Hejnol and Lowe, 2015; Liebeskind et al., 2016; Ryan, 2014). Ctenophores have neurons and a diffuse nerve net, which are absent in sponges and placozoans (Bullock and Horridge, 1965) (Fig. 2). Therefore, the animal nervous system either has a single origin and got independently lost in Porifera and Placozoa or evolved twice in Ctenophora and Cnidaria ​+ ​Bilateria (Jekely et al., 2015; Liebeskind et al., 2016; Moroz et al., 2014; Ryan, 2014; Ryan et al., 2013; Simion et al., 2017) (Fig. 3). Unfortunately, developmental studies on ctenophore neurogenesis that could shed light on this debate are very limited (Martindale and Henry, 1999; Norekian and Moroz, 2016). Gene expression studies in the ctenophore Mnemiopsis leidyi could not so far provide evidence for the homology of the developmental pathways involved in neurogenesis in ctenophores (Pang and Martindale, 2008; Schnitzler et al., 2014; Simmons et al., 2012). However, the absence of core bilaterian neurogenic regulators in ctenophore gene repertoires (Moroz et al., 2014; Ryan et al., 2013) and the failed attempt to identify neural cell types by using typical pan-neural orthologs (Sebé-Pedrós et al., 2018) suggest that if homologous, the gene networks controlling the specification and development of neurons might be significantly different between these animals and cnidarians and bilaterians. Similarly, the internal rearrangement of Bilateria has moved taxa that were once key to explain nervous system diversification (e.g. platyhelminths, annelids, onychophorans, nemerteans; see above) to internal nodes within the Ecdysozoa and Spiralia (Dunn et al., 2014) (Fig. 2). This, together with the vast diversity of CNS anatomies within Bilateria (Schmidt-Rhaesa et al., 2016), has blurred the identification of the ancestral bilaterian neuroanatomy (Fig. 2), and in particular of the sequence of events that led to the emergence of the vertebrate CNS, with an anterior brain and a medially condensed dorsal nerve cord.

Fig. 3. The evolution of the nervous system. Possible evolutionary scenarios for the evolution of neurons and a nervous system in Metazoa given the current position of Ctenophora. (A) If ctenophore, cnidarian and bilaterian nervous systems are homologous, neurons got independently lost in sponges and placozoans. (B) Alternatively, the nervous systems of ctenophores and cnidarians ​+ ​bilaterians evolved convergently.

As some traditional ideas have proven wrong, the generally robust new phylogenetic framework has set the foundations for a reinterpretation of animal evolution (Dunn et al., 2014; Giribet, 2016). For instance, the placement of xenacoelomorphs, which ancestrally solely possessed a diffuse basiepidermal nerve net (Achatz et al., 2013; Hejnol and Pang, 2016; Raikova et al., 2016), as the intermediate taxon between cnidarians and the remaining bilaterians decouples the evolution of a CNS from the emergence of Bilateria and suggests that neural condensations have evolved repeatedly within bilaterian animals.

1.3 The early development of the CNS in bilateria

With most traditional views on nervous system evolution falsified, a thorough comparative investigation of neurogenesis under an unbiased phylogenetic framework emerges as the alternative to reconstruct ancestral character states in the evolution of the nervous system. Are there comparable, and perhaps homologous processes in bilaterian neurogenesis that can illuminate the origin of a CNS? In most animal embryos, the primary committed cells that will hierarchically give rise to all differentiated cell types of an organism are spatially organized in relation to the animal-vegetal axis of the oocyte, which is defined by the site of extrusion of the polar bodies (i.e. animal pole) (Goldstein and Freeman, 1997; Martindale and Hejnol, 2009; von Baer, 1834). The earliest events of cell fate specification occur during cleavage, so that precursor cells immigrate and acquire their final embryonic destinations during gastrulation. Therefore, it is possible to identify the prospective embryonic areas, or even cells, that will eventually contribute to the formation of the nervous system at these very early embryonic stages. This can be inferred from the expression of upstream neurogenic genes (see below), but ideally cell fates should be demonstrated by actual cell tracking techniques (Amat and Keller, 2013; Hejnol and Schnabel, 2006; Sulston et al., 1983). Although limited to a handful of animal lineages, fate mapping gives a cellular ontogenetic context to molecular data, thus improving evolutionary comparisons, and can offer a general framework to understand common principles in nervous system development in animals.

In ctenophores, neuronal and ectodermal cell fates have a common developmental origin (Martindale and Henry, 1999), yet the specification and development of neurons appear to be a rather late process in development (Moroz et al., 2014; Norekian and Moroz, 2016). Differently from ctenophores, the cnidarian nerve net develops from both the ectoderm and the endoderm in a process that starts early in development, before the onset of gastrulation (Nakanishi et al., 2012; Richards and Rentzsch, 2014) (Fig. 4). Intracellular injections of fluorescent dyes in early blastomeres of the acoel Neochildia fusca demonstrate that the diffuse basiepidermal nerve net shares a common developmental origin only with the ectoderm (Henry et al., 2000) (Fig. 4). The sensory statocyst, however, appears to derive from, or at least require the presence of, the vegetal macromeres (endomesoderm) for its differentiation (Boyer, 1971). Indeed, the nervous system has a predominant ectodermal origin in all Bilateria.

Fig. 4. The diversity of neuroectodermal fate maps in Bilateria. Schematic fate maps of representatives of the major bilaterian lineages and cnidarians, with special emphasis on the origin of the nervous system and endoderm. In the cnidarian N. vectensis, the neurons that form the diffuse nerve net (red lines) develop from both the ectoderm and endoderm (purple). In Bilateria, the vast majority of neurons develop separate from the endoderm. The mode of specification and spatial position of neuronal progenitors is variable in Bilateria, even among lineages that share a common developmental program, such as spiral cleaving embryos (Trochozoa). Drawings are not to scale. Ma, macromeres; me, mesomeres.

In Spiralia, cell lineage studies have greatly concentrated in those animal groups that share the stereotypical quartet-spiral cleavage program (Hejnol, 2010; Henry and Martindale, 1999). This is a broadly conserved early embryonic program, probably ancestral to the whole Spiralia (Hejnol, 2010; Henry, 2014), where embryos get divided in four quadrants, named A to D, each roughly forming the left, anteroventral, right, and posterodorsal region of the animal respectively. The cell division from the 4- to the 8-cell stage is asymmetric, with the four blastomeres at the vegetal pole being larger (i.e. macromeres) than the cells at the animal top (i.e. micromeres). Before gastrulation, macromeres bud off four tiers of micromeres, which are named 1 to 4 (to represent the tier) and a to d (to represent the quadrant they come from) in the classic nomenclature used to describe spiralian cell lineages (Conklin, 1897). Generally, in groups such as polyclad flatworms, mollusks, annelids and nemerteans, the brain and anterior sensory organs (e.g. the larval apical tuft and photoreceptors) originate from the first tier, or quartet, of animal micromeres, which also form the head ectoderm (Henry and Martindale, 1999) (Fig. 4). Other components of the nervous system, such as the nerve cords, originate from derivatives of the second and third quartet micromeres (Henry and Martindale, 1999). For instance, derivatives of the first quartet micromers (1a¹-1d¹) form the brain in the annelid Capitella teleta, while the blastomere 2d gives rise to dorsal parts of the brain and the ventral nerve cord, and the 3a cell forms isolated neurons (Meyer et al., 2010; Meyer and Seaver, 2010). However, the first quartet micromers also contribute to anterior regions of the ventral nerve cord in the annelid P. dumerilii (Ackermann et al., 2005), and in the mollusk gastropod Crepidula fornicata the visceral nerve cords arise from the 2b blastomeres (Hejnol et al., 2007; Lyons et al., 2015). In the nemertean Cerebratulus lacteus, the larval nervous system originates from 1c, 1d, 2a, 2c, 2d, 3c and 3d (Henry and Martindale, 1998), and in the cyphonautes larva of the bryozoan Membranipora membranacea, which has modified spiral cleavage, the apical organ forms from the 1a–1d blastomeres (Vellutini et al., 2017). Therefore, there is a significant degree of interspecies variation underlying general spiralian developmental plans. This, together with the lack of detailed cell lineage investigations in gnathiferans (rotifers, gnathostomulids, and micrognathozoans) and gastrotrichs makes difficult to confidently infer the exact cellular mode of nervous system development for Spiralia.

The Ecdysozoa comprises three large monophyletic groups, namely Scalidophora, Nematoida, and Panarthropoda (Dunn et al., 2014). There is virtually no cellular data on the embryonic development of the nervous system in Scalidophora (i.e. priapulid worms, kynorynchs and loriciferans), where only the expression of the nervous system marker otx in the priapulid Priapulus caudatus indirectly suggests that the circumoral brain originates from ectodermal cells at the introvert-trunk boundary (Martin-Duran et al., 2012). This evident lack of knowledge stands in stark contrast with the high-resolution cell lineage of the nervous system in Nematoda, in particular in the developmental research system Caenorhabditis elegans. In this nematode, the nervous system is mostly formed from the ectodermal founder cell AB, with some contribution of the C blastomere, and even the mesodermal precursor MS (Sulston et al., 1983) (Fig. 4). During gastrulation, AB precursors spread from anterior to posterior, and trunk ventral neuroblasts finish the closure of the blastopore as they get internalized by dorsally expanding epithelial cells (Sulston et al., 1983). Unfortunately, less is known for most other nematode lineages, as well as for Nematomorpha, the sister group of nematodes. Within Panarthropoda, cell lineage data in the tardigrade Thulinia stephaniae demonstrates that the brain and the ventral nerve ganglia develop from separate neural precursors that delaminate from the ectoderm (Hejnol and Schnabel, 2005). Direct cell tracking is missing in onychophorans, but morphological and gene expression data indicate that delamination of neural progenitors from the neuroectoderm forms the nervous system (Mayer and Whitington, 2009). A similar, more organized and invariant process is seen in pancrustaceans (Fig. 4), where isolated neuroblasts that either delaminate (insects) or remain epithelial (crustaceans) divide asymmetrically to form the neurons (Dohle et al., 2004; Whitington, 1996). In addition, some parts of the nervous system (optic lobes, stomatogastric nervous system, neuroendocrine system) involve the ingression of larger groups of neuroectodermal precursors, a process that seems to be more prevalent in arthropod groups like chelicerates and myriapods (Stollewerk and Chipman, 2006). As in Spiralia, the lack of detailed studies in scalidophorans and nematodes with less derived nervous system development hampers inferring the ancestral mode of nervous system development in Ecdysozoa, and thus in Protostomia as a whole.

In Deuterostomia, direct tracing of the fate of individual blastomeres into neurons has not been performed in hemichordates. However, the basiepidermal nerve net probably derives from the animal mesomeres and macromeres, which also form the ectoderm of the embryo, in Saccoglossus kowalevskii (Colwin and Colwin, 1951) (Fig. 4). In line with their diffuse basiepidermal nerve net, early neurogenic markers like soxB and elav are ubiquitously expressed in the ectoderm of the gastrula of S. kowalevskii (Cunningham and Casey, 2014; Lowe et al., 2003). A similar fate map is observed in sea urchins (Cameron et al., 1987) (Fig. 4), but the canonical Wnt pathway and the Nodal and BMP2/4 pathway restricts neurogenesis to the anterior and the ciliary band neuroectoderm (Angerer et al., 2011). However, sea urchin larvae also form a subset of their neurons from pharyngeal endodermal derivatives (Wei et al., 2011). Chordates, on the other hand, exhibit a highly centralized nervous system, which develops from the dorsal neural plate, of ectodermal origin (Lumsden and Krumlauf, 1996) (Fig. 4). In cephalochordates and vertebrates there is no invariant blastomere lineage for the neural plate, but a defined set of blastomeres generates the neural plate in ascidians, in accordance with their invariant cell lineage (Nicol and Meinertzhagen, 1988; Nishida, 1987). Therefore, the situation observed in hemichordates is reminiscent of that of cnidarians and xenacoelomorphs, and might thus represent the ancestral condition for Deuterostomia (Holland, 2003; Lowe et al., 2015), albeit this interpretation is still debated (Arendt et al., 2016; Holland et al., 2013).

This brief outline of early bilaterian neurogenesis already highlights how diverse the formation of the nervous system is in animal embryos. The uncertainties about the homology between the ctenophore nervous system and the cnidarian/bilaterian neural tissues (Fig. 3) limit drawing far-reaching conclusions regarding the ancestral metazoan mode of neural development. However, the comparison between Cnidaria and Bilateria strongly suggests that the neurogenic potential became restricted to the ectoderm at the onset of Bilateria, and thus the rare cases of mixed germ-layer origins, like in nematodes (Sulston et al., 1983) and echinoderms (Wei et al., 2011), are probably secondary deployments of neurogenic programs in cellular lineages that give rise to mostly mesodermal (as in nematodes) and endodermal (as in echinoderms) derivatives. How the development of the nervous system became restricted to the ectoderm in Bilateria is still unclear (Martindale and Hejnol, 2009), and will require of detailed mechanistic investigations in key lineages of the animal phylogeny, such as cnidarians, xenacoelomorphs, hemichordates, scalidophorans and gnatiferans. In this regard, neurogenesis is largely unknown in many bilaterian taxa, which hampers the reconstruction of homologous developmental processes for the major bilaterian clades. Despite this paucity of knowledge, our current understanding of cnidarian and bilaterian neurogenesis has allowed refuting most traditional scenarios for CNS evolution (see above) and has set the grounds for a more accurate interpretation of an increasing amount of molecular data in a growing number of bilaterian taxa.

1.4 The molecular patterning of the bilaterian nervous system

The variation in the embryonic fate maps and cellular aspects of nervous system development summarized above ultimately relies on modifications of the underlying neurogenic processes. In this respect, the comparison of expression patterns of orthologous genes involved in the specification and regionalization of bilaterian neural tissues has transformed the study of CNS development and evolution (Arendt et al., 2016; Hejnol and Lowe, 2015; Holland, 2015; Holland et al., 2013; Puelles and Ferran, 2012; Strausfeld and Hirth, 2013; Tosches and Arendt, 2013). Although in many cases there is a lack of functional investigations of the genes used for comparisons, gene expression data has been widely used to propose evolutionary scenarios and working hypotheses (Arendt et al., 2016; Hejnol and Lowe, 2015; Holland et al., 2013; Lowe et al., 2003, 2006). For instance, the similar molecular profiles of the arthropod mushroom bodies with anterior neural condensations in vertebrates, the annelid Platynereis dumerilii, and other invertebrate taxa has grounded the hypothesis that complex brain centers were already present in the last common bilaterian ancestor (Tomer et al., 2010; Wolff and Strausfeld, 2015). As sequencing techniques and molecular methods improved, this comparative molecular approach incorporated a growing number of bilaterian lineages to CNS evolutionary discussions, thus encompassing a significant proportion of the diversity of neural anatomies and developmental modes found in Bilateria. As a result, many of the evolutionary interpretations originally based on data from a few bilaterian lineages have been either solidified or questioned (Hejnol and Lowe, 2015). In the following two sections, we review how increased taxon sampling has affected the use of anteroposterior and dorsoventral neuronal patterning to explain the evolution of the brain and nerve cords respectively.

1.4.1 Anteroposterior patterning

Similar to other morphological features, the nervous system of most bilaterian animals is regionalized along the anteroposterior axis (Bullock and Horridge, 1965). This regionalization occurs at both the morphological and the functional level and gets reflected on how upstream neural regulatory genes and downstream terminal differentiation genes are deployed along the anteroposterior axis of the neural tissue. A large number of transcription factors (e.g. six3/6, foxQ2, otx, otp, fezf, pax6, pax2/5/8, en, irx genes, Hox genes) and signaling pathways (e.g. the Wnt/β-catenin pathway, Hedgehog pathway, FGF pathway, EGFR pathway) are differentially expressed along the anteroposterior axis of the developing nervous system (Arendt et al., 2008; Holland et al., 2013; Vermeren and Keynes, 2001). Importantly, the expression of these genes often relates to anteroposterior neuroanatomical subdivisions, and as such, gene expression data has been widely used to homologize nervous system compartments across bilaterian lineages (Albuixech-Crespo et al., 2017; Arendt et al., 2008, 2016; Arendt and Nübler-Jung, 1999; Hirth et al., 2003; Hunnekuhl and Akam, 2014; Lowe et al., 2003; Marlow et al., 2014; Pani et al., 2012; Range, 2014; Sinigaglia et al., 2013; Steinmetz et al., 2011; Tessmar-Raible et al., 2007; Tomer et al., 2010; Tosches and Arendt, 2013).

One of the most relevant, and still debated examples of this comparative approach affects the anteroposterior patterning of the vertebrate CNS and the evolution of a complex brain. Early in vertebrate embryogenesis, the vertebrate neural plate becomes molecularly patterned into the basic units that will form the CNS, namely the forebrain, midbrain, hindbrain, and spinal cord (Puelles, 2013; Vermeren and Keynes, 2001). As development proceeds, secondary organizers such as zona limitans intrathalamica (ZLI) and the isthmic organizer (IsO) subdivide and specialized these primary neural regions. These areas, subregions, and organizers exhibit a combination of developmental genes robustly conserved among vertebrate lineages (Fig. 5), supporting the homology of the vertebrate CNS neuroanatomy (Puelles and Ferran, 2012). Consequently, the analysis of these molecular signatures in non-vertebrate chordates (i.e. tunicates and amphioxus) and hemichordates, the closest chordate outgroup, have been used to trace the evolutionary origins of the vertebrate CNS (Albuixech-Crespo et al., 2017; Pani et al., 2012). The analysis of an extensive gene dataset in the 7-somite mid neurula embryo of the European species Branchiostoma lanceolatum has shown that the amphioxus neural plate is already anteroposteriorly patterned, and that these primary areas have a direct topological correspondence with vertebrate CNS compartments (Albuixech-Crespo et al., 2017). However, amphioxus lacks the secondary organizers ZLI and IsO (Shimeld and Holland, 2005), whose emergence in the stem lineage of vertebrates could explain how a primary amphioxus-like neural plate pattern evolved into the more elaborated vertebrate CNS. Strikingly, expression and functional data of ZLI and IsO markers in the hemichordate Saccoglossus kowalevskii, together with the conservation of genomic Cis-regulatory regions between hemichordates and vertebrates, has given compelling evidence that these signaling centers predated chordates and the evolution of their complex brain (Pani et al., 2012; Yao et al., 2016). In S. kowalevskii, however, the ZLI and IsO are involved in the anteroposterior ectodermal pattern and the subdivision of the primary body axis in proboscis, collar, and trunk (Pani et al., 2012). These findings thus propose that the vertebrate CNS is a much recent innovation, evolved from the cooption of preexisting ectodermal patterning gene regulatory modules that apparently got lost in pre-vertebrate chordate lineages (Lowe et al., 2015; Pani et al., 2012) (Fig. 5).

Fig. 5. The evolution of anteroposterior neuronal patterning in Deuterostomia. Schematic representation of the anteroposterior expression domains of genes involved in patterning the vertebrate CNS and defining the major brain signaling centers in brachiopods (outgroup), hemichordates, cephalochordates, and vertebrates. The similarities in expression patterns and functional interrelationships of these genes between hemichordates and vertebrates suggest that the anterior neural ridge (ANR), the zona limitans intrathalamica (ZLI) and the isthmic organizer (IsO) are conserved deuterostomian signaling centers involved in general ectodermal patterning. These organizers were partially lost in cephalochordates and urochordates, and coopoted into neuroectodermal/brain patterning in vertebrates. Drawings are not to scale. Question marks indicate unknown expression and red crossed text indicates that gene expression is not related to the ectoderm and/or the nervous system. See main text for references. Ap, apical lobe; C, collar; CV, cerebral vesicle; D, diencephalon; H, hindbrain; M, midbrain; Mt, mantle lobe; NT, neural tube; P, proboscis; Pd, pedicle lobe; T, telencephalon; Tr, trunk.

Long-distance evolutionary comparisons of gene expression data have also been used to homologize animal brains generally (Arendt et al., 2008, 2016; Hirth et al., 2003; Holland, 2015; Holland et al., 2013; Strausfeld and Hirth, 2013; Tosches and Arendt, 2013). Indeed, the anteroposterior neural domains of transcription factors such as otx, pax2/5/8, and Hox genes have been used to homologize the tripartite organization of the arthropod brain (protocerebrum, deuterocerebrum and tritocerebrum) with the forebrain, midbrain, and hindbrain of vertebrates (Hirth et al., 2003). Similarly, comparable antibody immunoreactivity and gene expression data suggested the shared ancestry of complex sensory-associated brain centers, as those observed in arthropods, vertebrates, annelids, platyhelminths, and nemerteans (Tomer et al., 2010; Wolff and Strausfeld, 2015). Generally, these hypotheses suffer from low, and sometimes biased, taxonomic sampling, and thus it still remains to be seen how the inclusion of bilaterian lineages that lack elaborated brain centers affects them. For instance, immunoreactivity against pCaMKII, a protein enriched in arthropod mushroom bodies, is broadly detected in the larval nervous system of priapulid worms, whose nervous system comprises a simple circumoral nerve ring and a single caudal bipolar neuron as ventral nerve cord (Martin-Duran et al., 2016). A similar case is seen in the free-swimming larvae of the brachiopods Terebratalia transversa and Novocrania anomala (Fig. 5), which have only poorly developed anterior condensations, associated to an apical tuft in T. transversa (Santagata, 2011). In these larvae, genes associated to brain signaling centers like the ZLI and IsO and neural regional patterning like Hox genes are also expressed staggered along the anteroposterior axis (Martín-Durán et al., 2016; Santagata et al., 2012; Schiemann et al., 2017; Vellutini and Hejnol, 2016), but demarcating ectodermal domains rather than different areas of the CNS. Furthermore, the expression of anteroposterior patterning genes such as six3/6, otx, foxQ2, and gsc in bilaterian and cnidarian larvae suggest that brain-related genes are also generally involved in (neuro)ectodermal patterning (Hejnol and Martindale, 2008; Hiebert and Maslakova, 2015; Marlow et al., 2014; Martín-Durán et al., 2015; Nederbragt et al., 2002; Sinigaglia et al., 2013; Steinmetz et al., 2011; Wollesen et al., 2015). Therefore, scenarios that favor a complex brain as an ancestral feature of Bilateria (Arendt et al., 2008, 2016; Hirth et al., 2003; Holland, 2015; Holland et al., 2013; Strausfeld and Hirth, 2013; Tosches and Arendt, 2013), and thus extensive simplification in most bilaterian lineages, are confronted with data arguing for a convergent, stepwise evolution of bilaterian complex CNS architectures (Pani et al., 2012).

1.4.2 Dorsoventral patterning

Bilaterian lineages such as arthropods and chordates share in addition the presence of a medially unpaired condensed nerve cord (Bullock and Horridge, 1965; Schmidt-Rhaesa, 2007) (Fig. 2). Whereas the nerve cord is located ventrally in arthropods, it is dorsally positioned in chordates, which led the French naturalist Geoffroy Saint-Hilaire to propose the homology between the arthropod ventral side and the chordate dorsal side already on 1822 (Geoffroy Saint-Hilaire, 1822). Almost two centuries later, the finding that similar genes and signaling pathways (e.g. BMP pathway) were expressed in the embryonic territories giving rise to the nerve cords in the fruitfly D. melanogaster and vertebrates revived those classic ideas, and in particular those involving a dorsoventral inversion of the CNS in the stem lineage of Chordata (Arendt and Nübler-Jung, 1994; De Robertis and Sasai, 1996). In this context, orthologous members of the NK2.1, NK2.2, NK6, pax6, pax3/7, and msx gene families are deployed in a similar fashion along the dorsoventral axis of the nerve cord in insects, vertebrates and the polychaete annelid P. dumerilii (Denes et al., 2007) (Fig. 6). In insects and vertebrates, the expression of these genes is required for the proper patterning of the nerve cord (Cornell and Ohlen, 2000), and their expression is associated with the location of distinct neuronal populations in these three bilaterian lineages (Arendt et al., 2008). Although the upstream regulators and downstream effectors of dorsoventral patterning genes differ between Drosophila and vertebrates (Cornell and Ohlen, 2000; Winterbottom et al., 2010), the striking similarities observed in the dorsoventral patterning of the nerve cord between insects, vertebrates, and the annelid P. dumerilii solidified the idea that a CNS with a medially condensed ventral nerve cord is ancestral for Bilateria (Arendt et al., 2008, 2016; Denes et al., 2007; Holland et al., 2013; Tosches and Arendt, 2013).

Fig. 6. The evolution of dorsoventral nerve cord patterning in Bilateria. The genes NK2.1, NK2.2, NK6, pax6, pax3/7, and msx exhibit a similar combinatorial expression along the dorsoventral axis of vertebrates, arthropods, and the annelid P. dumerilii. This combinatorial expression is associated with the molecular patterning of the medially condensed nerve cord of these three bilaterian lineages and has been argued to support the presence of a single ventral nerve cord in the last common ancestor of Protostomia and Deuterostomia. However, a similar dorsoventral expression of these genes is absent in hemichordates, nematodes, and many spiralian taxa, such as the nemertean L. ruber and the annelid O. fusiformis, which similar to P. dumerilii, also shows a medially condensed unpaired nerve cord. Similarly, cephalochordates and urochordates, which do have a single dorsal nerve cord, do not exhibit the dorsoventral nerve cord patterning of vertebrates. These data favors more parsimonious scenarios that propose that the similarities in molecular patterning between vertebrates, Drosophila and the annelid P. dumerilii evolved by convergence.

Investigations in hemichordates and nematodes have however challenged this scenario (Kaul-Strehlow et al., 2017; Lowe et al., 2003, 2006; Okkema et al., 1997) (Fig. 6). The neural anatomy of these two lineages differs considerably from that of vertebrates, insects, and annelids. Hemichordates have a diffuse nerve net throughout the body, with one dorsal and one ventral nerve cord running along their trunks (Bullock and Horridge, 1965). Nematodes, on the other hand, have a main ventral and dorsal nerve cord, and additional pairs of lateral neurite bundles (Bullock and Horridge, 1965). In the hemichordates S. kowalevskii and Balanoglossus misakiensis, the genes NK2.1, NK2.2, pax6 and msx do not exhibit a staggered dorsoventral arrangement, but their expression is either confined to endoderm (e.g. NK2.2) or to particular ectodermal areas along the anteroposterior axis (e.g. NK2.1, pax6, and msx) (Kaul-Strehlow et al., 2017; Lowe et al., 2003, 2006). Similarly, only pax6 and msx are expressed in connection to the nervous system in the nematode C. elegans, (Chisholm and Horvitz, 1995; Du and Chalfie, 2001). Therefore, it is unclear whether the different expression of dorsoventral patterning genes in hemichordates and nematodes are derived situations, perhaps related to their different neuroanatomies and life styles (Arendt, 2018; Denes et al., 2007), or suggest that the similarities between lineages with a single medial nerve cord evolved convergently (Lowe et al., 2006; Martin-Duran et al., 2018).

A recent study on the expression of dorsoventral nerve cord patterning genes in xenacoelomorph worms and representatives of four major spiralian lineages has shed new light into this debate (Martin-Duran et al., 2018). Character state reconstructions suggest that a diffuse nerve net and one pair of ventral nerve chords are the most likely ancestral neuroanatomies of Xenacoelomorpha and Spiralia respectively (Hejnol and Lowe, 2015; Hejnol and Pang, 2016). However, trunk neuroarchitecture vary widely within these two bilaterian lineages, with acoelomorph species showing independently evolved neural condensations and spiralian lineages like annelids exhibiting medially condensed unpaired nerve cords (Bullock and Horridge, 1965; Hejnol and Lowe, 2015; Schmidt-Rhaesa, 2007) (Fig. 2). In line with this morphological diversity, the expression of dorsoventral nerve cord patterning genes varies significantly among xenacoelomorpha and spiralian lineages (Buresi et al., 2016; Forsthoefel et al., 2012; Franke et al., 2015; Janssen, 2017; Mannini et al., 2008; Martin-Duran et al., 2018; Martín-Durán et al., 2016; Vellutini et al., 2017), even between closely related species like the annelids Owenia fusiformis and Platynereis dumerilii that share the presence of a medially condensed unpaired nerve cord (Denes et al., 2007; Martin-Duran et al., 2018) (Fig. 6). Remarkably, a similar case is found in Chordata, where non-vertebrate chordates, such as amphioxus and tunicates, differ from vertebrates in the arrangement of dorsoventral nerve cord patterning genes, yet all of them share the presence of a dorsal neural tube (Holland et al., 1998; Ristoratore et al., 1999; Stolfi and Levine, 2011) (Fig. 6). It appears thus clear that dorsoventral nerve cord patterning and trunk neuroanatomy has evolved independently in several animal lineages, and that the diversity of expression arrangements of dorsoventral patterning genes is more the norm than the exception in Bilateria, supporting the evolutionary scenario that poses the similarities in dorsoventral patterning between vertebrates, Drosophila and some annelids as a case of convergence (Lowe et al., 2006; Martin-Duran et al., 2018). However, a more thorough investigation of the relationship between the dorsoventral patterning genes and nerve cord architecture is needed, in particular regarding the actual function of these transcription factors in the development of the nerve cords in most of the bilaterian lineages studied to date.

A The Pervasive Nervous System—Serving and Protecting the Body

The two most striking general features of the nervous system are its widespread distribution and its oneness. These cardinal attributes allow the system to keep track of events everywhere in the body and to respond in a manner for its general good. To illustrate these simple but paramount features, let us engage for a moment in fantasy.

Imagine that someone you know is almost completely transparent, a ghostly person in whom only the nervous system is visible (Fig. 2-1). Such a flight of fancy may seem better suited to science fiction. But it is more than a daydream. It is an important step in understanding what the nervous system is and what it does. It makes us realize how important the system is in the design and operation of the body.

Fig. 2-1. An ethereal form approaches. Its human form is readily distinguished, even though only in neural outline. Except for a few regions of the body less densely infiltrated by neural threads, the shapes of body parts—the head, torso and limbs—may easily be seen. If the finest nerve filaments were visible, the identifying characteristics of brow, ears, nose, cheeks, lips and chin could be made out.

The central masses and myriad outreaching threads of the nervous system delineate the body, its parts and features. Over 100,000 miles of nerve fibers run through and trace out all our inner and outer parts. It approaches the ubiquity of the vascular system. Both systems are nearly omnipresent, as well as somewhat complementary in function. By means of neural signals and blood-borne substances, the two systems integrate the activities of body parts, protect them in numerous ways, enhance their performances in effort or exercise, promote their growth and maintain their healthy tone and vigor. Both systems look after the body's well-being.

The density of nerve supply varies from place to place according to the needs of body parts. The degree of innervation of a part tells us how much that part needs to send messages to the system. Thus in places liberally supplied with nerve endings (such as the finger tips or lips), stimuli can be especially painful. In other regions (that are largely neglected by nerve fibers), we don't mind an intrusion as much. We would probably rather sit on a tack than step on one!

The nervous system is completely successful in planning and delivering its service. All body regions receive exactly the nerve supply they need, no more, no less. No region is far from an available line over which to send a message, even though messages from that place are seldom sent. And a message may be sent at any time; the lines are always open, the switchboard ever on call.

B Oneness of the Nervous System—Unifying the Body

We have seen the widespread and effective distribution of the nervous system. Let us now return to the phantom view and consider its oneness. We see that it is like a tree. Its trunk comprises the intricate, graceful mass of the brain and the slender, tubular spinal cord (Fig. 2-1). The important thing to remember about this tree is its continuity, the stringing together of all parts that allows each part to be in touch with every other. And you might pause to think (in case you had not already thought of it), “This tree inside me —this is where I live”.

The brain and spinal cord represent the central nervous system (CNS). Extending to either side of this neural tree trunk are numerous pairs of branches, the cranial and spinal nerves from brain and cord, respectively. These nerves, some thick and others thin, divide again and again into smaller and smaller branches. Eventually, they ramify into thousands of fine branchlets, ultimately into millions of terminal twigs and tips too small to be seen with the naked eye (and hence suggested by shading). This tremendous, far-flung system of branches is the peripheral nervous system(PNS). It reaches out everywhere, to the eyes, ears, lips, tongue, torso and limbs, fingers and toes—to the body's periphery.

The nervous system provides total connectivity; it offers bodywide communication. A message entering the PNS from any body part travels centrally over that tree of ever-combining, constantly enlarging and frequently interlacing twigs, branchlets and branches until it reaches the trunk—the brain or spinal cord, depending on where the message came from. The nervous system does not provide any direct “hookups” over which, let us say, the ring finger and right big toe could hold a private chat. Every message must go to the CNS. Once inside, the message is received and “studied” there. New signals are then generated and distributed to other places in the brain and cord. Some go to points nearby through local circuits, others to remote locations over long-distance lines. Certain of these places (notably the cerebral cortex) scrutinize the message further, others elaborate appropriate responses. Ultimately, messages are sent back to body muscles. Thus, while extensive deliberation accompanies every transaction, each part of the body, through the nervous system, can affect the activity of the entire body or of any other part of it. One world, one body!

C Flexibility of the Nervous System—Appropriate and Considered Responses

Great flexibility is noted in the components which handle signals. We know that if we make separate calls in succession from the same telephone, we will probably not get the same operator. Even when placing many calls in a row, we may never get a particular operator more than once. Which operator answers at any moment depends on how busy the switchboard is at that time and who is available to handle the call. Messages to the CNS from sensory neurons may be handled by various central neurons standing by to receive signals from a given body part. A message is processed by a team of neurons, never by a single one. The particular cluster of neurons responding would depend upon the activity of the entire nervous system at that time, as well as upon the activity of its individual cells from one moment to the next. Sometimes, certain cells are “tied up” and other cells nearby must serve in their place. At other times, cells may be under instructions from supervisory cells “higher up” to handle only certain types of messages, perhaps only urgent ones. Occasionally, if the body is in an emergency state of activity or a life-threatening situation, messages (even important ones) may not be studied at all, or responses may be deferred.

There are limitations, of course, in such flexibility. Not every neuron can handle every call. Cells that receive signals from the ear, for example, do not handle messages from the eye. But within any such sensory or motor subsystem, myriad numbers of neurons are receiving and sorting out signals from one moment to the next. At some level of data analysis, each subsystem reviews its messages on a priority basis that is changing constantly over a wide range. Stimuli that are pleasurable at certain times may become irritating at others, and vice versa. The entire nervous system, meanwhile, continually scans the reports of its many subsystems.

Such latitude of information processing by nerve cells and their connections shows that more is involved than a multiplicity of elements and circuits. Fixed circuits are employed flexibly. The nervous system offers flexibility in which systems and components come into play. This attribute allows us to function appropriately and intelligently in an ever-changing environment.

D Speed of Operation of the Nervous System—Instant Service

The nervous system couples flexibility with speed in a way that has no match. Speed and accuracy a computer has, but sometimes it can be anything but flexible! The old expression “as quick as thought” is a tribute to the lightning fast integrative properties of single nerve cells. Thoughts are quick, reflexes instantaneous. Responses are as fast as they need be. We may respond to a stimulus instantly, a short time later or perhaps only after a long period. Just when our response comes (if it comes at all) depends partly on the nature of the stimulus, as well as on the activity of our nervous system at that time or later. But it also depends on such factors as our training and experience, as well as on our personal, societal and genetic backgrounds.

E Reliability of the Nervous System—A Precise but Delicate Instrument

The reliability of our nervous system, like its flexibility, speed and predictability of response, again must be evaluated in terms of its individual cellular elements and of the entire system.

Nerve cells seldom make mistakes. The cellular specializations, cytoplasmic machinery and electrochemical events that they use to receive signals and generate responses are dependable. Furthermore, there are billions of neurons in the CNS and innumerable routes (including detours) over which they can send their messages. Thus, a loss of substantial portions of a population of neurons may not lead to serious disruption of service. What can happen is that residual neurons and alternative routes of communication are called into service. This service may not be as good as before, but at least it is better than nothing. The principle of redundancy provides reliability.

The nervous system is not always completely reliable. It does not always achieve the “foolproof” quality built into its respective components. Sometimes signals are inaccurate or distorted, messages are lost or misrouted and responses are inadequate or misdirected. Even poor decisions are made. The consequences of mistakes are often minor—a bit of clumsiness here, a wrong number there. The brain may become tired or disordered from overwork, lack of sleep, illness, want of essential nutrient or chemical substances or a variety of other causes. But even in these cases it performs admirably. It can ill afford to bungle emergency messages pertaining to the well-being of the body. It is part of the body itself and thus inseparable from the customers it serves. We see that our nervous system is more than a proficient integrator and regulator of the body's activities. It is a scholar and historian; it has highly developed capabilities for learning and memory. It is a wise planner; it weighs demands on the basis of need and in the framework of experience and constraints. It does more than merely react to and regulate events. Often, it initiates them. The nervous system thus not only “reserves the right to change its services and products at any time” but also “builds for the future.” It renews and remodels many of the parts of its cells. The brain indeed has a “mind of its own!”

F Summary of Cardinal Attributes of the Nervous System—Taking Stock

In a general way, we now see some of the major attributes of the nervous system. The PNS conveys signals from the environment (external and internal) to the CNS. There, with great flexibility, speed and reliability these signals are integrated and evaluated in relation to current and past events. A decision is made, a memory formed or a response initiated. The nervous system is much like a giant corporation: various jobs are assigned to experts, special cells in special structural relationships. It has many subdivisions—particular, highly individualized structures.

We shall turn now to an overview of the larger divisions of the human brain and from there to an analysis of functions of the parts of its marvelous whole.