Basic features of the ancestral chordate brain: a protochordate perspective.

Cross references:  Amphioxus PDFs     Chordates   Hemichordates  
Protochordates      Cephalochordate  
Amphioxus   Amphioxus Evolution   
Amphioxus Nervous System   


    Basic features of the anterior nerve cord in amphioxus larvae are summarized to highlight its essential similarity with the vertebrate brain. Except
for a pineal homolog, the amphioxus brain consists of a much simplified version of the ventral brainstem, including a region probably homologous
with the hypothalamus, and a locomotory control center roughly comparable to the vertebrate tegmentum and reticulospinal system. Amphioxus
has direct pathways for activating its locomotory circuits in response to mechanical stimuli via epithelial sensory cells, but this response is evidently
modulated by inputs from diverse sensory-type cells located in the putative hypothalamic homolog, and from the lamellar body, the pineal homolog.
This implies that a basic function of the amphioxus brain is to switch between locomotory activities, of which there are several, and the principal
non-locomotory one, namely feeding. A similar involvement in switching between behavioral modes may thus have been a core brain function in
ancestral chordates. Currently, however, incomplete knowledge of the physiology and behavior of amphioxus limits how effectively it can be used
as an evolutionary model. Eye evolution is briefly discussed to illustrate how a better understanding of living forms can inform the evolutionary
debate. An account of recent data on dorsoventral inversion is also included, as this bears directly on the question of where the chordate brain
originated in relation to other structures. It now appears likely that key components of the ancestral brain were originally located around the
mouth. A secondary repositioning of the latter would therefore have been required before a unitary brain could be assembled and internalized. This
association between the mouth and the evolving brain reinforces the idea of a fundamental early connection between core brain structures and the
control of feeding activity. 

1. Introduction

    When speculating about the events of early chordate evolution,
one needs to beware of taking any living protochordate too
literally as a model for the proximate invertebrate ancestor of
vertebrates. A close analysis of living organisms is certainly a
key source of information about the course of past evolution,
but all are modified from the ancestral form to a greater or lesser
degree. The problem, of course, is to identify which organisms,
and which characteristics, provide the most reliable guide to
the ancestral condition.    
    There are two groups of organisms to
consider here, cephalochordates (amphioxus) and urochordates
(tunicates). Of these, tunicates have evidently experienced much
more rapid genomic change, and the peculiarities of their life
history and morphology are considered to be secondary specializations
[6,9]. At the molecular level, even core features of
the developmental control program have been modified in tunicates,
probably due to the evolution of lineage-dependent control
mechanisms [6]. Thus, while fascinating subjects for studies of
the changes that can occur with evolution, tunicates are of somewhat
limited value as models of the ancestral form. In contrast,
cephalochordates appear to be less modified in both genetic and
morphological terms, and hence are potentially more directly
    When it comes to the evolution of the central nervous system
(CNS), there is a further problem with tunicates. This relates
to the extreme reduction of the nervous system in some forms
(larvaceans), and the alterations to it that have accompanied the
evolution of a sedentary habit in others (ascidians).   The life history
of ascidians imposes a nearly complete separation between
locomotion and feeding. During the larval phase the animal
swims but does not feed, while the reverse is true for the adult.
    The larval CNS is designed to perform a limited repertoire of
sensory and locomotory functions, but has no visceral control
functions. The adult has only the latter, the larval system having
degenerated at metamorphosis. There is thus no need for neural
circuitry to switch between locomotion and non-locomotory
    Amphioxus, in contrast, swims and feeds throughout
its post-embryonic life just as vertebrates do, and a key function
of the anterior brain-like region of its CNS[see below] appears to
be to modulate these activities and switch between them. There
is thus some justification for expecting the brains of amphioxus
and vertebrates to share basic organizational features, even to
the level of circuitry, so long as there has been at least a degree
of continuity through time in the core tasks they are required to
perform. Here I provide a brief summary of what is known of the
circuitry and organization of the brain-like region of the anterior
nerve cord of young amphioxus larvae, which, from molecular
data, is regionally subdivided in a comparable fashion to the
vertebrate brain. There follows a discussion of two related subjects.
    First, using the visual system as an example, I highlight
the need for more research on all aspects of amphioxus biology
as a necessary step towards understanding ancestral chordates as
fully functioning organisms.  
    Second, I illustrate how in-depth comparative analysis
can provide an unexpected resolution of
even the most puzzling evolutionary problems. The example
here comes from recent gene expression studies in hemichordates,
which are clarifying the role that dorsoventral inversion
has played in chordate evolution. A consequence of inversion
is that the mouth must shift its position before core components
of the brain can assemble into a unitary structure, and this
may, in itself, tell us something useful about the nature of those

2. The larval brain of amphioxus: organization and

    Past studies of the adult CNS in amphioxus have provided a
catalog of neuronal cell types, but the circuitry responsible for
observed behaviors has so far been examined in detail only in
young larvae. The latter are sufficiently small that their CNS
can be examined and reconstructed at the EM-level, a task that
occupied my own lab for much of the past 15 years [ref. [17]
and citations therein]. Of special interest is the anterior-most
ca. 90mu of the nerve cord (Fig. 1), comprising the cerebral
vesicle and the region just caudal to it to about the end of
myotome 1. This corresponds with the zone expressing genes
that are markers for vertebrate forebrain and midbrain, notably
Otx [5].

Fig. 1. 

Fig. 1. Organization of the anterior nerve cord in a 12-day amphioxus larva, seen here in lateral view, anterior to the left, and extending to just beyond the junction
between myotomes 1 and 2. The figure shows roughly 120 microns of tissue containing, in total, ca. 150 neurons at this stage. The infundibular cells (inf, dark
shading) mark the transition between the largely sensory anterior part of the cerebral vesicle and the posterior part, occupied dorsally by the lamellar body (light
shading, including cells within it) and ventrally by the post-infundibular neuropile. Sensory input, from both peripheral and centrally located sensory neurons enters
the cord via both the rostral and anterodorsal nerves and from caudal sources (not shown). Somewhat schematic; see refs. [7,17] for further details.

A ventral cluster of secretory infundibular cells serves
as a useful landmark at the transition between the anterior and
posterior parts of the cerebral vesicle. Immediately behind the
infundibular cells is a region of ventral neuropile, and behind
this is the primary motor center (PMC). There are qualitative
differences between the neurons in the anterior-most part of
the nerve cord to the level of the post-infundibular neuropile,
and those further back. The former, whether in the frontal eye
(pigment cells, photoreceptors and some neurons) or the
preinfundibular and infundibular region (sensory-type cells) have
neurites that mainly project to the neuropile, where they form
irregular varicosities. The varicosities are typically large, lack
synapses, and contain mixed populations of the vesicles, including
a large proportion of dense-cored vesicles. Cells of the
dorsal lamellar body, a putative pineal homolog, also contribute
to the neruropile, and their terminals also lack synapses. The
absence of synapses implies that the anterior neurons act largely
through paracrine release and, while the transmitters have not
been identified, the morphology indicates they likely include a
variety of neuropeptides. As one moves caudally into the primary
motor center, the most conspicuous neurons are sets of
large interneurons and somatic motoneurons with descending
projections. Among the former are three pairs of especially large
cells that, from their pattern of projections and synapses, function
as a pacemaker. The presynaptic terminals in this region are
of conventional appearance, and clear vesicles predominate. The
locomotory circuits thus probably depend on fast transmission
involving acetylcholine and various amino acid transmitters, a
conclusion supported by preliminary pharmacological studies.
    Dendrites from the pacemaker cells and some of the other
PMC neurons project forward to the post-infundibular neuropile.
There they receive a highly redundant input, via well-defined
synapses, from sensory axons entering via the rostral and
anterodorsal nerves and from fibers originating in the tail. The
input in young larvae appears to be predominantly if not entirely
mechanosensory, and larvae respond very strongly to touch by
initiating several types of escape response. They are also capable
of more prolonged periods of slow swimming, which may,
in nature, drive the diurnal vertical migrations observed in some
species. There is, in summary, ample morphological evidence for
the circuits needed to drive most of the locomotory behaviors
observed in larvae, though many details have yet to be worked
    The principal non-locomotory activity larvae engage in is
feeding, which takes place while they are suspended vertically
and more-or-less motionless at the water surface [16]. Swimming,
whether for escape or migration, is thus incompatible
with feeding. One can argue that, to optimize chances of survival,
there is an advantage to modulating locomotory behavior
depending on the available food supply and nutritional state of
the animal. In other words, when it is starved or surrounded by
abundant food, the best strategy might be to suppress locomotion
in order to feed with minimal interruption. While the synaptic
connections identified by EM do not indicate how this is done,
dendrites from the locomotory control cells do ramify within the
post-infundibular neuropile, which itself receives inputs from a
variety of sensory-type cells located dorsally and further forward
that could, in principle, act to modulate the locomotory
response. Some have cilia modified in ways that suggest they
monitor physical displacement, i.e. to act as a balance organ.
In most cases, however, function is not evident from the morphology.
Instead, because the cells reside in a region that, from
molecular data, is probably homologous with the hypothalamus,
one is left to speculate on their probable function based on
what hypothalamic neurons are known to do. The hypothalamus
in vertebrates monitors various physical and chemical stimuli,
including temperature, ionic balance, and concentrations of a
wide range of specific metabolites, and is involved in setting levels
of alertness, activity and appetite. As a working hypothesis,
it is reasonable to suppose that the closest amphioxus counterpart
of this brain region might do much the same by playing a
key role in modulating levels of activity in response to changes
in both external environment and internal conditions, including
the nutritional state of the animal.
    The importance of paracrine transmission to circuits originating
from the putative amphioxus homolog of the hypothalamus
also accords with a concept developed by Nieuwenhuys [14,15]
concerning the primitive paracrine (or limbic) core of the vertebrate
brain. The latter consists of a series of core brainstem
systems organized in a loose, reticular fashion and dependent
on paracrine release of a range of neuropeptides. The main
hypothalamic centers involved in homeostasis and metabolic
control are included among these, and the system as a whole
is largely ventral in location. Since it is predominantly counterparts
of ventral brainstem structures that are represented in
amphioxus (Fig. 2), it will be very interesting to see what further
research reveals these cells to be doing.    

Fig. 2.   

Fig. 2.  
    The current consensus, based on combined anatomical and molecular
data, for how the anterior nerve cord in amphioxus larvae maps to the vertebrate
brain. Most of the former finds its closest counterparts and probable homologs
in the ventral region (variously shaded) of the latter, with the exception of the
lamellar body, a pineal homolog. The Otx-positive zone in amphioxus, marking
the putative forebrain plus hindbrain homologs, extends through the cerebral
vesicle (c.v.) and primary motor center to a level somewhere in the front half of
myotome 2. Caudal to this point, the nerve cord becomes more hindbrain-like
in character, but the exact point of the transition is uncertain (indicated by ?)
(from ref. [17]).   

    Genomic analysis will provide a more complete
account of the neuropeptides and peptide
receptors used by amphioxus than is currently available,
and eventually these will be localized within the CNS. In the
author’s view, there is a reasonable prospect that many of the
basic homeostatic control systems that operate in vertebrates
will prove to be ancient and fundamental to CNS organization,
traceable to the protochordate level of organization at least, if
not earlier.   

3. Amphioxus photoreception: multiple receptors for
diverse needs

    We are used to thinking of the vertebrate visual system as consisting
first and foremost of paired, image-forming eyes, with at
most a few accessory structures to perform related functions,
notably the pineal. Explaining how this situation arose in evolution
requires some knowledge of the ancestral structures from
which the eyes evolved, and of how they were used.  
    Amphioxus has four known photoreceptor systems,
two of which are ciliary and two rhabdomeric [8].
The former have vertebrate homologs while the
latter evidently do not. The two ciliary receptors are
the frontal eye, which is almost certainly the homolog and precursor
of the paired eyes, and the pineal-like lamellar body. The
two differ structurally in amphioxus in ways that reflect their different
    Photoreceptor cells in the lamellar body have massive
arrays of lamellae to maximize light absorption. The
lamellar body is well developed in larvae but less so in the
adult, reflecting its probable role monitoring light levels during
vertical migration in the water column, which is primarily a
larval activity. At depth, there will be a reduction in light intensity,
so maximizing light absorption makes sense.  
    In contrast, the photoreceptors in the frontal eye in amphioxus are
simple unspecialized cells. There is, in fact, nothing morphological to
suggest they are photoreceptors except their close association
with the pigment cup, beside which they are ranged in rows that
form essentially a one-dimensional array. Behavioral observations
[16] indicate the frontal eye acts as a shadow detector to
orient the animal while suspended at the water surface, i.e. in
conditions of bright light.   
    How then might an image-forming eye evolve starting from
something like the amphioxus frontal eye? Since the image formation
for a grazing animal is chiefly concerned with identifying
approaching predators, any existing shadow detector could be
co-opted and improved for this purpose. Visual predators in
the Cambrian would have required light for hunting prey, so a
shadow detector used during daylight hours is the obvious choice
for an eye, which lets the lamellar body out on two counts: it is
non-directional and is probably too light-sensitive to operate at
the surface in daylight.    With regard to the sensitivity-enhancing
specializations of vertebrate rods and cones, it is clear these must
have come later, since it is only as the receptors are packed in
two-dimensional arrays that efficiency of photon capture again
becomes an issue.  
    While we are still a long way from a complete
story of eye evolution, my intent here is to illustrate how
an understanding of amphioxus behavior clarifies some aspects
of the earliest events of vertebrate eye evolution, including the
issue of why increasing predation produced a burst of adaptive
change affecting the frontal eye while leaving the lamellar body
essentially unchanged.   

4. Dorsoventral inversion and the implied linkage
between brain and mouth  

    For much of the past century, comparative zoologists designated
the ventral surface of the body as the one bearing the
mouth, and assumed this to be the correct orientation for comparing
animal body plans. Starting in the mid-1990s, new data
on the genes specifying the dorsoventral body axis reawakened
an interest in a hypothesis with roots in the 19th century, that
chordates are dorsoventrally inverted relative to non-chordates,
notably arthropods [1,2]. If true, then the position of the nerve
cord (ventral in arthropods, dorsal in vertebrates) is a more reliable
indicator of body orientation than the mouth (ventral in
both). This raises a further problem, however, for if the body
is inverted in chordates so that the mouth becomes dorsal, the
mouth has then to migrate or be otherwise shifted to what was
the dorsal surface and is now the ventral one. Recent work on
hemichordates goes a long way towards resolving this issue.
    Hemichordates are basal deuterostomes with a bilaterian body
plan, so they are exactly what is needed to bridge the gap between
protostomes and chordates. The available molecular data now
indicate strongly that hemichordates share the dorsoventral orientation
of protostomes [12], so it is chordates alone that seem to
be inverted relative to both non-chordate deuterostomes and protostomes.
    Further, and quite remarkably, the genes involved in
axial patterning are expressed in a series of circumferential bands
in hemichordates, providing a precise map of position along
the longitudinal body axis for comparison with other taxa [4].   

Fig. 3   

Fig. 3. Consequences of dorsoventral inversion for the site of origin and composition
of the chordate brain. (A) Embryo of an enteropneust hemichordate,
a close relative of basal chordates that has an epithelial nervous system. Circumferential
bands of gene expression mark axial position with considerable
precision in ways the closely mirror vertebrates. The expression zone corresponding
with the vertebrate mid-hindbrain boundary (MHB) lies behind the
mouth (m) as shown by the shaded box; shows also anterior pole marked an
apical plate (ap) which, though poorly defined in direct-developing species, is
evident in those with larvae. In the inverted ancestral chordate (B) the ancestral
(a-) dorsal surface is now on the underside of the body and the mouth is
on top. If axial markers are to be believed, however, the boxed region will still
be post-oral. The epithelial zone from which the anterior nervous tissue most
probably arises in chordates is shown as a U-shaped domain (shaded) embracing
the mouth and positioned such that the MHB and midbrain structures would lie
within the boxed post-oral domain. More anterior regions, including some or all
of the forebrain, would likely come from zones lateral to or even forward of the
mouth. However, in order for these structures to coalesce into a unitary brain,
the mouth must be moved out of the way. This evidently happened by having
the mouth shift laterally down the left side of the head to reach the new ventral
surface (see text), as in fact it does today during amphioxus development. 

    Notably, the genes expressed at the level of the mid-hindbrain
boundary in vertebrates are post-oral in hemichordates (Fig. 3),
so that part of the brain is evidently a post-oral structure in vertebrates
as well [10]. The source of more anterior brain structures
is not yet clear, but it is reasonable to suppose that at least some
could derive from sites further forward, along the sides of the
mouth, as shown in Fig. 3, and in front of it. It may be significant
that similarly positioned adoral neural centers are key 

components of the larval nervous system in echinoderms and
    There is a problem for any evolutionary scheme that derives
the mouth from an epithelial domain that also forms a significant
part of the CNS. Since a functional mouth requires an external
opening, the components of the CNS that surround it cannot
coalesce to form a unitary, internalized structure until the mouth
is moved out of the way. This implies an ancestral condition in
which the brain was not at first an internalized structure. There
are then two main options for moving the mouth. One, suggested
by Nielsen [13], is to have it move forward over the anterior end
of the animal. This requires that preoral structures like the apical
organ be displaced to a ventral position, below the mouth,
which forever breaks their connection with the brain. An alternative
is for the mouth to move laterally over the side of the head.
Preoral and apical structures could then retain some connection
with the evolving brain. Interestingly, among the structures in
question is the preoral ciliary organ of hemichordates, a putative
homolog of the adenohypophysis [3]. The latter retains associations
with both the mouth and brain throughout the vertebrates,
which could reflect a more ancient condition when its precursors
where in contact with both, at a time before the mouth had
shifted its position.
    Evidence that the mouth may have shifted laterally, down the
side of the head, comes from two sources. First, the hemichordate
data show that suppression of the dorsalizing signal allows
the mouth to expand laterally along both sides of the head until,
in some cases, the proboscis is completely separated from the
body [12]. This implies that there is a restricted circumferential
zone of tissue within which the dorsoventral extent of the
mouth is specified. Its position could in principle, be respecified
by altering relevant patterns of gene expression so as to
shift the orientation of the dorsoventral axis. This would move
the mouth down one side of the head, like a tram on a track,
and away from the dorsal midline, without altering its position
along the longitudinal axis. A second, less direct form of evidence
comes from an analysis of pharyngeal development in
amphioxus larvae [11]. The initial asymmetry of the larval head
is extreme, and to restore symmetry requires a process of repatterning
and differential growth that is exceedingly complex
so long as the mouth is assumed to be part of the ventral patterning
system. If it is not, i.e. if it originates dorsally, re-patterning
is much less of a problem. If this interpretation is correct, then
the initial appearance of the mouth on the left side of the larval
head of amphioxus, and its subsequent movement to the
ventral midline, is in part a recapitulation of a past evolutionary
event. A further consequence of inversion, relevant to the
circuitry analysis presented above, is that, by requiring a close
connection between the mouth and evolving CNS, it is consistent
with and, indeed, strongly reinforces the idea of a deep
connection between core centers in the brain in chordates and
the control of feeding-related activities. 


Supported by NSERC Canada. I thank Nick Holland and
Chris Lowe for insights and discussion, and the organizers of
this symposium. 


[1] D. Arendt, K. Nubler-Jung, Inversion of dorsoventral axis? Nature 371
(1994) 26. 

[2] D. Arendt, K. Nubler-Jung, Comparison of early nerve cord development
in insects and vertebrates, Development 126 (1999) 2309–2325. 

[3] S. Candiani, N.D. Holland, D. Oliveri, M. Parodi, M. Pestarino, Expression
of the amphioxus Pit-1 gene (AmphiPOU1F1/Pit-1) exclusively in the
developing preoral organ, a putative homolog of the vertebrate adenohypophysis,
Brain Res. Bull. 75 (2008) 324–330. 

[4] J. Gerhart, C. Lowe, M. Kirschner, Hemichordates and the origin of chordates,
Curr. Opin. Gen. Dev. 15 (2005) 461–467. 

[5] L.Z. Holland, N.D. Holland, Chordate origins of the vertebrate central
nervous system, Curr. Opin. Neurobiol. 9 (1999) 596–602. 

[6] L.Z. Holland, A chordate with a difference, Nature 447 (2007) 153–

[7] T.C. Lacalli, Sensory pathways in amphioxus larvae I. Constituent fibres
of the rostral and anterodorsal nerves, their targets and evolutionary significance,
Acta Zool. 83 (2002) 149–166. 

[8] T.C. Lacalli, Sensory systems in amphioxus: a window on the ancestral
chordate condition, Brain Behav. Evol. 64 (2004) 148–162. 

[9] T.C. Lacalli, Protochordate body plan and the evolutionary role of larvae:
old controversies resolved? Can. J. Zool. 83 (2005) 216–224. 

[10] T.C. Lacalli, Perspective protochordate homologs of vertebrate midbrain
and MHB, with some thoughts on MHB origins, Int. J. Biol. Sci. 2 (2006)

[11] T.C. Lacalli, Mucus secretion and transport in amphioxus larvae: organization
and ultrastructure of the food trapping system, and implications for
head evolution, Acta Zool., in press. 

[12] C.J. Lowe, M. Terasaki, M. Wu, R.M. Freeman, L. Runft, K. Kwan, S.
Haigo, J. Aronowicz, E. Lander, C. Gruber,M. Smith, M. Kirschner, J. Gerhart,
Dorsoventral patterning in hemichordates: insights into early chordate
evolution, PLoS Biol. 4 (2006) 1603–1619. 

[13] C. Nielsen, Origin of the chordate central nervous system and the origin of
chordates, Dev. Gen. Evol. 209 (1999) 198–205. 

[14] R. Nieuwenhuys, Chemoarchitecture of the Brain, Springer, Berlin,1985. 

[15] R. Nieuwenhuys, The greater limbic system, the emotional motor system
and the brain, Prog. Brain Res. 107 (1996) 551–580. 

[16] M.D. Stokes, N.D. Holland, Ciliary hovering in larval lancelets, Biol. Bull.
188 (1995) 231–233. 

[17] H. Wicht, T.C. Lacalli, The nervous system of amphioxus: structure,
development, and evolutionary significance, Can. J. Zool. 83 (2005) 122–