Introduction

The phylogenetic relationships of Tardigrada, a group of small metazoans with many unique and autapomorphic features, have always been controversially discussed (see Richters and Krumbach 1928; Giribet et al. 1996; Ax 1999; Budd 2001; Nielsen 2001; Kristensen 2003; Jenner and Scholtz 2005). Some authors favour close affinities to the nematodes and their allies whereas other investigations indicate that tardigrades belong to the arthropods. The latter view is currently more generally accepted. However, whether Tardigrada is the sister group of Euarthropoda, of Onychophora or of Onychophora plus Euarthropoda remains ambiguous in recent studies. The advent of the Ecdysozoa hypothesis (Aguinaldo et al. 1997) put new oil into the fire. Because this view allows to interpret the tardigrade organisation as a kind of intermediate between that of nemathelminths/cycloneuralians and of arthropods (Schmidt-Rhaesa et al. 1998; Giribet 2003).

The analysis of the nervous system of Tardigrada has plays an important role in the discussion of the phylogenetic affinities of the group (e.g. Hanström 1928; Marcus 1929; Kristensen and Higgins 1984a, b; Dewel and Dewel 1996, 1997; Nielsen 2001). Previous light microscopic investigations of the tardigrade nervous system distinguish a dorsal cerebral ganglion (supraoesophageal ganglion) joined by circumoesophageal connectives to a suboesophageal ganglion and a ladder-type chain of four paired ventral trunk ganglia (Doyère 1840; Greeff 1865; Plate 1889; Basse 1905; Thulin 1928; Marcus 1929; Greven 1980, Ramazzotti and Maucci 1983; Wiederhöft and Greven 1996; Dewel and Dewel 1997; Dewel et al. 1999). Together with the analysis of the nervous system, cephalic sense organs (Walz 1978; Wiederhöft and Greven 1996) have been described and based on these data a tripartite organisation of the tardigrade brain has been suggested by several authors (e.g. Kristensen and Higgins 1984a, b; Dewel and Dewel 1996; Nielsen 2001). This tripartite/three-segmented organisation has been related to the three sections of the euarthropod brain, namely the proto-, deuto-, and tritocerebrum (see Scholtz and Edgecombe 2006) by some authors (Kristensen and Higgins 1984a, b; Nielsen 2001). However, the various interpretations of tardigrade brains as being tripartite are very ambiguous, i.e. different regions and structures have been interpreted as corresponding to the three parts of the euarthropod brain (compare Kristensen and Higgins 1984a, b; Dewel and Dewel 1996; Nielsen 2001).

The ventral nerve cord of arthropods and annelids is characterised by a rope-ladder like organisation (Scholtz 2002; Müller 2006). The paired segmental ganglia are connected intrasegmentally by transverse commissures and intersegmentally by longitudinal connectives. These essential elements of a ladder-like central nervous system have also been described for Tardigrada and are found as general textbook knowledge (Kristensen 1982; Brusca and Brusca 2002; Ruhberg 2004).

In this study the structure of the nervous system of Macrobiotus hufelandi C.A.S Schultze, 1833 is described for the first time by means of modern morphological techniques such as fluorescence microscopy, confocal-laser-scanning-microscopy, and computer-aided 3D reconstruction. Some general aspects of the central nervous system of M. hufelandi correspond with older descriptions (Marcus 1929). However, our results reveal new details concerning the internal organisation of the brain, the connection of the brain to the ventral chain of ganglia, the question of existence of a suboesophageal ganglion, and the fact that no commissures exist between the ganglia of the ventral nerve cord. Furthermore, it is proposed that ganglia and neuropils of the tardigrade brain form a complex structure but that there is no indication for a tripartite/three-segmented brain comparable to that of euarthropods.

Materials and methods

Adult specimens of Macrobiotus hufelandi were extracted from moss collected from a roof in Keitum (Sylt), Germany and maintained at 6°C. The moss pads were regularly moistened with rainwater. For light microscopy unstained specimens were stretched maximally by removing oxygen (see Marcus 1929).

Fixation and permeabilization

For nuclei staining and whole mount antibody staining, the specimens were fixed in 3.7% paraformaldehyde in phosphate-buffered saline (1× PBS; 1.86 mM NaH2PO4, 8.41 mM Na2HPO4, 175 mM NaCl, pH 7.2) for 15–20 min at room temperature. The animals were washed in 1× PBS several times (three times for 5 min, four times for 30 min). In addition, a further fixation was applied using 2.5% glutardialdehyde in 1× PBS for 25–30 min followed by three 5-min and four 30-min washes in 1× PBS. The fixation in glutardialdehyde has proved to be more efficient for the antibody staining of the entire body, whereas the staining efficiency of formaldehyde fixed animals was higher either in the anterior or posterior part of the body. Before labelling, specimens were stretched by being placed in distilled water for 30 min at room temperature. For a better penetration of dyes, fixed specimens were sonicated for 60–75 s in an Elma ultrasonic bath model Transsonic 310 operating at 35 kHz.

Nuclear staining

For staining with the DNA-selective fluorescent dye Hoechst (Bisbenzimide, H33258) specimens were transferred into Hoechst solution (concentration 100 μg/ml Bisbenzimide in 1× PBS) and incubated for 15 min at room temperature. Unbound dye was removed by washing three times for 10 min in 1× PBS. Specimens were mounted on glass slides in DABCO-glycerol (2.5 mg/ml DABCO (1.4 diazobicyclo-[2.2.2.]-octane, Merk) in 90% glycerol–PBS).

Immunocytochemistry

Immunolabelling of microtubules, such as axonal bundles of neurons, was obtained with a monoclonal anti-acetylated α-tubulin antibody. For immunohistochemical processing, the fixed and sonicated specimens were first given three 10 min washes and four washes of 30 min each in PBT (1× PBS, 0.2% Bovine Serum Albumin, 0.1% Triton X-100) followed by two 30-min washes in PBT + N (5% normal goat serum). A monoclonal anti-acetylated α-tubulin antibody, clone 6-11B-1, Sigma diluted at 1/100 in PBT + N was applied over night at 4°C. The samples were next rinsed several times (three times for 5 min and four times for 30 min) in PBT followed by two 30-min washes in PBT + N. Next, tardigrades were incubated in goat anti-mouse serum (Sigma) conjugated to Cy3 over night at 4°C. This secondary antibody was diluted 1:200 in PBT + N. After rinsing frequently in PBS (three times for 5 min and six times for 30 min) specimens were counterstained with fluorescent dye Hoechst (Bisbenzimide, H33258) and mounted in DABCO-glycerol as described above. As negative controls, specimens of M. hufelandi were processed without incubation in the primary antibody. Such specimens exhibited no detectable fluorescence beside the autofluorescence of the cuticle, claws and buccal tube as visible in the documentations. For the applied phalloidin labelling, tardigrades were washed three times for 5 min and two times for 30 min in PBS after fixation and sonication. Next, samples were washed 1 h in PBT followed by incubation in 2 μl phalloidin [TRITC-labelled] (0.1 mg phalloidin/1 ml DMSO) in 1 ml PBS for one hr at room temperature. Unbound dye was removed by washing several times (three times for 5 min, four times for 30 min) in PBS. The mounting occurred as described above. Mounted specimens processed for Hoechst, α-tubulin, and phalloidin immunoreactivity were viewed either with a Zeiss Axioplan 2 fluorescence microscope equipped with a Zeiss digital camera (Axio Cam HRC) or a Leica SP2 confocal laser-scanning microscope. Fluorescence microscopic pictures were edited using Photoshop CS (Adobe). Images emerged from Image stacks obtained by scanning microscopy were processed and edited using the 3-D reconstruction software IMARIS (Bitplane AG, Zürich). All confocal images are based on stacks of between 50 and 170 optical sections of a z-series taken at intervals of 0.2–0.4 μm.

Discrimination between nerve cells and other cell types

To discriminate between nerve cells and muscle cells we applied a phalloidin staining for visualisation of the musculature especially in the head region (Fig. 1f, g). Two longitudinal muscle tracts dominate the dorsal side, in addition to the muscles of the buccal apparatus. A more complex structure of transverse running muscles as well as longitudinal ones is located ventral to the buccal tube. For a detailed description of the musculature system of tardigrades as resolved by phalloidin staining see Schmidt-Rhaesa and Kulessa (2007). The double labelling of anti-α-tubulin with Hoechst and the comparison with the phalloidin staining reveals that most of the nuclei dorsal to the buccal tube belong to nerve cells as well as the nuclei in the region of the ventral ganglia. Glia cells are largely ignored in our investigation in reference to the research of Greven and Kuhlmann (1972), according to which the nerve tissue of M. hufelandi possesses only very few glia cells.

Fig. 1
figure 1

Anti-acetylated α-tubulin immunoreactivity in the CNS of the tardigrade Macrobiotus hufelandi complemented with nuclear staining. Anterior is up. a Double labelling with the DNA-selective fluorescent dye Hoechst (blue) and the anti-acetylated α-tubulin antibody (red). Ventral view showing the segmentally repeated four ganglia (gI–gIV). The frame indicates a higher magnification of the second ganglion pair (one optical section). Nuclei of smaller ganglia (arrow in the fourth segment) and clawglands (cgl) can be seen in the posteriormost part of the legs (arrowhead). Claws (c), cuticle, buccal tube and placoids are autofluorescent. b Anti-acetylated α-tubulin labelling from (a). Fibre bundles dorsal to the buccal tube are highly reactive for anti-α-tubulin (arrow). The first ventral ganglion pair is connected to the outer lobes of the brain by the outer connective (oco). The four ventral ganglia are linked to each other by paired connectives (co) along the longitudinal axis. Claws, cuticle, buccal tube and placoids are autofluorescent. c Dorsal view of the brain (Hoechst), inverted confocal image. The nuclei of the dorsal brain part are arranged in three cell clusters, the anterior cluster (acl), the dorsal cluster (dcl), and the posterior cluster (pcl). Two bands of nerve cell nuclei link the clusters (arrowhead). Some nuclei are arranged circum-buccal (star). The position of the eye is indicated by arrowhead; first ganglion pair (gI). The buccal tube is autofluorescent. d Ventral view of the cerebral ganglion indicates a smaller amount of nuclei ventral to the cerebral ganglion (Hoechst). e Double labelling with Hoechst (blue) and anti-acetylated α-tubulin (red), single stack. Anterior to the first ganglion (gI) nuclei form the ventral cluster (dotted circle). A paired neuron was identified within the ventral cluster (arrowhead). f Phalloidin staining to mark the body musculature. Dorsal view shows two longitudinal muscles (arrowhead). g Phalloidin staining, ventral view. Scale bar a, b 50 μm; c, d 20 μm; e, f 30 μm; g 20 μm (see supplementary material movie 1)

Full size image

Results

The brain

The brain of M. hufelandi is characterized by a complex meshwork of nerve cell clusters and fibre bundles of various sizes (Fig. 1b, c). It forms a full circle around the buccal tube (Figs. 1, 2, 3). Anti-α-tubulin immunoreactivity and Hoechst labelling confirm the typical lobate appearance of the tardigrade brain which has been described by Marcus (1929). There are two pairs of caudally directed lobes, the outer dorsolateral lobes which bear the eyes (arrow in Fig. 1c) and the more ventral inner lobes (Figs. 1c, 2e). The largest part of the brain consists of three paired bilateral-symmetrically arranged neuronal cell clusters (Fig. 1c) and their corresponding plexus which are located dorsal and lateroventral to the buccal tube forming a saddle-like structure (Fig. 3b). Traditionally this part is called the supraoesophageal ganglion. In addition there is a small unpaired ventral cell group forming a mixture of muscles and nerve cells, the ventral cluster (Fig. 1e). Some of the numerous fibre bundles form large transverse tracts like the dorsal commissure, preoral commissure, and postoral commissure which connect the lateral halves of the brain (Figs. 2, 3).

Fig. 2
figure 2

Anti-acetylated α-tubulin immunoreactivity in the head represented in a series of single scans from dorsal (a) to ventral (f). Anterior is up. a Fibre bundles form a median plexus. A median tract (arrow) is associated with a small ganglion (star). Fibres curve between the median tract and fibres of the posterior cluster (arrowhead). Fibres of the dorsal cluster (dcl) expand dorsally. b The dorsal commissure (dcm) connects the fibres of the posterior clusters (inner lobes) (pcl). c The nerve 14 (n14) connects the small ganglion (see Fig. 2a) and the inner lobes (the posterior cluster) via two lateral branches (arrowheads). d Focus is still dorsal to the buccal tube. The preoral commissure (prcm) links the fibre bundles of the posterior clusters (inner lobes). Fibres of the posterior clusters (see e) emerge and project in posteroventral direction forming the paired circumbuccal connectives (cco). e Fibres of the cerebral ganglion are morphologically arranged as outer lobes (ol) and inner lobes (il). The circumbuccal connectives extends to the posterior margin of the inner lobe. f Focus is ventral to the buccal tube. Fibre bundles of the circumbuccal connectives extend to the midline forming the postoral commissure (pocm). Further fibre bundles of the circumbuccal connectives run in posterior direction representing the inner connectives (ico). Fibres of the anterior clusters extend ventrally (arrow). Scale bar a 20 μm applies to images b–f

Full size image
Fig. 3
figure 3

Anti-acetylated α-tubulin immunoreactivity in the head. Anterior is up unless indicated. a Dorsal view of fibre bundles located in the head. The small image shows the double labelling with Hoechst (blue) in the same anterior (a) posterior (p) orientation. The dorsal commissure (dcm) ranges between fibres of the posterior clusters (pcl). Fibres of the median tract (arrow) project anteriorly. They are described as nerve 13 (n13) which terminates in the median area of the forehead. Fibres of the dorsal commissure proceed in posterior direction, bend anteriorly and extend to the anterior margin of the head representing nerve 4 (n4). Multiple fibres described as nerve 5 (n5) link fibres of the anterior (acl) and posterior cluster (pcl). Nerve 6 (n6) emerges from nerve 5 (n5) and projects to fibres of the dorsal cluster (dcl) where it branches brush-like. 45° rotation of the image in anterior direction leads to the projection in figure (b). The frame indicates the position of (c). b View from behind showing the course of the dorsal commissure (dcm) and the preoral commissure (prcm). The small image shows the double labelling with Hoechst (blue) in the same dorsal (d) ventral (v) orientation. Position of the buccal tube is indicated by an asterisk. c Dorsolateral view Nerves 7–10 (n7, n8, n9, n10) emerge from the plexus of the anterior cluster and project anteriorly to the forehead. Scale bar a, b 20 μm, c 10 μm (see supplementary material movie 2)

Full size image

The anteriormost of three three pairs of dorso- and ventrolateral nerve cell arrangements are the anterior cell clusters which are connected by a band of nerve cells dorsal to the buccal tube (Fig. 1c). The most central paired nerve cell groups are represented by the posterior cell clusters which again are joined by a transverse dorsal cell band and which form the inner lobes of the brain (Figs. 1c, 2e, 3a). The posterior- and dorsalmost brain neurons are arranged in the dorsal clusters which form the outer lobes (Figs. 1c, 2e, 3a, b). Double labelling with anti-α-tubulin and Hoechst shows a distinct arrangement of fibre bundles in the regions corresponding to the cell clusters (Fig. 3). Fibres of the posterior clusters are located more central, fibres of the anterior clusters extend ventrolaterally to the buccal tube (arrow in Fig. 2f). A serial of single stacks of anti-α-tubulin immunoreactivity (Fig. 2a–f) reveals the complex pattern of fibres located in the head dorsal, dorsolateral, lateral, and lateroventral to the buccal tube. The dorsalmost part of the brain consists of a median plexus which gives rise to a strongly labelled longitudinal tract linked to a small ganglion (Figs. 2b, 3a) consisting of five to eight cells (data not shown). Various fibres connect the small ganglion with fibres of the posterior neuropils (Fig. 2b).The median and ventral part of the brain is composed of fibres laterally arranged to the buccal tube (Figs. 2d, e). Compared to the amount of nerve cell nuclei located dorsal to the buccal tube, there are much less nuclei on the ventral side (Fig. 1d, e).

The few neurons of the ventral cluster form, together with the postoesophageal commissure, the ventral part of the circumbuccal brain (Fig. 1e). The ventral cluster comprises in total around 25–35 nuclei. The muscle staining with phalloidin shows quite a complex structure of transverse and longitudinal tracts ventral to the buccal tube (Fig. 1g). Hence it is likely that most of the nuclei ventral to the buccal tube belong to muscle cells. We identified at least the somata of one pair of neurons within the ventral cluster (Fig. 1e) which are linked to the circumbuccal connectives at the transition to the postoral commissure (Fig. 4a, c). These neurons send their axons in anterior direction towards the mouth contributing to nerve 11. More medial, the postoral commissure gives rise to a pair of axonal fibre bundles that project anteriorly towards the mouth characterized as nerve 12 (Figs. 4a, c). The ventral cluster corresponds to the structure that has traditionally been named suboesophageal ganglion. However, it does not contain a large amount of fibres and it is not connected via connectives to the first ventral ganglion (see below).

Fig. 4
figure 4

Anti-acetylated α-tubulin immunoreactivity in the head region. Anterior is up. a Ventral view, projection indicates the level of fibres from ventral (v) to dorsal to (d) by colour code. The inner connective (ico) extends between fibre tracts of the inner lobe and the connective of the ventral ganglia chain (co). The outer connective (oco) leads between fibre tracts of the outer lobe (ol) and peripheral fibres of the first ganglion (gI). Very faint fluorescent signal marks the outer connectives (arrowhead). The postoral commissure (pocm) is located ventral to the buccal tube and sends nerves 11 and 12 (n11, n12) in anterior direction towards the mouth opening. Nerve 1 (n1) is associated with the outer connective projecting into the first leg. b Ventrolateral view. The inner connective (ico) arises from the circumbuccal connective (cco) and shows a dorsal-ventral course. The outer connective (oco) is linked to the outer lobe (ol). c Scheme of (a) and (b). Abbreviations: cco circumbuccal connective; co ventral connective; gI first ganglion; ico inner connective; n1 nerve 1; n11-n12 nerves 11 and 12; oco outer connective; pocm postoral commissure. Scale bar a 25 μm (see supplementary material movie 3)

Full size image

Dorsal commissure, preoral commissure, circumbuccal ring, and postoral commissure

The brain contains a voluminous transverse tract, the dorsal commissure. The dorsal commissure connects the fibre regions of the posterior cell clusters (inner lobes) (Figs. 2b, 3a, b). In addition to the dorsal commissure, a second more delicate fibre bundle, the preoral commissure is found dorsal to the buccal tube (Figs. 2d, 3b). It lies ventral to the dorsal commissure and connects also the inner lobes (Fig. 3b). Fibres deriving from the preoral commissure run anteriorly to the mouth cone and posteriorly to the stylets muscles. Paired fibres of the dorsal commissure project posteroventrally and cross the inner lobes of the brain (Figs. 2d, e). These fibre tracts pass the buccal tube laterally while taking course posteroventrally. They form the circumbuccal (circumoesophageal) connectives which are the lateral parts of the circumbuccal nerve ring (Figs. 2d, e, 4b). Each circumbuccal connective branches into two delicate fibre bundles ventrolateral to the buccal tube (Fig. 2f). A transverse fibre tract beneath the buccal tube forms the postoral commissure, the ventral part of the circumbuccal ring (Figs. 2f, 4a). The postoral commissure connects the inner lobes via the circumbuccal connectives. Recently obtained data of anti-tubulin immunoreactivity in 72 h old embryos have revealed a circumbuccal nerve ring that is ventrally linked to paired connectives extending to the posterior end (data not shown). Paired fibre tracts, the inner connectives, proceed from the circumbuccal connective and project posteroventrally to the first pair of trunk ganglia (Figs. 2f, 4b). They continue into the connectives of the ventral cord. These are thus directly connected to the inner lobes (posterior cluster) of the brain by the circumbuccal connectives. In addition to the inner connectives, the outer lobes of the brain give rise to the outer connectives which merge into the paired plexus of the first ganglia (Fig. 4a, b). A delicate nerve n1 connects the outer connective and the first leg (Fig. 4a). In summary, the brain is joined with the ventral chain of ganglia by two paired connectives, (1) the inner connectives originating in the inner brain lobes (posterior cluster) and (2) the outer connectives originating from the outer brain lobes (dorsal cluster) (Fig. 4b).

Head nerves

Marcus (1929) listed and numbered the large head and trunk nerves of tardigrades. Here we follow his account with some modifications based on our observations. Nerves n1 to n3 name the paired nerves of ventral segmental ganglia (see below) (Fig. 5a–g). Nerves n4 to n14 are brain nerves (Figs. 2, 3, 4). Except for the median nerves n13 and n14 all other nerves are paired.

  • Nerve 4 (n4): originates from cells of the posterior cluster near the dorsal commissure. It projects posteriorly, curves anteriorly, and then branches multiple times and merges into the plexus of the anterior cluster (Fig. 3a).

  • Nerve 5 (n5): exits the posterior cluster in anterolateral direction. The nerve splits into two branches of which one passes to the dorsal cluster (see nerve 6); the other branch splits into delicate fibres merging into the anterior cluster (Fig. 3a).

  • Nerve 6 (n6): arises from nerve n5, projects dorsally to the plexus of the dorsal cluster branching in a brush-like manner near the epidermis (Fig. 3a).

  • Nerves 7–10 (n7–n10): exit the plexus of the anterior cluster and terminate flanking the epidermis forming a homogenous field (Fig. 3c). Nerves n7 and n8 are located centrally, nerves n9 and n10 are arranged laterally.

  • Nerve 11 (n11): originates from the postoral commissure, extends horizontally in anterior direction innervating the mouth cone ventral to the buccal tube (Fig. 4a).

  • Nerve 12 (n12): runs parallel and median to n11 (Fig. 4a).

  • Nerve 13 (n13): is a median nerve emerging from the fibres of the dorsomedian tract (arrow in Fig. 2a). It passes the dorsal commissure leading in anterior direction in parallel to the buccal tube to the epidermis of the forehead (Fig. 3a).

  • Nerve 14 (n14): is a median nerve which connects the inner lobes and the small dorsomedian ganglion (Fig. 2a) It starts from the inner lobes (posterior clusters) as paired fine fibre tracts which unite to one nerve which passes the dorsal commissure ventrally and leads in posterodorsal direction to the dorsomedian ganglion (arrowhead in Fig. 2c).

Fig. 5
figure 5

Anti-acetylated α-tubulin immunoreactivity in the peripheral nervous system. Ventral view anterior is up. a–d Peripheral nerves of the first three body segments. e–g Peripheral nerves of the fourth body segment. a The first ganglion (gI) sends three segmental nerves (n1, n2, n3). Rostral fibres branch representing the nerves 1a (n1a) and nerve 1b (n1b). Nerve 1a extends dorsally and merges to the outer lobe, it forms the outer connective (oco). Nerve 1b projects to the first leg. The central nerve 2 branches and runs to the periphery of the first leg. The caudal nerve 3 (n3) innervates the first leg. Fibres linked to the rostral nerve of the first and second ganglion are highly reactive for anti-α-tubulin (arrowheads). b Higher magnification of the second ganglion in (c). No transverse fibre bundles can be seen which are reactive for anti-α-tubulin. c Second ganglion (gII) with peripheral nerves. Nerve 1 (n1) extends to the dorsal musculature. Nerves 2 and 3 (n2, n3) innervate the second leg. d Scheme of the segmental nerves 1a, 1b, 2, 3 (n1a, n1b, n2, n3) and outer connective (oco) of the ganglia one to three (gI–gIII). e Inverted confocal image. Fibres of the connectives (co) extend posteriorly across the fourth ganglion (gIV) representing nerve 2 (n2) which gives rise to nerve 3 (n3). Nerve 1 (n1) is located near the anus and linked to fibres which are high reactive for anti-α-tubulin (arrowheads). f Inverted confocal image. Double labelling with Hoechst (blue). Nerve 3 (n3) joins the fourth ganglion (gIV) with a small ganglion (arrowhead) showing fibre bundles reactive for anti-α-tubulin. Clawglands (clg) are indicated. g Scheme of the segmental nerves 1, 2, 3 (n1, n2, n3) of the fourth ganglion (gIV) with corresponding neuropils (arrowhead). Scale bar a 25μm; c 20μm, e, f 25 μm

Full size image

The ventral ganglia

The chain of ventral ganglia comprises four paired ganglia which are connected by paired connectives along the anterior-posterior axis (Fig. 1a). The segmental ganglia are arranged slightly anterior to the corresponding segmental appendages (Fig. 1a). Nuclei of the four ganglia show a bilateral-symmetrical arrangement along the longitudinal axis comprising also some unpaired nuclei. A higher magnification of the nuclei of ganglion two is shown in Fig. 1a (inset). The number of nuclei in each ganglion differs within one animal and between individuals. Counting the nuclei of ten individuals revealed that the first ganglion possesses the largest number of nuclei (between 50 and 75 nuclei). The second and the third ganglion comprise averagely 28% less nuclei than the first ganglion (between 45 and 50 nuclei). In comparison to the first ganglion the fourth shows 55% less nerve cells (15–30 nuclei). The ventral connectives contain various fibre bundles which do not form a homogenous tract (Fig. 5b). The ganglia are clearly separated from the connectives which are exclusively fibrous and lack perikarya of nerve cells (Fig. 1a). Transverse commissures connecting the lateral halves of segmental ganglia could not be detected (Fig. 5b). Each plexus within one ganglion is separated from the other. Even in the third and fourth ganglion where the connectives lie close to each other we could not identify transverse tracts.

Peripheral nervous system of the trunk

Each of the segmental ganglia gives rise to three pairs of segmental nerves extending laterally to the periphery (Fig. 5d). The segmental nerves have a common origin within the plexus of the ganglia. The rostral nerve n1 extends dorsally, the central nerve n2 and the caudal nerve n3 lie along the ventral body side terminating in the segmental limbs (Fig. 5c). Fibres in the distalmost part of each limb, which are linked to the central and caudal nerves, are highly reactive for anti-α tubulin (Fig. 5a, e). These fibre bundles represent small ganglia of unclear function (Fig. 5f). The corresponding nerve cell perikarya are found in close proximity to the cells of the clawglands in legs one to three only in the fourth limb pair they are spatially dissociated (Fig. 1a). The nerves of the first ganglion show a modified course (Fig. 5a). Rostral fibres give rise to two tracts of which nerve n1a projects dorsally and merges into the outer lobes of the brain and nerve n1b innervates the first limb (Fig. 5a). Nerve n1a forms the outer connective. A delicate fibre bundle arises from the central nerve n2 of the first ganglion terminating in the first limb (Fig. 5a). Also, the peripheral nerves of the fourth ganglion show a modified arrangement (Fig. 5e–g). The ventral connectives extend posteriorly and give rise to two nerves travelling in caudal direction forming a small plexus with corresponding nuclei (Fig. 5e, f). The paired distal nerve is likely to be homologous to the central nerve n2 of the ganglia one to three because of its course to the margin of the fourth limb. The paired proximal nerve is likely to be homologous to the caudal nerve n3. A paired nerve originates from the central part of the fourth ganglion leading in posterior direction innervating the anus (Fig. 5e). Due to the fact that this nerve does not innervate the limb like the rostral nerves of ganglia one to three, we assume that the central nerve of the fourth ganglion is homologous to the rostral nerves of ganglia one to three.

Discussion

Modern morphological techniques shed new light on the brain anatomy of tardigrades – all brain commissures are associated with the inner lobes

The general organisation of the nervous system of Macrobiotus hufelandi has been described in detail by Marcus (1929). With the modern morphological techniques applied by us we basically confirm many of Marcus’ results. However, in addition our investigation led to new data concerning the shape and the internal organisation of the supraoesophageal ganglion, the position of the brain, and the connection between the brain and the ventral nerve cord.

We identified the typical lobate appearance of the brain comprising two outer dorsolateral lobes extending posteriorly and two smaller inner lobes which lie in a more ventral position. We could not detect the third unpaired posterior lobe which has been described by Marcus (1929). The dorsolateral and inner lobes correspond largely to defined nerve fibre regions and their grouped cell somata which are concentrated dorso- and ventrolateral to the buccal tube. Furthermore, we identified several tracts and fibre bundles in the brain lobes which have not been described before. However, despite the compartmental organisation and the occurrence of well defined massive tracts and a complex pattern of fibre arrangements we could not identify brain centres that correspond to neuropils such as the corpora pedunculata (mushroom bodies), olfactory lobes, or a central body as are typical for arthropod brains (see Loesel 2005). This result is in agreement with previous studies of the brain of M. hufelandi using transmission electron microscopy (Greven and Kuhlmann 1972).

Three transverse fibre bundles, the dorsal commissure, the preoral commissure, and the postoral commissure, connect the left and the right brain hemispheres. All three transverse tracts are associated with the two (inner) lobes of the posterior cluster which form the most central areas of fibres in the brain. Furthermore, the commissures are clearly defined along a dorsoventral axis. The dorsal commissure forms the strongest tract and has also been described by Dewel and Dewel (1996) in Heterotardigrada. The preoral commissure is the second transverse tract lying dorsal to the buccal tube. Nerves run from the preoral commissure in anterior and posterior directions towards the mouth and the oesophagus. These nerves are assumed to be involved in the innervation of the stomatogastric nervous system. The postoral commissure, which also sends nerves into the mouth region, forms the circumbuccal ring together with the anterior sections of the circumbuccal connectives (Fig. 6). A clearly defined connection between the preoral and the postoral commissure was not detected. According to Dewel and Dewel (1996) four commissures exist in the brain of Echiniscus viridissimus Peterfi, 1956, the dorsal commissure, the stomatogastric nerve system which consists of a dorsal and a ventral commissure encasing the buccal tube, and a single suboesophageal commissure. However, more recent investigations of Dewel et al. (1999) describe only one postoral brain commissure in tardigrades.

Cephalic nerves and sense organs: cirri and clavae are lost within eutardigrades but corresponding sensory areas remain

Arthrotardigrada among the Heterotardigrada possess eleven head appendages with apparent sensory function including the median cirrus (see Kristensen and Higgins 1984a, b). This condition has been considered to be ancestral within Tardigrada whereas the appendage-less condition in Eutardigrada seems to be derived (Marcus 1929; Ramazzotti and Maucci 1983). M. hufelandi as a representative of the Eutardigrada does not show any cephalic appendages. However, already Walz (1978) identified various sense regions in the head of this species. According to Thulin (1928) M. hufelandi possesses 19 nerves in the head. In contrast to this we identified 10 paired and 2 unpaired nerves (in total 22) which are associated with the innervation of specific sense regions in the head.

The branching and attachment of nerves to distinct epidermal regions found by us supports the studies of Walz (1978) and Wiederhöft and Greven (1996) on the characterisation of an anterolateral and posterolateral sense region in Eutardigrada. Moreover, the present investigation reveals that each cephalic nerve or defined nerve region in M. hufelandi corresponds to a cephalic sense organ of Heterotardigrada (see Table 1). The identified nerve n13 in M. hufelandi is attached to the epidermis in a similar position to the median cirrus in Arthrotardigrada. The median cirrus of the arthrotardigrade Styraconyx nanoqsunguak Kristensen and Higgins 1984 is innervated by a small neuropil near the centre of the brain (Kristensen and Higgins 1984b). In M. hufelandi the unpaired nerve n13 is correspondingly connected to a small triangular ganglion. The course of the paired nerve n4 which runs into the area under the epidermis of the forehead correlates with the position of the internal cirri in Heterotardigrada. Further correspondences exist between nerves n7–n10 of the fibres of the anterior cluster in M. hufelandi and the external cirri and secondary clavae in Heterotardigrada. The position of the lateral cirri and the primary clavae in Arthrotardigrada fit with the location of the nerve n6 and fibres with the same course in M. hufelandi. In summary, it is obvious that the sensory regions of the eutardigrade head correspond to the cephalic sensory appendages of heterotardigrades such as cirri and clavae. If the heterotardigrade situation is considered as plesiomorphic then, with the notable exception of the members of the Milnesiidae, the Eutardigrada have apomorphically reduced their head appendages. In spite of this reduction, the arrangement of distinct nerves in the head is maintained and the areas of the appendages became transformed to cuticular sensory fields. This scenario is supported by recent molecular and morphological phylogenetic studies (Garey et al. 1999; Jørgensen and Kristensen 2004; Nichols et al. 2006). Although a sister group relationship between Heterotardigrada and Eutardigrada is suggested, Milnesiidae is the sister taxon to the remaining eutardigrades (Nichols et al. 2006).

Table 1 Correspondence between the regions of cephalic appendages in Heterotardigrada and the cephalic nerves in Macrobiotus hufelandi
Full size table

The connection between the brain and the ventral nerve cord: there is no suboesophageal ganglion

A ventral suboesophageal ganglion connected by a circumoesophageal ring to the dorsal supraoesophageal ganglion and by paired connectives to the first ventral pair of ganglia is described as a general feature of the central nervous system of Tardigrada (Doyère 1840; Plate 1889; Hanström 1928; Thulin 1928; Marcus 1929; Kristensen and Higgins 1984a, b; Kinchin 1994; Dewel and Dewel 1997; Dewel et al. 1999; Nielsen 2001). More specifically, Dewel et al. (1999) identify neuropils concentrated ventrolateral to the buccal tube which form a ventral commissure and which are suggested to represent the suboesophageal ganglion of the brain.

In contrast to these descriptions, we found that nerve cell nuclei and nerve fibres in the head of M. hufelandi form a circumbuccal ring whereby the saddle-like supraoesophageal ganglion extends lateroventrally to the buccal tube and is associated with a ventral group of nerve cell nuclei embedded in muscle cells, which we call the ventral cluster. We can show that in the domain of the ventral cluster there are no massive fibres or neuropil-like structures ventrolateral or ventromedial to the buccal tube. The only structures ventral to the buccal tube that show reactivity for anti-α-tubulin are a few nerve cells in the ventral cluster, the postoral commissure and two pairs of nerves (nerves n11 and n12). Hence, the presence of nerve cell nuclei in a ventral position and the postoral commissure do not indicate the existence of a proper suboesophageal ganglion in M. hufelandi.

According to our investigations the connectives of the ventral chain of ganglia continue to the brain in the form of the inner connectives which are connected with the circumbuccal connectives lateral to the buccal tube (Fig. 6). No nerves exist between the first ventral ganglion and the ventral cluster. In addition, we found a connection between the outer lobes of the brain and the first ventral ganglion formed by a pair of connectives (outer connectives) (Fig. 6) (see also Marcus 1929; Dewel and Dewel 1996). However, in contrast to the older descriptions (see Marcus 1929, fig. 23) we identified these outer connectives as being formed by a side-branch of the rostral nerve (n1) of the first ventral ganglion. Since the rostral nerves of all ventral ganglia lead to the dorsal side, the connection between the first ganglion and the outer lobes of the brain formed by the outer connective might be a secondary feature due to the posterior extension of the outer lobes. This disputes the scenario of Dewel and Dewel (1996) about the evolution of head segmentation in Tardigrada.

Fig. 6
figure 6

Simplified drawing of the CNS and peripheral nerves of the head and the first trunk segment. a Dorsal view. The circumbuccal ring is formed by the dorsal commissure (dcm), the circumbuccal connectives (cco), and the postoral commissure (pocm). The cephalic nerves innervating sensory fields in the head are marked by arrowheads. The circumbuccal ring is linked to the first ganglion (gI) via the inner connectives (ico) in the area of the inner lobes (il). The outer connectives (oco) join the outer lobes (ol) and the first ganglion. The preoral commissure (prcm) lies ventral to the dorsal commissure connecting fibres of the posterior clusters. Nerves visible are: segmental nerves of the chain of ventral ganglia (n1–n3), head nerves 4–13 (n4–n13). Anterior cluster (acl), dorsal cluster (dcl), posterior cluster (pcl), eyes (e), small limb ganglion (arrow). b Lateral view. Nervous structures are the same as in (a). The position of the ventral cluster is indicated by an arrowhead

Full size image

In summary, even though nerve cell nuclei exist ventral to the oesophagus in the head of M. hufelandi they do not form an autonomous ganglion with neuropil. Rather they represent a ventral association of the supraoesophageal ganglion. This is supported by embryological investigations. Hejnol and Schnabel (2005), who described the development of the “suboesophageal ganglion” by growing out of the supraoesophageal ganglion whereas a neuronal progenitor cell, as in the four ventral ganglia, is not detectable for the suboesophageal region.

Tardigrade head segmentation – questioning the tripartite/three-segmented brain

As with euarthropod (see Scholtz and Edgecombe 2006) and onychophoran heads (Strausfeld et al. 2006), the structures and possible segmentation of the tardigrade head have been the subject of highly controversial interpretations (Plate 1889; Hanström 1928; Marcus 1929; Kristensen 1982; Kristensen and Higgins 1984a, b; Dewel and Dewel 1996; Wiederhöft and Greven 1996; Nielsen 2001). Some authors interpret the brain as consisting of only one unit corresponding to the protocerebrum of arthropods (Richters and Krumbach 1926; Hanström 1928; Dewel and Dewel 1996) whereas others consider the lobate brain organisation as indicative either for a bipartite (Plate 1889) or a tripartite brain comprising a protocerebrum, deutocerebrum and tritocerebrum homologous to the brain parts of euarthropods (Kristensen and Higgins 1984a, b; Nielsen 2001). Plate (1889) interprets the inner lobes as the protocerebrum and the outer lobes as the deutocerebrum—although this view implies the unique association of the eyes with the second head segment. According to this author, a tritocerebrum is not yet present as a brain part in tardigrades. The hypothesis of a tripartite (three-segmented) brain is mainly based on the three-lobed structure of the brain (as described by these authors, but see above) and the connections of theses lobes to the cephalic sense organs such as eyes, cirri, and clavae. However, the interpretation of what structures represent the protocerebrum, deutocerebrum, and tritocerebrum differs between the authors. The large dorsolateral lobes which innervate the primary clavae, the lateral cirri, and the eyes (if present) are interpreted as the protocerebrum by (Kristensen 1982; Kristensen and Higgins 1984a, b; Nielsen 2001). The deutocerebrum, however, which innervates the internal cirri, is described by Nielsen (2001) as a more ventral, rather compact structure, whereas Kristensen and Higgins (1984a, b) characterize it as paired smaller dorsal lobes. Nielsen (2001) identifies a circumoesophageal tritocerebrum which has a postoral commissure, while Kristensen and Higgins (1984a, b) described the tritocerebrum as paired lateral lobes that innervate the secondary clavae and the external cirri. Interestingly enough, Dewel and Dewel (1996, 1997) also suggest a tripartite brain based on similar evidence as Kristensen and Higgins (1984a, b) and Nielsen (2001). However, Dewel and Dewel (1996, 1997) conclude that the tardigrade brain corresponds only to the protocerebrum of euarthropods. Nevertheless these authors list a three-segmented brain as a putative synapomorphy for tardigrades and (eu)arthropods (Dewel and Dewel 1997: 114). These apparent contradictions in the interpretation of the segmental status of the tardigrade brain indicate that the evidence for the claims of a metameric composition of the brain is much too weak.

Our observations on M. hufelandi confirm that there are distinct fibre and cell somata regions in the brain which partly correspond to the lobes already described in the 19th century. However, these structures alone cannot be used as evidence for a segmental composition of the brain since several compartmental areas such as neuropils, cell clusters, lobes, tracts etc. are described for single euarthropod brain neuromeres such as the protocerebrum or deutocerebrum (e.g. Hanström 1928; Sandeman et al. 1993; Harzsch 2006; Strausfeld et al. 2006). Whether the serial arrangement of cephalic cirri and clavae is indicative for the fusion of several segments is also problematic since the occurrence of a median cephalic cirrus suggests that, despite the fact that in the trunk cirri are often segmentally arranged, cirri occur independent of segmental affinities. The total lack of limb-like head appendages allows no further conclusions either. If we assume that the dorsal cluster represented the protocerebrum, the posterior cluster the deutocerebrum and the anterior cluster the tritocerebrum, then all major transverse tracts connecting the brain hemispheres were found in the hypothetical deutocerebrum and none in the hypothetical protocerebrum or tritocerebrum. Furthermore, the hypothetical tritocerebrum had no connection to the ventral nerve cord and the major tracts connecting the three brain parts do not allow putting the hypothetical proto-, deuto-, and tritocerebrum in a proper anteroposterior sequence (even if, as is occurring in many euarthropods, an upfolded anterior brain area is considered). All this would be a very atypical pattern for an arthropod tripartite brain (see Scholtz and Edgecombe 2005, 2006; Strausfeld et al. 2006).

In summary, we think that the case for a tardigrade head and brain formed by several metameric units is not convincing. On the contrary, our brain anatomy data indicate a compact, unsegmented brain. The compartmentalisation of brain areas and the concentration of sensory structures might simply reflect functional specialisations rather than the fusion of segments. The view of an ametameric brain is consistent with the results of older and more recent embryological studies. Already Marcus (1929) reports that the head mesoderm is formed by a single pair of coelomic pouches as is the case with the segmental mesoderm in the trunk. However, the strongest evidence comes from the analysis of the expression of the segmentation genes Engrailed and Pax3/7 in embryos of the eutardigrade species Hypsibius dujardini (Doyère 1840) (Gabriel and Goldstein 2007). Both genes show a clear segmental pattern with one major expression domain in the brain region. Pax3/7 is expressed in the segmental ganglia of the trunk and in the brain anlage forming a ring around the stomodaeum. Engrailed expression occurs in pairs of v-shaped stripes in the posterior region of each trunk segment, as a pair of semicircular stripes at the posterior margin of the forming brain, and in some cells in the anterior brain area (Gabriel and Goldstein 2007). In particular, the Engrailed expression in the head region of H. dujardini resembles the ocular/protocerebral Engrailed expression of euarthropods (see Scholtz and Edgecombe 2005: Fig. 2) suggesting that the tardigrade brain might correspond to the anteriormost brain part in euarthropods, the protocerebrum (see also Dewel and Dewel 1996). The tardigrade brain, however, forms a complete circumoesophageal ring composed of fibres and nerve cell perikarya. This stands in contrast to the protocerebrum of euarthropods which is considered to be a purely praeoesophageal neuromere, and the posterior boundary of the euarthropod circumoesophageal nerve ring is thought to be composed of deutocerebral and tritocerebral commissures (Harzsch 2004a; Scholtz and Edgecombe 2006).

The ventral ganglia: three segmental nerves and the absence of segmental commissures

To our great surprise we could not detect segmental commissures in the four ganglia of the ventral nerve cord in M. hufelandi. Nevertheless, it is obvious from the cell arrangement, the lateral swellings, and the three segmental nerves that the ganglia are bilaterally formed. We found in some cases nerve fibres projecting towards the midline in the ganglia but they never form a complete connection between the two segmental ganglionic plexus. To our knowledge the absence of segmental commissures has not been reported before in tardigrades. Of course, the lack of commissures could be due to the method, i.e. we simply did not detect the fibres. However, we consider this as being unlikely given that with the techniques applied we found very fine and detailed fibre structures in the brain and in the peripheral nervous system. Accordingly, it appears that M. hufelandi does not exhibit commissures in the ventral part of the central nervous system. This absence of segmental commissures might be due to the close proximity of the lateral halves of the ventral ganglia as has been suggested for millipedes which also lack segmental commissures as adults (Harzsch 2004b) despite the early anlagen of commissures during development (Dove and Stollewerk 2003). Although the central nervous system of tardigrades is generally described as belonging to the rope-ladder like type comprising ganglia with connectives and commissures, unambiguous documentation of commissures is almost absent in the literature. Only Marcus (1929: 40, fig. 26) describes and depicts segmental commissures in the ventral nerve cord of Hypsibius (Ramazottius) oberhaeuseri (Doyère 1840). However, this documentation is far from being satisfactory, leaving the possibility open that tardigrades in general lack commissures in the ventral part of the nervous system. Whether segmental commissures are transiently formed during embryonic development is unknown. Accordingly, more investigations on adult neuroanatomy and neurogenesis are badly needed to clarify the issue of segmental commissures in Tardigrada.

As is described in the older literature (Marcus 1929) our investigation reveals that each ventral ganglion pair possesses three large paired segmental nerves (n1 to n3). Our findings also confirm that n1 runs to the dorsal body region (Marcus 1929). In contrast to the previous descriptions, according to which only n3 projects into the limb, we found that n3 and n2 connect to the segmental appendage.

Phylogenetic implications

Neuroanatomy has been used for phylogenetic inferences for many decades (e.g. Holmgren 1916; Hanström 1928; Bullock and Horridge 1965; Sandeman et al. 1993; Harzsch 2006). In particular, the recent morphological techniques of confocal-laser-scanning-microscopy in combination with computer-aided 3-dimensional reconstruction offer a degree of anatomical resolution unmatched by previous methods. This allows comparisons at great detail with sometimes surprising results (e.g. Fanenbruck et al. 2004; Strausfeld et al. 2006; this paper).

Recent phylogenetic analyses of tardigrade affinities almost universally agree that Tardigrada are part of the Arthropoda (e.g. Garey et al. 1996; Giribet et al. 1996; Garey 2001; Mallatt and Giribet 2006). Although in some molecular and morphological analyses tardigrade representatives sometimes appear outside Arthropoda or cluster with cycloneuralians (e.g. Moon and Kim 1996; Bergström and Hou 2001; Giribet 2003; Jenner and Scholtz 2005; Mallat and Griribet 2006; Park et al. 2006), and Ax (1999) considers the relationships of Tardigrada to any other bilaterian group as unresolved. However, even if one accepts tardigrade affinities to arthropods whether tardigrades are the sister group to Euarthropoda (Dewel et al. 1999; Edgecombe et al. 2000; Budd 2001; Maas and Waloszek 2001; Nielsen 2001), to Onychophora plus Euarthropoda (see discussion in Schmidt-Rhaesa et al. 1998; Bergström and Hou 2001; Schmidt-Rhaesa 2001) or to Onychophora (e.g. Plate 1889; Mallatt and Giribet 2006) is still under debate.

The present study is the first to apply immunocytochemical techniques to the study of tardigrade neuroanatomy, and accordingly it is difficult to generalise our findings for the taxon Tardigrada as a whole (see above). However, a putative scenario considering the competing views of the Articulata (e.g. Scholtz 2002) and Ecdysozoa (e.g. Giribet 2003) hypotheses might be allowed with respect to the data at hand.

If the Ecdysozoa hypothesis is correct, then the absence of commissures in M. hufelandi might also represent a character state resembling the condition of cycloneuralians, where apart from lateral commissures to our knowledge no commissures connecting the midventral nerve cords of priapulids, kinorhynchs, and nematodes have been reported (e.g. Storch 1991; Wright 1991; Nebelsick 1993; White et al. 1986) whereas commissures, although in a different pattern (see Ruhberg 2004; Mayer and Harzsch 2007; Whitington 2007), occur in euarthropods and onychophorans. Furthermore, the type of unsegmented but a circumoesophageal ring forming brain found by us in M. hufelandi resembles the condition in cycloneuralians, which also exhibit a ring shaped brain around the oesophagus—although the detailed organisation of the cycloneuralian brain with its typical arrangement of cell bodies, neuropil, and cell bodies as three layers and the absence of an enlarged supraoesophageal part with lobe-like neuropils is different (Wright 1991; Storch 1991; Nebelsick 1993; Neuhaus 1994; Schmidt-Rhaesa et al. 1998; Nielsen 2001; Ax 2001). Moreover, the apparent absence of typical arthropod brain neuropil regions such as mushroom bodies, central complex etc. is shared by Tardigrada and Cycloneuralia. Greven and Kuhlmann (1972) report further similarities between the nervous system of M. hufelandi and nematodes such as the low number of glia cells and the absence of a perilemma surrounding the ganglia. Depending on the polarisation of the characters shared between tardigrades and cycloneuralians these characters might either represent apomorphies supporting cycloneuralian affinities of Tardigrada or they are plesiomorphies (see, for instance, Nielsen 2001, who considers a cricumoesophageal brain as part of the protostome ground pattern) with respect to onychophorans and euarthropods which were retained by the Tardigrada.

Under the assumption that Articulata is a valid taxon and that Tardigrada are related to Arthropoda, the lack of commissures in M. hufelandi would be an evolutionary loss since commissures occur in annelids, onychophorans, and euarthropods (Scholtz 2002; Müller 2006; Strausfeld et al. 2006). If the position of tardigrades as sister group to a clade comprising onychophorans and euarthropods is assumed, the absence of clearly defined segmental ganglia in adult Onychophora (Schürmann 1995; Eriksson et al. 2003; Ruhberg 2004; Mayer and Harzsch 2007; Whitington 2007) would be a secondary feature.

In summary, the characters of the central nervous system of M. hufelandi, in particular the lack of a tripartite/three-segmented brain, speak in favour of a phylogenetic position of Tardigrada outside of a clade comprising Onychophora and Euarthropoda, irrespective of the Articulata or Ecdysozoa hypotheses. However, whether Tardigrada are the sister group to Onychophora plus Euarthropoda or whether they show closer affinities to Cycloneuralia remains to be tested. In any case, this preliminary consideration already reveals that further use of more neuroanatomical data of Tardigrada will contribute to our understanding of tardigrade phylogenetic affinities in a more global approach considering a variety of morphological and molecular data.