The extracellular matrix molecule tenascin-R (TN-R) and the proteoglycans of the lectican family show an overlapping distribution in the developing brain, have been implicated in similar cellular processes and form a complex network of interactions. Previously, we have demonstrated that TN-R induces microprocesses along neurites and enlarged growth cones of tectal cells by interacting with the cell adhesion molecule contactin 1.

Here, we describe competition and cooperation between TN-R, lecticans and contactin 1, and their functional consequences for tectal cells. Aggrecan, brevican and neurocan inhibit the effects of TN-R on microprocess formation and growth cone size. This blocking effect is due to competition of lecticans with binding of TN-R to its neuronal receptor contactin 1, as shown by a sandwich-binding assay. Interaction of aggrecan with TN-R fibronectin type III domains 4-A is necessary for its inhibitory effect on both microprocess formation and TN-R binding to contactin 1. However, the chondroitin sulfate chains are not involved. Time-lapse video microscopy showed that aggrecan has no acute effect on motility and morphology of microprocesses and growth cones but induces long-term neurite retraction after pre-treatment with TN-R.

In contrast to the competition described above, TN-R cooperates with brevican and neurocan to induce attachment of tectal cells and neurite outgrowth, probably by forming a bridge between the lectican substrate and contactin 1 as the neuronal receptor.

Our findings suggest that a complex network of protein-protein interactions within the brain extracellular matrix, as shown here for TN-R and lecticans, is important for the fine-regulation of developmental processes such as microprocess formation along the neurite and neurite outgrowth.

Introduction

The establishment and refinement of neuronal connections depend on dynamic modifications of developing axons in response to extrinsic factors. Filopodia, lamellipodia and microprocesses are motile structures at the axonal growth cone and neurite shaft that are crucial for the growth and pathfinding of axons. For instance, they might be important for facilitating synaptogenic contact at early phases of development, for the correction of pathfinding errors and for terminal arborization. However, the cellular and molecular mechanisms underlying these processes are poorly understood.

Recently, the extracellular matrix (ECM) protein tenascin-R (TN-R) has been described as an extracellular cue that stimulates the generation of microprocesses along the neurite shaft and of enlarged and complex growth cones in chick tectal neurons. The immunoglobulin superfamily member contactin 1, previously termed F11 by us, was identified as the neuronal receptor mediating the effects of TN-R (Zacharias et al., 2002).

TN-R is one of the smallest members of the tenascin family of ECM glycoproteins (for a review, see Jones and Jones, 2000). The multidomain and trimeric structure of TN-R together with its elastic properties suggest that TN-R serves to link cell-surface molecules of different cells and is involved in network formation within the ECM (Oberhauser et al., 1998; Zacharias et al., 1999; Lundell et al., 2004). A variety of ligands and receptor proteins have been described for TN-R including immunoglobulin-like proteins such as contactin 1 (Rathjen et al., 1991; Brummendorf et al., 1993; Norenberg et al., 1995), and ECM proteins such as chondroitin sulfate proteoglycans (CSPG) including the lectican family members neurocan, brevican, versican and aggrecan (Aspberg et al., 1997; Milev et al., 1998; Asher et al., 2002). Interestingly, the high-affinity association between tenascins and lecticans is mediated by protein-protein-interaction involving the fibronectin type III domains (FNIII domains) 4-5 of TN-R or tenascin-C and the C-type lectin domain of lecticans (Aspberg et al., 1997; Rauch et al., 1997; Milev et al., 1998; Day et al., 2004; Lundell et al., 2004). Like TN-R and tenascin-C, CSPGs are abundant in the neuronal ECM and have been shown to be involved in similar cellular processes including neuronal cell adhesion and migration, axon pathfinding, synaptogenesis, and structural organization of the ECM (for reviews, see Yamaguchi, 2000; Rauch, 2004; Schwartz and Domowicz, 2004). In the adult, CSPGs are generally thought to be inhibitory for tissue plasticity and to have a barrier function during regeneration (Moon et al., 2001; Bradbury et al., 2002). However, in the developing central nervous system, CSPGs show a functional dualism: depending on the neuronal cell type and the way they are presented to neurons, both growth inhibitory and growth promoting effects have been described (Iijima et al., 1991; Emerling and Lander, 1996; Garwood et al., 1999; Akita et al., 2004; Wu et al., 2004). The inhibitory properties of CSPGs have mostly been attributed to their glycosaminoglycan (GAG) chains (Friedlander et al., 1994; Asher et al., 2002; Masuda et al., 2004; Sivasankaran et al., 2004), however, the core protein may also act as an inhibitor (Schmalfeldt et al., 2000; Monnier et al., 2003; Ughrin et al., 2003).

In the adult nervous system, TN-R and lecticans are colocalized in specialized ECM structures called perineuronal nets (PNN), which contribute to maintain the structural stability of the mature brain (Bruckner et al., 2000; Yamaguchi, 2000). TN-R functions as a molecular crosslinker of hyaluronan-lectican complexes (Lundell et al., 2004) and, thereby, participates in the macromolecular organization of PNNs. Accordingly, TN-R-knockout mice and lectican-knockout mice display disturbed PNN (Weber et al., 1999; Bruckner et al., 2000; Murakami and Ohtsuka, 2003). TN-R and lecticans are also present in the developing brain. For instance, aggrecan and versican have been described to be expressed in the developing chick optic tectum (Yamagata et al., 1995; Schwartz and Domowicz, 2004) where TN-R is also present (Zacharias et al., 2002). However, the role of their potential interaction during development is less understood.

TN-R and lecticans interact with each other, show an overlapping distribution in the brain and are involved in similar biological phenomena. Therefore, depending on the developmental context, lecticans are likely to collaborate with TN-R (e.g. during the formation of PNNs), or to inhibit the action of TN-R (e.g. by specifically interfering with TN-R binding to cell surface or ECM molecules).

To explore the role of the coordinated action of different ECM molecules with respect to the structural plasticity of the neurite, we analyzed the interaction of TN-R and lecticans in cellular assays for neurite outgrowth and microprocess formation. Here, we describe two new functions of their competitive and cooperative interactions. Aggrecan, neurocan and brevican were found to inhibit the TN-R-induced formation of microprocesses at the neurite shaft and of enlarged growth cones. This blocking effect is due to competition of lecticans with binding of TN-R to its cellular receptor contactin 1 and involves the interaction with TN-R FNIII domains 4-A. Time-lapse video microscopy showed that aggrecan has no acute effect on motility and morphology of shaft microprocesses and growth cones, but induces long-term neurite retraction after pretreatment with TN-R. In contrast to these inhibitory effects, TN-R cooperates with brevican and neurocan to induce tectal cell attachment and neurite outgrowth probably by forming a bridge between the lectican substrate and contactin 1 as the cellular receptor.

Results

Dynamic behavior of TN-R-induced microprocesses and complex growth cones

Previously we have shown that TN-R increases the number of microprocesses and side branches, giving chick tectal neurites a studded appearance on a contactin 1 substrate. In addition, TN-R stimulates the size and complexity of the growth cone. Both effects are mediated by cellular contactin 1 (Zacharias et al., 2002). Here, time-lapse video microscopy of single tectal neurons was used to follow the induction of microprocesses and enlarged growth cones by TN-R and to visualize the motility of these structures. Forty hours after seeding, tectal cells were exposed to TN-R and images were taken at 1-minute intervals (a representative example is shown in Fig. 1). Before bath application of TN-R, tectal neurons on immobilized contactin 1 mostly extended smooth and unbranched neurites with small, motile growth cones. The addition of TN-R had no acute effect on the morphology and motility of growth cones and microprocesses of tectal cells (data not shown). However, microprocess formation and growth cone enlargement became evident after about 2 hours in the presence of TN-R (Fig. 1). The effects on neurites and growth cones were more prominent after a 4-hour incubation period with TN-R (Fig. 1). TN-R induced microprocesses appeared quite dynamic, showing a high degree of motility in a time scale of minutes (Fig. 1, compare left with right image). Additionally, the increase in size and complexity of the growth cone by TN-R was not correlated with a reduced motility: large and complex growth cones showed a highly dynamic behaviour (Fig. 1, compare left with right image). We therefore conclude that morphological changes stimulated by TN-R are due to an increase in the number of microprocesses and the size of the growth cone, but not the stability of these structures.

Lecticans inhibit the TN-R-induced formation of shaft microprocesses and of enlarged growth cones

To investigate whether other constituents of the ECM modulate the effects of TN-R on microprocess formation and growth cone morphology, we became interested in the lectican family of proteoglycans. Aggrecan, versican, neurocan and brevican share the same binding site in TN-R: the FNIII domains 4-5 (Aspberg et al., 1997; Day et al., 2004). Additionally, aggrecan and versican were found to be expressed in the chick optic tectum (Yamagata et al., 1995; Schwartz and Domowicz, 2004).

Aggrecan was added together with TN-R for a period of 6 hours to tectal cells cultivated on a contactin 1 substrate. To quantify the effects of TN-R, tectal-cell morphology was scored as (1) smooth neurite, (2) smooth neurite with an enlarged growth cone, (3) neurite with microprocesses or (4) neurite with microprocesses and an enlarged growth cone (see Materials and Methods). Examples of typical neurons for each category are shown in Fig. 2. When added simultaneously, aggrecan blocked the TN-R-induced formation of microprocesses and of enlarged growth cones in a dose-dependent manner (Fig. 2A). Other members of the lectican family, neurocan and brevican, had a similar inhibitory effect (Fig. 2B). In the presence of lecticans, neurites appeared smooth, few microprocesses were formed along the neurite shaft, and growth cones were small and elongated with only few filopodia resembling neurons in the absence of TN-R. Aggrecan, neurocan and brevican did not affect microprocess formation and growth cone size of tectal cells in the absence of TN-R (Fig. 2A,B). To investigate the possible involvement of GAG side chains, aggrecan was pretreated with chondroitinase ABC. This treatment hardly reduced the blocking effect of aggrecan (Fig. 2B), suggesting an inhibitory role of the core protein.

The lectican-binding site has recently been mapped to TN-R FNIII domains 4-5 (Lundell et al., 2004). To study whether these domains are necessary for the inhibitory effect of aggrecan on the TN-R-induced formation of microprocesses and of enlarged growth cones, antibodies against TN-R FNIII domains 4-A were used. These antibodies did not affect TN-R-induced microprocess formation and growth cone enlargement in the absence of aggrecan (Fig. 3). However, antibodies against TN-R FNIII domains 4-A neutralized the inhibitory effect of aggrecan. Blocking the interaction between TN-R and aggrecan with antibodies against TN-R FNIII domains 4-A resulted in microprocess formation and in an increase in growth cone size comparable to the effect of TN-R alone (Fig. 3). Antibodies directed against other TN-R domains had no effect (data not shown).

Lecticans block the interaction between TN-R and its neuronal receptor contactin 1

The cell adhesion molecule contactin 1 has been identified as the cellular receptor mediating the effects of TN-R on microprocess formation and growth cone size (Zacharias et al., 2002). One explanation for the finding that lecticans inhibit TN-R-induced microprocess and growth cone formation might be that lecticans block the interaction between TN-R and contactin 1. The binding sites for contactin 1 and lecticans have been mapped to adjacent regions of TN-R, FNIII domains 2-3 (Norenberg et al., 1995) and FNIII domains 4-5 (Aspberg et al., 1997; Lundell et al., 2004), respectively. To address this question, the interactions between contactin 1, TN-R and lecticans were analyzed in a sandwich binding assay that reflects the situation in chick tectal cell cultures. In this assay, contactin-1-expressing COS cells were incubated with contactin-1-coated microspheres in the absence or presence of TN-R as illustrated in Fig. 4G. As described previously (Brummendorf et al., 1993; Zacharias et al., 1999), contactin 1 did not show homophilic binding (Fig. 4A). However, due to its trimeric structure, TN-R can most likely interact simultaneously with two individual contactin 1 polypeptides, resulting in the formation of a molecular bridge (Fig. 4B) (Zacharias et al., 1999). Addition of aggrecan together with TN-R led to a complete loss of contactin-1-bead binding (Fig. 4C), suggesting that aggrecan competes with contactin 1 for binding to TN-R. Similarly, brevican, neurocan and chondroitinase-ABC-treated aggrecan blocked the interaction between contactin 1 and TN-R (Fig. 4D-F). In parallel to the sandwich binding assay described above, competition by lecticans was also observed in a direct binding assay by using TN-R-coated beads and contactin-1-expressing COS cells (results not shown).

To investigate the role of the established lectican-binding site within TN-R, antibodies against TN-R FNIII domains 4-A and a TN-R fragment comprising the FNIII domains 4-A were included in the sandwich binding assay (Fig. 5). Both antagonists had no effect on TN-R-bridge formation in the absence of aggrecan (data not shown). However, both completely reversed the inhibitory effect of aggrecan. If the interaction between TN-R and aggrecan was blocked by antibodies against TN-R FNIII domains 4-A or a TN-R fragment comprising the FNIII domains 4-A, a TN-R bridge between contactin 1 polypeptides could have formed again (Fig. 5C,D).

Aggrecan stimulates long-term neurite retraction in the presence of TN-R

In the above described experiments TN-R and lecticans were simultaneously applied to tectal cells extending neurites on contactin 1. In the following set of experiments, we tested the effects of lecticans after the induction of microprocesses by pretreatment with TN-R for 20 hours. Bath application of aggrecan had no acute effect on morphology and motility of microprocesses and enlarged growth cones of tectal cells in the presence of TN-R, i.e. aggrecan induced neither immediate growth cone collapse nor acute retraction of microprocesses. A typical example is depicted in Fig. 6A. Images taken at 1-minute intervals illustrate the motility of TN-R-induced microprocesses and growth cones before and after exposure to aggrecan for 15 minutes and 1 hour, respectively. Quantification of growth cone size of fixed tectal cells (see Materials and Methods) that had been cultured for 20 hours in the presence of TN-R alone before treatment with aggrecan for 15 minutes or 1 hour, confirmed the lack of aggrecan-induced growth cone collapse (Fig. 6B). Additionally, aggrecan had no effect on the morphology and motility of growth cones and microprocesses in the absence of TN-R (data not shown).

Time-lapse microscopy was used to follow the long-term effects of aggrecan on single tectal cells after pretreatment with TN-R for 20 hours. Images were taken at 5-minute intervals during a 4-hour observation period, before and after bath application of aggrecan. Frequently, aggrecan induced retraction of tectal cell neurites. Neurite retraction was a rather slow process and microprocesses at the neurite shaft and growth cones mostly retained their morphology and motility during retraction. A representative example is shown in Fig. 7B. Neurite retraction usually started 1-2 hours after treatment with aggrecan, however, occasionally the onset of retraction was already observed after 30 minutes. The extent of retraction varied greatly - from partial to complete retraction of the neurite. Sometimes, even detachment of the cell soma was observed. For quantification, neurites were considered as retracted when neurite length was reduced by more than 20% of their initial total length within 4 hours. After pretreatment with TN-R for 20 hours, aggrecan induced neurite retraction in about 40% of tectal cells, whereas neurite retraction was observed only in about 10% of control cells (without TN-R pretreatment) either in the presence or in the absence of aggrecan (Fig. 7D). Chondroitinase-ABC-treated aggrecan induced neurite retraction to a similar extend (Fig. 7C,D).

TN-R cooperates with lecticans to induce neurite outgrowth

In contrast to the competition between TN-R, lecticans and contactin 1 described above, these molecules were able to cooperate in neurite induction in a different experimental setting. Brevican, immobilized as a substrate, was not permissive for neurite outgrowth of tectal cells (Fig. 8A). However, when TN-R was added together with tectal cells to immobilized brevican, strong cell attachment and neurite outgrowth were induced (Fig. 8B). Tectal cells displayed a complex morphology, including microprocesses along their neurites and enlarged growth cones. A similar promotion of neurite outgrowth was observed after preincubation of immobilized brevican with TN-R, suggesting that binding of TN-R to the brevican substrate is sufficient (data not shown). The direct interaction between immobilized brevican and TN-R was verified in a dot assay using TN-R-coated microspheres (data not shown).

To identify the neuronal receptor responsible for neurite outgrowth on immobilized brevican in the presence of TN-R, the involvement of various cell-surface proteins was tested by the application of blocking antibodies. Antibodies against contactin 1 completely inhibited TN-R-induced neurite outgrowth on immobilized brevican (Fig. 8D). Additionally, antibodies against TN-R FNIII domains 4-A, the established lectican-binding site in TN-R, strongly reduced the neurite-outgrowth-promoting effect of TN-R (Fig. 8E). A similar inhibition of neurite outgrowth was observed in the presence of aggrecan (Fig. 8F), probably due to aggrecan competition with brevican for their common binding site in TN-R. As illustrated in Fig. 8C, TN-R-induced neurite outgrowth most likely depends on the simultaneous interaction of TN-R with both the brevican substrate and the contactin 1 receptor on the tectal cell because of the trimeric structure of TN-R. A similar TN-R-bridge formation has been observed between two contactin 1 polypeptides in the COS cell sandwich binding assay described above (see Fig. 4G).

Moreover, cooperation is also possible between TN-R and immobilized neurocan, resulting in slight neurite outgrowth in tectal cells, that was also sensitive to antibodies against contactin 1, TN-R FNIII domains 4-A as well as to aggrecan (data not shown). However, neither aggrecan nor chondroitinase-ABC-treated aggrecan immobilized as a substrate were able to cooperate with TN-R to induce neurite outgrowth (data not shown).

Overlapping expression of TN-R, contactin 1 and CSPGs in the developing chick optic tectum and in tectal cell cultures

To investigate whether TN-R, contactin 1, and CSPGs subserve the functions described above for the chick optic tectum, we compared the protein expression pattern of these molecules by immunohistochemistry (Fig. 9A). In cryostat sections of embryonic day 9 (E9), E11 tectum and E13 the laminar organization of the tectum was revealed with a nuclear counterstain and laminae were classified according to Yamagata et al. (Yamagata et al., 1995). Contactin 1, neurocan, and chondroitin sulfate (as revealed by staining with mAb CS-56) were broadly expressed throughout the tectum with a slight preference for the plexiform layers at all stages investigated. TN-R and aggrecan showed an increase of expression levels at E13 compared with E9, and a more restricted localization. To show colocalization in more detail, cryostat sections of E13 tectum were double-stained for neurocan and contactin 1 or TN-R as indicated (Fig. 9B). All three molecules were co-expressed in the stratum griseum et fibrosum superficiale (SGFS) lamina H. These data further support possible interactions between TN-R, contactin 1 and lecticans based on their distinct but overlapping expression pattern in the developing tectum.

Interestingly, immunocytochemical staining of cultured tectal cells revealed that these cells express neurocan, contactin 1 and TN-R themselves (Fig. 9C). Contactin 1 showed the typical cell-surface localization of a cell adhesion molecule. The ECM molecule TN-R was apparently retained on the tectal cell surface, resulting in a similar cell-associated staining pattern, for comparison see Rathjen et al. and Kappler et al. (Rathjen et al., 1991; Kappler et al., 2002). Moreover, tectal cells were able to secrete neurocan, which showed a pericellular and substrate-associated distribution concentrated around tectal cells. Deposition of neurocan was observed in the vicinity of the cell soma, the neurite and/or the growth cone (Fig. 9C and data not shown). A similar distribution in the surroundings of the cell has been described for neurocan expressed by astrocytes (Asher et al., 2000).

Discussion

The active contribution of the extracellular matrix to differentiation and morphogenesis of neural tissues is increasingly acknowledged. ECM molecules of the lectican and tenascin families are multidomain and oligomeric proteins allowing them to establish a complex network of interactions within the ECM and with adjacent cells. Certain interactions are possible in parallel, whereas other interactions exclude each other. For example, lecticans and TN-R collaborate during the formation of PNNs in the adult, where TN-R is supposed to organize hyaluronan-lectican complexes by acting as a molecular cross linker (Bruckner et al., 2000; Lundell et al., 2004). Moreover, here, we describe that TN-R cooperates with brevican and neurocan to induce tectal cell attachment and neurite outgrowth, probably by forming a bridge between the lectican substrate and contactin 1 as the cellular receptor.

In addition to these examples for cooperation, we observe a competition between TN-R, lecticans and contactin 1 in a cellular assay for microprocess formation during development. Aggrecan, neurocan and brevican interfere with the binding of TN-R to its cellular receptor contactin 1, and thereby inhibit TN-R-induced formation of microprocesses as well as enlarged and complex growth cones in tectal cells. These examples illustrate that the interaction between TN-R, lecticans, and contactin 1 can result in cooperation or in competition, depending on the developmental context, the function analyzed, the assay design and the relative concentrations of the interaction partners. The coordinated action of different ECM molecules, such as TN-R and lecticans is one example how a complex network of protein-protein interactions allows the fine-regulation of developmental processes beyond the regulation of gene and protein expression.

The inhibitory properties of CSPGs have frequently been attributed to their GAG chains (Friedlander et al., 1994; Asher et al., 2002; Masuda et al., 2004; Sivasankaran et al., 2004). However, three lines of evidence suggest that the inhibitory effect of lecticans on TN-R-induced formation of microprocesses and enlarged growth cones is mediated by protein-protein interaction. (1) Chondroitinase-ABC-treated aggrecan is equally effective in blocking TN-R-induced microprocess formation, inhibiting TN-R binding to contactin 1 and inducing neurite retraction as aggrecan-bearing GAG chains. (2) In contrast to aggrecan and neurocan, brevican secreted by HEK 293 cells is not modified with GAG chains but exerts similar blocking effects. (3) Aggrecan inhibition can be specifically reversed by antibodies and by a protein fragment interfering with TN-R FNIII domains 4-5, the known binding site for the C-type lectin domain of lecticans (Aspberg et al., 1997; Lundell et al., 2004). The binding site for contactin 1 has been mapped to an adjacent region of TN-R, FNIII domains 2-3 (Norenberg et al., 1995). Accordingly, the inhibitory effect of lecticans on binding of TN-R to contactin 1, and on TN-R-induced formation of microprocesses are probably due to sterical hinderance because of the large size of the lectican molecules.

Time-lapse analysis of tectal neurite and growth cone dynamics shows that aggrecan does not seem to be inhibitory to tectal cells per se. Exposure of tectal cells (grown in the presence or absence of TN-R) to aggrecan has no acute effect on the morphology and motility of growth cones and shaft microprocesses. However, aggrecan induces long-term neurite retraction of tectal cells after pretreatment with TN-R. Unlike interactions of growth cones with other inhibitory guidance molecules, such as ephrins and semaphorins (Gallo and Letourneau, 2004), growth cones that contact CSPGs rarely collapse (Ughrin et al., 2003; Masuda et al., 2004; Schweigreiter et al., 2004). Otherwise, an avoidance reaction towards CSPG in a choice situation, such as in a stripe assay, has frequently been described (Snow et al., 1994; Walz et al., 2002; Monnier et al., 2003; Marler et al., 2005). These data suggest that the mechanisms of inhibition of growth cone migration by CSPGs differ from those of other repulsive guidance molecules and may allow a more subtle and more local response of the growth cone.

The most commonly used explanation for the inhibitory effects of CSPGs on neurite growth is their capability to interfere with the binding of the neurons to the substrate. CSPGs might simply block cell-matrix or cell-cell contacts, either directly by masking adhesive interactions (Friedlander et al., 1994; Oleszewski et al., 2000) or indirectly by acting as an extracellular matrix binding site for inhibitory molecules (Emerling and Lander, 1996; Kantor et al., 2004). In all cases the result would be an imbalanced adhesion of the growth cone to its surrounding environment, leading to neurite retraction. However, CSPGs have also been described to initiate an intracellular signaling cascade after interaction with specific receptors on the surface of the growth cone (Li et al., 2000; Monnier et al., 2003; Sivasankaran et al., 2004; Wu et al., 2004).

In contrast to the competitive interactions of TN-R, lecticans and contactin 1 discussed above, TN-R cooperates with brevican and neurocan to induce tectal cell attachment and neurite outgrowth, probably by forming a bridge between the brevican or neurocan substrate and contactin 1 as the neuronal receptor. Immobilized brevican and neurocan are not permissive for neurite outgrowth of tectal cells, however, neurite outgrowth is strongly induced in the presence of TN-R. Binding of TN-R to brevican and neurocan might therefore neutralize growth inhibitory properties of these molecules. However, immobilized brevican and neurocan might act as extracellular matrix attachment sites that present growth promoting cues - like TN-R - to the responding cell.

TN-R and lecticans are abundant in the brain ECM, implicated in similar biological phenomena, and undergo a complex network of competitive and cooperative interactions. The physiological significance of possible interactions between TN-R and lecticans during development of the chick optic tectum is supported by the overlapping distribution of TN-R with chondroitin sulfate, neurocan and aggrecan, and with its neuronal receptor contactin 1 in embryonic tectal sections. The node of Ranvier is discussed as another example for the coordinated action of lecticans and tenascins. Versican and hyaluronan colocalize with TN-R, contactin 1 and voltage-gated sodium channels at the node of Ranvier in the central nervous system and might contribute to the formation of multimeric nodal complexes (Srinivasan et al., 1998; Kazarinova-Noyes and Shrager, 2002; Oohashi et al., 2002). Based on the role of versican as an inhibitor of neurite outgrowth (Schmalfeldt et al., 2000; Asher et al., 2002), it has been speculated that versican acts as an inhibitor of axon branching at the node of Ranvier (Melendez-Vasquez et al., 2005). Additionally, contactin-1-knockout mice show morphological abnormalities at the axo-glial junction of the paranode of the myelinated peripheral nerve and in different cell types of the developing cerebellum (Berglund et al., 1999; Boyle et al., 2001).

It has been suggested that TN-R, via its neuronal receptor contactin 1, induces a transition from long-distance growth of tectal interneurons to differentiation, including the upregulation of their ability to form microprocesses and side branches. The identification of extracellular signals such as TN-R involved in the differentiation of neurite shafts and their interplay with other constituents of a complex ECM, such as lecticans, is essential for the understanding of the generation of neuronal connectivity during development.

Materials and Methods

Proteins and antibodies

TN-R and contactin 1 were purified from adult chicken brain by immunoaffinity chromatography (Zacharias et al., 1999). The TN-R fragment comprising FNIII domains 4-A was expressed as a glutathione-S-transferase-fusion protein (Norenberg et al., 1995). Antibodies to these proteins have already been described (Norenberg et al., 1995; Zacharias et al., 1999).

Affinity-purified rabbit polyclonal antibodies (pAb) against chick neurocan and chick aggrecan were kindly provided by S. Hoffman (Charleston, SC) (Zanin et al., 1999). Anti-chondroitin sulfate monoclonal antibody (mAb) CS-56 (Avnur and Geiger, 1985), chondroitinase ABC and aggrecan were purchased from Sigma. Alexa Fluor-488- and Alexa Fluor-594-labelled secondary antibodies were used in all experiments (Molecular Probes).

For treatment with chondroitinase ABC, aggrecan (50 μg) was incubated with 0.1 U/ml enzyme for 1 hour at 37°C according to the manufacturer's instructions. Removal of GAG chains was confirmed by SDS-PAGE and by loss of reactivity with mAb CS-56.

Neurocan- and brevican-containing culture supernatants of HEK 293 cells were collected under serum-free conditions and, if necessary, dialysed against Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum (FCS). The transfected rat neurocan and rat brevican cDNAs had been modified by changing the endogenous 5′ untranslated sequences and the signal peptide against the corresponding sequences of BM40 as described (Feng et al., 2000; Talts et al., 2000). CSPG content of the supernatants was verified by SDS-PAGE and staining with Coomassie Blue (supplementary material Fig. S1). Neurocan secreted by HEK 293 cells was modified with GAG chains, whereas brevican was not. Culture supernatants from mock-transfected cells were used as a control.

Morphological analysis of tectal cells and immunocytochemistry

Cultivation of chick embryonic day 6 (E6) tectal cells at low-density on contactin 1 immobilized to Petriperm dishes (Greiner) has been described (Zacharias et al., 1999; Zacharias et al., 2002). Tectal cells reveal a robust neurite outgrowth on a contactin 1 substrate using NrCAM as the neuronal receptor (Morales et al., 1993). After 24 hours, cultures were washed twice with DMEM supplemented with 10% FCS to remove non-adherent cells. During the last 6 hours of an additional 24-hour incubation period, tectal cells were treated with TN-R (10 μg/ml), aggrecan (5-20 μg/ml), chondroitinase-ABC-treated aggrecan (20 μg/ml), neurocan- or brevican-containing HEK 293 cell supernatants and Fab fragments of pAb against TN-R FNIII domains 4-A (500 μg/ml) in various combinations. Cell cultures were fixed and stained with mAb A2B5 by indirect immunofluorescence using Alexa Fluor-594 conjugated to goat anti-mouse pAb (1:200). For quantification of TN-R-induced morphological changes, tectal neurons were scored as containing a (1) smooth neurite, (2) smooth neurite with an enlarged growth cone, (3) neurite with microprocesses or (4) neurite with microprocesses and an enlarged growth cone similar to Zacharias et al. (Zacharias et al., 2002). Neurons with more than five microprocesses along the neurite shaft were considered to belong in category 3 or 4, depending on the morphology of their growth cone. For quantification of TN-R-induced changes in growth cone morphology, growth cones were considered enlarged when they were larger than 15 μm in diameter, regardless of the morphology of the neurite (which corresponds to the sum of categories 2 and 4). Between 50 and 100 neurons were scored for each experimental condition. Data were compiled from at least five independent experiments.

To investigate the endogenous expression of contactin 1, TN-R and neurocan, chick tectal cells were cultivated for 48 hours on a mouse contactin 1 (F3) substrate as described (Zacharias et al., 2002). Living cells were incubated on ice with anti-contactin 1 mAb F11 number 0 (10 μg/ml) or anti-TN-R mAb 23-13 (10 μg/ml), fixed and stained with Alexa Fluor-488-conjugated anti-mouse pAb (1:200). Afterwards, neurocan expression was revealed with affinity-purified pAb against neurocan (5 μg/ml) and Alexa Fluor-594-conjugated anti-rabbit pAb (1:200).

Time-lapse video microscopy

Chick E6 tectal cells were grown at low-density on contactin-1-coated Petriperm dishes, as described above, for 40 hours, placed in an Incubator XL mounted to a Zeiss AxioVert 200M inverted microscope and maintained under CO2- and temperature-controlled atmosphere. To avoid acute effects on the morphology of neurites and growth cones only a small part of the medium was removed in all time-lapse experiments, the respective substance added and the medium carefully returned to the cells. Tectal neurons were viewed with a 63× 1.4 PlanNeofluar oil objective under phase-contrast optics and imaged with an AxioCam camera under the control of AxioVision 4.3 digital imaging system (Zeiss). Image processing and analysis were performed using Zeiss AxioVision 4.3 software and Adobe Photoshop.

To analyze the formation of microprocesses, TN-R (10 μg/ml) was added during time-lapse recording and images were taken within 1-minute intervals. To follow aggrecan induced morphological changes, aggrecan (20 μg/ml) or chondroitinase-ABC-treated aggrecan (20 μg/ml) was added during time-lapse recording to tectal cells cultivated for 20 hours in the absence or presence of TN-R (10 μg/ml). Single tectal cells were imaged in the continuous presence of TN-R for acute effects of aggrecan on morphology and motility of growth cones and microprocesses (imaged after 15 minutes and 1 hour) and for neurite retraction (imaged after 2 hours and 4 hours). More than 50 cells were analyzed for morphology and motility of growth cones, microprocesses and neurites under each experimental condition.

For quantification of neurite retraction, neurites qualified as retraced when neurite length was reduced by more than 20% of their total length within 4 hours. The percentage of retracted neurites was calculated from at least five independent experiments each comprising about 20 cells.

Neurite outgrowth on immobilized lecticans

Petriperm dishes were coated overnight with brevican or neurocan and blocked, and chick E6 tectal cells were cultivated at low-density for 40 hours with TN-R (10 μg/ml), Fab fragments of pAb to contactin 1 (500 μg/ml), Fab fragments of pAb to TN-R FNIII domains 4-A (500 μg/ml) or aggrecan (20 μg/ml) in various combinations. Cell cultures were fixed and stained with mAb A2B5 as described above. For preincubation studies, immobilized brevican was incubated after blocking with TN-R (10 μg/ml) for 1 hour at 37°C.

COS cell sandwich binding assay

Transfection of COS cells and sandwich binding assay have been described previously (Zacharias et al., 1999). In brief, COS cells were transfected with contactin 1 and incubated with contactin-1-coated red-fluorescent microspheres in the presence of TN-R (10 μg/ml), aggrecan (20 μg/ml), chondroitinase-ABC-treated aggrecan (20 μg/ml), neurocan- or brevican-containing HEK 293 cell supernatants, TN-R FNIII domains 4-A fragment (100 μg/ml) or Fab fragments of pAb to TN-R FNIII domains 4-A (500 μg/ml) in various combinations. After fixation, contactin 1 expression on the COS cell surface was revealed by staining with anti-contactin 1 mAb F11 number 0 (10 μg/ml) and Alexa Fluor-488-conjugated anti-mouse pAb (1:200). Double fluorescence detection was performed with a confocal microscope (MRC 1024, Bio Rad Laboratories) and images were analyzed for red fluorescence (indicating microsphere binding to the cell surface) and for green fluorescence (indicating contactin 1 expression).

Immunohistochemistry

For immunohistochemical localization of contactin 1, TN-R and CSPGs, formaldehyde-fixed cryostat sections of E9, E11 and E13 chick optic tectum were incubated with anti-contactin 1 mAb F11 number 0 (10 μg/ml), with anti-TN-R mAb 23-13 (10 μg/ml), with anti-chondroitin sulfate mAb CS-56 (13 μg/ml), affinity purified pAb against neurocan (5 μg/ml) and aggrecan (10 μg/ml). Alexa Fluor-594-conjugated anti-mouse or Alexa Fluor-594-conjugated anti-rabbit pAb (1:400) were used as secondary antibodies. For double-labeling, primary antibodies were used in various combinations together with Alexa Fluor-594-conjugated anti-rabbit and Alexa 488-conjugated anti-mouse secondary antibodies as indicated. Nuclear counterstaining was performed using DRAQ5 (10 μg/ml, 10 minutes at room temperature) according to the manufacturer's instructions (Biostatus).

Acknowledgements

We thank F. G. Rathjen for continuous support and Hannelore Drechsler for technical assistance. S. Hoffman (Charleston, SC) kindly provided us with antibodies against chick neurocan and chick aggrecan. These studies were supported by grants from the Deutsche Forschungsgemeinschaft (Za 167/2) and the European community to U.Z. and from the Swedish Research Council, the Alfred Österlunds stiftelse, the H. och J. Forssmans fond, the Greta och Johan Kocks stiftelser, and the Crafoordska stiftelsen to U.R.

References

Akita, K., Toda, M., Hosoki, Y., Inoue, M., Fushiki, S., Oohira, A., Okayama, M., Yamashina, I. and Nakada, H. (
2004
). Heparan sulphate proteoglycans interact with neurocan and promote neurite outgrowth from cerebellar granule cells.
Biochem. J.
383
,
129
-138.
Asher, R. A., Morgenstern, D. A., Fidler, P. S., Adcock, K. H., Oohira, A., Braistead, J. E., Levine, J. M., Margolis, R. U., Rogers, J. H. and Fawcett, J. W. (
2000
). Neurocan is upregulated in injured brain and in cytokine-treated astrocytes.
J. Neurosci.
20
,
2427
-2438.
Asher, R. A., Morgenstern, D. A., Shearer, M. C., Adcock, K. H., Pesheva, P. and Fawcett, J. W. (
2002
). Versican is upregulated in CNS injury and is a product of oligodendrocyte lineage cells.
J. Neurosci.
22
,
2225
-2236.
Aspberg, A., Miura, R., Bourdoulous, S., Shimonaka, M., Heinegard, D., Schachner, M., Ruoslahti, E. and Yamaguchi, Y. (
1997
). The C-type lectin domains of lecticans, a family of aggregating chondroitin sulfate proteoglycans, bind tenascin-R by protein-protein interactions independent of carbohydrate moiety.
Proc. Natl. Acad. Sci. USA
94
,
10116
-10121.
Avnur, Z. and Geiger, B. (
1985
). Spatial interrelationships between proteoglycans and extracellular matrix proteins in cell cultures.
Exp. Cell Res.
158
,
321
-332.
Berglund, E. O., Murai, K. K., Fredette, B., Sekerkova, G., Marturano, B., Weber, L., Mugnaini, E. and Ranscht, B. (
1999
). Ataxia and abnormal cerebellar microorganization in mice with ablated contactin gene expression.
Neuron
24
,
739
-750.
Boyle, M. E., Berglund, E. O., Murai, K. K., Weber, L., Peles, E. and Ranscht, B. (
2001
). Contactin orchestrates assembly of the septate-like junctions at the paranode in myelinated peripheral nerve.
Neuron
30
,
385
-397.
Bradbury, E. J., Moon, L. D., Popat, R. J., King, V. R., Bennett, G. S., Patel, P. N., Fawcett, J. W. and McMahon, S. B. (
2002
). Chondroitinase ABC promotes functional recovery after spinal cord injury.
Nature
416
,
636
-640.
Bruckner, G., Grosche, J., Schmidt, S., Hartig, W., Margolis, R. U., Delpech, B., Seidenbecher, C. I., Czaniera, R. and Schachner, M. (
2000
). Postnatal development of perineuronal nets in wild-type mice and in a mutant deficient in tenascin-R.
J. Comp. Neurol.
428
,
616
-629.
Brummendorf, T., Hubert, M., Treubert, U., Leuschner, R., Tarnok, A. and Rathjen, F. G. (
1993
). The axonal recognition molecule F11 is a multifunctional protein: specific domains mediate interactions with Ng-CAM and restrictin.
Neuron
10
,
711
-727.
Day, J. M., Olin, A. I., Murdoch, A. D., Canfield, A., Sasaki, T., Timpl, R., Hardingham, T. E. and Aspberg, A. (
2004
). Alternative splicing in the aggrecan G3 domain influences binding interactions with tenascin-C and other extracellular matrix proteins.
J. Biol. Chem.
279
,
12511
-12518.
Emerling, D. E. and Lander, A. D. (
1996
). Inhibitors and promoters of thalamic neuron adhesion and outgrowth in embryonic neocortex: functional association with chondroitin sulfate.
Neuron
17
,
1089
-1100.
Feng, K., Arnold-Ammer, I. and Rauch, U. (
2000
). Neurocan is a heparin binding proteoglycan.
Biochem. Biophys. Res. Commun.
272
,
449
-455.
Friedlander, D. R., Milev, P., Karthikeyan, L., Margolis, R. K., Margolis, R. U. and Grumet, M. (
1994
). The neuronal chondroitin sulfate proteoglycan neurocan binds to the neural cell adhesion molecules Ng-CAM/L1/NILE and N-CAM, and inhibits neuronal adhesion and neurite outgrowth.
J. Cell Biol.
125
,
669
-680.
Gallo, G. and Letourneau, P. C. (
2004
). Regulation of growth cone actin filaments by guidance cues.
J. Neurobiol.
58
,
92
-102.
Garwood, J., Schnadelbach, O., Clement, A., Schutte, K., Bach, A. and Faissner, A. (
1999
). DSD-1-proteoglycan is the mouse homolog of phosphacan and displays opposing effects on neurite outgrowth dependent on neuronal lineage.
J. Neurosci.
19
,
3888
-3899.
Iijima, N., Oohira, A., Mori, T., Kitabatake, K. and Kohsaka, S. (
1991
). Core protein of chondroitin sulfate proteoglycan promotes neurite outgrowth from cultured neocortical neurons.
J. Neurochem.
56
,
706
-708.
Jones, F. S. and Jones, P. L. (
2000
). The tenascin family of ECM glycoproteins: structure, function, and regulation during embryonic development and tissue remodeling.
Dev. Dyn.
218
,
235
-259.
Kantor, D. B., Chivatakarn, O., Peer, K. L., Oster, S. F., Inatani, M., Hansen, M. J., Flanagan, J. G., Yamaguchi, Y., Sretavan, D. W., Giger, R. J. et al. (
2004
). Semaphorin 5A is a bifunctional axon guidance cue regulated by heparan and chondroitin sulfate proteoglycans.
Neuron
44
,
961
-975.
Kappler, J., Baader, S. L., Franken, S., Pesheva, P., Schilling, K., Rauch, U. and Gieselmann, V. (
2002
). Tenascins are associated with lipid rafts isolated from mouse brain.
Biochem. Biophys. Res. Commun.
294
,
742
-747.
Kazarinova-Noyes, K. and Shrager, P. (
2002
). Molecular constituents of the node of Ranvier.
Mol. Neurobiol.
26
,
167
-182.
Li, H., Leung, T. C., Hoffman, S., Balsamo, J. and Lilien, J. (
2000
). Coordinate regulation of cadherin and integrin function by the chondroitin sulfate proteoglycan neurocan.
J. Cell Biol.
149
,
1275
-1288.
Lundell, A., Olin, A. I., Morgelin, M., al Karadaghi, S., Aspberg, A. and Logan, D. T. (
2004
). Structural basis for interactions between tenascins and lectican C-type lectin domains: evidence for a crosslinking role for tenascins.
Structure
12
,
1495
-1506.
Marler, K. J., Kozma, R., Ahmed, S., Dong, J. M., Hall, C. and Lim, L. (
2005
). Outgrowth of neurites from NIE-115 neuroblastoma cells is prevented on repulsive substrates through the action of PAK.
Mol. Cell Biol.
25
,
5226
-5241.
Masuda, T., Fukamauchi, F., Takeda, Y., Fujisawa, H., Watanabe, K., Okado, N. and Shiga, T. (
2004
). Developmental regulation of notochord-derived repulsion for dorsal root ganglion axons.
Mol. Cell Neurosci.
25
,
217
-227.
Melendez-Vasquez, C., Carey, D. J., Zanazzi, G., Reizes, O., Maurel, P. and Salzer, J. L. (
2005
). Differential expression of proteoglycans at central and peripheral nodes of Ranvier.
Glia
52
,
301
-308.
Milev, P., Chiba, A., Haring, M., Rauvala, H., Schachner, M., Ranscht, B., Margolis, R. K. and Margolis, R. U. (
1998
). High affinity binding and overlapping localization of neurocan and phosphacan/protein-tyrosine phosphatase-zeta/beta with tenascin-R, amphoterin, and the heparin-binding growth-associated molecule.
J. Biol. Chem.
273
,
6998
-7005.
Monnier, P. P., Sierra, A., Schwab, J. M., Henke-Fahle, S. and Mueller, B. K. (
2003
). The Rho/ROCK pathway mediates neurite growth-inhibitory activity associated with the chondroitin sulfate proteoglycans of the CNS glial scar.
Mol. Cell Neurosci.
22
,
319
-330.
Moon, L. D., Asher, R. A., Rhodes, K. E. and Fawcett, J. W. (
2001
). Regeneration of CNS axons back to their target following treatment of adult rat brain with chondroitinase ABC.
Nat. Neurosci.
4
,
465
-466.
Morales, G., Hubert, M., Brummendorf, T., Treubert, U., Tarnok, A., Schwarz, U. and Rathjen, F. G. (
1993
). Induction of axonal growth by heterophilic interactions between the cell surface recognition proteins F11 and Nr-CAM/Bravo.
Neuron
11
,
1113
-1122.
Murakami, T. and Ohtsuka, A. (
2003
). Perisynaptic barrier of proteoglycans in the mature brain and spinal cord.
Arch. Histol. Cytol.
66
,
195
-207.
Norenberg, U., Hubert, M., Brummendorf, T., Tarnok, A. and Rathjen, F. G. (
1995
). Characterization of functional domains of the tenascin-R (restrictin) polypeptide: cell attachment site, binding with F11, and enhancement of F11-mediated neurite outgrowth by tenascin-R.
J. Cell Biol.
130
,
473
-484.
Oberhauser, A. F., Marszalek, P. E., Erickson, H. P. and Fernandez, J. M. (
1998
). The molecular elasticity of the extracellular matrix protein tenascin.
Nature
393
,
181
-185.
Oleszewski, M., Gutwein, P., von der Lieth, W., Rauch, U. and Altevogt, P. (
2000
). Characterization of the L1-neurocan-binding site. Implications for L1-L1 homophilic binding.
J. Biol. Chem.
275
,
34478
-34485.
Oohashi, T., Hirakawa, S., Bekku, Y., Rauch, U., Zimmermann, D. R., Su, W. D., Ohtsuka, A., Murakami, T. and Ninomiya, Y. (
2002
). Bral1, a brain-specific link protein, colocalizing with the versican V2 isoform at the nodes of Ranvier in developing and adult mouse central nervous systems.
Mol. Cell Neurosci.
19
,
43
-57.
Rathjen, F. G., Wolff, J. M. and Chiquet-Ehrismann, R. (
1991
). Restrictin: a chick neural extracellular matrix protein involved in cell attachment co-purifies with the cell recognition molecule F11.
Development
113
,
151
-164.
Rauch, U. (
2004
). Extracellular matrix components associated with remodeling processes in brain.
Cell Mol. Life Sci.
61
,
2031
-2045.
Rauch, U., Clement, A., Retzler, C., Frohlich, L., Fassler, R., Gohring, W. and Faissner, A. (
1997
). Mapping of a defined neurocan binding site to distinct domains of tenascin-C.
J. Biol. Chem.
272
,
26905
-26912.
Schmalfeldt, M., Bandtlow, C. E., Dours-Zimmermann, M. T., Winterhalter, K. H. and Zimmermann, D. R. (
2000
). Brain derived versican V2 is a potent inhibitor of axonal growth.
J. Cell Sci.
113
,
807
-816.
Schwartz, N. B. and Domowicz, M. (
2004
). Proteoglycans in brain development.
Glycoconj. J.
21
,
329
-341.
Schweigreiter, R., Walmsley, A. R., Niederost, B., Zimmermann, D. R., Oertle, T., Casademunt, E., Frentzel, S., Dechant, G., Mir, A. and Bandtlow, C. E. (
2004
). Versican V2 and the central inhibitory domain of Nogo-A inhibit neurite growth via p75NTR/NgR-independent pathways that converge at RhoA.
Mol. Cell Neurosci.
27
,
163
-174.
Sivasankaran, R., Pei, J., Wang, K. C., Zhang, Y. P., Shields, C. B., Xu, X. M. and He, Z. (
2004
). PKC mediates inhibitory effects of myelin and chondroitin sulfate proteoglycans on axonal regeneration.
Nat. Neurosci.
7
,
261
-268.
Snow, D. M., Atkinson, P. B., Hassinger, T. D., Letourneau, P. C. and Kater, S. B. (
1994
). Chondroitin sulfate proteoglycan elevates cytoplasmic calcium in DRG neurons.
Dev. Biol.
166
,
87
-100.
Srinivasan, J., Schachner, M. and Catterall, W. A. (
1998
). Interaction of voltage-gated sodium channels with the extracellular matrix molecules tenascin-C and tenascin-R.
Proc. Natl. Acad. Sci. USA
95
,
15753
-15757.
Talts, U., Kuhn, U., Roos, G. and Rauch, U. (
2000
). Modulation of extracellular matrix adhesiveness by neurocan and identification of its molecular basis.
Exp. Cell Res.
259
,
378
-388.
Ughrin, Y. M., Chen, Z. J. and Levine, J. M. (
2003
). Multiple regions of the NG2 proteoglycan inhibit neurite growth and induce growth cone collapse.
J. Neurosci.
23
,
175
-186.
Walz, A., Anderson, R. B., Irie, A., Chien, C. B. and Holt, C. E. (
2002
). Chondroitin sulfate disrupts axon pathfinding in the optic tract and alters growth cone dynamics.
J. Neurobiol.
53
,
330
-342.
Weber, P., Bartsch, U., Rasband, M. N., Czaniera, R., Lang, Y., Bluethmann, H., Margolis, R. U., Levinson, S. R., Shrager, P., Montag, D. et al. (
1999
). Mice deficient for tenascin-R display alterations of the extracellular matrix and decreased axonal conduction velocities in the CNS.
J. Neurosci.
19
,
4245
-4262.
Wu, Y., Sheng, W., Chen, L., Dong, H., Lee, V., Lu, F., Wong, C. S., Lu, W. Y. and Yang, B. B. (
2004
). Versican V1 isoform induces neuronal differentiation and promotes neurite outgrowth.
Mol. Biol. Cell
15
,
2093
-2104.
Yamagata, M., Herman, J. P. and Sanes, J. R. (
1995
). Lamina-specific expression of adhesion molecules in developing chick optic tectum.
J. Neurosci.
15
,
4556
-4571.
Yamaguchi, Y. (
2000
). Lecticans: organizers of the brain extracellular matrix.
Cell Mol. Life Sci.
57
,
276
-289.
Zacharias, U., Norenberg, U. and Rathjen, F. G. (
1999
). Functional interactions of the immunoglobulin superfamily member F11 are differentially regulated by the extracellular matrix proteins tenascin-R and tenascin-C.
J. Biol. Chem.
274
,
24357
-24365.
Zacharias, U., Leuschner, R., Norenberg, U. and Rathjen, F. G. (
2002
). Tenascin-R induces actin-rich microprocesses and branches along neurite shafts.
Mol. Cell Neurosci.
21
,
626
-633.
Zanin, M. K., Bundy, J., Ernst, H., Wessels, A., Conway, S. J. and Hoffman, S. (
1999
). Distinct spatial and temporal distributions of aggrecan and versican in the embryonic chick heart.
Anat. Rec.
256
,
366
-380.