The vertebrate central nervous system (CNS) develops from a small homogeneous population of neural precursor cells, the neuroepithelium, of the embryonic neural tube. Members of the fibroblast growth factor (FGF) and Wnt family are known to play an important role in the proliferation and self-renewal of neural precursor cells and in the development of different neuronal and glial types (Vaccarino et al., 1999a; Ragsdale and Grove, 2001). FGF2 and FGF8 are particularly important in the early phases of patterning, proliferation, and neurogenesis in the developing brain. FGF2 stimulates the production of neurons and glia both in vitro and in vivo through effects on survival, proliferation, and differentiation (Murphy et al., 1990; Kilpatrick et al., 1995). Recent studies have shown that the major effects of FGF2 loss are found in dorsal regions of the brain (the cortex), where there is a loss of the large neurons in deep cortical layers (Raballo et al., 2000). FGF8 is a key signalling molecule in establishing patterning of the brain, in particular in the forebrain and at the midbrain–hindbrain junction (Heikinheimo et al., 1994; Crossley et al., 1996, Fukuchi-Shimogori, 2001 #1201). FGF8 has been directly implicated in regulation of proliferation of precursor cells in the midbrain (Lee et al., 1997; Shamim et al., 1999). The Wnt gene family of secreted glycoproteins also have important roles in the development of the central nervous system (Bally-Cuif and Wassef, 1995; Ikeya et al., 1997; Lee et al., 2000). Wnt 3A regulates precursor cell proliferation in the cortical hem, which gives rise to the hippocampus (Lee et al., 2000), and both Wnt1 and Wnt3A have been implicated in neural crest precursor cell proliferation (see Patapoutian and Reichardt, 2000).
FGF and Wnt signalling is regulated by heparan sulfate proteoglycans (HSPGs). HSPGs are localised to cell surfaces or within the extracellular matrix of tissues where they bind FGFs and many other HS binding growth factors, including the Wnts (Bernfield et al., 1999; Ornitz, 2000). Accumulating evidence demonstrates that HSPGs play a key role in shaping extracellular morphogen gradients (Wnts, bone morphogenetic and Hedgehog protein families) in tissues (Christian, 2000; Strigini and Cohen, 2000; Giraldez et al., 2002). Genetic studies have provided compelling evidence that HSPGs are essential for normal development (see Lander and Selleck, 2000). The HS chains play a key role in both orchestrating the formation of and also stabilizing the signalling by FGF:FGFR signalling complexes (Rapraeger et al., 1991; Kan et al., 1993; Guimond and Turnbull, 1999; Turnbull et al., 2001). Studies in Drosophila have demonstrated that HSPGs are involved in receptor complexes with the Wnt/Wingless ligands and Frizzled receptors that modulate the short- and long-range activities of Wingless (Lin and Perrimon, 1999; Tsuda et al., 1999; Baeg et al., 2001).
We were interested in determining which specific HSPGs are involved in stimulating neural precursor cell proliferation. Previous studies from our laboratory showed that perlecan, a member of the extracellular matrix family of HSPGs, was expressed in the developing neuroepithelium (Joseph et al., 1996). Other studies have also revealed that different members of the cell-surface families are expressed in the developing CNS (Bernfield et al., 1999; Erickson and Couchman, 2000; Yamaguchi, 2001). Recent data has provided support for a direct role of cell-surface membrane-bound HSPGs rather than basement membrane localised HSPGs in FGF signalling (Filla et al., 1998; Chang et al., 2000; Zhang et al., 2001). To determine which HSPGs could potentially be involved in signalling events regulating early brain development, we have comprehensively analysed the expression of these proteins during the proliferative phase in the developing mouse brain.
We have used reverse transcription—polymerase chain reaction (RT-PCR) and immunohistochemistry to determine the relative abundance and localisation of the syndecans and glypicans in the embryonic day 10 (E10) mouse brain. We also provide a comparative localisation of perlecan. Of the syndecans, syndecan-1 is the most abundantly expressed and localises exclusively to neuroepithelial cells in the ventricular zone. Syndecan-4 is expressed at a much lower levels, but also localises exclusively to cells in the ventricular zone. Of the glypicans, glypican-4 is the most abundant and localises to neuroepithelial cells in the ventricular zone and to early born neurons in the marginal zone. In comparison, we find perlecan localises to the basement membrane surrounding the neural tube and is not present within the neuroepithelium. We also demonstrate using in vitro culture that HSPGs are required for FGF-stimulated proliferation of brain precursor cells. Thus, the expression of syndecan-1, syndecan-4, and glypican-4 within the proliferative zone of the brain suggests that they may be involved in regulation of early brain development, such as FGF-stimulated proliferation.
RT-PCR Analysis of Syndecan Expression in the E10 Mouse Brain
We wished to determine which specific HSPG core proteins are expressed in the mouse brain at the time of neural precursor cell proliferation. In this particular study, we have compared the relative abundance of all currently known cell-surface localised HSPGs, the transmembrane syndecans (syndecans 1–4), and the GPI-linked glypicans (glypicans 1–6) at E10, a developmental stage at which the neural tube is predominantly composed of proliferating neural precursor cells. We initially performed RT-PCR expression analysis on the syndecans to determine which members were expressed. In addition, we incorporated a semiquantitative approach to obtain an appreciation of the relative abundance of the different family members. Our results (Fig. 1) show that all four syndecans are expressed in the E10 brain. Our analysis indicates that syndecan-1 mRNA is expressed at significantly higher levels than syndecan-3. This analysis also indicates that mRNAs for syndecan-2 and -4 are expressed at much lower levels than the mRNAs for the other two syndecans.
Comparative Immunolocalisation Studies Show That Syndecan-1 Localises to the Cell Surface of Neuroepithelial Cells
Whereas our RT-PCR analysis indicated that all syndecans were present and at distinctly different levels, this information was only semi-quantitative and lacked spatial resolution within the neuroepithelium. To obtain an accurate picture of syndecan protein levels and their relative localisations, we undertook a comprehensive immunohistochemical analysis of all four HSPGs. In addition, we compared their immunolocalisation with that of perlecan, because our previous studies had suggested that a perlecan-related molecule (Joseph et al., 1996) was expressed in the ventricular zone of the neural tube.
We initially compared syndecan-1 staining with perlecan. For these studies, we used antibodies that are well characterised and have demonstrated specificity for their respective HSPGs (Dziadek et al., 1985; Hayashi et al., 1987). To distinguish newly differentiated neurons from neuroepithelial cells, we also stained the sections with the early neuronal marker TUJ1 (Menezes and Luskin, 1994). Parasagittal sections of E10.5 mouse embryos were simultaneously immunostained with anti–syndecan-1 (red), anti-perlecan (rabbit polyclonal 906, coloured blue), and with TUJ1 (green) and examined using confocal fluorescence microscopy. Figure 2 shows that all neuroepithelial cells in the neural tube are positive for syndecan-1 (Fig. 2A,A′). High-power views of the neuroepithelium show a stronger degree of fluorescence at the luminal side of the neural tube when compared with the basal side. The luminal edge contains the most actively dividing cells and the rounded mitotic cells are intensely stained (Fig. 2A′, closed arrows indicate mitotic cells lining the lumen). The ectoderm is also strongly positive (Fig. 2A′). At this time in development, no TUJ1-positive cells are apparent in the telencephalon, as revealed by the lack of green cells in this region (Fig. 2A,A′). TUJ1-positive cells become apparent more caudally in the marginal zone of the midbrain and hindbrain (Fig. 2B–B′′,C–C′′). There are no cells coexpressing both syndecan-1 and TUJ1 present (which would be yellow). The TUJ1 positive neurons (green) are clearly distinct and, in fact, occupy areas that otherwise appear negative for syndecan-1 (arrows, compare Fig. 2B,B′′ with Fig. 2C,C′′). These observations suggest that the expression of syndecan-1 is restricted to neuroepithelial cells and is down-regulated when the cells differentiate into neurons.
Perlecan Localises to the Basement Membrane of the Neural Tube and the Overlying Ectoderm
Perlecan immunoreactivity (blue) is intense in the basement membrane that underlies the neural tube (Fig. 2A). It is not expressed on the neuroepithelial cells within the tube, nor is it present in the matrix surrounding these cells. It is also strong in the basement membrane of the overlying ectoderm (Fig. 2A,A′), where it appears pink in places due to the superimposition of the red immunofluorescence from the closely juxtaposed syndecan-1–positive ectodermal cells. Perlecan is also present in the mesodermal matrix outside the tube (open arrows Fig. 2A). Thus, whereas syndecan-1 is present throughout the neuroepithelium, perlecan is exclusively present in the basement membrane.
Syndecan-4 Localises in Discrete Puncta on the Surface of Neuroepithelial Cells
Syndecan-4 was immunolocalised by using an affinity purified goat polyclonal antibody (N-19, Santa Cruz) with demonstrated specificity for syndecan-4 (Paul Goetinck, personal communication; Saoncella et al., 1999). In our initial experiments, we also used a nuclear stain (SYTO Green, Molecular Probes, coloured blue) to visualise the nuclei of the densely packed neuroepithelial cells within the neural tube. Syndecan-4 immunostaining is present at low levels but in bright puncta that are predominantly found along the luminal edge of the neural tube (Fig. 2D, closed arrows). They are also present elsewhere in the neuroepithelium, the mesoderm, and the ectoderm overlying the brain (Fig. 2D, open arrows). These puncta are reminiscent of the immunostaining recently demonstrated for syndecan-4 when the HSPG was induced to cluster into raft membrane domains (Tkachenko and Simons, 2002). To examine the nature of the syndecan-4 staining in more detail, parasagittal sections of E10.5 mouse embryos were simultaneously immunostained with anti–syndecan-4, anti–syndecan-1 (to delineate the cell surface of the neuroepithelial cells), and also with TUJ1 (to distinguish neurons from neuroepithelial cells). Figure 2 E–E′′ shows high-power views of single optical sections through mitotic cells along the luminal edge of the neuroepithelium. Syndecan-4 (red) is present as a ring of bright red puncta on the surface of these cells (Fig. 2E, closed arrows) and often appear to be present in areas where syndecan-1 is either absent or very low (Fig. 2E–E′′, closed arrows). There are, however, some instances where the two HSPGs do appear to colocalise (yellow, open arrows, Fig. 2E′′). High-power views of neurons in the marginal zone reveal that, although syndecan-4 is present on the neuroepithelial cells, it is not present on neurons (Fig. 2F). The negative control for this series is presented in Figure 2G. Taken together, these observations suggest that the expression of syndecan-4 is restricted to the cell surface of neuroepithelial cells, but generally distinct from syndecan-1 and is down-regulated when the cells differentiate into neurons.
Immunolocalisation Studies Show That Syndecans -2 and -3 Localise to Neurons in the Developing Neural Tube
Parasagittal sections of E10.5 mouse embryos were also immunostained for syndecan-3 (red) by using a specific rabbit polyclonal antibody, Synd-3C (Hsueh and Sheng, 1999). The sections were simultaneously immunostained with TUJ1 to mark neurons (green) and SYTO Green (coloured blue) to reveal cell nuclei (Fig. 3A–C series).Syndecan-3–expressing cells (red) colocalise with TUJ1 (green) -positive cells and are present in the marginal zone throughout the neural tube: the midbrain, Figure 3B–B′′, and the hindbrain, Figure 3C–C′′. Furthermore, it would also appear from the merged images (Fig. 3B′′ or C′′) that this HSPG is predominantly localised to the processes emanating from the neuron cell body and is not present on the cell soma. Parasagittal sections of E10.5 mouse embryos were immunostained for syndecan-2 (red) by using a specific rabbit polyclonal antibody, Synd-2C (Hsueh and Sheng, 1999), and TUJ1 (green). Syndecan-2 immunoreactivity was only found in the marginal zone of the hindbrain (Fig. 3D–D′′), in a similar region to that depicted by box C in Figure 3A. Immunoreactivity was low and colocalised with TUJ1 (Fig. 3D′′ arrows, pale yellow neurons). Thus, our results indicate that syndecan-3 is strongly expressed on the subpopulation of newly differentiated neurons in the neuroepithelium at this age. Syndecan-2 has a more limited expression compared with syndecan-3 and is only expressed at low levels on a fraction of the neuronal subpopulation.
Thus, our studies indicate that syndecan-1 is the most widely expressed in the neuroepithelium and that syndecan-4, although also expressed by the neuroepithelium, is considerably more restricted in its localisation and relative abundance. Syndecan-3 is restricted to the neurons in the marginal zone but is abundantly expressed. In comparison, syndecan-2 is only found at low levels on a small subpopulation of neurons in the hindbrain. These immunohistochemical studies compare very well with the relative abundance obtained in the RT-PCR studies.
RT-PCR Analysis of Glypican Expression in the E10 Mouse Brain
Our RT-PCR analyses show that mRNA for five of six known glypicans are expressed in the E10 mouse brain (Fig. 4); glypicans 1–4 and 6, but not glypican-5. We showed we could detect a PCR product derived from glypican-5 mRNA by using our RT-PCR procedure on RNA isolated from postnatal brain (data not shown), which is known to express glypican-5 (Saunders et al., 1997). Our analysis indicates the glypican-4 mRNA is expressed at significantly higher levels than all other glypicans.
Characterisation of a Glypican-4 Antisera
To immunolocalise glypican-4 in the developing brain, we initially used five different rabbit antisera that had been raised against a series of peptides in human glypican-4 (Siebertz et al., 1999). Our preliminary immunohistochemistry studies showed that the anti–peptide-1 (SEQCNHLQAVFASR, amino acids 88-101) antisera gave strong staining in the developing mouse brain (data not shown). There is only one amino acid difference between the human and the mouse sequences for this peptide, V to I at position 97 (Siebertz et al., 1999), and glypican-4 is the only gene found in the entire mouse genome that contains this peptide sequence. We further characterised this antisera by using Western blotting. Whole E10 embryos were lysed and proteins from the clarified lysate were run on 4–20% gradient polyacrylamide gel electrophoresis (PAGE) gels, untreated (-) or treated with heparitinase I and heparinase/heparitinase III (+) to remove HS chains. The gels were blotted and probed either with affinity-purified anti–peptide-1 antisera or with 3G10, a monoclonal antibody that recognises the HS stubs generated by heparitinase treatment (David et al., 1992; Fig. 5A). Without heparitinase digestion (Fig. 5A), the anti–peptide-1 antibody reacts with a high molecular weight smear typical of HSPGs (arrow, > 200 kDa) and with two major bands at 62 kDa and 36 kDa. Heparitinase digestion results in loss of the high molecular weight smear and the appearance of a third band at 67 kD. This band is presumably the full-length core protein, as it becomes apparent upon heparitinase treatment and its size is consistent with full-length glypican-4 core protein (Watanabe et al., 1995). This band is also recognised by the monoclonal antibody 3G10 data only after heparitinase digestion (see Fig. 5, 3G10 + lane), which indicates that it is a core protein with a HS stub. The 36-kDa band has been reported previously for glypican-4 (Watanabe et al., 1995) and is believed to arise from a proteolytic cleavage site in the middle of the core protein. In the initial characterisation of glypican-4, c-myc tagged glypican-4 constructs were expressed in cell lines. When homogenates from these cell lines were analysed by sodium dodecyl sulfate (SDS) -PAGE gels under reducing conditions and Western blots probed with c-myc antibodies, this N-terminal 36-kDa fragment was the only band present. It was not apparent under nonreducing conditions (Watanabe et al., 1995).
To confirm this origin for the 36-kDa fragment detected with our anti–peptide-1 antisera, we undertook the following experiment. An E10 mouse embryo was placed directly into sample buffer immediately upon dissection and boiled, either with (R) or without (NR) reducing agents. This approach was done to minimize proteolytic degradation of the sample. Under reducing conditions (R), the 36-kDa band is the major band present and, under NR conditions, there is no band of this size (Fig. 5B). This evidence is in accord with the findings of Watanabe et al. (1995) and further demonstrates that the anti–peptide-1 antisera is specific for glypican-4. A high molecular weight smear is also present, probably corresponding with fully glycosylated glypican-4 (Fig. 5B, lane R, black filled arrows). The 62-kDa band, which was a major band in untreated E10 lysates (Fig. 5A, - lane), was only just detectable under our minimal proteolytic degradation conditions (Fig. 5B, lane R, open arrow). This finding indicates that it is also a proteolytic degradation product. That it is not detected by 3G10 (Fig. 5A, 3G10, – and + lanes) demonstrates that it does not contain HS stubs. This finding suggests that this fragment is likely to derive from a proteolytic cleavage event occurring in the region just N-terminal to the GAG attachment sites. Figure 5C is a diagrammatic illustration of the glypican-4 HSPG showing the location of the peptide-1 sequence and indicating the suggested proteolytic cleavage sites and proposed origin of the various fragments detected by Western analysis. There are two other minor bands at 74 kDa and 123 kDa, which are recognised by both anti–peptide-1 antibody and 3G10 (Fig. 5A, open arrows), but are not seen when proteolysis was minimized (Fig. 5B). It follows that these are likely to be degraded forms of the glypican-4 HSPG.
Taken together, these data demonstrate that the anti–peptide-1 antibody is specifically detecting glypican-4. Therefore, we have used this antibody, subsequently designated as anti–glypican-4, to localise glypican-4 in the E10 mouse brain.
Immunolocalisation Studies Show That Glypican-4 Localises to Neural Precursor Cells in the Ventricular Zone and Also to Neurons in the Marginal Zone in the E10.5 Mouse Brain
Parasagittal sections of E10.5 mouse embryos were simultaneously immunostained with anti–glypican-4 (red) and TUJ1 (green) and examined by using confocal fluorescence microscopy. Figure 6A shows the neuroepithelium of the neural tube in the mouse brain is strongly immunoreactive to anti–glypican-4. All cells are anti–glypican-4 immunoreactive (Fig. 6B). Glypican-4 localisation is similar to that of syndecan-1 (compare Fig. 6B with Fig. 2A′) and appears more intense at the luminal side of the tube and decreases toward the pial surface (Fig. 6B). However, in contrast to syndecan-1 immunostaining (Fig. 2A′), the ectoderm is only weakly stained. Again, no TUJ1-positive cells are apparent in the dorsal telencephalon as revealed by the lack of green cells in this region (Fig. 6A,B). TUJ1-positive cells become apparent more caudally in the marginal zone of the midbrain and hindbrain (Fig. 6C,D series, arrows). The merged images (Fig. 6C′′,D′′) reveal that the TUJ1 immunoreactivity colocalises with glypican-4 (yellow cells) in many, but not all, early neurons. These observations demonstrate that glypican-4 is expressed on the undifferentiated neural precursor cells of the neuroepithelium, as well as small populations of early neurons at the marginal zone of the E10 neural tube.
Parasagittal sections of E10.5 mouse embryos were also immunostained for glypican-2 (red) by using a specific rabbit polyclonal antibody, and TUJ1 (green; Fig 6. E–E′′). Previous studies have shown that glypican-2 localises exclusively to neurons (Carey et al., 1992; Stipp et al., 1994; Gould et al., 1995; Ivins et al., 1997; Kinnunen et al., 1998). At this age, glypican-2 immunoreactivity was only found in a restricted part of the hindbrain (similar region to that depicted by box D in Fig. 6A). In addition, expression was restricted to a subpopulation of neurons in the marginal zone (Fig. 6E′′). Thus, glypican-2 has a very restricted expression pattern at this age. Our immunohistochemical studies on these two members of the glypican family therefore compare very well with their relative abundance as determined by our RT-PCR studies.
Neural Precursor Cell Growth Is Blocked by Chlorate Inhibition of Sulfation and Rescued by the Addition of Exogenous HS
To demonstrate a functional role for the HSPGs on the neuroepithelial cells at this stage in development, we undertook in vitro culture experiments. We examined whether the FGF2-regulated proliferation of these cells (while well characterised) was dependent on HSPGs. We used sodium chlorate treatment to block sulfation of HS (Keller et al., 1989). The loss of sulfation on endogenous HS inhibits FGF signalling through its receptor (Guimond et al., 1993; Rapraeger and Ott, 1998). An examination of the morphology of the MAP2-stained cultures (Fig. 7A) showed chlorate concentrations 10 mM (and above) severely affected the growth of these cells and also their attachment to the substrate. At 10 mM chlorate, there are a few tight balls of cells remaining with occasional processes stretched taught between those that still remain attached to the substratum. These effects are not due to ionic strength imbalance from the addition of chlorate, as the addition of 40 mM NaCl to the cultures has no ill effects and the morphology of these cultures are identical to FGF2-treated cultures (data not shown). With decreasing chlorate concentrations (5 mM and less, data not shown), the cultures return to a morphology more characteristic of cells growing in FGF2.
The total cell counts for the different treatments are given in Figure 7B. In control cultures (see Fig. 7A, no FGF2 panel), an average of 350 cells were present after 2 days in culture. Addition of FGF2 resulted in a fivefold increase in cell number, consistent with our previous results (also see Fig. 7A, FGF2 panel). In cultures treated with both FGF2 and 10 mM chlorate, cell numbers were very similar to those in control cultures, indicating that FGF2 stimulation of cell growth is abrogated in the presence of chlorate. When cultures are treated with either sodium sulfate (10 mM, which restores normal sulfation of HS) or heparin (100 ng/ml, a highly sulfated structural analogue of HS), in addition to 10 mM chlorate, the cell numbers are similar to control cultures treated with FGF2 alone (Fig. 7B). The normal morphology of the cultures is also restored (see Fig. 7A, + heparin panel, results were also similar for + sulfate, data not shown). Overall, these data show that endogenous HSPGs are required for FGF2 stimulation of neural precursor cell growth.
We have determined the major HSPGs present during the proliferative phase of brain development. Our current analyses indicate that syndecan-1 and glypican-4 are the dominant cell surface HSPGs within the neuroepithelium. In addition to these HSPGs, a third HSPG, syndecan-4, is expressed at lower levels on neuroepithelial cells. The extracellular matrix (ECM) HSPG, perlecan localises exclusively to the basement membrane. Thus several cell surface HSPGs are candidate co-receptors for regulation of neural precursor cell proliferation.
Syndecans in the Early Mouse Brain
We have confirmed and expanded upon a previous study by Nakanishi et al. (1997) in which it was demonstrated that syndecan-1 was expressed in the developing rat neural tube. Our study is focussed on the mouse brain at a time when proliferation predominates. We find expression within the neuroepithelium and demonstrate that syndecan-1 is the predominant member of the syndecan family expressed at this time. We also demonstrate, by using the neuronal marker TUJ1, that syndecan-1 expression is exclusive to undifferentiated neuroepithelial cells. Our study also reveals a gradient of syndecan-1 with high expression on the luminal side decreasing to the basal side of the neural tube, which is particularly apparent in the dorsal telencephalon (from which the cortex arises).
The only other syndecan that is expressed in the neuroepithelium is syndecan-4 and its expression is much lower than syndecan-1. The localisation of syndecan-4 as puncta on the cell surface, particularly the neuroepithelial cells toward the luminal edge of the neural tube, is surprising and intriguing. This type of staining pattern is reminiscent of that shown in a recent in vitro culture study, where clusters of syndecan-4 were shown to be present in raft membrane domains (Tkachenko and Simons, 2002). Rafts are lipid-ordered microdomains, which have been suggested to act as platforms for signalling (Galbiati et al., 2001), and punctate staining has been demonstrated for several molecules known to be associate with rafts (Stuermer et al., 2001).
The two other syndecans, syndecan-2 and syndecan-3, are expressed in the developing brain but are exclusively neuronal. Syndecan-2 is present at very low levels and only on neurons in the hindbrain. Syndecan-3 is expressed at higher levels when compared with syndecan-2 and is found on neuronal processes throughout the neural tube. Our findings for syndecan-2 on early developing neurons may represent the initial expression of this HSPG, which later localises to dendrites and synapses (Hsueh and Sheng, 1999; Yamaguchi, 2001). Our data for syndecan-3 is consistent with previous studies that have shown that this HSPG is most prominently expressed in differentiated regions in the developing mouse and chick brain (Carey et al., 1992; Gould et al., 1995; Kinnunen et al., 1998; Hsueh and Sheng, 1999).
Glypicans in the Early Mouse Brain
Of the glypicans, we find that glypican-4 is the most abundant transcript in the neural tube at a time of predominant proliferation. Hagihara et al. (2000) also find glypican-4 by in situ hybridisation (Hagihara et al., 2000). Their studies show that, at E10, there is strong expression in the telencephalon with weaker expression in the hindbrain. Later (after E11), expression becomes restricted to the ventricular zone in the dorsal telencephalon. Our in vivo immunolocalisation shows strong immunoreactivity on the neuroepithelial cells throughout the head. Expression is highest at the luminal side of the tube, similar to syndecan-1. Thus, like syndecan-1, glypican-4 is most intensely expressed in regions of active cell division. Our data are, therefore, consistent with findings of Hagihara et al. (2000) in that glypican-4 is expressed on proliferating precursor cells. In cultures of E13 rat cortical precursor cells, Hagihara et al. (2000) also found that glypican-4 was predominantly expressed on proliferating precursor cells and a small proportion of newly differentiated neurons (TUJ1 positive cells) but was completely absent on more mature (MAP2-positive) neurons. Our studies reveal that, in more caudal regions of the brain where neuronal differentiation has just begun, glypican-4 staining is present on most of the newly differentiated TUJ1-positive neurons present.
We also found glypican-1 by RT-PCR expression analysis. However, it was at much lower levels compared with glypican-4. Although our RT-PCR analysis is only semiquantitative, our immunohistochemical analysis of the syndecan family demonstrates that the relative levels we see by using RT-PCR closely match the levels of the HSPG protein. Thus, it is probable that glypican-1 is expressed at lower levels than glypican-4. Previous studies show glypican-1 expression in the developing rat brain (Litwack et al., 1998). It is difficult to determine the precise location of glypican-1, but the immunoreactivity appears mainly along the luminal surface of the ventricular zone and there is diffuse low level staining through the neuroepithelium (Litwack et al., 1998). At later stages, it is expressed on neurons (Karthikeyan et al., 1994; Litwack et al., 1994, 1998). We obtained an antiserum against glypican-1 (Richard Margolis) but were unable to obtain specific immunostaining by using a variety of methods.
Glypican 2 (or cerebroglycan) is only expressed by neurons (Stipp et al., 1994; Ivins et al., 1997), and our immunolocalisation studies are consistent with these previous studies. Studies have shown that glypican-3 and glypican-5 are most abundant in the postnatal forebrain (Watanabe et al., 1995; Saunders et al., 1997). Our RT-PCR data did not detect expression of glypican-5 but did reveal moderate levels of glypican-3 in our E10 preparations of neuroepithelium. However, a detailed study using in situ hybridisation (Pellegrini et al., 1998) indicates that this HSPG is not expressed in the early mouse brain, but is strongly expressed in the mesenchyme surrounding the neural tube. Our expression data, therefore, may reflect some contamination of our neuroepithelial preparations by mesenchymal tissue. These cells could associate as loose cells with the neuroepithelial sheets we dissect for the preparation.
In summary, these data provide good evidence that glypican-4 is the dominant member of this family expressed in the neuroepithelium and that it is expressed by proliferating precursor cells. In addition, expression of this protein persists to an early stage of neuronal differentiation. Glypican-1 is also present at much lower levels and other studies show that it is expressed in the ventricular zone of the forebrain.
Perlecan in the Early Mouse Brain
Our previous studies suggested that perlecan was a major HSPG in the developing CNS (Joseph et al., 1996) and that it (or rather a perlecan splice variant called PRM for perlecan-related molecule) was localised both to the basement membrane and within the neuroepithelium. In light of our new findings, we have re-considered our original perlecan findings. First, our primary evidence that perlecan was expressed within the neuroepithelium was based on immunohistochemistry using a polyclonal antiserum called anti-PRM. This antibody was generated by using a crude HSPG preparation from a transformed neuroepithelial cell line (known as 2.3D); therefore, although the dominant HSPG may have been perlecan, it is probable that other HSPGs were present. Northern analysis of mRNA isolated from this cell line showed that, in addition to perlecan transcripts, syndecan-1 and syndecan-3 were also expressed (Joseph et al., 1996). Therefore, this antibody may not have been perlecan-specific but also detected other HSPGs. The anti-PRM antibody is no longer available; therefore, we cannot determine its full antigenicity. Second, characterisation of the anti-PRM antibody suggested it was detecting a 45-kDa perlecan variant in HSPG preparations from the 2.3D cell line. However, the well-characterised perlecan antibody used in the current study also recognised this perlecan variant (this antibody was called anti-PAb in Joseph et al., 1996), but showed no immunoreactivity within the neuroepithelium (in our current study). It follows that the immunoreactivity detected within the neuroepithelium by anti-PRM was not perlecan or a perlecan-related molecule. Finally, all other studies of perlecan show it to be a classic ECM HSPG, a major protein of the basement membrane and extracellular matrices (Noonan and Hassell, 1993; Iozzo et al., 1994). Taken together, these data are most consistent with perlecan localising exclusively to the basement membrane surrounding the neural tube.
HSPGs and the Regulation of Proliferation in the Early Mouse Brain
We have demonstrated that syndecan-1 and glypican-4 are strongly expressed in the neural tube when precursor cells are in a proliferative and self-renewal phase. Our immunohistochemical studies revealed a high concentration of syndecan-1 and glypican-4, and lesser amounts of syndecan-4, on the actively dividing cells at the luminal edge of the neural tube. At this time, the neuroepithelium is essentially one cell thick, but the nuclei are migrating up and down the cell body, between the lumen and pia, during the cell cycle. The cells round up and divide at the luminal surface during mitosis.
Both FGF2 and FGFR1 are localised to the luminal surface at this time (Raballo et al., 2000). FGF2 is thought to work at early G1 phase, when the cells are close to the lumen to promote commitment to a subsequent cell cycle (DeHamer et al., 1994; Raballo et al., 2000). In the current study, we have shown, by using chlorate inhibition of sulfation, that HSPGs are required for FGF2- stimulated proliferation of E10-derived neural precursor cells. Our studies are consistent with other studies within the central nervous system that have also demonstrated the importance of HSPGs for FGF2-induced responses of neural precursor cells for particular cell lineages (Bansal and Pfeiffer, 1994; Gomez-Pinilla et al., 1995; Caldwell and Svendsen, 1998). We also recently have demonstrated that the HS chains made by E10 neural precursor cells in vitro have the ability to potentiate signalling of FGF2 and FGF8 through specific FGF receptors (Ford-Perriss et al., 2002). Given that syndecan-1, glypican-4, and syndecan-4 are the only HSPGs expressed on neuroepithelial cells at this time, it follows that one or more of these HSPGs may be involved in FGF-regulated signalling events.
Several in vitro studies have demonstrated that each of these molecules can regulate FGF2 binding and signalling (Mali et al., 1993; Steinfeld et al., 1996; Filla et al., 1998; Hagihara et al., 2000). Several of these studies have also demonstrated that syndecan-1 potentiation of signalling occurs through FGFR1 (Steinfeld et al., 1996; Filla et al., 1998). Glypican-4 is able to bind FGF2 and when added to E13 precursor cells can moderately inhibit their proliferation in response to FGF2, presumably through sequestration of the FGF (Hagihara et al., 2000). In vitro culture studies (Volk et al., 1999; Horowitz et al., 2002) have indicated that syndecan-4 can mediate FGF2 signalling and Tkachenko et al. (2002) have suggested that the process of clustering into raft domains might play an important role in syndecan-4–mediated FGF2 signalling. These studies, therefore, suggest that all of the HSPGs we show expressed at this time in the neuroepithelium have the capacity to participate in FGF signalling.
We also demonstrate that perlecan is restricted to the basement membrane surrounding the neural tube at the early stages of development. There is no doubt that perlecan plays a key role in brain development, as approximately 40% of perlecan null mutant mice die at E10.5 (the age we are interested in) with defective cephalic development (Arikawa-Hirasawa et al., 1999; Costell et al., 1999). The principle effect of the loss of perlecan seems to be exencephaly, probably due to disruption of basement membrane integrity. Several studies have demonstrated that perlecan can regulate FGF2 activity through its HS chains (Aviezer et al., 1994; Sharma et al., 1998; Knox et al., 2002). It is unlikely, however, that perlecan is involved in neural precursor proliferation, because the extruded neuroepithelium in the perlecan null mutant mice continues to proliferate and differentiate (Arikawa-Hirasawa et al., 1999; Costell et al., 1999). Furthermore, from E10 onward in the developing brain, the basement membrane becomes progressively more separated from the dividing cells as the cortical plate and other cell layers become progressively thicker.
Current studies on the cell-surface HSPG null mutant mice are not informative with respect to a functional role in neural precursor cell proliferation. Syndecan-1 null mutant mice are apparently normal and fertile, and there have been no reports of defects in the brains of these mice (Alexander et al., 2000). Similarly, syndecan-4 null mutant mice are reported to be fertile and apparently normal, and there are no reports of defects in the brains of these mice (Echtermeyer et al., 1999). However, there is delayed wound repair and impaired angiogenesis in these mice, which has also been found for FGF2 null mutant mice (Dono et al., 1998; Ortega et al., 1998) and indicates a possible connection between these two genes. There have been no published studies to date on glypican-4 null mutant mice, although reports indicate that these mice are viable (Leighton et al., 2001; see http://socrates.berkeley.edu/∼skarnes). The brain histology has not been reported for any of these cell-surface HSPG mutant mice and, even if it appears superficially normal, it may be necessary to apply the kind of approach undertaken with the FGF2 null mutant mice to reveal more subtle effects (Vaccarino et al., 1999b; Raballo et al., 2000). Examination of the brains of these animals will reveal if there are any defects in precursor proliferation. If no phenotype is seen in the brains of individual nulls, then presumably more than one HSPG can participate and compensate for the loss of the other. Clearly, combinations of these gene deletions in a single animal will be necessary if this is the case. It seems unlikely that none of these HSPGs would be involved. Such studies should be informative and help to unravel the question as to which of these HSPGs are involved in regulation of early brain development.
Isolation of Neuroepithelial RNA From Mouse Embryos
Neural tubes containing the region encompassing the telencephalon through to, and including, the mesencephalon were dissected from embryonic day 10 (E10) mice. Total RNA was extracted from the dissected neuroepithelium by using Trizol (Invitrogen, Australia), according to the manufacturer's instructions. RNA was treated with DNA-free (Ambion, Genesearch, Australia) to remove residual DNA. RNA was quantified using the RiboGreen Quantification Kit (Molecular Probes, Bio Scientific, Australia), according to the manufacturer's instructions, and an fmaxfluorescence platereader (Molecular Devices, Sunnyvale, CA), by using filter set 485 nm:530 nm and software, as provided by the manufacturer.
RT-PCR Semiquantitative Expression Analysis
Sequences of the oligonucleotide primers used for RT-PCR of mouse syndecans 1–4 and glypicans 1–6 and the PCR product sizes generated are as follows: syndecan-1, forward (F)-5′-GACTCTGACAACTTCTCTGGCTCT, reverse (R)-5′-GCTGTGGTGACTCTGACTGTTG, internal (I)-5′-TGTGCTCCATGTAGAAGCAGAG, 319 bps; syndecan-2, F-5′-CAGGAGCTGATGAAGACATAGAGA, R-5′-ATGAGGAAAATGGCAAAGAGAA, I- 5′-GTTGAAGACACAAAGCATTACACC, 324 bps; syndecan-3, F-5′-TCGTTTCCTGATGATGAACTAGAC, R-5′-GTGGTAGATGTGGTGGTAGAGATG, I-5′-GAAAGAGGTCGTAGAAGAGTCCAG, 301 bps; syndecan-4, F-5′-ACCTCCTGGAAGGCAGATACTT, R-5′-GTGCTGGACATGGATACTTTGTT, I-5′-AACTGGAAGAGAATGAGGTCATTC, 294 bps; glypican-1, F-5′-GCTACATCTCCATCTTCCTTGAC, R-5′-AACACACATTATCCACTGACACC, I-5′-CTAGGGGATGCTGAGTTGCTATAT, 566 bps; glypican-2, F-5′-TACAGTTTCCCTCCTGATTTCCT, R-5′-ACCTTGGCTGACACCTTTACAC, I-5′-GGTGAAGATCTCAGAAGGTTTGAT, 482 bps; glypican-3, F-5′-GGATGGTGAAAGTGAAGAATCAAC, R-5′-GAGAGAAAGAGAAAAGAGGGAAAC, I- 5′-ATTAAGTAGGAGACTAACCGCGTG, 522 bps; glypican-4, F-5′-CATGGCACGCTTAGGCTTGCTCGC, R-5′-TGGTTGCACTGTTCGCTGACCACG, I-5′-CTGTCCCCAGGATTATACATGCTG, 279 bps; glypican-5, F-5′-CACGTGCTCCTGAACTTCCACTTG, R-5′-AACACAAAGCTGGTCAGCCAGGCC, I-5′-ACTTGCCAACAGAAGAAAAGAATT, 282 bps; and glypican-6, F-5′-GCTGTGATTCTTCCTCTCTCCGGG, R-5′-GTACAGCATCCCGTAGGTCCGGAC, I-5′-GGAAGACAAGCTGAGTCAACACAG, 378bps. cDNA was generated from 1 μg of total RNA by using the gene-specific reverse primer and the Superscript Preamplification System (Invitrogen). The syndecan family were assayed by using Platinum Taq polymerase (Invitrogen), whereas the glypican family required the use of 1×PCR Enhancer (Invitrogen) with Platinum Taq. PCR reactions (20 μl) were run at 2 mM MgCl2 in the following manner: 94°C for 1 min, followed by 30 cycles of 94°C for 30 sec, 60°C for 30 sec, and 68°C for 30 sec. A final synthesis step was performed for 10 min at 68°C. Ten-microliter samples were removed at 25 and 30 cycles and run on 1.2% agarose gel. Previous studies had shown that the amplification of the PCR product was still in the linear phase at 25 cycles. Southern analysis of the PCR products was as detailed in Ford et al. (1997). To quantify the expression levels of each gene, the intensities of the PCR product bands at 25 cycles were measured by exposing the blots to a PhosphorImager screen (Fujix Imaging Plate, Type BAS-IIIS) and using MacBasII software. The RT-PCR expression analysis was performed on three independent RNA samples. The values presented are the mean ± SD on three samples.
Western Analysis of Neuroepithelial Proteoglycans
E10 embryos were dissected and lysed in modified universal buffer (50 mM Tris pH 7.5, 150 mM NaCl, with detergents 0.2% Triton X-100, 0.3% Ige pal CA-630, and protease inhibitors 10 mM ethylenediaminetetraacetic acid (EDTA), 10 mM benzamidine) on ice. The lysate was spun at 10,000 × g for 10 min at 4°C and the supernatant removed to a fresh tube. Protein concentration was determined by using the Bio-Rad protein assay. For Western analysis, 10 μg were run treated (see below) or untreated with lyases on 4 to 20% SDS-PAGE gradient gels (Novex, Amrad Pharmacia Biotech, Australia). Gels were blotted onto polyvinylidene difluoride (Immobilon-P, Millipore) membranes, according to manufacturer's protocols and using the Novex Xcell II Mini Cell protein system. Protein markers were the MultiMark Multi-coloured standards from Novex or the biotinylated SDS-PAGE protein standards (broad range) from Bio-Rad. Blots were blocked in 5% skim milk in phosphate-buffered saline (PBS) + Tween 20 (0.1%) for 1 hr at room temperature. Primary antibody (anti–peptide-1/glypican-4 at 1:2,000, 3G10 at 1:250) was added for 1 hr at room temperature. Blots were washed in PBS + 0.1% Tween 20 and biotinylated secondary antibody (anti-rabbit and anti-mouse from Vector, 1:1,000) was added for 1 hr at room temp. Blots were washed in PBS + Tween 20 (0.1%), and streptavidin avidin–biotin complex (ABC) complex/horseradish peroxidase (HRP) was added for 1 hr at room temperature. Blots were washed in PBS + Tween 20 (0.1%), and immunoreactive bands visualised by using enhanced chemiluminescence (Amersham), according to manufacturer's instructions. Blots were exposed for between 10 and 30 sec, and images were scanned and processed by using Adobe Photoshop 5.5.
Heparitinase I and heparitinase III (heparanase) from Seikagaku (Japan) were reconstituted in 100 mM Na acetate, 100 μM Ca acetate, 1 mg/ml bovine serum albumin pH 7.0. For HS digestion, 10 μg of E10 embryo protein lysate was diluted into 100 mM Na acetate pH 7.0, 0.1 M Ca acetate plus protease inhibitors, 10 mM benzamidine, 10 mM EDTA, 5 mM N-ethylmaleimide, and 1 mM pepstatin to which 2.5 mU of Hep1 and 2.5 mU of Hep III were added and incubated overnight at 37°C. Proteins were precipitated by the addition of 3 × volumes of acetone, left overnight at −20°C, spun at 10K for 10 min at 4°C, followed by a wash with cold 100% ethanol, briefly dried, and resuspended in sample buffer for SDS-PAGE.
Mouse E10.5 embryos were fixed with 4% paraformaldehyde in PBS for 3.5 hr at room temperature. They were then immersed in 10% sucrose in PBS for 2 to 3 days at 4°C. After embedding in OCT compound (Tissue-Tek, Zakura, USA), 10-μm sections were cut on a cryostat and placed onto AES-coated slides (6%) and air-dried for several hours before immunohistochemistry. For immunostaining, all incubation steps and washes were performed at room temperature. Sections were blocked with CasBlock (Zymed, Australia) for 1 hr, and primary antibodies were added at the following dilutions: perlecan (906, rabbit polyclonal 1:25,000); anti–syndecan-1 (281-2, rat monoclonal, 1:500); anti–syndecan-2 (Synd-2C, rabbit polyclonal, 1:250); anti–syndecan-3 (Synd-3C, rabbit polyclonal, 1:250); anti–syndecan-4 (N19 Santa Cruz, goat polyclonal, 1:250); anti–glypican-4 (anti–peptide-1, rabbit polyclonal, 1:7,500; Seibertz et al, 1999); anti–glypican-2 (521-2, rabbit polyclonal, 1:750); and TUJ1 (BabCo, USA, mouse monoclonal, 1:1,000). All primary antibodies were incubated for 1 hr at room temperature. Slides were washed three times in PBS + 0.1% Triton X-100, followed by rinses in PBS and detected in the following manner: for syndecan-1, perlecan, and TUJ1 triple immunostaining, syndecan-1 was detected with Alexa Fluro 594 (Molecular Probes, BioScientific, Australia) goat anti-rat at 1:400; perlecan was detected by using Alexa Fluro 488 (Molecular Probes) goat anti-rabbit at 1:400; and TUJ1 was detected by using biotinylated horse anti-mouse (Vector, ESP Inc., TX) at 1:400, followed by FluoroLink Cy 5–labelled streptavidin (Amersham Biotech, UK) at 1:1,000. For syndecan-4, syndecan-1, and TUJ1 triple immunostaining, syndecan-4 was detected with Alexa Fluro 594 (Molecular Probes) donkey anti-sheep at 1:100, syndecan-1 was detected by using donkey FITC anti-rat at 1:100, and TUJ1 was detected as above. For DNA staining with SYTO Green (Molecular Probes), the manufacturer's instructions were followed. Sections were washed with PBS and cover-slipped with fluorescent mounting media (DAKO, Australia). Slides were viewed and imaged by confocal fluorescent microscopy (Bio-Rad). Images were processed and merged using Confocal Assistant software, as provided by manufacturer, Adobe Photoshop 5.5, and CorelDraw 10. For optimal presentation of colocalisation, TUJ1 immunoreactivity detected using FluoroLink Cy 5–labelled streptavidin (Amersham Biotech, UK) was pseudocoloured green and the DNA staining with SYTO Green (Molecular Probes) was pseudocoloured blue.
Chlorate Inhibition Experiments
Microtest plates (Terasaki plates, 60 well, Greiner) were coated overnight with 0.5 mg/ml poly-ornithine (Sigma, Australia). Wells were washed with PBS and then with water. The wells were then coated with 20 μg/ml laminin (Invitrogen) in HEM (Hepes buffered Eagles media), incubated at 37°C for 1 hr and then washed three times in HEM plus 10% fetal calf serum before various treatments. Cells were plated in Neurobasal (Brewer et al., 1993) supplemented with glutamine, 2% B27 and gentamicin (GibcoBRL Life Technologies, Australia). Neurobasal is also low in sulfate and NaCl, and is, therefore, an optimal medium for use in chlorate-induced inhibition of sulfation where low sulfate is required. E10 neural precursor cells were dissected according to Murphy et al. (1990) and plated at 1,500 cells per well (density of 3 × 105 cells per ml), with or without FGF2 at 10 ng/ml. Cells were cultured for 24 hr, and treatments were as follows: no FGF; FGF2 at 10 ng/ml; FGF2 with sodium chlorate at concentrations between 0 and 30 mM; FGF2 and chlorate at 10 mM and heparin at 100 ng/ml (determined by dose response experiments within the concentration range 1 μg/ml to 1 ng/ml); and FGF2 with chlorate at 10 mM and sodium sulfate at 10 mM. Cells were grown for a further 24 hr with treatments. Cells were then fixed with 1%(v/v) paraformaldehyde in MTPBS for 20 min at room temperature. The plates were then rinsed in MTPBS and fixed for a further 30 min at −20°C in methanol. After fixation, the plates were immunostained with anti-MAP2 (Sigma, monoclonal HM-2) and MAP2-stained cells were then visualised by using the streptavidin ABC/HRP complex and diaminobenzidine substrate chromogen (Dako). After staining, cells were briefly counterstained with haematoxylin to visualise nuclei of unstained (non-neuronal) cells for total cell counts. Neurons were counted separately from the blue-stained nuclei of non-neuronal cells, and the numbers summed for total cell counts per well. To determine total cell numbers in chlorate treated cultures, the cells were stained with haematoxylin only. Cultures were photographed by inverted light microscopy on a Leica DMIRB/E microscope and images scanned and processed by using Adobe PhotoShop. Results are the mean and standard deviation for the four wells and are representative of at least three separate experiments.
We thank Professor Merton Bernfield's laboratory for the gift of the syndecan-1 antibody, Professor Richard Margolis for the glypican-1 antisera, Professor Arthur Lander for the gift of the anti–glypican-2 antisera, Dr Yi-Ping Hsueh for the gift of the anti–syndecan-2 and -3 antisera, and Professor Paul Goetinck for advice regarding the syndecan-4 antibody. We would also like to thank Dr Heather Young and Yvette Wilson for advice regarding the confocal microscopy. This work is supported by grants from the National Health and Medical Research Council, Australia (M.F.-P. and M.M.) and the Medical Research Council (Senior Research Fellowship to J.T.) and The Royal Society (travel grant to S.G), United Kingdom.