Activation of the Wnt/β-catenin Signaling Reporter in Developing Mouse Olfactory Nerve Layer Marks A Specialized Subgroup of Olfactory Ensheathing Cells

Wnt reporter TOPgal activated cells were found in early embryonic olfactory system

We first observed characteristic X-gal staining for the βgal enzymatic activity at embryonic day (E) 12 in the front tips of rostral brains underneath the skull (red arrow in Fig. 1B). The X-gal stained signals were also reproducibly present in the cortical hem of the telencephalon and the developing facial structures (asterisks in Fig. 1B) in which Wnt signaling plays essential roles in developing neocortex (Zhou et al., 2006), hippocampal formation (Zhou et al., 2004b), and the orofacial primordia (Zhou lab, in preparation). We also found a tangentially oriented single-cell layer of the intensely immunolabeled βgal⁺ cells along the migratory route on the pia surface of OB anlage (green in Fig. 1C). At this stage, the olfactory axons immunolabeled with the antibodies to NCAM (neural cell adhesion molecule) extend from olfactory epithelium to the pia surface of OB anlage (red in Fig. 1C). NCAM is expressed by olfactory sensory neurons and their axons as well as OECs in the embryonic olfactory system (Aoki et al., 1995; Franceschini and Barnett, 1996). These βgal⁺ cells were restricted to a “connecting or docking zone” where the axonal bundles extend along the pia for future connections with CNS axons. At E14, we found the intensive βgal⁺ cells in the connecting zone between OB and the thick migratory mass (which consists of the migrating OECs intermingled with the olfactory axons) (Fig. 1D). All of these βgal⁺ cells in the connecting zone were co-immunolabeled with NCAM (Fig. 1C,D). In addition, we found that the tangentially oriented βgal⁺ cells were also co-immunolabeled with Nestin (arrows in Fig. 1E–F2). Nestin is expressed in neural lineage cells including OECs (Wang et al., 2007). We noted many Nestin⁺ cells in the middle region between two olfactory bulbs (Fig. 1E1,E2) and also inside the migratory mass (Fig. 1F1–F2). The tangentially organized βgal⁺ cells in these regions were co-immunolabeled with Nestin weakly or intensely (arrows in Fig. 1E–F2). To distinguish the tangentially organized OEC-like TOPgal labeled cells in the connecting zone from the classic OECs inside or surrounding the migratory mass, we arbitrarily propose a temporary name for these OEC-like TOPgal labeled cells in the “connecting or docking zone” between the olfactory migratory mass (the future olfactory nerve layer) and the OB at the early developmental stage as “connecting zone” (CZ) cells.

The above findings provide evidence that Wnt signaling is active in early embryonic olfactory neural lineage cells, particularly in a distinct cell population that is tangentially located in the connecting zone where olfactory sensory axons attached to the brain surface, and suggest that CZ cells are a unique subpopulation of OECs in which there is intense activation of the Wnt reporter TOPgal.

The TOPgal labeled “connecting zone” cells were found in the inner olfactory nerve layer during glomerulus formation and axonal connections

The development of glomeruli starts at E18 in the mouse OB (Valverde et al., 1992), when olfactory axons enter the OB from the ONL. We performed X-gal staining for E18.5 brains of TOPgal mice. X-gal staining that labels Wnt reporter activated cells was found in entire olfactory bulb surface with a sharp ending border between the OB and the major forebrain regions (Fig. 2A). To further assess the localization of X-gal staining, we prepared sagittal and transverse 150 mm vibratome sections of the OB obtained from E18.5 TOPgal embryos and stained them for βgal enzymatic activity. Intense X-gal staining was evident only on the surface layer of the OB on sagittal (Fig. 2A1) and coronal (Fig. 2A2) sections. This demonstrates that the canonical Wnt pathway is consistently active in the forming ONL during prenatal development.

From the immunolabeling on sagittal sections at E18.5, we found that the single-cell-layered βgal⁺ CZ cells expanded as multiple-cell-layer (arrowheads in Fig. 2B,Fig. 3A,3B) in the connecting zone of the migratory mass/forming ONL that was evident by NCAM-immunolabeling. We then examined immunolabeling for βgal on horizontal sections of the OB of P3 ~ P14 TOPgal mice (Fig. 2C–E1) and observed a significant change in the multiple-cell-layered βgal⁺ CZ cells in the ONL during glomerulus formation. The TOPgal labeled CZ cells were significantly increased in density and numbers with a peak around P3 to P10 but declined by P14 onwards when most glomeruli formed (Fig. 2C–E1). The OMP-immunolabeled olfactory axons were dramatically increased in the outer ONL from P3 to P14. No CZ cells were observed inside of the glomeruli in any of these developmental stages (Fig. 2C1–E1). However, we observed the TOPgal labeled CZ cells forming capsule-like structures around some glomeruli at P3 (Fig. 2C1). The CZ cell-capsules were less apparent at P5 (Fig. 2D1) and were not present in the OB of P14 mice (Fig. 2E1). We also examined the OB of P28 TOPgal mice and found small numbers of TOPgal labeled CZ cells in the ONL (data not shown). These findings suggest that the CZ cells are transiently activated by canonical Wnt signaling in the inner ONL, which may play important roles in related developmental processes.

TOPgal labeled “connecting zone” cells are a phenotypically unique subgroup of OECs in postnatal ONL that ensheath olfactory sensory axons connected with glomeruli

To determine the properties of TOPgal labeled CZ cells during late embryonic and early postnatal development, we double-immunolabeled βgal with representative lineage markers at E18.5 to P14 (Fig. 3,Fig. 4). On coronal sections of E18.5 olfactory bulb at higher magnification, we observed the CZ cells also changed in shape, with thick processes and sheet-like structures evident by intense βgal immunolabeling (Fig. 3A,B). Many of the βgal⁺ cells and processes were co-immunolabeled or tightly associated with NCAM in the forming inner ONL (ONLi) (Fig. 3A–B2). The low affinity neurotrophin receptor p75 has been shown to be specifically expressed in the OECs during development and can be used as a specific marker for the OECs in the olfactory system (Gong et al., 1994; Franceschini and Barnett, 1996). In contrast to that most βgal⁺ cells were located in the ONLi, the intense p75⁺ OECs were distributed in the ONLo with particular density surrounding the large axonal bundles (arrowhead and asterisk in Fig. 3C–C2). Some weakly labeled p75⁺ cells were found in the ONLi that were co-immunolabeled or tightly associated with NCAM (arrows in Fig. 3D–D2). We also detected some scattered βgal⁺ cells in the outer ONL that were intensely p75⁺ co-immunolabeled (arrows in Fig. 3E–E2). We further found that most βgal⁺ cells in the ONLi were co-immunolabeled with Nestin (Fig. 3F–F2). The p75⁺ OECs were also co-immunolabeled with Nestin (data not shown). These observations suggest that the CZ cells are a distinct subgroup of OECs based on their distributions and antigenic characters at E18.5.

S100β is a widely accepted OEC marker particularly in the ONLi (Au et al., 2002). At P10, we found that the βgal⁺ CZ cells were weakly co-immunolabeled with S100β in the inner ONL (Fig. 4A–C). At this age, we were able to recognize three distinct OEC subgroups in the ONL by combined immunoreactivities of P75 and S100β (Fig. 4D,L). In addition to p75⁺ OECs restricted in the ONLo, intense S100β⁺ OECs were distributed widely from the outer (where co-immunolabeled with p75) to middle (where p75 is negative) ONL (Fig. 4D), and very weakly S100β-immunolabled OECs were found in the inner ONL (arrows in Fig. 4D) where they were co-immunolabeled with Wnt reporter TOPgal (arrows in Fig. 4B,C). These data demonstrate that the TOPgal labeled CZ cells are a subgroup of OECs with down-regulated antigenic expression of p75 and S100β in the postnatal ONL. These OECs have extensive processes and sheet-like branches with intense βgal-immunolabeling that form a network in the inner ONL immediately adjacent to the glomerular layer (Fig. 4A,B,E–J). We further excluded the possibility that the majority of CZ cells were other neural lineage cells by co-immunolabeling of βgal with O4 for oligodendrocyte/Schwann lineage cells (Fig. 4E), GFAP for astrocytes (Fig. 4F), NeuN for neuronal lineage cells (Fig. 4G), F4/80 for microglia (Fig. 4H), and saw no co-immunolabeling except that a few CZ cell processes were tightly linked with GFAP⁺ astrocyte processes in the periglomerular region (arrow in Fig. 4F). In addition, we found that a small population of the TOPgal labeled CZ cells is proliferative by immunolabeling for the proliferating cell nuclear antigen (PCNA) at P7 (arrow in Fig. 4I).

Neuropeptide Y (NPY) is expressed in the ONLi OECs (Ubink and Hokfelt, 2000; Au et al., 2002). Although we could not find a good NPY antibody for co-immunolabeling with βgal in the ONLi, we detected intensive NPY mRNA expression that was overlapped with the βgal⁺ cells in the ONLi on the adjacent neighboring section (Fig. 4J,K). The distribution and antigenic properties of the βgal⁺ OECs (or OEC-like cells) and other OEC subgroups in the developing ONL were summarized for a better understanding (Fig. 4L).

To determine whether CZ cells actually ensheath olfactory sensory axons between ONLi and the forming glomerular layer, we performed confocal microscopy at P3~14 (Fig. 5). At P3, we observed a long extension formed by TOPgal⁺ CZ cell processes surrounding the edge of a forming glomerulus (green in Fig. 5A indicated by arrows in serial adjacent confocal images). At P5, we found the capsule-like CZ cell processes surrounding a developing glomerulus (arrows in Figs. 5B,C) and a tunnel-like structure wrapping a big olfactory sensory axonal bundle that projected to a glomerulus (arrowheads in Figs. 5D,E). At P14, we found that the CZ cell processes formed multiple tunnel-like structures bridging multiple axonal bundles from ONLi into a single glomerulus (arrowheads in Fig. 5F). In the adjacent area on the same section, we found the net-like processes of a single CZ cell wrapping up multiple axonal bundles that were cut transversely (No. 5 cell in Fig. 5F). Above observations definitively demonstrated that CZ cells are the specialized OECs with intense Wnt/β-catenin reporter activation in the ONLi, and their networked tunnel-like processes ensheath multiple olfactory sensory axonal bundles connected with individual glomeruli.

Wnt signaling components in the developing olfactory bulb

It has been reported that Wnt1, Wnt5a, and Wnt7a are expressed in the glomerular and mitral cell layers of postnatal olfactory bulbs of mice (P0~P20) (Shimogori et al., 2004). We have also confirmed the expression of Wnt1, Wnt3, and Wnt5a in the OB of P7 mice and observed their expression in the embryonic OB and OE (Supplement S1). Interestingly, while the Wnt signaling molecules are widely expressed in the OB with their expression domains occupying multiple OB layers, we have only observed the Wnt reporter TOPgal-activated cells in the ONL. Indeed, both Wnt receptor Fzd7 and transcription factor Lef1 have been exclusively expressed in the early postnatal ONL (Shimogori et al., 2004), which correlates well with the βgal immunolabeling in our study. On the other hand, Zaghetto et al. has demonstrated that more than 10 Wnts are detectable by RT-PCR in the embryonic OB and OE, and that the in situ expression of Wnt5b and Fzd7 are closely related to the BATgal labeled cells in the embryonic olfactory bulb (Zaghetto et al., 2007). Further studies will clarify which sets of Wnt signaling molecules are functionally activated in these OECs.

Wnt reporter activated OECs/CZ cells and their possible roles in the developing ONL

In this study we have observed TOPgal-activated cells in the OB primordium beginning around E12. Developing olfactory axons emerge from the OSNs in the OE and are accompanied by OECs, course the olfactory mucosa, and reach the prospective OB by E12 (Aoki et al., 1995). This timing coincides well with the activation of the Wnt signaling in the “connecting zone” (CZ), suggesting that the canonical Wnt signaling and its responsive CZ cells are the key for olfactory axon docking to a designated brain surface at the early developmental stage. The majority of olfactory axons form a dense band, the migratory mass, immediately below the OB around E14. We found that the Wnt reporter labeled CZ cells line up in front of the migratory mass around this age, suggesting that Wnt signaling may be also required to construct a permissive zone for the axons to enter the OB.

The glomeruli in the OB can be identified by P1, with the majority of the glomeruli developed by P14 and P21 (Valverde et al., 1992). We observed intense βgal staining indicative of Wnt signaling in presumably differentiated CZ cells in the ONL, particularly the inner ONL, with a peak around P3 to P10. From P14 onwards, the Wnt reporter labeled cells were decreased and restricted to a narrow zone immediately adjacent to the glomeruli. Olfactory axon sorting is thought to take place in the ONLi (Mombaerts, 2006). OSNs expressing the same odorant receptor are dispersed within a wide area of the olfactory epithelium. Therefore, thousands of axons from OSNs expressing the same odorant receptor are intermingled with many other axons, which do not express this receptor, in the axonal bundles of the olfactory nerve. The convergence of OSN axons into specific glomeruli requires the defasciculation of axonal bundles and their extensive sorting and rearrangement. Subsets of glial cells in the sorting zone and nerve layer of the moth, equivalent to OECs in vertebrates, are thought to be involved in the chemo attraction and repulsion of the OSN axons navigating the ONL (Tucker et al., 2004). Although the expression patterns of many chemoattractive and chemorepulsive molecules have been studied, the navigation and sorting of the OSN axons are still not understood (Nedelec et al., 2005). Recently, canonical Wnt/β-catenin signaling has been demonstrated to act as directional guidance cues for axons in anterior-posterior axis of the vertebrate spinal cord (Zou, 2004) and to control the migration of growth cones and neuronal cell bodies as well as neuronal polarity in anterior-posterior axis of worms (Hilliard and Bargmann, 2006). The Drosophila Wnt5 receptor Drl has been shown to function in glial cells of the olfactory system through its extracellular but not intracellular domain to antagonists Wnt5 to regulate midline crossing of sensory axon branches and the glomerular patterning in flies (Yao et al., 2007). Therefore, Wnts may also play an essential role in olfactory axon guidance in mammals, and the intense activation of canonical Wnt signaling in the ONLi may be directly responsible for OSN axon navigating and sorting into their designated glomeruli. These exciting possibilities are also supported by the fact that the canonical Wnt signaling is activated transiently and dominantly only in the front line of the olfactory axon projections during a critical developmental period but weakly in the normal mature olfactory system. Indeed, two recent studies using the TOPgal or a similar Wnt reporter BATgal transgenic mouse line have found unique Wnt-responsive cells in the embryonic OB that are similar to our CZ cells found in the TOPgal ONL, and have shown that canonical Wnt signaling in these cells may be essential for olfactory axonal connections in a tissue culture system (Zaghetto et al., 2007). This is not a surprise, as we have clearly demonstrated that the Wnt reporter TOPgal activated CZ cells are a specialized subgroup of OECs in the developing ONL. Embryonic CZ cells may contribute to guiding pioneering olfactory axons into the forebrain, and postnatally differentiated CZ cells may contribute to glomerulus formation and convergent sorting of olfactory sensory axons by their richly extended processes forming tunnel-like structures that ensheath and guide olfactory sensory axons into glomeruli. Further studies using genetic approaches will be required to address the role of canonical Wnt signaling in specialization or modulation of these OECs and their unique functions.

Phenotypic subpopulations of OECs in the ONL

Previous studies have demonstrated that several phenotypic subpopulations of the OECs reside in the ONL (Au et al., 2002). The OECs in the ONLi were also intensely immunolabeled by S100β in adult OB (Au et al., 2002). However, we noted that the intensity of S100β immunolabeling in early postnatal ONLi is much weaker than in adult ONLi or postnatal ONLo and ONLm. It has been reported that the OECs in the entire ONL were immunolabeled with the O4 antibody (Franceschini and Barnett, 1996). OEC antigenic phenotypes have been further described in vitro studies by OEC primary cultures derived from the postnatal OBs (Au and Roskams, 2003). However, we detected only a small number of O4⁺ cells in the ONL that were not co-immunolabeled with βgal⁺ OECs. Others have also reported that OECs in the ONLi can be immunolabeled with antibodies to GFAP and NPY. We have noted a few processes that may have been co-immunolabeled for βgal and GFAP, but we failed to find a good antibody for NPY to immunolabel OECs in the ONLi. However, we found that the mRNA expression of NPY in embryonic (Ubink and Hokfelt, 2000) and postnatal OB overlaps with the TOPgal activated ONLi cells. We provided partial evidence for the overlapped expression pattern of βgal immunostaining and NPY in situ expression on the adjacent neighboring sections.

Controversial identity of the Wnt reporter activated cells in the ONL

Two recent publications reported no co-immunolabeling of βgal of the Wnt reporter with any of their selected markers including Nestin and NPY in the ONL, and concluded mysteriously unknown identity of these cells (Zaghetto et al., 2007; Booker-Dwyer et al., 2008). Our results challenge these findings. For instance, we clearly demonstrate that βgal⁺ CZ cells were co-immunolabeled with Nestin at E14 and E18. Zaghetto et al. did not test this marker, but Booker-Dwyer et al. showed a negative result at P16. We suspect they would reach at the same conclusion as we have if they perform double immunolabeling for βgal and Nestin on embryonic OB sections. In addition, Booker-Dwyer et al. reported no co-immunolabeling of βgal with NPY at P16. However, the image from their Figure 2B might tell the positively co-immunolabeled cells with intense βgal but weak NPY immunoreactivities. Indeed, they may obtain a result comparable to ours if they perform NPY in situ hybridization. Furthermore, we have demonstrated the co-immunolabeling of βgal with S100β in the ONLi. Neither groups showed S100β in their papers. Finally, we definitively demonstrated that these βgal⁺ OEC processes ensheath the olfactory sensory axonal bundles in early postnatal ONL by high power confocal microscopy. We argue that the disagreements between these two groups and our conclusions are due to different selection of OEC markers, sample ages, and ways of detection. As a future plan, we will use a genetic fate mapping approach to demonstrate the lineage origin of these βgal⁺ CZ cells/OECs in the developing ONL.

In conclusion, using Wnt/β-catenin reporter TOPgal mice we have characterized a significantly specialized and dynamically developing OEC subgroup that may be directly involved in glomerulus formation and convergent sorting of olfactory sensory axons in the ONL.

Animals and Sampling

The Wnt reporter TOPgal transgenic mice were obtained from the Jackson Laboratories (USA) and were housed in the vivarium of the UC Davis Medical School (Sacramento, CA). Pregnant, timed-mated TOPgal mice were euthanized with CO₂ gas prior to cesarean section. The embryos were immersion fixed in 4% paraformaldehyde (PFA). Embryos were collected at embryonic day (E) 11.5~18.5; the day of conception is designated E0. Postnatal mice were perfused with 4% PFA in PBS, and the brains with intact olfactory bulbs were collected at postnatal day (P) 3~P28. All research procedures using mice were approved by the UC Davis Animal Care and Use Committee and conformed to NIH guidelines. The embryos and brains were frozen sectioned with a Leica cryostat.

X-gal staining

The TOPgal reporter activity was detected in whole-embryos of E12.5 and brains of E18.5 stained for 18 hours with 5-bromo-4-chloro-3-indolyl-D-galactopyranoside (X-gal) substrate solution (1 mg/ml X-gal in 5 mM K₄Fe(CN)₆, 5 mM K₃Fe(CN)₆, and 5 mM MgCl₂) for β-galactosidase activity. E18.5 brains were also embedded in 3% agarose and sectioned at 100 µm with a Vibratome (Leica VT1000S, Leica Microsystems, Germany) for X-gal staining.

Antibodies and Immunohistochemistry

The tissue sections were preincubated in a blocking solution of 10% lamb serum in PBS for 2 hours at room temperature to block nonspecific binding. They were then reacted with one or two of the following antibodies: (1) The rabbit antiserum to β-galactosidase (1:1000, Cappel, MP Biomedicals #55976, Solon, Ohio) was produced by hyperimmunizing rabbits with the enzyme β-galactosidase from E. coli, and the antibody was tested on Western blot to ensure specificity; the antibody detected a single band of β-galactosidase with no reaction to other E. coli proteins (manufacturer’s technical information). We actually tested the specificity of this antibody on wild type mouse sections and obtained no any immunolabeling (data not shown). (2) The mouse monoclonal (IgG2a) to β-galactosidase (1:10, Hybridoma Bank #JIE7, Iowa, IA) was prepared against E. coli β-galactosidase. This antibody was developed by Mason T.L. and Partaledis J.A. at the University of Massachusetts. The binding affinity and specificity seem a little less comparable with the Cappel’s rabbit anti-β-galactosidase, and we used it only for double immunolabeling with p75 (as it was a rabbit antibody) and obtained an identical staining pattern. (3) The rabbit antiserum to p75 nerve growth factor receptor (1:100, Chemicon #AB1554, USA) was prepared against extracellular fragment generated from sequences from the third exon of mouse p75 (amino acids 43–161). We obtained an identical immunolabeling pattern in the olfactory system with the published data (Ubink and Hokfelt, 2000; Balmer and LaMantia, 2005). (4) The mouse monoclonal (IgG2a) to S100β (1:100, Abcam #ab16959, Cambridge, MA) was prepared against full length native purified cow protein; the antibody stains a single band of ~10 kDa on Western blot (manufacturer’s technical information). We obtained an identical immunolabeling pattern in the olfactory system with the published data (Ubink and Hokfelt, 2000). (5) The mouse monoclonal (IgG1) to Neural Cell Adhesion Molecule (NCAM) (1:10, Hybridoma Bank #5B8, Iowa, IA) was prepared (by Jessel TM and Dodd J at the Columbia University) against NCAM from rat embryonic spinal cord membranes; the antibody cross-reacts to mouse and rat NCAM (manufacturer’s technical information). We obtained an identical immunolabeling pattern in the olfactory system with the published data (Aoki et al., 1995; LaMantia et al., 2000; Balmer and LaMantia, 2005). (6) The goat antiserum to olfactory marker protein (OMP) (1:1000, Wako Chemicals #544-10001, Richmond, VA) was prepared by multiple immunizations with rodent OMP; this antiserum is highly specific for mature olfactory neurons and their axons in tissue sections of many vertebrate species including rodents (manufacturer’s technical information). We obtained an identical immunolabeling pattern in the olfactory system with published data (Bailey et al., 1999; Treloar et al., 1999; Balmer and LaMantia, 2005). (7) The rabbit antiserum to glial fibrillary acidic protein (GFAP) (1:100, DakoCytomation #Z0334, Denmark) was prepared against GFAP isolated from cow spinal cord; the antibody shows one distinct precipitate (GFAP) with cow brain extract and cross-reacts with GFAP in mouse (manufacturer’s technical information). We obtained an identical immunolabeling pattern with published data (Bailey et al., 1999; Saghatelyan et al., 2004). (8) The mouse monoclonal (IgG1) to neuronal nuclei (NeuN) (1:1000, Chemicon #MAB377, USA) was prepared against purified cell nuclei from mouse brain; the antibody recognizes 2–3 bands (46–48 kDa) on a Western blot (manufacturer’s technical information). (9) The mouse monoclonal (IgG1) to Nestin (1:10, Hybridoma Bank, Rat-401, Iowa, IA; developed by Hockfield S at Yale University) was prepared against homogenized SD rat spinal cord; the antibody detects a single 200 kDa band on a Western blot (manufacturer’s technical information). We obtained an identical immunolabeling pattern in both PNS and CNS (Hockfield and McKay, 1985; Friedman et al., 1990). (10) The rat monoclonal (IgG2a) affinity purified antibody to mouse F4/80 antigen (1:10, eBioscience #14-4801, clone BM8, San Diego, CA,) was prepared against purified mouse F4/80 antigen; the antibody detects 140 kDa band on a Western blot (manufacturer’s technical information). (11) The mouse monoclonal (IgM) antibody to the oligodendrocyte marker O4 was prepared against homogenated of white matter of corpus callosum from bovine brain (1:1000, Chemicon #MAB345). (12) The mouse monoclonal antibody (mAb) to PCNA (proliferating cell nuclear antigen) was produced by immunizing mice with a recombinant protein A-PCNA fusion protein obtained from pC2T (1:1000, Cell Signaling #2586, Beverly, MA, USA). Dilutions of all primary antibodies were made in 1% lamb serum in PBS. After overnight incubation at 4°C with primary antibodies, sections were rinsed 3 times for 10 minutes each with PBS and incubated for 2 hours at room temperature with the secondary antibodies. All secondary antibodies were diluted in 1% lamb serum in PBS and used at dilution of 1:1000. The following secondary antibodies were used: (1) Alexa Fluor 594 goat anti-rabbit IgG, (2) Alexa Fluor 488 goat anti-mouse IgG, (3) Alexa Fluor 594 goat anti-mouse IgG, (4) Alexa Fluor 568 donkey anti-goat IgG, and (5) Alexa Fluor 488 donkey anti-rabbit IgG (all secondary antibodies from Molecular Probes, Inc., Eugene, OR). After three 10-minute washes with PBS, sections were counterstained with DAPI (1 mg/ml) and mounted with Vectashield (Vector Laboratories, CA).

Fluorescence and confocal microscopy

All tissue sections were assessed using a Zeiss Axiophot 2 fluorescent microscope equipped with an AxioCam digital camera (Carl Zeiss, Inc., North America). Confocal microscopy was carried out with a laser-scanning spectral confocal microscope system (Nikon Eclipse TE2000-E2).

In situ hybridization

Nonradioactive in situ hybridization histochemistry was performed using digoxigeninlabeled riboprobes as described previously (Zhou et al., 2004a; Zhou et al., 2004b; Zhou et al., 2006).