Dab1 (Disable Homolog-1) Reelin Adaptor Protein Is Overexpressed in the Olfactory Bulb at Early Postnatal Stages

Dab1 mediates reelin signalling and plays critical roles in early brain development such as the stereotypical positioning of neurons in the brain. The olfactory bulb undergoes a prominent layering reorganization, but shows not apparent differences between wild type and reeler in the layer organization. Therefore, an accurate regional and cellular simultaneous analysis of these molecules becomes essential to clarify the role played by Dab1 upon Reelin effect. The present study reveals a strong and consistent Dab1 mRNA and protein expressions, throughout the olfactory bulb layers in both wild type and reeler mice. In addition, noteworthy is the pattern of Dab1 location within cell nuclei in both strains. Furthermore, a temporal increment of Dab1 expression levels is detected from P0 to P15 in both strains, being the protein quantity higher in reeler than in wild type mice. Altogether, our results revealed that Reln acts directly from projection neurons via the production of different Reln fragments. Changes in the pattern of Dab1 expression could reflect an alternative Reln function in postnatal and adult stages, besides a possible regulation of Dab1 by other molecules distinct to Reln.

At earlier embryonic stages, Reln and Dab1 mRNA become to be expressed in the forebrain [3,19], being the Dab1 mRNA regionally detected underneath cortical Cajal-Retzius cells [18,19]. In the OB, Reln protein is detected in deeper layers and from postnatal day 5 its expression is restricted to mitral and periglomerular cells [3,11,20]. Thus, although the distribution and relationship of Reln with its intracellular adaptor Dab1 has been fully addressed [3,17,[19][20][21][22], the relationship between the absence of Reln (reeler mice) and the presence of Dab1 is currently unknown . There is ample evidence on the layering alterations in the reeler OB [8,23], but the precise localization and changes of Dab1 protein, within the synaptic circuits of the OB, requires further investigation. Thus, we analyzed Dab1 mRNA and protein expression patterns during the early postnatal development of the OB comparing wild type (wt) with reeler mice. Moreover, using different glial and neuronal markers we characterized the molecular phenotype of Dab1 expressing cells. While immunohistochemical data revealed no differences in Dab1 distribution throughout the OB layers, western blot measurements showed differences between the two strains. Remarkably is the unexpected Dab1 expression in cell nuclei in the OB, suggesting alternative functions of Dab1 protein.

Animals
Wild type C57 (wt) and Reln mutant (reeler) mice were obtained from the Cajal Institute mouse breeding facility. All procedures followed the guidelines for animal care of the European Community Council (86/609/CEE) and were approved by the Bioethical Committee at the Spanish National Research Council (CSIC). Animals were deeply anesthetized by hypothermia (P0 and P3) or with i.p. equithesin at lethal dose (P7 and P15) and then transcardially perfused with different solutions according to the experiment. Reeler mice were genotyped by PCR.

Western blot
Anesthetized mice were transcardially perfused with saline, OB were dissected, stripped of meninges and homogenized using a Then, tissue was centrifuged (800 g, 15 min) to obtain the crude cytoplasmic fraction (CF) in the supernatant and the nuclear fraction (NF) in the pellet. NF was resuspended in homogenization buffer and centrifuged 700 g to obtain the crude NF. CF and NF protein concentrations were measured by Bradford method using a dye reagent for protein assay (BIO-RAD Laboratories GmbH, München, DE) and then, adjusted at the same quantity with 5X loading buffer (b-mercaptoethanol containing Laemmli buffer).  Then, 24 mg of protein were loaded per well in a 7.5% SDSpolyacrylamide gel using a Mini-PROTEAN electrophoresis system (BIO-RAD Laboratories GmbH). Molecular weights (MW) were determined with precision plus protein dual color standards (BIO-RAD Laboratories GmbH). Proteins were transferred to 0.22 mm Protran TM nitrocellulose membranes (What-man, GE Healthcare) for immunoblotting. Membranes were incubated with phosphate buffered saline (PBS) supplemented with 2% bovine serum albumin (BSA, Sigma-Aldrich Co) to block unspecific binding sites, and overnight incubated at 4uC with specific primary antibodies diluted in PBS-2% BSA (see Table 1). Antibody binding was detected using goat anti-mouse IgG and   goat anti-rabbit IgG peroxidase conjugate antibodies (Jackson Immunoresearch, West Grove, PA), diluted 1:10.000 in PBS for 2 h at room temperature (RT). Bands were visualized using ECL Plus western blotting detection reagents (Amersham GE Healthcare) and exposed to X-ray films (Amersham GE Healthcare). Binding of antibodies was stripped from membranes by Restore PLUS Western Blot Stripping Buffer (Thermo Scientific, Rockford, US). Protein levels were quantified by densitometry of bands, for which X-ray films were digitalized using a GS800 Densitometer (BIO-RAD Laboratories GmbH). OD levels for each sample were subtracted to the background and then Dab1 levels were normalized against load control proteins, b-tubulin and P62 for cytoplasm and nuclei Dab1 levels respectively. Statistical differences between Dab1 OD intensities (n = 3 for each strain and age) were assured applying a One-way and a Factorial ANOVAs respectively, followed by a post-hoc Newman-Keuls test.

Isolated OB nuclei staining
A portion of the nuclei suspension (see above) was used for intranuclear staining with different markers (Table 1). Nuclei were centrifuged at 800 g and then resuspended in ice-cool 4% paraformaldehyde (PFA) to fix at RT during 15 min. Then, nuclei were washed twice with PBS +0.1% TritonX100 (PBST) and 1 h incubated with rabbit anti-Dab1 at RT. Nuclei were washed twice with PBST and incubated 1 h with the specific secondary antibody Alexa 488 conjugated goat anti-rabbit IgG (Molecular Probes-Invitrogen, Leiden, NL). Nuclei were counterstained with Hoechst (1 mg/ml, Sigma-Aldrich Co). Staining was visualized using a confocal microscope TS5 (Leica).

In situ hybridization
Wt and reeler mice (n = 3 for each age and strain) were transcardially perfused with 1 ml/g of ice-cool 4% PFA in 0.1 M diethyl-pyrocarbonate (Sigma-Aldrich Co.) phosphate buffer treated (DEPC-PB). Then, OB were removed and postfixed overnight, cryoprotected in 30% sucrose in DEPC-PB and then sectioned 20 mm thick at the sagittal plane, using a TC1900 cryostat (Leica). In situ hybridization (ISH) was performed using an InsituPro VS device (Intavis Bioanalytical Instruments, AG) with the following protocol: sections were postfixed 10 min at RT with 4% PFA-DEPC, rinsed with PBS, digested 5 min at RT with 1 mg/ml proteinase K (Sigma-Aldrich Co.), fixed again 5 min and acetylated at RT for 10 min. After PBST permeabilization for 30 min, sections were pre-hybridized for 2 h with hybridization solution (1X Denharts, 50% formamide, 250 mg/ml baker yeast RNA, 500 mg/ml herring sperm DNA in 0.2X saline sodium citrate buffer (SSC) at RT. Dab1 probes were hybridized in a hybridization solution containing 350 ng/ml digoxigenin (DIG)-Dab1 RNA probe (preheated to 80uC for 5 min and iced) overnight at 72uC. Unspecified binding sites were blocked for 1 h in 10% NGS (Millipore) supplemented Tris buffered saline (TBS) and then incubated with alkaline peroxidase anti-DIG primary antibody (Roche, Mannheim, DE) at RT overnight. Antibody binding was detected using BM Purple AP substrate precipitating reactive (Roche).

Immunohistochemistry
Mice (n = 5 labeling for each age and strain) were transcardially perfused, first with heparinized saline solution and then with 1 ml/g of ice-cool 4% PFA. Brains were cryoprotected in 30% sucrose and 20 mm-sectioned at the sagittal plane, using a TC1900 cryostat (Leica). Frozen sections were air dried, permeabilized with PBST and boiled in 10 mM citrate buffer pH 6 to unmask the antigen. Unspecific antibodies binding was blocked with 0.1% PBST supplemented with 10% NGS +0.1% BSA for 1 h at RT and then incubated overnight at 4uC using specific primary antibodies, all diluted in 0.1% PBST +1% NGS (Table 1). Antibody binding was detected with the following secondary antibodies, diluted 1:1000 in PBST: Alexa 488 goat anti-rabbit IgG and Alexa 568 goat anti-mouse IgG (Molecular Probes-Invitrogen). RC2 antibody was detected by incubating 2 h at RT with biotin conjugated goat anti-mouse IgM (Jackson Immunoresearch) and then incubated 1 h at RT with Alexa 568 conjugated streptavidin (Molecular Probes-Invitrogen). Nuclei were counterstained with Hoechst (1 mg/ml, Sigma-Aldrich Co).
compartments. (c) Comparison of relative levels of Dab1 in CF and NF of wt and reeler mice. CF amount of Dab1 is statistically higher in wt at P15, while in reeler the levels are higher at P7 and P15. In NF, Dab1 levels are statistically higher exclusively at P15. (d) Comparison of Dab1 quantities between wt and reeler animals. Dab1 reeler content is significantly higher in CF at all studied ages. In NF this difference is restricted to P15. (* p,0.05; ** p ,0.01; *** p ,0.001; N = 3) doi:10.1371/journal.pone.0026673.g004 Staining was visualized using a confocal microscope TS5 (Leica).

Results and Discussion
Reln molecule translocates its signal activating phosphorylation of the intracellular adaptor protein Dab1 [24]. These two molecules are closely interrelated, so that one regulates the expression of the other [25,26]. Thus, Reln-Dab1 interaction is critical regulator of key steps during CNS development, acting in a time-dependent manner in different regions of the brain [19,20]. Then, regional and cellular quantified distribution of these molecules becomes essential for understanding their role during development. Since the OB is a microcircuitry, with a heterogeneous interneuron population generated throughout development and adulthood, it represents a perfect model to analyze layering and developmental cell maturation in a specific time windows. This led us to address in detail the Reln/Dab1 postnatal expression in reeler and wt mice and its phenotypical consequences.

Spatiotemporal expression of Dab1 mRNA and protein in the olfactory bulb
Localization of Dab1 transcripts was strikingly similar throughout the OB layers in both wt and reeler. Prominent, Dab1 expression was located in periglomerular and mitral cells, as well as in the granular cell layer (Fig. 1). Although our data revealed Dab1 transcripts in the whole glomeruli from P0 to P7, the expression pattern was similar to those presented in the Allen Brain Atlas [27] (www.brain-map.org).
As occurs with Dab1 transcripts, its protein also showed a homogeneous distribution throughout OB layers in both strains (Fig. 2). These similarities were unexpected, since Dab1 traslocates the Reln signal after is bounded to their receptors. This might represents an alternative role of Dab1 to the Reln pathway, as proposed in the rostral migratory stream cells where Dab1 is triggered by Trombospondin-1 and F-spondin [28,29]. Other possibilities could be related to the presence of Dab1 isoforms, similar to those phosphorylated independently of Reln in embryonic retinal cells [30,31] or by mechanisms similar to those underlying the GABAergic interneurons layering [32].
Specifically, Dab1 immunoreactivity was mostly located within three different OB regions (Fig. 2): i) in neuronal processes, primarily in dendrites of the glomerulus compartment (GL), and in both external and internal plexiform layers (EPL and IPL); ii) in cell bodies from both GcL and subependimal zone (SEZ), with an intense labeling in OB radial glial cells which slightly decreases from P0 to P15 (see below); iii) an unpredicted Dab1 expression in cell nuclei of MCL, EPL and GL, as well as in the neuroblasts that reach the OB through the rostral migratory stream. This agrees with the specific nuclear transporting sequences of Dab1 molecule, that facilitates the nuclear translocation describing Dab1 as a nucleocytoplasmic shuttling protein [33]. However, this specific nuclear location of Dab1 protein occurs when the nuclear exporting receptor CRM1 (able for binding to Dab1 sequences) is blocked with leptomycin B [34]. Our data suggest a new nuclear function for Dab1 molecule, since always appeared located in the nucleus, even in the absence of blocking agents for the nuclear exporting receptor CRM1. Moreover, the pattern of a clearly Dab1 nuclear labeling in mitral and periglomerular neurons in vivo sections makes possible its use as a specific marker of these cell populations.
Because the intriguing immunostaining pattern in the cell nuclei (Fig. 2), we further analyzed this intra-nuclear localization of Dab1 protein on isolated nuclei. As a result, Dab1 protein always appeared clustered in a dotted manner in both wt (Fig. 3 a, e, c, g, i, m, k, o) and reeler mice (Fig. 3 b, f, d, h, j, n, l, p). Thus, this nuclear pattern (Fig. a-d, i-l) correlated with the in vivo labeling (Fig.3 e-h, m-p).
Quantification of Reln and Dab1 levels were independently performed in both cytosolic (CF) and nuclear (NF) fractions of OB from wt and reeler tissues. Absence of proteins contamination in tissue extracts was guaranteed by detection of nuclear and cytoplasm proteins in CF and NF samples respectively (Fig.  S1a). In addition, cytoplasmic markers (Aconitase and GFAP) were also tested on isolated nuclei (Fig.S1b-g). While aconitase labeling completely avoid cell nuclei (Fig. S1b-d), GFAP shows a labeling circle specifically confined to the nuclear membrane ( Fig.  S1e-g). This was probably due to the nuclear membrane anchorage [35]. Reln western blot allowed the identification of immunoreactive bands of molecular weights of 450, 370 and 180 kDa (Fig. 4a), corresponding with the full length and two cleavage fragments of the molecule. Full length Reln molecule was detected at P0, the cleavage Reln fragments appeared at P3 and P7, while the 370 kDa fragment was missing at P15. These fragments probably corresponded to a time-course dependent cleavage mechanism, likely produced after C-terminus cleavage by extracellular metalloproteinases [36]. These results suggest that Reln selective fragments might act in a distinct development events via the activation of different metalloproteinases, as occurs for embryonic stages [37]. Further, Reln cleavage would signal cells for a correct positioning in the OB, as suggested for hippocampal granule cells [38].
With respect to cytoplasmic and nuclear Dab1 fractions, protein levels increased from P0 to P15 in both strains (Fig. 4a,  c). Statistical analyses of CF-Dab1 levels revealed and increment at P7 and P15 in reeler, whereas this increment was restricted to P15 in wt (Fig. 4a, c). By other hand, analyses of NF-Dab1 levels showed an increment at P15 in both strains (Fig. 4a, c). The quantity of CF-Dab1 protein was higher in reeler than in wt in all studied ages, revealing a protein accumulation in OB cells (Fig. 4b, d), while the NF-Dab1 level was just higher in reeler at P15. Dab1 increment in both strains strongly suggests an augmented Dab1 function, due to other signaling pathways distinct to the Reln. The higher levels of Dab1 in reeler than in wt could be due to the requirement of Reln to degrade Dab1 [26,39]. Thus, in absence of Reln, Dab1 should be accumulated in the cytoplasm, in an unphosphorylated state [16,19].
To strengthen the Dab1 nuclear and cellular processes localization with the antibody used in this study (from Sigma-Aldrich), we provided supplementary data (Fig. S2) using another anti-Dab1 (from Chemicon). Comparing the labeling of both antibodies, western blot revealed the presence of an 80 kDa band in tissue fractions (Fig. S2a) along with the immunolabeling of isolated nuclei (Fig. S2b). Unexpectedly, cell nuclei and glomerular processes were not labeled in olfactory bulb sections with anti-Dab1 from Chemicon (Fig. S2c). These differences could be related with the immunogens used to produce the antibody, which are different in length. Thus, the immunogen peptide sequence used by Chemicon (aminoacid residues from 400 to 555 of the mouse Dab1 protein, GeneID: 13131; UNIPROT number P97318.2) is longer than that used by Sigma-Aldrich (aminoacid residues from 538 to 555 of the rat Dab1 protein, GeneID: 266729; UNIPROT number Q8CJH2), and contains the total sequence. This could explain the differences in the epitopes recognized by both antibodies. Although both antibodies recognized correctly the Dab1 protein, we suggest that differences in labeling patterns could be related due to both antibodies recognize different epitopes in Dab1 isoforms.

Molecular characterization of Dab1-expressing cells in the olfactory bulb
To analyze the phenotypic profile of Dab1 expressing cells, we performed double immunohistochemistry with different neuronal and glial markers in both strains ( Table 2). Nuclei of mitral and some periglomerular cells, identified by Reln labeling at postnatal stages [3,20], coexpressed Dab1 at all ages ( Fig. 5a-d). Moreover, Dab1/NeuN coexpression pattern varied from P0 to P15 (Fig. 5ep). At P0-P3, double labeled cells were mainly located in the GcL-IPL along with some sparsely labeled cells in the MCL (Fig. 5e-j,  arrows). From P7 to P15, Dab1/NeuN cells were also located in EPL and GL (Fig. 5k-p arrows), which could be related to OB layer refinement [40,41]. This layering process is summarizes in the cartoons (Fig. 5 g, j, m, p). Combining Dab1 nuclear labeling with NeuN marker allows to observe superficial granular cells forming a monolayer just beneath the MCL. This correspond to the granule cells projecting to the most external part of the OB [42], and whose disposition was disrupted in reeler [8].
Coexpression of neuronal marker Map2a,b with Dab1 was observed in both, processes of the GL, EPL and IPL, and in neuronal somas of MCL-GcL at P0-P3, (Fig. 6a-b, f-g). From P7, Map2a,b/Dab1 colocalized in the proximal portions of large processes, mostly located in GL, EPL and IPL (Fig. 6k,l). However, some others Dab1 processes were negative for Map2a,b, but intensively labeled with the radial glial cell marker RC2 (Fig. 6c-e, h-j, m-o, r-t). This marker showed radial glial cell bodies mainly located in the SEZ with thin and large cell processes that extended to the most external OB layers at P0 (Fig. 6). From P3 onwards, RC2/Dab1 positive cells were located in upper OB layers, mostly surrounding the glomeruli, adopting a stellate morphology with shorter cellular processes than those observed at P0. From P7, none RC2 cell expressed Dab1. Specific neuronal markers, calbindin-CB, parvalbumin-PV and tyrosine hydroxylase-TH, identified the neurochemical phenotype of Dab1 positive cells in the glomeruli in both genetic backgrounds. Dab1 colocalized with CB and PV cells from P7 onwards (Fig. 7a-d, g-j, m-p, s-v), being negative in earlier ages. Dab1/TH coexpression was restricted to the nucleus at P0-P3 while at P7 the colabeling was detected in some TH-positive processes inside the glomeruli (Fig. 7e-f, k-l, q-r, w-x). None glial marker (GFAP, RIP) colocalized with Dab1 cells in both wt and reeler mice (Fig. 8).
In summary, our data revealed that Reln, secreted by mitral and periglomerular cells, suffered a cleavage process during the postnatal OB development probably related to cell lamination. Its main adaptor protein, Dab1, showed an undistinguishable transcript and protein expression patterns between wt and reeler mice. This suggests alternative roles in the Dab1 function not linked to Reln during the OB postnatal development. This idea is reinforced by the higher quantity of Dab1 in reeler respect to wt. Additionally, the increment in Dab1, from P0 to P15, suggests further roles of this molecule in the OB postnatal development. Lastly, the unexpected presence of Dab1 labeling in cell nuclei is probably linked to alternative Dab1 functions in the nuclear activity. Figure S1 Control western blot to test protein contamination in tissue fractions. (a) Immunodetection of cytoplasm and nuclear proteins in NF and CF respectively in OB extracts from both wt and reeler mice. (b) Immunohistochemistry on isolated nuclei with cytoplasm proteins at P0 (b, e), P7 (c, f) and P15 (d, g). By western blot are detected bands of nucleoporin P62 in the CF and a slight labeling of both aconitase and GFAP in the NF (a). Aconitase marker is completely absent in isolated nuclei (b-d),

Supporting Information
whereas GFAP labeling appears delimiting the nuclear membranes (e-g), Scale bar: 5 mm. (TIF) Figure S2 Dab1 expression using two different anti-Dab1 antibodies by western blot and immunohistochemistry. (a) Western blot of wt OB extracts using two anti-Dab1 antibodies from Chemicon and Sigma-Aldrich. Both labeled the specific 80 kDa band correspond to Dab1 protein. Antibody from Chemicon is not being able to detect the increase in protein levels from P0 to P15 as detected by Sigma-Aldrich antibody. (b) Labeling of isolated nuclei with both anti-Dab1 antibodies show a similar dotted pattern inside the nuclei at P0, P7 and P15. (c) Labeling of OB sagittal sections using the anti-Dab1 from Chemicon. In this case the labeling is broadly similar to that observed with the Sigma-Aldrich antibody, which is mainly found in periglomerular cells, MCL and in GcL. A difference respect to the Sigma-Aldrich antibody is the absence of nuclear and cell processes labeling with the Chemicon antibody. Scale bars: 100 mm and 1 mm in the inserts. (TIF)