Results

Annexin A7 is expressed in the early mouse embryo
First we examined the expression of Annexin A7 in ES cells (Bruce4, established from C57BL/6J mice) and the early stages of mouse embryonic development at the mRNA level by northern blot analysis and at the protein level by Western blotting and immunohistochemistry, respectively. Northern blot analysis shows in ES-cells and at embryonic days E7, E11, E15, and E17 two mRNA species of 1.8 kb and 2.4 kb, which result from alternative splicing in the untranslated 3'end (Fig. 1A) [11]. We found similar Annexin A7 mRNA levels in the four embryonic stages. Reprobing with a β-actin probe verified equal loading; the appearance of a faster migrating mRNA species in addition to the 2.0 kb species is characteristic for β-actin [12]. On the protein level, ES cells express only the smaller Annexin A7 isoform of 47 kDa (Fig. 1B, neuro-2a cells and heart tissue are included for control).
Figure 1  Expression of Annexin A7. (A) Northern blot analysis of RNA from ES-cells and embryos E7, E11, E15, E17 probed with a full length annexin A7 cDNA. Transcripts of 1.8 and 2.4 kb are detected in all stages examined (upper panel). Reprobing with a β-actin probe verified equal loading (lower panel, 2.0 kb band). (B) Murine ES cells (ES) express only the small Annexin A7 isoform. Both isoforms are expressed in neuroblastoma neuro-2a cells (N2A), heart (HRT) and brain tissue (CNS) from adult mice. Proteins from neuro-2a and ES cells, heart and brain were resolved by SDS polyacrylamide gel electrophoresis (12% acrylamide) and transferred to nitrocellulose membranes. The Western blot was probed with mAb 203–217 followed by incubation with a peroxidase coupled secondary antibody. Detection was with enhanced chemiluminescence. Immunohistochemistry confirmed the presence of the protein during early development. Annexin A7 immunoreactivity using mAb 203–217, which specifically recognizes both isoforms of Annexin A7, can be observed in the two cell types of the cylinder stage at E5 with a weaker Annexin A7 expression in the ectoderm (Fig. 2A–C). At E8 Annexin A7 can be detected in almost all tissues of the embryo (data not shown). A correlation between the presence of Annexin A7 and the origin of a cell type from one of the three germ layers is not apparent. We specifically noted the localization of the protein in the neuroepithelium of the neural fold and tube (Fig. 2D–F, proximal, E8; Fig. 2G–I, distal, E13). At the subcellular level, Annexin A7 was present in punctate structures mainly in the cytosol of all examined cells (Fig. 2E,F). This distribution was also observed during subsequent development.
Figure 2  Annexin A7 immunoreactivity in early mouse embryos. (A) Phase contrast, embryo E5: The egg cylinder consists of an inner cell mass (a) representing the ectoderm and an outer layer of endoderm cells (b). (B) Immunostaining of the paraffin section was performed using purified mAb 203–217 and Cy3-conjugated anti-mouse IgG. Annexin A7 is expressed in both cell types of the egg cylinder with a strong staining of the endoderm and a weaker staining of the ectoderm. The nuclei are devoid of immune reactions. (C) Negative control using the secondary Cy3-antibody only. (D-F) Annexin A7 expression in the proximal neural tube (D) and nearby neural fold (E,F), embryo E8, transverse section. Immunolabeling of Annexin A7 was performed with purified mAb 203–217 and visualization was with an Alexa Fluor 488-conjugated anti-mouse IgG. (D) An intense Annexin A7 immunostaining is detectable in the neuroepithelium of the neural tube (a, lumen of neural tube). (E,F) Higher magnifications of the neuroepithelium show that Annexin A7 is expressed in the cytosol. Arrowheads point to Annexin A7 immunoreactivity in the cytosol. (G) Phase contrast, embryo E13, caudal neural tube. (H) Immunostaining of the paraffin section was performed using purified mAb 203–217 and Alexa Fluor 488-conjugated anti-mouse IgG. (I) Negative control using the secondary Alexa Fluor 488-antibody only. Bar, 20 μm.

Annexin A7 changes its subcellular localization in the embryonic brain at E16
Next we analysed the cellular and subcellular distribution of Annexin A7 in the developing mouse brain (Fig. 3). At E13–E16 in transverse sections of the embryonic brain, most of the immunoreactive cells are present in the ventricular germinative zone surrounding the lateral ventricle (Fig. 3A, overview E16; Fig. 3H, overview E13). A closer inspection revealed that Annexin A7 is mainly localized in the cytosol at E13 (Fig. 3B,E). During further development the immature cells have rounded up two days later, but they retain Annexin A7 in the cytosol at E15 (Fig. 3C,F). At E16 we observed the first prominent nuclear staining of Annexin A7 (Fig. 3D,G) in cells of the intermediate zone (Fig. 3I, oval), which contains neurons radially migrating towards the growing neopallium cortex [13,14] and also glial cells forming the white matter of the adult cortex. Cells located marginally in the neopallial cortex also exhibit a nuclear stain. Labelling in the ventricular germinative zone is less prominent and the cytosol is no longer more strongly stained than the nucleus.
Figure 3  Subcellular localization of Annexin A7 in embryos E13, E15 and E16. Paraffin sections of embryonic brain were stained with purified mAb 203–217. Annexin A7 was visualized with Alexa Fluor 488-conjugated secondary antibody. (A) Overview, at E16 the immature GFAP-negative cells of the forming cerebral neocortex (b) surrounding the lateral ventricle (a) are strongly stained; square, a higher magnification of this area is given in (I). (B) Higher magnification of the earlier stage E13 (H) shows, that the cells are stained in the cytosol (arrowhead). (C) Two days later at E15 the cells have rounded up, and Annexin A7 stays in the cytosol. (D) At E16 the first nuclear staining becomes apparent (arrowhead) in cells of the intermediate zone located between ventricular germinative zone and marginal neopallial cortex as seen in (I), oval. (E-G) Confocal microscopy confirms the results in B-D. (H) Overview, at E13 the immature GFAP-negative cells are strongly stained; a, lateral ventricle. (I) Higher magnification of a cortical section of (A, square) demonstrating ventricular, intermediate, and marginal zones; oval, a higher magnification of this area is given in (D,G).

Mature neurons in adult mouse brain show an intense nuclear staining of Annexin A7
Both Annexin A7 isoforms are found in adult brain tissue (Fig. 1B). In general, we observed two characteristic staining patterns for Annexin A7, a prominent nuclear stain in neurons (determined by lack of GFAP-immunoreactivity, localization, morphology) and a cytoplasmic and nuclear stain in astrocytes (GFAP-positive) in all areas of the mature murine brain. In the neocortex (isocortex) Annexin A7 was strongly enriched in nuclei of neurons of all six cortical laminae (Fig. 4A, section derived from cortex temporalis). For control, in the adult brain type-1 astrocytes can be characterized by a positive GFAP-stain (Fig. 4B). Fig. 4F demonstrates the presence of nuclear Annexin A7 in a neocortical neuron of the external granular layer (layer II) and in an astrocytic cell of the neocortical molecular layer (layer I) which additionally exhibits an intense signal in the cytoplasm and in cellular branches. A further cell type of the isocortex showed a strong immunoreactivity in both, nucleus and cytosol, and was identified as pyramidal neuron based on morphology and absence of GFAP immunoreactivity. However, Annexin A7 apparently is more enriched in the nucleus (Fig. 4E).
Figure 4  Annexin A7 is present in neurons and astrocytes of the cortex temporalis and hippocampal formation of 10-weeks-old mice. (A) Low magnification of the cortex temporalis presents an Annexin A7 expression in cells of the pial border, in neurons of all six isocortical laminae, and a weak signal in the adjacent white matter. (B) Corresponding section stained with GFAP. (C) Staining in the Stratum pyramidalis (a) and in the dentate gyrus of the hippocampus; square, a higher magnification of acorresponding area is given in (G,H). An intense Annexin A7 immunostaining is detectable. (D) Corresponding section stained with secondary antibody only. (E) Presence of Annexin A7 in pyramidal neurons (lamina pyramidalis externa) of the isocortex temporalis. These neurons were identified based on their morphology, distribution and lack of GFAP staining. AnnexinA7 exhibits a punctate staining, which is pronounced in the nucleus (arrowhead). (F) Higher magnification of image (A) also shows Annexin A7 in nuclei of neurons (lamina granularis externa (corpuscularis), arrowhead) and in the cytoplasm and nuclei of astrocytes (lamina molecularis, arrow; GFAP-confirmed). (G) Higher magnification of the pyramidal neurons in the hippocampus confirms the presence of the Annexin A7 protein in the nucleus (arrowhead) of mature neurons. (H) To further confirm this, a similar section derived from an AnxA7-/- mouse was stained with the annexin specific antibody and lacked the nuclear signal. The residual stain of the tissue is unspecific, as it is also observed in controls of the AnxA7-/- brain omitting the primary antibody (data not shown). All paraffin sections were stained with mAb 203–217 (A, C, E, F, G) or anti-GFAP-antibody (B). The hippocampal control section (D) lacks the primary antibody. In the hippocampal formation we detected prominent Annexin A7 immunostaining in the stratum pyramidalis and in the dentate gyrus and also a weak astrocytic staining (Fig. 4C). The astrocytes are mainly localized in the area between cells exhibiting nuclear Annexin A7 staining and show the protein also in the nucleus and the cytoplasm (data not shown). When we analyzed the pyramidal neurons in the hippocampus at a higher magnification we found that Annexin A7 again mainly localized to the nuclei (Fig. 4G). Polyclonal antibodies gave a similar staining pattern (data not shown). The intense nuclear staining was absent in controls either stained only with the secondary antibody (Fig. 4D) or in brain sections derived from AnxA7-/- mice stained for Annexin A7 (Fig. 4H). The residual staining seen in the AnxA7-/- brain is unspecific as it is also present in corresponding negative controls lacking the primary antibody. The absence of Annexin A7 protein in brain and other tissues of the AnxA7-/- mouse has been verified in [10]. In another actual study of the knock out mouse we detected a thickened basal membrane and a widened intercellular space. This may cause unspecific binding of the secondary antibody.
An equal staining of Annexin A7 was also found in the cerebellum (Fig. 5A). Nuclei of neurons in the stratum granulosum show the most prominent staining (Fig. 5C,D). In the stratum moleculare, which is poor in cell bodies of neurons, only a few dots appear which also correspond to nuclei of neurons. Astrocytes are also positive for Annexin A7 in the nuclei and cytoplasm (data not shown). The signal of the pial border and the prominent stain of the white matter tracts (laminae medullares) are due to Annexin A7 positive astrocytes (Fig. 5 A–C). Higher magnification of the boundary layer between stratum moleculare and granulosum (Fig. 5D) revealed an Annexin A7 signal in nuclei, but also in the cytoplasm and at the plasma membrane of Purkinje cells (Fig. 5E). The typical pair of dendrites points to the margin of the cerebellar cortex. The specificity of the Annexin A7 signal was further confirmed in similar brain sections of the AnxA7-/- mouse (Fig. 5F). Fig. 5G shows Annexin A7 in neurites (axons) connecting the laminae medullares with the Purkinje-cell layer. Most likely these are the axons of the Purkinje-cells. Apart from these efferent neurites the laminae medullares (lamina alba, white matter) contain afferent mossy and climbing fibers. Thus, the intense stain of the cerebellar white matter arises from Annexin A7 located in astrocytes and neurites.
Figure 5  Annexin A7 immunostaining in the cerebellum of adult mice. (A) Low magnification of the cerebellum presents an Annexin A7 expression mainly in cells of the stratum granulosum and laminae medullares. (B) Corresponding section stained for GFAP. (C) Higher magnification of folia of the cerebellum, where a polyclonal anti-AnnexinA7 antibody was used; square, a higher magnification of acorresponding area stained with mAb 203–217 is given in (D). Between stratum granulosum and stratum moleculare the layer of Purkinje-cells (stratum neuronorum piriformium (ganglionare)) can be observed. (D) Staining of the band of Purkinje-cells (arrows). The positive Annexin A7-stain in the stratum granulosum is due to staining of the nuclei of neurons. (E) In addition to an intense staining of the nucleus, the cell body is AnnexinA7-positive including both dendrites of the Purkinje-cell shown. (F) Corresponding section from an AnxA7-/- mouse. (G) AnnexinA7 staining of axons (arrowheads) running from the laminae medullares to the Purkinje-cell layer located in the round end of a convolution. Sections A, D, E, F, G were stained with mAb 203–217.

Expression of Annexin A7 in the adult human isocortex
In the human parietal neocortex of aged individuals without any neuropathological alterations, subpial astrocytes exhibited a staining of Annexin A7 in the cytoplasm. A nuclear presence of Annexin A7 was limited to single astrocytes (Fig. 6B). Pyramidal neurons, predominantly those of layer V exhibited Annexin A7 at the plasma membrane of their perikaryon as well as of the apical dendrite (Fig. 6A,C). The neurons lacked a signal for Annexin A7 in their nuclei. Apical dendrites within the molecular layer also indicated a positive staining for Annexin A7 (Fig. 6B). The staining pattern of Annexin A7 in the human autopsy brain did not change after pre-treatment with trypsin.
Figure 6  Annexin A7 immunostaining in human isocortex. (A) Pyramidal neurons (bold arrow) and apical dendrites (small arrows) were clearly labeled with the antibody against Annexin A7 in the human parietal cortex. (B) A dendritic staining was also seen in the molecular layer (small arrows). In addition the subpial astrocytes were weakly labeled (double headed arrows). The staining was cytoplasmic, only few astrocytes showed a nuclear staining as well (inset). (C) Enlargement of (A): Annexin A7 was seen at the cell membrane of the perikaryon (bold arrow) and the apical dendrite (small arrows). (D) A blank control lacking the primary antibody did not show a specific labeling, but autofluorescent lipofuscin was detectable in the neurons (arrowheads). Bar, (A) 70 μm, (B) 50 μm, (B-inset, C) 17 μm, (D) 20 μm.

Presence of Annexin A7 in nuclei from neuronal and astroglial cells
The nuclear localisation of Annexin A7 in mice observed in immature cells at E16 and differentiated adult neurons could be only detected as a strong signal when the brain sections were pre-treated with trypsin before antibody staining. This procedure is thought to allow the antibodies to access epitopes masked by the formaldehyde fixation. Methanol fixation of cultured cells led to inconsistent results. Antigen retrieval in formalin-fixed and paraffin-embedded tissue sections employs various heating or proteolytic pre-treatment methods [[15-17]; Ein Handbuch für die Histologie, dianova GmbH, Hamburg, Germany]. These methods can result in moderate or strong specific antibody staining, but the detectability of other antigens might be decreased. Moreover, the optimal pre-treatment has to be individualised for each antigen.
To verify the presence of endogenous Annexin A7 in nuclei, we used a biochemical approach and purified nuclei from neuro-2a, PC-12, and C6 cells and treated them with a hypotonic extraction buffer to obtain the nucleoplasm (Fig. 7A). The nucleoplasm and the remaining nuclei, from which the nucleoplasm has been extracted partially, were subjected to SDS-PAGE and Western blotting. Western blots probed with mAb 203–217 showed Annexin A7 in the nucleoplasm of neuronal (neuro-2a, PC-12) and astroglial (C6) cell lines (Fig. 7A), however, the large isoform could only be clearly extracted with nucleoplasm from C6 cells. For control, antibodies against LAP2α, Emerin and tubulin were used to verify that the nucleoplasm (marker:LAP2α, which is anon-membrane-bound isoform of LAP2) was successfully extracted and also not contaminated by nuclear membranes (marker:Emerin) or by cytoplasm (marker:tubulin). Likewise, in these neuronal and astroglial cell lines Annexin A7 had been observed in the nucleus by immunofluorescence (Fig. 7B and data not shown). We noticed no difference between PC-12 and neuro-2a cells, however, as for the brain sections, we found an increase of the Annexin A7 staining after pre-treatment with trypsin.
Figure 7  Nuclear localization of Annexin A7. (A) Extraction of Annexin A7 from nuclei of neuronal (PC-12, N2A) and an astroglial (C6) cell line using a hypotonic buffer. Samples of total cells (c), and corresponding amounts of nuclear membranes (m) and nucleoplasm (p) were subjected to SDS-PAGE and Western blotting. Annexin A7 (AnxA7) isoforms are extractable from the nucleus of the indicated cells (47 kDa and 51 kDa isoform, arrowhead). For control, immunoblotting of LAP2α (75 kDa; non-membrane-bound isoform), Emerin (EM, 34 kDa), and tubulin (TB, 55 kDa) which are specific for nucleoplasm (LAP2α), nuclear membranes (Emerin), and the cytoplasm (tubulin) are shown. Note that the nucleoplasm is only partially extractable. (B) Immunofluorescence images of the PC-12 cells used. Cells were fixed with paraformaldehyde, permeabilized with Triton X-100, pretreated with trypsin as indicated, and stained with Annexin A7 specific mAb 203–217, DAPI in blue, bar 10 μm.