Analysis of Cdh22 expression and function in the developing mouse brain


  • Jonna Saarimäki-Vire,

    1. Department of Biosciences, University of Helsinki, Helsinki, Finland
    2. Institute of Biotechnology, Viikki Biocenter, University of Helsinki, Helsinki, Finland
    Search for more papers by this author
  • Annamari Alitalo,

    1. Institute of Biotechnology, Viikki Biocenter, University of Helsinki, Helsinki, Finland
    Search for more papers by this author
  • Juha Partanen

    Corresponding author
    1. Department of Biosciences, University of Helsinki, Helsinki, Finland
    2. Institute of Biotechnology, Viikki Biocenter, University of Helsinki, Helsinki, Finland
    • Department of Biosciences, P.O. Box 56, University of Helsinki, Helsinki 00014, Finland
    Search for more papers by this author


Classical cadherins are important cell adhesion molecules specifying and separating brain nuclei and developmental compartments. Cadherin-22 (Cdh22) belongs to type II subfamily of classical cadherins, and is expressed at the midbrain-hindbrain boundary during early embryogenesis. In Fgfr1 mutant mouse embryos, which have a disturbed midbrain-hindbrain border, Cdh22 is down-regulated. Here, we studied expression of Cdh22 in developing mouse brain in more detail and compared it to expression of related family members. This revealed both complementary and overlapping patterns of Cdh22, Cdh11, Cdh8, and Cdh6 expression in distinct regions of the forebrain and midbrain. We used a mutated allele of Cdh22 to study its function in brain development. Loss of Cdh22 caused reduced postnatal viability. Despite strong Cdh22 expression in the developing brain, we did not observe defects in compartmentalization or abnormalities in the midbrain and forebrain nuclei in Cdh22 mutants. This may be explained by functional redundancy between type II cadherins. Developmental Dynamics 240:1989–2001, 2011. © 2011 Wiley-Liss, Inc.


Cadherins are a large family of calcium-dependent cell adhesion molecules (Takeichi,1995). In the brain, they participate in cell sorting, boundary formation, generation, and remodelling of synapses and neuronal circuits (Redies and Takeichi,1996; Takeichi,2007). Cadherins have been classified to classical cadherins, desmosomal cadherins, protocadherins, and other cadherins. The classical cadherins are involved in homophilic binding, and they are divided in two subgroups: type I and type II cadherins (Nollet et al.,2000). These two groups have similar basic protein structures consisting of five extracellular, immunoglobulin-like cadherin-binding domains (EC1-5), a transmembrane domain, and an intracellular catenin-binding domain. The binding specificity of these two groups is mainly provided by EC1. In type I cadherins, the EC1 contains tripeptide HAV, whereas type II cadherins have corresponding tripeptide sequences, which are important for the homophilic cadherin interactions.

Cell adhesion has been shown to be important for establishing and maintaining cellular boundaries between brain regions. Also, several cadherins have been shown to be involved in boundary formation. For example, cadherins are expressed differently in certain brain compartments or nuclei as seen for Cdh6 in rhombomeres (Inoue et al.,1997). Moreover, R-cadherin and Cdh6 are guiding cell sorting in telencephalon between the forming striatum and cerebral cortex (Inoue et al.,2001).

At the midbrain–hindbrain border, cells form a compact signalling center called the isthmic organizer (Wurst and Bally-Cuif,2001). The isthmic organizer expresses signalling molecules Wnt1 and Fgf8 in discrete cell populations in posterior midbrain and anterior hindbrain, respectively. In addition, our earlier studies identified a narrow boundary cell population at the midbrain–hindbrain border (Trokovic et al.,2005; Kala et al.,2008). These boundary cells are found at both sides of the midbrain–hindbrain border defined by Otx2/Gbx2 expression, and they have a unique gene expression profile. Compared to the surrounding cells, the boundary cells express less positive cell cycle regulators, such as CyclinD1 and CyclinD2, but more inhibitory regulators, Jumonji and p21. This correlates with slower cell proliferation at the boundary. Loss of Fgfr1 function and FGF signalling at the boundary disrupts the coherent midbrain–hindbrain border, possibly due to a cell-sorting defect (Trokovic et al.,2003). Interestingly, Cdh22 (also known as PB-cadherin; Sugimoto et al.,1996), was found to be expressed in the boundary cells and down-regulated in Fgfr1 mutants (Trokovic et al.,2003; Kala et al.,2008). Thus, Cdh22 is a candidate cell adhesion molecule regulating boundary cell properties and compartmentalization at the midbrain–hindbrain border.

Cdh22 belongs to the type II subfamily of classical cadherins. It has been shown to be expressed at the midbrain–hindbrain boundary, forebrain, and limb buds during early embryogenesis (Kitajima et al.,1999). Later in development and in postnatal brain, it is expressed in cerebellum and several forebrain areas such as hypothalamus, amygdala, several nuclei in subcortical areas, hippocampus, and some neocortical layers (Mayer et al.,2010). In addition, it has been reported that short-type Cdh22, which lacks the catenin-binding domain and is generated by alternative splicing, promotes survival of gonocytes in neonatal rats (Wu et al.,2005). Cdh22 might participate in the regulation of cell survival and maintenance of stem cell population also in other tissues. However, related type II cadherins might be expressed in the same cell populations and have redundant functions.

In this report, we describe the expression pattern of Cdh22 in the developing brain in detail and compare it to related family members. In addition, we used a deletion allele of Cdh22 (Turakainen et al.,2009) to study its functions in the developing mouse brain. We found that Cdh22 deficiency results in reduced post-natal survival. However, we observed no changes in early patterning of the midbrain–hindbrain border, and Cdh22-expressing brain nuclei appeared normal throughout the brain in Cdh22del mutants. This might be explained by functional redundancy between type II cadherins during development of these brain nuclei.


Expression of Cdh22, Cdh11, Cdh8, and Cdh6 in the Midbrain and Diencephalon of Early Embryos

Cdh22 is down-regulated in conditional Fgfr1 mutants, which show aberrant cell mixing at the midbrain–hindbrain boundary (Trokovic et al.,2003). Therefore, we wanted to study more closely the spatio-temporal pattern of Cdh22 expression in the midbrain and anterior hindbrain. We observed that Cdh22 expression started at E8.0 in same region with Fgf8 and Wnt1 (Fig. 1A–F), and formed an expression gradient in the midbrain-hindbrain boundary (Fig. 1G, H, J, and M). We also observed that Cdh22 was co-expressed with a cell-cycle inhibitor p21 in a narrow boundary cell population, which has been shown to proliferate slower than the surrounding cells (Fig. 1J–O, Trokovic et al.,2005). Another type II cadherin, Cdh11 (also known as OB-cadherin), was also expressed in this region (Fig. 1T, Kimura et al.,1995,1996), and it was also down-regulated in the midbrain–hindbrain boundary of Fgfr1cko mutants (Jukkola and Partanen, unpublished data).

Figure 1.

Cdh22 expression co-localizes with midbrain–hindbrain border-specific genes in the early embryos. Whole mount in situ hybridizations with Cdh22, Fgf8, Wnt1, and Cdh11 probes in wild-type embryos from E8.0 to E11.5 A–I: Section in situ hybridizations with Cdh22, p21, and Otx2 probes at E10.5 (J–O). Midsagittal sections (J–L) and parasagittal sections (M–O). Red arrows mark the Cdh22 expression in the isthmus region, forebrain, and limb buds, and black arrow marks Cdh11 expression in the isthmus. Arrowhead indicates Cdh22 expression in the ventral midbrain. Red dashed line marks area for E11.5 and E12.5 sections in Figure 2. Black line marks the border of midbrain and hindbrain (J–O). Scale bars = 100 μm.

At E10.5–E11.5, we detected Cdh22 expression in additional areas of the central nervous system (Fig. 1G and H). Interestingly, Cdh22 was expressed in ventral midbrain near the place where ventral midbrain dopaminergic neurons are born (Fig. 1H, arrowhead). We examined this ventral midbrain expression more closely at the beginning of neurogenesis. At first (E10.5), Cdh22 was not detected in anterior part of the midbrain (aMB, Fig. 2A), but was expressed throughout the ventrolateral region in the posterior midbrain (pMB, Fig. 2B). At E11.5 and E12.5, Cdh22 was expressed in postmitotic regions of all ventral midbrain compartments m3–m7 (gabaergic, Fig. 2I and J; Nakatani et al.,2007; Kala et al.,2009) and also in postmitotic dopaminergic cells in m7 (Fig. 2I and J, arrows). In addition, we detected weaker Cdh22 expression in mitotic cells, especially in m6, next to the ventricular surface of the third ventricle. Interestingly, in addition to Cdh22, Cdh11, Cdh6 (Inoue et al.,1997), and Cdh8 (Korematsu and Redies,1997b) are known to be expressed in the embryonic brain. At E10.5, we observed Cdh11 expression in a ventrolateral region of the midbrain (Fig. 2C and D). In addition, Cdh11 was strongly expressed in underlying mesenchymal tissue. Cdh6 was strongly expressed in ventral midbrain and near the midbrain–hindbrain boundary (Fig. 2E and F). Cdh8 was expressed in dorsal midbrain, and weakly in the antero-ventral midbrain, but was absent from the midbrain–hindbrain boundary (Fig. 2G and H, data not shown). At E11.5 and E12.5, Cdh11 expression was strongest in the m6 region where cholinergic and glutamatergic neurons differentiate and was also observed in m7 and m5–m3, where it was more prominent in proliferative regions (Fig. 2K and L). Cdh11 was not detected in the most ventral part of the midbrain floor plate. Cdh6 was strongly expressed in the ventral midbrain region involved in dopaminergic neurogenesis (m7), but also in the ventricular zone of m6 and m5. Some postmitotic populations in m6 and m7 also expressed Cdh6 (Fig. 2M and N, arrowheads). Cdh8 expression was weak in the ventral midbrain region and concentrated to the dorsal parts (Fig. 2O, data not shown), but its expression was also detected in a postmitotic population in m6 (Fig. 2O and P, arrowheads) distinct from the Cdh6-expressing cells.

Figure 2.

Cdh22, Cdh11, Cdh6, and Cdh8 expression in the ventral midbrain. Section in situ hybridizations with Cdh22, Cdh11, Cdh6, and Cdh8 probes in wild-type coronal sections at E10.5 (A–H), E11.5, and E12.5 (I–P). Black lines indicate ventral midbrain compartments identified by comparing expression of Cdh22 to marker genes of neuronal populations: m7 is Lmx1b+, m6 is Pou4f1+, m5–m3 are Gad1+. Arrows indicate Cdh22 expression in m7 at E11.5 and E12.5. Arrowheads indicate postmitotic Cdh6 and Cdh8 expression in m6 and m7 at E11.5 and E12.5. aMB anterior midbrain, pMB posterior midbrain. Scale bars = 100 μm.

Next, we studied expression of Cdh22, Cdh11, Cdh6, and Cdh8 in the diencephalon (Fig. 3), where distinct expression domains were also observed. Cdh22 was detected primarily in postmitotic populations (Fig. 3A), whereas the other three cadherins were expressed in both the proliferative ventricular zone and postmitotic mantle zone. In the most caudal part of the developing diencephalon, prosomere 1 (P1, presumptive pretectum), we observed three distinct postmitotic populations. Anterior P1 co-expressed Cdh22, Cdh11, and Cdh8 (Fig. 3A, B, and D). The medial P1 population expressed Cdh22 and Cdh11 (Fig. 3A and B) and the most caudal population expressed only Cdh6 (Fig. 3C). In addition, Cdh8 expression was detected throughout the ventricular zone of P1 whereas Cdh11 expression was concentrated to more posterior and Cdh6 to more anterior regions (Fig. 3B–D). In prosomere 2 (P2, presumptive thalamus), Cdh22, Cdh11, and Cdh8 were expressed in an anterior postmitotic population (pTHr, Fig. 3A, B, D). In addition, Cdh6 and Cdh8 were expressed in the ventricular zone of P2 (Fig. 3C,D). Cdh11, Cdh6, and Cdh8 were co-expressed in the signalling center, zona limitans intrathalamica (ZLI, Fig. 3B–D). In prosomere 3 (P3, presumptive prethalamus), Cdh8 was expressed both in the ventricular zone and in a postmitotic population (Fig. 3D), whereas Cdh6 expression was strongest in the ventricular zone (Fig. 3C). Cdh22 and Cdh11 were not expressed in this part of the P3 area (Fig. 3A and B). In the developing hypothalamus, we detected Cdh11 expression in a small population in the ventricular zone and Cdh6 throughout the ventricular zone (Fig. 3B and C). We also detected expression of Cdh6 and Cdh8 in postmitotic cells of the lateral hypothalamus and Cdh8 in a medial domain (Fig. 3C, D). Cdh22 was not detected in this hypothalamic region (Fig. 3A).

Figure 3.

Cdh22, Cdh11, Cdh6, and Cdh8 expression in the diencephalon at E12.5. Section in situ hybridizations with Cdh22 (A), Cdh11 (B), Cdh6 (C), and Cdh8 (D) probes in wild-type coronal sections E12.5. Dashed black lines mark the borders of different compartments in the developing thalamus. H, hypothalamus; P1, prosomere 1; P2, prosomere 2; P3, prosomere 3; aP1, anterior prosomere 1; aP2, anterior prosomere 2; cP1, caudal prosomere1; mP1, medial prosomere1; ZLI, zona limitans intrathalamica. Scale bar = 200 μm.

Cdh22 and Cdh11 Expression in the Late Embryonic Brain

Next, we analyzed expression of Cdh22 and Cdh11 during maturation of the nervous system at later stages. In the midbrain and isthmus region, we observed weak Cdh22 expression throughout the area and stronger expression in some specific brain nuclei including substantia nigra pars compacta (SNpc), ventral tegmental area (VTA), arcuate nucleus (AN), ventral periaqueductal gray (VPAG), interpeduncular nucleus (IPN), nucleus of lateral lemniscus dorsal/ventral (NLLd/v), and locus coeruleus (LC, Fig. 4A–D'). Expression of Cdh11 was analysed on adjacent sections (Fig. 4E–H'). Both Cdh22 and Cdh11 were expressed in midbrain reticular formation (MRF) and pontine reticular formation (PRF). Cdh11 was also expressed in Cdh22-negative areas, such as subcomissural organ (SCO), substantia nigra pars reticulata (SNpr), dorsal cochlear nucleus (DCN), and cerebellar hemisphere. Cdh11 was expressed in oculomotor nucleus, which is negative for Cdh22. Interestingly, these two cadherins were also expressed in distinct layers of superior colliculus, as Cdh22 was expressed in the superficial layer (SFL), deep layer of superior colliculus (DL), and dorsal periaqueductal gray (DPAG), whereas Cdh11 was expressed in the supraficial layer of superior colliculus (SuFL; Fig. 4B'' and F''). Tyrosine hydroxylase (TH) and serotonin (5HT) expression detected on adjacent sections was used to mark dopaminergic and serotonergic cell populations (Fig. 4I–L, blue dash lines in A'–H').

Figure 4.

Cdh22 and Cdh11 expression in the midbrain and anterior hindbrain at E17.5. A–H': Section in situ hybridisations with Cdh22 and Cdh11 probes in wild-type brains. I–L: TH and 5HT immunohistochemistry on adjacent sections. Boxed area marks area in close-ups (A'–H'). B'' and F'' are close-ups from superior colliculus in B and F. Blue dashed line marks TH+ or 5HT+ area. Scale bars = 500 μm. AN, arcuate nucleus; CHml, cerebellar hemisphere molecular layer and purkinje cell layer; DCN, dorsal cochlear nucleus; DL, deep layer of superior colliculus; DPAG, dorsal periaqueductal gray; DR, dorsal raphe nucleus; IC, inferior colliculus; IPN, interpeduncular nucleus; Is, Isthmus; LC, locus coeruleus; MMN, medial mammillary nucleus; MRF, midbrain reticular formation; NLLd/v, nucleus of lateral lemniscus dorsal/ventral; OMN, oculomotor nucleus; PRF, pontine reticular formation; SC, superior colliculus; SCO, subcommissural organ; SCR, superior central raphe nucleus; SFL, superficial layer of superior colliculus; SNpc, substantia nigra pars compacta; SNpr, substantia nigra pars reticulata; SON, superior olivary nucleus; SuFL, Supraficial layer of superior colliculus; VPAG, ventral periaqueductal gray; VTA, ventral tegmental area.

In the forebrain, we detected Cdh22 expression in rather defined areas including indesium griseum (IG), medial habenular nucleus (MHN), ventrolateral geniculate nucleus (VLGN), amygdala (A), piriform cortex (PC), paraventricular nucleus of thalamus (PVNt), entorhinal cortex (ERC), reuniens nucleus (ReN) and several nuclei in hypothalamus such as paraventricular nucleus (PVNh), dorsomedial nucleus (DMNh), ventromedial nucleus (VMN), and posterior hypothalamic area (PHA; Fig. 5A–E'). Cdh22 was weakly expressed in bed nucleus of stria terminalis (BNST), triangular septal nucleus (TSN), ventromedial nucleus of thalamus (VMNt), CA3–CA1 layers in hippocampus, and medial lemniscus (ML). Cdh11 was also expressed in forebrain but in distinct brain nuclei including cingulate cortex (CC), infragranular part of cortical plate (Cpi), fornix (F), dorsolateral geniculate nucleus (DLGN), zona incerta (ZI), subcomissural organ (SCO), hypothalamic neuroepithelium (HNE), strionuclear neuroepithelium (SNN), and hippocampus (HC). Cdh11 expression was also detected in several nuclei in thalamus including lateral habenuclear nucleus (LHN), laterodorsal nucleus (LDN), centromedial nucleus (CMN), ventrolateral nuclear complex (VLNC), dorsolateral geniculate nucleus (DLGN), and posterior nuclear complex (PNC, Fig. 5F–J''). We detected both Cdh22 and Cdh11 expression only in IG, Hippocampus CA3–CA1 area, ventrolateral geniculate nucleus (VLGN), ventromedial nucleus of thalamus (VMNt), and anterior pretectal nucleus (PTN). Cdh22 could be detected in postmitotic areas such as striatal subventricular zone (SSVZ), and CA3–CA1 areas in hippocampus (Fig. 5A and C), whereas Cdh11 was also detected in mitotic regions such as SNN and HNE (Fig. 5F–I'). The distinct expression patterns suggest roles for type II cadherins in separation of brain nuclei.

Figure 5.

Cdh22 and Cdh11 expression in the forebrain at E17.5. Section in situ hybridizations with Cdh22 (A–E') and Cdh11 probes (F–J'). Black box marks area in close-ups (A'–J'). Scale bars = 500 μm. A, amygdala; AVN, anteroventral nucleus; BNST, bed nucleus of stria terminalis; CC, cingulate cortex; CMN, centromedial nucleus; cp, choroid plexus of third ventricle; Cpi, cortical plate infragranular part; DLGN, dorsolateral geniculate nucleus; DMNh, Dorsomedial nucleus of hypothalamus; ERC, entorhinal cortex; F, fornix; HC, hippocampus; HNE, hypothalamic neuroepithelium; IG, indesium griseum; LDN, laterodorsal nucleus; LHN, lateral habenular nucleus; MHN, medial habenular nucleus; ML, medial lemniscus; PC, piriform cortex; PHA, posterior hypothalamic area; PNC, posterior nuclear complex; POR, preoptic region; PTN, anterior pretectal nucleus; PVNh, paraventricular nucleus of hypothalamus; PVNt, paraventricular nucleus of thalamus; ReN, reuniens nucleus; SCO, subcommissural organ; SNN, strionuclear neuroepithelium; SSVZ, striatal subventricular zone; TSN, triangular septal nucleus; VLGN, ventrolateral geniculate nucleus; VLNC, ventrolateral nuclear complex; VMNh, Ventromedial nucleus of hypothalamus; VMNt, ventromedial nucleus of thalamus; ZI, zona incerta.

Inactivation of Cdh22 Causes Decreased Postnatal Survival

To study the function of Cdh22, we generated a mutated allele of Cdh22 (Cdh22del) where exon3 was removed (Turakainen et al.,2009). This exon encodes a tripeptide, which is necessary for the binding of type II cadherins (QAR in case of Cdh22, Kitajima et al.,1999). Deletion of exon3 was also expected to shift the reading frame of the Cdh22del transcript resulting in a truncated protein product (Fig. 6A). We designed RT-PCR primers to analyse if expected transcripts were produced from the Cdh22del allele or if loss of exon3 can induce aberrant splicing to more downstream exons. We only detected truncated transcripts, in which only exon3 was deleted and did not find evidence for splicing to exons further downstream (exons 5–7, Fig. 6B). The open reading frame of this Cdh22del transcript encodes for the signal peptide and a small part of first extracellular domain. The rest of the extracellular domains, transmembrane domain, and intracellular catenin-binding domain were not translated (Fig. 6A). Thus, the Cdh22del allele could not produce a functional Cdh22 protein.

Figure 6.

Schematic presentation of Cdh22del transcript and protein. A: Schematic view of wt and mutated Cdh22del allele, transcript and protein. B: Semiquantitative PCR analysis of Cdh22 RNA. Samples were reverse transcriptase processed total RNAs from the midbrain–hindbrain area at E12.5. Cdh22 cDNA clone was used as a positive control and total RNA (RT-) as a negative control of PCR. Cdh22del mutants encoded truncated transcript, which lacks exon 3. Wild type (wt) mRNA is 1.20 kb and truncated, Cdh22del RNA, 0.95 kb. CB, catenin-binding domain; EC, extracellular domain; QAR, tripeptide sequence; SP, signal peptide, TM, transmembrane domain. Primer1 shows the place for 5′ primer, and primer 2 for 3′ primer in semiquantitative PCR.

We then studied if Cdh22 mutation affected viability of embryos or mice. We analysed genotypes of altogether 170 embryos and 298 postnatal mice from F1-intercrosses. At the embryonic stages (E10.5, E12.5, and E17.5–E18.5), the genotypes followed expected mendelian ratios (Table 1). In contrast, the number of Cdh22del/del mutants was clearly smaller than expected in postnatal litters. Although we did not observe dead newborn pups, the genotype ratios presented in Table 1 are consistent with perinatal death of a portion of Cdh22del mutants. The surviving Cdh22del/del mutants appeared phenotypically normal and showed no obvious changes in their size, weight, or behaviour. Also, both males and females were fertile and their litter sizes were normal. The reason for the apparent reduction in the post-natal viability of Cdh22del mutant mice is currently unknown.

Table 1. Ratio of Genotypes in Cdh22del Litters
  1. aWith 3 groups df is 2, and with this df χ2 value should be over 5.99 so that p value would be less than 0.05.

E10.5 (expected)29 (25.25)44 (50.5)28 (25.25)1011.44∼0.5
E12.5 (expected)7 (8.5)21 (17)6 (8.5)341.94∼0.4
E17.5-18.5 (expected)10 (8.75)17 (17.5)8 (8.75)350.33>0.8
All E (expected)46 (42.5)82 (85)42 (42.5)1700.400.8
P3 (expected)11 (7.75)15 (15.5)5 (7.75)312.60∼0.3
P10 (expected)9 (5)8 (10)3 (5)204.40∼0.1
P21 (expected)76 (61.75)128 (123.5)43 (61.75)2479.15∼0.01**
All P (expected)96 (74,5)151 (149)51 (74,5)29812.54<0.001***

Patterning of the Midbrain-Hindbrain Boundary Is Normal in Cdh22del Mutant Embryos

To study compartmentalization and patterning of the midbrain–hindbrain region, we analyzed expression of midbrain–hindbrain boundary-specific genes by whole-mount mRNA in situ hybridization. Signalling molecules Wnt1 and Fgf8 orchestrate development of the midbrain–hindbrain region and are expressed in adjacent stripes in the posterior midbrain and anterior hindbrain, respectively (Wurst and Bally-Cuif,2001). Otx2 is a homeobox gene expressed anterior to the midbrain–hindbrain border. This transcription factor is important for positioning of the isthmic organizer at the midbrain–hindbrain boundary (Broccoli et al.,1999;, Garda et al.,2001). Cyclin-dependent kinase inhibitor p21 is a negative regulator of the cell cycle and marks the slowly proliferating boundary cell population at the midbrain–hindbrain border (Trokovic et al.,2005). All Fgf8, Wnt1, Otx2, and p21 were normally expressed in Cdh22del mutants, and these results suggest that there is no cell mixing across the midbrain–hindbrain boundary (Fig. 7A–H). Thus, early development and compartmentalization of the midbrain–hindbrain region is unaltered in the Cdh22del mutants.

Figure 7.

Patterning of midbrain–hindbrain boundary remains normal in Cdh22del mutants at E10.75. Whole mount in situ hybridization analysis with Fgf8 (A, B), Wnt1 (C, D), Otx2 (E, F), and p21 (G, H) probes. I,J: Whole-mount immunohistochemistry with anti-neurofilament antibodies shows oculomotor (III) and trochlear (IV) cranial nerves in wild-types and Cdh22del mutants. Arrowheads indicate the place of granial nerve nuclei and arrows indicate the projections.

In addition to its role in neuroepithelial patterning, the midbrain–hindbrain boundary also is important for later neuronal development. Trochlear motor neurons send projections anterodorsally along the isthmus using FGF8 for their guidance (Irving et al.,2002). Furthermore, FGF signalling and cadherins have been shown to interact in regulation of axon guidance (Lom et al.,1998). Therefore, we studied development of cranial nerves in Cdh22del mutants using whole-mount immunohistochemistry. We observed no defects in development and axon pathfinding of these nerves, including the IV nerve (trochlear, Fig. 7I and J).

Midbrain and Forebrain Development Is Unaffected in Cdh22del Mutants

As Cdh22 was expressed later in specific brain nuclei, phenotypic changes might appear during later brain development in the Cdh22del mutants. Therefore, we analyzed gabaergic (Gad1), glutamatergic (Vglut2), dopaminergic (TH), and serotonergic (5HT) neuron populations in the Cdh22del mutant midbrain at E18.5. All these neuronal populations were found in the Cdh22del mutants (Fig. 8A–P). The truncated Cdh22 mRNA could be detected in the same areas in Cdh22del mutants as the normal Cdh22 in wild-type embryos (Fig. 8E, G, M, and O). The shape of some Cdh22-expressing populations, such as interpeduncular nucleus, was slightly different, but these differences were not found consistently and are likely due to variation in sectioning.

Figure 8.

Midbrain compartmentalization remains unchanged in Cdh22del mutants at E17.5. Section in situ hybridization analysis with Gad1 (gabaergic, A–B', I–J') and Vglut2 (glutamatergic, C–D', K–L') probes. E,M:Cdh22 expression in wild-type. In Cdh22del mutants, a truncated Cdh22 transcript could be observed (G, O). TH and 5HT immunohistochemistry on adjacent sections (F, H, N, P). Scale bars = 500 μm. DR, dorsal raphe nucleus; IPN, interpeduncular nucleus; MGN, medial geniculate nucleus; MRF, midbrain reticular formation; NLLv/d, nucleus of lateral lemniscus ventral/dorsal; OMN, oculomotor nucleus; SNpc, substantia nigra pars compacta; RN, red nucleus; VPAG, ventral periaqueductal gray; VTA, ventral tegmental area.

As Cdh22 was abundantly expressed in specific forebrain nuclei (see above), we studied them in Cdh22del mutants. We used GFAP antibody to recognize forebrain glial structures. We observed normal glial structures such as indesium griseum and glial wedge in Cdh22del mutants (Fig. 9A, B). Also cholinergic populations, expressing Islet1, in arcuate nuclei could be seen in Cdh22del mutants (Fig. 9C, D). Gad67 and Vglut2 immunohistochemistry revealed that the main gabaergic and glutamatergic populations in forebrain were still present in Cdh22del mutants (Fig. 9E–T). The thickness of the glutamatergic layer in cortical plate appeared normal in Cdh22del mutants (Fig. 9I–J'). Notably, the nuclei where Cdh22 was expressed, such as medial habenular nucleus, amygdala, ventrolateral geniculate nucleus, and ventral diencephalic nuclei, were found in the right positions.

Figure 9.

No apparent alterations in the forebrain region in Cdh22del mutants at E17.5. GFAP visualizes glial structures in dorsal forebrain in wild-types and Cdh22del mutants (A, B). ISLET 1 visualizes ventral cholinergic populations in wild-types and Cdh22del mutants (C, D). GAD67 (gabaergic) or VGLUT2 (glutamatergic) positive neuronal populations are seen in wild-types and Cdh22del mutants (E–T). I', J': Close-ups from I and J. The white box marks the area for the close-ups. Scale bars = 500 μm. A, mygdala; AND, anterodorsal nucleus; AN, arcuate nucleus; AVN, anteroventral nucleus; BNST, bed nucleus of stria terminalis; Cpi, Cortical plate infragranular part; DLGN, dorsolateral geniculate nucleus; DMN, dorsomedial nucleus; GW, glial wedge; HPT, habenulopeduncular tract; IG, indesium griseum; LHT, lateral hypothalamus; LOT, lateral olfactory tract; LPN, lateral posterior nucleus; MGN, medial geniculate nucleus; MHN, medial habenular nucleus; MPOA, medial preoptic area; PTN, anterior pretectal nucleus; PVNh, paraventricular nucleus of hypothalamus; RcN, reticular nucleus; S, striatum; SI, substantia innominata; SMN, supramammillary nucleus; ST, stria terminalis; STN, subthalamic nucleus; SPOA, superior preoptic area; VBNC, ventrobasal nuclear complex; VLGN, ventrolateral geniculate nucleus; VNM, ventromedial nucleus; ZI, zona incerta.

Limb and Testicular Development in Cdh22del Mutants

As Cdh22 is also expressed in limb buds during early embryogenesis (Kitajima et al.,1999) and later in testicular gonocytes (Wu et al.,2003,2005), we briefly studied the morphology of limbs and testis in Cdh22del mutants. Adult Cdh22del mutant limbs appeared normal (data not shown). In the adult Cdh22del mutant males, the morphology of testis was normal. The males were also fertile and produced offspring in normal size litters (data not shown).


In this report, we studied the expression and function of Cdh22, a predicted homophilic cell adhesion molecule. Expression of Cdh22 was activated early during establishment of the midbrain–hindbrain boundary. Later, expression of Cdh22 and related cadherin family members was detected in specific yet overlapping regions in the embryonic midbrain and diencephalon. At birth, Cdh22 and Cdh11 were detected in distinct nuclei in forebrain, midbrain, and anterior hindbrain. Expression of Cdh22 and related cadherins in distinct brain areas and nuclei suggests a role in their specification or separation. Postnatal viability of Cdh22 null mutants was decreased, but surviving Cdh22del mutant mice were phenotypically normal and fertile. Brain development appeared normal in Cdh22del mutants suggesting functional redundancy among type II cadherins.

Cdh22 at the Midbrain–Hindbrain Boundary

Isthmic organizer orchestrates patterning of the midbrain and hindbrain. Coherence of this localized signalling center needs to be maintained by cell adhesion molecules. Cdh22 is expressed at the midbrain–hindbrain boundary (Kitajima et al.,1999), which represents a true compartment border in the brain (Zervas et al.,2004; Sunmonu et al.,2011). In addition, expression of Cdh22 was down-regulated at the midbrain–hindbrain border in Fgfr1cko mutants, which have lost FGF signalling and show cell mixing at the midbrain–hindbrain boundary (Trokovic et al.,2003). Our previous (Kala et al.,2008) and current studies show that Cdh22 marks a specific boundary cell population in the most posterior midbrain and most anterior hindbrain. These cells proliferate less, express Fgfr1, and might control brain compartmentalization (Trokovic et al.,2005). If the changes in cell adhesion properties in Fgfr1 mutants are caused by loss of Cdh22 expression, similar compartmentalization defects should have been visible in Cdh22del mutants. However, we could not detect any midbrain–hindbrain boundary defects in Cdh22del mutants. Thus, Cdh22 alone is not required for segregation of neuroepithelial cells between the midbrain and hindbrain.

In addition to Cdh22, Fgf signalling might also regulate expression of other cadherins at the midbrain–hindbrain boundary. At least Cdh6 (Inoue et al.,1997), Cdh8 (Korematsu and Redies,1997b), and Cdh11 (Kimura et al.,1995,1996; Redies and Takeichi,1996) are expressed at the midbrain–hindbrain region during embryogenesis. Indeed, we detected Cdh6 and Cdh11 expression at the midbrain–hindbrain boundary similar to Cdh22 (Fig. 2). Moreover, Cdh11 is also down-regulated in Fgfr1cko similarly to Cdh22 (Jukkola and Partanen, unpublished data). In addition to cadherins, the expression of additional adhesion molecules might be regulated by FGFs. For example, the expression of a cell adhesion molecule Cepu1 has been demonstrated at the midbrain–hindbrain boundary in chicken embryos in a pattern similar to Wnt1 (Jungbluth et al.,2001).

In addition to transcriptional regulation, Fgfrs may also interact with cadherins more directly by modulating their adhesion properties. In the case of widespread cadherin expression, Fgfrs might stabilize cadherin-based adhesion and promote binding specificity and affinity in specific areas. Cadherins can also stimulate neurite outgrowth by binding to Fgfrs. Fgfrs interact with cadherins, at least with N-cadherin (Williams et al.,2001; Sanchez-Heras et al.,2006) and Cdh11 (Boscher and Mege,2008), through their extracellular domain to induce neurite outgrowth and elongation. Moreover, inhibition of FGF signalling via N-cadherin (Lom et al.,1998) disrupts pathfinding of trochlear motoneurons and projection in the isthmus region (Irving et al.,2002). Interestingly, loss of zebrafish N-cadherin does not affect antero-posterior or dorso-ventral patterning during early embryogenesis but leads to abnormal positioning of neurons and axon guidance defects in the midbrain–hindbrain region (Lele et al.,2002). The critical role of N-cadherin might indicate that the other cadherins could play just a modulatory role in this process.

Cadherins Specifying Distinct Brain Nuclei

Cdh22 and related cadherin family members were expressed throughout the ventral midbrain including the dopaminergic, gabaergic, and glutamatergic nuclei, both at early and late stages of embryonic development. Interestingly, Cdh22 and Cdh11 showed largely complementary expression patterns in several regions of the forebrain, midbrain, and hindbrain. This suggests a role in specification and separation of distinct neuronal populations and layers. Also in the developing spinal cord, different type II cadherins are expressed in separate motoneuron populations and regulate segregation of these motoneuron pools (Price et al.,2002). Similar to Cdh22del mutants, Cdh6 mutants are viable and fertile and show no alterations in brain morphology. However, at the cortico-striatal boundary region in endogenous cadherin-expressing background, Cdh6 and R-Cdh overexpressing cells are sorted to lateral ganglionic eminence or cortex, respectively, according to the expression boundaries of these genes (Inoue et al.,2001). This sorting of over-expressing cells does not occur in Cdh6 mutants. Thus, expression of different cadherins can regulate cell sorting at the boundary region and separation of brain nuclei. Similar mechanisms could apply to other cadherins, such as Cdh22, in other neuronal populations in the central nervous system. However, we were unable to detect any major changes in neural populations in Cdh22del mutant brains.

In embryonic diencephalon, expression of type II cadherins may indicate specification of diencephalic nuclei. For example, Cdh22, Cdh11, and Cdh8 were expressed in postmitotic cells at anterior P2, which later gives rise to ventrolateral geniculate nucleus (Puelles and Rubenstein,2003; Vue et al.,2007; Kataoka and Shimogori,2008), where these cadherins were also expressed. Similarly, at E18.5 we observed Cdh22, Cdh11, and Cdh8 expression in the pretectum, anterior pretectal nucleus, which could be the most anterior postmitotic population in P1 at E12.5. Postmitotic P3 population, where Cdh8 was expressed, could be developing subthalamic nucleus, where Cdh8 is also expressed later (Korematsu and Redies,1997a; Suzuki et al.,1997; Korematsu et al.,1998).

Some type II cadherins, such as Cdh11 (Manabe et al.,2000) and Cdh8 (Korematsu and Redies,1997a), participate in neuronal specification and contact formation in the limbic system. Expression of Cdh22 in medial habenular nucleus, amygdala, ventral thalamus, hippocampus, hypothalamus, and interpeduncular nucleus suggests a role in the limbic system. As Cdh22 was expressed in several developing nuclei in the same neuronal circuitry, Cdh22 might also participate in axon guidance and synaptogenesis as suggested for other type II cadherins (Suzuki et al.,1997; Inoue et al.,1998; Korematsu et al.,1998; Redies,2000).

The cadherins are a large group of adhesion molecules, and many of them are expressed in similar areas in the brain. In many cases, abolishing a function of a single cadherin is not enough to result in a phenotypic change. Despite specific expression patterns, there also was significant overlap in the expression of Cdh22, Cdh11, Cdh8, and Cdh6. In regions like indesium griseum, bed nucleus of stria terminalis, pyramidal cell layer in hippocampus (CA3–CA1), ventrolateral geniculate nucleus, and anterior pretectal nucleus, Cdh22 and Cdh11 were co-expressed and might function together. Similarly, expression of other cadherin genes shows overlap and the Cdh22-positive regions always seemed to co-express at least one other type II cadherin, Cdh11, Cdh8, or Cdh6 (Korematsu and Redies,1997b; Suzuki et al.,1997). Also an evolutionarily closely related cadherin-12 appears to be expressed in some of these populations (Mayer et al.,2010). There are several additional type II cadherins, Cdh7, Cdh9, Cdh10, and Cdh20 (Fushimi et al.,1997; Kools et al.,1999; Bekirov et al.,2002; Takahashi and Osumi,2008), expressed in forebrain and midbrain during embryogenesis, but their expression still remains to be characterized in more detail. It is also possible that Cdh22 inactivation leads to compensatory up-regulation of one or several other family members by an unknown mechanism. Redundancy with related type II cadherins could explain why no phenotypic alterations were seen in Cdh22del mutants.



Construction of Cdh22del (Turakainen et al.,2009) allele has been described previously. Transposon-technique was used to introduce loxP sites around exon 3 of Cdh22. Matings with PGK-Cre mice were used to remove the neo-cassette, and to generate the Cdh22del allele. For the genotyping of wt Cdh22 alleles, the following primers were used Cdh22_1 (5′GGATGCCCTCTCACACCCTCC3′), Cdh22_2 (5′GGGAACACAGAGAGAC CCAGAAGC3′) and for Cdh22del allele Cdh22_1 and Cdh22_3 (5′GTGGCACT AGAGAAGGGACACGG3′). Noon of the day of vaginal plug was designated as embryonic day 0.5 (E0.5). Embryonic age was determined more precisely by counting somites. All experiments were approved by the National Committee of Experimental Animal Research in Finland.

In Situ mRNA Hydridizations and Immunohistochemistry

For section in situ analysis and immunohistochemisty, embryos were fixed by 4% PFA in PBS at least for overnight, dehydrated and embedded in paraffin. Sections were cut at 5 μm.

Whole mount in situ hybridization was performed by a modified protocol (Henrique et al.,1995) using digoxigenin-labelled antisense RNA probes Cdh22 (3′ side, IMAGE clone UI-M-BH4-azf-e10-o-ui; full cDNA gift from Dr. Nakamura), Fgf8 (Crossley and Martin,1995), Wnt1 (a gift from Klaus Schughart), Otx2 (Acampora et al.,1997), p21 (Trokovic et al.,2005), and Cdh11 (IMAGE clone 4035346). Section in situ hybridizations were performed by a modified protocol (Wilkinson and Green,1990) using 35S-labelled antisense RNA probes for Cdh22, p21 , Otx2 ,Cdh11, Gad1 (RZPD IRAV p968 M67D6), Vglut2 (Guimera et al.,2006), Cdh6 (IMAGE clone IRCLp5011G0820D), and Cdh8 (IMAGE clone IRAVp968E01116D).

Immunohistochemisty for paraffin sections was performed as described earlier (Jukkola et al.,2006). The following antibodies were used: mouse anti-tyrosine hydroxylase (TH, 1:500, Millipore, Billerica, MA), rabbit anti- serotonin (5-HT, 1:5000, Immunostar, Hudson, WI), rabbit anti-GFAP (1:500, Sigma, St. Louis, MO), mouse anti- ISLET1 (1:200, Developmental Studies Hybridoma Bank, Iowa City, IA), mouse anti-GAD67 (1:500, Millipore), rabbit anti-VGLUT2 (Slc17a6, 1:1,000, Sigma). Neuronal populations were identified according to the Atlas of Prenatal Rat Brain Development (Altman and Bager,1995) and the Electronic Prenatal Mouse Brain Atlas (

For whole mount neurofilament immunohistochemistry, E10.5 embryos were fixed in methanol:DMSO 4:1 overnight at 4°C and washed 5 × 10 min with 100% methanol. Embryos were treated with 6% H2O2 in 100% methanol at 4°C overnight , rehydrated, and blocked in 80% FCS:20% DMSO for 2 hr at RT. Embryos were incubated with mouse anti-neurofilament antibody (2H3 supernatant 1:100, Developmental Studies Hybridoma Bank) overnight in blocking solution, washed 10×30 min with PBT at RT, and incubated with HRP-conjugated goat anti-mouse IgG (1:200, Jackson, Santa Cruz, CA) in blocking solution. Embryos were washed 10 times with PBT at RT and incubated with diaminobenzidine (0.5 mg/ml) and 0.1% H2O2 in PBS for 20–30 min. Embryos were washed in PBS, dehydrated, and cleared in 1:1 mix of benzyl alcohol:benzyl benzoate. The embryos were bisected vertically for visualization. All in situ hybridization and immunohistochemical analyses were repeated at least once.

Semi-Quantitative PCR

For semi-quantitative PCR, the midbrain and anterior hindbrain area of E12.5 embryos were dissected and frozen in liquid nitrogen. Total RNA was extracted with the Omega Total RNA Kit I (Omega Bio-Tek, Frederick, MD). Reverse transcription was performed at + 37°C for 1.5 hr with oligo-dT primers (Promega, Madison, WI) and MMCV-RTaase (Promega). Primers for semi-quantitative PCR were 5′ TCGCCTGTGCTGCTGTTTCT (recombinant for region in exon 2) and 3′ ATCAATGGCGTACCGGACG (recombinant for region in exon7 and 8). Total RNA control was used to detect genomic DNA contamination.


We thank Eija Koivunen, Outi Kostia, and Raija Ikonen for their technical assistance. This work was supported by the Finnish Cultural Foundation (J S.-V.) and Helsinki Graduate School in Biotechnology and Molecular Biology (J.S.-V.).