The Reelin Receptor ApoER2 Recruits JNK-interacting Proteins-1 and -2*

Correct positioning of neurons during embryonic development of the brain depends, among other processes, on the proper transmission of the reelin signal into the migrating cells via the interplay of its receptors with cytoplasmic signal transducers. Cellular components of this signaling pathway characterized to date are cell surface receptors for reelin like apolipoprotein E receptor 2 (ApoER2), very low density lipoprotein receptor (VLDLR), and cadherin-related neuronal receptors, and intracellular components like Disabled-1 and the nonreceptor tyrosine kinase Fyn, which bind to the intracellular domains of the ApoER2 and VLDL receptor or of cadherin-related neuronal receptors, respectively. Here we show that ApoER2, but not VLDLR, also binds the family of JNK-interacting proteins (JIPs), which act as molecular scaffolds for the JNK-signaling pathway. The ApoER2 binding domain on JIP-2 does not overlap with the binding sites for MLK3, MKK7, and JNK. These results suggest that ApoER2 is able to assemble a multiprotein complex containing Disabled-1 and JIPs, together with their binding partners, to the cell surface of neurons. This complex might participate in ApoER2-specific reelin signaling and thus would explain the different phenotype of mice lacking the ApoER2 from that of VLDLR-deficient mice.

The most recently discovered members of the receptor family are mammalian ApoER2 and its avian homologue LR7/8B (4,5). Their common structure resembles that of the LDLR and the very low density lipoprotein receptor (VLDLR), but multiple alternative splicing events result in a complex array of receptor proteins (6,7). ApoER2 may contain a 59-amino acid insertion in its cytoplasmic domain, which is encoded by an extra exon that is differentially spliced in distinct variants of the receptor. The insertion is proline-rich and contains two potential SH3 binding sites (8). The prevalent site of expression of ApoER2 and LR7/8B is the brain, particularly neurons and cells that are components of brain barrier systems, such as the epithelia of the choroid plexus and of the arachnoidea, as well as endothelial cells of blood vessels (9,10).
Recently, the finding that cytosolic adaptor proteins like Dab1 and Fe65 bind to the intracellular receptor domains, in particular to the NPXY motif of LDLR and the LDLR-related protein (11,12), suggested that the current view of endocytosis of nutrients or removal of spent proteins as the main function of these receptors is only part of the story. Signal transduction should be considered to be another part of the functional spectrum that some or all of these receptors might participate in (13). First proof for this hypothesis was generated by the results of targeted disruption of both the VLDLR and the ApoER2 in mice (14). This genetic approach has identified both receptors as components of a signaling pathway that relays the reelin signal into migrating neurons. In addition, binding experiments showed that reelin interacts with the extracellular domains of both VLDLR and ApoER2 (15,16). Reelin is secreted by Cajal-Retzius cells in the outermost layer of the developing cerebral cortex and orchestrates the migration of neurons along radial fibers, thus forming distinct cortical layers in the cerebrum (for review see Ref. 17). Similar mechanisms are used to establish the characteristic neuronal arrangement in the hippocampus as well as the correct positioning of Purkinje cells in the cerebellum. A naturally occurring genetic defect in reelin creates the phenotype of the reeler mouse, which is characterized by an inversion of the cortical layers and absence of cerebellar foliation with an abnormal distribution of the Purkinje cells. Naturally occurring disruption or targeted deletion of the DAB1 gene produces phenotypes indistinguishable from that of the reeler mouse, suggesting that Dab1 is another component in the reelin signaling pathway (18,19). In addition, a family of cadherinrelated cell surface proteins termed the cadherin-related neuronal receptor (CNR) family have been identified to bind reelin (20). These proteins associate with Fyn, a nonreceptor tyrosine kinase of the Src kinase family (21), which in turn interacts with Dab1 (22). The current model suggests that reelin acts by direct binding to VLDLR and/or ApoER2 and to CNRs, thereby leading to phosphorylation of Dab1, which is associated with the tails of the LDLR relatives. Tyrosine phosphorylation of Dab1 may then start kinase cascade(s) controlling cellular motility and shape by acting on the neuronal cytoskeleton (14,20).
However, other experimental evidence suggests that the Reelin/Dab1 signaling pathway might not be the only signaling event involving members of the LDLR family. Apolipoprotein E, a universal ligand for these receptors, exerts an antithrombotic activity by inhibiting platelet aggregation through the L-arginine:nitric oxide pathway (23), supposedly mediated by ApoER2 (24,25). Furthermore, the stimulation of androgen synthesis in rat ovarian theca cells by apoE also seems to be mediated by members of the LDLR family (26).
A key feature of development of atherosclerosis is the activation of endothelial cells (27). Evidence is accumulating to indicate that activation of the transcription factor activator protein-1 via the c-Jun NH 2 -terminal kinase (JNK) pathway plays a pivotal role in endothelial cell activation by cytokines, hypoxia, shear stress, and most interestingly, by LDL (28 -30). The JNK-signaling pathway is involved in a variety of processes including cellular responses to environmental stress, cell proliferation, apoptosis, and morphogenesis (for review see Ref. 31). Recently it became evident that this pathway plays an important role in regulating region-specific apoptosis of neurons during early brain development (32,33).
The purpose of the current study was to search for proteins interacting with the proline-rich insertion in the cytoplasmic tail of ApoER2 and to examine if this region is involved in signaling pathways. Here we demonstrate that this domain interacts with the JNK-interacting protein (JIP)-1 and JIP-2, both members of the JIP group of mitogen-activated protein kinase scaffolding proteins (34). Furthermore, we show that the occurrence of the proline-rich domain-containing splice variant of ApoER2 and the expression of JIP-2 coincide during differentiation of neurons. The results demonstrate a molecular link between ApoER2 and the JNK signaling pathway.

EXPERIMENTAL PROCEDURES
Yeast Two Hybrid Screen and cDNA Cloning-The cDNA coding for the 59-amino acid insertion of the murine ApoER2 tail was generated using the following primers: sense, 5Ј-GGA ATT CGC AAT CAG CAA CTA TGA TCG C and antisense, 5Ј-TTG GAT CCT TAC TTG CAC TTG ACG ACA GGC. The PCR product was cloned via internal EcoRI restriction sites of the primers into pLexA. Fidelity of the construct was confirmed by sequencing, and the construct was subsequently used as a bait for screening a mouse brain Matchmaker R LexA Yeast Two Hybrid library (CLONTECH) according to the manufacturer's instructions. Putative positive colonies were streaked out to create master plates, and the corresponding plasmid insertions were selectively amplified by colony PCR using vector-specific primers and were directly sequenced.
The full-length cDNA of murine JIP-2 was cloned by two consecutive rounds of 5Ј-rapid amplification of cDNA ends PCR. Using primer a (5Ј-ACTGGCCTCGCAGCTGCTGTCC, annealing temperature, 58°C), a 1678-base pair fragment was obtained. Then, using primer b (5Ј-GTCGTCCGTGATCTCAGACAGG, annealing temperature, 58°C), another 81 base pairs of the 5Ј-sequence were obtained. The ATG assigned to represent the start codon for the mouse JIP-2 cDNA is located within a sequence consistent with the consensus sequence of translational start sites (35) and corresponds to the published sequence of human JIP-2 (34).
cDNA Constructs-An IB1 (JIP1-b) cDNA was obtained from Dr. G. Waeber, Lausanne, France. IB1 and JIP-2 were tagged at the 5Ј-end with a 9xmyc epitope and cloned into pCIneo (Promega). The partial cDNA clone obtained by Yeast Two Hybrid screening (corresponding to amino acids 607-830, Fig. 3B) was tagged at the 5Ј-end with an influenza hemagglutinin (HA) epitope and cloned into pCIneo. GST fusion proteins of the respective domains of the ApoER2 tail ( Fig. 1) were prepared as described (9) using the following primers: A) sense, 5Ј-CGG GAT CCG CCG GAA GAA CAC CAA GAG C; antisense, 5Ј-GGA ATT CTC AGG GCA GTC CAT CAT C (tail with and without proline-rich insertion; the resulting two bands were separately subcloned and verified by sequencing); and B) sense, 5Ј-CGG GAT CCG AGC AAT CAG CAA CTA TGA TC; antisense, 5Ј-GGA ATT CTC ACT TGC ACT TGA CGA C (proline-rich insert). GST-Dab1 is described in Ref. 14; fulllength Dab1 was tagged at the 5Ј-end with the HA epitope and cloned into pCI neo. The cDNA constructs corresponding to the SH3 (270 base pairs) and the PID (430 base pairs) domains of JIP-2 were amplified via PCR from the full-length JIP-2 cDNA using the following primers: SH3 sense, 5Ј-GAATTCAATGTCAACAGCACGTCTCGATCC; SH3 antisense, 5Ј-CCAGCAGGTCCTTGGCAGGACC (annealing temperature 60°C); PID sense, 5Ј-GAATTCAAGGTGGATCGCTTCGATGTGC, PID antisense, 5Ј-GTAGATATCCTCTGTAGGGCAGGC (annealing temperature 60°C). The PCR fragments were subcloned into a T/A cloning vector (pCR2.1, Invitrogen), then cut out via EcoRI sites present in the primers, and cloned into the mammalian expression vector pCIneo, which contained a 9xmyc-tag 5Ј of the insertion. Fidelity of all constructs was verified by sequencing.
Northern Blot Analysis-Total RNA was prepared from different tissues of a male and female Balb/C mouse using TRI REAGENT (Molecular Research Center, Inc.). 30 g of total RNA was separated by electrophoresis on a 1.5% agarose gel and transferred onto a nylon membrane. After UV cross-linking, hybridization was performed using the partial JIP-2 cDNA originally found in the Yeast Two Hybrid screen. Methylene Blue staining was used as quality and loading control.
Antibodies-The antibody against mouse ApoER2 was produced against the complete intracellular domain expressed as a GST fusion protein as described previously (9). Anti-JIP-2 was produced against a synthetic peptide (see Fig. 3A) corresponding to the cloned mouse JIP-2 cDNA as described (4). The following antibodies were obtained commercially from the sources indicated: antimitogen-activated protein 2 (Sigma), anti-HA tag (16B12, Babco), anti-Myc (9E10, used as hybridoma supernatant from the corresponding cells, ATCC), Oregon Green 488labeled goat anti-mouse IgG (Molecular Probes), Alexa 594-labeled goat anti-rabbit (Molecular Probes), and goat anti-rabbit-biotinylated IgG (Sigma).
Cell Culture and GST Pull-down Assays-293 cells were maintained in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) supplemented with 10% fetal calf serum and 584 mg/liter glutamine and transfected with Lipofectin reagent (Life Technologies, Inc.). Cellular extracts for Western blots and precipitation experiments were prepared by lysing the cells in HUNT buffer (20 mM Tris, pH 8.0, 100 mM NaCl, 1 mM EDTA, 0.5% Nonidet P-40, 0.1 mg/ml phenylmethylsulfonyl fluoride) for 15 min on ice and centrifugation to remove insoluble material. For pull-down experiments, 30 l of cell extract were diluted with 150 l of HUNT buffer. After the addition of ϳ10 l of glutathione-Sepharose with bound fusion protein, the mixture was incubated for 2 h at 4°C on a rotary shaker. Subsequently, the beads were washed three times with 200 l of HUNT buffer. Bound proteins were eluted by incubating the beads for 3 min at 95°C under reducing conditions in SDS loading buffer and subjected to SDS-polyacrylamide gel electrophoresis. Electrophoresis, transfer to nitrocellulose membranes, and Western blotting were performed as described previously (36).
Immunohistochemistry-Mice were euthanized with CO 2 and perfused via the left ventricle with 25 ml of PBS, followed by 25 ml of a solution containing 75 mM L-lysine, 75 mM sodium phosphate, pH 7.3, 2% paraformaldehyde, and 2.4 mg/ml sodium meta-periodate. The brain, testis, and epididymus were then removed and stored in the fixation solution for 4 h at 4°C. Embedding and specimen preparation was done as described (9). Incubation with the primary antibodies (dilutions: anti-JIP-2, 1:500; anti-ApoER2, 1:100) was performed at 4°C for 20 h in blocking solution. After five washes in PBS, the following incubations were performed at 23°C: goat anti-rabbit biotinylated IgG diluted 1:500 in blocking solution for 1 h, five washes with PBS, peroxidase-labeled avidin (Sigma) for 1 h diluted 1:200 in 1% nonfat dry milk in PBS, and a final wash with PBS (five times). For the color reaction, slices were incubated in 0.1 M sodium acetate, pH 5.1, containing 150 l of 30% peroxide and 20 mg of 3-amino-9-ethylcarbazole, per 100 ml of buffer. The staining process was followed under the microscope (Zeiss Axiovert 135) and stopped by immersing the slides in water. Nuclei were counterstained with Harris-modified hematoxylin solution (Sigma).
Neuronal Stem Cell Culture-Neuronal stem cell culture and differentiation into neurons was performed according to Ref. 37. Embryos 15-17 days old were removed and put into PBS plus chlortetracycline (Life Technologies, Inc.). Embryo heads were opened sagittally under sterile conditions, and the brains were removed with forceps and transferred into HBSS (Life Technologies, Inc.). After three washes with HBSS, brains were triturated in Hanks' balanced salt solution and centrifuged at 4000 rpm for 4 min, and the cells were suspended in prewarmed culture medium (Dulbecco's Nut Mix F12 (Life Technologies, Inc.) supplemented with 2 mM L-glutamine, B27 supplement (Life Technologies, Inc.), antibiotic/antimycotic (Life Technologies, Inc.)), grown on 10-cm cell culture dish coated with poly-L-ornithine (Sigma) at 37°C with 5% CO 2 . Stem cells were grown in the presence of 10 ng/ml human recombinant bovine fibroblast growth factor (Sigma). Growth factor was added every day, and the medium was changed every second day. To start differentiation into neurons, in addition to 10 ng/ml basic fibroblast growth factor, 10 ng/ml recombinant platelet-derived growth factor-BB (Sigma) was added to the medium. Two days later only platelet-derived growth factor-BB was added daily. At 50% confluency, cells were split by rinsing once with PBS and incubating with 1 ml of Trypsin (Life Technologies, Inc.) for 10 -20 min at 37°C until cells detached. Trypsinization was stopped by the addition of medium supplemented with 1 mg/ml chicken egg white trypsin-inhibitor (Sigma).
Immunofluorescence-293 cells were grown on culture slides (Becton Dickinson) coated with poly-D-lysine to a confluency of 50 -70%. Mouse brain cells were transferred to coated 8-well slides 2-6 days prior to the staining. Slides were coated with poly-L-ornithine (Sigma) and fibronectin (Sigma). After washing twice with PBS, cells were fixed with a mixture of equal amounts of methanol and acetone at Ϫ20°C for 4 min (293 cells) or at Ϫ20°C for 30 -60 min (mouse brain cells). Fixed cells were rehydrated for 15 min at room temperature with PBS, incubated with the primary antibody in blocking solution (1% nonfat dry milk in PBS) for 1 h at room temperature, washed three times with blocking solution, and incubated at room temperature with the secondary antibody (Alexa 594 1:500 or Oregon Green 488 1:500, Molecular Probes) for 1 h. After two washes with blocking solution and two final washes with PBS, microscopy was performed.

RESULTS
The intracellular domain of ApoER2 is highly homologous to that of LDLR and VLDLR and binds the intracellular adaptor protein disabled via its conserved NPXY motif (11). However, because of the differential splicing of an additional exon in the ApoER2 gene, a region coding for a 59-amino acid insertion is present in the cytoplasmic tail of a significant fraction of the receptor population (7). As depicted in Fig. 1a, this proline-rich domain is inserted 10 amino acids before the carboxyl terminus of the protein. To identify proteins interacting with this specific domain of ApoER2, we used the yeast two-hybrid method based upon the LexA system (38). A Matchmaker LexA mouse brain cDNA library was screened using the 59-amino acid insert as bait (Fig. 1b). Out of 10 8 transformants screened, four clones scored positive for an interaction of the receptor insert with products of transcripts encoded by the library. Two of these clones represented sequences interrupted by stop codons in all reading frames; the remaining two clones were identical and homologous to human JIP-1 and JIP-2 (34). To verify the interaction between ApoER2 and the candidate protein at the biochemical level, we performed "pull-down" experiments by incubation of extracts prepared from 293 cells expressing the HA-tagged protein with bacterially expressed fusion proteins between GST and various receptor tail constructs as shown in Fig. 1, c-e. Protein complexes were precipitated with glutathione-Sepharose beads, eluted, separated on SDS-polyacrylamide gel electrophoresis, and subjected to Western blotting with a monoclonal anti-HA antibody (Fig. 2, lanes 2-4). The same antibody was used to demonstrate the expression of the tagged cDNA product in transfected 293 cells by direct Western blotting of the respective cell extracts (Fig. 2, lane 1). Clearly, the HA-tagged product of the candidate clone interacted with the GST fusion proteins that contained either the proline-rich insert alone ("i," lane 2) or the full-length tail ("ti," lane 3) of the receptor. Importantly, the interaction is specific for the insert, because the tail without insert ("t," lane 4) did not interact with the HA-tagged protein. The specificity for the insert is also documented by the fact that the tagged protein does not bind to the cytosolic tail of the VLDLR (data not shown). To further verify this result, we used an HA-tagged fragment of Dab1, which previously had been shown to bind to the ApoER2 tail independently of the proline-rich insert (14), in a similar experiment. Indeed, Dab1 binds to both the short and long cytosolic domains but not to the proline-rich insert itself (Fig. 2,  lanes 5-8). In contrast to the novel protein identified by the screen, it was previously shown that Dab1 binds to the VLDLR tail, too (14).
Having biochemically verified the interaction of the product of the candidate cDNA obtained in the two-hybrid screen, we cloned the corresponding full-length cDNA by 5Ј-rapid amplification of cDNA ends (Fig. 3). Data base searches and se- quence alignment clearly identified the novel cDNA as the murine homologue of human JIP-2 (Fig. 3A). JIP-2 belongs to a group of proteins that bind to JNK, MKK7, and MLKs, thereby providing a scaffold to route JNK activation by this pathway (39,40). The domain structure of JIP-2, which is very similar to that of JIP-1 (34), is illustrated schematically in Fig. 3B. The most prominent features of this protein are (i) the JNK binding domain residing in the amino-terminal part and (ii) a SH3 domain and a PID domain clustered in close vicinity to each other in the carboxyl-terminal half of the protein. The clones originally obtained in the two-hybrid screen and represented in the cartoon as "prey" (Fig. 3B) contained both the SH3 and the PID domain. Because both modules are involved in the assembly of signaling complexes (41), it was of interest to determine whether one or both of these domains mediate binding of JIP-2 to the receptor. To answer this question we tested the interaction directly by employing partial constructs containing either domain tagged with 9xmyc (Fig. 3B) in pull-down assays as above. As shown in Fig. 4A, only the peptide containing the PID domain but not the SH3 domain binds to the ApoER2 tail.
Previous studies demonstrated that the SH3 and PID domains of human JIP-2 and JIP-1 are 66 and 64% identical, respectively (34). Therefore, we speculated that JIP-1 might also be a binding partner of ApoER2. To test possible binding of JIP-1 to ApoER2, we used 9xmyc-tagged full-length IB1 in a similar pull-down experiment as above. IB1 corresponds to the longer splice variant of rat JIP-1 (JIP-1b) containing the intact PID domain (42). Interestingly, IB1 also binds to the receptor and, moreover, displays the same specificity for the proline-rich insert (Fig. 4C) as JIP-2 does (see Figs. 2C and 4A).
Northern blot analysis (Fig. 5) of different murine tissues revealed an expression pattern of JIP-2 similar to that in man, with strong expression in brain and testis. This appears to be of particular significance, because these two organs are also the major sites of ApoER2 expression in mammals (5). To further substantiate this finding, we analyzed the expression of both interaction partners at the cellular level using immunohistochemistry. For this purpose, we produced a polyclonal rabbit antipeptide antibody against murine JIP-2. The synthetic peptide used for immunization (indicated sequence in Fig. 3A) was designed to represent a region in JIP-2 maximally different from JIP-1 to avoid cross-reactivity. Western blot experiments using extracts from cells expressing full-length JIP-2 and IB1 as tagged proteins showed that this antibody indeed recognizes JIP-2, but not JIP-1 or any other protein in the extracts (not shown). Immunohistochemical analysis revealed JIP-2 expression in all neurons. Representative for all major regions of the brain is a section of the lateral hypothalamus (Fig. 6A). Neurons are prominently stained by anti-JIP-2 antibody; glial cells, however, did not show any detectable expression. Together with the previously reported expression of LR7/8B in chicken brain (9) and ApoER2 in mouse brain (10), these data demonstrate that JIP-2 and ApoER2 are both expressed in neurons. A very similar situation became evident when the expression of both proteins was examined in the testis; both JIP-2 (Fig. 6B) and ApoER2 (Fig. 6C) are strongly expressed in the principal cells of the epididymus. Principal cells are the most abundant cells of the epididymal epithelium, which is responsible for the transport, maturation, and storage of spermatozoa (43). The most striking feature of this analysis is the strictly apical expression of ApoER2 in these cells, which is reminiscent of the expression of ApoER2 in the epithelium of the choroid plexus (9). JIP-2 is detectable throughout the principal cells; however, expression appears to be concentrated at the apical side of the cells. Because JIP-2 is a cytoplasmic protein, this result suggests that a considerable fraction becomes localized to the apical surface of the principal cells by interaction with the cytoplasmic tail of ApoER2. The impressive overlap of JIP-1 and -2 expression with ApoER2 in adult brain and epididymus (this study) and during embryonic development (51) provides strong circumstantial evidence for common biological functions  1 and 2), the Myc-tagged PID domain of JIP-2 ( lanes  3 and 4), the Myc-tagged SH3 domain of JIP-2 ( lanes 5 and 6), and full-length Myc-tagged IB1 (JIP-1b) (lanes 7-10) were either directly loaded onto the gel (lanes 1, 3, 5, and 7) or incubated with a GST fusion protein containing the complete ApoER2 tail (lanes 2, 4, 6, and 9), the proline-rich insert alone (lane 8), or the tail without the insert (lane 10) prebound to gluthathione-Sepharose for 2 h at 4°C. Bound proteins were eluted from the Sepharose and analyzed by electrophoresis and Western blotting with the anti-Myc antibody.
in which these genes participate.
Finally, we studied the expression of the relevant splice variant of ApoER2 and its interacting partner JIP-2 during brain cell differentiation. Not only are neurons, astrocytes, and oligodendrocytes derived from a single class of stem cells during embryonic brain development, but such stem cells also reside in the adult brain, where they can give rise to and/or replace differentiated brain cells in vivo (44,45). These stem cells can be isolated from embryonic mouse brains, expanded, and differentiated in vitro into neurons, astrocytes, and oligodendrocytes (37). Using reverse transcriptase-PCR and presented in Fig. 7A, we could demonstrate that expanded murine brain stem cells express very little ApoER2 (lane 1). Under the conditions applied, we could detect only receptor transcripts lacking the exon coding for the proline-rich insertion. Concomitantly, these cells do not express any detectable levels of transcripts for JIP-2 (lane 2) and only trace amounts of transcripts for JIP-1 (lane 3). However, upon differentiation into neurons, strong expression of JIP-2 (lane 5) and JIP-1 (lane 6) was evident. Importantly, expression of ApoER2 in differentiated cells is not only induced severalfold, but in addition to the receptor variant present in stem cells, neurons also express the longer receptor variant containing the 59-residue JIP-interaction domain (lane 4) (7). As control and further refinement of these results we used these differentiated neurons for immunofluorescent studies. As demonstrated in Fig. 7B, panels 1 and 2, cells staining for JIP-2 also express microtubule-associated protein 2, a commonly used neuronal marker (37). Neurons derived from expanded stem cells also express ApoER2 in a similar fashion as described for primary rat neurons (9) (Fig.  7B, panel 3). Because both antibodies used in this study, anti-ApoER2 and anti-JIP-2, are polyclonal antibodies produced in rabbits, it is not possible to show a colocalization of both endogenous proteins in differentiated neurons. To overcome this problem, we expressed ApoER2 together with a 9xMyc-tagged version of full-length JIP-2 in 293 cells. Using a monoclonal anti-Myc antibody and anti-ApoER2, we could indeed demonstrate the colocalization of both proteins in a cellular context (Fig. 7B, panels 4 -6). These results strongly support the hypothesis that in neurons the proline-rich domain in ApoER2 may mediate a signal transduced by the receptor via the JIP proteins to JNK and its target genes.

DISCUSSION
ApoER2 and VLDLR together are required for the correct embryonic development of the brain (14). Targeted disruption of both genes results in a phenotype virtually indistinguishable from that seen in reeler and scrambler mice. The phenotype arising from the absence of both receptors is a result of the inability of migrating neurons to relay the extracellular reelin signal into a cellular response. Apparently, such neurons, which differentiate normally, do not find their correct positions in the cerebral cortex. In addition, Purkinje cells are not arranged in the normal pattern but, are dispersed in ectopic clusters throughout the rudimentary and unfoliated cerebel- FIG. 5. Northern blot analysis of ApoER2 expression in murine tissues. A, 30 g of total RNA from various tissues from male and female Balb/C mice were separated by electrophoresis on a 1.5% agarose gel and transferred onto a nylon membrane. After UV cross-linking, hybridization was performed using the partial JIP-2 cDNA originally found in the yeast two hybrid screen. Methylene Blue staining was used as loading control (B). A 1-kilobase marker from Life Technologies, Inc. was used to calibrate the gel.
FIG. 6. Immunohistochemical analysis of JIP-2 and ApoER2 in mouse brain and epididymus. Immunohistochemistry was performed on the sagittal sections of the brain and the epididymus from a male Balb/C mouse as described under "Experimental Procedures" using anti-JIP-2 antibody (A and B) and anti-ApoER2 (C). Nuclei are counterstained with hematoxylin (blue). A, section of the lateral hypothalamus; B and C, sections of the epididymus. lum. Interestingly, the single knock-out of either one of the two receptors produces subtle, but different, phenotypes. This suggests that each receptor compensates extensively, but not completely, for the loss of the other. This may not be surprising, because both receptors appear to bind reelin equally well (15,16), and both receptors interact with the intracellular adaptor Dab1 (14), which becomes phosphorylated upon reelin binding (46). Subtle effects of VLDLR deletion are found mainly in the cerebellum, whereas lack of ApoER2 predominantly affects the positioning of neurons in the neocortex. In addition, differences in the neuroanatomical phenotype produced by the lack of either receptor within the neocortex suggest that one or both receptors may interact specifically with adaptor proteins different from Dab1. This speculation is strengthened by the fact that a subpopulation of ApoER2 differs significantly in its cytoplasmic tail from that of VLDLR. Because of differential splicing, ApoER2 is facultatively expressed with a 59-amino acid proline-rich insertion in its cytoplasmic domain. This do-main has not been found in any other member of the LDL receptor family, including the VLDLR. These findings provided the rationale for our search for proteins that interact specifically with the proline-rich domain in ApoER2.
in IB1 (JIP-1b) and JIP-2, partially overlaps with the SH3 and the PID domains, respectively (Fig. 8A) (42); this domain could be responsible for the interaction of IB1 with the GLUT2 promoter. Because of their high homology in the carboxyl-terminal parts, not only JIP-2, but also JIP-1 binds to ApoER2. Mapping of the domain of JIP-2 responsible for the interaction with the proline-rich domain in ApoER2 localized the binding site to within the PID domain. This is surprising, because the prolinerich domain in the receptor contains two consensus target sites for the SH3 domains (PXXP); it is the NPXY motif present in the cytosolic domains of both ApoER2 variants and the VLDLR, which is recognized as the consensus site for interaction with PID domains (41). Because, however, the PID domain in JIP-2 does not bind to the NPXY motif in the receptor tail (as shown in this study and in Ref. 51), it is possible that JIP-2 and Dab1 may bind simultaneously to the long form of ApoER2 as part of a multimeric signaling complex.
Previously, the binding sites for MKK7 and MLK3 on JIP-1 were found to be independent (40). As indicated in the cartoon presented in Fig. 8A, MLK3 was shown to interact with JIP-1 in the region between residues 471 and 660. This stretch contains the SH3 domain but does not overlap with the PID domain shown here to interact with ApoER2 (Fig. 8A). The MKK7 and JNK binding sites on JIPs are located even closer to the amino terminus of the protein. All of these interaction domains show no overlap with each other, suggesting that binding to ApoER2 does not interfere with binding of JIP-2 to JNK, MKK7, and MLK3 (Fig. 8A). In addition, an interaction of JIP-1 with rhoGEF, an exchange factor for the small GTPase rhoA, has also been reported recently (48). RhoGEF is highly expressed in neural tissues and via its interaction with rhoA might be involved in cytoskeletal rearrangements of neurons Binding of reelin to CNRs activates Fyn, which is associated with the intracellular domains of the CNRs. Fyn tyrosine phosphorylates Dab1, which is bound to the NPXY motif present on the intracellular domain of ApoER2, and triggers activation of downstream components. Independently of the interaction with Dab1, ApoER2 is able to recruit JIP proteins to the plasma membrane. This interaction is suggested to interfere with JNK signaling, possibly by inhibiting translocation of activated JNK to the nucleus. In addition, the activity of rhoGEF, which associates with JIP, might be modulated by this complex. (49). However, it remains to be seen whether binding of rho-GEF to JIP-1 interferes with JIP-1 binding to the receptor. These results suggest that ApoER2 can assemble a large signaling complex around its cytoplasmic domain including Dab1, JIPs, MLK3, MKK7, JNK, and possibly rhoGEF (Fig. 8B).
In the context of reelin signaling, it is important to note that this complex is specific for ApoER2. Because VLDLR only binds to Dab1 but not to JIPs, the interaction of reelin with ApoER2 and CNRs may lead not only to fyn-catalyzed tyrosine phosphorylation of Dab1 and to a reorganization of the cytoskeleton via Cdk5 acting downstream of Dab1, but also, via JIPs, to additional interaction with the cytoskeleton mediated by rho-GEF. Furthermore, binding of ApoER2 to JIPs links the reelin signal to the JNK pathway and thereby may modulate cell proliferation, apoptosis, and tissue morphogenesis (31). Recent studies have shown that the JNK family is involved in early brain development by regulating region-specific apoptosis (33); furthermore, constitutively activated c-Jun can induce neuronal differentiation of PC12 cells (50). Finally, the interaction of JIP-1 with rhoGEF might directly link the reelin signal to a pathway involved in the organization of the neuronal cytoskeleton. Again, because the interaction with the JIPs is specific for ApoER2 and does not apply to the reelin signaling pathway transmitted via VLDLR, these additional pathways could explain the distinct phenotypes of mice lacking either the VLDL receptor or the ApoER2 alone.
Previous results regarding the expression of ApoER2 in the epithelium of the choroid plexus and of blood vessels (9) together with our results on ApoER2 expression in the epididymus suggest to us that the receptor might have another function yet, e.g. it could display endocytic activity. The most striking feature of the expression of ApoER2 in epithelial cells in these tissues is its prominent and strictly apical expression. Furthermore, we and others have shown that ApoER2 is an endocytosis-competent receptor (5,9). The uptake of components from the cerebrospinal fluid, or in the case of the epididymus from the seminal fluid, does not at all exclude that ligand endocytosis may be accompanied by signal transmission into these cells. This is an intriguing possibility for epithelial cells of the epididymus, because these cells also express JIP-2, as demonstrated here. In consideration of the broad ligand spectrum of most members of the LDLR family, the identification and characterization of hitherto unknown ligands present in the cerebrospinal and/or seminal fluid may still hold further surprises.
In Ref. 51 we have further extended the spectrum of physiological roles in which LDL receptor family members may function in vivo. Using yeast two-hybrid and GST pull-down approaches, an extended set of cytoplasmic adaptor and scaffold proteins that differentially interact with the cytoplasmic tails of the various family members was identified. This set includes the JIP proteins that have been the focus of the present study. In addition, other adaptor and scaffold proteins that were identified in this screen suggest that LDL receptor family members regulate not only cellular kinase pathways but may also be involved in the modulation of signaling events mediated by G-proteins, NO, and possibly ion channels. Whereas the present studies point toward a range of novel biological roles for this ancient family of multifunctional receptors, genetic studies in animals will likely be required to unambiguously decipher the physiological importance of the individual interactions.