Professor: Masaharu Noda

　We have been studying the molecular and cellular mechanisms underlying the development and functioning of the vertebrate central nervous system. The scope of our interests encompasses regional specification in the retina, neuronal differentiation, cellular migration, path-finding and target recognition of axons, formation and refinement of specific synapses, and also various functions of the matured brain.

I. Molecular mechanism for topographic retinotectal projection

A. Topographic projection

　Topographic maps are a fundamental feature of neural networks in the nervous system. Understanding the molecular mechanisms by which topographically ordered neuronal connections are established during development has long been a major challenge in developmental neurobiology. The retinotectal projection of lower vertebrates including birds has been used as a readily accessible model system. In this projection, the temporal (posterior) retina is connected to the rostral (anterior) part of the contralateral optic tectum, the nasal (anterior) retina to the caudal (posterior) tectum, and likewise the dorsal and ventral retina to the ventral and dorsal tectum, respectively. Thus, images received by the retina are precisely projected onto the tectum in a reversed manner.
　Since 1992, we have been devoting our efforts to searching for topographic molecules which show an asymmetrical distribution in the embryonic chick retina. In the first-round screening using a cDNA subtractive hybridization technique, we identified two winged-helix transcriptional regulators, CBF-1 and CBF-2, expressed in the nasal and temporal retina, respectively. Furthermore, our misexpression experiments using a retroviral vector showed that these two transcription factors determine the regional specificity of the retinal ganglion cells, namely, the directed axonal projections to the appropriate tectal targets along the anteroposterior axis. We next examined whether CBF-1 and CBF-2 control the expression of EphA3, a topographic guidance molecule in the retina using in ovo electropolation. We obtained results suggesting that these two transcription factors affect the topographic expression of EphA3, ephrin-A2 and -A5 indirectly.
To further search for topographic molecules in the embryonic retina, we next performed a large-scale screening using a new cDNA display system called Restriction Landmark cDNA Scanning (RLCS) in 1997. With the assistance of a computer image-processing software, we successfully identified 33 molecules along the nasotemporal axis and 20 molecules along the dorsoventral axis, with various asymmetrical expression patterns in the developing retina. We have elucidated the primary structures of all these cDNA clones and examined their expression patterns during development. Included were many novel together with known molecules: transcription factors (CBF-2, COUP-TFII, etc.), receptor proteins (EphA3, EphB3, etc.), secretory factors, intracellular proteins, and so on.
　Among them, we identified a novel retinoic acid (RA)-generating enzyme, RALDH-3, which is specifically expressed in the ventral region of the retina, together with a dorsal-specific enzyme RALDH-1. In chick and mouse embryos, the expression of Raldh-3 is observed first in the surface ectoderm overlying the ventral portion of the prospective eye region earlier than that of Raldh-1 in the dorsal retina, and then the Raldh-3 expression shifts to the ventral retina. Furthermore, we found that Raldh-3 is a downstream target of Pax6 which is known to be the master gene for the development of the eye in many species. RA is essential for the eye to develop. These results suggest that RALDH-3 and RALDH-1 are the key enzymes for the dorsoventral patterning at the early stage of retinal development.
　Furthermore, we recently identified a novel secretory protein, Ventroptin, which has BMP-4-neutralizing activity. Ventroptin is expressed in the retina with a ventral high-dorsal low gradient in the early stages. This expression pattern is complementary to that of BMP-4. At a later stage (E6-8), a nasal high-temporal low gradient expression pattern is also detected. Ventroptin thus shows a double-gradient expression profile along the dorsoventral and anteroposterior axes. Misexpression of Ventroptin altered expression patterns of several topographic genes: BMP-4, Tbx5 and cVax along the dorsoventral axis and ephrin-A2 along the anteroposterior axis. Consistently, in these embryos, projection of the retinal ganglion cell axons to the tectum was also changed along both axes. The topographic retinotectal projection along the dorsoventral and anteroposterior axes thus appears to be controlled not separately but in a highly concerted manner by Ventroptin.

B. Layer-specific projections

　After reaching their appropriate target zones along the rostrocaudal and dorsoventral axes of the tectum, retinal axons begin to seek their appropriate termination sites among 15 distinct laminae within the tectum, of which only three or four receive retinal projections. The molecular and cellular basis of such a discrete choice is poorly understood.
　In 1994, we screened for monoclonal antibodies that recognize one of these retinal termination laminae. We found three clones, TB5, TB2 and TB4, which labeled laminae B, D and F, respectively. cDNA cloning and immunochemical analysis revealed that the TB4 antigen was ezrin, a cytoskeletal-membrane linker molecule belonging to the ezrin-radixin-moesin family. Ezrin was selectively expressed in a subset of retinal ganglion cells that project to lamina F. A similar subset-selective expression and resultant laminae-selective distribution of ezrin were observed in the lamina-specific central projections from the dorsal root ganglia. The staining pattern for TB4 in the dorsal root ganglia and spinal cord indicated that the expression of ezrin was restricted to cutaneous sensory neurons. Since ezrin modulates cell morphology and cell adhesion profiles by linking specific membrane proteins with the cytoskeleton, it was suggested that ezrin may be involved in the formation and/or maintenance of lamina-specific connections for neuronal subsets in the visual and somatosensory systems.

C. Genetic labeling of specific axonal pathways

　Very recently, we generated transgenic mouse lines in which several sensory systems in the brain are specifically visualized genetically. We employed GAP-lacZ as an axon-targeted reporter protein that was constructed by fusing the membrane-anchoring domain of the GAP-43 protein to lacZ. The reporter gene was introduced into the genome under the control of a promoter element of Brn3b transcription factor to establish transgenic mouse lines. The individual lines thus generated afforded clear images of specific axonal pathways of the visual, vomeronasal, pontocerebellar, and auditory systems. The reporter protein labeled the entire axonal process as well as the cell body of developing and mature neurons on staining with X-gal. We showed that these lines facilitate the developmental and anatomical study of these neural systems. This strategy should be applicable to a variety of neural systems by using various specific promoter elements.

　We expect that our studies will lead to elucidation of the molecular mechanism underlying the formation of the retinotectal projection, and ultimately help to uncover the basic principles for establishing complicated but extremely precise neural networks in the nervous system.

II. Functional roles of protein tyrosine phosphatase ζ and γ

　Protein tyrosine phosphorylation plays crucial roles in various aspects of brain development and function. The level of tyrosine phosphorylation is determined by the balance between the activities of protein tyrosine kinases and protein tyrosine phosphatases. Many types of receptor-like protein tyrosine phosphatases (RPTP) have been cloned and characterized. On the other hand, proteoglycans are a family of proteins bearing sulfated glycosaminoglycans, which bind many extracellular matrix components and growth factors through their core protein and glycosaminoglycan portions. We have been interested in the functional roles of the brain-specific proteoglycans in the development of the nervous system because they are the major extracellular matrix components in the tissue.
　In 1994, we found that PTPζ/RPTPβ, abundant in the nervous system, is expressed as a chondroitin sulfate proteoglycan in the brain. The extracellular region of PTPζ consists of a carbonic anhydrase-like domain, a fibronectin-type III-like domain, and a serine-glycine-rich region which is considered to be the chondroitin sulfate attachment region. There exist three splice variants of this molecule: the full-length transmembrane form (PTPζ-A); the short transmembrane form which has a deletion in the extracellular region compared with PTPζ-A (PTPζ-B); and the soluble form (PTPζ-S), which is also known as 6B4 proteoglycan/phosphacan. PTPζis expressed from the early developmental stage to adulthood. This suggests that the gene plays variegated roles in the development and function of the brain.
　The RPTP subfamily, R5 has another member, RPTPγ, which shows a high degree of structural similarity to PTPζ. To reveal its molecular diversity, we cloned rat RPTPγ cDNAs and identified four splicing variants in 1997. We designated these forms RPTPγ-A, -B, -C and -S. RPTPγ-A was the longest species and had a similar structure to human and mouse RPTPγ. RPTPγ-B was devoid of the intracellular juxtamembrane 29 amino acids of RPTPγ-A. RPTPγ-C had a single phosphatase domain, and RPTPγ-S was a secretory-type RPTPγ. mRNAs of these four species were expressed in the brain, kidney, lung and heart. RPTPγ is thus comparable to PTPζ with regard not only to structure but also to the presence of both secretory and transmembrane isoforms. However, no RPTPγ variants are expressed as proteoglycans, in contrast to PTPζ variants.
　We began by searching for ligand molecules of PTPζ using 6B4 proteoglycan-Sepharose. We found in 1996 that PTPζ binds pleiotrophin/HB-GAM and midkine, closely related heparin-binding growth factors which share many biological activities. The chondroitin sulfate portion of PTPζ is essential for the high affinity binding (Kd = ~0.25 nM) to these growth factors, and removal of chondroitin sulfate chains results in a marked decrease of binding affinity (Kd = ~13 nM). We further revealed that chondroitin sulfate interacts with Arg78 in Cluster I, one of the two heparin-binding sites in the C-terminal half domain of midkine. This is the first demonstration that chondroitin sulfate plays an important regulatory role in growth factor signaling.
Next, we examined the roles of pleiotrophin/midkine-PTPζ interaction in neuronal migration using the glass fiber assay and Boyden chamber cell migration assay. Pleiotrophin and midkine on the substratum stimulated the migration of neurons in these assays. Experiments using various midkine mutants with various affinities for PTPζ indicated that the strength of binding and the neuronal migration-inducing activity are highly correlated. These results suggest that PTPζ is involved in migration as a neuronal receptor for pleiotrophin and midkine.
　In order to reveal the intracellular signaling mechanism of PTPζ, we performed yeast two-hybrid screening using the intracellular region of PTPζ as bait in 1998. We found that PTPζ interacts with PSD-95/SAP90 family members, SAP102, PSD-95/SAP90 and SAP97/hDlg, which are concentrated in the central synapses mediating protein-protein interactions to form large synaptic macromolecular complexes. Here, the C-terminus of PTPζ binds to PSD-95/SAP90 proteins through the second PDZ domain. This suggests that PTPζ is involved in the regulation of synaptic function. However, PSD-95/SAP90 family members are not likely to be the substrate for PTPζ because they are not tyrosine-phosphorylated.
　To identify the substrate molecules of PTPζ, we recently developed a yeast substrate-trapping system. This system is based on the yeast two-hybrid system with two essential modifications: conditional expression of v-src to tyrosine-phosphorylate the prey proteins and screening using a substrate-trap mutant of PTPζ as bait. Using this system, we isolated a number of candidate clones for substrate molecules or interacting molecules. Among them, we first identified GIT1/Cat-1 as a PTPζ substrate. GIT1/Cat-1 has ADP-ribosylation factor-GTPase-activating protein (ARF-GAP) activity. When B103 neuroblastoma cells were stimulated with pleiotrophin-coated beads, a transient increase in tyrosine phosphorylation of GIT1 was observed after 30 min. GIT1 is known to be tyrosine-phosphorylated by FAK and Src, although its physiological role is not clear. Overexpression of GIT1 in fibroblast and epithelial cells causes a loss of paxillin from the Rho-containing focal complex and stimulates cell motility. Cat-1 modulates Pak signaling, a serine threonine kinase which serves as a target for the small GTP-binding proteins, Cdc42 and Rac, and is implicated in a wide range of cellular events including cell adhesion and cell morphological change. Pleiotrophin, PTPζ and GIT1/Cat-1 may regulate the neuronal migration and neurite extension by controlling the Pak signaling pathway. We are continuing efforts to characterize the other candidate clones.
　To further study the physiological function of PTPζ in vivo, in 1997 we generated PTPz-deficient mice in which the PTPζ gene was replaced with the LacZ gene. First, we examined the cell types expressing PTPζ by investigating the expression of LacZ in heterozygous PTPζ-deficient mice. Throughout development from the early stage of embryogenesis, LacZ staining was essentially restricted to the nervous system. On embryonic day 12.5 (E12.5), LacZ staining was observed in the forebrain, midbrain, hindbrain and spinal cord. Examination of the cerebral cortex at higher magnification indicated that subsets of neurons including pyramidal neurons expressed LacZ. At the early postnatal stages, subsets of neurons and astrocytes in the brain including pyramidal cells in the hippocampus expressed LacZ. Both neurons and astrocytes were positive for LacZ in primary cultures of cells from the fetal cerebral cortex. From these results, we concluded that many neurons as well as astrocytes express PTPζ.
　We are currently studying the phenotype of PTPζ-deficient mice biochemically, anatomically, physiologically and ethologically, and have already found abnormalities in behavior, learning and memory, etc.

III. Physiological roles of Nax sodium channels

　The Nax channel, formerly called NaG/SCL11 (rat), Nav2.3 (mouse), and Nav2.1 (human), has been classified as a subfamily of voltage-gated sodium channels, although in primary structure, Nax is significantly different from other voltage-gated sodium channels, including in those regions important for voltage sensing and inactivation. However, the functional properties of the channel have been enigmatic, because all attempts at the functional expression of Nax in heterologous systems have failed. We recently generated Nax gene-knockout mice by inserting the lacZ gene in-frame, and found that this channel is expressed in neurons in the circumventricular organs (CVOs) like the subfornical organ (SFO) and organum vasculosum laminae terminalis (OVLT) which are known to be important regions for the control of body fluid balance. The Nax-deficient mice showed hyperactivity of neurons in these areas under thirst conditions and ingested excessive salt as compared with wild-type mice. This led us to hypothesize that Nax is involved in the sodium-level sensing mechanism in the brain.
　We verified this possibility by imaging analysis of changes in the intracellular sodium-ion concentration [Na+]i when [Na+]o was raised stepwise from the normal level. When [Na+]o was increased from the control level of 145 mM to 170 mM (high sodium solution) by bath application of NaCl solution, [Na+]i of the dissociated SFO neurons derived from wild-type mice showed a marked increase from ~10 mM to ~ 30 mM (41 of 70: 58.6%): all of the [Na+]i response-positive cells were Nax-immunoreactive cells (41 of 41: 100%). The threshold value for the [Na+]o was 157 mM. In contrast, none of the SFO neurons derived from Nax-deficient mice showed such responses (0 of 70: 0%). The findings were confirmed by the whole-cell current clamp technique. The inward currents with an average amplitude of 8.4 pA were observed only from the wild-type neurons. The current amplitude was consistent with that estimated from the ion-imaging study (6.7 pA).
　These results strongly suggest physiological roles for the Nax channel. It is well known that the sodium level and plasma osmolarity increase by 5-10% under thirst conditions. The sensitivity and threshold of Nax for [Na+]o is in the range of physiological change. SFO and OVLT are regions where the blood-brain barrier is deficient and directly accessible to monitor body fluid conditions. Thus, Nax is a newly identified type of sodium channel that is sensitive to an increase in the extracellular sodium concentration, and is likely to be the sodium-level sensor of body fluids in the brain.