Expression of the tyrosine kinase receptor gene trkB is confined to the murine embryonic and adult nervous system

An important objective in the field of vertebrate developmental biology is the identification of genes whose products mediate the regulatory signals required for the intricate process of embryogenesis. Various approaches have been undertaken in efforts to identify such genes. For example, the search for mammalian homologues of invertebrate embryonic regulatory genes has led to the isolation of a number of genes that are related by sequence to the homeodomain-containing genes of Drosophila (McGinnis et al. 1984). The discovery of murine homeobox (Hox)-containing genes has provoked investigation into their potential role in embryogenesis, which gained considerable support once restricted embryonic patterns of Hox expression were observed. For example, distinct Hox genes are transcribed at specific rostro-caudal levels of the neural tube and spinal cord as development proceeds, suggesting a role for Hox genes in the spatial alignment of the nervous system (reviewed by Holland and Hogan, 1988; Gaunt et al. 1989; Wilkinson et al. 1989). Another approach toward searching for developmentally important regulatory genes stems from the existence of conserved sequence motifs among transcriptional factors (ie., zinc fingers, POU-domains, etc.). The presence of such sequences within a gene is widely accepted as evidence for a gene having a role in gene regulation (Herr et al. 1988). Krox-20 and mKr2 are examples of genes encoding zinc finger proteins that are expressed in the embryonic nervous system. Krox20 is transcribed in the second and fourth neuromeres during early neurogenesis (Wilkinson et al. 1988), and mKr2 is expressed in various tissues of the developing nervous system (Chowdhury et al. 1988). Similarly, a series of genes that encode POU sequences have been shown to have neural specific patterns of expression during development and in the adult mouse (He et al. 1989). It has been suggested that these genes may be participants in specifying cellular phenotypes during development of the nervous system.

Cell surface receptors containing tyrosine kinase (TK) catalytic domains have been extensively studied for their role in cell growth regulation and oncogenic transformation (Hunter and Cooper, 1985; Carpenter, 1987; Yarden and Ullrich, 1988). However, evidence for their involvement in embryonic development has only recently been obtained. The mouse locus W, which affects the proliferative or migratory properties of primordial germ cells, melanoblasts and hematopoietic stem cells, encodes a TK receptor (Silvers, 1979; Geissler et al. 1988; Chabot et al. 1988). In Drosophila, the genes sevenless and torso encode putative TK receptors. By mutation analysis, it is known that the sevenless TK receptor is required for the determination of the R7 photoreceptor cell in the retina during embryogenesis (Basler and Hafen, 1987; Banerjee etal. 1987; Hafen et al. 1987). Similarly, the torso TK receptor is a maternally encoded member of the terminal class of genes required for the determination of the terminal antero-posterior structures of the embryo (Sprenger et al. 1989). We report here the neural specific embryonic expression pattern of another gene that codes for a TK receptor, trkB. The trk family of proto-oncogenes consists of at least two loci in the mouse: trk and trkB (Klein et al. 1989, and Martin-Zanca et al. 1990). Trk was first identified as an oncogene following transfection of NIH-3T3 cells with DNA obtained from a human colon carcinoma (Martin-Zanca et al. 1986). Northern analysis using a human Trk cDNA probe to hybridize RNA from mouse tissues revealed a weakly hybridizing, complex pattern of bands in brain. Subsequent studies have shown that these brain transcripts originate from a locus distinct from that of the trk gene, which we have designated trkB (Klein et al. 1989). Organization of the predicted trkB transmembrane, extracellular and catalytic TK domains is colinear with that of the trk protein and is described in detail elsewhere (Klein et al. 1989). Furthermore, we have shown, in a preliminary study, that trkB transcripts can be found in the mouse embryo nervous system. In this study, we present a more detailed RNA in situ analysis for the trkB gene in the embryo and in the adult brain. Our data demonstrate exclusive expression for this gene in tissues of neuroepithelial and neural crest origin, which are destined to constitute the central and peripheral nervous systems (CNS and PNS) from the time of morphogenesis, through development, and into maturity.

Mouse embryos were derived from C57BL/6 NCR ×C3H/ HeN MTV^- F2 litters and were staged under the dissecting microscope by counting somites through midgestation. In older embryos, external markers such as state of limb development were used (Theiler, 1972). The morning of vaginal plug was considered day 0.5 (E0.5).

Embryos were dissected under the microscope, frozen in liquid nitrogen and stored at – 70°C. RNA was extracted using RNAzol™ (Cinna/Biotecx) following the manufacturer’s recommendations. For Northern analysis, 20 μg of total RNA were electrophoresed through 1.2% agarose gels containing 0.37 M formaldehyde as described (Maniatis et al. 1982), transferred to ZETABIND membranes (CUNO) and hybridized at 65°C according to Church and Gilbert (1984). Washes were carried out at 65°C in 0.5% BSA, 40 mM phosphate buffer pH 7.2, 1% SDS.

In situ hybridization protocols were as described in detail elsewhere (Klein et al. 1989; Martin-Zanca et al. 1990).

RNA (Krieg and Melton, 1987) and DNA probes were prepared from the 482 bp insert in pFRK16 (see Fig. 1A). This insert encompasses sequences encoding a portion of the putative extracellular domain (nucleotides 1181-1663 in Klein et al. 1989). This region was chosen to avoid possible recognition of related TK sequences.

Our previous analysis had revealed the existence of at least six independent transcripts in adult brain whereas only a single abundant 2.5 kb transcript was detected in E14 and E18 mouse embryos (Klein et al. 1989). We wished to extend this study and to determine whether additional trkB transcripts were present in the embryo that had been below our sensitivity of detection in prior analyses (Klein et al. 1989). We therefore performed Northern analysis of total RNA isolated from mouse embryos of stages E6.5 to E17.5. A 0.5 kbp DNA fragment from the trkB cDNA that corresponds to a portion of the region encoding the putative ligandbinding extracellular domain was used for preparing DNA and RNA probes (Fig. 1A; see Materials and methods). As shown in Fig. 1B, a 2.5 kb mRNA species is first observed in E8.5 embryos, albeit at extremely low levels. This 2.5 kb transcript and a smaller 2.0 kb species that appears at stage E10.5 are the most abundant transcripts throughout gestation, while additional transcripts that may exist below our level of detection early, are evident only at later stages of development. For example, at stages E12.5 to E13.5, two RNA bands of approximately 8.0 and 9.0 kb appear preferentially in RNA from heads of embryos, although trace amounts can be detected in RNA from trunks. By stage E15.5, the larger transcript is no longer visible in heads while the 8.0 kb transcript remains present at low levels in trunks. In E16.5 embryos, the complexity of trkB transcripts increases, approaching that observed in the adult (Klein et al. 1989). The existence of multiple size trkB encoded transcripts ranging from 2.0kb to 9.0 kb has important implications for the interpretation of the results discussed below.

To determine whether temporal appearance of trkB expression coincided spatially with particular developmental events, in situ hybridization analysis was performed on embryo sections corresponding to all stages tested by Northern analysis. trkB expression was first seen in E9.5 embryos. At this stage, the embryo body plan is well established, about 25 somites are formed, the rudiments of several organs are appearing and the CNS and PNS are in the process of differentiating into anatomically recognizable structures. These include the primary divisions of the brain into forebrain, midbrain and hindbrain, and the formation of the cranial and spinal ganglia. We detect trkB transcripts in forebrain, caudal midbrain, hindbrain, spinal cord and differentiating neural crest cells, which form dorsal root ganglia

(DRG). Expression is also present along the dorsal aorta, which is utilized by the presympathetic neural crest cells as a migratory pathway (Fig. 2A and B). A magnified view of the isthmus between telencephalon and mesencephalon demonstrates that trkB mRNA expression neither correlates with the location of the mitotic regions (ependymal layer) nor does it reflect potential artifacts due to variable cell density, but rather with expression in a specific set of morphologically indistinguishable cells located at the fold (Fig. 2C and D). Similarly, a frontal section through the myelencephalon of an E9.5 embryo further illustrates the localized expression of trkB within a subset of undifferentiated neuroepithelial cells. Here, trkB expression displays a bilaterally symmetric profile along the lateral neuroepithelium. Additionally, trkB hybridization is found within the forming fifth cranial (trigeminal) ganglion. By day 10.5, the neural crest cells that were seen along the descending aorta a day earlier are in place, and begin forming the sympathetic network. trkB expression can be detected within the morphologically discernible neural crest-derived sympathetic chains and plexuses of the autonomic nervous system (data not shown). Thus, at stage E9.5, trkB expression is confined to morphologically indistinguishable regions of the developing CNS and PNS and, a day later, expression is found in differentiating structures of the sympathetic and autonomic nervous system.

The onset of organogenesis coincides with the period of maximal neuronal proliferation in the sensory ganglia of the PNS and with the extension of axonal bundles and migration of pre-Schwann cells (Altman and Bayer, 1982). trkB expression is found in regions of the PNS in addition to continued expression in the CNS. This is illustrated by a frontal section of an E13.5 embryo (Fig. 3), where the pattern of trkB hybridization appears to reflect the entire network of the developing CNS and PNS. The lateral walls of the telencephalon contain high levels of trkB transcripts and the trajectories of the ophthalmic and maxillary branches of the trigeminal nerve can be traced to their respective sites of innervation, the eye and the palate (Fig. 3A and B). In the same figure, the caudal view of the embryo shows a cross section at the level of the hind limb. For easier analysis, this caudal area is shown magnified in Fig. 3D and E. Here, the pattern of trkB hybridization can be used to trace the extension of the PNS from the spinal cord to the lateral and visceral innervation sites as well as to the migrating pathways of the crest cells that form the autonomic ganglia (compare panels C and E). Moreover, the restricted expression in DRG to a subset of neurons is well illustrated in this figure.

In late fetal development, the scaffolding for the entire nervous system network is fully constructed. As can be seen in a sagittal view of an E16.5 fetus, trkB expression is maintained throughout the network. Continued high levels of expression are visible in the CNS, notably in the spinal cord, the medullar auricle, the olfactory lobe and the ependymal layer of the fourth ventricle. The facial structures display a complex pattern of expression that outline innervation pathways (Fig. 4B). A myosin heavy chain probe, which hybridizes to skeletal muscle, was used as a control for probe specificity in a section from the same embryo, resulting in a completely distinct profile from that obtained with trkB probes (Fig. 4B and C).

trkB transcripts were initially detected by Northern analysis in adult mouse brain and the probes used for our studies have been derived from an adult brain cDNA library (Klein et al. 1989). To determine the spatial distribution for trkB in adult CNS, and whether clues about physiological function might be provided, serial brain sections were surveyed by in situ hybridization. These experiments confirmed that expression of this gene was sustained in the adult and that at least some form of trkB mRNA is found in most cell types of the forebrain. A coronal section through a dorsal region of the cerebrum demonstrates the presence of trkB transcripts throughout the cortical layers, the thalamus, and the hippocampus (Fig. 5A,B). Regions of strongest hybridization include the pyramidal layer of the dentate gyrus, the medial habenular nucleus, the lateral choroid plexus, the ependymal lining of the third and lateral ventricles, and the mammilary recess. In the hippocampus, the variation in concentration of hybridizing grains is at least partially attributable to the distribution of cell bodies as can be seen by direct comparison of the bright-field plate (Fig. 5A), where darker staining reflects greater cell density, and the dark-field plate where the highest grain densities correlate with the nuclear rich pyramidal layer (Fig. 5B). It is worth noting that in deviation from the apparent generalized expression, there is marked absence of trkB transcripts in the corpus callosum (not shown), the parafasicular and reticular thalamic nuclei, and in the fimbria hippocampus, at this level of sensitivity. A more restricted pattern of trkB expression is seen in the cerebellum with high transcript levels in the Purkinje cell layer and in the caudal peduncle while hybridization to other cerebellar neurons (Fig. 5C) is below detection. In the brainstem, expression is varied with notably high levels in the motor and spinal tracts of the trigeminal nerve, and in the vestibular nuclei. Thus, in the adult mouse brain, many neuroblasts, neurons and glia alike express at least some form of trkB mRNA. However, expression of this gene is not uniform, with areas of high mRNA, such as the dentate gyrus and the ependyma, and areas that are devoid of detectable transcripts such as certain thalamic nuclei (Fig. 5B). In the cerebellar folds, a unique type of neurons, the Purkinje cells, present high levels of transcripts, while other neurons and glia have significantly lower levels, and no transcripts are detected in the granule cells (Fig. 5C).

All well-characterized transmembrane receptors that contain tyrosine kinase catalytic domains mediate specialized growth, differentiation, or morphogenic signals at the cell surface (for references, see Hunter and Cooper, 1985; and Carpenter, 1987). In Drosophila, TK receptors are known to play fundamental roles in the differentiation of embryonic terminal structures (Sprenger et al. 1989) and of photoreceptor cells (Basler and Hafen, 1987). These affirmations have been possible through the powerful genetic analysis available with this organism. In vertebrates, while the importance of TK receptors in cell growth control and oncogenesis is well established, evidence of their involvement in developmental processes is only now emerging. Compelling evidence has recently been reported indicating that the affected gene in the mouse developmental mutation, W, is the homologue of the protooncogene c-kit, a gene that encodes a TK receptor

(Geissler et al. 1988; Chabot et al. 1988). Structural analysis of the mutations in the c-kit locus indicate that the severity of phenotype correlates with mutations in the tyrosine kinase catalytic domain (P. Besmer, personal communication).

Among the nonreceptor tyrosine kinases, expression of the c-src and c-abl proto-oncogenes has been localized to tissues of neuronal origin. In mouse, a variant form of the c-src protein that encodes a pp60^c-src with six additional amino acid residues, is expressed specifically in the brain (Martinez et al. 1987), while during Drosophila embryogenesis, the abl protein is localized in the axons of the CNS (Gertler et al. 1989). No associated mutations have been described.

In this study, we show that the trkB gene, encoding a TK transmembrane receptor, is expressed in tissues of neuroectodermal origin during mouse embryogenesis and postnatally. We find transcripts from this gene in cells derived from the neuroepithelium and neural crest, which is in contrast to the expression profile of the highly related trk proto-oncogene whose expression is confined to sensory neurons of neural crest origin (Martin-Zanca et al. 1990). Although we detect expression from the trkB locus in the neural tube, brain and spinal cord, beginning with the onset of neurulation, we have been unable to detect trkB transcripts in the neural crest per se, prior to cell migration. Many TK receptors such as EGF, CSF-1 and PDGF receptors have been associated with mitogenic properties (Yarden et al. 1988). The location of trkB transcripts in the embryo, within CNS or PNS structures, does not correlate with known sites of cell proliferation. In adult brain, for example, expression is present in postmitotic, terminally differentiated cells (ie., Purkinje cells). These observations suggest that the trkB product does not function primarily in a mitogenic pathway.

A diversity of expression is observed for the trkB gene in the CNS and PNS among distinct cell types and even within apparently similar cell types. Thus, in the CNS, at stage E9.5 there already exist differences in gene expression within the apparently homogeneous neuroepithelium, with regions of high trkB expression and regions of low or undetectable expression. Likewise, in the PNS, within DRG and the trigeminal ganglion, the presence of trkB transcripts appear confined to a portion of the cell bodies from the earliest stages of morphogenesis. These observations indicate that trkB is a marker for a subset of neurons that may have diverged functionally, while retaining morphological homogeneity within these structures, from the earliest stages of differentiation. Further, it indicates the existence of cell types common to both structures, a phenomenon also observed with the closely related trk proto-oncogene (Martin-Zanca etal. 1990).

During embryonic development, as the nervous system diversifies, the spatial pattern of trkB expression becomes accordingly complex. The available data may best be reconciled by the appearance of multiple transcripts ranging from 2.0 to 9.0 kb, beginning with the onset of neural differentiation. The existence of more than one trkB promoter would provide the capacity to target and compartmentalize distinct transcripts to specific cell types, while alternative splicing of coding exons could result in the generation of diverse receptors from one transcriptional unit. This latter possibility is supported by the fact that the smallest transcripts (2.0 and 2.5 kb) do not have the capacity to encode gpl45^trkB, the only protein characterized to date from this locus (Klein et al. 1989), suggesting that smaller receptor forms may exist. It is therefore possible that the trkB gene encodes for several independent receptors that may either bind different ligands or that may interact with different intracellular messenger pathways. Resolution of these alternatives and their implications is in progress with the generation of probes that recognize unique or specific subsets of the diverse trkB transcripts observed.

The c-kit proto-oncogene represents the first reported example of a TK receptor with a role in mammalian embryonic development. It is likely that additional examples will follow, as has been the case in Drosophila. The exclusive and early expression of the trk (Martin-Zanca et al. 1990) and trkB proto-oncogenes in the mouse nervous system suggests that they are good candidates for mammalian TK receptors that contribute to the appearance of specific neuronal cell phenotypes.

We thank Gretchen White and Aliki Grammatikakis for their assistance with embryo preparations and are grateful to the members of the Mammalian Genetics Laboratory for their encouragement and discussions. In particular, we thank Brian Stanton, Peter Donovan and Leslie Lock for their helpful comments, Neal Copeland for critically reading the manuscript, and Robin Handley for help in its preparation. This research was supported in part by the National Cancer Institute, DHHS, under contract N01-CO-74101 with BRI. The NCI-Frederick Cancer Research Facility is fully accredited by the American Association for Accreditation of Laboratory Animal Care.