The cell fate of precursor cells within the vertebrate central nervous system is in large part determined by the intersection of a dorsal-ventral patterning system with a rostral-caudal patterning system (Lumsden and Krumlauf, 1996; Jessell, 2001). The dorsal-ventral system is derived from an antagonistic relationship between ventralizing signals from the floor plate and notochord and dorsalizing signals from the surface ectoderm and the dorsal neural tube. This antagonistic relationship results in a variety of transcription factors, including Pax-3, Pax-6, and Pax-7, displaying regionally restricted expression patterns in the dorsoventral axis of the neural tube. Compartmentalization within the rostrocaudal axis depends, at least in part, on secreted signals from the anterior visceral endoderm (Tam and Steiner, 1999), which stimulate the formation of rostral structures, and on signals from the primitive streak and caudal paraxial mesoderm, which stimulate the formation of caudal structures (Muhr et al., 1999). Rostrocaudal patterning is readily discernible through the rostrocaudally restricted expression patterns of Hox genes. The result of these two axes of compartmentalization is a grid-like pattern, in which the coordinates of a particular progenitor cell within the grid plays a prominent role in the determination of its eventual mature neural identity (Jessell, 2001). For example, the dorsomedial roof of the third ventricle (DMRTV) of the postgastrulation vertebrate embryo includes the primordia for an almost contiguous rostral to caudal string of five organs, the subfornical organ (SFO), the organum vasculosum of the lamina terminalis (OVLT), the diencephalic choroid plexus (DCP), the pineal gland (PG), and the subcommissural organ (SCO). These organs represent significant portals between blood and neural tissue (Hofer, 1959; Johnson and Gross, 1993), or between cerebrospinal fluid (CSF) and neural tissue (Schreiber et al., 1990). As a result, these organs have all been categorized as circumventricular organs (CVOs) (Weindl, 1973; Tsuneki, 1986), although inclusion of the DCP is controversial (Johnson and Gross, 1993; Ganong, 2000). The CVOs are of considerable medical importance, due to their endocrine roles (including their control of salt and water balance in the body) and their roles in producing the CSF. As a result, the morphology and adult physiology of these organs has been extensively studied. However, the number of marker genes of the primordia of these structures to date remains small (Duan et al., 1991; Higuchi et al., 1995; Louvi and Wassef, 2000). Therefore, we have developed a screen for genes preferentially expressed in the avian DMRTV compared with the rest of the forebrain. This screen led us to identify several novel interesting genes, the first of which, DNTNP (dorsal neural tube nuclear protein), encodes a nuclear protein widely expressed throughout the dorsal neural tube. As we will demonstrate, DNTNP, is a member of a novel family of proteins, all of which contain elements resembling nuclear localization signals (NLS). We will also demonstrate that the potential NLS of DNTNP is a functional NLS, capable of targeting green fluorescent protein (GFP) to the nucleus.
Although our screen has not produced genes specific to the DMRTV, it has identified candidate genes that may play important roles in its development, possibly by contributing to the molecular definition of dorsal neural tube identity.
RESULTS AND DISCUSSION
Isolation and Expression of cDNTNP
Recognizable differentiation of CVOs from surrounding neural tissue begins around stage 18 in the chicken embryo. It is at this stage that the pineal primordium becomes identifiable as an evagination of the dorsal diencephalic neuroepithelium. The DMRTV spans the dorsomedial neural tube from the caudal telencephalon (including the SFO and the OVLT), through the diencephalon (including the DCP and the PG), and into the pretectum of the midbrain (including the SCO) (Fig. 1A). To isolate sequences preferentially expressed in the DMRTV in the stage 17–25 chicken embryo, suppressive subtractive polymerase chain reaction (PCR) was used to isolate sequences more highly expressed in the DMRTV relative to the ventral diencephalon and the telencephalic cortices. The cDNA fragments generated by this approach were subsequently screened for their expression pattern in the stage 18 chicken embryo by using whole-mount RNA in situ analysis. Of the 70 clones screened for expression, 5 showed expression patterns of interest in the DMRTV. Of these five clones, one denoted 1-122 (data not shown) was shown by sequencing to be a fragment of the previously reported Frizzled-10 gene, a known marker of dorsal avian neural tube (Kawakami et al., 2000). A second clone, initially denoted 1-73 displayed the interesting expression pattern depicted in Figure 1C–E. 1-73 is expressed throughout the dorsal central nervous system (CNS), both within the brain and the spinal neural tube. Strongest expression is observed within the tectum and the cerebellar primordium. Within the region of the pretectum (the posterior border of the DMRTV), and also the tectum, the expression of 1-73 extended more laterally than in other regions of the CNS. Outside the neural tube, expression is also observed in the somites, the heart, the branchial arches, and the limb buds.
An essentially similar pattern can be observed for the expression of 1-73 at an earlier point in development, stage 12 (Fig. 1B). At this stage, expression can already be observed in the dorsal neural tube, as well as the somites and the heart. Expression is notably absent from the presomitic mesoderm but turns on rapidly after segmentation.
Although the expression of this gene is clearly not specific to the DMRTV, and in fact, is highest in the tectum and the cerebellar primordium, its expression pattern indicates that it may play roles in the specification of the dorsal neural tube, which is a likely important step in the development of the DMRTV.
Rapid amplification of cDNA ends (RACE) experiments allowed us to isolate its full-length cDNA sequence with an open reading frame of 1206 bp (Genbank accession no. AF396666). As we will demonstrate that this protein localizes to the nucleus, we have named this protein as cDNTNP (chicken-dorsal neural tube nuclear protein). The remaining three clones identified by the screen are still being characterized and will be reported elsewhere. Like Frizzled-10 and cDNTNP, these three clones also display patterns consistent with dorsal neural tube specification, rather than being restricted to the DMRTV (data not shown).
cDNTNP Belongs to a Family of Uncharacterized Proteins
The open reading frame in cDNTNP potentially encodes a protein of 418 amino acids in length. It possesses a putative NLS between residues 285 and 293 (Fig. 2A). Other noteworthy features include a small hydrophobic amino terminus (residues 1 through 6) and a highly acidic carboxy-terminal domain from residue 323 to the carboxy-terminus. Comparison of the full-length sequence of cDNTNP to the databases by using the NCBI BLAST programs (Altschul et al., 1990, 1997) yielded two human sequences (Fig. 2A), c5ORF6 (NP057689) and KIAA0140. c5ORF6 is a gene discovered at 5q31, which encodes a putative nuclear protein expressed in hematopoietic cells and tissues (Lai et al., 2000). C5ORF6 was also isolated by random sequencing of cDNAs from retinoblastoma tissue. KIAA0140 is a putative nuclear protein (based on the presence of basic regions resembling NLS) derived by random sequencing of cDNAs from the KG-1 myeloblast cell line (Nagase et al., 1995). Amongst the related sequences in the DBEST database, there are also sequences representing a mouse gene (AK010537 from the RIKEN mouse ES cell library and AK017321 from the RIKEN mouse neonate head library [Shibata et al., 2000]) showing a particularly high level of similarity to cDNTNP. We used these mouse sequences to design primers and, thereby, reverse transcriptase-PCR (RT-PCR) amplify the gene from mouse cerebrocortical cDNA. The sequence of the mouse gene (AF396665) encodes a protein shown in Figure 2A. At the amino acid level, the mouse sequence shares 59% identity, KIAA0140 shares 33% identity, and c5ORF6 shares 26% identity to cDNTNP. This finding indicates the mouse sequence may be an orthologue of cDNTNP, and for the rest of the study, it will be referred to as mDNTNP. However, we recognize that a more extensive analysis of the family will be necessary before this relationship can be confirmed. The closeness of the mouse and chicken sequences, however, does preclude either of the human sequences from being orthologues of cDNTNP or mDNTNP (see Fig. 2B).
The family ranges from 392 amino acid residues to 422 amino acid residues in length. Given the alignment in Figure 2A, identical amino acid residues occupy 65 positions in all four sequences. The positions at which identity is completely conserved are highly clustered at two locations: 21 of the 65 occur between residues 112 and 157 (homology region I, HR I), whereas another 28 occur between residues 254 and 309 (HR II) (numbering according to cDNTNP). The putative NLS are located within HR II. Although all four members display a small hydrophobic amino terminus, only cDNTNP and mDNTNP display a strongly acidic carboxy-terminal domain.
cDNTNP Has a Functional NLS
To determine whether the putative NLS of DNTNP is indeed functional, we fused cDNTNP to green fluorescent protein (GFP-cDNTNP). Expression of the fusion protein in COS-7 cells resulted in its nuclear localization (Fig. 3A). Similarly, a fusion of a 40 amino acid fragment containing the NLS (residues 267-307) to GFP (GFP-cNLS) resulted in nuclear localization (Fig. 3C), establishing the putative NLS as a functional nuclear localization signal. Although the nuclear distribution of GFP-cNLS was uniform, that of GFP-cDNTNP frequently showed a restricted subnuclear distribution, reminiscent of nuclear bodies. Fusion of GFP to an amino-terminal fragment (residues 1-307) or a carboxy-terminal fragment (residues 267-418), both of which retain the putative NLS, also resulted in nuclear localization (data not shown). As expected, GFP alone did not show nuclear localization (Fig. 3E).
Screen for Genes Preferentially Expressed in the DMRTV
We used the suppressive subtractive PCR approach (Diatchenko et al., 1996) to identify genes preferentially expressed in the DMRTV. This approach allows the identification of genes preferentially expressed in one particular tissue (referred to as the tracer) in comparison to a second tissue (referred to as the driver). DMRTV tissue was dissected from 200 chicken embryos ranging in development from stage 17 to stage 25 (as defined by Hamburger and Hamilton, 1951) and combined for use as the tracer tissue. From the same embryos we also dissected out telencephalic cortical tissue and ventral diencephalic tissue, which were combined to form the driver tissue. Total RNA was isolated separately from the tracer and the driver tissue by using the single step method (Chomczynski and Sacchi, 1987; Kingston et al., 1994b). Tracer and driver RNA was converted to cDNA and amplified as outlined in the instructions for the SMART PCR amplification kit (Clontech) (Endege et al., 1999). Amplified cDNAs were then subjected to suppressive subtractive PCR, as outlined in the instructions for the PCR-select subtraction kit (Clontech).
Product from the suppressive subtractive PCR was cloned into pSLAX13B, a derivative of the pSLAX13 vector (Hughes et al., 1987) with an altered polylinker site. Of the resulting clones, 200 were picked and checked for insert size. Seventy clones that displayed inserts in the range of 200 to 700 bp were subjected to sequence analysis. The expression pattern of each of these clones in stage 17 to 19 chicken embryos was then determined through whole-mount RNA in situ analysis (Tribioli et al., 1997). Digoxigenin-UTP– labelled RNA probes were generated through transcription of individual clones. As the library was not oriented, probes were generated for each clone by using both T3 and T7 RNA polymerases, and tested separately. Each treatment generated probe corresponding to the full length of the insert.
In situ analysis of cDNTNP was performed by using probe generated from the original 1-73 clone. 1-73 DNA was linearized by HindIII and then used as template for RNA synthesis by T3 RNA polymerase. The resulting antisense probe spanned the region from nucleotide residue 150 to residue 438 of the sequence submitted to the databases, which encodes residues 51 to 146 of the amino acid sequence displayed in Figure 2A. Identical results were obtained by using an antisense RNA to sequences in the 3′ noncoding region of cDNTNP. The sense control RNA (generated by using XbaI-linearized 1-73 DNA as a substrate for T7 RNA polymerase) gave no signal.
RACE and RT-PCR Experiments
Total RNA from chick fetal brain (embryonic day 14) and mouse cerebrocortex was extracted with TRIZOL reagent (Life Technologies, Gaithersburg, MD). RACE and RT-PCR experiments were performed according to standard procedures. For cDNTNP, primer O06 (5′-CCACCGCTGTTGTATCGTCTCTTTG-3′) and O07 (5′-AAACCTCAACGAAAACTCGGGTCAG-3′) were used for 5í and 3í RACE, respectively. The full-length cDNTNP was RT-PCR amplified from total RNA by using primer O31 (5′-GCCTCCAACACTTCTGTGTG-3′) and O32 (5′-ATGCATGGCAAAGGTAATCC-3′). The mouse transcript was amplified by using primer O28 (5′-AGAGCCCGGCCATGGTCAC-3′) and O29 (5′-TCAGCCAAAGTCAGTTGTTC-3′). All amplified products were cloned into pBSK+ vector and sequencing analyzed.
The full-length cDNTNP cDNA was cloned as a C-terminal fusion to GFP by means of BamHI cloning in pEGFP-C1 (Clontech). Cells were transfected with 0.5 μg of each plasmid construct per 3.5-cm plate, by using the calcium phosphate method (Graham and van der Eb, 1973; Kingston et al., 1994a), and then plated on coverslips. Forty hours after transfection, the cells were rinsed with phosphate buffered saline and then fixed with 4% paraformaldehyde for 5 min. Coverslips were then mounted with VectaShield mounting medium containing DAPI (Vector Laboratories, Burlingame, CA). Cell were analyzed and photographed by using a ZEISS Axioplan microscope.
We thank Edward Johnson, Jonathan Licht, Francesca Cole, Tom Lufkin, Philip Mulieri, and David Wolfe for helpful discussions. We also thank Lucy Skrabanek, of the Institute of Computational Biomedicine at the Mount Sinai School of Medicine, for her help with the phylogenetic analysis. This work was funded by grants to A.D.B. from the Whitehall Foundation and the National Institute for Child Health and Development.