Transgenic Analyses Reveal Developmentally Regulated Neuron- and Muscle-specific Elements in the Murine Neurofilament Light Chain Gene Promoter*

We report here the developmental activity of regulatory elements that reside within 1.7 kilobases of the murine neurofilament light chain (NF-L) gene promoter. NF-L promoter activity is first detected at embryonic day 8.5 in neuroepithelial cells. Neuron-specific gene expression is maintained in the spinal cord until embryonic day 12.5 and at later developmental stages in the brain and sensory neuroepithelia. After day 14.5, the promoter becomes active in myogenic cells. Transgene expression in both neurons and muscle is consistent with the detection of endogenous NF-L transcript in both neuronal and myogenic tissues of neonates by reverse transcriptase-polymerase chain reaction. Neuron- and muscle-specific activities of the NF-L promoter decrease and are nearly undetectable after birth. Thus, the 1.7-kilobase NF-L promoter contains regulatory elements for initiation but not maintenance of transcription from the NF-L locus. Deletion analyses reveal that independent regulatory elements control the observed tissue-specific activities and implicate a potential MyoD binding site as the muscle-specific enhancer. Our results demonstrate that the NF-L promoter contains distinct regulatory elements for both neuron- and muscle-specific gene expression and that these activities are temporally separated during embryogenesis.

Neurofilaments (NFs) 1 are a class of intermediate filaments that comprise the major cytoskeletal component of differentiated neurons. The mature filaments are assembled from three polypeptide subunits: light (NF-L), medium (NF-M) and heavy (NF-H) (for review, see Ref. 1). Each subunit is encoded by a distinct gene and is regulated differently during development, with NF-L and NF-M being expressed early at the time of neuronal cell differentiation and NF-H expressed much later (2,3). The levels of NF gene expression are low during embryonic development with an up-regulation of transcriptional activity at postnatal time points (4 -6). The neuronal selectivity of neurofilaments has resulted in numerous studies aimed at identifying molecular mechanisms that regulate NF gene expression in neurons; nevertheless, little is understood about these processes.
One approach to elucidate the regulation of gene expression is to fuse the genetic elements of interest to a reporter gene and monitor its expression in transfected cells or transgenic mice. In this way, the 5Ј-promoter regions of both the mouse and human NF genes have been shown to contain multiple positive and negative regulatory sites (7)(8)(9)(10). However, in vivo attempts to identify neuron-specific elements within NF genes in transgenic mice have proven difficult (9,(11)(12)(13)(14)(15)(16). For example, one report claims that 300 bp of the human NF-L gene promoter sequence are sufficient for neuronal expression (9) while another report finds this region insufficient for neuron specificity (11). Such discrepancies have been interpreted as a relative weakness of the NF-L gene promoter regions in combination with a requirement for intragenic regulatory elements (9, 11, 16 -18). Further complications in understanding the tissue specificity of NF gene expression arise from in vitro studies that report efficient expression of either reporter genes, or NF genes themselves, under the control of various NF promoters when transfected into non-neuronal cells (7, 13, 19 -23). In addition, the identification of elements in the mouse NF-L gene 3Ј-untranslated region that stabilize the NF-L mRNA transcript suggests that both transcriptional and post-transcriptional mechanisms play a role in NF gene expression (24).
The majority of these in vivo studies were conducted in adult mice and do not address the developmental regulation of NF gene expression during embryogenesis. We reasoned that the initiation of transcription during development was a crucial step in attaining neuron-specific gene expression. As a model, we chose the promoter from the murine NF-L gene that has previously been used to direct the neuronal expression of several transgenes in adult animals (25)(26)(27)(28). In this study, we directed embryonic expression of a lacZ reporter gene under the control of the murine NF-L promoter using a 2-tiered system of gene regulation in transgenic mice (25,29). Expression of the reporter transgene in this system is dependent on the potent viral transcriptional transactivator, VP16. We report here the developmental activity, structure, and identification of critical tissue-specific regulatory regions of the murine NF-L gene promoter. Our results demonstrate that 1.7-kb upstream of the mouse NF-L promoter contain temporally separable and distinct enhancer elements that regulate transcription in neuronal and myogenic tissues during embryonic development.

EXPERIMENTAL PROCEDURES
Generation of Transgenic Mice-Transgenic mice were established following the standard protocols described in Hogan et al. (30). The NF-TIF4 and IE-LacZ16 transgenic mouse lines have been previously described (25,29), and all other lines were newly generated. The plasmid NFL-TSV (a gift from G. Byrne, Nextran, Inc., Princeton, NJ) was created by ligating the 0.87-kb SalI/BamHI fragment containing the SV40 virus splice region, intron and poly(A) site (pMSG, Pharmacia Biotech Inc.) into the EcoRI site of pNF-TIF (25). The resulting insert DNA was released by a ClaI/BamHI restriction digest. Upon complete digestion, the DNA fragment was purified over a 20 -40% sucrose gradient in a near vertical rotor and dialyzed extensively against 10 mM Tris-Cl, pH 7.3, 0.25 mM EDTA. 2 Purified DNA was diluted to a final concentration of 1.5-2.0 g/ml for microinjection. FVB mice (31) were used for microinjection and CD-1 mice were used as foster mothers.
Identification of Transgenic Mice-Genomic DNA from potentially transgenic founder mice was purified from tail biopsies (32). The polymerase chain reaction (PCR) was used to identify transgenic mice by virtue of the presence of the VP16 gene. A 494-bp VP16-specific PCR product was amplified using the primers 5Ј-CTATGTACCATGCTC-GATACCTGGA-3Ј and 5Ј-CCATAGGATCTCGCGGGTCAAAAAT-3Ј for 35 cycles (30 s at 94°C, 1 min at 60°C, 30 s at 72°C). DNA from PCR-identified founder animals was subjected to Southern blot hybridization to determine transgene copy number and to ascertain the integrity of the transgene (33). Samples were hybridized with a digoxygeninlabeled 1.2-kb SalI probe corresponding to the VP16 gene. Specific hybridization was detected using the Genius system (Boehringer Mannheim).
Embryo Isolation and ␤-Galactosidase Staining-Embryos at various developmental stages were isolated after natural matings. The morning of the presence of a vaginal plug was designated as 0.5 days post-coitum (dpc). The embryos were removed from the uterus and rinsed in phosphate-buffered saline (PBS). The yolk sac was separated from each embryo and used as a source of embryonic genomic DNA for PCR analyses. Embryos younger than 9.5 dpc were fixed in 4% paraformaldehyde in PBS for 30 min at 4°C, those between 10.5 and 12.5 dpc were fixed for 2 h, and those older than 13.5 dpc were fixed for 4 h. The fixative was rinsed away with three washes of 15 min at 37°C in a substrate buffer containing 100 mM NaH 2 PO 4 , pH 7.3, 2 mM MgCl 2 , 0.1% deoxycholate, 0.2% Nonidet P-40. Staining proceeded in the same buffer supplemented with 5 mM K 3 Fe(CN) 6 ⅐3H 2 O, 5 mM K 4 Fe(CN) 6 , and 1 mg/ml 5-bromo-4-chloro-3-indolyl ␤-D-galactopyranoside (X-gal). The staining reaction was incubated for 2 h at 37°C and was monitored by the appearance of staining in the apical ectodermal ridge of embryos from the IE-LacZ16 line. This ectopic staining served as an internal control for the ␤-galactosidase staining reaction and is independent of VP16 transactivation (29). The staining was stopped by rinsing with PBS and post-fixing in 4% paraformaldehyde, 0.25% glutaraldehyde in PBS. Embryos were dehydrated in increasing gradations of ethanol and cleared in xylene. Photographs were taken with a 35-mm camera (Wild) on a Leica Stereo Zoom M3Z microscope. Histologic sections were obtained by embedding the embryos in paraffin wax and microtome sectioning at 10 m thickness. Sections were analyzed and photographed with a Leica DMRB microscope.
RNA Isolation and Reverse Transcriptase-PCR-RNA was prepared from newborn mice 10 days after birth. Brain, spinal cord, muscle, skin, liver, spleen, kidney, and heart tissues were isolated and flash frozen in liquid nitrogen. The tissue was thawed in the presence of Tri Reagent (Molecular Research Center, Cincinnati, OH) and homogenized. Total RNA was prepared according to the manufacturer specifications. Contaminating DNA was removed by treating 2 g of each RNA sample with DNaseI (Life Technologies, Inc.) in the presence of RNase Inhibitor (CLONTECH). First strand cDNA synthesis proceeded with 500 ng of RNA using the primers dT 16 VA, dT 16 VT, dT 16 VG, and dT 16 VC (12.5 M each), where V ϭ G, C, or A, and 100 units of Superscript II reverse transcriptase (Life Technologies, Inc.) for 50 min at 40°C, 15 min at 70°C. The manufacturer protocols were followed for all enzymatic reactions. For each primer set, a one-twentieth volume of cDNA was subjected to PCR amplification using cycling conditions identical to those stated above for 33 cycles. When primers were used to detect ␤-actin, 30 total cycles were used. Positive and negative control samples were carried out for each amplification reaction. The primers used were VP16 (5Ј-CAACTGACGCCAGCTCTCCAGGTCGC-3Ј and 5Ј-CTCGTT-TCTTCCACGCCGAGCTAC-3Ј), lacZ (5Ј-ACCGAATTCAGTTGGTCTG-GTGTC-3Ј and 5Ј-GTGAAGCTTGTGATGCTATTGCTT-3Ј), NF-L (5Ј-AGCAGAATGCAGACATTAGCGCC-3Ј and 5Ј-TGGTCTCTTCGCCT-TCCAAGAGT-3Ј), and ␤-actin (5Ј-GTGGGCCGCTCTAGGCACCA-3Ј and 5Ј-TGGCCTTAGGGTGCAGGGGG-3Ј). Amplification products were separated by 1% agarose gel electrophoresis and visualized by ethidium bromide staining. Three independent cDNA preparations were analyzed.
Sequence Analysis of the NF-L Promoter-The sequence of the murine NF-L promoter was obtained using the plasmid NFL-TSV as a double-stranded DNA template. This plasmid contains 1.7 kb of the mouse NF-L promoter (Ϫ1.6-kb HindIII/ϩ74 SmaI). Automated DNA sequencing was performed on an Applied Biosystems 373A DNA Sequencer (Foster City, CA). Primers were specifically designed to provide sequence data from both DNA strands. Subsequent sequence analyses were carried out using the programs of the University of Wisconsin Genetics Computer Group (Madison, WI) and McMolly Tetra (Soft Gene GmbH, Berlin).
Generation of Deletions within the NF-L Promoter-The plasmid NFL-TSV was linearized by ClaI digestion. Deletions within the NF-L promoter were subsequently generated by partial digestion with one of the following enzymes: HaeII (at base Ϫ1215), BstXI (base Ϫ941), NspI (base Ϫ524), and NruI (base Ϫ328). The fragment of interest was purified from agarose gels and used to generate transgenic embryos as described above. Embryos were isolated at 13.5 dpc and stained for ␤-galactosidase activity as described previously.

RESULTS
The Neurofilament Light Chain Gene Promoter Directs lacZ Expression in a Binary Transgenic Mouse System-To examine the developmental activity of the murine neurofilament light chain gene promoter, we have employed a binary system of gene expression in transgenic mice (25,29). This system ( Fig.  1A) consists of two independent transgenic components. One transgenic parent, the transactivator, carries the gene for the transcriptional activator VP16 from herpes simplex virus under the control of the 1.7-kb murine NF-L promoter. The second transgenic parent, the transresponder, provides the reporter gene, lacZ, linked to the viral immediate early (IE) ICP4 gene promoter. The IE promoter contains the response elements for VP16-mediated transactivation but is not transactivated in the absence of VP16 (25,29). Progeny that receive both transgenes will exhibit lacZ expression in those cells where the NF-L promoter is active. The structure of the transgene DNA constructs used in these studies is presented in Fig. 1B.
Previous studies, using an IE-CAT transresponder mouse mated with the NF-TIF4 transactivator line, demonstrated CAT activity in homogenates of the brain and spinal cord of double transgenic adult progeny (25). We identified and established six NFL-VP16 mouse lines with the ability to transactivate the expression of an IE-LacZ transresponder gene. The activity of each NFL-VP16 transgene was ascertained by mating a male NF-L-transactivator to a female IE-LacZ16 mouse. Embryos were isolated at 10.5 dpc, and their genotypes were determined by PCR using transgene-specific primers on yolk sac genomic DNA (not shown). Transgenic progeny from several independent NF-L-transactivators showed the same pattern of expression of the IE-LacZ transgene. All six independent NF-L-transactivator lines produced identical staining patterns, indicating that lacZ expression reproducibly reflected the activity of the NF-L promoter (data not shown). With each of these transactivator lines, ␤-galactosidase activity was restricted to the cells of the developing brain and spinal cord. These results demonstrate that expression of the IE-LacZ transresponder transgene was specifically activated in the nervous system when an NFL-VP16 transactivator transgene was present. In addition, they reveal that the 1.7-kb NF-L promoter is active in embryogenesis and reproducibly directs reporter gene expression in the developing nervous system. The 1.7-kb Murine Neurofilament Light Chain Gene Promoter Is Active and Neuron-specific Early during Neurogenesis-To elucidate the onset and progression of NF-L promoter activity in the nervous system, we isolated embryos from crosses between NF-L-transactivators and IE-LacZ16 transresponders at different stages of development. Fig. 2 depicts representative embryos carrying both transgenes. Both whole-mount embryo stainings and histologic sections from paraffinembedded embryos were analyzed. The NF-L promoter first became active at 8.5 dpc when ␤-galactosidase activity was restricted to neuroepithelial cells ( Fig. 2A). No promoter activity was detected in embryos isolated at 6.5 or 7.5 dpc (not shown). At 9.5 dpc (Fig. 2B), staining was restricted to the developing nervous system. In the brain, staining was observed in the neuroepithelium of the hindbrain, the midbrain, and weakly in the forebrain (Fig. 2B). LacZ activity could also be detected in hindbrain rhombomeres as well as the trigeminal (V) and facial-acoustic (VII-VIII) complexes on histologic sections of the embryo (not shown). The neuroepithelium of the spinal cord exhibited ␤-galactosidase activity throughout its entire anterior-posterior axis (Figs. 2B and 3A). This staining pattern persisted at 10.5 dpc with increased activity in the midand forebrain (Fig. 2C). Both the trigeminal and facial-acoustic ganglia were stained, as were the primitive optic apparatus and the otocysts (not shown). The neural tube continued to stain as did the early condensations of presumptive neural crest cells. The latter cells, which give rise to dorsal root ganglia, could be seen adjacent to the neural tube at the level of the forelimb bud (not shown). Neural tube staining appeared stronger in more ventral regions (Fig. 3B). There was little difference in the staining pattern at 11.5 dpc in whole-mount embryos; ␤-galactosidase activity was observed in the midbrain and forebrain (not shown). The trigeminal ganglion and the retina were also positively stained. In the developing spinal cord, the number of stained cells decreased (compare Fig. 3, B and C) and remained mostly confined to cells of the ventricular layer. Cells in the dorsal root ganglia were also stained ( Fig.  3C). However, by 12.5 dpc, ␤-galactosidase activity within the nervous system was noticeably weaker (Fig. 2D). The walls of the hindbrain, midbrain, and forebrain continued to stain, and cells of the trigeminal ganglia, retina, and roof of the neopallial cortex were also positive (not shown). Within the spinal cord at the mid-trunk region (Fig. 2D), staining persisted in neuroepithelial cells in ventrally and, weakly, in dorsally located areas (Fig. 3D). Some cells in the dorsal root ganglia also exhibited ␤-galactosidase activity (Fig. 3D). Similar staining patterns were observed with at least three different, independently derived NFL-transactivator lines. These results demonstrate that the 1.7-kb NF-L promoter specifically directs transgene expression to neuroepithelial cells.
We further characterized the developmental activity of the NF-L promoter by examining ␤-galactosidase activity in embryos at later stages. At 14.5 dpc, neuron-specific ␤-galactosidase activity persisted in several developing sensory organs (  A, the VP16-based binary transgenic system principally consists of two independent strains of transgenic mice (25,29). One transgenic parent, the transactivator, carries a transgene encoding the herpes simplex virus transcriptional activator, VP16, under the control of a specific promoter (striped mouse). Here, the murine NF-L promoter is used to specifically express VP16 in neurons. The second transgenic parent, the transresponder, harbors the lacZ reporter gene downstream of the VP16 IE-response element (shaded mouse). The IE promoter is silent in the absence of VP16. Therefore, expression of the gene of interest will only occur in offspring that receive a copy of both transgenes (striped and shaded mouse). B, structure of DNA fragments used to generate transgenic mice. Several independent NFL-TSV transgenic mouse lines were generated. NF-L, 1.7-kb murine neurofilament light chain gene promoter; SV40 intron/poly(A), polyadenylation signal and splice acceptor site from SV40 virus (Pharmacia); IE, immediate early promoter from the ICP4 gene of herpes simplex virus. The cells of the developing semicircular canals in the auditory complex also exhibited ␤-galactosidase activity (Fig. 4C) as did the cells of the developing olfactory epithelia (Fig. 4, D and E). It is interesting to note that cells in the olfactory epithelia stained in a non-uniform manner with varying degrees of intensity (Fig. 4, D and E).
The examination of whole-mount stained embryos at later developmental time points revealed a general decline in neuron-specific activity from the 1.7-kb NF-L promoter (Fig. 5). By 13.5 dpc, neuron-specific ␤-galactosidase staining became restricted to the neopallial cortex, midbrain, retina, and cerebellar primordia (Fig. 5A). The spinal cord was only stained in the most posterior region, as were the dorsal root ganglia (not shown). The spinal cord and dorsal root ganglia in anterior regions were essentially negative (Fig. 5B). By 14.5 dpc, staining within the developing brain was limited to the cerebellar anlage, neopallial cortex, olfactory epithelia, and the developing auditory and optic apparati (Figs. 4 and 5C). Together, these data reveal that the 1.7-kb NF-L promoter is neuronspecific during early neurogenesis. The noticeable decrease in neuronal gene expression implies that while the 1.7-kb NF-L promoter contains regulatory elements for initiation of transcription early in development, other elements residing elsewhere may be required to maintain NF-L gene expression in neurons at later stages.
Muscle-specific Activity of the NF-L Promoter at Later Developmental Stages-Surprisingly, at later developmental stages, strong ␤-galactosidase staining was observed in non-neuronal cells. At 13.5 dpc, the premuscle masses of the fore and hind limbs as well as skeletal muscle precursors were labeled (Fig.  5A). Thus, from 13.5 dpc on, the NF-L-promoter was active in myogenic cells in addition to specific neuron subsets. One day later, at 14.5 dpc, intense staining for ␤-galactosidase was observed in the developing abdominal and limb musculature, intercostal muscles, and some facial muscles (Fig. 5C). The examination of histologic sections from these embryos confirmed the identity of the lacZ positive cells as myogenic cells (Fig. 5, D and E). Staining within the extrinsic ocular muscle is presented in Fig. 4B where both neuronal (lens and optic nerve) and myogenic cells exhibited ␤-galactosidase activity. In addition, cells of the developing intercostal muscles also displayed strong ␤-galactosidase activity (Fig. 5D). Using a second, independent NF-L-transactivator line, NFL-TSV207, the staining pattern was similar to that observed with the NF-TIF4 line (not shown). Fig. 5E shows ␤-galactosidase activity within the extensor muscles of the developing fore limb surrounding the cartilage primordia of the humerus and the ulna from the NFL-TSV207 line.
Embryos examined at 16.5 and 18.5 dpc exhibited no ␤- ␤-galactosidase activity is detected only within nervous tissue. As development proceeds, there is a general decrease in lacZ expression within the neural tube. By 12.5 dpc, ␤-galactosidase activity is restricted to only a few neuroepithelial cells in the spinal cord with stronger intensity in more ventral regions. Staining within the dorsal root ganglia also declined with only a small number of stained cells detectable by 12.5 dpc. galactosidase activity in either whole-mount specimens or in isolated brain or muscle tissue (not shown). Furthermore, no ␤-galactosidase activity could be detected in tissues (neuronal and non-neuronal) isolated from mice at postnatal day 10 and at 1 and 3 months old (not shown). Taken together, these data reveal the presence of regulatory elements within the 1.7-kb NF-L promoter that direct gene expression to myogenic cells at later developmental time points. However, similar to the transient activity observed in neurons, NF-L promoter activity in muscle tissue does not continue through late developmental stages or into adulthood. Thus, the NF-L promoter contains temporally regulated DNA elements that can initiate but do not maintain transgene expression in different cell types at various developmental stages.
Detection of Transgene Expression in Neonatal Tissue-The lack of ␤-galactosidase activity in either late embryos or in tissues isolated from neonatal mice contradicted an earlier report that the NF-TIF4 transactivator line could transactivate an IE-CAT transresponder transgene in the brains of neonatal mice (25). To resolve this discrepancy, we employed sensitive reverse transcriptase-PCR (RT-PCR) methodologies to analyze the expression of the NF-L-transactivator and the IE-LacZ transgenes in neonates at day 10. Mice with the following genotypes were analyzed: animals carrying both the NF-TIF4 and IE-LacZ16 transgenes or the NFL-TSV207 and IE-LacZ16 transgenes, and control animals harboring only the NF-TIF4 or the IE-LacZ16 transgene. As predicted, the expression of VP16 was only detected in animals that carried a transactivator transgene, either NF-TIF4 or NFL-TSV207 (Fig. 6A). VP16 could be detected in brain, spinal cord, and muscle tissue in the NF-TIF4 line with the addition of skin in the NFL-TSV207 line. Primers specific for the lacZ transgene detected lacZ expression only in mice who received a copy of both the IE-LacZ transgene and an NF-L-transactivator transgene (Fig. 6A). No lacZ expression could be detected in animals that contained the IE-LacZ16 transgene alone, confirming that the IE promoter is silent in the absence of VP16 (Fig. 6A). With two different NF-L-transactivators, lacZ transcripts were consistently present in the brains of double transgenic animals (Fig. 6A). In spinal cord samples, lacZ transcript was detected in 4 of 6 experiments, and no lacZ transcript was ever detected in postnatal muscle or skin tissue. Although lacZ expression was detected at the RNA level in the brains from both double transgenic lines that we examined, the level of protein expression was insufficient to produce detectable ␤-galactosidase activity. In summary, these results confirm the specificity of our binary transgenic mouse system in that transactivation of the lacZ gene occurred only in animals that harbored both a transactivator and transresponder transgene. In addition, lacZ expression coincided with VP16 expression in tissues where the NF-L promoter is active. The weak ␤-galactosidase activity in neonatal tissues may reflect the decrease in transcriptional activity of the NF-L promoter. The embryos were photographed with dark field optics. Histologic cross-sections at a thickness of 10 m were obtained from stained and paraffin-embedded embryos from crosses between either NF-TIF4 (D) or NFL-TSV207 males (E) and IE-LacZ16 females at 14.5 dpc. Staining is weak in the developing spinal cord, but ␤-galactosidase activity is present in the developing intercostal muscles (D) and the extensor muscles of the developing forelimb (E). R, cartilage primordium of the ribs; H, cartilage primordium of the humerus; U, cartilage primordium of the ulna.
Transcriptional Activity of the Endogenous NF-L Gene Locus-To compare the transcriptional activity of the transgenic NF-L promoter to that of the endogenous gene, we analyzed the expression of the endogenous NF-L gene locus in multiple tissues. Eight tissues from 10-day old neonatal mice were analyzed by RT-PCR using primers that flanked a region encompassing the first two introns of the NF-L gene (see "Experimental Procedures"). As expected, expression from the endogenous NF-L locus was readily detectable in the brain and spinal cord samples (Fig. 6B, lanes B and Sc). In addition, we consistently detected NF-L expression in skeletal muscle tissue (Fig. 6B, lane M). No signal was found for skin, liver, spleen, kidney, or heart (Fig. 6B). The detection of NF-L transcript in muscle tissue coincides with that observed for VP16 expression from our NFL-transactivator transgenes (Fig. 6, A and B). The detection of endogenous NF-L mRNA in muscle tissue supports our observation of NF-L promoter activity, as visualized by ␤-galactosidase activity and confirmed by RT-PCR analyses, in myogenic cells in our transgenic animals (Figs. 4 and 5). Together, these results show that the endogenous NF-L gene is transcriptionally active in skeletal muscle tissue. In addition, they demonstrate that the transcriptional activity of the endogenous NF-L gene locus is consistent with the activity of the transgenic 1.7-kb NF-L promoter.
Structure of the Murine NF-L Promoter-To identify putative DNA regulatory elements that control NF-L gene expression, we determined the primary structure of the NF-L promoter. We sequenced 1328 bp of the promoter, 1095 bp of which were novel. From bases Ϫ513 to Ϫ279, our sequence is identical to and confirms the previously published data (13). Data base searches revealed a strong similarity of the mouse NF-L promoter to the NF-L promoters from both the rat (34) and human (10).
The promoter contains several repeated elements (Fig. 7): repeats of TAAAGGTTTTA at Ϫ1271 and Ϫ708, both of which are conserved in the rat and mouse promoter; repeats of ATC-CTCCCCTT at Ϫ114 and ϩ17, both of which are conserved in the human, rat, and mouse sequence; and, an indirect repeat of AGTGGGGTGGGGTGA at Ϫ404, which is present in both rodent promoters. In addition, a dinucleotide repeat of (TC) 27 is present at Ϫ1031. Many putative DNA regulatory sites reside within the 1.7-kb promoter. The conservation of these potential regulatory sites among the mouse, rat, and human NF-L promoters is indicated in Fig. 7. Sequences proximal to the transcription initiation site and the TATA box, at base Ϫ31 (13), contain potential binding sites for general transcription factors such as AP2, Sp1, and ATF/CREB (35) (Fig. 7). In the distal region of the promoter, there are a number of putative binding sites for tissue-specific and developmentally regulated transcription factors together with additional putative sites for AP2 and Sp1. Similarities to regulatory sequences found in other neuron-specific genes include: an OL-1 binding site, which had been implicated in lens-specific gene expression (36); five PEA-3 sites (37); and a zif268/egr-1/krox24 site (35). Four potential binding sites for the embryonically expressed POUdomain transcription factor Pit-1 (38) are located within the distal promoter region. Of particular interest is the presence of putative binding sites for several myogenic and basic-helixloop-helix (bHLH) transcription factors including two MyoD/E box sites at Ϫ1186 and Ϫ928 (39), which are conserved in the human, rat, and mouse NF-L promoters. Also conserved in the rat and mouse promoters are an MLC1f/3f site at Ϫ515 (40), an MCBF site (with one mismatch) at Ϫ509 (41), an E box at Ϫ1283 (35), and a potential myocyte enhancer factor 2 (MEF2) binding site (42) overlapping the TATA box of the NF-L promoter at position Ϫ32 with 9 of 10 bases identical to the consensus (Fig. 7). The conservation of these sites within the three species suggests that they may play a role in the normal expression of the NF-L gene. The presence of putative myogenic transcription factor binding sites may explain the expression of the lacZ transgene in developing and neonatal myogenic FIG. 7. Sequence analysis of the murine NF-L promoter. Numbering begins with the transcription initiation site as ϩ1. Upper case letters refer to nucleotides sequenced in this study while lower case letters refer to the published partial sequence of the mouse NF-L promoter (13). Repeated elements and potential binding sites for various transcription factors are indicated. Sequences unique to the mouse NF-L promoter are underlined; those identical in mouse and rat NF-L promoters are boxed; and those identical in mouse, rat, and human NF-L are shaded.

tissue (Figs. 5 and 6B).
Deletions within the NF-L Promoter Identify Muscle-specific Regulatory Elements-To identify genetic regulatory elements that contribute to the observed muscle-specific expression pattern at 13.5 dpc, a series of deletions in the NF-L promoter were generated. The first deletion truncated the promoter at base Ϫ1215, eliminating the putative Sp1 binding site at Ϫ1490 and an E-box at Ϫ1283. Still, this construct was able to direct transgene expression in both neuronal and myogenic tissues with intense ␤-galactosidase activity in the brain and spinal cord as well as in facial, limb, and trunk musculature (Fig. 8A). Deleting further to base Ϫ941 removes potential AP-2, PEA-3, and OL-1 binding sites as well as the dinucleotide CT repeat at Ϫ1031 and the putative MyoD binding site at Ϫ1186. This fragment was capable of activating the lacZ transresponder gene in both neurons and muscles (Fig. 8B). However, deletion to base Ϫ524 removes a putative MyoD binding site at Ϫ928 and was unable to produce transgene expression in muscle tissue (Fig. 8C). Weak neuronal staining was still observed in the retina, cerebellar anlage, the most posterior aspects of the spinal cord (Fig. 8, C and E), and otic apparatus (not shown). Upon examination of histologic crosssections through these embryos, no ␤-galactosidase staining was observed in muscle tissue (Fig. 8F). These experiments indicate that a muscle-specific enhancer resides between Ϫ941 and Ϫ524 of the NF-L promoter and suggest that the putative MyoD binding site at Ϫ928 may confer the muscle-specific promoter activity. This site is conserved in both the rat and human NF-L promoters. In addition, the dramatic decrease in neuronal staining demonstrates that this region of the promoter (between bases Ϫ941 and Ϫ524) also contains elements important for its level of activity or efficiency in neurons. It is interesting to note that within this fragment, the 90-bp between bases Ϫ777 and Ϫ688 comprise a region that is 90% identical to the rat NF-L promoter (34) and 68% identical to the that of the human (10). Although the sequence does not contain any known transcription factor binding sites, its high degree of conservation is suggestive of a regulatory function. Deleting further to base Ϫ328, which removes all putative myogenic transcription factor binding sites, resulted in only weak neuronal staining where ␤-galactosidase activity was restricted to the retina (Fig. 8D) and olfactory epithelia (not shown). These deletion analyses identify distinct functional roles between the proximal and distal promoter regions. Although the proximal promoter was able to direct gene expression in certain neuronal subsets (Fig. 8, C and D), it clearly was unable to recapitulate the full pattern of neuronal activity when compared with either the full 1.7-kb promoter (Fig. 5A) or with deletion fragments containing more distal promoter elements (Fig. 8, A and B). The distal region of the promoter, between bases Ϫ941 and Ϫ524, thus contains regulatory elements necessary for both the specificity and intensity of neuronal activity. The muscle-specific activity of the promoter was lost when only proximal promoter fragments were used (Fig. 8, C and D). Critical muscle-specific regulatory elements therefore reside in the distal promoter region, between Ϫ941 and Ϫ524, or cooperate with more proximal downstream enhancers. These data identify relevant tissue-specific regulatory elements in the distal promoter region of the mouse NF-L promoter. Furthermore, this deletion analysis shows that the regulatory element conferring muscle-specific activity is physically separable from the elements responsible for neuronal activity. DISCUSSION With this study, we have characterized the activity of the 1.7-kb murine NF-L promoter during embryonic development using a unique binary transgenic mouse system. This system uses the potent viral transcriptional transactivator, VP16. The VP16 gene is expressed under the control of the 1.7-kb murine NF-L promoter and is used to transactivate a lacZ reporter gene, the transresponder transgene, which is linked to the VP16-responsive IE promoter (25,29). We generated several NFL-transactivator transgenic mouse lines to characterize the embryonic expression pattern of the mouse NF-L promoter by virtue of ␤-galactosidase activity resulting from transactivation of an IE-LacZ transresponder transgene. Embryos that inherited both a transactivator and transresponder transgene demonstrated specific lacZ expression within the developing nervous system. Staining patterns were faithfully reproduced with six independent NFL-transactivator mouse lines in addition to the previously established NF-TIF4 transactivator line.
The developmental activity of the promoter was characterized by staining for lacZ gene expression at different developmental time points. ␤-galactosidase activity was initially found to be specific for cells of the developing nervous system and could be detected at 8.5 dpc in neuroepithelial cells. The first appearance of NF-L protein has been detected by immunohistochemistry in the hindbrain between 9.0 -9.5 dpc and at 10.5 dpc in the spinal cord (2). The detection of lacZ expression before NF-L protein synthesis is consistent with the onset of promoter activity prior to or concomitant with the terminal cell FIG. 8. Deletions within the NF-L promoter identify a potential myogenic regulatory region. Double transgenic founder embryos were generated using DNA fragments harboring different deletions within the NF-L promoter and isolated at 13.5 dpc. Both neuronand muscle-specific promoter activity was detected with a deletion to base Ϫ1215 (A) or to base Ϫ941 (B). Weak staining was detected in neuronal subsets upon further deletion to base Ϫ524 (C) or to Ϫ328 (D). Histologic cross-sections through the embryo in panel C identify ␤galactosidase activity in neuronal cells including the lens and the retina (E); however, no ␤-galactosidase activity is detected in developing muscle tissue (F). S, scapula; Su, subscapularis muscle; T, trapezius muscle. division of neuronal precursors. Differentiated neurons, such as the ventrally located motor neurons, however, did not stain for ␤-galactosidase at later developmental stages, suggesting that the NF-L promoter does not maintain its activity in mature neurons. By 14.5 dpc, neuronal NF-L promoter activity was restricted to the developing optic, otic, and olfactory apparati with some weak persistent activity in the fore-and midbrain. Taken together, these data show that the 1.7-kb NF-L promoter initiates transcription in a neuron-specific fashion at early stages of neuronal differentiation.
Previous studies using the human NF-L promoter report that either a 307-bp or 2.3-kb fragment of the promoter could direct neuron-specific gene expression in transgenic embryos at 13.0 dpc (16,18). However, in those experiments, there was variability among independently derived transgenic embryos, suggesting that the promoter fragments used were relatively weak and could be influenced by the transgene integration sites. Independently derived transgenic founder embryos exhibited varying patterns of neuronal expression. The failure to identify a strong neuron-specific enhancer element in the promoter regions of the NF genes has led to the proposition that intragenic regions of the NF-L gene are necessary for accurate nervous system expression in transgenic mice (11,16,18). Our own preliminary results in transgenic embryos indicate the possible presence of additional neuron-specific regulatory elements in the second intron of the mouse NF-L gene. 2 Collectively, our data document a consistent pattern of NF-L promoter activity in transgenic mice. Multiple independent transactivator transgenic lines that could transactivate the IE-LacZ transgene gave a similar pattern of ␤-galactosidase activity. Thus, we conclude that the 1.7-kb promoter directs reporter gene expression restricted to the developing nervous system in a reproducible fashion. Several explanations may account for the differences to results reported in the literature. First, our incubation time for the ␤-galactosidase staining reaction was limited to 2 h. This would select only those transgenic lines that exhibit the strongest expression. Second, the murine NF-L promoter, although very similar to that from human and rat, might contain additional regulatory elements that could affect gene expression. Third, we employed 1.7 kb of promoter sequence while other studies either used much larger or minimal promoter sequences. The presence or absence of additional regulatory elements could alter the expression pattern from the promoter. Fourth, we employed a transcriptional transactivator, VP16, to activate the reporter gene in a binary transgenic mouse system. VP16 interacts with cellular transcription factors (43)(44)(45)(46)(47)(48), and it is therefore possible that VP16 potentiates reporter gene expression. The conclusion from our results is that the 1.7-kb upstream fragment of the mouse NF-L promoter is developmentally active specifically in the nervous system during early neurogenesis.
The developmental activity of the NF-L promoter was monitored through neonatal time points. Previously, the NF-TIF4 line was shown to be capable of transactivating an IE-CAT transresponder transgene in the brain of neonatal mice (25). However, using an IE-LacZ transresponder transgene, ␤-galactosidase activity could not be detected in nervous tissue at developmental stages later than 15.5 dpc or at three different postnatal time points. RT-PCR documented the presence of VP16 transcripts in 10-day old postnatal brain, spinal cord, and muscle tissues. lacZ transcripts were also consistently detected in the brain with weaker signals for the spinal cord of double transgenic neonates. These results suggest that the 1.7-kb NF-L promoter is weak at this time point or that only a few cells may be expressing the transgene; in either case, the LacZ protein expression was too low to be detected using the X-gal staining assay. In combination with the staining data from earlier embryonic stages, our results indicate that the 1.7-kb NF-L promoter contains elements for initiating but not maintaining gene expression within neurons.
Surprisingly, NF-L promoter activity was also detected in developing muscle tissue as visualized by strong ␤-galactosidase staining of 13.5-and 14.5-dpc embryos. These results demonstrate that the NF-L promoter is active in non-neuronal cell types in vivo, namely in developing muscle cells. Although the promoter was transcriptionally active in these cells, the expression of NF-L protein could not be detected by immunohistochemistry in non-neuronal tissues at 9.5, 12.5, 14.5, 16.5, and 18.5 dpc in the mouse (2). 2 This is consistent with data from cell transfection experiments that detected NF transcripts but not protein upon introduction of NF genes into non-neuronal cell types (7, 13, 19 -22). Our data suggest that post-transcriptional mechanisms regulate the tissue-specific expression of NF proteins; however, they also reveal that muscle-specific regulatory elements appear to be present within the NF genes. Nonneuronal expression from NF-promoters has been previously reported in vivo in transgenic mice. A 5-kb promoter fragment from the rat NF-L gene could direct expression of a CAT reporter gene in adult nervous and skeletal muscle tissue (14), and transgene-derived neurofilaments have been observed in skeletal muscle by electron microscopy (27,49). It is also interesting to note that, in rabbit, NF-M mRNA has been isolated from precursors of heart conduction myocytes. 3 We now show that elements regulating NF expression in muscle tissue reside within the 1.7-kb NF-L promoter.
The transcriptional activity of the endogenous NF-L gene locus was assayed by RT-PCR and was detected in brain, spinal cord, and skeletal muscle tissue from every experimental animal examined. Here we show that in vivo, the endogenous NF-L locus is transcriptionally active in non-neuronal cells, specifically in muscle tissue. NF-L transcripts were not detected in any other tissue studied. These results are consistent with our finding that the NF-L promoter is active in muscle cells during late embryogenesis. The transient presence of ␤galactosidase activity in myogenic cells at 14.5 dpc but not later stages of development suggests that, as in neurons, the 1.7-kb NF-L promoter is only capable of initiating but not maintaining tissue-specific gene expression. These data imply that molecular cues for non-neuronal NF gene expression exist within the 1.7-kb NF-L promoter and that post-transcriptional mechanisms may restrict the expression of NF proteins to neurons.
To identify potential regulatory elements involved in NF-L gene expression, we sequenced 1095 bp of novel sequence from the murine NF-L promoter. The overall sequence similarity was very high between the mouse and rat NF-L promoters, and many domains were highly conserved with the human NF-L promoter (Fig. 7). Sites with similarities to known transcription factor binding sites appear to be clustered into two domains: the proximal promoter region, from Ϫ328 to ϩ105; and the region distal to the transcription start site, from Ϫ1608 to Ϫ329. Notable in the proximal promoter is the presence of two putative PEA-3 binding sites (37), one zif268/egr-1/krox24 site (10,13,35) and three potential AP2 sites (35). These factors have recently been shown to bind, in combination, to the promoter of the neuron-specific human synapsin II gene (50). PEA-3 mRNA expression has been detected in the brain and epididymis in adult mice (51) and in subsets of spinal motor neurons in the chicken. 4 Thus, it is possible that these sites are involved in regulating NF-L gene expression in the nervous system. However, our own promoter deletion studies demonstrated that the proximal promoter was able to direct transgene expression only in certain neuronal subsets (Fig. 8D) and confirmed that additional, distal promoter regions are required for the neuronal activity of the NF-L promoter.
The promoter deletion analyses also confirm that the distal promoter region contains elements necessary for both the neuron-and/or muscle-specific activity of the promoter. Several putative binding sites for myogenic factors of the bHLH class are present in the distal region: two sites for MyoD binding (39), one site for MLC1f/3f (40), one site for MCBF (41), and an E box (35). Deletion of the promoter from Ϫ941 to base Ϫ524 or Ϫ328 resulted exclusively in neuronal staining (Fig. 8, B-D), implicating the putative MyoD binding site at Ϫ928 as the muscle-specific enhancer. The remaining neuronal staining was extremely weak and restricted to certain neuronal subsets. These data demonstrate that the promoter region between Ϫ941 and Ϫ328 contains distinct and independent elements for muscle-and neuron-specific gene expression. Furthermore, it appears that the level of enhancer activity in neurons requires the interaction of elements in the proximal, Ϫ328 to ϩ105, region of the promoter with those further upstream.
It is interesting to note that the temporal pattern of transgene expression in muscle in our transgenic embryos does not coincide with the onset of expression of any known myogenic bHLH genes (52). The Mef2 gene family is both spatially and temporally regulated in its expression during skeletal muscle development (53,54) and, as development progresses, Mef2 gene expression is expanded to include the brain (55). This progression from muscle to nervous tissue has been taken as an example of accumulating evidence for parallel transcriptional mechanisms in neuron-and muscle-specific gene expression. The 1.7-kb NF-L promoter constitutes another example of a regulatory element in muscle and neurons with a temporally regulated switch in tissue specificity. However, in this situation, expression shifts from neurons to muscle cells. Taken together, these data support the notion that temporally separable enhancers are present in these promoters that control specific gene expression in distinct tissues during embryonic development.
In summary, our study demonstrates that a region 1.7 kb upstream of the murine NF-L promoter reproducibly directs nervous system-specific gene expression during embryogenesis in transgenic mice. Characterization of the developmental activity of this promoter revealed intricate spatial and temporal regulation that accompanies a switch to muscle-specific gene expression during later developmental stages. Deletion analyses indicate that distinct elements within the distal promoter region regulate the neuron-and muscle-specific promoter activity. The transient nature of promoter activity in both neuronal and muscle cells suggests that initiation of transcription from the NF-L promoter is possibly linked to cell division or differentiation state and indicates that additional regulatory elements may be required to maintain NF-L transcription in terminally differentiated cells.