Address correspondence and reprint requests to Dr. P. A. Wight at Department of Physiology and Biophysics, Mail Slot 750, University of Arkansas for Medical Sciences, 4301 W. Markham St., Little Rock, AR 72205, U.S.A.
Abstract: Antisilencer or antirepressor elements have been described, thus far, for only a few eukaryotic genes and were identified by their ability not to augment gene expression per se but to override repression mediated via negative transcription regulatory elements. Here we report the first case of antisilencing for a neural-specific gene, the myelin proteolipid protein (PLP) gene (Plp). PLP is the most abundant protein found in CNS myelin. The protein is synthesized in oligodendrocytes, and its expression is regulated developmentally. Previously we have shown that a PLP-lacZ transgene (which includes the entire sequence for Plp intron 1) is regulated in mice, in a manner consistent with the spatial and temporal expression of the endogenous Plp gene. In the present report, we demonstrate by transfection analyses, using various PLP-lacZ deletion constructs, that Plp intron 1 DNA contains multiple elements that collectively regulate Plp gene expression in oligodendrocytes. One of these regulatory elements functions as an antisilencer element, which acts to override repression mediated by at least two negative regulatory elements located elsewhere within Plp intron 1 DNA. The mechanism for antisilencing appears to be complex as the intragenic region that mediates this function binds multiple nuclear factors specifically.
The combinatorial effect of positive and negative transcriptional regulators forms the foundation for cell type-and developmental stage-specific expression of many genes. The myelin proteolipid protein (PLP) gene (Plp) affords such a model system. The Plp gene, which encodes the most abundant protein present in CNS myelin, is expressed primarily in oligodendrocytes, although low levels of the mRNA have been detected in Schwann cells (Puckett et al., 1987; Kamholz et al., 1992), cardiac myocytes (Campagnoni et al., 1992), and fetal thymus and spleen (Pribyl et al., 1996). Besides cell type-specific expression, Plp gene expression is also regulated temporally (Kronquist et al., 1987; Gardinier and Macklin, 1988; LeVine et al., 1990; Schindler et al., 1990). In the developing brain, DM20 mRNA, an alternatively spliced product of PLP pre-mRNA, precedes the appearance of PLP-specific transcripts and is found in relatively low abundance (Ikenaka et al., 1992; Timsit et al., 1992). As development proceeds, PLP mRNA becomes evident, superceding the amount of DM20-specific transcripts, with peak quantities observed during the period of active myelination (LeVine et al., 1990).
To begin to address the parameters that are responsible for restricting Plp gene expression primarily to oligodendrocytes and those that dictate the varying capacity of its expression in oligodendrocytes during development, it is necessary to identify the regulatory elements that control cell type-specific and temporal expression. Previously, we have demonstrated that transgenic mice that carry a PLP-lacZ fusion gene, which includes the entire sequence for Plp intron 1 DNA, regulate the fusion gene in a manner consistent with the spatial and temporal patterns of endogenous Plp gene expression (Wight et al., 1993). Transfection of this fusion gene [PLP(+)Z] or a similar construct [PLP(-)Z], which is missing all 8,140 bp of mouse Plp intron 1 DNA, resulted in high levels of β-galactosidase (β-gal) activity only for the “intronless” construct when nonglial cells were utilized. However, transfection of either construct in the N20.1 oligodendroglial cell line (which transcribes the Plp gene) resulted in high levels of β-gal activity that were similar (Wight and Dobretsova, 1997). From those results, it appeared that Plp intron 1 DNA confers cell type-specific expression probably through a transcriptional repression mechanism in nonexpressing cell types, but that the intronic sequence does not influence Plp gene activity in oligodendrocytes. Yet transfection results presented here, utilizing related PLP-lacZ constructs that contain partial deletion of Plp intron 1 sequences, demonstrate that the intron contains multiple elements that collectively modulate Plp gene expression in oligodendrocytes. One of these regulatory elements functions as an antisilencer element. Antisilencer or antirepressor elements have been described, thus far, for only a few eukaryotic genes (Stover and Zehner, 1992; Frenkel et al., 1994; Goping and Shores, 1994; Jahroudi and Lynch, 1994; Dirks et al., 1996; Lopez et al., 1997) and were identified by their ability to override repression mediated via negative transcriptional regulatory elements. We have also identified other regions of Plp intron 1 DNA that contain negative regulatory elements, whose activities are overcome by way of antisilencing in the N20.1 cells. Our present findings indicate that Plp gene expression in oligodendrocytes is regulated by a complicated interplay of antisilencer and negative regulatory elements located within the first intron.
MATERIALS AND METHODS
Dr. Anthony Campagnoni (University of California at Los Angeles) generously provided the mouse N20.1 cell line (Verity et al., 1993), which was derived by immortalization of primary oligodendrocytes with a temperature-sensitive form of SV40 large T antigen. N20.1 cells were grown in Ham's F-12/Dulbecco's modified Eagle's low-glucose medium (1:1; Irvine Scientific, Santa Ana, CA, U.S.A.) supplemented with 5.75 g/L glucose, 3.75 g/L HEPES, 1.2 g/L sodium bicarbonate, 100 μg/ml G-418, and 10% fetal bovine serum (Intergen, Purchase, NY, U.S.A.). Cultures were maintained at 34°C in an atmosphere of 5% CO2. The mouse liver cell line +/+ Li (ATCC, Rockville, MD, U.S.A.) was grown in Dulbecco's modified Eagle's medium (Life Technologies, Gaithersburg, MD, U.S.A.) supplemented with 10% fetal bovine serum (Intergen) and maintained at 37°C in an atmosphere of 10% CO2.
Construction of plasmids PLP(+)Z (Wight et al., 1993) and PLP(-)Z (Wight and Dobretsova, 1997) has been described previously. Both plasmids contain mouse Plp DNA, which includes the proximal 2.4 kb of 5′ flanking DNA, all of exon 1, and the first 37 bp of exon 2, fused in-frame to the lacZ reporter gene. In addition, PLP(+)Z also contains the complete sequence for Plp intron 1 DNA encompassing 8,140 bp.
PLP(+)Z was used to derive constructs that contain internal deletion of Plp intron 1 sequences. The resulting constructs were named according to their deletions based upon numbering the first intron from positions 1 through 8,140 (Wight and Dobretsova, 1997). PLPΔ809-5807 was generated by complete digestion of PLP(+)Z with PstI and subsequent ligation of the distal ends using T4 DNA ligase. Similarly, PLPΔ809-4058 and PLPΔ4065-5807 were generated by partial PstI digestion and PLPΔ1488-4300 by partial digestion with XmnI. PLPΔ795-1891 was produced by digestion of PLP(+)Z with AvrII (partial digestion) and NheI followed by ligation of the resulting compatible ends. PLPΔ1079-2236 was created by digestion of PLP(+)Z with SexAI, fill-in of the recessed 3′ ends with Klenow enzyme, followed by ligation of the ensuing blunt ends. Analogously, PLPΔ809-7299 was produced by digestion with BstBI and contains a PstI site at the deletion junction through the addition of a linker (5′-GCTGCAGC-3′).
PLPΔ1265-1889 was produced by a two-step process. Initially, plasmid KK1.38 (Macklin et al., 1991), which contains Plp 5′ intron 1 sequence, was cleaved with BsgI, treated with T4 DNA polymerase to remove 3′ overhangs, and ligated to an NheI synthetic linker (5′-GGCTAGCC-3′). Thereafter, the SanDI-NheI fragment, which contained Plp 5′ intron 1 DNA, was isolated from the resulting plasmid and substituted for the corresponding sequence in PLP(+)Z.
PLPΔ1265-1889 was used to generate a nested set of deletions by cleavage with NheI, followed by Nuclease Bal-31 treatment, fill-in of staggered ends with Klenow enzyme, subsequent ligation to XhoI linkers (5′-CCTCGAGG-3′), XhoI digestion, and ligation of the ensuing ends. The 5′ deletion endpoints at positions 1,068, 1,115, 1,141, 1,161, 1,183, and 1,265 were determined by DNA sequencing using the methods of Sanger and co-workers (1977). All of these constructs have 3′ deletion endpoints at position 2,243.
PLPΔ809-8068 was generated by BstBI digestion of PLP(+)Z, followed by Nuclease Bal-31 treatment, ligation to PstI linkers, digestion with PstI, and subsequent ligation of the resulting ends. PLPΔ12-7515 was generated in an analogous fashion using SanDI-linearized PLPΔ809-7299 DNA. PLPΔ324-5807 was generated by digestion of PLP(+)Z with SanDI, fill-in of the recessed 3′ ends with Klenow enzyme, ligation with PstI linkers, followed by cleavage with PstI and subsequent ligation of the resulting ends. PLPΔ324-8068 is a hybrid construct between PLPΔ324-5807 and PLPΔ809-8068 that was generated by digestion of the parental constructs with PstI and NotI and subsequent exchange of PstI-NotI fragments. PLPΔ12-8068 was generated in a similar fashion using PLPΔ12-7515 and PLPΔ809-8068.
N20.1 cells were seeded at a density of 2.5-2.8 × 105 cells per 35-mm well (six-well dishes; Costar, Cambridge, MA, U.S.A.) 20-24 h prior to transfection in 2 ml of growth medium. Likewise, +/+ Li cells were seeded at a slightly lower density (2.0-2.3 × 105 cells). The cells within a given well were transfected with 2 μg of total plasmid DNA and 5 μl of LipofectAMINE (Life Technologies) according to the manufacturer's specifications. Cells were transfected with either vector DNA (pBluescript SK+; Stratagene, La Jolla, CA, U.S.A.) alone or with a given PLP-lacZ construct (plus vector DNA to hold the molar concentration of the test plasmid constant) and 0.35 μg of Rous sarcoma virus (RSV)-luciferase (Chang and Brenner, 1988) to control for differences in transfection efficiency. Each construct was transfected in duplicate per experiment. Cells were incubated with the DNA/LipofectAMINE mixture for 6 h. Lysates were prepared approximately 48 (+/+ Li) to 72 (N20.1) h post DNA addition with 190 μl of Reporter Lysis Buffer (Promega, Madison, WI, U.S.A.) as recommended by the supplier. Aliquots of lysate were evaluated for reporter gene expression using the Galacto-Light Plus Kit (Tropix, Bedford, MA, U.S.A.) and the Luciferase Assay System (Promega), respectively, for determination of β-gal and luciferase activities, as previously described (Wight et al., 1997). Luminescence was measured as relative light units (RLU) using a Monolight 2010 luminometer (Analytical Luminescence Laboratory, San Diego, CA, U.S.A.). β-Gal RLU obtained from PLP-lacZ transfectants ranged from 103 to 105 powers per 10 μl of lysate tested. Likewise, RLU obtained for the luciferase assays typically exhibited magnitudes of 105-106. Initially, the background luminescence (as determined from lysates prepared from vector-only transfected cells) was subtracted from those obtained with the various PLP-lacZ transfected cells. Adjusting the RLU obtained for β-gal from a given lysate to a constant amount of luciferase activity negated differences in transfection efficiency. Values obtained from PLP(-)Z transfected cells were set arbitrarily at 100% for each experiment. The values obtained for the other PLP-lacZ transfectants from a given experiment were adjusted accordingly. Results are presented as the means ± SD of the relative β-gal activity for three or more experiments.
Electrophoretic mobility shift assay (EMSA)
A Sau3AI fragment containing Plp intron 1 DNA (Wight and Dobretsova, 1997) from positions 1,083 to 1,203 was inserted into the BamHI site of pBluescript SK(+). The resulting subclone was cleaved with BsmAI and SpeI and the fragment, which contained Plp intron 1 DNA from positions 1,101 to 1,203, isolated. The purified DNA (“antisilencer” probe) was 32P-labeled with Klenow enzyme by fill-in reaction.
Nuclear extracts were prepared from N20.1 cells grown to ∼80% confluence in 162-cm2 flasks by the methods of Dignam et al. (1983). Multiple preparations of nuclear extracts were made and tested in the EMSA experiments. Initially, binding reactions were assembled at 0°C in a total volume of 20 μl and contained 2 × 104 cpm of labeled probe, 1 μg of protein (nuclear extract), and 1 μg of poly(dI-dC) (Sigma, St. Louis, MO, U.S.A.) in a buffer containing 20 mmol/L potassium phosphate (pH 7.9), 50 mmol/L KCl, 1 mmol/L EDTA, and 4% glycerol. Some reactions also contained unlabeled competitor DNA, with two different double-stranded oligonucleotides to the β2-adrenergic receptor promoter serving as heterologous competitors. The DNA sequences for the sense strands of the heterologous oligonucleotides were 5′-CCCGGGCCAAC-CCCAGGAGGAGGCGGG-3′ (βAR-5′) and 5′-CGGGCTAG-GCAAGGAGGGTTGGCCCTT-3′ (βAR-3′). Subsequently, the binding reactions were incubated at room temperature for 20 min and then fractionated on a 4% nondenaturing polyacrylamide gel, and protein-DNA complexes were visualized by autoradiography.
Total RNA was isolated from N20.1 cells transiently transfected with an equimolar amount of either PLPΔ12-7615 or PLPΔ809-7299, as well as from untransfected cells, using TriZOL reagent (Life Technologies) as recommended by the supplier. The resulting RNA was treated with 1 U of RQ 1 RNase-free DNase (Promega) per microgram of RNA at 37°C for 1 h. RT was performed at 42°C for 50 min with 7.5 μg of RNA, 75 ng of oligo(dT)12-18 (Life Technologies), and 1 U of SuperScript II RNase H- RT (Life Technologies) in a 20-μl reaction volume according to the methods of Nathan and colleagues (1995). Subsequently, RT was inactivated by incubation at 70°C for 15 min, followed by treatment with 1.2 U of RNase H (Life Technologies) at 37°C for 20 min. Optimization of PCR conditions was achieved by the following methods. Initially, PCR was performed with 1.0 μl of the RT reaction using the eLONGase Amplification System (Life Technologies) in a total volume of 50 μl. Amplifications (30 s at 90°C, 5 s at 59°C, and 30 s at 68°C) were performed using Plp sense (5′-GATCCTTCCAGCTGAGCAA-3′) and lacZ antisense (5′-CCTCTTCGCTATTACGCCA-3′) primers for 26, 28, 30, and 32 cycles. A β-actin primer pair was also included in the same reaction, sense (5′-TCTAGGCACCAAGGTGTGA-3′) and antisense (5′-CAGGCAGCTCATAGCTCTT-3′). Subsequently, 0.125, 0.25, 0.5, 1, and 2 μl of the RT reaction were amplified for 28 cycles to determine the linear range. Finally, 0.5 μl of the RT reaction was amplified for 28, 30, 32, and 34 cycles. PCR reactions with an equivalent amount of RNA not treated with RT were amplified for 34 cycles under similar conditions. Twenty-microliter aliquots from the PCR reactions were fractionated on a 1.2% agarose gel and the products visualized by ethidium bromide staining.
Identification of region within intron 1 that “positively” modulates Plp gene expression in oligodendrocytes
As illustrated in Fig. 1A, Plp intron 1 DNA constitutes a substantial portion of the gene. To test the influence of intron 1 sequences on Plp gene expression, deletion constructs were derived from PLP(+)Z (Fig. 1A) that contain internal deletion of Plp intron 1 sequences. These deletion constructs were named by numbering the first intron from positions 1 through 8,140; e.g., PLPΔ809-5807 is missing Plp intron 1 DNA from positions 809 through 5,807. Transfection of a PLP nonexpressing liver cell line (+/+ Li) with either PLP(+)Z, several related deletion constructs, or PLP(-)Z (which does not contain any Plp intron 1 DNA) did not demonstrate significant levels of β-gal activity except in the PLP(-)Z transfected +/+ Li cells (Fig. 1B). These results suggest that negative regulatory element(s) still remain within the Plp intron 1 DNA present in these partial deletion constructs, which actively function to repress Plp gene expression in the +/+ Li cells.
Unlike in the liver cells, transfection of N20.1 oligodendroglial cells with either PLP(+)Z or PLP(-)Z demonstrated high levels of β-gal activity that were comparable (Fig. 1B). Solely considering these results, it appears that Plp intron 1 DNA does not contain any regulatory elements that govern Plp gene expression in oligodendrocytes. However, transfection of N20.1 cells with intron 1 partial deletion constructs suggests otherwise. Deletion of two PstI fragments from Plp intron 1 (PLPΔ809-5807) resulted in a dramatic decrease in β-gal activity compared with that obtained from PLP(-)Z transfected cells (7 and 100%, respectively; Fig. 1B). Deletion of sequences within the 5′PstI fragment was responsible for the reduction of β-gal activity as PLPΔ809-4058 (8%), but not PLPΔ4065-5807 (95%), demonstrated diminished levels compared with PLP(-)Z transfected cells (Fig. 1B).
Additional deletion constructs were generated to further refine the location of the “positive” regulatory element(s) that were deleted in PLPΔ809-4058. As shown in Fig. 2A, constructs with deletion of Plp intron 1 sequences from positions 795 to 1,891 or from positions 1,079 to 2,236 demonstrated reduced β-gal activities in transfected N20.1 cells (4 and 7%, respectively). However, both PLPΔ1265-1889 (105%) and PLPΔ1488-4300 (93%) transfected cells showed similar levels as those obtained from PLP(-)Z transfectants (100%). Taken together, these data suggest that a “positive” regulatory element resides between Plp intron 1 positions 1,079 and 1,265.
A nested set of deletion constructs was generated to map the 3′ boundary of the “positive” regulatory element located in Plp intron 1. As shown in Fig. 2B, β-gal activities obtained from PLPΔ1265-2243 transfected N20.1 cells were similar to those obtained from PLP(-)Z transfectants. Further deletion of 5′ sequences to position 1,183 (PLPΔ1183-2243) actually augmented β-gal activity (189%) compared with PLP(-)Z transfected cells (100%). Additional deletion of 5′ sequences to positions 1,161 (PLPΔ1161-2243), 1,141 (PLPΔ1141-2243), 1,115 (PLPΔ1115-2243), and 1,068 (PLPΔ1068-2243) dramatically decreased the relative β-gal activities to 27, 23, 6, and 4%, respectively. Thus, it appears that sequences between intron 1 positions 1,115 and 1,183 are critical for Plp gene expression in oligodendrocytes.
Specific binding of nuclear factors to intronic region that “positively” modulates Plp gene expression
EMSAs were performed to see if factors present in N20.1 nuclei actually bind to the “positive” regulatory region identified by transfection analyses. Initially, binding to a double-stranded DNA fragment encompassing Plp intron 1 positions 1,101-1,203 was examined (Fig. 3). Multiple protein-DNA complexes were detected using this ∼100-bp probe (Fig. 3, lane 2). To investigate the specificity of the protein-DNA complexes, competition analysis was performed. The slower-migrating complexes were competed for by the addition of unlabeled homologous DNA (Fig. 3, lanes 3-5). However, addition of unlabeled heterologous (β2-adrenergic receptor) DNA did not result in competition (Fig. 3, lanes 6-9). Thus, N20.1 cells contain nuclear proteins that specifically bind to Plp intron 1 DNA between positions 1,101 and 1,203.
To further define the regions of protein binding, seven overlapping double-stranded oligonucleotides (27 bp each) were generated that collectively span Plp intron 1 DNA from positions 1,083 through 1,211 (Table 1). These oligonucleotides were used as unlabeled competitors for protein binding to the ∼100-bp probe (positions 1,101-1,203). As shown in Fig. 4 (lane 3), even a mixture of all seven oligonucleotides, at 200-fold molar excess per each oligonucleotide, was unable to compete with the fastest-migrating complex, suggesting that the binding is nonspecific. When the oligonucleotides were tested individually as unlabeled competitors (200-fold molar excess), only oligonucleotides I, III, and V (Oligos I, III, and V) demonstrated significant competition with the ∼100-bp probe. Oligo I and Oligo V specifically competed with the two slowest-migrating complexes (Fig. 4, lanes 4 and 9, respectively). In addition, Oligo III also competed with the two slowest-migrating complexes, although to a lesser extent. Moreover, Oligo III specifically competed with several faster-migrating complexes (Fig. 4, lane 6). Therefore, two sites (between positions 1,117-1,143 and 1,151-1,177) that bind nuclear factors were found within the “positive” regulatory region between Plp intron 1 positions 1,115 and 1,183.
Table 1. Oligonucleotides used as unlabeled competitors in EMSA
The “positive” regulatory element located between Plp intron 1 positions 1,115 and 1,183 is no longer necessary for high β-gal activity with PLP-lacZ constructs that lack most of first intron
To understand why Plp intron 1 DNA from positions 1,115-1,183 was required for significant levels of β-gal expression in N20.1 cells transfected with the partial deletion constructs tested thus far, but not for PLP(-)Z (which lacks the first intron), additional PLP-lacZ constructs were analyzed (Fig. 5). These constructs contain very large deletion of Plp intron 1 sequences and are missing the “positive” element, mapped to positions 1,115-1,183 in the intron. Remarkably, N20.1 cells transfected with either PLPΔ12-7615 or PLPΔ12-8068 resulted in high levels of relative β-gal activity (117 and 128%, respectively). These data suggest that additional regulatory elements (which function in a negative fashion) might have been deleted from these constructs. In contrast, transfection of N20.1 cells with either PLPΔ324-5807 or PLPΔ324-8068 led to widely disparate levels of β-gal activity (10 and 171%, respectively). Taken together, these results suggest that a negative regulatory element resides between Plp intron 1 positions 5,807 and 8,068. As either PLPΔ12-7615 or PLPΔ12-8068 transfected cells showed high levels of β-gal activity, the location of this negative regulatory element can be refined to positions 5,807-7,615. The 5′ region of the intron, between positions 12 and 324, does not appear to harbor a negative regulatory element, as both PLPΔ12-8068 and PLPΔ324-8068 transfected cells demonstrated high levels of reporter gene expression. However, N20.1 cells transfected with the less deleted construct, PLPΔ809-7299, demonstrated low β-gal activity (10%), which was similar to that obtained from PLPΔ809-5807 transfected cells (Fig. 1B). Further deletion of 3′ sequences to position 8,068 (PLPΔ809- 8068) resulted in an increase of β-gal activity to 39% (Fig. 5), which suggests that although a 3′ negative regulatory element may have been removed from PLPΔ809-8068, another negative regulatory element (located between intron 1 positions 324 and 809) still remains within this construct. On the whole, these data imply that there are at least two negative regulatory elements present in Plp intron 1 DNA. One of these elements is located in the 5′ region between positions 324 and 809, whereas the other element appears to reside in the 3′ portion of the intron, between positions 7,299 and 7,615.
Thus, the “positive” regulatory element located between Plp intron 1 positions 1,115 and 1,183 may act as an antisilencer element, which is required to overcome repression mediated by negative elements located elsewhere within the intron. These negative regulatory elements would have been deleted in PLP-lacZ constructs that lack most or all of the first intron, and therefore the presence of the antisilencer element is mute. However, an alternative possibility is that the “positive” regulatory element might function as a splicing element that is no longer required for accurate splicing of pre-mRNAs produced by the extremely deleted constructs. To test this possibility, the integrity of PLP-lacZ mRNAs was examined from transfected N20.1 cells.
Low levels of β-gal activity are not due to splicing aberrations of PLP-lacZ pre-mRNAs
As the translation start site is encoded in Plp exon 1, the expression of β-gal is dependent upon accurate splicing of PLP-lacZ pre-mRNAs. To see if the regulatory region located between Plp intron 1 positions 1,115 and 1,183 functions as an element necessary for accurate splicing of the relatively large first intron, PLP-lacZ RNA was analyzed from PLPΔ12-7615 and PLPΔ809-7299 transfected N20.1 cells (Fig. 6). Both constructs gave rise to RT-PCR products of the predicted size based upon accurate RNA splicing, albeit to varying amounts. These products were the same size as that obtained from PLP(+)Z transfectants (data not shown). No larger products were evident, as would be expected from unspliced transcripts or from mRNAs generated utilizing internal cryptic splice sites. The RT-PCR products for PLP-lacZ were at least 10-fold more abundant in PLPΔ12-7615 than PLPΔ809-7299 transfectants, relative to those obtained for endogenous β-actin, which were used to control for differences in RT-PCR efficiencies (Fig. 6). These results are consistent with the transfection results (Fig. 5), whereby PLPΔ12-7615 transfectants showed higher levels of β-gal activity when compared with those obtained with PLPΔ809-7299. Results from other transfected constructs (data not shown) showed a similar pattern, whereby constructs that resulted in high β-gal activities also resulted in higher amounts of the PLP-lacZ RT-PCR product. Thus, the “positive” regulatory element appears to affect Plp gene expression at a stage prior to RNA splicing (presumably at transcription), although it is remotely possible that the discrepancy in RT-PCR product amounts could simply reflect the differences due to variable transfection efficiencies.
The emerging role of regulatory elements located within Plp intron 1 DNA that control Plp gene expression in oligodendrocytes is summarized in Fig. 7. This model depicts several regions within the intron that contain either antisilencer or negative regulatory elements. The presence of the antisilencer element, which at least partially if not wholly resides between Plp intron 1 positions 1,115 and 1,183, is essential to overcome repression mediated by negative regulatory elements located within the 5′ (positions 324-809) and 3′ (positions 7,299-7,615) regions of the intron. The function of the antisilencer element is no longer necessary when these negative regulatory elements have been deleted, as in constructs PLPΔ12-7615, PLPΔ12-8068, and PLPΔ324-8068.
In oligodendrocytes, the antisilencer appears to influence gene expression at a step prior to RNA splicing (presumably transcription), as (a) RT-PCR analysis did not reveal any deviation in the length of the product from that expected, (b) EMSA analysis showed that nuclear factors isolated from N20.1 cells specifically bound to double-stranded DNA encompassing the antisilencer region, and (c) deletion constructs missing the antisilencer element showed intermediate levels of β-gal activity when only the 5′ negative regulatory region was present (e.g., PLPΔ809-8068) compared with similar constructs that contained both the 5′ and 3′ negative regulatory regions (e.g., PLPΔ809-7299), suggesting that repression mediated through the negative regulatory elements may be cumulative.
Thus far, antisilencer elements have been mapped in a relatively small subset of genes, including the vimentin (Stover and Zehner, 1992), osteocalcin (Frenkel et al., 1994), carbamyl phosphate synthetase I (Goping and Shores, 1994), von Willebrand factor (Jahroudi and Lynch, 1994), βB2-crystallin (Dirks et al., 1996), and interferon-α (Lopez et al., 1997) genes. Many of these antisilencer elements are located in the promoter region and hence must modulate transcription rather than post-transcriptional processes. The antisilencer region described in this report constitutes the first case of a neural-specific gene containing this type of regulatory element and does not show any apparent similarity to the antisilencer elements from the genes cited above.
The mechanism for Plp gene antisilencing is likely to be complex, because multiple specific protein-DNA complexes were formed by EMSA analyses. Three of the oligonucleotides (Oligo I, Oligo III, and Oligo V) tested as unlabeled competitors resulted in specific competition with the labeled probe (positions 1,101-1,203). Interestingly, Oligo I (positions 1,083-1,109) and Oligo V (positions 1,151-1,177) showed similar patterns of competition, suggesting that the proteins that bind these DNA fragments may be similar. Comparison of the DNA sequences for these oligonucleotides revealed a region of 75% identity over an 8-bp span: positions 1,091-1,098 and positions 1,156-1,163 in the opposite orientation. Whether these regions of similarity are important in protein binding remains to be determined. Alternatively, sequences from the distal 3′ end of Oligo I might be responsible for part of the specific complexes, as the labeled probe used in EMSA covered Plp intron 1 positions 1,101-1,203. However, this possibility seems unlikely, because Oligo II, which overlaps this region of Oligo I, did not demonstrate competition in the EMSA analysis. As mentioned earlier, Oligo III (positions 1,117-1,143) also resulted in specific competition with the same probe; however, two faster-migrating complexes were primarily affected. Oligo IV also slightly competed with these complexes, suggesting that the overlapping regions between Oligo III and Oligo IV may be important for protein binding.
A computer search (Quandt et al., 1995) for potential transcription factor binding sites revealed sequences for TFIID-Inr (Smale et al., 1990) and AP-4 (Mermod et al., 1988) in Oligo I and AP-1 (Lee et al., 1987; Cousin et al., 1991), GCN4 (Arndt and Fink, 1986), and Oct-1 (Groenen et al., 1992) in Oligo III. Both Oligo III and Oligo IV contain consensus binding sites for the nuclear zinc finger protein Gfi-1 (Zweidler-McKay et al., 1996), with the core binding sequence located at the terminal 5′ end of Oligo IV. Perhaps this factor is responsible for the high-mobility doublet detected by gel shift analysis with Oligo III and to a lesser extent with Oligo IV. Oligo V contained potential DNA binding sites for SOX-5 (Denny et al., 1992) and SRY (Harley et al., 1994) as well as similarity to a stress response element found in yeast (Schueller et al., 1994). Interestingly, both Oligo I and Oligo V contained sequence similarity to the target site for RFX1 (Emery et al., 1996). Coupled with the strikingly similar EMSA results obtained with these oligonucleotides as competitors, RFX1 emerges as a promising candidate for binding to the antisilencer region. Whether any of these homologies are relevant is currently under investigation. Alternatively, the binding sites may interact with novel factors.
Although we have mapped the antisilencing element between Plp intron 1 positions 1,115 and 1,183, it is probable that multiple elements are required for antisilencing. N20.1 cells transfected with deletion constructs PLPΔ1161-2243 or PLPΔ1141-2243 resulted in modest (∼25%) β-gal activities, whereas PLPΔ1115-2243 and PLPΔ1068-2243 transfectants showed only minimal (∼5%) activities compared with that obtained from PLP(-)Z transfected cells. These data suggest that multiple elements are required for complete antisilencing. Furthermore, it is possible that other elements within Plp intron 1 DNA, besides those located between positions 1,115 and 1,183, may be required for antisilencing. For example, an element located 5′ to position 1,115 (for instance, sequences contained within Oligo I) may be necessary for antisilencing but not sufficient to promote derepression in the absence of other elements present in Oligo III and Oligo V. Experiments to rescue antisilencing by the addition of Plp intron 1 sequences into PLP-lacZ deletion constructs, which originally showed very low levels of β-gal activity, are currently in progress. These experiments will be important in further delimiting the antisilencer element(s).
In this report we have presented evidence that the regulation of myelin Plp gene expression in oligodendrocytes is tightly controlled by the interplay of silencing and antisilencing mechanisms. As N20.1 cells transfected with either PLP(+)Z or PLP(-)Z demonstrated comparable levels of β-gal activity, it is plausible (although unlikely) that the positive and negative elements located within Plp intron 1 DNA mediate activities that serve to counterbalance each other. However, this prospect appears improbable because other studies (Wight and Dobretsova, 1998) showed that six SstI fragments, which collectively encompass the entire intron, all failed to either augment or diminish Plp promoter activity. Thus, in oligodendrocytes, the antisilencer element acts to override repression mediated through negative regulatory elements located elsewhere within the intron, presumably at the level of gene transcription. The prominent question remains: Why is Plp gene expression regulated in such a complicated manner? We postulate that the combinatorial effects of the antisilencing and silencing mechanism may govern the precise stage-specific regulation of PLP expression. For instance, factors that bind to the antisilencer element may be temporally regulated or functionally modified during the period of active myelination and consequently are able to overcome repression mediated through the negative elements present in the first intron. Experiments are currently in progress to test this possibility.
It is known that mutations in the Plp gene result in leukodystrophies with variable severity in humans. For example, Plp gene mutations have been found in people with Pelizaeus-Merzbacher disease (PMD). The mutations include complete deletion of the gene (Raskind et al., 1991), gene duplication (Cremers et al., 1987), as well as point mutations in the coding region (Hodes et al., 1993). Interestingly, Gow and Lazzarini (1996) have shown that protein trafficking of Plp gene products with missense mutations behaves differently in transfected COS-7 cells and may be the basis for the severity of PMD. However, at least one-half of all persons diagnosed with PMD do not show any defect in the Plp coding sequence. Consequently, other explanations besides entrapment of misfolded protein in the endoplasmic reticulum must be the cause of disease in these people. Based upon the current studies, we predict that an ineffective or compromised antisilencing mechanism in oligodendrocytes would lead to underexpression of the Plp gene and hence result in a dysmyelinating disorder. Thus, critical mutations within the antisilencer element(s) or in genes that encode the antisilencer factor(s) may be the cellular basis for PMD in some patients that do not show mutations in their Plp coding region.
Whether or not antisilencing is possible in the +/+ Li cells remains to be determined. Preliminary results suggest that the liver cells also contain nuclear factors capable of binding to the antisilencer region, implying that the antisilencer factors may not be restricted solely to oligodendrocytes. Additional preliminary data suggest that other negative regulatory elements are functional in the +/+ Li cells, which are located 3′ of position 7,615. Thus, it appears that independent or supplementary negative regulatory elements located within Plp intron 1 DNA function to repress Plp gene expression in the nonglial cells. Whether cell type-specific regulation is due to an inability to mount an antisilencing response in PLP-nonexpressing cells or a dominant effect mediated through cell type-specific negative regulatory elements or a combination thereof is currently under investigation.