Analysis of a Gene Encoding Two Glycine Transporter Variants Reveals Alternative Promoter Usage and a Novel Gene Structure*

The rat GLYT-1 gene encodes two glycine transporter variants, GLYT-1a and GLYT-1b, that differ in NH2 termini and 5′-noncoding regions as well as in tissue distribution. The GLYT-1 gene contains 15 exons, with the first two specific for GLYT-1a and the third specific for GLYT-1b. By combining RNase protection and rapid amplification of cDNA ends analysis, we have determined transcription start sites for GLYT-1a and GLYT-1b. By using a functional luciferase reporter assay, we demonstrate that distinct promoters regulate the expression of these transporters in several cell lines. Serially truncated GLYT-1b promoter constructs reveal a basal promoter within 304 base pairs of the transcription start site, possible negative regulatory elements between −304 and −1310, and additional positive regulatory elements between −1310 and −5264. The GLYT-1 gene contains three sets of dinucleotide repeats, two AC repeats, and one TG repeat which may form stem-loop structures to either facilitate or interfere with transcription of one of the transporter isoforms. The potential use of dinucleotide repeats in this manner would represent a novel mechanism for gene splicing. The use of distinct promoters for GLYT-1a and GLYT-1b suggests that these transporters have unique regulatory requirements that may reflect the differential tissue-specific expression patterns in white matter (GLYT-1b) and gray matter (GLYT-1a).

Glycine is the major inhibitory neurotransmitter in the brainstem and spinal cord and acts as a co-agonist at the ionotropic glutamate NMDA 1 receptors throughout the central nervous system. A role for NMDA receptors, and therefore, glycine, has been implicated in numerous disease states including glutamate neurotoxicity associated with ischemia (1, 2) and schizophrenia (3,4). Several studies have suggested that partial glycine agonists may be effective in limiting damage due to glutamate excitotoxicity (5). The primary means of regulating extracellular levels of glycine is by re-uptake through high affinity glycine transporters. Recent evidence demonstrates that glycine transporter activity can modulate NMDA receptor activity when both are co-expressed in Xenopus oocytes (6) and that, in the presence of low levels of glycine, peptides such as dynorphin can regulate NMDA receptors (7). In addition to removing glycine from the extracellular space, Ca 2ϩ -independent glycine release may occur through reversal of the transporter (8,9). In light of these data, regulation of glycine transporter activity may represent an important therapeutic avenue by which to alter extracellular glycine levels.
Neurotransmitter transporters are membrane-bound proteins that actively transport released neurotransmitters back into presynaptic neurons and surrounding glia, thereby terminating the activity of monoamine and amino acid neurotransmitters, and helping replenish presynaptic pools of neurotransmitters. On the basis of sequence homology, ion dependence, and predicted topology, transporters have been classified into several families (10,11). Included in the largest of these, the Na ϩ /Cl Ϫ -dependent family, are transporters for the monoamine neurotransmitters dopamine, norepinephrine, and serotonin, as well as amino acid neurotransmitters ␥-aminobutyric acid, glycine, proline, taurine, and betaine.
We have previously demonstrated by both in situ hybridization histochemistry and Northern analysis that GLYT-1a and GLYT-1b have distinct distributions in the rat. Whereas GLYT-1a mRNA is found in gray matter in the central nervous system as well as a number of peripheral tissues, GLYT-1b mRNA was detected only in white matter in the central nervous system (12). By using RNase protection assays, mRNA expression of these two isoforms is temporally distinct in the developing mouse embryo (16). Thus, these transporters appear to be alternatively spliced or alternatively promoted from * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) U28948 for the 5Ј-flanking region and the first exon of GLYT-1 and U28975 for introns [1][2][3][4]  a single gene, in a tissue-specific and developmentally specific manner. Based on the determination of multiple transcription start sites for both mouse GLYT-1a and GLYT-1b, it was suggested, but not demonstrated, that alternative promoter usage rather than alternative splicing mediates the production of these two isoforms (16). Although tissue culture models suggest similar pharmacologic profiles for these transporters, the limited number of pharmacologic tools for glycine transporters makes it difficult to definitively address this issue (13,14). Whereas GLYT1 proteins have been identified by immunohistochemistry in the rat brain (20,21), available antibodies do not distinguish between the two isoforms, GLYT-1a and GLYT-1b. The differences in the NH 2 termini of these transporters may reflect differences in the ability of these transporters to interact with different cellular proteins and/or to target to different subcellular compartments. In addition, glycine transporters found in gray matter (GLYT-1a) and glycine transporters found in white matter (GLYT-1b) may have different physiologic roles with distinct regulatory requirements.
In the present study, we have determined transcription start sites for the rat GLYT-1a and GLYT-1b transporters by combining RNase protection assays with rapid amplification of cDNA ends (RACE) analysis. Furthermore, by using functional reporter assays, we have identified distinct promoters that regulate the expression of GLYT-1a and GLYT-1b. Analysis of the genomic structure of the GLYT-1 gene suggests a novel mechanism for gene splicing.

EXPERIMENTAL PROCEDURES
DNA Cloning and Sequence Analysis-All probes and primers described in the text are illustrated in Fig. 1. Two overlapping clones from a Wistar rat genomic cosmid library (Stratagene, La Jolla, CA) that hybridized to a probe common to GLYT-1a and GLYT-1b were isolated (probe 1). One clone (RC-GLYT-1-1) hybridized to a probe specific for GLYT-1a (probe 3) and a probe specific for GLYT-1b (probe 4) but not to a common 3Ј probe (probe 2). A second clone (RC-GLYT-1-2) hybridized only to the common 3Ј probe. Probes were labeled with [␣-32 P]dCTP using a random priming kit (Boehringer Mannheim). Southern analysis was performed as described previously (12). The nucleotide sequence was determined from overlapping fragments subcloned into pBluescript II (Stratagene, La Jolla, CA) by the dideoxynucleotide method using Sequenase polymerase (U. S. Biochemical Corp.), or directly from the cosmid clones using SequiTherm Cycle Sequencing polymerase (Epicentre, Madison, WI). Nucleotide sequences were assembled using the GCG Sequence Analysis Software Package (22).
5Ј-Rapid Amplification of cDNA Ends (5Ј-RACE)-The 5Ј ends of GLYT-1 were amplified using a 5Ј-AmpliFINDER RACE kit (CLON-TECH, Palo Alto, CA). Poly (A) ϩ RNA (2 g) from rat cerebellum or rat brainstem and a primer common to GLYT-1a and GLYT-1b (primer A or primer B) were annealed at 65°C for 5 min. First strand cDNA was synthesized with avian myeloblastosis virus reverse transcriptase at 52°C for 30 min. The RNA was then hydrolyzed with NaOH at 65°C for 30 min, and excess primer and nucleotides were removed using a glass matrix support (GENO-BIND, NEN Life Science Products). A singlestranded AmpliFINDER anchor (CLONTECH, Palo Alto, CA) was ligated to the 3Ј end of the cDNA overnight at room temperature using T4 RNA ligase. The anchored cDNA was amplified with Taq DNA polymerase (Boehringer Mannheim) using a primer complementary to the anchor (AmpliFINDER anchor primer) and either a GLYT-1a-specific primer (primer C), a GLYT-1b-specific primer (primer D), or a primer common to GLYT-1a and GLYT-1b (primer E). All PCR components except for primers were heated to 82°C. Primers were then added followed by 35 cycles of amplification (94°C for 45 s, 55°C for 45 s, and 72°C for 2 min) and a final 7-min extension at 72°C. PCR products were blunt-ended with T4 polynucleotide kinase (New England Biolabs, Beverly, MA), and subcloned into pCR-Script SK(ϩ) (Stratagene, La Jolla, CA). Clones were screened with 32 P-end-labeled oligonucleotide probes specific for GLYT-1a (probe 5) or GLYT-1b (probe 6). Positive clones were sequenced to determine transcription start sites.
RNase Protection Assay-To make complementary RNA (cRNA) probes, GLYT-1a-and GLYT-1b-specific genomic fragments were PCRamplified (using primers F and G for GLYT-1a, and primers D and H for GLYT-1b) and subcloned into pBluescript SK(ϩ). The GLYT-1a-specific fragment was from Ϫ181 to Ϫ579, and the GLYT-1b-specific fragment was from ϩ44 to Ϫ236, relative to their respective translation start sites (Fig. 2). To make the antisense cRNA probes for GLYT-1a (probe 7) and GLYT-1b (probe 8), plasmids (1 g) were linearized with NotI and transcribed in vitro with T7 RNA polymerase and [␣-32 P]CTP, using a MAXIscript T7/T3 kit (Ambion, Austin, TX). Sense probes were transcribed with T3 RNA polymerase from EcoRI-digested GLYT-1a or Xho-digested GLYT-1b plasmids. Labeled probes were purified on 5% acrylamide, 8.4 M urea denaturing gels and eluted overnight at 37°C. RNase protection assays were carried out using the RPA II kit (Ambion, Austin, TX). Sense and antisense cRNA probes (1 ϫ 10 6 cpm) were hybridized overnight at 55°C with either 8 g of poly (A) ϩ RNA from total rat brain or 10 g of yeast RNA and then digested with RNase A (0.5 units) and RNase T1 (20 units) for 30 min at 37°C. Following ethanol precipitation, protected fragments were separated on a 6% acrylamide, 8.4 M urea denaturing gel. The sizes of protected fragments were determined by comparison to an adjacent dideoxynucleotide sequencing reaction.
Luciferase Reporter Constructs-Luciferase (LUC) reporter plasmids were constructed by subcloning portions of the GLYT-1 gene into the polylinker site upstream from the firefly luciferase cDNA in pGL2-Basic (Promega, Madison, WI). pGLYT-1a-LUC contained a 2.9-kb NotI-SacII fragment that included the 5Ј-most region of the isolated cosmid and part of the first exon (Ϫ32683 to ϩ210 relative to the GLYT-1a transcription start). pGLYT-1b-LUC contained a 5.4-kb BamHI-EarI fragment assembled from a 4.9-kb BamHI fragment and adjacent 0.5-kb BamHI-EarI fragment (Ϫ5264 to ϩ130 relative to GLYT-1b transcription start). pGLYT-1b-LUC was digested with NheI, SacI, or KpnI and re-ligated to produce the following truncated constructs: pGLYT-1b-1310-LUC (Ϫ1310 to ϩ130), pGLYT-1b-785-LUC (Ϫ785 to ϩ130), and pGLYT-1b-304-LUC (Ϫ304 to ϩ130), respectively. Constructs were ver- Transfection and Luciferase Assay-The rat neuroblastoma cell B104, the rat glioma C6 glioma, and the rat oligodendrocyte type 2 astrocyte progenitor cell CG4 were plated at 2 ϫ 10 5 to 5 ϫ 10 5 cells per well in 6-well tissue culture plates 1 day before transfection. Two days after plating at low density, CG4 cells were differentiated into oligodendrocytes or type 2 astrocytes (23) and allowed to differentiate for 6 days prior to transfection. Reporter constructs (2 g of DNA in 100 l of DMEM) and LipofectAMINE reagent (Life Technologies, Inc.; 6 g in 100 l of DMEM) were mixed, incubated at room temperature for 45 min, and then added to cells along with 800 l of DMEM. After 4 -5 h, 1 ml of DMEM with 20% fetal bovine serum was added to each well, and cells were fed with complete growth medium 20 h later. Seventy-two hours after transfection, cells were washed twice with phosphate-buffered saline and lysed in 200 l of lysis buffer (Promega, Madison, WI). Following a brief centrifugation to remove cell debris, triplicate 10-l samples were used to determine luciferase activity (Berthold Lumat LB 9501). An additional 10-l sample was reserved for protein determination (BCA kit, Pierce). All transfections were performed in triplicate 2-4 times. Transfection efficiencies for each cell line were determined by ␤-galactosidase staining following transfection with pCMV ␤-galactosidase. Briefly, cells were fixed for 5 min at 4°C with a solution of 2% formaldehyde and 0.2% glutaraldehyde and then incubated overnight at 37°C with 5-bromo-4-chloro-3-indolyl-␤-galactopyranoside (1 mg/ml) in 5 mM potassium ferricyanide, 5 mM potassium ferrocyanide, 2 mM MgCl 2 . Transfection efficiencies were 40-50% for all cells except oligodendrocyte-differentiated CG4 cells, which had less than 5% transfection efficiency.

RESULTS
Intron-Exon Structure and 5Ј-Flanking Region of the Rat GLYT-1 Gene-There are 15 exons in the rat GLYT-1 gene (Fig.   2). Fig. 2A illustrates the gene structure of GLYT-1 and the mRNA structure of GLYT-1a and GLYT-1b. The intron-exon consensus splice sites and exon sizes are also shown (Fig. 2B). The first two exons are specific for GLYT-1a; the third exon is specific for GLYT-1b, and the remaining 12 exons are common to both. Exon 1, which contains only noncoding sequences, was not detected in previously published cDNAs. RC-GLYT-1-1 contained 2.9 kb upstream (5Ј) from the first exon. Translation start sites for GLYT-1a and GLYT-1b are found in exons 2 and 3, respectively.
Determination of Transcription Initiation Sites-Transcription initiation sites can be analyzed either by RNase protection analysis or by 5Ј-RACE. Prediction of transcription start sites based on analysis of RACE products alone can be misleading because of the potential for premature reverse transcription stops. Similarly, RNase protection analysis can be misleading because it does not reveal information about exon size or number. We integrated these two approaches to more reliably predict the transcription start sites for GLYT-1a and GLYT-1b.
5Ј-RACE analysis using a GLYT-1a-specific primer (primer C) revealed a GLYT-1a-specific exon approximately 8 kb upstream from the second GLYT-1a-specific exon. Of the 25 RACE products that hybridized to a GLYT-1a-specific probe (probe 5), 10 extended into the first exon. As indicated by the inner arrows on the left panel of Fig. 3, 6 of these predicted start sites clustered over a 10-base range, from 238 to 247 bases upstream from the translation start of GLYT-1a. Three prod- ucts predicted start sites 20 -52 bases downstream from this region. One RACE product predicted a transcription start site 358 bases upstream from the translation start (Fig. 3, left). Of the 24 RACE products that hybridized to a GLYT-1b-specific probe (probe 6), 11 were clustered over a 10-base region 98 -107 bases upstream from the GLYT-1b translation start site (Fig. 3, right, inner arrows). Eight products were clustered over a 14base region 65-78 bases upstream from the GLYT-1b translation start (Fig. 3, right). Four additional RACE products were 44, 51, 122, and 140 bases upstream from the GLYT-1b translation start site.
RNase protection analysis on rat brain poly (A ϩ ) RNA using a cRNA probe containing part of exon 1 and its 5Ј-flanking region (probe 7) identified multiple transcription start sites for GLYT-1a over a 100-base region, with a major protected band in exon 1, 338 bases upstream from the GLYT-1a translation start site (Fig. 3, right, outer arrows). Several smaller, fainter protected bands suggest multiple start sites. The clustering of RACE products around the smallest protected band is most likely due to the failure of reverse transcriptase to complete transcription. None of these bands were protected when this probe was hybridized with yeast RNA or when the sense probe was hybridized with rat brain poly (A) ϩ RNA (data not shown). RNase protection analysis of rat brain poly (A ϩ ) RNA using probe 8, which contains part of exon 3 and its 5Ј-flanking region, predicted multiple transcription start sites for GLYT-1b (Fig. 3, right). This probe protected several bands over a 60base range. The major protected band predicted a start site 107 bases upstream from the GLYT-1b translation start site and coincided with the major clustering of RACE products. Protected bands 76 and 68 bases from the translation start site coincided with other RACE products. None of these bands were protected when this probe was hybridized with yeast RNA or when the sense probe was hybridized with rat brain poly(A) ϩ RNA (data not shown).
Features of Sequence of the 5Ј-Flanking Regions of GLYT-1a and GLYT-1b-The sequence of the first 2 exons of GLYT-1a as well as part of the 5Ј-flanking region of GLYT-1a is shown in Fig. 4A. The sequence is numbered with first potential transcription start site as 1. There are no apparent TATA or CCAAT boxes near the transcription start sites. Instead, as is typical for TATA-less promoters, there is a clustering of consensus sites for SP1 and one overlapping liver suppressor factor site located 13-28 bases upstream of the first predicted start site. In addition, there is a consensus site for AP-1 58 bases upstream from the first start site, and two consensus sites for a transcriptional initiator protein (24) 81 and 136 bases upstream from the first start site. Further 5Ј to this region are 3 sites for SP1, and single sites for C/EBP, AP-3 and GRE. Also 5Ј to the transcription start site is a 46-base pair dinucleotide repeat (see below). Fig. 4B shows the first exon of GLYT-1b (exon 3) and 1 kb of its 5Ј-flanking region. The first potential transcription start site is numbered 1. As was the case for GLYT-1a, no clear TATA box is found near the transcription start sites. Unlike GLYT-1a, there is no clustering of SP1 sites near the start site. Instead, there is a 54-base dinucleotide repeat 160 bases upstream from the first predicted start site (see below). Immediately upstream from this repeat are consensus sites for Z-DNA (25) and a consensus site for a transcriptional initiator protein.
There is also a CRE binding site within the multiple start site region. Within the 6-kb intron between exon 2 and exon 3 there are numerous consensus sites for various transcription factors, some of which are illustrated in Fig. 7. Fig. 4C illustrates the 5Ј end of the GLYT-1 gene and highlights the promoter regions, the first potential transcription start sites, and the translation start sites for GLYT-1a and GLYT-1b.
Functional Analysis of the GLYT-1a and GLYT-1b Promoters-Although it has been demonstrated that GLYT-1a and GLYT-1b are encoded by a single gene, it remains to be determined whether they are alternatively spliced from a single promoter or the result of alternative promoter usage. To investigate this, we measured the promoter activities of the 5Јflanking regions of GLYT-1a and GLYT-1b by using a transient transfection system with luciferase as a reporter gene. The constructs used in these studies consisted of fragments of genomic DNA from the 5Ј-flanking regions of GLYT-1a or GLYT-1b inserted upstream of the gene encoding firefly luciferase in the vector pGL2-Basic (Promega) (Fig. 5, bottom).
As shown in Fig. 5 (left bars), a 2.9-kb fragment from the 5Ј-flanking region of GLYT-1a, pGLYT-1a-LUC, had very strong promoter activity in all cells tested. A 5.4-kb fragment from the 5Ј-flanking region of GLYT-1b, pGLYT-1b-LUC, had significant promoter activity in differentiated CG4 cells but not in CG4 progenitors (Fig. 5, second bars from left). We recently determined by both RT-PCR and in situ hybridization that GLYT-1a is expressed in all three states of CG4 cells, whereas GLYT-1b is expressed only very weakly in the progenitor cells but more strongly in the differentiated cells. 2 Thus, the 5Јflanking regions of both GLYT-1a and GLYT-1b have promoter activity in cells expressing these transporters. These results demonstrate that GLYT-1a and GLYT-1b arise by alternative promoter usage rather than alternative splicing from a common promoter.
We also tested the promoter activities of the 5Ј-flanking 2 B. Borowsky and B. J. Hoffman, unpublished observations.

FIG. 3. Identification of transcription start sites for GLYT-1a and GLYT-1b.
RNase protection analysis was performed on 8 g of poly(A) ϩ RNA from total rat brain using 32 P-labeled cRNA probes for GLYT-1a (left, probe 7) and GLYT-1b (right, probe 8). Outer arrows indicate protected bands with the number being relative to the translation start for each. The inner arrows indicate where individual RACE products predict the start sites for GLYT-1a and GLYT-1b to be. regions of GLYT-1a and GLYT-1b in the rat glioma cell line C6 glioma and the rat neuroblastoma cell line B104 (Fig. 5, top). We have demonstrated by RT-PCR that both of these cell lines express GLYT-1a but not GLYT-1b. 2 In both cell lines, the 5Ј-flanking region of GLYT-1a had very strong promoter activity. Surprisingly, the 5Ј-flanking region of GLYT-1b had weak, yet significant, promoter activity in these cells. One possible explanation for this is that these cell lines express low levels of GLYT-1b that were not detected by RT-PCR. Alternatively, the 5Ј-flanking region of GLYT-1b may have a basal promoter activity that would normally be inhibited in cells not expressing the transporter.
As a first attempt at exploring this possibility, we generated serially truncated reporter constructs containing only the 3Јmost sequence of the GLYT-1b construct as follows: pGLYT-1b-1310-LUC, pGLYT-1b-785-LUC, and pGLYT-1b-304-LUC. In CG4 oligodendrocytes, CG4 astrocytes, C6 glioma cells, or B104 cells, the 304-base pair flanking region and the full 5264-base pair flanking region had the strongest promoter activities (Fig.  6). In contrast, the 785-and 1310-base pair flanking regions had significantly weaker promoter activities in these cells. These findings suggest that a functional promoter exists within the 434 bases closest to exon 3 and that negative regulatory elements exist within the 304 bases upstream from this region and perhaps also in the adjacent 525 bases further upstream. The recovery of promoter activity in the full-length construct could be accounted for by additional positive regulatory elements in the more 3Ј sequences or by a disinhibition of the negative regulation.
In all cell lines tested, the promoter activity of the GLYT-1b construct was weaker than that of the GLYT-1a construct. It should be noted, however, that the promoter activities of both of these constructs were quite high. In C6 glioma cells, for example, the activities of the GLYT-1a and GLYT-1b promoters were 343,184 and 81,603 light units/g of protein, respectively, whereas the SV40 control promoter was 239 light units/g of protein (Fig. 5). DISCUSSION In the present study, we have isolated and characterized the gene encoding two rat glycine transporter variants, GLYT-1a and GLYT-1b. We have used two strategies to determine the transcription start sites for these transporters and employed a functional reporter assay to determine regions of the gene important in regulating transcription. Our results demonstrate, for the first time, that transcription of these transporter variants is mediated by alternative promoter usage rather than alternative splicing off of a single promoter. In addition, our findings suggest several novel mechanisms that may potentially be important in regulating the alternative expression of these two transporters.
Structure and Analysis of the GLYT-1 Gene-The gene encoding the neurotransmitter transporters GLYT-1a and GLYT-1b consists of 15 exons. The first two exons are specific for GLYT-1a, the third specific for GLYT-1b, and the remaining exons are common to both transporters. The genomic organization of the rat GLYT-1 gene is similar to those of both the g-aminobutyric acid transporter GAT1 (26) and the mouse Glyt-1 gene (16). The major apparent difference between the genomic organization of the rat GLYT-1 gene reported here and that of the mouse homologue reported previously (16) is that we have found no evidence for a GLYT-1c-specific exon. Kim and colleagues (27) reported the isolation of a human GLYT-1c isoform consisting of the third exon, normally found in GLYT-1b, and an additional exon found between the third exon and the first common exon. Although we have found a region in the rat GLYT-1 gene that is homologous to this exon, we have not seen evidence that it is retained in mature mRNA. None of the RACE products that we obtained, using primers from the common region, contained this exon. Furthermore, Northern analysis, using an oligonucleotide probe complementary to this region, failed to detect any signal from rat brain or peripheral tissues (data not shown). It should be noted that, whereas Adams et al. (16) include a GLYT-1c exon in their description of the GLYT-1 gene, they too have failed to detect any RACE clones containing this exon. Therefore, it appears that the only evidence for a GLYT-1c isoform is in human substantia nigra RNA (27).
Combining RNase protection with RACE analysis, we demonstrate multiple transcription start sites for both GLYT-1a and GLYT-1b. The heterogeneity of transcription starts may be explained by the absence of TATA and CCAAT boxes for these genes. While potential start sites have been reported for the mouse GLYT-1 transporter, these were based only on RACE analysis (16). Because this type of analysis relies on reverse transcriptase, there is always the potential that premature transcription termination will lead to identification of false start sites. In the present study, we found a clustering of RACE products for GLYT-1a that were about 100 bases downstream from the major protected band, clearly illustrating the neces-C6 glioma, B104, undifferentiated CG4 cells, and CG4 cells differentiated into type 2 astrocytes or oligodendrocytes with constructs containing portions of the GLYT-1 gene upstream from of the gene encoding firefly luciferase. The response to constructs containing a 2.9-kb fragment from the 5Ј-flanking region of GLYT-1a, a 5.4-kb fragment from the 5Ј-flanking region of GLYT-1b, an SV40 promoter, and the luciferase gene alone are compared (left to right). Numbers on the constructs represent nucleotide location relative to the first transcription start sites for GLYT-1a or GLYT-1b.
FIG . 5. Functional analysis of the GLYT-1a and GLYT-1b promoters. Promoter activities were measured by transiently transfecting sity for using multiple approaches.
Promoter Activities of GLYT-1a and GLYT-1b-GLYT-1a is localized in gray matter within the central nervous system, as well as in macrophages in a number of peripheral tissues (12). In contrast, GLYT-1b is found only in fiber tracts in the central nervous system (28). These differences make it likely that GLYT-1a and GLYT-1b have distinct regulatory requirements. Based on the sequences of GLYT-1a and GLYT-1b, we had previously suggested that the expression of these transporters was the result of either alternative splicing from a single promoter or alternative promoter usage (12). Several lines of evidence suggest the latter. First, if alternative splicing alone mediates the expression of GLYT-1a and GLYT-1b, then there would have to be a common first exon to which the third exon could splice to generate GLYT-1b. In addition, the presence of multiple start sites for exon 3 suggests that it is a site of transcription initiation rather than simply an internal exon. Finally, evidence for alternative promoter usage comes from the functional reporter assays described in this study. We have demonstrated that the 5Ј-flanking regions of both GLYT-1a and GLYT-1b have strong promoter activity in transfected cells. Interestingly, although GLYT-1a and GLYT-1b have strong promoters, the levels of expression of these mRNAs in tissues are low, perhaps due to the presence of an endogenous inhibitor or to the instability of the mRNAs.
GLYT-1a Promoter Activity-By transiently transfecting cells with a portion of the 5Ј-flanking region of GLYT-1a linked to a luciferase reporter gene, we have demonstrated very strong promoter activity in the region from ϩ210 to Ϫ2683 (relative to the first transcription start for GLYT-1a). Within this region, the only canonical sites for a TATA box are 878 and 1103 bases upstream from the first predicted transcription start site. There are a clustering of SP1 sites and an overlapping liver suppressor factor site (29) 13-28 bases upstream from the first start site. Such a clustering is typical for TATAless promoters (30). There are also two consensus sites for a transcriptional initiator protein ((C/T)(C/T)A(T/A)(C/T)(C/T)) located 81 and 136 bases upstream from the start sites. Whereas these have been shown to be sufficient for determining start sites in TATA-less promoters, the first A is usually located between Ϫ5 and ϩ5 of the start site (24).
GLYT-1b Promoter Activity-Transient transfection of a luciferase reporter gene containing serially truncated portions of exon 3 and the upstream intron revealed basal promoter activity in a 434-base pair region including 304 bases upstream and 130 bases downstream from the first potential transcription start site. This finding indicates that GLYT-1b is regulated by a separate promoter from GLYT-1a. There is neither a clear TATA box nor a clustering of SP1 sites in the 5Ј-flanking region of GLYT-1b. However, the results of the present reporter studies have helped to define the regions of 5Ј sequence that may be important in regulating the transcription of GLYT-1b. Fig. 7 shows the consensus sites for representative transcription factors and other potentially interesting structures located 5Ј to exon 3 and also illustrates the boundaries of the reporter constructs used in this study. The only recognizable sites within pGLYT-1b-304-LUC are for a transcription initiation protein, a Z-DNA, and a 54-base dinucleotide repeat. These sites are located 159 -232 bases upstream from the first predicted start site, with the first two being adjacent to one another and the dinucleotide repeat beginning 5 bases downstream. The strong promoter activity seen with this construct suggests that these three elements or a portion of them may be sufficient for basal promoter activation of GLYT-1b. It is possible that secondary structures formed by the Z-DNA, which can cause negative supercoiling (25), and the TG repeat could affect transcription. Of course, it remains possible that other yet to be described transcription factors are also involved. Our results further suggest that there are negative regulatory elements within pGLYT-1b-785-LUC and pGLYT-1b-1310-LUC and either additional positive regulatory elements or disinhibiting factors within pGLYT-1b-LUC.
One of the most interesting features of the GLYT-1 gene is the presence of several sets of dinucleotide repeats. As shown in Fig. 8A, there are 23 pairs of ACs 5Ј of exon 1, 27 pairs of TGs 5Ј of exon 3, and 2 sets of AC repeats, 17 and 21 pairs, 3Ј of exon 3. Several possible scenarios in which secondary structures formed by these repeats might enable or disable the transcription of GLYT-1a or GLYT-1b are illustrated in Fig. 8. Although a definitive role for these repeats has yet to be demonstrated, it is tempting to speculate that they may result in secondary structures that could help to determine which isoform might be transcribed. While a role for these repeats in splicing or promotor choice has not been demonstrated, a role for secondary structures has been demonstrated. Secondary structure can influence alternative splicing as has been suggested for ␣-tropomyosin (31-33) and may preclude the recognition of one exon over another (34). For example, a stem loop formed between the first AC repeat and the TG repeat could ensure the exclusion of exons 1 and 2 from GLYT-1b (Fig. 8B). Similarly, a stem-loop formation between the TG repeat and one (Fig. 8C) or both (Fig.  8D) of the 3Ј AC repeats could help ensure the exclusion of exon 3 from GLYT-1a. In addition to bringing exons to be spliced closer together physically, the secondary structure formed by these stem loops might also result in transcriptional interference of one or the other transporter. Although the exact roles of these dinucleotide repeats remain to be demonstrated, those proposed here represent a novel and intriguing mechanism for gene splicing.
The present demonstration that GLYT-1a and GLYT-1b are transcribed from the same gene by different promoters suggests that it is critical for cells to be able to differentially regulate the levels of these transporters. The most likely rationale for this requirement is the ability to modulate the levels of these transporters in different regions or at different times during development. In support of this, we have previously demonstrated that GLYT-1a and GLYT-1b have quite distinct distributions in the rat, with the former present in gray matter in the central nervous system as well as several peripheral tissues, and the latter detected only in white matter in the central nervous system (12). The limited number of pharmacologic tools available to assess glycine transporters makes it difficult to compare these transporters functionally, but it is likely that the demands on and requirements for a transporter in white matter are unique from those of a transporter in gray matter. We have recently demonstrated that progenitor oligodendrocyte type 2 astrocyte cells express high levels of GLYT-1a but very low levels of GLYT-1b, whereas mature oligodendrocytes derived from these cells express high levels of both transporter variants. 2 Other potential differences between these transporters include targeting to different subcellular compartments and differential post-translational regulation by a casein kinase II phosphorylation site found only in GLYT-1b.