The Drosophila U1 and U6 gene proximal sequence elements act as important determinants of the RNA polymerase specificity of small nuclear RNA gene promoters in vitro and in vivo.

Transcription of genes coding for metazoan spliceosomal snRNAs by RNA polymerase II (U1, U2, U4, U5) or RNA polymerase III (U6) is dependent upon a unique, positionally conserved regulatory element referred to as the proximal sequence element (PSE). Previous studies in the organism Drosophila melanogaster indicated that as few as three nucleotide differences in the sequences of the U1 and U6 PSEs can play a decisive role in recruiting the different RNA polymerases to transcribe the U1 and U6 snRNA genes in vitro. Those studies utilized constructs that contained only the minimal promoter elements of the U1 and U6 genes in an artificial context. To overcome the limitations of those earlier studies, we have now performed experiments that demonstrate that the Drosophila U1 and U6 PSEs have functionally distinct properties even in the environment of the natural U1 and U6 gene 5'-flanking DNAs. Moreover, assays in cells and in transgenic flies indicate that expression of genes from promoters that contain the "incorrect" PSE is suppressed in vivo. The Drosophila U6 PSE is incapable of recruiting RNA polymerase II to initiate transcription from the U1 promoter region, and the U1 PSE is unable to recruit RNA polymerase III to transcribe the U6 gene.


From the Department of Chemistry and Molecular Biology Institute, San Diego State University, San Diego, California 92182-1030
Transcription of genes coding for metazoan spliceosomal snRNAs by RNA polymerase II (U1, U2, U4, U5) or RNA polymerase III (U6) is dependent upon a unique, positionally conserved regulatory element referred to as the proximal sequence element (PSE). Previous studies in the organism Drosophila melanogaster indicated that as few as three nucleotide differences in the sequences of the U1 and U6 PSEs can play a decisive role in recruiting the different RNA polymerases to transcribe the U1 and U6 snRNA genes in vitro. Those studies utilized constructs that contained only the minimal promoter elements of the U1 and U6 genes in an artificial context. To overcome the limitations of those earlier studies, we have now performed experiments that demonstrate that the Drosophila U1 and U6 PSEs have functionally distinct properties even in the environment of the natural U1 and U6 gene 5-flanking DNAs. Moreover, assays in cells and in transgenic flies indicate that expression of genes from promoters that contain the "incorrect" PSE is suppressed in vivo. The Drosophila U6 PSE is incapable of recruiting RNA polymerase II to initiate transcription from the U1 promoter region, and the U1 PSE is unable to recruit RNA polymerase III to transcribe the U6 gene.
Genes coding for most of the small nuclear RNAs (snRNAs) 1 are transcribed by RNA polymerase II, but U6 genes are transcribed by RNA polymerase III. In all cases, however, the promoters of these genes have features that are very similar to each other yet distinct from "classical" RNA polymerase II and RNA polymerase III promoters. In most organisms studied, transcription of U1, U2, U4, and U5 genes by RNA polymerase II or U6 genes by RNA polymerase III requires a unique proximal sequence element (PSE) that is located upstream of position Ϫ40 relative to the transcription start site (Fig. 1A) (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12).
In vertebrates, the PSEs of U1 and U2 genes are functionally interchangeable with the PSEs of U6 genes (13,14). That is, if the PSE of the U6 promoter is replaced with the U1 or U2 PSE, there is no effect on the RNA polymerase III specificity of the U6 promoter. Likewise, the U6 PSE can functionally substitute for the U1 or U2 PSE in the vertebrate U1 and U2 promoters for transcription by RNA polymerase II. In echinoderms and plants, the U1, U2, and U6 PSEs (called USEs in plants) are similarly functionally interchangeable with each other (15)(16)(17).
The RNA polymerase III specificity of vertebrate U6 snRNA genes is determined by the presence of a TATA box at a fixed distance downstream of the PSE (2,13,18). Paradoxically, the promoters of the vertebrate snRNA genes transcribed by RNA polymerase II lack TATA boxes (Fig. 1A). In plants, on the other hand, snRNA genes transcribed by both RNA polymerases have essential TATA boxes in their promoters. In this case, the choice of RNA polymerase is determined by a 10-base pair difference in the spacing between the USE and the respective TATA box (Fig. 1A) (16,17,19).
In contrast to the organisms described above, in vitro experiments carried out in the Drosophila melanogaster system indicated that the sequence of the PSE itself plays a major role in determining the RNA polymerase specificity of snRNA gene promoters in insects (20). The basal promoter elements of Drosophila snRNA genes are shown in the last two lines of Fig. 1A. The Drosophila proximal sequence element A (PSEA) is a 21base pair conserved sequence that is analogous to the vertebrate PSE in that it is required for a basal level of transcription and is involved in specifying the transcription start site (7). The U6 genes of Drosophila, like those of other organisms, contain canonical TATA boxes. The proximal sequence element B (PSEB), which is found in Drosophila RNA polymerase IItranscribed snRNA genes, has no obvious counterpart in other organisms. Finally, the spacing of the PSEA from the PSEB (8 base pairs) or from the TATA box (12 base pairs) differs between the two classes of Drosophila genes. Fig. 1B shows typical sequences of the various Drosophila U1 and U6 promoter elements.
In a previous study, the cis-acting determinants of RNA polymerase specificity in the Drosophila system were studied by transcribing artificial templates that contained all possible combinations of U1 or U6 PSEA, PSEB, or TATA box and 8 or 12 base pair inter-element spacing (20). The results of those in vitro transcription experiments indicated that the PSEB and TATA sequences were essentially interchangeable in terms of their effects on RNA polymerase specificity (20). Similarly, the difference in the spacing between the elements affected primarily the efficiency of transcription rather than RNA polymerase specificity. The major determinant of RNA polymerase speci-ficity lay within the sequence of the PSEA itself (20), even though the U1 and U6 PSEAs differed by only a few nucleotides (Fig. 1B).
To thoroughly investigate the role of the PSEA as a determinant of RNA polymerase specificity, we felt that it was important to interchange the U1 and U6 PSEAs within their natural promoter contexts and to examine the effects of such an exchange on promoter activity in living cells. These experiments validate the notion that the U1 and U6 PSEAs are functionally distinct in terms of their abilities to selectively promote transcription by either RNA polymerase II or RNA polymerase III, respectively. Unlike the in vitro situation, however, the accumulation of transcripts from promoters that contain the "wrong" PSEA is suppressed in vivo.

EXPERIMENTAL PROCEDURES
In Vitro Transcription of U1 and U6 Gene Promoter Templates-The templates that contained the wild type U1 and U6 promoter sequences have been described previously (7,21). The U1 plasmid template contained sequences from the Drosophila U1-95.1 gene from position Ϫ391 to position ϩ32. The U6 in vitro transcription template contained sequences of the Drosophila U6-2 gene from position Ϫ409 to position ϩ59. To change each PSEA sequence to that of the other gene, each template was cut at natural restriction enzyme sites that flanked the respective PSEA, and the intervening sequence was replaced with synthetic double-stranded oligonucleotides that contained the PSEA sequence of the other gene (constituting 5 base changes in the 21-base pair PSE). In vitro transcription reactions and analysis by primer extension were performed as described previously (7,22).
Expression of U6 Maxigene Constructs by Transfection of Drosophila Tissue Culture Cells-The plasmid pU6-maxi was a gift from Deborah Johnson (Departments of Molecular Pharmacology and Biochemistry, University of Southern California). This plasmid contained an insertion of 25 base pairs at position 66 in the U6 RNA coding region. To replace the U6 PSEA in this construct with the U1 or mutant PSEA, the plasmid was digested with AflII and BclI, which cut on opposite sides of the PSEA; the intervening DNA was then replaced with synthetic oligonucleotides that contained either the U1 PSEA sequence or a mutant sequence that was altered at 20 nucleotide positions relative to the U6 PSEA. These plasmids were then modified by deleting 15 base pairs of the original maxigene insertion sequence; this resulted in a shorter insertion (GCGCGGATCG) between nucleotides 66 and 67 of the U6 coding region. The modified maxigene derivatives were used to transfect Drosophila S2 cells maintained in Schneider's medium supplemented with 10% fetal bovine serum (Life Technologies, Inc.).
Cells were grown to 40 -60% confluence in Falcon 25-cm 2 tissue culture flasks and transfected using Promega's Profection System. Each flask was cotransfected with 6 g of one of the U6 maxigene plasmids, 2 g of an internal control plasmid that contained the firefly luciferase gene driven by a Drosophila heat shock promoter, and 12 g of pBlue-scriptSK(ϩ) vector (Stratagene). Cells were harvested 40 h after transfection. A 0.25 aliquot of each flask of cells was separated out and used for luciferase assays (see below) to normalize for relative transfection efficiency. RNA was isolated from the remainder of the cells using Promega's RNAgents Total RNA Isolation System, and samples were assayed for maxigene expression using Promega's Primer Extension System. The first 13 nucleotides of the maxigene-specific primer (5Ј-GTGTCATCCTTGCCGATCC-3Ј) were complementary to the wild type U6 snRNA sequence, and the last 6 were complementary to the maxigene insertion sequence. Products of the primer extension reactions were resolved on 8% denaturing polyacrylamide gels, and band intensities were quantified by phosphorimager analysis.
Luciferase Gene Expression Driven by the U1 Promoter in Transient Transfection Assays-U1-95.1 gene sequences between positions Ϫ381 and ϩ32 were cloned into the promoter-less luciferase vector pGL2-Basic (Promega). Nucleotide substitutions corresponding to those found in the U6 PSEA were introduced into the U1 PSEA at positions 7, 14,16,19, and 20 (either as individual substitutions or in groups, see Figs. 1 and 4) by digesting the U1 promoter region with restriction enzymes MluI and BlpI that cut on either side of the U1 PSEA. Then doublestranded synthetic DNA oligonucleotides that contained the desired sequences were inserted to generate seven different chimeric U1-luciferase variants (with U1 or U6 or hybrid U1/U6 PSEAs in the U1 promoter) as indicated in Fig. 4.
For assays to measure luciferase activities, S2 cells were transfected in triplicate with 3 g of one of the chimeric U1/pGL2 constructs, 3 g of copia-lacZ plasmid (obtained from D. Johnson) as an internal control for transfection efficiency, and 14 g of pBluescriptSK(ϩ) plasmid to bring the total amount of DNA to 20 g. Approximately 48 h after transfection, cells were harvested using Promega's Reporter Lysis Buffer. Luciferase activities were determined using BD PharMingen's Enhanced Luciferase Assay kit.
For primer extension assays, S2 cells were transfected in triplicate with 3 g of one of the chimeric U1/pGL2 constructs, 0.9 g of a cytomegalovirus-Renilla luciferase construct as an internal control (obtained from R. Tjian, Department of Molecular and Cell Biology, University of California, Berkeley, CA), and 16 g of pBluescriptSK(ϩ). Primer extension reactions were carried out as described above using a primer (5Ј-CGTCTTCCATTTTACCAACAGTACC-3Ј) complementary to DNA sequences near the 5Ј end of the luciferase gene. An extension product 84 nucleotides in length was expected for transcripts initiated at position ϩ1 of the U1 gene. Relative expression levels from the various constructs were determined by phosphorimager analysis of the primer extension products.
NeoR Gene Expression Driven by the U1 Promoter in Transient Expression Assays-The neoR gene was obtained from the pP{hsneo} Pelement vector (23) (also known as pUChsneo (24)) by polymerase chain reaction with primers that flanked the neoR gene and contributed an upstream BclI site and a downstream BamHI site. This fragment was cloned into the BamHI site of pBluescriptSK(ϩ) such that the SpeI site of the vector was upstream of the neoR gene, producing the construct pSK/neo. A fragment containing wild type U1 promoter sequences extending from positions Ϫ381 to ϩ32 was recovered from the U1(wild type)/pGL2 construct described in the preceding section by digestion with NheI and XbaI. This fragment was cloned in both the forward and reverse orientations into the SpeI site of the pSK/neo construct to generate the constructs pSK/U1(wild type)/neo and pSK/U1(reverse)/ neo. The construct pSK/U1(U6PSEA)/neo, in which the U6 PSEA replaced the U1 PSEA in the U1 promoter region, was prepared in a similar manner.
Transfection of S2 cells was carried out in triplicate using 3 g of pSK/U1(wild type)/neo, pSK/U1(reverse)/neo, pSK/U1(U6PSEA)/neo, or pP{hsneo}. The pBluescriptSK(ϩ) vector was added to bring the total amount of DNA to 20 g. Thirty hours following transfection, cells were transferred to and subsequently maintained in medium containing 0.3 mg/ml G418 (Geneticin, Life Technologies, Inc.). The survival and growth of the cells were monitored by light microscopy over a period of 5 weeks, at which point the experiment was terminated.
P-element-mediated Transformation with U1-Neo Constructs and Determination of G418 Resistance of Transgenic Flies-Chimeric P-element constructs for transformation of flies with the neoR gene driven by the U1 promoter were prepared as follows. The constructs pSK/U1(wild type)/neo and pSK/U1(U6PSEA)/neo described above were each di- gested with SalI and NotI. This released fragments that had the neoR gene downstream of the U1 promoter, which contained either the U1 or U6 PSEA. These two fragments were then cloned separately between the NotI and XhoI sites of the P-element vector pP{CaSpeR-4} (Gen-Bank TM accession number X81645) that carries a functional allele of the white gene (23).
Transformation of flies (25) was carried out by microinjection of yw embryos at the pre-blastoderm stage with one of the two P-element constructs. Eye color in the F1 generation flies was used to identify transformants. Appropriate crosses then generated lines homozygous for the transgene insertion.
To test the resistance of the transformed fly lines to G418, eight males and 15 females of the same line were placed in bottles and allowed to lay eggs on food that contained 0.4 mg/ml G418. Adult flies were then removed, and the number of first generation offspring surviving to adulthood were counted and totaled.

Exchanging the U1 and U6 PSEAs Indicates That They Are
Not Functionally Equivalent in Vitro-Previous transcription studies that examined the role of cis-acting elements in determining RNA polymerase specificity utilized synthetic DNA templates that contained only the essential Drosophila U1 or U6 promoter elements and lacked most of the remaining sequences present in the wild type U1 or U6 gene 5Ј-flanking DNA (20). Those studies indicated that the U1 and U6 PSEAs could act as the major determinants of RNA polymerase specificity in vitro. However, we also wished to determine whether this would hold true in the context of the natural U1 and U6 gene 5Ј-flanking DNAs. To examine the effect of substituting the U6 PSEA into the context of the wild type U1 promoter, or conversely substituting the U1 PSEA into the context of the U6 promoter, the constructs illustrated at the top of Fig. 2 were prepared. These were transcribed in vitro as described previously using a soluble nuclear protein fraction competent for both RNA polymerase II and RNA polymerase III transcription (22). The transcripts were detected by primer extension analysis.
As noted previously (7,22), the wild type U1 gene was transcribed in vitro by RNA polymerase II as indicated by the sensitivity of the transcription signal to the RNA polymerase II inhibitor ␣-amanitin but its resistance to the RNA polymerase III inhibitor tagetitoxin (Fig. 2, lanes 1-4). When the U1 PSEA was converted to a U6 PSEA via a 5-base pair substitution, the transcription signal decreased, but, more importantly, the promoter specificity was changed to RNA polymerase III (Fig. 2, lanes [5][6][7][8]. This was indicated by the sensitivity of the transcription product to tagetitoxin but its resistance to ␣-amanitin. Thus, a 5-base pair substitution that altered the U1 PSEA to a U6 PSEA was sufficient to switch the promoter specificity from RNA polymerase II to RNA polymerase III, even though the remainder of the promoter and the transcription start site was derived entirely from the U1 gene. The results of the converse experiment are shown in Fig. 2, lanes 9 -16. As observed previously (22), the wild type U6 gene was transcribed exclusively by RNA polymerase III (lanes 9 -12). When the U6 PSEA was converted to a U1 PSEA, transcription by RNA polymerase II was activated (compare lanes 11 and 15). However, transcription of this construct by RNA polymerase III was also still observed (lane 14). This is not surprising since the same result was obtained previously from an analogous artificial promoter construct (20). The reason for the RNA polymerase III activity of this template is 2-fold. First, the TATA sequence TTTATATA by itself promotes substantial transcription by RNA polymerase III (as well as by RNA polymerase II) that is independent of the PSEA (21). Second, the U1 PSEA functions very ineffectively at a 12-base pair spacing from either the downstream PSEB or TATA box (20). Thus, when a TATA-containing construct is expressed in the Drosophila in vitro transcription system, there often exists a high background of TATA-mediated transcription by both RNA polymerases II and III that is independent of the PSEA. This obscures potential effects of the U1 PSEA when substituted into the U6 promoter.
To overcome this limitation of the in vitro transcription assay, we turned to an in vivo assay. Although a TATA box alone can be sufficient for transcription by RNA polymerases II or III in nuclear extracts, the formation of a stable, active transcription complex in vivo normally requires the presence of additional promoter or activator elements besides the TATA box. Therefore, in the experiments described below we investigated the functional interchangeability of the U1 and U6 PSEAs under the more stringent conditions of transcription in vivo.
The U1 and U6 PSEAs Are Not Functionally Interchangeable for RNA Polymerases II and III Promoter Activity in Transfected Tissue Culture Cells-We next examined whether the Drosophila U1 PSEA could functionally substitute for the U6 PSEA to promote transcription of the U6 gene in Drosophila S2 cells. For this purpose, the three constructs illustrated at the top of Fig. 3 were prepared. The upper construct contained the wild type U6 promoter and complete U6 gene with a 10-base pair insertion between nucleotides 66 and 67 of the U6 RNA coding region. The insertion permitted the differentiation of the transfected gene product from the endogenous U6 RNA. The second construct was identical to the first except for five nucleotide substitutions that converted the U6 PSEA to a U1 PSEA, and the third construct contained a highly mutated PSEA with 20 of 21 base positions altered. These three constructs were used to transfect S2 cells. Two days after transfection, RNA was isolated from the cells and assayed by primer extension for the presence of the U6 maxigene transcript.
RNA from cells transfected with the construct that contained the wild type U6 promoter yielded a primer extension product of the size expected (89 nucleotides) for initiation of transcription at position ϩ1 of the U6 gene (Fig. 3, lane 2). In contrast, substitution of the U1 PSEA into the U6 promoter reduced transcription activity to almost undetectable levels (Fig. 3, lane  3). Upon prolonged exposure, a very weak band was visible in some experiments, but phosphorimager analysis indicated that transcription was reduced more than 65-fold relative to the wild type U6 promoter. This level was barely above the background obtained either with a mock transfection or with a construct that contained the extensively mutated PSEA (Fig. 3,  lanes 1 and 4). From these results it is clear that the U1 PSEA could not effectively substitute for the U6 PSEA as a component of the U6 promoter. The promoter activity of the U1 PSEA was apparently suppressed in the context of the surrounding U6 DNA.
We next performed a reciprocal set of experiments to examine whether the U6 PSEA could functionally substitute for the U1 PSEA in the U1 promoter. Since the U1 promoter normally recruits RNA polymerase II, we used luciferase as a reporter gene for these experiments. The simplicity and sensitivity of the luciferase assay allowed us to examine a greater number of constructs that contained several additional variants of the PSEA. Besides replacing the U1 PSEA with the wild type U6 PSEA, constructs were also prepared using "hybrid" PSEA sequences that contained a mixture of U1 and U6 nucleotides. A generalized schematic illustration of this family of constructs is shown at the top of Fig. 4. For the hybrid PSEAs, we employ a nomenclature that indicates whether a U1 or U6 nucleotide is present at positions 7, 14, 16, 19, or 20 within the PSEA (Fig.  1B). For example, 11166 indicates that U1 nucleotides are present at positions 7, 14, and 16 but U6 nucleotides at positions 19 and 20.
Following transfection of S2 cells with the indicated constructs, extracts were prepared and luciferase activities determined. The wild type U1 promoter efficiently drove expression of the luciferase reporter (Fig. 4B). However, 5 base changes to those of the U6 PSEA (66666) resulted in the reduction of reporter gene expression essentially to background levels. Thus, the U6 PSEA was unable to functionally substitute for the U1 PSEA to recruit RNA polymerase II to the U1 promoter.
With the constructs containing hybrid PSEAs (11161, 11166, 61166, 16166, 11666), the RNA polymerase II activity of the promoter decreased as the U6 character of the PSEA increased.
The right-hand column in Fig. 4B recapitulates results from previous work that determined the in vitro RNA polymerase specificity of synthetic constructs that contained the same set of hybrid PSEAs (20). Significantly, the decrease in transcription seen in the transfection assays paralleled the switch from RNA polymerase II to RNA polymerase III specificity observed previously in the in vitro determinations.
Primer extension analyses of several of the U1/pGL2 constructs that spanned the range of transcriptional activities were carried out to confirm the luciferase reporter results at the level of the RNA produced (Fig. 4C). Cells transfected with the construct that contained the wild type U1 promoter yielded a primer extension product of 84 nucleotides, the size expected for initiation of transcription at position ϩ1 of the U1 gene (Fig.  4C, lane 1). In contrast, substitution of the U6 PSEA into the U1 promoter reduced the accumulation of transcripts to near background levels (Fig. 4C, lanes 5 and 6). Constructs that contained the hybrid 11161 and 11166 PSEAs produced intermediate levels of transcription products (lanes 2 and 3); the 11666 construct (lane 4) was essentially no more active by primer extension analysis than the 66666 construct. For each construct, the relative level of expression was somewhat lower when measured by primer extension (Fig. 4C) than when measured by luciferase activity (Fig. 4B). This may reflect the possibility that the luciferase enzyme assay is more sensitive than the primer extension assay at the lower transcription levels.
Altogether, these results indicate that the U6 PSEA was unable to functionally substitute for the U1 PSEA in the U1 promoter to effectively recruit RNA polymerase II to transcribe the chimeric reporter gene. Furthermore, there was a gradual decrease in transcript accumulation in cells as the U6 character of the PSEA was increased. Thus, the primer extension data provide no evidence for activation of RNA polymerase III transcription in vivo when the U6 PSEA is substituted into the U1 promoter; instead, transcription seems to be suppressed.
The U1 Promoter Efficiently Drives Expression of the Neomycin Resistance Gene in Vivo, but Not upon Substitution of the U6 PSEA-As an alternative method of examining the properties of the U1 and U6 PSEAs in living cells, the neoR gene, which confers resistance to the drug G418, was cloned downstream of the wild type U1 promoter. A similar construct was made in which the U1 PSEA was replaced with the U6 PSEA (Fig. 5). Two additional constructs were used as controls: the first (negative control) contained the U1 promoter cloned in reverse orientation relative to the neoR gene; the other (positive control) utilized the heat shock promoter to drive neoR gene expression (Fig. 5).
Drosophila S2 cells were transfected with these four constructs, and 30 h later growth medium containing 0.3 mg/ml G418 was added to the cells. The fate of the cells was monitored visually by light microscopy over a period of several weeks. On all plates, the majority of the cells stopped dividing and began to die after a few days. Within 2 weeks, viable colonies of cells were no longer visible in the plates that were transfected with the constructs that contained either the U6 PSEA substitution or the entire U1 promoter in reverse orientation (Fig. 5, right  column). In contrast, colonies of cells continued to grow and divide on plates transfected with constructs that contained the neoR gene under the control of either the wild type U1 or heat shock promoters. When grown continuously on medium containing 0.3 mg/ml G418, these latter cells continued to divide and remained robust and healthy until the experiment was terminated 5 weeks after the initial transfection. From these results we conclude that the wild type U1 promoter can effectively drive expression of the neoR gene. However, the U6 PSEA was unable to substitute for the U1 PSEA for this purpose, presumably because it was unable to recruit the RNA polymerase II required for production of a functional mRNA.
To confirm these results in a truly in vivo situation, similar neoR gene constructs were introduced into the germ line of yw flies by P-element transformation. Homozygous transgenic fly lines were then isolated that contained the neoR gene either under the control of the wild type U1 promoter or under the control of a U1 promoter in which the PSEA had been converted to a U6 PSEA. Several of these transgenic fly lines were then tested for in vivo expression of the neoR gene by allowing flies to lay eggs on food that contained 0.4 mg/ml G418. Significant numbers of G418-resistant offspring were produced from parental flies that contained genomic copies of the neoR gene downstream of the wild type U1 promoter (Table I). In contrast, the flies that contained the substitution of the U6 PSEA in the U1 promoter produced no viable offspring when subjected to the same G418 selection (Table I). These results indicate that the U1 promoter can efficiently drive expression of the neoR gene, but substitution of the U6 PSEA effectively inactivated the RNA polymerase II activity of the U1 promoter in vivo. DISCUSSION Previous in vitro studies suggested that the RNA polymerase specificity of Drosophila snRNA genes was determined primarily by a few nucleotide differences within the 21-base pair PSEA sequence. That conclusion was based upon transcription of synthetic promoter constructs that contained various mixand-match features of the U1 and U6 gene promoters. However, it was important to validate those findings both in cells and in the context of the natural promoter environment of the U1 and U6 genes. The data reported here reveal that the Drosophila U1 and U6 PSEAs are functionally distinct in vivo as well as in vitro: the U1 PSEA is compatible with transcription only by RNA polymerase II, and the U6 PSEA supports transcription only by RNA polymerase III. However, there are some differences as discussed below.
Interchanging the U1 and U6 PSEAs Results in RNA Polymerase Switching in Vitro but Suppression of Transcription in Vivo-Earlier published data (20), as well as those presented here (Fig. 2), indicated that the PSEA can act as a dominant element in vitro to determine the RNA polymerase specificity of Drosophila snRNA gene promoters. That is, exchanging the U1 and U6 PSEAs resulted in switching the RNA polymerase  Fig. 4B is included to summarize earlier data on the in vitro RNA polymerase specificity of promoter constructs that contained the same hybrid PSEAs (20). C, the activities of several of the chimeric U1-luciferase promoter constructs were assayed by primer extension analysis following transient transfection. An autoradiogram of the results is shown, and the column on the right lists the relative activities of the constructs as determined by phosphorimager analysis.
FIG. 5. The U1 promoter can drive expression of a neomycin resistance gene, but not if the promoter contains a U6 PSEA substitution. Schematic illustrations of four constructs used to transfect S2 cells are shown. The first construct contained a wild type U1 promoter in front of the neoR gene, and the second contained a 5-base substitution that converted the U1 PSEA to a U6 PSEA. The third and fourth constructs were negative and positive controls, respectively. Following transfection, cells were grown on medium containing the drug G418. The column to the right indicates whether G418-resistant cell lines were obtained from flasks of cells that were transfected with the indicated chimeric constructs. Transfections with each construct were carried out in triplicate, and each repetition gave the same results. specificity. In transfection experiments, however, transcription from promoters that contained the wrong PSEA was suppressed rather than switched (Figs. 3 and 4). Several factors could play a role in this apparent difference between the in vivo and in vitro observations. First, the requirements for transcription are undoubtedly more stringent in vivo than in vitro. The more permissive in vitro reactions contained only a single type of promoter template that was present in a large number of copies (ϳ10 11 ) per reaction tube. A detectable signal can be produced in vitro even though only a small percentage of the template molecules are actively transcribed (26). In vivo, on the other hand, the transfected copies of the gene must compete against thousands of other genes within the cell. If the stability of the transcription pre-initiation complex is even partially compromised by the switch of the PSEAs, the introduced genes may not compete effectively with endogenous genes for available transcription factors.
Chromatin structure may also play a role in the suppression of transcription in vivo from constructs that contain the wrong PSEA. Positioned nucleosomes have been implicated in both the activation and repression of transcription from snRNA gene promoters (27)(28)(29)(30). It is possible that the chromatin environment required for optimal RNA polymerase III transcription is different from that required for optimal RNA polymerase II transcription. The modification of chromatin structure is one of several effects that can be attributed to the binding of transcriptional activator proteins and coactivators to upstream sequence elements. In vertebrates, at least two upstream activator proteins (Oct-1 and SBF/Staf) involved in transcription of snRNA genes by RNA polymerase II are also utilized for transcription of U6 genes by RNA polymerase III (31)(32)(33)(34)(35). At the present time, however, there is no information available about comparable upstream activators for Drosophila snRNA genes. It is possible that the Drosophila system could utilize different sets of upstream activator proteins that differentially modify the chromatin environment of the U1 and U6 genes. This could restrict polymerase usage in cells. Alternatively, different activator proteins may function effectively to recruit only a given RNA polymerase or polymerase-specific general transcription factor. In this regard, it is interesting that transcription of U6 genes in Drosophila utilizes the TBP-related factor TRF1 instead of the classical TATA-binding protein TBP (36).
Finally, it is also important to consider the fact that the absence of detectable transcripts from U1 or U6 promoters that contain a switched PSEA may be due to post-initiation events as well as to effects on transcription initiation. For example, if RNA polymerase II does initiate transcription of the U6 maxi-gene that contains the U1 PSEA substitution, these transcripts would very likely neither be terminated nor processed correctly in cells. Such aberrant transcripts would probably be subject to more rapid degradation than normal cellular RNAs. Similarly, potential RNA polymerase III-initiated transcripts that arise from U1 promoter constructs that contain the U6 PSEA may be rapidly turned over in cells. As a result, our ability to detect transcripts initiated in vivo by the "wrong" polymerase is to some extent limited. On the other hand, there is probably very little if any preferential degradation of abnormal transcripts synthesized in vitro.
Despite the above discussed limitations on the interpretation of the data, the main premise of the findings is clear: the Drosophila U1 and U6 PSEAs are functionally distinct and are not interchangeable either in vivo or in vitro. The U1 PSEA can effectively support only RNA polymerase II transcription, and the U6 PSEA functions effectively only for RNA polymerase III transcription.
Efficient Expression of Protein-coding Genes from the Drosophila U1 Promoter-Our results with the luciferase and neoR genes as reporters indicate that the U1 promoter can efficiently drive expression of functional mRNAs in Drosophila cells. In our hands, the Drosophila U1 promoter was more active than the copia promoter in the production of luciferase activity but less active than the heat shock promoter. 2 In vertebrate systems, an early report indicated that the U1 promoter could not drive the expression of a functional mRNA (37). Later, however, it was demonstrated that sea urchin U1 promoters could be used to produce fully functional histone mRNAs (38). More recent studies have demonstrated that even vertebrate RNA polymerase II snRNA promoters can indeed produce functional mRNAs if care is taken to avoid cryptic 3Ј end formation signals within the coding region of the gene (39,40). We have noticed that Drosophila U1, U2, and U4 genes have a well conserved sequence (consensus CATTTATAAATAAATATNNA) in their 3Ј-flanking DNA that may act in 3Ј end formation of Drosophila snRNAs. Sequences similar to these are not present internal to the luciferase or neoR genes. 2 Thus, it is not surprising that the Drosophila U1 promoter can be highly effective for the production of functional mRNAs encoding either luciferase activity or G418 resistance.
Previously published data from our laboratory indicated that the D. melanogaster PSEA-binding protein (DmPBP) interacts differently with the DNA in vitro depending upon whether it is bound to a U1 or U6 PSEA sequence (41,42). We therefore TABLE I G418 resistance of transgenic fly lines that contain the neoR gene driven by the U1 gene promoter Transgenic fly lines were tested for their ability to produce offspring surviving to adulthood when the larvae were raised on food containing 0.4 mg/ml G418. Two bottles were prepared for each fly line; each bottle was seeded with 8 males and 15 females. Lines designated U1-Neo contained transgenes with the wild type U1 promoter; lines designated U1(U6 PSEA)-Neo contained transgenes with a substitution of the U6 PSEA for the U1 PSEA in the U1 promoter.  hypothesize that DmPBP assumes different conformations upon binding to a U1 or U6 PSEA and that these distinct conformations are each compatible with the downstream recruitment of only RNA polymerase II or RNA polymerase III, respectively. Our current results pave the way toward the use of Drosophila transgenic and genetic technologies to define the molecular interactions that contribute to RNA polymerase specificity. As one example, it may be feasible to obtain suppressor-like mutations in DmPBP that alter the RNA polymerase specificity normally associated with the U1 or U6 PSEAs.