Transcriptional control of Ca2⁺-activated K⁺ channel expression: identification of a second, evolutionarily conserved, neuronal promoter

Nerve cells encode and transmit information in the form of electrical signals. The production of these impulses depends upon the cooperative effort of a number of distinct ion channels. These channels conduct the ionic currents responsible for the cell’s electrical properties: its resting potential, its sensitivity to stimulation and the shape and duration of its action potential. In short, they shape the cell’s input–output properties. Any given neuron will typically have one or more different inward currents and perhaps four or more outward currents. Together, these currents, each conducted by a different channel, give rise to the overall electrical properties of the cell. Not only do these channels shape their electrical environment, they also sense it and alter their activity in response to it. From this complex interplay between electrical environment and channel activity arises the cell’s electrical character.

Eukaryotic genomes encode a staggeringly large number of distinct ion channels (Jan and Jan, 1990; Wei et al., 1996). The choice of which channels to express is an extremely significant cellular decision because it delimits the range of electrical properties that a cell can produce. It is therefore important to determine how this decision is made. Our model for studying the mechanics and consequences of channel gene regulation has been the slowpoke gene of Drosophila melanogaster. The Drosophila slowpoke gene was the first Ca²⁺-activated K⁺ channel gene to be cloned (Atkinson et al., 1991) and it is the Drosophila homolog of the vertebrate BK-type Ca²⁺-activated K⁺ channel gene. BK channels are well-known for their roles in determining the firing pattern of neurons and for modulating the contractile properties of muscles (Rudy, 1988; Latorre et al., 1989; Brayden and Nelson, 1992; Hille, 1992; Robitaille et al., 1993; Issa and Hudspeth, 1994; Nelson et al., 1995).

The complete slowpoke transcriptional control region was previously determined to be contained within 11 kb of genomic DNA (Brenner et al., 1996). Using a lacZ reporter gene and transgenic flies, we demonstrated that this genomic DNA could reproduce the entire slowpoke expression pattern. Within this region, tissue-specific promoters were mapped. Separate promoters for neural and midgut expression were identified, while muscle and tracheal cell expression were shown to arise from a shared promoter (Brenner and Atkinson, 1996; Brenner et al., 1996; Thomas et al., 1997). Presumably, tissue-specific promoters enable the gene to tailor the sequence or abundance of a channel to the specific needs of the cell.

To simplify the identification of sequences that drive slowpoke expression in specific tissues, we chose to rely on the fact that evolution favors the conservation of functionally important sequences. Here, we begin the hunt for sequences that direct expression in structures of the central and peripheral nervous systems. Using conservation as our metric, we have identified a new neural promoter and a large number of potential control elements.

A 414 base pair (bp) BamHI/ApaI fragment from the slowpoke cDNA Z54 (Becker et al., 1995), that contained exon C1 and C3, was used to probe a Drosophila hydei genomic library (O’Neil and Belote, 1992) carried in λEMBL4 (generously provided by Dr John Belote, Syracuse University) under reduced stringency. Hybridization and washing conditions were as follows. Hybridization: 20 % (v/v) formamide, 6× SSPE (prepared from a 20× stock solution containing 3.6 mol l⁻¹ NaCl, 0.02 mol l⁻¹ disodium EDTA and 0.2 mol l⁻¹ NaPO₄, pH 7.7), 10× Denhardt’s solution [prepared from a 50× stock solution containing 1 % Ficoll, 1 % bovine serum albumin, 1 % poly(vinylpyrrolidone)], 0.2 % SDS and 200 μg ml⁻¹ salmon sperm DNA at 42 °C; wash: 2× SSPE, 0.1 % SDS at 65 °C. Of four purified clones, one also hybridized to an XhoI–BamHI fragment containing the neural promoter C1 from Drosophila melanogaster. This one was chosen for further study.

DNA fragments from Drosophila hydei were subcloned into pBluescriptII using convenient restriction enzyme sites. Nested deletions were introduced using the enzyme Bal31 (Sambrook et al., 1989) and sequenced using the dideoxy chain termination method (Sanger et al., 1977). Accession numbers for the D. melanogaster and D. hydei sequences are U40221 and AF210728, respectively.

All the reporter genes described have a lacZ gene inserted into the 3′-most ApaI site of the slowpoke transcriptional control region (Fig. 1). Expression of all reporter genes were assayed by β-galactosidase staining as described in Brenner et al. (1996). The construction of reporter gene constructs P1 and P3 has been described (Becker et al., 1995; Brenner et al., 1996), and they are carried in the vector pCaSperβgal (Thummel et al., 1988). P1 contains the 11 kb full-length slowpoke transcriptional control region as presently defined. P3 was derived from P1 by deleting a region that contained all the slowpoke promoters except Promoters C0 and C1. P3 has been shown to reproduce the complete slowpoke neuronal expression pattern. P12 and P13 were derived from P3 using the ExoIII/S1 method (Sambrook et al., 1989). The construction of P12 has been described by Thomas et al. (1997). For both, the P3 plasmid was digested using SphI and SpeI to generate a protected 3′ overhang at the SphI site and a 5′ overhang at the SpeI site. Unidirectional deletions were then made using the ExoIII/S1 enzyme combination. Deletions were ligated, transformed into bacteria and then screened using restriction digests to identify those of proper length. The extent of the deletion was confirmed by sequence analysis. P12 contains a 950-nucleotide deletion that includes Promoter C0. P13 contains a 620-nucleotide deletion that does not remove this promoter.

The transformation constructs (1 μg μl⁻¹) and the helper plasmid pπ25.7 (200 ng μl⁻¹) were co-injected into w¹¹¹⁸Drosophila embryos (Spradling, 1986). Transformants were identified on the basis of complementation of w¹¹¹⁸ by the white gene of pCaSperβgal.

Total RNA was prepared from 0–24 h embryos, wandering third-instar larvae and whole adult animals. Reverse transcription (RT) was performed using the RABRT1 primer (see below) which specifically anneals to exon C3, an exon common to all known slowpoke transcripts.

Conditions for reverse transcription were as follows: 10 μg of total RNA and 40 pmol of RABRT1 primer were mixed, incubated at 70 °C for 10 min and then allowed to cool to 42 °C. The reaction was initiated by adding NEB M-MuLVRT buffer (50 mmol l⁻¹ Tris-HCl, pH 8.3, 8 mmol l⁻¹ MgCl₂, 10 mmol l⁻¹ dithiothreitol, DTT), 2.5 mmol l⁻¹ each of dGTP, dATP, dTTP and dCTP (dNTPs) and 12.5 units of Moloney murine leukemia virus reverse transcriptase (New England Biolabs). Following incubation at 42 °C for 45 min, the cDNA from each tissue type was subjected to the polymerase chain reaction (PCR) using primers RABP1 and GAMMA5 in the presence of 1.5 or 2.5 mmol l⁻¹ MgCl₂ using standard conditions. Primer RABP1 anneals to exon C3 immediately upstream of the RABRT1 binding sites. GAMMA5 anneals to homology block 5 (see Fig. 2). Primer annealing temperature was determined using the OLIGO program (National Biosciences, Inc.). The primers used in this study were as follows: RABRT1, 5′-CGGCGTCGAATGGTGAATCTGTTGG-3′; RABP1, 5′-AATGATTCGACAGTGCTTTGAT-3′; GAMMA4, 5′-AAT-GTTATTTTTGTTGCTTCCT-3′; GAMMA5, 5′-ATTGTA-TACGCTGCTGACGAGA-3′.

The RNA-ligase-mediated rapid amplification of cDNA ends procedure (RLM-5′-RACE) was based on the procedure described by Schaefer (1995), taking into consideration the recommendations of Frohman (1995). Total DNAase-digested RNA (100 μg) was incubated with 30 units of calf intestinal alkaline phosphatase (New England Biolabs) in 100 μl of 1× digestion buffer (50 mmol l⁻¹ Tris-HCl, pH 7.9, 10 mmol l⁻¹ MgCl₂, 100 mmol l⁻¹ NaCl, 1 mmol l⁻¹ DTT) at 42 °C for 1 h. The reaction was terminated with a single 300 μl phenol:chloroform:isoamyl alcohol (50:49:1; PCIA) extraction. The nucleic acid was precipitated with 2.5 mol l⁻¹ ammonium acetate and 200 μl of ethanol, washed once with 70 % ethanol, and resuspended in 100 μl of water. The 7-methyl guanosine triphosphate cap was removed with 10 units of tobacco acid pyrophosphatase (TAP; Epicentre) in 50 mmol l⁻¹ sodium acetate (pH 6.0), 1 mmol l⁻¹ EDTA, 0.1 mmol l⁻¹ DTT, 0.01 % Triton X-100 and 0.1 % β-mercaptoethanol at 37 °C for 1 h. The mixture was extracted with PCIA, once with chloroform:isoamyl alcohol (49:1; CIA), precipitated with 2.5 mol l⁻¹ ammonium acetate and two new volumes of ethanol, washed once with 70 % ethanol and resuspended in 50 μl of water. A 135-nucleotide RNA anchor was generated by in vitro transcription (Krieg and Melton, 1987; Krieg, 1990). The RNA was digested with DNAase I (37 °C for 15 min) and purified by agarose gel electrophoresis, PCIA- and CIA-extracted and precipitated, dried and resuspended in 100 μl of RNAase-free water. A sample (10 μl) of the anchor was added to 10 μg of TAP-treated RNA and ligated using T4 RNA ligase (New England Biolabs) in the buffer recommended by the manufacturer at 18 °C for 16 h. The ligation product was PCIA-extracted, precipitated, washed and resuspended in 10 μl of water. RT-PCR was performed using the ThermoScript RT-PCR kit (Gibco-BRL). Reverse transcription was performed using 15 units of thermoscript reverse transcriptase (Gibco-BRL) in a volume of 30 μl using a primer RABRT1 (see above) that annealed to exon C3. A sample (2 μl) of the cDNA synthesis was used to seed the PCR reaction. Two rounds of LA-PCR (long and accurate PCR) were used to amplify the product (Barnes, 1994; Cheng et al., 1994). The first round used the gene-specific primer RABP1 (see above) and an anchor-specific primer GL1 (5′-CACCTCAGGTTCAGGCTCTT-3′). The second round of LA-PCR used the gene-specific primer 184 (5′-CCGTCTTGATCGATAGTTGTTCGTTC-3′) and the anchor primer GL2 (5′-ATTGCTGCCTTTGAAGTCTCCA-3′). The products were Southern-blotted, and the exon-C0-containing products were identified by hybridization. This band was gel-purified, cloned into the vector pBlunt (Invitrogen) and identified by colony hybridization.

RNAase protection assays (RPA) were performed using the Ambion Maxiscript kit and Ambion RPAII kit (Ambion, Austin, TX, USA). A single RPA probe which contained exon C0 and C1 sequences was used to identify Promoter C0 and Promoter C1 products. Digestion products were electrophoresed on 8 mol l⁻¹ urea, 5 % acrylamide gels. DNA sequencing ladders derived from the template were used to determine the size of the products. The relative abundance of the products was determined by densitometrically scanning the lanes and determining the areas under each peak. Areas were normalized for the number of radiolabeled nucleotides incorporated into each protection product.

The previously mapped slowpoke promoters (C1, C1b, C1c and C2) are shown within the transcriptional control region in Fig. 1 (Brenner et al., 1996). RT-PCR, RPA (RNAase protection assay) and deletion analysis experiments indicate that Promoter C1 is active in the Drosophila nervous system (Brenner and Atkinson, 1996; Brenner et al., 1996). Thomas et al. (1997) provided evidence for a second neuronal promoter by deletion analysis. However, these expression studies were performed only in the embryo, and the position of this promoter was only crudely mapped to a 5 kb region 5′ of Promoter C1. This promoter was called Promoter Ce because it was responsible for slowpoke expression in the embryonic central nervous system. Here, we show that it is also expressed in the adult and larval brain in a pattern largely overlapping with Promoter C1. Since it is not specific to the embryo, we have renamed it Promoter C0 in keeping with the numerical numbering of slowpoke promoters.

To help identify sequences important for normal slowpoke expression, we have cloned parts of the slowpoke transcriptional control region from Drosophila hydei. The two species, D. melanogaster and D. hydei, diverged from a common ancestor approximately 60 million years ago (Patterson and Stone, 1952). We expect that control elements will be conserved between the species and that sequences not important for control of expression will have diverged.

In this paper, our analysis has been limited to a genomic sequence beginning approximately 1.5 kb upstream of neuronal Promoter C1 and terminating in the downstream intron that abuts exon C1. To identify conserved regions, we used the Macaw program (National Center for Biotechnology Information). Macaw ranks each conserved block of sequence on the basis of its overall length and similarity. As expected, we observed blocks of similar or identical sequence separated by strikingly dissimilar regions (Figs 1B, 2). We have identified each block of conserved sequence by a number (1–39) that reflects its ranking with respect to the other blocks. Blocks with lower numbers received a higher similarity score than blocks with higher numbers.

These blocks are also conserved in another manner. For both D. melanogaster and D. hydei, the relative position of all the blocks with respect to one another and to Promoter C1 is conserved (Fig. 2). Blocks whose position was not conserved are not shown. Such blocks tended to be tiny and to be composed of simple sequence. Within the region compared, we observed no evidence that chromosome rearrangements have occurred since the evolutionary separation of the two species. Such rearrangements would reorganize the transcriptional control region.

These blocks are likely to have been conserved because they represent functionally important transcription factor binding sites. Conservation of position would be selected for if the groups of factors that bind to the blocks interact with one another and if productive interactions require a particular order and spacing. This added layer of conservation adds credence to the hypothesis that these elements are important for proper slowpoke expression.

One of the first features that we looked for was conservation of Promoter C1. The D. melanogaster Promoter C1 was previously mapped by 5′-RACE, by RPA and by cDNA cloning (Brenner et al., 1996). In D. melanogaster, Promoter C1 is located between conserved blocks 6 and 12 (Figs 1B, 2). We used the NNPP program (Reese, 1994) to search the D. hydei sequence in this area for potential TATA boxes followed by reasonable transcription start sites. TATA boxes direct transcription initiation to a unique nucleotide. The best match was found at the 2228th nucleotide in the D. hydei sequence. This aligns nicely with a TATA box in the D. melanogaster sequence that is 30 bp 5′ to the physically mapped D. melanogaster transcription start site (Fig. 2). Other criteria (see below) support this identification.

Exon C1 was the most strongly conserved block of sequence even though it contributes only 5′ untranslated region (UTR) to the slowpoke transcript. Therefore, it occurred to us that some of the other conserved blocks might represent undiscovered slowpoke exons. Each conserved block was examined to determine whether it contained a consensus splice donor and therefore might represent an exon. Solely on the basis of sequence analysis, two blocks appeared to be good exon candidates. These are blocks 4 and 5 (Figs 1B, 2). To determine whether these blocks encoded exons, we used RT-PCR to determine whether different batches of mRNA contained transcribed versions of blocks 4 and 5. Since all known Drosophila slowpoke cDNAs include exon C3, we employed a reverse transcription primer and a 3′ PCR primer within exon C3. The 5′ PCR primer was specific for block 4 or block 5. Using this primer set, RT-PCR was performed on RNA samples purified from embryos, larvae, pupae and adults. The RT-PCR reaction using the block 4 primer was non-productive, indicating that block 4 does not serve as an exon in these developmental stages. However, the reaction using the block 5 primer amplified a band of approximately 500 bp from RNA purified from all developmental stages (Fig. 3). The only other PCR product was a small artifactual band that appeared when the PCR was carried out at very high Mg²⁺ concentration (not shown). Both fragments were cloned and sequenced. Sequence analysis of the 500 bp band indicated that the amplification product was actually 495 bp. The exon contained wholly or partially within conserved block 5 will from henceforth be referred to as exon C0. It should be noted that block 5 and exon C0 are not identical since block 5 also contains conserved intronic sequences that are probably required for splicing. DNA sequencing of the band amplified in a solution containing a high concentration of MgCl₂ showed that it was composed of primer concatomers.

To map the transcription start site of Promoter C0, we used RLM-5′-RACE. This approach positively selects for full-length messages by ligating an RNA linker of known sequence only to the 5′ end of mRNAs that retain their 5′ CAP. The 5′ CAP provides a unique identifier of the first nucleotide transcribed by RNA polymerase II. mRNA fragments that do not have a 5′ CAP are dephosphorylated with calf intestinal alkaline phosphatase so that they cannot participate in a subsequent ligation reaction. Treatment with tobacco acid pyrophosphatase converts the 5′ CAP, which is found only on the first nucleotide of the mRNA, into a 5′ phosphate group. Since, productive ligation of RNA linkers will only occur at these remaining phosphates, one can selectively attach the linker to mRNAs that represent full-length products. RT-PCR using linker- and gene-specific primers will selectively amplify products derived from 5′-CAP-containing mRNAs. Since, the linker is added to the mRNA, only full-length reverse transcription products are available for PCR amplification.

Using this approach, we were able to generate and clone 5′-RACE products that contained sequences derived from exon C0. Eight independently generated C0-containing clones were identified by colony hybridization using an oligo that anneals to exon C0. All eight had the same 5′ end and, by aligning the sequence of these cDNAs to the genomic sequence, we were able to map the transcription start site of Promoter C0 (identified in Figs 1, 2). Unlike Promoter C1, the transcription start site of Promoter C0 is not preceded by a recognizable TATA box. TATA-less transcriptional promoters are associated with both house-keeping genes and genes with tissue-specific expression patterns (Latchman, 1998).

The four previously mapped 5′ exons (C1, C1b, C1c and C2) directly splice to exon C3. Their splicing patterns can be summarized as C1:C3, C1b:C3, C1c:C3 and C2:C3 (Fig. 1A). One might anticipate that the new exon would also splice directly to exon C3. However, the RT-PCR-amplified cDNA fragments and the 5′-RACE products showed a novel splicing pattern: C0:C1:C3 (Fig. 4). Our surprise was compounded by the fact that this splicing pattern retains all but the first two nucleotides of exon C1. The exon C0 5′ splice site was indeed the site identified by examination of the genomic sequence. Its 3′ splice acceptor site was the third nucleotide of exon C1.

As previously noted, Fig. 3 presents the results of an RT-PCR experiment using RNA purified from embryos, larvae, pupae and adults. The amplification was performed using a primer set that would amplify only mRNAs that contain both exon C0 and C3. The 495 bp product representing the C0:C1:C3 splice variant was amplifiable from all developmental stages (Fig. 3).

To quantify the relative expression levels of Promoter C0 and C1, we performed RNAase protection assays using a probe composed of portions of exon C0 and C1. This 273-nucleotide probe contains sequences from both exon C0 and C1 and is derived from an actual Promoter C0 transcript. When used, this probe detects a transcript starting either at Promoter C0 (producing a 220-nucleotide protection product) or at Promoter C1 (generating a 196-nucleotide protection product). Fig. 5 shows that the C0 product is expressed at higher levels than the C1 product in embryos, larvae and adults. After normalizing for the number of labeled nucleotides in each protected product, we determined that, in all developmental stages, the ratio of Promoter C0 to Promoter C1 transcripts is approximately 12:1. Therefore, at a gross level, Promoter C0 is responsible for most of the expression in the adult.

We had previously shown that a reporter gene called P3 (Fig. 6) reproduces the slowpoke neuronal expression pattern but is not expressed in other tissues (Brenner et al., 1996). The P3 construct includes both Promoter C0 and C1, but it does not contain any of the other slowpoke tissue-specific promoters (Fig. 6). Therefore, the newly discovered Promoter C0 must be neuronal-specific. In the larval brain, P3 is expressed in the brain lobes, central brain, mushroom bodies and ventral nerve cord (Fig. 7A). In the adult, P3 is expressed in the optic lobes, central brain, mushroom bodies and eyes (Fig. 7B). Expression in the eye is believed to be in photoreceptor cells (Brenner et al., 1996). To help determine the relative contributions of Promoter C0 and Promoter C1 to larval and adult neuronal expression, we employed two derivatives of P3: P12 and P13. The expression patterns of transformed flies were determined by β-galactosidase staining. The transformed animals being compared were stained for the same time in the same solution so that the relative expression level could be crudely compared. P12 contains a 950 bp deletion that removes Promoter C0 but is otherwise identical to P3 (Fig. 6). In the absence of Promoter C0, P12 should report the expression pattern of Promoter C1. The P13 construct is essentially identical to P12 except that P13 has a slightly smaller deletion whose 3′ end is 121 nucleotides 5′ of the Promoter C0 transcription start site. The 5′ end of the deletion in both constructs is identical.

The P12 construct showed an expression level that was substantially reduced compared with P3, although all areas of the larval and adult brain seem to be represented (Fig. 7). Only in the adult eye does the P12 deletion cause a loss of expression. P13, which contains Promoter C0, shows essentially the same expression pattern and expression level as the P3 (wild-type) construct in both larval and adult brain (Fig. 7).

The evolutionary sequence conservation between D. melanogaster and D. hydei transcriptional control regions provides a detailed map that will speed the identification of important control elements. In this study, it helped to determine the position of a previously unmapped transcriptional promoter. This homology map will be used to guide future deletion analysis experiments aimed at identifying the control elements that regulate slowpoke expression.

The transcriptional control of the slowpoke BK type Ca²⁺-activated K⁺ channel gene is remarkably complex. We have previously mapped a muscle/tracheal cell promoter, a midgut promoter and a central nervous system (CNS)-specific promoter (Brenner et al., 1996). These were all initially identified by isolating cDNAs representing transcripts from the gene. As an alternative approach, we mapped evolutionarily conserved portions of the transcriptional control region as a means of identifying promoters. This approached helped identify the new 5′ exon called exon C0 and the promoter that produced it: Promoter C0. The remaining blocks of conserved sequence are likely to represent control elements that regulate promoter activity. Functional testing will be required to determine the purpose of the conserved blocks. Our data indicate that Promoter C0 and Promoter C1 are active in all developmental stages and are responsible for almost all expression in the CNS. Of the two, transcripts arising from Promoter C0 are approximately 12 times more abundant than transcripts produced by Promoter C1.

Why does the fly require two neuronal promoters? The first translation start site in transcripts produced by either Promoter C0 or Promoter C1 is the second codon of exon C3. That is, all of exon C0 and exon C1 represent 5′ untranslated regions (5′ UTRs) of the mRNA. Therefore, promoter choice does not affect the sequence of the encoded polypeptide. Perhaps the two neuronal promoters provide a simple way for distinct cells to express the gene at different levels and thereby to produce cell membranes with different channel densities. The control of channel density can be just as important as channel type in determining a cell’s electrical properties (Baro et al., 1997). Unfortunately, cell-to-cell differences in promoter use are not detectable in our gross histological assays. The facts that Promoter C0 is TATA-less and that Promoter C1 is preceded by a good match to a consensus TATA box support the idea that these two promoters are differentially regulated.

We were intrigued by the splicing pattern of Promoter C0 transcripts. The other four slowpoke transcripts that we have characterized all begin with an exon that is directly spliced to exon C3. However, transcripts that begin with exon C0 splice first to exon C1 and then to exon C3. The Promoter C1 transcription start site and the splice acceptor site of exon C1 are only separated by two nucleotides. The cell seems to go to great lengths to ensure that almost all the untranslated exon C1 is included. It may be that untranslated sequences in exon C1 serve some important function. The fact that untranslated exon C1 contains the largest and most strongly conserved block of homology (block 1) lends support to this interpretation. These sequences might be important for mRNA stability or translatability or, alternatively, these sequences might be involved in targeting the mRNA to a specific portion of the endoplasmic reticulum. When Promoter C0 is used to drive expression, it appends 28 nucleotides of 5′ UTR to the message. The inclusion of these sequences might modulate one of these properties in a cell-specific manner.

It is known that deletion of the C2/C3 intronic region (see Fig. 1) causes a complete loss of adult neuronal expression from slowpoke reporter genes (Brenner and Atkinson, 1996). This had previously been interpreted to mean that Promoter C1 was dependent on these sequences for activity. With the discovery that a portion of CNS expression arises from Promoter C0, this interpretation can be expanded to mean that both Promoter C0 and Promoter C1 require the presence of the C2/C3 intronic region for activity. We interpret this to mean that transcription elements within the C2/C3 intronic region act on both these promoters.

In addition, Brenner et al. (1996) determined that a 1.3 kb region, named the CNS box (Fig. 1A), was required for neuronal expression in all developmental stages. At that time, no promoters had been mapped to this area and it was therefore postulated that this region contained regulatory elements required for neuronal activation of Promoter C1. We now know that Promoter C0 maps to the 5′ half of the 1.3 kb CNS box (Fig. 1A). Brenner et al. (1996) observed that deletion of these sequences results in a complete loss of adult brain expression and an almost complete loss of larval brain expression. Does the removal of Promoter C0 alone account for this? The P12 deletion indicates that it does not. P12 removes approximately 1 kb of sequence from the 5′ side of the CNS box including Promoter C0. Even though Promoter C0 has been removed, P12 expression in the adult brain persists, albeit at a much reduced level. The simplest interpretation is that elements not removed by the P12 deletion, the 3′-most 300 bp of the CNS box (open box, Fig. 1A), contain sequences required for the normal activity of another neural promoter, presumably Promoter C1.

Which conserved sequences might represent these elements? The original 1.3 kb CNS box defined by Brenner et al. (1996) includes conserved blocks 26, 14, 5, 11 and approximately half of homology block 9. The loss of this 1.3 kb of DNA causes a loss of expression in the adult and larval brain (Brenner et al., 1996). The P12 deletion, however, removes only blocks 26, 14, 5 and nine base pairs of block 11 and does not eliminate expression. This suggests that blocks 11 and 9 are sequences required for Promoter C1 activity in the CNS.

We did not observe any reasonable similarity between the Drosophila sequence and the human slowpoke transcriptional control region recently characterized by Dhulipala and Kotlikoff (1999), who identified a single transcription start site and studied 1675 bp upstream of this site. The tissue-specificity of this promoter is not yet known. It may be that the human homolog to Promoters C0 and C1 awaits discovery or that vertebrates and invertebrates employ different mechanisms of regulation.

The remaining blocks of evolutionarily conserved sequence may also represent transcriptional control elements that modulate and direct the activity of the slowpoke promoters. While some showed similarities to known transcription factor binding sites, the most striking similarities were not to known transcription factor binding sites but to the transcriptional control regions of other genes (Table 1). These may represent undescribed transcriptional control elements. Some of these homologies were shown to exist between two different examples of the same gene. The position of these similarities is noted in Fig. 2. This type of analysis is of course speculative and should not be over-interpreted. However, it does provide an alternative method of ranking the conserved sequences for future deletion analysis studies and may be of use to others.

This research was supported by the National Science Foundation Grant IBN-9724088 to N.S.A. Accession numbers for DNA sequences are U40221 and AF210728. A portion of this work was reported at the 1998 meeting of the New York Academy of Sciences.