Enhanced expression of the eosinophil-derived neurotoxin ribonuclease (RNS2) gene requires interaction between the promoter and intron.

The eosinophil-derived neurotoxin (EDN/RNS2) is a member of the mammalian ribonuclease gene family and is one of four proteins found in the large specific granules of human eosinophilic leukocytes. The gene encoding EDN consists of two exons, including a noncoding exon 1, separated by a single intron from the coding sequence in exon 2. We have identified a functional promoter of the EDN gene and shown that optimal expression depends on interaction between the promoter and one or more sequence elements found in the single intron. Cells of the clone 15 eosinophilic variant of the human promyelocytic HL-60 cell line were transfected with constructs that included the promoter region of the EDN gene alone, promoter with exon 1, and promoter with both exon 1 and the intron positioned 5′ to the chloramphenicol acetyltransferase (CAT) reporter gene (constructs referred to as PrCAT, PrExCAT, and PrExInCAT, respectively). Although reporter gene activity from either PrCAT or PrExCAT was only 2-3 fold higher than baseline (CAT alone), inclusion of the single intron (PrExInCAT) resulted in a 28-fold increase in reporter gene activity in uninduced clone 15 cells, and an 80-fold in activity when clone 15 cells were induced to differentiate toward eosinophils with butyric acid. The intron-mediated enhancer activity was reproduced in other human hematopoietic cell lines (K562, Jurkat, U937, and HL-60), but was not found in human 293 kidney cells, suggesting that the function of the enhancer element(s) may be tissue-specific. A significant portion of the observed enhancer activity resides in the first 60 base pairs the the intron, which includes consensus binding sites for both AP-1 and NF-ATp transcription factors, and a 15-base pair segment that is identical to a sequence found in the promoter of the gene encoding the neutrophil granule protein, lactoferrin. The noncoding exon 1/single intron/coding exon 2 genomic structure is a common feature among the mammalian ribonucleases; this finding suggests the possibility of a conserved mechanism of regulation in this gene family.

The regulation of gene expression is critical to the develop-ment of cells and organisms. Although the regulation of gene expression during hematopoeisis has been under intense scrutiny, few studies have examined the eosinophil lineage directly. Whereas the regulatory regions of several eosinophil genes have been characterized (1)(2)(3)(4), the specific molecular events underlying commitment to and differentiation of the eosinophil lineage remain unknown.
The eosinophil-derived neurotoxin (EDN) 1 is a small, cationic granule protein synthesized during the promyelocyte stage of eosinophil development (5,6). The cDNA sequence and complete open reading frame identified EDN as a member of the ribonuclease gene family (7,8). The gene encoding EDN (1.2 kilobases, designated RNS2) contains two exons separated by a 230-bp intron, with the entire coding sequence residing on exon 2 (9). This gene structure (noncoding exon 1/single intron/ coding exon 2) is shared by least three additional ribonuclease genes, including eosinophil cationic protein (ECP) (9), angiogenin (10,11), and pancreatic ribonuclease (12,13), and appears to be a consistent feature of this gene family.
The studies to be described were performed by transfection of various segments of the EDN gene into clone 15 cells; clone 15 is a subline of the human promyelocytic leukemia cell line, HL-60, which was isolated on the the basis of its propensity to develop into cells resembling mature eosinophils (14). Features of this cell line include formation of Luxol-fast blue staining granules (14,15), biosynthesis of eosinophil major basic protein and eosinophil peroxidase (15), and the ability to express cell surface receptors for interleukin-5 (16,17). Most recently, we have shown that differentiated clone 15 cells also synthesize immunoreactive ECP and EDN (18).
In the work presented here, we have identified features within the EDN gene that are responsible for promoting gene expression in human cell lines, including at least one functional enhancer element within the characteristic single intron.

EXPERIMENTAL PROCEDURES
Cell Lines-The human cell lines used in this investigation were obtained from the American Type Tissue Culture Collection (Rockville, MD) and maintained as indicated. The HL-60, (promyelocytic leukemia), U937, (histocytic lymphoma), Jurkat, (acute T cell leukemia), and K562 (chronic myelogenous leukemia) were all grown in RPMI 1640 medium (Biofluids, Inc., Rockville, MD) supplemented with 10% heatinactivated fetal bovine serum (Hyclone Laboratories, Inc. Logan, Utah or Life Technologies, Inc.), 2 mM L-glutamine, 100 units/ml penicillin, and 100 g/ml streptomycin at pH 7.2. The HL-60 clone 15 promyelocytic leukemia cells were grown in the same medium supplemented * 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  with 25 mM HEPPSO (Sigma) to maintain the pH at 7.6. The human 293 cells, a transformed primary embryonal kidney cell line, were grown in Dulbecco's modified Eagle's medium (Biofluids, Inc.) supplemented as above. All cells were grown at 37°C and 5% carbon dioxide in a humidified incubator.
Cellular Differentiation-The clone 15 cells were induced to differentiate toward the eosinophil lineage with the addition of 0.5 mM butyric acid (BA) (Sigma) to freshly cultured cells at a concentration of 0.5 ϫ 10 6 cells/ml as described by Fischkoff and colleagues (16,17).
Primer Extension-Total RNA harvested from clone 15 cells induced to differentiate with BA for 48 h was used for primer extension experiments. Messenger RNA was purified by polyadenosine selection using a Poly(A) Quik Kit (Stratagene) following the manufacturer's procedure. Poly(A)-selected mRNA was then used in primer extension experiments with the Primer Extension System (Promega Corporation, Madison, WI). RNA (EDN mRNA or kanamycin-positive kit control) was reverse transcribed using avian myeloblastosis virus-reverse transcriptase and an antisense EDN-specific primer, EDN-1a (see Fig. 1B), corresponding to nucleotides 192 to 175 (5Ј-TTGCTGGGAGGT-CATATT-3Ј) of the EDN cDNA (7). EDN-1a was end-labeled with [␥-32 P]ATP (ICN, Costa Mesa, CA) and annealed to the poly(A) selected mRNA prior to reverse transcription. The primer extension reaction was electrophoresed in a denaturing 6% TBE polyacrylamide gel along with known DNA sequence for size determination. The number of nucleotides from the primer to the 5Ј end of the mRNA was determined.
Rapid Amplification of cDNA Ends-To determine the precise transcriptional start site, rapid amplification of cDNA ends (RACE) was performed using mRNA described above with the 5Ј-AmpliFINDER RACE Kit (Clontech Laboratories, Inc., Palo Alto, CA) following the manufacturer's protocol. The cDNA was first synthesized from mRNA using avian myeloblastosis virus-reverse transcriptase and antisense EDN-specific primers. These primers corresponded to nucleotides 291-274, 5Ј-ACAAACATTAACTACGTT (EDN-2a) of the EDN probe, nucleotides 341-296, described above, of the EDN cDNA sequence (7). An anchor of known sequence was ligated to the 5Ј end of the cDNA. Polymerase chain reaction (PCR) amplification (described below) of the anchor ligated cDNA was performed using the supplied anchor primer and a nested primer for each of the antisense EDN primers used to synthesize the cDNA. The nested primer used with the antisense oligonucleotide EDN was EDN-2a, and for EDN-2a the nested primer was EDN-1a, as shown in Fig. 1B. The PCR were performed in a Perkin-Elmer Gene Amp PCR System 9600 using Taq DNA polymerase (5 units) (Boehringer Mannheim) in 10 mM Tris, pH 8.3, buffer containing 50 mM KCl, 2.5 mM MgCl 2 , 2 uM each dATP, dCTP, dGTP, and dTTP, 100 ng of each oligonucleotide primer, and 10 -50 ng of DNA template. The DNA was denatured for 3 min at 95°C, followed by a three-step PCR reaction: denaturation at 95°C for 30 s, annealing at an appropriate temperature for each primer pair for 30 s, and extension at 72°C for 1 min. After 35 cycles the reactions were extended for an additional 5 min at 72°C followed by cooling to 4°C. PCR products were subcloned into pBluescript (Invitrogen, San Diego, CA) for dideoxy sequence analysis.
Isolation of Genomic Clones-EDN genomic clones were isolated from a human lymphocyte genomic library (Clontech Laboratories, Inc.) by plaque hybridization with an EDN cDNA radiolabeled probe (7) and confirmed by dot blotting with the EDN-1a oligonucleotide (see Fig. 1B) and by dideoxy sequencing. The nucleotide sequence obtained agreed with previously published data (9).
Electroporation of CAT Constructs into Human Cell Lines-Cell lines as described above were grown to a density of 0.5 to 1.0 ϫ 10 6 cells/ml. Cells were harvested by centrifugation at 2000 rpm for 5 min and resuspended to 30 ϫ 10 6 cells/ml in complete medium as described above for each cell line. Ten to twenty micrograms uncut purified plasmid DNA (using either Qiagen or Promega Maxiprep kit followed by phenol/chloroform extraction) were added to an electroporation cuvette with a 0.4 cm electrode gap (Bio-Rad) with 15 ϫ 10 6 cells. The cells were electroporated at a capacitance of 960 microfarads at 250 volts using a Gene Pulser (Bio-Rad). The cells were placed on ice for 15 min prior to the addition to 30 ml of complete medium and incubated for 2 days at 37°C, 5% CO 2 in a humidified incubator. Cells were also electroporated with either 10 g of uncut control pSV-␤-galactosidase (Promega) or pCMV-␤-galactosidase (Stratagene) vector to control for variation in electroporation efficiency. The level of ␤-galactosidase activity was determined spectrophotometrically (␤-galactosidase assay kit, Promega). CAT activity was normalized to the level of expression of this control vector.
CAT Assays-CAT assays were performed essentially as described (22). Transfected cells were harvested after 2 days of incubation by centrifugation at 2000 rpm for 5 min, washed twice at room temperature with phosphate-buffered saline, and resuspended in 0.5 ml of 40 mM Tris, pH 7.4, with 1 mM EDTA and 150 mM NaCl. Cells were then harvested by a 1-min centrifugation in a microfuge and resuspended in 100 -150 l of 25 mM Tris, pH 8, and disrupted by freeze/thaw, three times on dry ice/37°C water bath. The debris was removed by centrifugation; clarified extract was added to the following reaction mixture Analysis of Intronic Region-Computer analysis of the intronic regions of the EDN gene for consensus transcription factor binding sites was performed using the MacVector sequence analysis program. Com-parison of intronic regions with known sequences was performed using the FASTA algorithm of the Wisconsin Genetics Computer Group program on-line at the National Institutes of Health.

RNA Expression in the Clone 15
Cell Line-Expression of mRNA for the EDN in the clone 15 cell line is demonstrated in Fig. 1A. Messenger RNA encoding EDN was detected in uninduced clone 15 cells (lanes 1-7); the level of mRNA accumulating in these cells increased 6 -7-fold within 48 h after the addition of 0.5 mM BA (lane 9) as determined by densitometric analysis. Hybridization with a ␤-actin probe (below) demonstrated relative loading of each lane.
Primer Extension-Primer extension experiments were undertaken to determine the precise location of the transcriptional start site. RNA from clone 15 cells induced for 2 days with 0.5 mM BA was used for this analysis; a control RNA reaction was included, as was a known DNA sequence for determination of molecular size (Fig. 1C). The band in the ϩEDN mRNA lane indicated by the arrowhead is the reversetranscribed product. Comparison of the nucleotides in the known sequence (Std) with the mobility of the band indicated in the ϩEDN mRNA lane demonstrates that the distance from the primer to the 5Ј end of EDN is 219 nucleotides. This result suggests that there are 25 additional nucleotides in the mRNA sequence that had not previously been identified by cDNA cloning (7,8).
Rapid Amplification of cDNA Ends-It was unclear as to whether these additional 25 nucleotides represented a direct extension of exon 1 or a completely distinct exon; RACE was undertaken to make this determination. Using the poly(A)selected mRNA described for the primer extension experiments, a cDNA strand complementary to the 5Ј region of the EDN mRNA was reverse transcribed using the EDN or EDN-2a primers (Fig. 1B), PCR amplified, and subcloned. Numerous colonies were obtained; 18 random colonies were selected for sequencing. The results from a representative sequence are depicted in Fig. 1D. The additional 25-base pair sequence obtained, which follows the anchor sequence and precedes the first exon sequence, is 5Ј-AGCTGCCCCTGAAC-CCCAGAACAAC-3Ј. This sequence matches the published genomic sequence of Hamann and colleagues (9) that is found immediately 5Ј to the previously designated exon 1. These results suggest that the additional nucleotides represent a direct extension of exon 1 as opposed to another distinct exon. The transcriptional start site defined by both primer extension and RACE is located 25 nucleotides 3Ј from a consensus TATA box (9). Of the 18 clones, 11 sequences were as described in Fig.  1D, two were 4 nucleotides shorter at the 5Ј end, and five contained unidentified sequences between the PCR primers. The five unidentified clones are assumed to be the result of incorrect priming, because they contained no sequences representing either exon 1 or exon 2.
CAT Activity in Clone 15 Cells-The sequence of the EDN gene positioned 5Ј to CAT used in the following experiments are shown in Fig. 2A. The EDN promoter region alone (PrCAT) supported a relatively low level of activity in clone 15 cells, representing 2-3-fold over that of the promotorless CAT (Fig.  2B). Surprisingly, the EDN promoter-exon-intron construct (PrExInCAT) supported significantly greater degree of reporter gene activity: 28-fold over CAT in uninduced clone 15 cells and 80-fold over CAT in clone 15 cells grown in the presence of BA (Fig. 2, B and C). The EDN promoter-exon (PrExCAT) construct produced low levels of CAT activity, similar to those of PrCAT alone, suggesting that the dramatic enhancement seen with PrExInCAT was due to sequence elements found in the intron. The intron alone (InCAT) did not demonstrate reporter gene activity above the basal levels in these cells, whereas the EDN promoter-intron combination (PrInCAT) without the first exon provided a 12-fold increase in activity over CAT alone. The increase in activity with promoter and intron and the absence of activity with the intron alone suggests that the enhanced activity depends on interactions between sequence elements in both the promoter and intron. Fig. 3. PrExInCAT was capable of promoting reporter gene activity 20 -40-fold over that observed with the CAT in the K562, Jurkat, U-937, and HL-60 human hematopoietic cell lines (Fig. 3). The EDN promoter alone (PrCAT) supported only the lower, basal levels of expression in all cell lines tested. To evaluate regulation in a nonhematopoietic cell line, the PrCAT and PrExInCAT constructs were introduced into the human transformed 293 kidney cell line. Similar to the observations made with the hematopoietic cells, PrCAT induced a basal (2-3-fold) increase in reporter gene activity over CAT in the human kidney cell line. Interestingly, in contrast to the 20 -40-fold increase in activity observed in hematopoietic cells, PrExInCAT induced only a 3-fold increase in the kidney cells. This activity was not significantly greater than that induced by PrCAT, suggesting that the enhancer activity provided by the intron may function in a tissue-specific fashion.

Expression of EDN-CAT Constructs in Other Cell Lines-Analysis of CAT activity in several other cell lines is shown in
Dissection and Analysis of the Intron-In order to define the region or regions of the intron that enhance gene expression, truncated promoter-intron constructs were evaluated in clone 15 cells (Fig. 4). The PrEx0.25InCAT and PrEx0.5InCAT constructs, which contain the first 60 and first 120 base pairs of the intron, respectively, were both found to increase reporter gene activity 8-fold over CAT alone. These results suggest that there are one or more functional enhancer elements in the first 60 base pairs of the intron. In contrast, the PrEx0.75InCAT construct, containing the first 180 bp of the intron, produces only a 2.5-fold increase in activity over CAT, suggesting that this region (120 and 180 bp) contains sequence elements that neutralize the effects of the aforementioned functional enhancer(s). Full activity (28-fold over CAT) is restored with the complete intron (PrExInCAT), suggesting the possibility of additional functional enhancers in this final segment of the intron. A sequence map of the EDN gene with a focus on the intron is shown in Fig. 5A. The first 60 bp of the intron contain consensus binding sites for both AP-1 and NF-ATp transcription factors. In addition, there is a segment of 15 bp that is identical to a segment found in the promoter of the lactoferrin gene (see "Discussion") (24). Point mutations were introduced into the sequence of this segment (Fig. 5B), and the PrExIn-CAT constructs both with and without mutations were evaluated for their ability to support reporter gene activity. Neither set of mutations (lfn X or lfn Y; Fig. 5, B and C) altered the intron-enhancing activity to any significant degree (21-and 35-fold over CAT alone, as compared with 28-fold for the wild type). DISCUSSION In the initial phase of this study, we determined the transcriptional start site of the EDN gene. Both primer extension and RACE extended the length of the mRNA to include an additional 25 nucleotides. This additional sequence matched the region of genomic sequence that was directly 5Ј to sequence identified as exon 1 (9). Thus, exon 1 has been enlarged to 67 nucleotides, and the two-exon structure of this gene is confirmed. The transcriptional start site is situated appropriately at 23 and 101 base pairs 3Ј to consensus TATA and CAAT boxes, respectively.
Although significantly more active than either PrCAT or PrExCAT, we found that PrInCAT was not as effective as PrExInCAT in producing reporter gene activity. It is possible that the noncoding exon 1, in conjunction with the intron, plays a specific role in regulating activity of the EDN gene. However, it is also possible that the spatial rearrangement caused by deletion of this 67-base pair segment affects the ability of promoter and intron binding proteins to interact with one another as they interact with the DNA. This point may be clarified once the binding proteins mediating transcription of the EDN gene have been identified. Furthermore, it is not clear whether the enhancer elements present in the EDN intron coordinate specifically with the EDN promoter, or whether they might function equally effectively with other unrelated gene promoters. Although not completely independent (InCAT produced little reporter gene activity in clone 15 cells), the specificity of the interaction between the intron and the EDN promoter has not been established.
Investigators have been examining the promoter regions of granule protein genes (1-4, 24, 25) in hopes of identifying factors promoting their tissue-specific expression. Our initial results suggest that the function of one or more of the intronic enhancer elements may be tissue-specific. In contrast to the results obtained with the human hematopoietic cell lines, the PrExInCAT construct supported no additional reporter gene activity over that of the promoter alone (PrCAT) in the human kidney cell line. Although preliminary, the increase in activity of PrExInCAT in the hematopoietic cells but not in the kidney cell line suggests that one or more of the functional enhancer elements may interact with a transcription factor expressed in a limited range of human cell types.
Sequence analysis of the EDN intron revealed the presence of several consensus sequences for transcription factor binding (Fig. 5). In the first 60 bp of the intron, the region in which a large portion of the enhancing activity resides, there are consensus sequences representing binding sites for transcription factors AP-1 (26) and NF-ATp (27). In addition, a 15-base pair segment that is identical to a sequence found in the 5Ј promoter region of the neutrophil-specific granule protein, lactoferrin (24), is highlighted. The coordinate control of eosinophil and neutrophil granule protein biosynthesis was suggested by a series of studies examining the rare genetic disorder known as neutrophil specific granule deficiency (28,29). Lomax and colleagues (30) determined that the defective biosynthesis of lactoferrin observed in this disorder resulted from a defect in mRNA transcription affecting specifically cells of the neutrophil lineage. Rosenberg and Gallin (31) showed that the biosynthetic defect could extend to include eosinophils; EDN was among the eosinophil granule proteins affected. Specific mutations introduced into this "lactoferrin site" resulted in no significant alteration in reporter gene activity, suggesting that this segment is not crucial to this specific aspect of EDN gene regulation. Additional investigations using electrophoretic mobility shift assays coupled with DNA footprinting and mutational analysis of additional sites will be necessary to determine the source of the enhancing activity.
In summary, this investigation has demonstrated that the intron of the EDN gene contains one or more functional en- FIG. 5. A, nucleotide sequence of the 5Ј promoter (Ϫ287 to Ϫ1), exon 1 (1-67), intron (68 -297) and the beginning of exon 2 (298 onward). Consensus CAAT and TATA box promoter sequences are as indicated. Consensus transcription factor binding sites (AP-1, NF-ATp, and PU.1) are single overlined; the arrowheads denote the division points for PrExIn0.25CAT, PrEx0.5InCAT, and PrEx0.75InCAT, respectively. The 15-bp segment that is double overlined is identical to a segment found in the 5Ј promoter region of the neutrophil specific granule protein, lactoferrin (24). B, mutations introduced into the 15-bp lactoferrin segment in the intron to create lfn X and lfn Y in PrExIn. Points at which lfn X and lfn Y differ from the wild type are indicated over each sequence with filled circles. C, reporter gene activity of each construct transfected into clone 15 cells, relative to that of CAT (no promoter). The bars represent densitometric analysis of duplicate samples. hancer elements and that interaction between the promoter and these intronic elements is required for the optimal gene expression. Although intronic enhancer elements have been described previously (32)(33)(34)(35)(36)(37)(38)(39), this is the first evidence that this mechanism exists within the mammalian ribonuclease gene family. This finding takes on considerable significance because the noncoding exon/single intron/coding exon gene structure is shared by all members of this gene family whose gene structures have been determined (9 -13). Although the nucleotide sequences of introns of each of the characterized ribonuclease genes are not overtly homologous to one another (save for EDN and ECP, whose introns are virtually identical (9)), the potential for this as a shared mechanism of gene expression is intriguing.