Tandemly reiterated negative enhancer-like elements regulate transcription of a human gene for the large subunit of calcium-dependent protease.

Calcium-dependent protease (CANP, Calpain) is an intracellular protease involved in essential cellular functions mediated by calcium. To understand the mechanism regulating the expression of CANP at the transcriptional level, we isolated a human gene for the large subunit of mCANP (CANP mL) and analyzed its 5'-region. The transcription initiation sites were mapped to multiple positions (-142 to -103, A of initiation ATG as +1). The upstream region lacks typical promoter elements such as TATA and CAAT boxes and is characterized by its high GC content (-300 to -20, 70% GC content). Functional analyses of the 5'-region by a transient expression assay on HeLa cells revealed that the region (-202 to -80) has a promoter activity. The upstream half of the promoter region (-202 to -130) acts as an upstream promoter element in an orientation-independent manner. Upstream of the promoter region are tandemly reiterated multiple regulatory regions (-2.5k to -690, -690 to -460, -460 to -260, and -260 to -202), each of which negatively regulates the CANP mL gene promoter as well as heterologous promoters in an orientation-independent manner. The presence of a cellular factor(s) mediating the action of these positive (promoter) and negative regulatory elements was suggested by an in vivo competition assay. The negative regulation of transcription mediated by these reiterated cis-acting elements and trans-acting factor(s) may play an essential role in the expression of the CANP mL gene.

Calcium-dependent protease (CANP, Calpain) is an intracellular protease involved in essential cellular functions mediated by calcium. To understand the mechanism regulating the expression of CANP at the transcriptional level, we isolated a human gene for the large subunit of mCANP (CANP mL) and analyzed its 5"region. The transcription initiation sites were mapped to multiple positions (-142 to -103, A of initiation ATG as +l). The upstream region lacks typical promoter elements such as TATA and CAAT boxes and is characterized by its high GC content (-300 to -20, 70% GC content). Functional analyses of the 5"region by a transient expression assay on HeLa cells revealed that the region (-202 to -80) has a promoter activity. The upstream half of the promoter region (-202 to -130) acts as an upstream promoter element in an orientation-independent manner. Upstream of the promoter region are tandemly reiterated multiple regulatory regions (-2.5k to -690, -690 to -460, -460 to -260, and -260 to -202), each of which negatively regulates the CANP mL gene promoter as well as heterologous promoters in an orientation-independent manner. The presence of a cellular factor(s) mediating the action of these positive (promoter) and negative regulatory elements was suggested by an in vivo competition assay. The negative regulation of transcription mediated by these reiterated cis-acting elements and trans-acting factor(s) may play an essential role in the expression of the CANP mL gene.
Calcium ions regulate various cellular functions as messengers of extracellular stimuli through interaction with various calcium-binding proteins. Calcium-dependent protease (CANP,Calpain,EC 3.4.22.17), an intracellular protease requiring calcium for catalytic activity, is one such calcium-binding protein (1). CANP hydrolyzes proteins of limited classes in uitro, including epi-* This investigation has been supported in part by research grants from the Ministry of Education, Science and Culture and the Ministry of Health and Welfare, Japan. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper hos been submitted to the GenBankTM/EMBL Data Bank with accession numbeds) 504700.
$ To whom correspondence should be addressed. The abbreviations used are: CANP, calcium-activated neutral protease; bp, base pair(s); Pipes, 1,4-piperazinediethanesulfonic acid; CAT, chloramphenicol transferase; Hepes, 4-(2-hydroxyethyl)-l-piperazineethanesulfonic acid; IFN, interferon. dermal growth factor receptor, platelet-derived growth factor receptor, and protein kinase C, suggesting that CANP is involved in cellular functions through down-regulation or activation of these specific substrates. Ubiquitous distribution of CANP among tissues of higher animals and conservation of the primary sequence among mammalian species support the idea that CANP is involved in essential cellular functions (1).
At least two isozymes (pCANP and mCANP) with different calcium requirements exist. Both are heterodimers composed of L (large, catalytic, 80 kDa) and S (small, regulatory, 30 kDa) subunits (1). They share identical S subunits, and thus differences in their various properties arise from the L subunits (pL and mL) (2). Recent molecular cloning experiments have revealed primary structures of three human CANP proteins (pL, mL, and s) (3-6). They are encoded by distinct genes on distinct human chromosomes,2 although sequence homologies demonstrate their evolutionary relationship (6). Knowledge of the mechanism underlying the regulation of genes for ubiquitously expressed intracellular proteins (housekeeping proteins) is quite limited compared with that of genes which are expressed in a tissue-specific manner. One of the reasons is that these genes are usually expressed at a low level making it difficult to isolate cDNA clones and analyze the gene expression.
In this paper, we describe analyses of the promoter region of a human CANP gene (CANP mL) and reveal the presence of a tandemly reiterated array of cis-acting negative regulatory elements upstream of positive regulatory (promoter) elements.

EXPERIMENTAL PROCEDURES
Screening of the Library-Charon 4A human genomic library (provided by Dr. Y. Sakaki) was constructed essentially as described (7) using high molecular weight DNA from human lymph node partially digested with Hue111 and AluI. The hybridization probe used for the screening was a 0.35-kilobase ApaI fragment (probe Al) of the human CANP mL cDNA (p21-5), corresponding to the most N-terminal355 bp (4). The cDNA was labeled with [a-3ZP]dCTP (-3000 Ci/mmol, Du Pont-New England Nuclear) by a multiprime DNA labeling system (Amersham Corp.). Hybridization was carried out according to the standard method (7).
DNA Sequencing-The insert of the phage clone was further dissected to generate smaller overlapping subclones in the plasmid Bluescript KS or pUC18. DNA sequencing was performed by the dideoxy chain termination method on denatured plasmid DNA (8).
at 37 'C in 5% COz. DNA transfections were done essentially as described (15) with some modifications. Approximately lo6 cells were plated onto 10-cm tissue culture dishes 24 h before transfection with a change of medium 4 h before addition of DNA precipitate. A DNA-CaClZ mixture (15 pg of plasmid DNA in 0.5 ml of 240 mM CaC12) was added dropwise to 0.51 ml of a transfection mixture (4 mM Hepes, 269 mM NaC1, 1.4 mM Na2HP04, 1.4 mM NaHzP04 adjusted to pH 6.95) with vortexing. After 30 min at room temperature, the precipitates were applied to the cells and incubated at 37 "C. Four hours later the cells were treated with a glycerol solution (15% (v/v) glycerol in 8 mM NazHP04, 1.5 mM KHzPO4 adjusted to pH 7.4) for 1 min and incubated for 44 h at 37 "C. Each transfection experiment was repeated at least four times with two different plasmid preparations. To obtain reproducible results, all plasmids were prepared by alkaline lysis and banded by ultracentrifugation on a cesium chloride-ethidium bromide equilibrium gradient.
Assay for CATActiuity-Cells incubated for 48 h after transfection were harvested and disrupted by freeze-thawing. Protein concentrations of extracts were determined with a protein assay kit (Bio-Rad). The CAT activity was measured essentially by the method previously described (16). In order to confirm the reproducibility of transfection efficiency, we cotransfected the plasmid pCHllO coding for #?-galactosidase as an internal reference plasmid and determined #?-galactosidase activity of each sample. #?-Galactosidase activity was determined spectrophotometrically at 420 nm using o-nitrophenyl-#?-Dgalactoside (Wako Purechemicals) as a substrate (17). CAT assay was performed on protein equivalents, because protein concentrations changed parallel to the #?-galactosidase activity. Cell extract equivalent to 15-90 pg of protein was incubated for 30 min at 37 "C in an assay mixture (final volume 150 pl) containing 0.2 pci of D-threo-[dichloroacetyl-l-14C]chloramphenicol (54 mCi/mmol, Amersham Corp.), 0.5 mM acetyl coenzyme A (Sigma), and 0.47 M Tris-HC1 (pH 7.8). The reactions were terminated by extraction with 1 ml of ethyl acetate. The extracted chloramphenicol was dried and dissolved in 10 p1 of ethyl acetate. The acetylated chloramphenicol was separated from the nonacetylated form by thin-layer chromatography (HPTLC Silica Gel 60, Merck) developed in ch1oroform:methanol (95:5, v/v).
After autoradiography, radioactive spots were quantitated by a scintillation counter to determine the acetylated forms. The CAT activity of each construct was determined by at least four independent transfection experiments, and the average value represented percent acetylation (W acetylation = 100 X counts in acetylated products (cpm)/total counts of acetylated and unreacted chloramphenicol In the competition assay, cotransfection of the internal reference plasmid pCHllO was used to normalize a transfection efficiency of each sample. CAT activity was determined by an additional method (18). Cell extract (50 pl in 100 mM Tris-HC1, pH 7.8) was heated to 70 "c for 15 min, and 200 p1 of CAT mixture (1.25 mM chloramphenicol, 100 mM Tris-HC1, pH 7.8) and 10 pl of [butyryl-l-"C]butyryl coenzyme A (4.0 mCi/mmol, Du Pont-New England Nuclear) were added. After 90 min of incubation at 37 "C, butyrylated chloramphenicol was extracted with 5 ml of liquid scintillator Econofluor (Du Pont), and the CAT activity was directly quantitated by a scintillation count of the butyrylated products. The latter CAT assay method was faster and its sensitivity was almost the same as the former method.
Isolation and Sequencing of the 5"Region of the Human CANP mL Gene and Determination of the Transcription Znitiation Site-Using a cDNA fragment of human CANP mL (4) containing a 144-bp 5"flanking and a 211-bp proteincoding sequence as a probe, a genomic clone, M1-3, was isolated from a human genomic library. Restriction mapping and Southern hybridization analysis of M1-3 using the cDNA fragment as a probe identified an exon-containing region, and the nucleotide sequence of this region with its 5"flanking sequence (1154 bp in total) was determined (Fig. 2). The cDNA sequence from -131 to +236 (A of the initiation codon as +1) exists in this region in a complete match, identifying the location of the putative first exon. The presence of an intron at +236 was also shown for the chicken CANP L gene (19).
An S1 nuclease mapping experiment using a genomic DNA fragment (-460 to -20) 5'-end-labeled at position -20 identified multiple signals (between positions -137 and -103) corresponding to the 5'-end of the exon ( Fig. 3A and 2B, closed triangles). The fact that the cDNA starts at around -130 and that no splicing acceptor sequence (AG) is seen in the near upstream region of the 5'-ends indicates that the exon is actually the first exon and that transcription starts from multiple sites. This is supported by functional analysis showing the presence of promoter activity just upstream of the cluster of the putative transcription initiation sites.
The upstream region of the transcription initiation sites does not contain a TATA-like sequence and is extremely rich in GC residues (-300 to -20, 70% GC content). These structural features may explain the presence of multiple transcription initiation sites.
Deletion Analysis of the Promoter Region-To identify DNA sequences involved in expression of the human CANP mL gene, a series of 5'-and 3"deletion fragments of the 5"region of the CANP gene was ligated to the 5"upstream region of the bacterial CAT gene in a CAT expression plasmid, pKSCAT (Fig. 1). These CAT constructs were tested for promoter activity by transfection into HeLa cells. Fig. 4 shows structures of various deletion mutants and their relative promoter activities determined by the transient expression assay. The longest construct, p-2.5k/-20CAT, showed relatively low promoter activity. However, as the deletion proceeds to -202 (p-460/-20CAT, p-260/-20CAT, and p-202/-20CAT), the levels of CAT expression gradually increased. Complete removal of a -2.5 kilobase pair to -202 region (p-202/-20CAT) resulted in a 13-fold increase in CAT expression. This indicates that a -2.5 kilobase pair to -202 region negatively regulates the promoter activity of the downstream region (-202 to -20).
Primer extension analysis of the CAT poly(A)+ RNA obtained from cells transfected with p-202/-20CAT demonstrated that the 5'-end of the CAT mRNA locates -142 and -141 (Fig, 3B and Fig. 2B, open triangles). These positions are essentially the same as those determined for the wild type

RNA. B, primer extension analysis of the 5'-end of the CANP-CAT fusion transcript isolated from HeLa cells transfected with the CANP-CAT fusion gene (p-202/-20CAT). Poly(A)+ RNA (5 pg) from HeLa cells (24 h after transfection) was annealed using a synthetic oligonucleotide complementary to the 5'-end of the CAT gene as a primer. Lane 1 is the primer extension products. Lanes G, A, T, and C are the sequencing ladders of the nCANP gene. Protected bands are indicated by arrows.
CANP gene using S1 nuclease (Fig. 2B). Slight differences in these results may reflect a difference in downstream sequences or detection methods.
Deletion of -202 to -160 resulted in a drastic decrease in CAT expression (Fig. 4, A and B, closed circles) indicating that this region is involved in promoter function. The 3'deletion of -80 to -20 (p-202/-80CAT) (Fig. 4, A and B, open circles) indicates that this region is not required for promoter function. However, removal of -130 to -80 (p-202/-130CAT) resulted in a 70% decrease in promoter activity indicating that this region is required for full promoter activity. These results demonstrate that the -202 to -80 region is a promoter region for the CANP mL gene.

6407
Functional Dissection of the Promoter Region-To further characterize the promoter region (-202 to -80), various chimeric promoters were constructed, and their activities were analyzed by means of CAT expression (Fig. 5). When fused upstream of the TK promoter (-105 to +56) which contains both upstream (CAAT box) and downstream (TATA box) promoter elements (pTKCAT), the CANP promoter region (-202 to -80) did not produce a significant increase in the TK promoter activity (p-202/-80TKCAT) (Fig. 5A). This was confirmed by another CAT assay using a smaller amount of cell extract (equivalent to 54 pg of protein). The relative CAT activity of the CANP promoter to the TK promoter was 1.14 (71.2 and 62.5% acetylation for p-202/-80TKCAT and pTKCAT, respectively). However, when fused upstream of the IFN promoter sequence (-55 to +19), which contains the TATA box and the transcription initiation site (pIFNCAT), the CANP promoter region (-202 to -80) showed a marked increase in CAT expression (p-202/-80IFNCAT) (Fig. 5A). A similar effect was also shown using half of the promoter region, P2, in an orientation-independent manner (p-202/-130IFNCAT, p-l30/-202IFNCAT). This indicates that the region from -202 to -130 (P2) contains upstream promoter element(s), whereas the region from -130 to -80 (Pl) contains downstream element(s).
Negative Regulatory Elements Act on Heterologous Promoters in an Orientation-independent Manner-Analysis of the 5'-deletion mutants of the CANP-CAT fusion constructs shown in Fig. 4 suggested the presence of a control element(s1 which negatively regulates the downstream promoter (P1 + P2). We next examined whether the upstream sequence acts on heterologous promoters. As shown in Fig. 5B, when inserted upstream of the TK promoter, each of the upstream regions (N4, N3, and N1) repressed TK promoter activity (p-2.5k/-690TKCAT, p-690/-460TKCAT, and p-260/-202TKCAT). Interestingly, each of the upstream regions acts in an orientation-independent manner (p-690/-2.5kTKCAT, p-460/-690TKCAT, and ~-202/-260TKCAT). The degree of repression, however, depends on the region inserted. Further, N3 also repressed the activity of the SV40 early promoter/enhancer (~-690/-460SVCAT, p-460/-690SVCAT) in an orientation-independent manner (Fig. 5C). These results clearly indicate that the CANP gene contains at least three (N4, N3, and N1) negative regulatory elements which independently repress promoters of different origins in an orientation-independent manner. Moreover, N2 was also identified as a negative element as described below.
Competition of Trans-acting Cellular Factors Thut Bind to CANP Gene-As a step toward identifying cellular factors that mediate the effects of the promoter and negative regulatory elements on transcription of the CANP gene, in vivo competition experiments were carried out. A constant amount of a test plasmid containing positive or negative elements (pT-1, p-260/-20CAT; pT-2, p-202/-20CAT) fused upstream of the CAT gene was cotransfected into HeLa cells with increasing amounts of competitor DNA (PC-1-5) (Fig. 6A). Plasmid vector Bluescript KS lacking both the CAT gene and the CANP gene was used to normalize the amount of DNA transfected into the cells. If cellular factors mediate the function of a cis-acting element, CAT expression of the test plasmid would be increased or decreased in the presence of competitor plasmid.
First we examined whether additional copies of the CANP positively controlled region decreased its own activity. When the test plasmid pT-2 (p-202/-20CAT), containing the positive element of the CANP gene, was transfected into cells with competitor plasmid PC-5, a decrease in CAT activity with an increasing amount of competitor was observed (Fig.   6B). At a 17-fold excess of competing positive element, the level of CAT activity decreased to 30% of the initial level. Control experiments using the plasmid containing the upstream fragment (-260 to -202) caused no decrease in CAT expression (Fig. 6D, open bars). These results strongly suggest the presence of a limiting amount of cellular factor(s) interacting directly or indirectly with the promoter element, P2, and that the factor(s) is essential for promoter activity.
We next examined the effect of cellular factor(s) acting on the negative elements N4 to N1. The test plasmid pT-1 (p-260/-20CAT), containing the promoter region and negative element (Nl), was transfected into cells with competitor plasmid PC-4. Increasing amounts of PC-4 resulted in an increase in CAT activity (Fig. 6C). At a 17-fold excess of competitor plasmid, the level of CAT activity of pT-1 increased to 440% of the initial level. These results suggest that the binding of one or more cellular factors present in a limiting amount in HeLa cells is essential for the function of the negative element (Nl) of the CANP gene.
To examine whether the trans-acting factor(s) which interacts with the negative element (Nl, -260 to -202) is common to other negative elements (N4, N3, and N2) located upstream of the positive element (P1 + P2), we transfected the test plasmid pT-1 with each of the competitor plasmids PC-1, -2, -3, or -5. Surprisingly, all the competitor plasmids containing N2, N3, and N4 increased the CAT activity to 400% of the initial level, equivalent to that observed with N1 (Fig. 6D,  striped bars). On the other hand, PC-5, which contains the P2 region, did not increase the CAT activity. These results suggest that all four negative elements recognize the same or similar cellular factor(s).

DISCUSSION
In this study, we isolated and characterized the upstream region of the human CANP mL gene. Functional analysis of the CANP mL gene upstream region by means of a transient expression assay on HeLa cells using CAT constructs identified four negative regulatory regions (N4, N3, N2, and N1) tandemly reiterated just upstream of the promoter region (P2 and Pl).
The promoter region (P2 + P1; P2, -202 to -130; P1, -130 to -80) is extremely rich in GC residues, lacks a TATA box, and contains multiple transcription initiation sites which cluster between -142 and -103. These are structural features common to many genes of housekeeping proteins such as hypoxanthine phosphoribosyltransferase (20,21), adenosine deaminase (22), and hydroxy-3-methylglutaryl coenzyme A reductase (23). The gene for the small subunit of CANP (CANP S) also shares similar structural characteristics in its 5"flanking region (24). Although the CANP promoter region does not contain typical promoter motifs such as TATA and CAAT boxes, it possesses a promoter activity whose strength is comparable to that of the TK gene. The P2 region alone exerts full promoter activity when fused upstream of the TATA box of IFN promoter in either orientation (p-202/-130IFNCAT, p-l30/-202IFNCAT), although the P2 region by itself is not sufficient for full promoter activity (p-202/-130CAT). Thus, P2 seems to correspond functionally to an orientation-independent upstream promoter element found in several genes (25). In fact, P2 contains a sequence known to act as an upstream promoter element in, for example, the 8globin gene (CCACACCCCG, starting at -166) (26) and consensus sequence for Spl recognition (GGGGCCGGGC, starting at -155) (27). Competition experiments indicate that the promoter activity of P2 is mediated by cellular factor(s) specifically interacting with the P2 sequence. Cellular factors such as Spl and a protein in HeLa cells which interacts with the P-globin upstream element are candidates for mediating the promoter function of the P2 region. It should be noted that, whatever the factors are, the amount of the molecule regulating the promoter function of P2 is limited in cells.
In contrast to the P2 region, the P1 region of the CANP mL promoter may correspond functionally to the downstream promoter element which defines the direction and location of transcription initiation. The IFN gene promoter containing the TATA box can be substituted for the P1 region without any reduction in promoter activity (p-202 f -80CAT versus p-202/-130IFNCAT). A characteristic sequence in the P1 region, a direct repeat of a 15-bp sequence CGCT/CCGCAGC/ TGGCG/CG, may be involved in the function of PI. The presence of a sequence conserved in the AP-1 binding sites in the Pl-P2 junction of the CANP mL gene (TGAATCA, starting at -132) suggests the involvement of AP-1 in the regulation of CANP mL gene transcription. A potential AP-1 binding sequence (TGAGTCA, starting at -108) is also seen in the corresponding region of the CANP S gene. Since CANP irreversibly activates protein kinase C (l), it would be quite interesting to examine whether the CANP mL gene (and S gene) responds to 12-0-tetradecanoylphorbol-13-acetate via a protein kinase C-mediated process. One of the most intriguing results obtained from the functional analysis of the CANP gene is the elucidation of multiple negative regulatory elements. Serial deletions of these negative regulatory regions (N4, N3, N2, and N1) result in a gradual increase in the promoter activity (Fig. 4B). At least three of these negative elements (Nl, N3, and N4) also repress heterologous promoter activity, such as the TK gene, when inserted just upstream of the promoter in either orientation. Because the negative action of N1 disappears in the presence of an excess amount of these negative regions, the negative action of N1 is mediated by cellular factor(s), and all four negative elements recognize the same or very similar factor(s). The presence of a common sequence GGC/GCCGTC/G in regions N3, N2, and N1 (the complete sequence of N4 has not yet been determined) may explain the notion that the cellular trans-acting factor(s) interacting with these negative elements are identical or similar. The presence of cis-acting elements which negatively regulate promoter activity has been reported for several genes including rat a-fetoprotein (28), rat insulin 1 (29), rat growth hormone (30), human IFN-0 (31), and human apolipoprotein CIII gene (32), although their precise nature, including the presence of a trans-acting factor, is unknown in most cases. The conserved sequence in some of these negative regulatory elements for the human IFN-0 and apolipoprotein CIII genes appears to be absent in the CANP gene negative regulatory regions.
Tandem reiteration of positive (or inducible) regulatory elements is seen in several eukaryotic genes and thought to be a rather general feature of eukaryotic gene regulation. For example, a 12-base pair sequence element of the human metallothionein gene (33) is sufficient to confer metal inducibility upon a heterologous promoter, and the level of induction by heavy metals depends on the number of copies present. It is conceivable that the number of the reiterations will control the level of the regulatory action depending on the concentration of the trans-acting factor. This may explain the difference in the gene expression level between different types of cells where the concentration of the trans-acting factor is different. Interestingly, the negative elements of the CANP mL gene share several features with positive regulatory elements of inducible genes such as the metallothionein gene. Both regulatory sequences are composed of tandemly reiterated multiple elements, each element acting on heterologous promoters in an orientation-independent manner, whose regulatory function is mediated by trans-acting factors which exist in limited amounts.
Taking into consideration an ubiquitous distribution of CANP among tissues of higher animals, expression of the CANP gene may be explained by the presence of rather general promoter elements, whose activity is mediated mainly by an ubiquitous trans-acting factor such as Spl, and tandemly reiterated negative regulatory elements, whose activities are mediated by a ubiquitous trans-acting factor whose concentration is cell type-specific. Negative regulatory elements may also be responsible for the feedback regulation of genes where the concentration of the gene product must be strictly maintained. This kind of regulation may be required for genes whose products form a heteromeric structure with other gene products. The presence of stretches of conserved sequences in corresponding regions of the large and the small subunits of human CANP genes (-360 to -60,50% sequence homology including potential Spl and AP-1 recognition sequences, data not shown) suggests that expression of the genes coding for the two CANP subunits is co-regulated at the level of transcription and that their regulation is mediated by the same factor(s) acting on the negative regulatory elements. Future experiments on the cellular factor(s) acting on the CANP gene negative regulatory elements will explore these issues and eventually lead to the elucidation of a general mechanism regulating the expression of intracellular housekeeping genes.