The Promoter for Constitutive Expression of the Human ICln Gene CLNS1A *

The ICln protein is expressed ubiquitously in mammals. Experiments designed to knock down the ICln protein in NIH 3T3 fibroblasts as well as in epithelial cells led to the conclusion that this protein is crucially involved in volume regulation after cytoplasmic swelling. Reconstitution of the ICln protein in lipid bilayers revealed the ion channel nature of ICln. Here we describe a new human promoter sequence, composed of 89 nucleotides, which is responsible for a highly constitutive expression of the ICln protein. The promoter sequence lacks a TATA box, and the transcription can be effected at multiple sites. In addition to the starting sites, upstream sequence elements are mandatory for an efficient transcription of the ICln gene (CLNS1A). These new nucleotide elements were defined by site-directed mutagenesis.

The ICln protein is expressed ubiquitously in mammals. Experiments designed to knock down the ICln protein in NIH 3T3 fibroblasts as well as in epithelial cells led to the conclusion that this protein is crucially involved in volume regulation after cytoplasmic swelling. Reconstitution of the ICln protein in lipid bilayers revealed the ion channel nature of ICln. Here we describe a new human promoter sequence, composed of 89 nucleotides, which is responsible for a highly constitutive expression of the ICln protein. The promoter sequence lacks a TATA box, and the transcription can be effected at multiple sites. In addition to the starting sites, upstream sequence elements are mandatory for an efficient transcription of the ICln gene (CLNS1A). These new nucleotide elements were defined by site-directed mutagenesis.
The increase of the cell volume results in the activation of pathways that effect the reduction of osmotically active molecules and ions in the cytoplasm, thus leading to compensatory cell shrinkage (1,2). The loss of potassium and chloride caused by the activation of swelling-activated channels (SAC) 1 plays a major role in regulatory volume decrease, evidenced by the consistent activation of SAC in mammalian cells after swelling (2). Using the expression cloning technique, we identified a protein (ICln) that, expressed in Xenopus laevis oocytes, leads to a current that is similar to SAC found in several cell types in terms of its kinetics, pharmacology, and relative selectivity for anions (4). The assumption that ICln is functionally tightly linked to SAC is supported further by the finding that the knock-down of the ICln protein in fibroblasts and epithelial cells by the use of antisense oligodeoxynucleotides specific to ICln mRNA can seriously hamper the activation of swellingdependent anion channels (5,6). Using fluorescence in situ hybridization, we identified two different human gene loci that carry the coding region for ICln. One locus at position 6p12, termed CLNS1B, contains an intronless gene (7), whereas the second gene locus at position 11q13.5-14.1, termed CLNS1A, is segmented by introns. The exon sequences of CLNS1A are identical with the cloned human cDNA, suggesting transcription of this gene (8). Because ICln is expressed ubiquitously and transcribed constitutively in mammalian cells (9), and volume regulation is an inherent regulatory entity of living cells, it seems likely that this protein is part of one of the housekeeping regulatory machineries of the cells. However, preliminary experiments indicate that this constitutive transcription of the CLNS1A gene is modulated by the volume stress placed upon the cells and, in addition, during their progression in the cell cycle. A prerequisite of the attempt to investigate these regulatory mechanisms for CLNS1A transcription is the identification of the promoter for the constitutive transcription of this gene. Furthermore, the constitutive expression of the ICln protein is remarkably high. Therefore, the identification of the minimal promoter region needed for its transcription should lead to the identification of a highly efficient human promoter that could also be used for an effective expression of other proteins in human cells. Such a promoter would be very instrumental in avoiding the use of viral promoters for the expression of proteins in human cells. To characterize the minimal promoter responsible for the constitutive expression of the ICln gene, we subcloned and analyzed the 5Ј-flanking region of the human CLNS1A gene.

Cloning of the ICln Promoter Region
By screening a human genomic P1 library (Ressourcen-Zentrum/ Primä r Datenbank RZPD library (10)) with a 700-bp cDNA probe specific to ICln, we isolated a clone termed P1R3 (the corresponding RZPD reference number is ICRFP700E1924Q). The 3Ј-end of this clone was sequenced and corresponded to the 5Ј-end of the human CLNS1A gene (7,8). The P1R3 clone was digested with KpnI and SacI. The restriction enzyme products were separated on a 2% agarose gel and transferred to a nitrocellulose membrane for hybridization. They were subsequently probed using the 5Ј-end of the ICln open reading frame (ORF). This procedure led to the isolation of a Ϸ2.3-kilobase fragment from the P1R3 clone, which was cloned into a pBlueskript SK IIϩ vector thereafter. Sequencing analysis revealed that the clone carried 2151 nucleotides upstream from the first coding ATG and parts of exon 1 of the ICln gene.

PCR Amplification
Standard PCR protocols with Taq or Pfu polymerase (Roche Molecular Biochemicals, Stratagene) were used for parts of the deletion mutations as well as for the site-directed mutations.
* This work was supported in part by Austrian Science Foundation Grants P10393-MED, P12337-MED, and P13041-MED, Austrian National Bank Grant 6994/1, European Commission Grant BMH4-CT96-0602, Gastein Foundation Grant FP46, and by the Ö sterreichische Gesellschaft fü r Lungenerkrankungen und Tuberkulose (to M. P.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AF148460 (human) and AF148459 (monkey).
¶ To whom correspondence should be addressed.

Construction of the Vectors for the Promoter-Reporter System
The 5Ј-flanking region of the 2151-bp fragment described above was cloned into the BglII and HindIII restriction sites of the pGL3-basic vector (Promega) containing the gene for the firefly luciferase (isolated from Photinus pyralis). A size reduction of the 2151-bp fragment was obtained by using the restriction enzyme PvuII, the Erase-a-Base kit (Promega) according to the manufacturer's instructions, or the PCR technique (see above). Site-directed mutagenesis was performed by introducing the respective mutations into the primer used for PCRs. All promoter fragments were sequenced to test for the correct sequence before applying them to the reporter gene assays. To normalize the changing transfection efficacy, the different promoter constructs were cotransfected in HEK 293 T cells with the pRL-TK vector (Promega), containing the gene for the Renilla luciferase (isolated from Renilla reniformis) controlled by the viral TK promoter. All values of the reporter assay are given as the ratio obtained from the firefly and Renilla luciferase readings (luc/ren).

Transient Transfection
The luciferase reporter gene constructs together with the pRL-TK plasmid were transfected into HEK 293 T cells by calcium phosphate precipitation (11,12). For this reason Ϸ10 5 HEK 293 T cells were spread on cell culture dishes with a diameter of 30 mm the day before transfection. For the transfection, 150 ng of promoter plasmid and 75 ng of pRL-TK vector were mixed with buffer A (0.5 M CaCl 2 , 100 mM HEPES, pH 6.95; adjusted with NaOH) and incubated for 10 min at room temperature, before adding buffer B (0.28 M NaCl, 0.75 mM NaH 2 PO 4 , 0.75 mM Na 2 HPO 4 , 5 mM HEPES, pH 6.7; adjusted with NaOH). After a further incubation period of 10 -20 min at room temperature, the transfection mix was spread over the cells. The next day the cells were washed twice with culture medium (see below). On the 3rd day the luciferase assays were carried out.

Luciferase Reporter Gene Assays
To remove traces of culture medium, cells were washed with ice-cold phosphate-buffered saline, then lysed by adding 250 l of lysis buffer (25 mM glycylglycin, 15 mM MgSO 4 , 8 mM EGTA, 2% Tween 20, 1 mM dithiothreitol), and the cells were scraped off the culture dishes. The lysates were analyzed in a luminometer (EG&G Berthold) for both firefly and Renilla luciferase by mixing 50 l of lysate with 350 l of firefly or Renilla assay buffer and 200 l of substrate-buffer for the firefly or Renilla luciferase, respectively.

5Ј-RACE
From the human mRNA, the first strand cDNA was made by using an ICln-specific primer (5Ј-TGT AGG GAT GTC CCC CTG TCC TTG-3Ј) and avian myeloblastosis virus reverse transcriptase (Promega). The reaction was heat inactivated and was then subjected to an RNase A treatment. Unincorporated nucleotides and enzymes were removed using a QIAquick PCR purification column (Qiagen). Polyadenosine tails were added to the 5Ј-end by using terminal deoxynucleotidyl transferase (MBI). Second strand synthesis was carried out by PCR using Taq polymerase (MBI) and the primer RACE-N (5Ј-GCA TCG ATC GCG CGA CTC TTT TTT TTT TTT TTT TT(AGC)-3Ј). For the first PCR, the gene-specific primer (5Ј-AGG TTC AAC ATC ATC ATC ACT GTC-3Ј) and the RACE-N primer were added to the reaction above, and a standard PCR protocol was carried out. To enhance specificity further, the PCR products were subjected to a second nested PCR using the primer RACE-N2 (5Ј-GCA TCG ATC GCG CGA CTC-3Ј, corresponding to the 5Ј-end of the primer RACE-N), and a second gene-specific nested primer (5Ј-AGA GCC ATC TAA CCA AGA CA-3Ј). The fragments obtained were subcloned and sequenced.

Primer Extension
Using Radioactive-labeled Primer-The synthetic oligonucleotide primer (5Ј-ACA AAT GCT CTC CTA GAC AGT C-3Ј, starting within exon 2) was labeled at the 5Ј-end with [␥-32 P]ATP using T4 polynucleotide kinase (MBI). The labeled primer was purified and annealed with poly(A) ϩ RNA from HEK 293 T cells. For the extension, Moloney murine leukemia virus reverse transcriptase (Promega) was added. After RNase A treatment, the samples were loaded on a denaturing 6% polyacrylamide gel.
Using Fluorescence-labeled Primer-A primer starting at position ϩ87 (5Ј-CCC GTT CAG CAC AGC CTC-3Ј; ϩ1 being the adenosine of the first ATG) labeled with IRD-800 was used for the reverse transcription, and the products were sequenced by an automatic sequencer (LiCor Gene ReadIR 4200, MWG Biotech).

Nuclease Protection Assay
The IRD-800-labeled primer mentioned above, starting at ϩ87, was employed for unidirectional PCR with a template ending at Ϫ224. This product was used as the antisense probe for the nuclease protection assay (Multi-NPA; Ambion), and the length of the products was analyzed by an automatic sequencer (LiCor Gene ReadIR 4200).

Identification of the Monkey ICln Promoter Sequence
Monkey genomic DNA was purchased from CLONTECH. For the PCR two primers were used which were identical to the human ICln gene. With respect to the human gene, the primer started at position Ϫ783 (forward; 5Ј-GAA GAT CTT CTG ATT GGT TGG GTG GGA GAT G-3Ј) and ϩ24 (reverse; 5Ј-CGG GAA ACT TTT GAG GAA GCT CAT-3Ј). The amplified fragment was subcloned and sequenced (see below).

Promoter Efficacy
To test for the promoter efficacy of the Ϫ163/Ϫ74 fragment compared with the CMV or RSV promoter we analyzed the ICln protein expression in HEK 293 T cells using the corresponding promoter regions. The pGL3-basic vector was used after modifying the reporter region. The luc ϩ gene was replaced by the ICln coding region after attaching the nucleotide triplets coding for a sequence of five histidines prior to the starting methionine of ICln. This allowed us to purify the ICln protein derived from the translation of the extrachromosomal transcribed plasmid cDNA. The pGL3-basic vector replaced by His-ICln was termed His-ICln-basic and used for the subsequent admission of the respective promoter sequences. The Ϫ163/Ϫ74 fragment was cloned into the SmaI site, whereas the 654-nucleotide fragment of the CMV promoter was cloned into the NheI/XhoI sites and the 627-nucleotide fragment of the RSV promoter was cloned into the SacI/XhoI sites. The respective vectors were transfected into HEK 293 T cells by calcium phosphate precipitation (11,12). For this reason Ϸ10 6 HEK 293 T cells were spread on cell culture dishes with a diameter of 90 mm the day before transfection. For the transfection, 7 g of promoter plasmid was mixed with buffer A and incubated for 10 min at room temperature before adding buffer B. After a further incubation period of 10 -20 min at room temperature, the transfection mix was spread over the cells. After 18 h the cells were washed twice with culture medium, and after an additional 6 h the cells were harvested, and cell lysis was performed using 200 l of lysis buffer (1% Triton X-100, 1 mM phenylmethylsulfonyl fluoride, 20 g/ml of each leupeptin, pepstatin A, antipain, and aprotinin, in phosphate-buffered saline at pH 8.00). A total volume of 200 l of supernatant was used for the extraction of His tag-fused ICln protein (Ni-NTA spin columns; Qiagen). Imidazole (400 mM imidazole, 50 mM K 2 HPO 4 , pH 7.40) was used for eluting the His-ICln. The extract was subjected to gel electrophoresis, blotted, and analyzed using antibodies specific for the ICln protein. The different Western blots were quantified using the ImageQuant (version 1.0) software (Molecular Dynamics), and the values were normalized according the total protein measurements (Bradford).

DNA Sequencing
All plasmids used were sequenced using an automatic sequencer (LiCor Gene ReadIR 4200) with the protocols suggested by the manufacturer.

Chemicals
All chemicals used were of pro analysis grade.

Statistical Analysis
Where applicable, data are expressed as arithmetic means Ϯ S.E. Statistical analysis was made by t test, where appropriate. Significant difference was assumed at p Ͻ 0.05.

RESULTS
Cloning of the 5Ј-Untranslated Region of CLNS1A and Transcription Start Estimation-By screening a human genomic P1 library as described under "Experimental Procedures" with a 700-bp cDNA probe specific for ICln, we isolated a clone termed P1R3 (the corresponding RZPD reference number is ICRFP700E1924Q). The 3Ј-end of this clone was sequenced and corresponded to the 5Ј-end of the human ICln ORF (7,8). Using fluorescence in situ hybridization, we located the P1R3 clone at the human chromosomal position 11q13.5-14.1, as expected. The P1R3 clone carries a Ͼ50-kilobase segment containing the 5Ј-untranslated region of the CLNS1A gene. To study the promoter activity of CLNS1A, we subcloned and sequenced a 2.259-bp KpnI/SacI fragment from P1R3. This fragment was identified by hybridization with the 5Ј-end of the CLNS1A ORF and corresponded to position Ϫ2151 to ϩ108 of the CLNS1A gene (Ϫ1 being the first 5Ј-untranslated nucleotide of the human ICln cDNA, and ϩ1 being the adenosine in the starting ATG of the ORF).
Several human ICln cDNA clone sequences have been reported so far. For the different clones, 5Ј-untranslated regions of varying lengths have been described, i.e. Ϫ88 (accession number 4502890 (13)), Ϫ84 this sequence was submitted by Lamb et al.), 2 Ϫ73 (HSU17899 (14)), and Ϫ30 (AF005422 (15)). Except for the sequence published by Anguita et al. (14), all of the other sequences are identical to our KpnI/SacI fragment. The sequence published by Anguita et al. (14) is perfectly matched by P1R3 up to position Ϫ41; however, the following 32 nucleotides located further upstream from this position show no homology to the P1R3 sequence or the cDNAs published by Buyse et al. (13) or Lamb et al. 2 At present we cannot clearly determine whether the initial 32 nucleotides of the sequence published by Anguita are related to the ICln mRNA or whether they are the result of the cloning procedure used.
The different lengths of the 5Ј-untranslated regions of the reported ICln cDNAs could indicate that multiple starting sites are used for the ICln gene transcription. To test this possibility, we performed several tests including primer extension, nuclease protection, and 5Ј-RACE using poly(A) ϩ RNA from HEK 293 T cells as a template. For the primer extension, the mRNA was annealed with a 5Ј-end 32 P-labeled primer, designed to anneal within exon 2 of ICln, and the primer was extended using a Moloney murine leukemia virus reverse transcriptase. Autoradiographic analysis of the primer extension products separated by polyacrylamide gel electrophoresis revealed products with a transcription start at positions Ϫ78 and Ϫ49. Using a primer starting at position ϩ87, labeled with IRD-800, a predominant start for transcription at position Ϫ48 was measured. By using the 5Ј-RACE, we obtained three major products, which were subsequently sequenced. The respective transcripts started at positions Ϫ95 (two independent clones were isolated and sequenced), Ϫ78, Ϫ77, and Ϫ48, respectively. The transcription starts at the Ϫ78 and Ϫ48 clusters were also verified by using the nuclease protection assay. From these experiments we conclude that there are several starting points for human ICln gene transcription, thus providing an explanation for the varying 5Ј-untranslated sequences published for ICln so far.
Determination of the Promoter Sequences Responsible for Base-line ICln Expression-The ICln protein is expressed ubiquitously in mammals (4,9,13). To identify the promoter re- sponsible for this base-line expression, we used the luciferase reporter system (Promega) by transfecting HEK 293 T cells with promoter-luciferase vector constructs together with the Renilla pRL-TK vector for normalization.
The first sequence tested was generated by PCR using the KpnI/SacI fragment as a template. The sequence was limited by the positions Ϫ2151 and Ϫ1 and was cloned into the luciferase reporter vector pGL3-basic. As shown in Fig. 1, this construct gives a signal that is more than 40-fold higher compared with the pGL3-basic vector containing no promoter sequences or foreign DNA (1.011 Ϯ 0.1144 luc/ren, n ϭ 49 and 0.023604 Ϯ 0.0072199, n ϭ 51, respectively). The efficacy of the Ϫ2151/Ϫ1 fragment in driving expression is close to the efficacy of the viral TK promoter. This is supported by experiments in which the pGL3-TK vector (luciferase gene under the control of the TK promoter), cotransfected with the Renilla pRL-TK vector applying the same molar ratio as used for the fragment analysis, shows an activity that is only five times higher (5.291 Ϯ 0.524 luc/ren, n ϭ 6). The cloning direction of the Ϫ2151/Ϫ1 fragment into the pGL3 vector is, as expected, important because a reporter construct containing a fragment (Ϫ568/ϩ108) in the reverse direction showed no promoter activity (0.0041 Ϯ 0.0005 luc/ren, n ϭ 9).
The reduction of the Ϫ2151/Ϫ1 fragment from the 5Ј-as well as from the 3Ј-ends leads to the fragment Ϫ783/Ϫ35, whose promoter activity is not significantly different from that of the Ϫ2151/Ϫ1 fragment (Fig. 1). Nested deletions from the 5Ј-end of the Ϫ783/Ϫ35 clone produced further five fragments, all of which display activities comparable to the activity of the Ϫ2151/Ϫ1 fragment (Fig. 1). These experiments suggest that the base-line promoter activity of ICln has to be associated with a fragment confined by the positions Ϫ167 and Ϫ35. Accordingly, the two fragments starting at position Ϫ162 and ending at position Ϫ783 and Ϫ2151, respectively, show no activity (Fig. 1).
After the restriction of the Ϫ167/Ϫ35 fragment to Ϫ163/Ϫ56, there was no significant loss of activity compared with the former level (Fig. 2). However, further reduction of the 5Ј-end to position Ϫ154, Ϫ148, Ϫ135, or Ϫ126 decreased the activity by half. Therefore, the region between Ϫ163 and Ϫ150 was examined by sitedirected mutagenesis to detect the nucleotides responsible for the Ϸ50% reduction of the promoter activity. As shown in Fig. 3, the mutants AϪ150G, TϪ154G, and Ϫ156CCϪ154/Ϫ156TTϪ154 show no reduction of the signal, whereas the mutant Ϫ159TTTϪ157/Ϫ159GCGϪ157 is associated with an activity not significantly different from that of the Ϫ154/Ϫ56 fragment (0.69 Ϯ 0.1 luc/ren, n ϭ 15 for the mutant and 0.60 Ϯ 0.1 luc/ren, n ϭ 26 for the Ϫ154/Ϫ56 fragment). These experiments indicate that the thymidines at positions Ϫ157, Ϫ158, and Ϫ159 are most likely essential for half of the transcriptional capacity observed. The remaining half can be annihilated by further reduction of the 5Ј-end of the Ϫ126/Ϫ56 clone.
As shown in Fig. 4, the reduction of the 5Ј-end to the positions Ϫ113, Ϫ99, or Ϫ81 leads to values not significantly different from those of the empty pGL3 vector. Therefore, the thymidine triplet together with nucleotides comprising the string from Ϫ126 to Ϫ113 is necessary for the activity observed. By mutating the sequence between Ϫ126 and Ϫ113, further reduction of the essential sequence to single nucleotides was possible. As shown in Fig. 5, the mutant AϪ119G/TϪ117G was not followed by a decrease in activity. However, the activity of the mutant Ϫ122CAϪ121/Ϫ122GGϪ121was significantly impaired. Accordingly, the reduction of the 5Ј-end to position Ϫ120 (fragment Ϫ120/Ϫ56) leads to an activity that does not deviate significantly from the value obtained by the transfec- tion of the Ϫ122CAϪ121/Ϫ122GGϪ121 mutant. The activity produced by the transfection of the Ϫ120/Ϫ56 fragment can be annihilated by further reduction at the 5Ј-end to position Ϫ117 (Fig. 5).
The experiments show that the base-line activity measured after the transfection of fragment Ϫ2151/Ϫ1 can be obtained when the thymidine triplet (Ϫ159TTTϪ157), the cytosine CϪ122 and adenosine AϪ121, and the guanine GϪ120 and thymidine TϪ118 are present. In addition to these nucleotides mandatory for the promoter activity, the presence of starting sites is also necessary because the fragment embraced by the positions Ϫ163 and Ϫ113 shows no activity compared with the empty pGL3 vector (0.0137 Ϯ 0.00056 luc/ren, n ϭ 6). As mentioned above, major transcription starts were identified at positions Ϫ95, Ϫ78/Ϫ77, and Ϫ49/Ϫ48. Surprisingly, the restriction of the 3Ј-end of the tested sequences to position Ϫ74, which therefore excludes a transcription start at position Ϫ49/ Ϫ48, has a promoter activity similar to that of the Ϫ163/Ϫ1 fragment (Fig. 6), indicating that the critical sites for the transcription start are located upstream from position Ϫ74. This is also evidenced by the finding that four out of the five clones we have identified start at the Ϫ95 cluster (two clones) and the Ϫ78 cluster (two clones). Accordingly, the fragment starting at position Ϫ163 and ending at position Ϫ95, i.e. exactly at the most proximal end of the cDNA identified by 5Ј-RACE, leads to a signal not different from that of the empty pGL3 vector (Fig.  6). This indicates that the critical sequence for the transcription start is between positions Ϫ95 and Ϫ74. This stretch of nucleotides contains two pyrimidine-rich clusters with the CT-TCC sequence. As shown in Fig. 7, these pyrimidine-rich clusters are perfectly conserved from humans to monkeys and mice (16). If the Ϫ78 cluster is omitted (fragment Ϫ163/Ϫ88), the transcription efficacy is half of that of the Ϫ163/Ϫ74 fragment (Fig. 6). The reduction of the same amount can be observed when the Ϫ78 cluster is present, and a mutation was introduced in the Ϫ95 cluster (Ϫ96CTTCϪ93/Ϫ96GGGGϪ93), thus suggesting that both pyrimidine-rich clusters are equally im-portant for an effective transcription.
Efficacy of the Ϫ163/Ϫ74 Promoter Compared with the Viral Promoters RSV and CMV-To estimate the efficacy of expression driven by the Ϫ163/Ϫ74 ICln promoter fragment, we compared the ICln fragment with the expression governed by the viral promoters RSV and CMV. Because the ICln protein is endogenously expressed in HEK 293 T cells, we made promoter constructs harboring the coding region of an ICln protein fused to a string of five histidines, allowing enrichment of the protein extract for this newly expressed protein. We quantified the expressed ICln protein using a polyclonal antibody made against a peptide composed of the 24 C-terminal amino acids of the ICln protein. The Ϫ163/Ϫ74 ICln promoter fragment leads to an expression of the histidine-tagged ICln protein, which is only four times or three times lower compared with the CMV or RSV promoter, respectively (Ϫ163/Ϫ74 ICln promoter fragment expression ϭ 0.69 Ϯ 0.21, n ϭ 4, RSV ϭ 2.22 Ϯ 0.31, n ϭ 3 and CMV ϭ 2.76 Ϯ 0.26, n ϭ 4).

DISCUSSION
The cytoplasmic volume of cells is tightly regulated, and the mechanisms involved comprise ion transporters and/or channels (2) as well as cellular osmolyte production, uptake, or exit (17,18). The transcriptional regulation of transporters for compatible osmolytes activated after increasing the extracellular osmolality is well understood (17,18). The control of rENaC sodium channel expression, again in response to hypertonicity, is in the process of investigation (19). However, little is known regarding the transcription of genes involved in the ion conductive pathways implicated in volume regulation after swelling.
When whole-cell patch-clamp experiments are made using symmetrical CsCl concentrations, a chloride current can be measured after the reduction of the extracellular osmolality (5,20,21). Similar findings have been reported for a variety of different cells (2). The channels activated under hypotonic conditions are, however, more permeable to other anions than to chloride i.e. SCN Ϫ , Br Ϫ , and I Ϫ . They are also permeable, even though to a lower degree, to cations and, in addition, to osmolytes (22)(23)(24)(25). Therefore the habit to use terms such as "swelling-induced," "volume-regulated," or "volume-sensitive outwardly rectifying" chloride channels seems inappropriate. It is probably more suitable to name these ionic pores swellingactivated channels. Several proteins are thought to be molecular targets for SAC. The list of the candidates comprises ClC-2 (26), mdr-1 (27), phospholemman (28,29), ClC-3 (30), and ICln (4,31). Because the expression of these different proteins was done in cells bearing endogenous SAC, the experiments cannot unambiguously clarify whether these proteins are the channels responsible for the currents observed after cytoplasmic swelling. When using antisense oligodeoxynucleotides that specifically impair the ICln production in NIH 3T3 fibroblasts (5) or epithelial cells (32), the activation of SAC is reduced dramatically after the decrease of extracellular osmolality. This indicates that the ICln protein is critical to the appearance of the SAC current under hypotonic conditions. Based on experiments using immunohistochemistry and Western blots, we developed the hypothesis that ICln is a constitutively expressed protein that can be transposed from a water-soluble form in the cytosol into the membrane, thus leading to an ionic current (31). The fact that ICln can indeed act as an ion channel was confirmed by functionally reconstituting the protein into lipid bilayers (31,33). In the absence of calcium, the current obtained in bilayers is more selective for cations than for anions; however, the addition of calcium shifts the selectivity toward chloride. The reconstituted ICln current is, just like the current effected by native SAC, rectifying and can be blocked by nucleotides. Therefore, we conclude that ICln is the molecular entity of SAC or a substantial part thereof (31).
As mentioned above, little is known regarding the transcriptional regulation of SAC. It has been shown that the transcription of the ClC-2 protein in rat lung is modulated by Sp1 and Sp3 (34); however, nothing is known about the transcription regulation of potential SAC candidates in human cells. This has prompted us to study the transcription regulation of human ICln, for which an important role in regulatory volume decrease was defined. These studies might bring about a better understanding of the cross-talk between plasma membrane and nucleus. Experiments done so far indicate that the constitutive expression of ICln might be modulated by the necessity for regulatory volume decrease (35,36) and seems to be dependent on cell cycle progression. A number of different consensus regions for the binding of a variety of transcription factors can be identified 5Ј of the first coding ATG of the CLNS1A gene. Among them also a Sp1 site identical to the one found in the rat ClC-2 promoter was identified; however, the role of these sites needs to be scrutinized before a definitive function can be assigned. The characterization of the nucleotide elements needed for the constitutive expression of ICln is a prerequisite for the in-depth analysis of these regulatory mechanisms. Furthermore, it may also facilitate the definition of a human promoter sequence that could allow a high copy expression of foreign proteins in human cells without the need for viral promoters. The experiments summarized in this paper were therefore made to define the nucleotides necessary for the constitutive expression of human ICln.
Four different cDNAs coding for the human ICln have been reported so far, each having a 5Ј-untranslated region of varying length (Ϫ88, Ϫ84, Ϫ73, and Ϫ30) and indicating that the transcription of the ICln gene starts at multiple sites. With the aid of primer extension, nuclease protection, and 5Ј-RACE, we were able to define three different initiator regions (Inr) in close proximity to the 5Ј-untranslated sequences published.
The Inr closest to the ORF of the ICln gene is similar to the canonical Inr sequence described as Py 2 ANT/APy 2 (the underlined A corresponds to position ϩ1 of the transcript (37)). In the human CLNS1A gene this site is located at position Ϫ50 and comprises the nucleotides CGCATTGCT. The identical sequence is also found in the ICln gene of monkeys (Fig. 7). In mice this sequence is slightly changed to GCGATTGCG but reveals a thymidine at the crucial position of ϩ3 (A being at position ϩ1 and therefore confining the predicted transcription start (37)). Testing the efficacy of this Inr revealed that this starting point is probably not mandatory for base-line ICln gene transcription because the deletion of this nucleotide cluster does not reduce the constitutive expression of the reporter system we used. In contrast, two clusters at position Ϫ95 and Ϫ78 seem to be critical for the transcription of the human ICln gene. The nucleotide sequences of both clusters are identical, reading CTTCC, which are in addition completely conserved in mouse, monkey, and humans. Despite the fact that the first base of the mRNA tends to be adenosine, weight-matrix analysis of an extensive number of Inr sequences revealed that the transcription start can be efficiently effected at a cytosine embraced by pyrimidines. An effective Inr tested was built by the sequence GTTCTTCC (the underlined C would be the predicted start of the transcription (38)). This consensus region for Inr is identical to the most proximal cluster we identified as a possible Inr for ICln. 5Ј-RACE in HEK 293 T cells led to two clones that start within this Inr cluster, suggesting that this sequence is functionally important to the ICln transcription. This hypothesis is supported by the finding that mutations made within this sequence lead to a reduction of the promoter activity by 50%. The third Inr cluster (position Ϫ78 in the human ICln gene) is similar in sequence (CCTCTTCC in humans and monkeys and CTGCTTCC in mice) and function because the deletion of this sequence leads again to a drop in the promoter activity by 50%.
Sequence analysis of the fragments used in this study failed to identify any obvious TATA box. Similar results were obtained when analyzing the promoter of the ClC-2 chloride channel (34), which is also a possible candidate for SAC (26). The products of genes lacking a TATA box are often involved in housekeeping functions. However, TATA-less genes could also be identified when tissue-specific expression is observed, i.e. the ClC-K1 chloride channel (39), CFTR (40,41), or the betaine ␥-amino-n-butyric acid transporter (42). The lack of a TATA box usually results in multiple starting sites, as also shown in this study, and does not rule out that sequences upstream from the starting sites are necessary for the binding of proteins needed for effective transcription. It was shown for the TATAless simian virus 40 major late promoter that effective transcription is only possible when the cloned human TATA boxbinding protein hTFIID functionally binds to an upstream sequence element reading 5Ј-TACCT-3Ј. The mutation of both thymidines reduced transcription more than 50%, whereas the exchange of the two cytosines in thymidines enhanced tran-scription 8-fold (43). We performed similar mutation experiments in a region of the ICln promoter (5Ј-TTCCT-3Ј), which is closely related to the 5Ј-TACCT-3Ј region of the simian promoter, and we found that the mutation of the first thymidine was also followed by a reduction of the transcription. The mutation of the second thymidine, however, showed no effect. Moreover, mutating the two cytosines in thymidines was not followed by an increased transcription. Therefore it is not certain at the moment whether the binding of hTFIID is obligatory for the ICln gene transcription. Beside the described thymidine, which is part of the Ϫ159 triplet, a second nucleotide triplet at position Ϫ122 was identified being crucial for ICln transcription. These triplets have identical sequences in humans and monkeys and are similar in mice (Figs. 7 and 8). Interestingly enough, these nucleotide triplets are in close proximity to a sequence stretch with a length of 11 nucleotides, which is identical in all three organisms. Further experiments are needed to elucidate whether this region of high homology is functionally important for transcription.
The expression driven by the Ϫ163/Ϫ74 ICln promoter fragment is highly efficient because the viral RSV or CMV promoters lead to an expression efficacy, which is only three or four times higher compared with the ICln fragment.
In conclusion, we identified promoter elements and Inr sites, which are essential for the constitutive expression of the ICln protein (Fig. 8). The minimal sequence stretch showing full efficacy compared with the Ϫ2151/Ϫ1 fragment has a length of 89 nucleotides and is limited by the positions Ϫ163 and Ϫ74, related to the human CLNS1A. The efficacy of this fragment to express the reporter system is high; and given the fact that no cell type has been identified so far which would lack ICln expression, it might appear possible to use this 89-nucleotide minimal promoter sequence for the expression of other proteins in human cells, when the use of viral promoters is not desired. Koenig, M., Gschwentner, M., Ritter, M., Schmarda, A., Botta, G., Benz, R., Deetjen, P., and Paulmichl, M. (2000) Pfluegers Arch., in press).