ORE, a eukaryotic minimal essential osmotic response element. The aldose reductase gene in hyperosmotic stress.

Organisms, almost universally, adapt to hyperosmotic stress through increased accumulation of organic osmolytes but the molecular mechanisms have only begun to be addressed. Among mammalian tissues, renal medullary cells are uniquely exposed to extreme hyperosmotic stress. Sorbitol, synthesized through aldose reductase, is a predominant osmolyte induced under hyperosmotic conditions in renal cells. Using a rabbit renal cell line, we originally demonstrated that hyperosmotic stress induces transcription of the aldose reductase gene. Recently, we cloned the rabbit aldose reductase gene, characterized its structure, and found the first evidence of an osmotic response region in a eukaryotic gene. Now, we have progressively subdivided this 3221-base pair (bp) region into discrete fragments in reporter gene constructs. Thereby, we have functionally defined the smallest sequence able to confer hyperosmotic response on a downstream gene independent of other putative cis-elements, that is, a minimal essential osmotic response element (ORE). The sequence of the ORE is CGGAAAATCAC(C) (bp −1105/−1094). A 17-bp fragment (−1108/−1092) containing the ORE used as a probe in electrophoretic mobility shift assays suggests hyperosmotic induction of a slowly migrating band. Isolation of trans-acting factor(s) and characterization of their interaction with the ORE should elucidate the basic mechanisms for regulation of gene expression by hyperosmotic stress.

Organisms, almost universally, adapt to hyperosmotic stress through increased accumulation of organic osmolytes but the molecular mechanisms have only begun to be addressed. Among mammalian tissues, renal medullary cells are uniquely exposed to extreme hyperosmotic stress. Sorbitol, synthesized through aldose reductase, is a predominant osmolyte induced under hyperosmotic conditions in renal cells. Using a rabbit renal cell line, we originally demonstrated that hyperosmotic stress induces transcription of the aldose reductase gene. Recently, we cloned the rabbit aldose reductase gene, characterized its structure, and found the first evidence of an osmotic response region in a eukaryotic gene. Now, we have progressively subdivided this 3221-base pair (bp) region into discrete fragments in reporter gene constructs. Thereby, we have functionally defined the smallest sequence able to confer hyperosmotic response on a downstream gene independent of other putative cis-elements, that is, a minimal essential osmotic response element (ORE). The sequence of the ORE is CGGAAAATCAC(C) (bp ؊1105/؊1094). A 17-bp fragment (؊1108/؊1092) containing the ORE used as a probe in electrophoretic mobility shift assays suggests hyperosmotic induction of a slowly migrating band. Isolation of trans-acting factor(s) and characterization of their interaction with the ORE should elucidate the basic mechanisms for regulation of gene expression by hyperosmotic stress.
In nature, one of the most prevalent types of stress is that caused by prolonged exposure of an organism to a hyperosmotic environment. Although the cellular responses to hyperosmotic stress are among the most profound, the molecular mecha-nisms involved have only recently begun to be addressed. Thus, contrary to the earnest study of heat shock at the molecular level, relatively little is known about the cascade of signals between the initial extracellular stimulus (hyperosmolality) and the ultimate adaptative response. It is recognized that, across the evolutionary spectrum, organisms have developed a universal adaptation process to cope with hyperosmotic stress; that is, increased accumulation of osmotically active organic solutes (organic osmolytes) (1). Cells accumulate high concentrations of organic osmolytes in place of inorganic ions. This is because, unlike equivalent concentrations of inorganic ions, the organic osmolytes apparently are not perturbing to cellular macromolecules (1).
Some of the predominant organic osmolytes are: betaine in bacteria (e.g. Escherichia coli) (2), glycerol in yeast (e.g. Saccharomyces cerevisiae) (3,4), and sorbitol and betaine in cells of the mammalian renal medulla, the only mammalian tissue routinely exposed to extreme hyperosmotic stress in normal physiological conditions (5). In bacteria, where the molecular mechanisms for accumulation of organic osmolytes have been studied most extensively, hyperosmotic stress induces transcription of the proU operon. proU encodes the transport system involved in accumulation of betaine (2). Osmotic control of proU transcription is exerted, at least in part, by an "upstream activating region" that is currently being characterized by several groups (reviewed in Ref. 2). In S. cerevisiae, studies of the high-osmolarity glycerol (HOG) response have concentrated on the signal transduction pathway immediately following hyperosmotic stress and preceding induction of target genes that directly control accumulation of glycerol (reviewed in Ref. 4). Thus, only one ultimate gene target in the HOG pathway has been identified. That is GPD1, which encodes glycerol-3-phosphate dehydrogenase, one of the two enzymes that catalyze the synthesis of glycerol. Most recently, osmotic stress was shown to increase glycerol-3-phosphate dehydrogenase activity and mRNA levels (3). The identification of GPD1 gene regions that control the osmotic response has not been reported.
In mammalian renal medullary cells under hyperosmotic stress, the synthesis of sorbitol, catalyzed by aldose reductase (AR), 1 is increased (6). PAP-HT25 cells are a line of rabbit inner medullary cells (7) that accumulate large amounts of sorbitol (8) and other organic osmolytes (9) under hyperosmotic conditions. Using this line, we originally demonstrated that hyperosmotic stress induces transcription of the AR gene (10), resulting in increased AR mRNA levels (11), followed by a rise in AR protein synthesis rate (12) and, ultimately, increased sorbitol accumulation (13). Recently, we cloned the rabbit AR gene, characterized its structure, and found the first evidence of an osmotic response element (ORE) within a eukaryotic gene (14). A 3221 base pair (bp) fragment of the 5Ј-flanking region of the aldose reductase gene was shown to confer osmotic response to a downstream luciferase gene with the AR promoter as well as with a heterologous, B19, promoter (14). Here, we present the identification of the minimal essential ORE that controls induction of AR transcription by hyperosmotic stress. * 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.
‡ To whom correspondence should be addressed: National Institutes of Health, Bldg. 10 Reporter Gene Expression Analysis of Transient Transfectants-Expression vectors, 007Luc, ARLuc, and B19CAT, were as described (14). 007Luc is promoterless and contains the luciferase gene, whereas AR-Luc contains the rabbit AR promoter (bp Ϫ208/ϩ27) in unique XhoI-HindIII sites immediately upstream of the luciferase gene. ARLuc was previously demonstrated to exhibit basal promoter activity (as construct ARLuc6 in Ref. 14). B19CAT contains the B19 promoter immediately upstream of the chloramphenicol acetyltransferase (CAT) gene (14,15). ARL and TKL were constructed to obtain the ability to subclone AR fragments directionally in vectors containing either the AR or a heterologous promoter from the polyoma virus thymidine kinase (TK) gene, respectively, upstream of the luciferase gene. ARL and TKL vectors were constructed by using PCR primers synthesized with appropriate restriction enzyme sites toward their 5Ј ends. ARL contains the AR promoter (bp Ϫ208/ϩ27) flanked immediately upstream by XhoI, NcoI, and KpnI sites and immediately downstream by AflII and HindIII sites. TKL contains the same restriction enzyme sites flanking the mutant polyoma virus enhancer PYF441 and the thymidine kinase promoter PCR-amplified as a unit from pSSC-9 (16). The AR and TK promoters were subcloned directionally into unique XhoI-HindIII sites in 007Luc to produce ARL and TKL.
AR fragments longer than 70 bp to be subcloned into expression vectors were PCR-amplified from genomic clone gAR56 -5 (GenBank™ accession number is U12317) (14) using primers synthesized with appropriate restriction enzyme sites toward their 5Ј end. Fragments less than 70 bp, inclusive of appropriate restriction enzyme sites, were created by annealing synthesized sense and antisense oligonucleotides (DNA/RNA Synthesizer, model 392, Applied Biosystems Inc.). Fragments were subcloned into the XhoI site of ARLuc, followed by selection of clones carrying the fragment in the forward direction. Alternatively, fragments were directionally cloned into the XhoI-KpnI sites of ARL or TKL. Construct sequences were verified by primer-directed, doublestranded plasmid sequencing (Sequenase DNA Sequencing Kit, U. S. Biochemical Corp.). To confirm osmotic regulation of a heterologous promoter by the ORE, a 17-bp fragment (bp Ϫ1108/Ϫ1092) containing the minimal essential ORE was subcloned in the forward direction into the XhoI-KpnI sites of TKL upstream of the TK promoter.
Transfection, Luciferase, and CAT Assays-Transfections were performed as described previously (14). Briefly, rabbit renal medulla PAP-HT25 cells (passage 72-80) were grown in isoosmotic medium (300 mosm/kg H 2 O) (7) in 150-mm dishes (Corning) and co-transfected with a given luciferase construct (3 g) and B19CAT (12-20 g) using Cell-Phect (Pharmacia Biotech Inc.). From each transfected 150-mm dish, cells were seeded into six 35-mm dishes and left overnight. The medium in three of the dishes was changed to fresh isoosmotic medium (300 mosm/kg H 2 O); the medium in the three other dishes was changed to the same medium made hyperosmotic (500 mosm/kg H 2 O) with NaCl. Twenty-four hours after changing the medium, cells were harvested by adding 100 l of lysis buffer (Enhanced Luciferase Assay Kit, Analytical Luminescence Laboratory).
Cell lysates were analyzed for total protein, luciferase activity, and CAT protein using the following kits as per manufacturers' instructions. Total protein was determined using the Bio-Rad Protein Assay

TABLE I Effect of hyperosmolality on reporter gene expression in transient transfectants
Transfected PAP-HT25 cells were maintained in isoosmotic medium (ISO) (300 mosm/kg H 2 O) or exposed to hyperosmotic medium (HYPER) (500 mosm/kg H 2 O) for 18 -24 h. Values are expressed as the ratio of HYPER divided by ISO. 007Luc is promoterless, ARLuc and ARL contain the rabbit AR promoter (bp Ϫ208 to ϩ27), and TKL contains the thymidine kinase (TK) promoter; all promoters are upstream of the luciferase gene. DNA fragments from the rabbit AR gene were inserted upstream of the AR promoter in ARLuc and ARL or upstream of the TK promoter in TKL; these fragments are shown directly below their corresponding promoter. Positions of nucleotides that define the AR gene DNA fragments are numbered with the first nucleotide of exon 1 as ϩ1. Positive numbers indicate nucleotides downstream of ϩ1; negative numbers are nucleotides upstream of ϩ1. For constructs demonstrating hyperosmotic response, the size of the DNA fragment is shown in parentheses. Tandem repeats of selected fragments are indicated (e.g. 3X). Cells were co-transfected with a given luciferase construct and B19CAT. B19CAT contains the B19 promoter upstream of the CAT gene. Luciferase activity in relative light units/g of cell protein was normalized by CAT protein in picograms/g cell protein (Luc/CAT). *HYPER/ISO is the ratio of Luc/CAT in hyperosmotic medium divided by Luc/CAT in isoosmotic medium (mean Ϯ standard error). **HYPER/ISO is the ratio of Luc/CAT in hyperosmotic medium divided by Luc/CAT in isoosmotic medium expressed relative to the construct containing the promoter alone (mean Ϯ standard error); these data are shown only for constructs demonstrating hyperosmotic response. n is the number of independent transfections. Transfection Data Analyses-Luciferase activity in relative light units/g of cell protein was normalized by CAT protein in picograms (from the co-transfected B19CAT construct)/g of cell protein.

RESULTS AND DISCUSSION
We first identified a 3221-bp region of DNA (Ϫ3429/Ϫ209) that contained a putative osmotic response element (as construct ARLuc9 in Ref. 14) (Table I). We have now proceeded to functionally identify within that fragment the smallest sequence that could confer osmotic response on a downstream gene independent of other putative cis-elements that may potentiate the response; that is, a minimal essential osmotic response element. Instead of traditional nested deletions, we increasingly subdivided the Ϫ3429/Ϫ209 region into discrete pieces and tested them individually. This allowed us to exam-ine all fragments for independent osmotic response as opposed to only those fragments that remain after deletion. All fragments were synthesized either by PCR amplification (Ն146 bp) or directly on a DNA synthesizer and tested for ability to confer osmotic response to a luciferase gene driven by the rabbit aldose reductase promoter (Ϫ208/ϩ27) (14,20) in transient transfection assays (Table I). As shown in Table I in the ARLuc constructs, the most upstream region (Ϫ3429/Ϫ2686 and Ϫ2705/Ϫ1152) of the 3221-bp fragment (Ϫ3429/Ϫ209) did not generate osmotic response. However, the downstream 962 bp (Ϫ1170/Ϫ209) retained osmotic response. Within this 962-bp fragment, osmotic response was confined to an upstream 679-bp fragment (Ϫ1170/Ϫ492).
Initially, for the ARLuc constructs (Table I), the DNA fragments were cloned into the only available unique restriction enzyme (XhoI) site immediately upstream of the AR promoter. After having narrowed down the ORE to 277 bp (Ϫ1170/Ϫ894), the construct ARL was produced by adding, to ARLuc, unique restriction enzyme sites that would allow us to subclone directionally (see "Experimental Procedures"). Since the added bases could affect function, we subcloned into ARL two of the fragments (Ϫ1170/Ϫ492 and Ϫ1170/Ϫ894) that had osmotic response in ARLuc. Indeed, these fragments continued to display osmotic response when in the ARL vector (Table I).
Osmotic response had been gradually decreasing with the size of the fragment, as seen in the only other sequence identified as containing a eukaryotic osmotic response element (19). To better evaluate osmotic response, we created tandem repeats of the 26-and 17-bp fragments, hypothesizing that this would magnify the response. As shown in Table I, tandem repeats of sequence containing the ORE (Ϫ1108/Ϫ1083 two times and Ϫ1108/Ϫ1092 three times) markedly increased osmotic response. In constrast, tandem repeats (Ϫ1104/Ϫ1091 three times) of a fragment having no osmotic response (Ϫ1104/ Ϫ1091) were unable to evoke the response. We conclude that the sequence Ϫ1108/Ϫ1092 alone can confer osmotic response. The gradual drop in magnitude with decreasing fragment size suggests the possibility that other cis-elements may potentiate the osmotic response.
FIG. 1. The aldose reductase minimal essential osmotic response element (ORE). Transfected PAP-HT25 cells were maintained in isoosmotic medium (Iso) (300 mosm/kg H 2 O) or exposed to hyperosmotic medium (Hyper) (500 mosm/kg H 2 O) for 18 -24 h. Values are expressed as the ratio of Hyper divided by Iso. ARL contains the rabbit AR promoter (bp Ϫ208/ϩ27) upstream of the luciferase gene. DNA fragments from the rabbit AR gene were inserted upstream of the AR promoter in ARL. Positions of nucleotides that define the AR gene DNA fragments are numbered with the first nucleotide of exon 1 as ϩ1. Negative numbers are nucleotides upstream of ϩ1. As indicated by an underline, bp Ϫ1102 was mutated from an A to a G and bp Ϫ1094 was mutated from a C to an A. Cells were co-transfected with a given luciferase construct and B19CAT. B19CAT contains the B19 promoter upstream of the CAT gene. Luciferase activity in relative light units/g of cell protein was normalized by CAT protein in picograms/g of cell protein (Luc/CAT). *Hyper/Iso is the ratio of Luc/CAT in hyperosmotic medium divided by Luc/CAT in isoosmotic medium (mean Ϯ standard error). **Hyper/Iso is the ratio of Luc/CAT in hyperosmotic medium divided by Luc/CAT in isoosmotic medium expressed relative to ARL (Hyper/Iso ϭ 2.4 Ϯ 0.09 and Table I) (mean Ϯ standard error); these data are shown only for constructs demonstrating hyperosmotic response.
[n] is the number of independent transfections.
By gradually eliminating base pairs from the 17-bp fragment (Ϫ1108/Ϫ1092) we continued to functionally define the minimal essential osmotic response element. As shown in Fig. 1, upstream bp Ϫ1108 to Ϫ1106 are unnecessary for osmotic response. However removal of bp Ϫ1105 eliminates osmotic response, thereby defining the upstream end of the ORE to be bp Ϫ1105. Relative to the downstream end of the element, construct Ϫ1108/Ϫ1094 retained osmotic response (Fig. 1). In addition, we had observed previously that when we subdivided the 47-bp fragment (Ϫ1117/Ϫ1071) into two pieces (Table I, ARL constructs Ϫ1120/Ϫ1096 and Ϫ1095/Ϫ1071), thereby splitting between bp Ϫ1096 and Ϫ1095, all osmotic response was lost. We concluded that bp Ϫ1095 was essential. If bp Ϫ1094, a cytidine (C), is substituted by an adenosine (A), osmotic response is unaffected. However, ARL construct Ϫ1108/ Ϫ1095 showed a Hyper/Iso response equal to 3.5 Ϯ 0.25 (1.5 Ϯ 0.10 relative to ARL, Fig. 1). We conclude that bp Ϫ1094 may be necessary, but it need not be a pyrimidine, and that the minimal essential ORE is defined by bp Ϫ1105/Ϫ1094.
We noted, particularly because of the concentration of purines at the 5Ј end of the ORE, the similarity between a nucleotide (nt) sequence (nt Ϫ1104/Ϫ1095; GGAAAATCAC) within it and the consensus sequence for the NF-B element (GGG(A/ G)NN(C/T)(C/T)(C/A/T)C) (21). However, the sequence Ϫ1104/ Ϫ1095 does not fit the NF-B consensus at nucleotide Ϫ1102 where all currently recognized NF-B elements have a guanosine (G) at the corresponding nucleotide (21). To determine whether the ORE would retain osmotic response if its sequence were modified to include an NF-B element consensus, we substituted base Ϫ1102, an adenosine (A), by a guanosine (G) (underlined in Fig. 1) as in an NF-B element. Osmotic response was lost (Fig. 1); we conclude that, at least based on cis-element sequence, the ORE does not contain an NF-B element.
The same experiment provided information relative to a putative ORE consensus. As shown below, the ORE shares six consecutive base pairs (Ϫ1104/Ϫ1099, GGAAAA) in common with TonE, the only other sequence identified as containing a eukaryotic osmotic response element (19). TonE regulates the hyperosmotic response of the dog renal Na ϩ -and Cl Ϫ -coupled betaine transporter responsible for accumulation of betaine, another organic osmolyte (22)(23)(24). In TonE, when five of the 6 bases (GAAAA) were simultaneously substituted by TCCCC, osmotic response was lost (19). By defining the minimal essential osmotic response element in the aldose reductase gene, we have shown that these 6 bases, although conserved, do not suffice to confer the response. However, as referred to above, we have also demonstrated that, substitution of only one of the six nucleotides (nt Ϫ1102 in the ORE), an adenosine (A), with another purine, a guanosine (G), eliminates osmotic response. Identification of other osmotic response elements should better define a eukaryotic osmotic response consensus, if there is one.
To test for interaction of the ORE with putative osmotically induced transcription factors, we performed electrophoretic mobility shift assays of the ORE in the presence of nuclear protein extracts prepared either from cells maintained in isoosmotic medium or cells exposed to hyperosmotic medium for 18 -24 h. We used bp Ϫ1108/Ϫ1092 as the probe. This fragment had been shown to confer osmotic response to the luciferase gene in transfection analyses (Table I). A slowly migrating but narrow band (arrow) was observed predominantly with extracts from hyperosmotically treated cells in the presence of 0.5 g of poly(dI-dC) (Fig. 2, lane 6). As shown in Fig. 2, a 50-fold molar excess of unlabeled (cold) probe reduces and a 100-fold molar excess virtually eliminates the narrow slowly migrating band (lanes 7 and 8, respectively). An additional, faster migrating and very broad band also remains in the presence of poly(dI-dC) but is eliminated by a 100-fold molar excess of the specific competitor (bp Ϫ1108/Ϫ1092). This band is also present in lanes containing extract from isoosmotically treated cells but to a lesser degree than that seen in lanes containing extract from hyperosmotically treated cells. These mobility shift assays represent exploratory evidence that hyperosmotic stress results in the induction of putative trans-acting factor(s) that associate with the aldose reductase ORE. Isolation of these trans-acting factor(s) and characterization of their interaction with the ORE should elucidate the basic mechanisms for regulation of gene expression by hyperosmotic stress.