Role of LFB3 in cell-specific cAMP induction of the urokinase-type plasminogen activator gene.

In previous work we suggested that a kidney-specific transcription factor LFB3 cooperates with cAMP-response element (CRE)-binding proteins within a cAMP regulatory unit comprised of three protein-binding domains and located 3.4 kilobase pairs upstream of the urokinase-type plasminogen activator (uPA) gene in LLC-PK cells (Menoud, P.-A., Matthies, R., Hofsteenge, J., and Nagamine, Y.(1993) Nucleic Acids Res. 21, 1845-1852). The two domains contain a CRE-like sequence, and the third domain is recognized by LFB3. The absolute requirement of LFB3 as well as the cooperation among the three domains for cAMP regulation were confirmed by transient transfection assays in F9 teratocarcinoma cells, in which the level of LFB3 was negligible. Suspecting a possible feedback regulation of LFB3 mRNA expression during cAMP-dependent uPA gene induction in LLC-PK cells, we measured LFB3 mRNA levels after cAMP treatment and found a strong reduction. This reduction was not due to a change in template activity of the LFB3 gene because run-on transcription showed no significant change in LFB3 gene transcription. RNA synthesis inhibitor-chase experiments indicated that the down-regulation was post-transcriptional. Interestingly, when the inhibitor was added at the same time as cAMP, the cAMP-induced decrease in LFB3 mRNA levels was abrogated, suggesting that on-going RNA synthesis is required for the decrease. Similar effects on LFB3 mRNA metabolism were observed with all agents that induce uPA mRNA in LLC-PK cells, including 12-O-tetradecanoylphorbol-13-acetate, okadaic acid, colchicine, and cytochalasin. We discuss the significance of this regulation in uPA gene expression.

Signal transduction, a process of successive activation of various molecules, is subject to various levels of regulation. In many cases, for the sake of homeostasis, activated molecules are sequestered from the pathway by desensitization of membrane-bound receptors (1,2), degradation of activated molecules (3,4), inactivation of activated molecules by dephosphorylation (5,6), or by a feedback mechanism (7). Cross-talk between different signaling pathways is also an important mechanism for bestowing flexible and versatile regulation on a given pathway. This can be either positive or negative and occurs at various steps in the pathway in a cell-specific manner (for reviews, see Refs. 8 -10). Therefore, in addition to the identification of successively activated components of a signaling pathway and the elucidation of the mechanism of activation of each component, it is also very important to know how the activity of each component is modulated by molecules not immediately upstream in the pathway. In this way, the nature of a signaling pathway may be understood in a more physiologically relevant context.
We have been studying urokinase-type plasminogen activator (uPA) 1 gene regulation in LLC-PK 1 cells, a cell line derived from pig kidney epithelia (11). In these cells, the uPA gene is induced through independent signaling pathways by various signals such as cAMP (12), 12-O-tetradecanoylphorbol-13-acetate (TPA) (13), the protein phosphatase 1/2A inhibitor okadaic acid (14,15), and cytoskeletal reorganization (13,16). The pig uPA gene has a cAMP-inducible enhancer located 3.4 kb upstream of the transcription start site (12). This enhancer is comprised of three protein-binding domains, A, B, and C. Domains A and B contain a core sequence of the cAMP response element (CRE) but require the adjoining C domain to confer full cAMP responsiveness on a heterologous promoter (12,17). The C domain has no CRE and cannot mediate cAMP responsiveness when used in isolation. We have purified the protein binding to the C domain (17) and found it to be the pig equivalent of mouse LFB3 (18). It is also known as HNF1␤ (19) or vHNF1 (20). LFB3 is a tissue-specific transcription factor highly expressed in kidney cells (18) with a structure closely related to the liver-specific transcription factor HNF1␣. Both HNF1␣ and LFB3 recognize the same DNA sequence, at least in vitro (17,21), although the domain C sequence is quite different from the consensus HNF1␣ recognition sequence. It is still not known which genes besides the uPA gene are the targets of LFB3 in kidney cells, or how the expression of LFB3 is regulated. As LFB3 is apparently involved in cAMP-dependent uPA gene regulation in LLC-PK 1 cells, we were interested to know whether cAMP-evoked signaling affected the expression of LFB3 in these cells. Indeed, we have shown that cAMP treatment strongly reduces LFB3 mRNA levels, suggesting a feedback mechanism via LFB3 in cAMP-dependent uPA gene regulation in LLC-PK1 cells (17). In the present study, we verify the involvement of LFB3 in cAMP-induction of the uPA gene and show that not only cAMP but also other agents that induce uPA gene expression strongly reduce the amount of LFB3 mRNA. These agents are 12-O-tetradecanoylphorbol-13acetate, okadaic acid, colchicine, and cytochalasin B. Our results suggest the involvement of LFB3 in uPA gene regulation by cAMP at different levels.

MATERIALS AND METHODS
Reagents-TPA, colchicine, and cytochalasin B were obtained from Sigma; 5,6-dichloro-1-␤-D-ribofuranosylbenzimidazole (DRB) from Fluka; 8-bromo-cAMP (Br-cAMP) from Boehringer Mannheim; and okadaic acid from Anawa. [␣-32 P]dATP (3000 Ci/mmol) was obtained from * 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.
Expression Vectors-In pTATA (constructed and provided by A. E. Sippel) the firefly luciferase gene is linked to a minimal promoter of the thymidine kinase gene (Ϫ46 to ϩ52) containing only the TATA box and the transcription initiation site. Mutated and nonmutated sequences derived from the cAMP-responsive enhancer ABC site, which is located 3.4 kb upstream of the transcription initiation site of the uPA gene, were inserted immediately 5Ј of the TATA box of pTATA (Fig. 1a). Similar constructs with nonmutated sequences, but with the SV40 early gene promoter, and the pig LFB3 expression vector have been described previously (17).
Transient Transfection Assays-Cells were seeded at 3 ϫ 10 5 (LLC-PK 1 ) or 2 ϫ 10 5 (F9) per 35-mm plate the day before transfection. Cells were transfected by calcium phosphate-mediated precipitation with 1-3 g of DNA. When cells were to be induced, cells were treated 20 h after transfection with 1 mM Br-cAMP for 6 h. Cell extracts were assayed for luciferase activity as described elsewhere (17) using a Lumiometer (Autolumat LB 953, Berthold).
RNA Isolation and Northern Blot Analysis-Total RNA was isolated according to Chomczynski and Sacchi (25) and analyzed for levels of specific mRNAs by Northern blot hybridization as described (23). To confirm the loading and transfer of similar amounts of RNA, ribosomal RNA was visualized on nylon filters by staining with methylene blue (26). After hybridization, filters were exposed to Kodak X-Omat AR film with an intensifying screen at Ϫ70°C. Levels of specific RNA were quantitated using a Molecular Dynamics PhosphorImager.
Determination of mRNA Stability-RNA stability was measured by the RNA synthesis inhibitor-chase method as described (27). Briefly, cells were treated with DRB (20 g/ml) to inhibit transcription, and total RNA was isolated at several subsequent time points. RNA was analyzed by Northern blot hybridization. mRNA levels were plotted using SigmaPlot (Jandel Scientific) and subjected to linear regression.
Nuclear Transcription-The isolation of nuclei, nuclear run-on transcription, and quantitation of specific transcripts by hybridization were performed as described previously (28).
Nuclear Extracts and Electrophoretic Mobility Shift Assays-Nuclear extracts were prepared from LLC-PK 1 cells and electrophoretic mobility shift assays were performed as described previously (13).

Cooperation of LFB3 in a cAMP-responsive Enhancer-A
cAMP-inducible enhancer located 3.4 kb upstream of the transcription initiation site of the uPA gene is composed of three protein binding domains, A, B, and C; domain C lacks a CRE sequence and is necessary together with domains A and B for full cAMP-inducible activity (12). Cooperation between domains A and B and domain C was reevaluated in the context of the minimum thymidine kinase gene promoter containing only a TATA box by transient transfection assays (Fig. 1b). To activate cAMP-dependent signaling, we used Br-cAMP or an expression vector of the catalytic subunit of cAMP-dependent protein kinase. As shown in Fig. 1b, templates with domains A and B alone or domain C alone did not exhibit significant inducibility compared with the control template, while the template with all three domains (pABC-TATA) showed strong inducibility. This inducibility was strongly reduced when the template pABC-TATA was mutated in any of the three do-mains. The mutations were introduced to the sites that had been shown to interact with nuclear proteins by methylation interference experiments (12). The induction with Br-cAMP elicited a stronger response than with the catalytic subunit, which may be due to a high concentration of free regulatory subunits in LLC-PK 1 cells.
We previously cloned the domain C-binding protein and found it to be the pig equivalent of mouse LFB3 (17). We therefore examined the effect of LFB3 on the above templates by transient coexpression assays in F9 cells, which have a negligible level of endogenous LFB3. We used only the catalytic subunit to activate the signaling because endogenous cAMPdependent protein kinase is not responsive to cAMP in F9 cells by an unknown mechanism (29). Fig. 1c shows that in F9 cells pABC-TATA was strongly induced by the catalytic subunit only when LFB3 was coexpressed. The control pTATA was not affected. These results unambiguously indicate the cooperation among three protein-binding domains and the involvement of LFB3 in cAMP regulation through the ABC site.
Effect of cAMP and Other uPA Inducers on LFB3 mRNA Levels-To confirm the previous observation that cAMP treatment reduces LFB3 mRNA in LLC-PK 1 cells, we compared Br-cAMP to other agents shown to induce uPA mRNA in the same cells, TPA, colchicine, cytochalasin B, and okadaic acid. The cells were incubated for the time optimal for uPA mRNA induction, i.e. 2 h for Br-cAMP and TPA and 4 h for the rest. As shown in Fig. 2, Br-cAMP as well as other agents strongly reduced LFB3 mRNA levels; all of them induced uPA mRNA. The greatest reduction in LFB3 mRNA levels was obtained with TPA and okadaic acid (85% by 2 and 4 h) and the least with cytochalasin B (60% by 4 h).
With the exception of Br-cAMP, all the other agents induce uPA gene via the activation of AP1, acting on the PEA3/AP1 site located 2 kb upstream of the transcription initiation site (13,15). Therefore, in the following experiments we compared in particular Br-cAMP and TPA.
Effect of Br-cAMP and TPA on Domain C Binding Activity-We tested whether the reduction of LFB3 mRNA levels after treatment with uPA inducers was reflected at the protein level. As specific antibodies against pig LFB3 were not available, we measured domain C binding activity in nuclear extracts. We performed electrophoretic mobility shift assays using crude nuclear extracts prepared from LLC-PK 1 cells pretreated for 7 h with Br-cAMP or TPA. Using a 32 P-labeled domain C oligonucleotide as a probe, we observed a single distinct band (Fig. 3), which could be competed by excess of the identical unlabeled oligonucleotide but not by an oligonucleotide carrying the same mutations as shown in Fig. 1a (data not shown). After treatment of the cells with TPA, the binding activity was reduced to about 50%, with Br-cAMP to 40% and with TPA and Br-cAMP added together to about 30%. These data indicate that the LFB3 binding activity is reduced by Br-cAMP or TPA treatment, suggesting that the reduced mRNA level affects the protein level. To see whether the observed reduction is specific for domain C binding, we tested other oligonucleotides recognized by different transcription factors using the same nuclear extracts. Although domains A and B were required for cAMP induction, proteins binding to these sites were not affected by treatment of the cells with Br-cAMP or TPA. With the SP1 oligonucleotide, two major specific bands were detected, but they did not change in intensity after this treatment. In contrast, with the mPEA3/AP1 oligonucleotide, which binds transcription factor AP1 (13), the intensity of the shifted band markedly increased on treatment with TPA and even more with Br-cAMP.
No Change in Transcription Rate of the LFB3 Gene-The mechanism leading to the reduction of LFB3 mRNA by uPA inducers may involve either transcriptional or post-transcriptional regulation of the LFB3 gene. To distinguish between these two possibilities, we first performed nuclear run-on transcription to assess changes in the LFB3 gene transcription rate. The results shown in Fig. 4 indicate that the transcription rate of the LFB3 gene did not change when cells were treated with TPA, Br-cAMP, colchicine, or cytochalasin B. As expected, these agents significantly enhanced the uPA gene transcription rate. Thus, the decrease in LFB3 mRNA levels seems not to be due to decreased de novo synthesis of LFB3 mRNA, suggesting that the reduction of LFB3 mRNA is a post-transcriptional event.
Induced Instability of LFB3 mRNA-If a post-transcriptional step is responsible for the induced decrease in LFB3 mRNA level, the most obvious mechanism could be an effect on mRNA stability. The stability of LFB3 mRNA was assessed by DRB-chase experiments. Since DRB specifically inhibits the synthesis of eukaryotic heterogeneous nuclear RNA and mRNA (30), chase of mRNA levels after its addition allows estimation of the decay rate of the mRNA. Because inhibition of mRNA synthesis may have some indirect influence on mRNA stability (31), we did chase experiments using two different schemes: in one experiment DRB was added at the same time as Br-cAMP or TPA, and in the other DRB was added 1 h after Br-cAMP or TPA treatment. The effect on LFB3 mRNA was independent of the presence of Br-cAMP when DRB was added at the beginning of the experiment (Fig. 5a). However, LFB3 mRNA decayed faster in the presence of Br-cAMP when DRB was added 1 h after Br-cAMP (Fig. 5b). Similar results were obtained using TPA (Fig. 5, c and d). These results indicate that the stability of LFB3 mRNA is reduced by uPA inducers, and that this requires on-going RNA synthesis for at least 1 h at the beginning of the treatment.
Effect of the Decrease in DNA Binding Activity of LFB3 on cAMP Induction-TPA and cAMP treatment reduced the DNA binding activity of LFB3. To test the biological relevance of this decrease in cAMP induction we asked whether TPA pretreatment could affect the cAMP-induction of pABC-TATA. As shown in Fig. 6, TPA pretreatment by itself had little effect on  3. Domain C binding activity. Domain C binding activity in the nucleus was tested by electromobility shift assays using crude nuclear extracts from LLC-PK 1 cells pretreated for 7 h with 1 mM Br-cAMP, 100 ng/ml TPA, or both. As controls, the same extracts were tested for domains A and B, mPEA3/AP1, and SP1 binding activities.  4. Nuclear run-on transcription. Nuclei were isolated from LLC-PK 1 cells untreated or pretreated with various agents for 90 min. Nuclear transcription was performed in the presence of a radioactive precursor and specific transcripts were analyzed by filter hybridization. basal expression but significantly reduced cAMP induction of the luciferase gene driven by ABC sites.

DISCUSSION
LFB3 is an enhancer-binding protein augmenting basal expression of a gene that contains its cognate cis-element. We found in the induction of the uPA gene by cAMP in LLC-PK 1 cells that LFB3 is a positive regulator cooperating with CREbinding proteins within a composite cAMP-responsive enhancer (17) (this work). Our results also suggest that LFB3 is involved in a down-regulating phase of cAMP-induced uPA gene expression. We have previously shown that uPA gene induction by cAMP is transient; the rate of uPA gene transcription reaches optimal after 2-4 h of cAMP treatment but declines thereafter (32). It may be that in uPA gene regulation LFB3 acts as a negative feedback regulator by decreasing its own concentration in response to cAMP. This throws new light on LFB3, which has been implicated as a factor coupling hormonal regulation and tissue-specific regulation of uPA gene expression in kidney epithelial cells (17).
The decrease in domain C binding activity seems to be due to a decrease in LFB3 protein levels. The decrease was also observed with TPA, and it may also be the case for colchicine, cytochalasin B, and okadaic acid, which all decreased LFB3 mRNA levels (see below). These agents induce uPA gene expression in LLC-PK1 cells via activation of the transcription factor AP1, although the mechanism of AP1 activation by each agent is different (13)(14)(15). Thus, in addition to the features mentioned above, LFB3 may mediate negative cross-talk between cAMP-dependent signaling and AP1-activating signaling pathways in uPA gene regulation. Indeed, pretreatment with TPA significantly reduced cAMP induction of the luciferase gene driven by an enhancer consisting of domains A, B, and C. The decrease in DNA binding by LFB3 in the cells seems to be due to the reduction in the protein levels. We cannot formally exclude the possibility that the decrease is due to a post-translational modification of the protein; however, this is in any case not the main cause because we also detected a strong reduction in LFB3 mRNA levels. The possible role of LFB3 in cAMP-dependent uPA gene regulation through the ABC site revealed by this work is summarized in Fig. 7.
The decrease in DNA-binding activity evoked by treatment with the uPA inducers in these cells was specific to the domain C binding protein, LFB3, and not a general effect, because DNA binding of the proteins recognizing domains A and B and of the ubiquitous transcription factor SP1 remained constant. Furthermore, the DNA-binding activity to the mutated PEA3/AP1oligonucleotide, which contains an active AP1 site mediating the action of TPA, colchicine, cytochalasin and okadaic acid, was increased by Br-cAMP as well as by TPA. We have not elaborated the mechanism of the increase in PEA3/AP1-binding activity, i.e. whether it is transcriptional or post-transcriptional. It is worthwhile to mention that the peptide hormone calcitonin, which raises intracellular cAMP concentrations, strongly enhances de novo synthesis of c-Fos and c-Jun (13), raising the interesting possibility of a cross-regulation of the TPA-dependent signaling pathway by the cAMP-dependent signaling pathway at the transcription step. The cAMP signal by itself does not utilize the PEA3/AP1 site to increase uPA gene expression (13). We do not know yet whether the enhancement of c-Fos together with c-Jun levels exerts positive effects on PEA3/AP1 site-mediated uPA gene expression, because the overexpression of c-Fos had no effect on uPA gene induction in NIH3T3 cells (33).
The decrease in LFB3 mRNA levels is mainly attributable to induced mRNA instability. We did not detect changes in the LFB3 gene transcription rate, but we did observe that LFB3 mRNA degradation increased in the presence of TPA or Br-cAMP. Interestingly, however, enhanced instability was observed only when DRB was added 1 h after TPA or Br-cAMP treatment, suggesting that some RNA transcripts or their translation products are involved in LFB3 mRNA metabolism. It may be that TPA or Br-cAMP induces a factor, RNA or protein essential for LFB3 mRNA degradation, or that an RNA or a protein of short half-life is involved in LFB3 mRNA degradation, at least at an early stage. A requirement for on-going RNA synthesis in mRNA degradation has been reported for several mRNAs, such as those for c-fos, (34), c-myc (35), collagenase (36), and the transferrin receptor (37). We have shown that an RNA instability-regulating site in the 3Ј-UTR of uPA mRNA requires on-going RNA synthesis for its activity (31) and that the importance of this site in overall uPA mRNA degradation may depend on cell type (38). In none of these cases is it known how on-going RNA synthesis contributes to mRNA degradation.
Several instability-determining sequences have been identified in many mRNAs. These include sequences located in the 3Ј-untranslated region, such as the iron-responsive element in the transferrin receptor mRNA (37,39), sequences in the un- it also mediates a negative feedback regulation by cAMP and TPA by decreasing its protein levels, which is due to enhanced degradation of LFB3 mRNA. stable yeast MFA2 mRNA (38) and AU-rich sequences in various oncogene and lymphokine mRNAs (40 -44). But instability-determining elements have also been identified in coding regions, e.g. c-myc (35) and c-fos (45) mRNAs. We tested the 3Ј-UTR and protein-coding regions of LFB3 mRNA in a system developed for the study of uPA mRNA degradation by inserting these sequences in an otherwise stable globin mRNA (31); however, the stability of recombinant globin mRNAs was not affected by TPA or Br-cAMP. 2 It may be that regulatory sequences reside in 5Ј-UTR or the 3Ј extreme which we have not tested or that the globin mRNA context interfered with TPAand Br-cAMP-induced mRNA degradation.
Whether the cAMP and TPA signals utilize the same mechanism to induce LFB3 mRNA destabilization is not yet established, although it is plausible considering that induced LFB3 mRNA instability by either agent requires ongoing RNA synthesis and that signal transductions induced by the two agents are related. In the cell, cAMP and TPA activate distinct signaling pathways but are otherwise quite related. Both agents trigger signaling by activating serine/threonine kinases, and the transcription factors that are eventually activated by these signals are also related; the cAMP and TPA signals activate CREB/ATF and AP1, respectively, which are highly related transcription factors containing basic/leucine zipper domains, recognize highly similar sequences, and can cross-dimerize (for reviews, see Refs. 8 and 46). A protein responsible for induced LFB3 mRNA degradation could be phosphorylated and regulated by cAMP-dependent protein kinase as well as by protein kinase C. Alternatively, the two different but related transcription factors may exert their effects at a post-transcriptional step by interacting with the same RNA sequence or RNAbinding protein. It should be remembered that colchicine, cytochalasin B, and okadaic acid also reduce LFB3 mRNA (Fig. 2) and that these agents do not require protein kinase C to activate AP1 and induce the uPA gene (13,15). Identification of regulatory sequences in LFB3 mRNA and the corresponding binding proteins should help answer these questions.
We have shown that uPA inducers reduce LFB3 mRNA levels. Is there any physiological significance in this apparent linkage, or is this reverse regulation fortuitous, using very common signaling pathways? uPA is a secreted protease which plays an important role in various extracellular proteolytic processes (for reviews, see Refs. [47][48][49][50], but its unchecked expression may have deleterious effects on producing organs or nearby organs (37). As LFB3 is an abundant transcription factor in kidney (18) 3 and is involved in uPA gene regulation, it may have evolved so that the kidney cells use LFB3 as one means to control the level of uPA expression.