Excitotoxic Brain Lesion Modifies Binding to a USF Binding Site Acting as a Negative Regulatory Element in theProtease Nexin-1Promoter

MCN

Molecular and Cellular Neuroscience 8, 28–37 (1996) Article No. 0041

Excitotoxic Brain Lesion Modifies Binding to a USF Binding Site Acting as a Negative Regulatory Element in the Protease Nexin-1 Promoter Henrik Ernø,1,2 Patrick Ku¨ry,2 Cordula Nitsch,* Jean-Pierre Jost, and Denis Monard3 Friedrich Miescher-Institut, P. O. Box 2543, CH-4002 Basel, Switzerland; and *Institute of Anatomy, University of Basel, Pestalozzistrasse 20, CH-4056 Basel, Switzerland

The expression of the serine protease inhibitor Protease nexin-1 (PN-1) is upregulated in glial cells following different types of lesion in the nervous system. A strong negative regulatory element has been shown by the missing nucleoside technique to be a CACGTG site (E-box) in the proximal part of the PN-1 promoter. The factor binding to this site is specifically recognized by antibodies directed against the human upstream stimulatory factor (USF). Point mutations in the E-box binding site which abolish USF binding in vitro increase the transcriptional activity of the PN-1 promoter. Cotransfection of a PN-1 promoter/reporter construct together with an expression vector for human USF1 confirms the negative regulatory function of this site. Finally, we show that the binding to this USF site changed after ibotenic acid-induced lesion of the caudate putamen in the rat brain.

INTRODUCTION Protease nexin-1 (PN-1), also known as glia-derived nexin (GDN), is a potent serine protease inhibitor. It was purified as a thrombin binding protein and as a factor promoting neurite outgrowth in mouse neuroblastoma cells as well as in cultured chick sympathetic and rat hippocampal neurons (Baker et al., 1980; Farmer et al., 1990; Guenther et al., 1985; Gurwitz and Cunningham, 1988; Monard et al., 1973; Zurn et al., 1988). The amino acid sequence shows domains of homology with other 1 Present address: Ecole Normale Supe ´ rieure, CNRS URA 1414, 46, rue d’Ulm, 75005 Paris, France. 2 These authors contributed equally to this study and are listed alphabetically. 3 To whom correspondence should be addressed. Fax: (061) 697 39 76. E-mail: [email protected].

28

serine protease inhibitors especially plasminogen activator inhibitors 1 and 2 (Gloor et al., 1986; McGrogan et al., 1988; Sommer et al., 1987). PN-1 inhibits the activities of thrombin, urokinase, tissue plasminogen activator, and trypsin (Stone et al., 1987). Since thrombin has been shown to cause neurite retraction through cleavage and activation of its receptor (Suidan et al., 1992), it seems likely that PN-1 promotes neurite outgrowth by inhibiting thrombin. In vivo, PN-1 is expressed at high levels in the brain at specific developmental stages (Mansuy et al., 1993; Reinhard et al., 1994), the highest mRNA expression being detected before Postnatal Day 12 (Gloor et al., 1986). However, in the olfactory system, a structure in which neuronal degeneration and regeneration continue throughout life, PN-1 expression remains high (Reinhard et al., 1988). PN-1 is upregulated at both the mRNA and the protein levels following lesion of the rat sciatic nerve (Meier et al., 1989). PN-1 protein increases after ischemia-induced death of hippocampal pyramidal neurons (Hoffmann et al., 1992). This upregulation remains detectable even 1 year after ischemia (Nitsch and Schaefer, 1990; Nitsch et al., 1993). Ibotenic acid or 6-hydroxydopamine injected in the substantia nigra (SN) have been shown to induce PN-1 expression in the astrocytes of the caudate putamen (CP), probably as a consequence of nerve terminal degeneration (Scotti et al., 1994). Ibotenic acid injected in the CP also increased PN-1 protein in the astrocytes of this region (C. Nitsch, unpublished results). These findings indicate that the regulation of PN-1 expression is of importance both during development and in pathological situations. The promoter of the rat PN-1 gene has recently been characterized (Ernø and Monard, 1993). It is highly GC 1044-7431/96 $18.00 Copyright r 1996 by Academic Press, Inc. All rights of reproduction in any form reserved.

29

Changes in USF Binding Site upon Ibotenic Acid Lesion

rich and contains binding sites for several transcription factors, such as Sp1, NGFI-A, NGFI-C, Krox-20, Wilms tumor factor (WT1), and E-box binding proteins. Here we demonstrate that a ubiquitous transcription factor binds to a CACGTG site in the proximal part of the PN-1 promoter. Several factors, including upstream stimulatory factor (USF; Gregor et al., 1990; Sawadogo, 1988; Sawadogo and Roeder, 1985; Sawadogo et al., 1988), members of the Myc family (Kerkhoff et al., 1991), members of the MyoD family (Davis et al., 1990), and other E-box binding proteins (Emerson, 1990), have been shown to bind to a CANNTG site. The factor identified in this study is solely recognized by specific anti-USF antibodies, thus defining the negative regulatory element as a USF binding site. The proximal promoter region was further analyzed for the presence of sites able to bind factors which would be regulated in response to a lesion. A specific change in the gel-shift pattern of proteins binding to the USF site was detected following ibotenic acid-induced rat brain lesion, a condition under which PN-1 is upregulated. This shows for the first time a change in a protein(s) binding to a USF site after lesion in the nervous system and suggests a role for this site in the subsequent in vivo upregulation of PN-1 synthesis.

RESULTS The CACGTG Site at -275 Binds a Repressor Protein To determine which transcription factors regulate the PN-1 promoter a deletion analysis and a series of gel-shift assays using various fragments of the proximal promoter sequence were performed (Ernø, 1994). These preliminary results suggested that a repressor protein(s) binds to the 2383 to 2238 fragment (see Fig. 2A; numbering according to Ernø and Monard, 1993). To determine the binding site(s) of the protein(s) the fragment was analyzed by the missing nucleoside technique using free hydroxyl radical chemistry. The end-labeled fragment was treated with free radicals as described under Experimental Methods; a gel-shift assay was then performed and the retarded band and the free probe were isolated (Fig. 1A). The DNA was purified and analyzed on a sequencing gel. The fragment was first labeled at the XhoI site at 2239; only one region from 2280 to 2273 was found to interact with proteins (Fig. 1B). The same result was obtained when the fragment was labeled at the SalI site at 2383; again only one area within the fragment was found to bind protein(s) (data not shown). The missing nucleosides CCACGTGC (2280 to 2273) contained the core consensus binding site CANNTG, called an E-box, for the family of basic

FIG. 1. Missing nucleoside experiment with the silencer fragment (2383 to 2238). (A) Gel-shift assay using the free-radical-treated silencer fragment using 20 µg of the 0.4 M KCl heparin–Sepharose fraction from C6 rat glioma nuclear extract. The bars b and f show the parts of the gel from which the DNA was eluted and subsequently run on the sequencing gel in B. (B) The silencer fragment end-labeled at the downstream XhoI site (2238). Lanes b, The silencer fragment with bound repressor protein(s); lane f, free (unbound) silencer fragment; and lane c, control fragment directly after free radical treatment.

helix-loop-helix (bHLH) transcription factors (Davis et al., 1990; Emerson, 1990).

The CACGTG Binding Protein Is a Negative Regulator of Protease Nexin-1 Expression In order to be sure that the CCACGTGC sequence is a protein binding site, and to investigate the function of this site, three constructs containing PN-1 promoter fragments controlling the luciferase gene were transiently transfected into different cell lines and primary cells. The constructs and the sequence of the fragment 2383 to 2238 are outlined in Fig. 2A. As shown for the transfected C6 rat glioma cells (Fig. 2B) mutation of the E-box binding site from CACGTG to AACGTT resulted in an approximately sevenfold increase, while deletion of the 2383 to 2238 region resulted in a fourfold increase of luciferase levels. This demonstrates that the CCACGTGC sequence protected in the missing nucleotide experiments (Fig. 1) is binding a repressor protein(s). Transfection of Rat-1 fibroblasts as well as several other cell lines showed similar or identical effects (results not shown).

30

Ernø et al.

FIG. 2. (A) Outline of the PN-1 promoter constructs. The PN-1 promoter fragment from 2383 to 1105 (designated p500 for the wildtype or p500mut for the mutant E-box) and the promoter fragment from 2238 to 1105 (with the silencer deleted and designated p350) were cloned in front of the luciferase gene (hatched box) using the SalI and BamHI sites. The sequence of the silencer fragment (2383 to 2238) is given below; the E-box is in boldface and underlined. (B,C) Effect of mutations in the USF binding site (CACGTG to AACGTT) of the wildtype PN-1 promoter compared to deletion of the whole fragment 2383 to 2238. The plasmids were transiently transfected in C6 rat glioma cells (B) and in cultured primary rat astrocytes (C). (D) Gel-shift assay of the 2383 to 2238 fragment with or without the AACGTT mutation in the USF-binding site. Lanes 1 and 2, wildtype silencer fragment; lanes 3 and 4, with the AACGTT mutation; lanes 1 and 3, without nuclear extract; lanes 2 and 4, with crude C6 rat glioma nuclear extract (2 µg). (E) The silencer fragment represses gene expression from a heterologous promoter. The SV40 early promoter (SV40) and two constructs with one (sil-SV40) or three (3sil-SV40) copies of the silencer fragment cloned upstream of the SV40 early promoter were transiently transfected in C6 rat glioma cells. Transfection experiments were performed in triplicate; activities are given in counts of light units per minute. The standard deviation is shown by the error bars above the columns.

To determine whether the CACGTG site would also bind a repressor protein in primary cells, cultures of primary rat astrocytes (prepared as described under Experimental Methods) were transfected with the same constructs. As seen in Fig. 2C the AACGTT mutation led

to a fourfold increase and the deletion of the silencer region to an approximately threefold increase in reporter gene activity. This shows that the CACGTG site binds a repressor not only in established cell lines but also in primary astrocytes.

31

Changes in USF Binding Site upon Ibotenic Acid Lesion

The effect of the CACGTG to AACGTT mutation on protein binding was tested by gel-shift assay using the silencer fragment (2383 to 2238); protein binding to the fragment was completely abolished by the mutation (Fig. 2D). The point mutations did not generate new protein binding activity. To determine whether this fragment (2383 to 2238) was also able to repress transcription from a heterologous promoter, it was cloned in one copy (sil-SV40) and in three copies (3sil-SV40) upstream of the SV40 early promoter in the pPALU plasmid (Artelt et al., 1991). This resulted in a 82 (sil-SV40) and 93% (3sil-SV40) reduction of luciferase expression compared to that obtained with the normal SV40 early promoter in transiently transfected C6 rat glioma cells (Fig. 2E) and Rat-1 fibroblasts (data not shown). The silencer fragment did not have any cellspecific effect since the extent of the reduction was the same in both types of cells, although there is 20-fold more PN-1 mRNA in C6 rat glioma cells (Ernø and Monard, 1993). Taken together, these results indicate that the negative regulatory effect must be attributed to the CACGTG binding protein(s). To characterize this repressor protein(s), we tested nuclear extracts from C6 rat glioma cells fractionated on a heparin–Sepharose column by stepwise elution with increasing KCl concentrations. Each fraction was concentrated using centricon microconcentrators and tested for binding activity by gel-shift assays. The binding protein(s) was predominantly found in the 0.4 M KCl fraction (data not shown). UV crosslinking of nuclear proteins of this 0.4 M KCl fraction to oligonucleotides with the E-box binding site, followed by SDS–PAGE was used to demonstrate the presence of two specific DNA– protein complexes of approximately 55 kDa (Fig. 3). No

FIG. 4. Gel-shift assays in the presence of antibodies. (A) Crude nuclear extract from C6 rat glioma cells was incubated with the following antibodies prior to the DNA binding reaction to doublestranded USF wt oligonucleotide: Lanes 1, No antibody; 2, anti-c-Myc (9E10); 3, anti-USF (No. 52.3, recognizing USF43); 4, anti-MyoD (N-terminus); 5, anti-MyoD (bHLH-domain); 6, anti-MyoD (Cterminus); 7, nonspecific IgG. (B) Gel-shift assay with immunodepleted crude nuclear extract from C6 rat glioma cells binding to USF wt oligonucleotide. The nuclear extract was incubated as described with PBS (lane 1), rabbit serum (lane 2), and rabbit anti-USF serum No. 52.3 (lane 3). (C) Supershift of repressor protein–DNA complex. Gel-shift assay of crude nuclear extract from C6 rat glioma cells in the absence (lane 1) and presence (lane 2) of rabbit anti-USF antibody (sc-229X, Santa Cruz Biotechnology, recognizing USF43 and USF44) binding to USF wt oligonucleotide. The arrow shows the supershifted band. (D) Immunoblot of nuclear extracts from C6 rat glioma cells (lane 1), whole rat brain (lane 2), and HeLa cells (lane 3) using the rabbit anti-USF serum No. 52.3.

specific DNA–protein complexes could be detected using the 0.2 M KCl or other fractions devoid of the repressor protein(s) (data not shown).

The Negative Regulator Is Recognized by Anti-USF Antibodies

FIG. 3. UV crosslinking of nuclear proteins from C6 rat glioma cells to USF wt oligonucleotide (sequence in legend for Fig. 5). The repressor protein(s) partially purified by heparin–Sepharose chromatography was crosslinked by UV irradiation to the double-stranded and end-labeled USF wt oligonucleotide and separated by SDS–PAGE (lane 1), as described under Experimental Methods. A 30-fold excess of unlabeled oligonucleotide (USF wt) was added as competitor (lane 2). Arrows show the presence of two specific DNA–protein complexes.

The E-box consensus site binds several different proteins, all belonging to a family of transcription factors of the basic helix-loop-helix (bHLH) type. USF (Gregor et al., 1990), c-Myc (Kerkhoff et al., 1991), and members of the MyoD family (Davis et al., 1990) are all known to bind to the CACGTG site identified here as a repressor binding site in the PN-1 promoter sequence. To determine the identity of the repressor protein(s) we used antibodies directed against several different bHLH proteins. Gel-shift experiments showed that preincubation of crude nuclear extracts from C6 rat glioma cells with antibodies directed against human USF (Fig. 4A,

32

Ernø et al.

lane 3), but not with antibodies against c-Myc and MyoD (Fig. 4A, lanes 2, 4, 5, and 6), either inhibited the binding of the repressor protein(s) to the E-box binding site or ‘‘supershifted’’ the repressor protein–DNA complex (Fig. 4C). Furthermore, crude C6 rat glioma cell nuclear extract immunodepleted by the rabbit USF antiserum revealed no specific binding activity while extract depleted by rabbit control serum retained the binding activity (Fig. 4B, lanes 3 and 2). Immunoblot analysis using rabbit USF antiserum (raised against USF43) showed that crude nuclear extracts from C6 rat glioma cells, rat brain cells, and HeLa cells contain a protein of the expected 43-kDa molecular weight (Fig. 4D). In addition, this protein was enriched in the heparin–Sepharose 0.4 M KCl fraction of the crude C6 rat glioma nuclear extract (data not shown). In summary, our findings support the identification of rat USF or a closely related protein as the protein binding in vitro to the CACGTG site in the PN-1 promoter.

Mutations in the USF Binding Site Prevents Protein Interaction The retarded DNA–protein complex present in crude nuclear extracts prepared from C6 rat glioma cells could be competed away by oligonucleotides containing the CACGTG sequence (Fig. 5A, USF wildtype (wt), but not by oligonucleotides containing a consensus binding site for the transcription factor Sp1 (Fig. 5A, Sp1). Oligonucleotides with a binding site that changed from CACGTG to AACGTT could not compete with the wildtype sequence (Fig. 5A, USF mt1) nor did they bind any protein (data not shown), thus establishing that these two nucleotides are necessary for binding of the protein. The two single substitutions AACGTG or CACGTT also reduced the competition effect although to a lesser degree than did the double mutation (Fig. 5A, USF mt2 and mt3). Substitution of CACGTG to CACTTG also reduced the competition effect (Fig. 5A, USF mt5) as did a CCACGTG to ACACGTG substitution (Fig. 5A, USF mt4). This latter effect showed that nucleotides adjacent to the CACGTG consensus also influence the binding of the protein. Oligonucleotides with substitutions outside the protected CCACGTGC sequence did compete for binding (data not shown). To further test if USF represses transcription from the PN-1 promoter we performed cotransfection assays in C6 rat glioma cells using an expression vector containing the human USF1 (USF43) cDNA. Cotransfection of this vector together with a plasmid containing the CAT reporter gene driven by a synthetic pyruvate kinase

FIG. 5. (A) Binding competition experiment. Crude nuclear extract (2 µg) from C6 rat glioma cells was incubated with 0.1 pmol of 32P-labeled silencer fragment (2383 to 2238) in the presence of 5 pmol of double-stranded competitor oligonucleotides with wildtype or mutated USF binding sites. The bands from a gel-shift assay (shown at the top) were quantified using a PhosphorImager (Molecular Dynamics); the counts are indicated above the columns in the histogram. The first slot on the left represents an incubation without nuclear extract. Upper strands of competitor oligonucleotides (mutated bases are in boldface): Sp1, GAAGGGCGGGGGCGGCGGCGG; USF wt, CCGGACGACCCACGTGCAGCTCTACCCCG; USF mt1, CCGGACGACCAACGTTCAGCTCTACCCCG; USF mt2, CCGGACGACCAACGTGCAGCTCTACCCCG; USF mt3, CCGGACGACCCACGTTCAGCTCTACCCCG; USF mt4, CCGGACGACACACGTGCAGCTCTACCCCG; USF mt5, CCGGACGACCCACTTGCAGCTCTACCCCG. (B) USF1 represses PN-1 promoter activity. C6 rat glioma cells were transiently transfected with 1.5 µg of PN-1 promoter/reporter plasmid (p500 or p500mut) together with increasing amounts of a USF1 expression vector (0.1 and 0.3 µg). The total amount of transfected DNA was kept constant by the addition of cDNA-deficient expression vector. Transfections were performed in duplicate and one representative experiment out of three is shown. The standard deviation is shown by the error bars above the columns.

promoter having four or eight CACGTG sites previously showed that USF43 can work as a strong activator (Lefrancois-Martinez et al., 1995). As seen in Fig. 5B cotransfection of the USF1 expression vector and the wildtype PN-1 promoter/reporter construct (p500) re-

Changes in USF Binding Site upon Ibotenic Acid Lesion

33

sulted in a weak dose-dependent repression of PN-1 promoter activity. This repression was dependent on the presence of a functional USF binding site in the PN-1 promoter, since cotransfection of the USF1 expression vector with a reporter plasmid containing the same PN-1 promoter fragment but with a mutated USF binding site (p500mut: CACGTG to AACGTT) did not show any repression but rather an activation (Fig. 5B). The molar ratio between USF1 expression vectors and luciferase reporter vectors was maintained low in order to prevent nonspecific DNA binding effects. This let us conclude that binding of USF43 to the E-box at 2275 can indeed lead to a downregulation of PN-1 promoter activity.

Binding to the USF Site Is Regulated after Brain Lesion PN-1 expression has been shown to be upregulated after lesions induced by injection of the excitotoxin ibotenic acid in rat brains (Scotti et al., 1994). To determine whether protein binding to the USF site is altered in this lesion-induced response, we analyzed nuclear extracts from caudate putamen and substantia nigra (ipsi- and contralateral) 6 days after unilateral ibotenic acid injection in the CP. Nuclear extracts from animals injected with saline and unlesioned animals were used as controls. The gel-shift analysis of nuclear proteins binding to the USF wt oligonucleotide of treated and control animals showed the presence of two bands A and B (Fig. 6). In C6 rat glioma cells band A was readily detected while band B was seen only when a large amount of nuclear extract was analyzed (results not shown). In rats treated with ibotenic acid, nuclear extracts from the caudate putamen of the lesioned left hemisphere showed an increase in binding activity for the lower band B (Fig. 6A, lanes 4L, 5L, 6L, and 7L). The excitotoxic lesion did not cause a change in the intensity of band A (Fig. 6A). The specific increase in the B band after lesion was quantified by PhosphoImager analysis. The results are shown in the bottom panel of Fig. 6A as the B/A ratio. Nuclear extracts of lesioned left brain hemispheres show an average B/A ratio of 0.65, whereas nuclear extracts of unlesioned right brain hemispheres show an average B/A ratio of 0.22, which is very similar to that of unlesioned or saline injected control hemispheres (0.23). No differences in gel-shift patterns were seen with nuclear extracts from the nonlesioned substantia nigra (data not shown). To determine whether the A and B bands could be identified, a supershift was performed with the rabbit polyclonal anti-USF antibody (sc-229X,

FIG. 6. (A) Effects of ibotenic acid-induced lesions on USF binding were investigated by gel-shift assays. USF binding to double-stranded USF wt oligonucleotide in nuclear extracts of caudate putamen from one unlesioned control rat (lane 1), two saline injected rats (lanes 2 and 3), and four rats injected with ibotenic acid in the left caudate putamen (lanes 4 to 7) was analyzed. L is the left (injected) side and R is the right (control) side. The A and B bands were quantified using a PhosphorImager (Molecular Dynamics). The bottom panel shows calculated B/A ratios. (B) Supershift of nuclear extracts from unlesioned and lesioned brains using the rabbit anti-USF antibody (sc-229X, Santa Cruz Biotechnology). Lanes 1 (L and R), without antibody; and lanes 2 (L and R), with antibody. The extracts are identical to those used in lane 4 (L and R) in part A. The arrow indicates the position of the supershifted bands. (C) Calpain treatment affects both A and B bands to the same extent. The nuclear extract used in lane 4R of part A was incubated for 5 min at 30°C with 5 3 1023 U m-calpain/ml in the presence of 5 mM CaCl2 prior to gel-shift analysis (lane 2) and compared to the same extract incubated without calpain (lane 1). The arrows indicate that both bands have undergone the same modification. (D) Lesion does not change the immunoblot signals obtained with anti-USF antibodies (sc-229X, Santa Cruz Biotechnology). Nuclear extracts (10 µg protein) identical to those in lanes 4L and 4R of part A were subjected to SDS–PAGE prior to immunoblot analysis. As above L is the left (injected) and R the right (control) side.

Santa Cruz Biotechnology, recognizing both USF43 and USF44) and the same extracts as used in lanes 4 (L and R) of Fig. 6A. As seen in Fig. 6B (lanes 2L and 2R) both the A and B bands were recognized by the antibody. The same change in binding activity to this site was seen in the caudate putamen after nerve terminal degeneration induced by injection of 6-hydroxydopamine in the substantia nigra (data not shown), another situation in which PN-1 expression was shown to be upregulated (Scotti et al., 1994). No differences in transcription factor

34 binding were detected using oligonucleotides containing Sp1 or NGFI-A/C binding sites of the PN-1 promoter sequence, demonstrating that the effect on the USF binding site was specific (data not shown). As USF can be modified by calpain (Watt and Molloy, 1993), the effect of this serine protease on the proteins binding to the CACGTG site was investigated. Calpain treatment of the nuclear extract prior to the gel-shift analysis affected the migration of both the A and the B bands to a similar extent (Fig. 6C), generating polypeptides which in the gel-shift gave bands different from A and B. In addition the intensity of band A was significantly decreased upon calpain treatment, a modification which was not detected upon excitotoxic lesion (Fig. 6A). Furthermore, immunoblot analysis of nuclear proteins derived from caudate putamen using the anti-USF antibody (sc-229X, Santa Cruz Biotechnology) reveals three bands of 44, 43, and 38 kDa. Lesions changed neither the number nor the intensity of the bands revealed by immunoblot analysis (Fig. 6D). These results indicated that proteolytic cleavage of band A is unlikely to be the cause of the increase of band B after lesion.

DISCUSSION By means of the missing nucleoside technique (Fig. 1) we could demonstrate that only one site in the whole PN-1 silencer fragment interacts with nuclear proteins. Together with the mutation analysis (Figs. 2 and 5A), it identifies the CACGTG site as a negative regulatory element of the PN-1 promoter. It was a surprise to realize that the protein(s) binding to this E-box is specifically recognized by anti-USF antibodies (Fig. 4). In fact, USF was originally described as an activator of the major late promoter of human adenovirus (Gregor et al., 1990; Sawadogo, 1988; Sawadogo and Roeder, 1985; Sawadogo et al., 1988). USF is also a positive regulator of a number of other promoters, among which are the p53 (Reisman and Rotter, 1993), the plasminogen activator inhibitor 1 (Riccio et al., 1992), and the vitellogenin II (Seal et al., 1991) promoters. In the last two cases, USF acts as a transcriptional activator and regulates transcription in response to TGF-b1 and estrogen, respectively. However, in both cases other factors were shown to be involved in this regulation as well (Riccio et al., 1992; Seal et al., 1991). It has only once been suggested that a USF binding site may be a negative regulator of the HIV promoter (Giacca et al., 1992). Our findings clearly indicate that the rat homologue of USF or a closely related protein binds to the CACGTG site acting as a

Ernø et al.

negative regulatory element and that overexpression of USF leads to a dose-dependent repression of the PN-1 promoter activity. This repression is dependent on the presence of a functional USF binding site (Fig. 5B). However, whether the extent of this negative regulatory effect depends on additional coregulators, as recently demonstrated by Halle et al. (1995), remains to be elucidated. Other transcription factors, i.e., promoter elements, can function as both positive and negative regulators. MyoD is generally thought to be a positive factor, although it has recently been shown to repress transcription from the c-fos promoter in cells undergoing muscle differentiation (Trouche et al., 1993). The Myc binding site can function both as a repressor site and as an activator site depending on which heterodimer is bound to it. The Myc protein forms a heterodimer with a bHLH protein, Max, acting as a positive regulator (Prendergast et al., 1991). Max can also form a heterodimer with the protein Mad; this Mad/Max heterodimer will, when bound to the E-box, antagonize the positive transcriptional activity of the Myc/Max protein dimer (Ayer et al., 1993). A model analogous to Myc/Max and Mad/Max may also be relevant for USF. Two forms of USF of 43 and 44 kDa (Sawadogo et al., 1988; Sirito et al., 1994) have been described and splice variants for both USF genes have been reported (Gregor et al., 1990; Sawadogo et al., 1988; Sirito et al., 1992). It is possible that in the case of the PN-1 promoter, a third protein, different from USF43 and USF44, may form a heterodimer with either of these proteins. Alternatively, USF protein binding to the PN-1 promoter could be modified, for example, by phosphorylation or by interaction with a cofactor as recently shown in Halle et al. (1995). Whatever makes the PN-1 USF binding site function as a negative regulatory element has to be found within the 144 bp of the silencer fragment since this fragment represses transcription when cloned upstream of the SV40 early promoter (Fig. 2E). It has been shown that the flanking sequences contribute to the binding and function of several bHLH factors (Fisher and Goding, 1992; Yutzey and Konieczny, 1992). The characterization of the DNA–protein interactions by the missing nucleoside (Fig. 1) and competition experiments (Fig. 5A) indicated that the USF protein not only contacts all the nucleotides in the CACGTG core sequence on both strands but also the Cs 58 and 38 to the CACGTG. The role of the flanking sequence in USF binding was recently investigated by Bendall and Molloy (1994), who demonstrated that the full USF consensus is R25Y24C23A22 C21 G11T12G13R14 Y15 . This is in accor-

35

Changes in USF Binding Site upon Ibotenic Acid Lesion

dance with the bases found to be necessary for USF binding in the missing nucleoside experiments (Fig. 1). Similar results have been reported for other E-box binding factors (Fisher and Goding, 1992; Yutzey and Konieczny, 1992). Differences in the flanking sequences may thus determine the preferential binding of positive or negative USF dimers, a suggestion which is supported by the fact that binding of Myc, but not Max or USF, is inhibited by the presence of a T at 24 (Bendall and Molloy, 1994). Since PN-1 expression is regulated in vivo following brain lesions, it was obvious to analyze the properties of the protein(s) binding to the identified USF site under such conditions. Injection of ibotenic acid in the caudate putamen causes an increased PN-1 expression only in the astrocytes of this structure (Scotti et al., 1994; C. Nitsch, unpublished results). Correspondingly the change in USF gel-shift pattern was observed with nuclear extracts from the caudate putamen (Fig. 6A) but not with those from the substantia nigra. This change was specifically detected in extracts from the left, lesioned caudate putamen, but not from the contralateral side nor from saline-injected or unlesioned control animals. The same characteristic increase in band B is seen in nuclear extract of the caudate putamen after injection of 6-hydroxydopamine into the substantia nigra (data not shown). This type of lesion causes nerve terminal degeneration in the caudate putamen without inducing necrosis. This implies that the increase of band B after lesion is not due to proteolytic activity associated with necrotic tissue. Since both the A and the B bands reveal binding to the CACGTG site and are recognized by specific antibodies against USF, it is probable that these bands contain rat USF or a closely related protein. Several different mechanisms might cause the increase in band B after ibotenic acid- or 6-hydroxydopamine-induced lesion. A posttranslational modification of the proteins might alter the binding or dimerization properties, leading to a change in the gel-shift pattern. Another possibility is the induction of a special isoform of USF, such as mini-USF2, which lacks a transactivation domain (Sirito et al., 1994). Finally, the increase in band B might have been caused by a lesion-specific proteolytic cleavage of USF or of a precursor protein. However, the absence of a change in band A after lesion (Fig. 6A) and the identical USF signal on the immunoblot of nuclear extracts of lesion and control caudate putamen (Fig. 6D) do not support this idea. The molecular mechanisms triggering the upregulation of PN-1 expression after lesion remain to be elucidated. Obviously many scenarios are feasible. However, at this stage, we propose that in astrocytes in which the USF binding site is a negative regulatory element, the

increase of a dominant-negative isoform such as described for mini-USF2 (Sirito et al., 1994) could contribute to the upregulation of PN-1 expression. To our knowledge this is the first time that changes in binding activity to an E-box site have been associated with brain lesions. Modification of the binding, dimerization, and transactivation properties of proteins of the USF family is therefore likely to be part of the events controlling the increase of PN-1 immunoreactivity in astrocytes bordering the zone of neuronal cell death.

EXPERIMENTAL METHODS Recombinant DNA Techniques Cloning and sequencing were performed according to Sambrook et al. (1989). For site-directed mutagenesis of the E-box the proximal promoter fragment (PstI–EcoRI) was cloned into the M13mp18 phage polylinker, and the AAGCTT oligonucleotide (USF mt1) was used as the mutagenic primer as described (Kunkel, 1985). For transfection and luciferase assays a vector was constructed by deleting the HindIII–HindIII fragment containing the SV40 early promoter of the pPALU plasmid (Artelt et al., 1991) and replacing the original polylinker by cloning two oligonucleotides in the HindIII and BamHI sites: 58-AGCTGGATCCGTCGACACTAGTGAGCTCTGCAGAAGCTT-38 and 58-GATCAAGCTTCTGCAGAGCTCACTAGTGTCGACGGATCC-38. The promoter fragments were cloned into this vector as shown in Fig. 2A.

Tissue Culture and Transfections Cells were grown in Dulbecco’s modified Eagle’s medium containing 10% fetal calf serum. Primary cultures of astrocytes were prepared from brain cortex of P2 rats as described by McCarthy and de Vellis (1980). The resulting cultures contained 97% GFAP-positive astrocytes as revealed by immunocytochemistry and were kept in culture for 11 days before being subcultured and transfected. Transfections and luciferase activity measurements were performed as described (Ernø and Monard, 1993). The human USF1 expression vector used in cotransfection assays was kindly provided by Dr. A. Kahn (Paris, France).

Preparation, Fractionation of Nuclear Extracts, and Gel-Shift Assays These assays were performed as previously described (Ernø and Monard, 1993). In the case of brain nuclear extracts 5 µg protein was used per binding reaction.

36 Gel-shift assays in the presence of antibodies were performed by preincubating 1 µg of crude C6 rat glioma nuclear extract or 5 µg of brain nuclear extract with 1 µl of antiserum for 1 h on ice. The antibodies to MyoD, c-Myc, and USF were kindly provided by Drs. G. Pedraza (Basel, Switzerland), M. Eilers (Heidelberg, Germany), and R. G. Roeder (New York). Poly(dI–dC) probe and binding buffer were then added and the gel-shift assay was performed as described (Jost and Saluz, 1991). Immunodepletion of crude C6 nuclear extract (7.5 µg) using 7.5 µl of antiserum and 15 µl of swollen protein A–Sepharose CL-4B (Pharmacia) was done as described (Ausubel et al., 1987). Out of this reaction 4 µl was used per binding assay. For the sequence of oligonucleotides see Fig. 5.

Free Radical Treatment and Missing Nucleoside Technique The DNA fragment was end-labeled on the upper or lower strand with [a-32P]ATP using the Klenow enzyme (Sambrook et al., 1989). One single random nucleoside per molecule was removed using the procedure of Hayes and Tullius (1989). The free-radical-treated DNA was incubated with fractionated nuclear extract from C6 rat glioma cells under standard gel-shift conditions and then run on a 5% native polyacrylamide gel. After exposing the gel for 30 min the bound and free fragments were cut out, eluted overnight into TE buffer, phenol extracted, ethanol precipitated, and analyzed on a 6% sequencing gel.

UV Crosslinking, SDS–PAGE, and Immunoblot The molecular weight determination of DNA–protein complexes by UV crosslinking combined with SDS– PAGE was carried out according to the protocols in Jost and Saluz (1991). Immunoblots were performed according to the protocol provided by Santa Cruz Biotechnology.

Lesion of Rat Brains Male Wistar rats were anesthetized and a total volume of 0.7 µl 63 mM ibotenic acid was injected in the left caudate putamen (Nitsch and Schaefer, 1990). After 6 days, the animals were sacrificed, the brains were rapidly removed, the caudate putamen and substantia nigra were dissected out, and nuclear extract was prepared as described previously (Ernø and Monard, 1993) in the presence of leupeptin (0.5 µg/ml), aprotinin (1 µg/ml), phenylmethylsulfonyl fluoride (0.5 µg/ml), pepstatin (0.7 µg/ml), benzamidine (0.313 µg/ml), and b-glycero-phosphate (4.2 µg/ml).

Ernø et al.

ACKNOWLEDGMENTS We thank G. Pedraza, M. Eilers, and R. G. Roeder for the kind gift of the antibodies against MyoD, c-Myc, and USF; A. Kahn for providing the USF1 plasmid; V. N. Murer for providing primary rat astrocytes; and A. Bruhat, J. Hofsteenge, and D. J. Heard for critical reading of the manuscript. H.E. was supported in part by a grant from the Danish Science Academy.

REFERENCES Artelt, P., Grannemann, R., Stocking, C., Friel, J., Bartsch, J., and Hauser, H. (1991). The prokaryotic neomycin-resistance-encoding gene acts as a transcriptional silencer in eukaryotic cells. Gene 99: 249–254. Ausubel, F. M., Brent, R., Kingston, R. E., Moore, D. D., Siedman, J. G., Smith, J. A., and Struhl, K. (1987). Current Protocols in Molecular Biology. Greene Publishing Associates and Wiley–Interscience, New York. Ayer, D. E., Kretzner, L., and Eisenman, R. N. (1993). Mad: A heterodimeric partner for max that antagonizes myc transcriptional activity. Cell 72: 211–222. Baker, J. B., Low, D. A., Simmer, R. L., and Cunningham, D. D. (1980). Protease-nexin: A cellular component that links thrombin and plasminogen activator and mediates their binding to cells. Cell 21: 37–45. Bendall, A. J., and Molloy, P. L. (1994). Base preferences for DNA binding by the bHLH-Zip protein USF: Effects of MgCl2 on specificity and comparison with binding of Myc family members. Nucleic Acids Res. 22: 2801–2810. Davis, R. L., Cheng, P. F., Lassar, A. B., and Weintraub, H. (1990). The MyoD DNA binding domain contains a recognition code for muscle-specific gene activation. Cell 60: 733–746. Emerson, C. P. (1990). Myogenesis and development control genes. Curr. Opin. Cell Biol. 2: 1065–1075. Ernø, H. (1994). Ph.D. Thesis, University of Basel, Switzerland. Ernø, H., and Monard, D. (1993). Molecular organization of the rat glia-derived nexin/protease nexin-1 promoter. Gene Expr. 3: 163– 174. Farmer, L., Sommer, J., and Monard, D. (1990). Glia-derived nexin potentiates neurite extension in hippocampal pyramidal cells in vitro. Dev. Neurosci. 12: 73–80. Fisher, F., and Goding, C. R. (1992). Single amino acid substitutions alter helix-loop-helix protein specificity for bases flanking the core CANNTG motif. EMBO J. 11: 4103–4109. Giacca, M., Gutierrez, M. I., Menzo, S., Di Fagagna, F. D., and Falaschi, A. (1992). A human binding site for transcription factor USF/MLTF mimics the negative regulatory element of human immunodeficiency virus type 1. Virology 186: 133–147. Gloor, S., Odink, K., Guenther, J., Nick, H., and Monard, D. (1986). A glia-derived neurite promoting factor with protease inhibitory activity belongs to the protease nexins. Cell 47: 687–693. Gregor, P. D., Sawadogo, M., and Roeder, R. G. (1990). The adenovirus major late transcription factor USF is a member of the helix-loophelix group of regulatory proteins and binds to DNA as a dimer. Genes Dev. 4: 1730–1740. Guenther, J., Nick, H., and D. Monard, D. (1985). A glia-derived neurite-promoting factor with protease inhibitory activity. EMBO. J. 4: 1963–1966. Gurwitz, D., and Cunningham, D. D. (1988). Thrombin modulates and

Changes in USF Binding Site upon Ibotenic Acid Lesion

reverses neuroblastoma neurite outgrowth. Proc. Natl. Acad. Sci. USA 85: 3440–3444. Halle, J. P., Stelzer, G., Goppelt, A., and Meisterernst, M. (1995). Activation of transcription by recombinant upstream stimulatory factor 1 is mediated by a novel positive cofactor. J. Biol. Chem. 270: 21307–21311. Hayes, J. J., and Tullius, T. D. (1989). The missing nucleoside experiment: A new technique to study recognition of DNA by protein. Biochemistry 28: 9521–9527. Hoffmann, M. C., Nitsch, C., Scotti, A. L., Reinhard, E., and Monard, D. (1992). The prolonged presence of glia-derived nexin, an endogenous protease inhibitor, in the hippocampus after ischemiainduced delayed neuronal death. Neuroscience 49: 397–408. Jost, J. P., and Saluz, H. P. (1991). A Laboratory Guide to in Vitro Studies of Protein–DNA Interactions, Biomethods, Vol. 5. Birkhaeuser, Basel/ Boston. Kerkhoff, E., Bister, K., and Klempnauer, K. H. (1991). Sequencespecific DNA binding by Myc proteins. Proc. Natl. Acad. Sci. USA 88: 4323–4327. Kunkel, T. A. (1985). Rapid and efficient site-specific mutagenesis without phenotypic selection. Proc. Natl. Acad. Sci. USA 82: 488–492. Lefrancois-Martinez, A. M., Martinez, A., Antoine, B., Raymondjean, M., and Kahn, A. (1995). Upstream stimulatory factor proteins are major components of the glucose response complex of the L-type pyruvate kinase gene promoter. J. Biol. Chem. 270: 2640–2643. Mansuy, I. M., van der Putten, H., Schmid, P., Meins, M., Botteri, F. M., and Monard, D. (1993). Variable and multiple expression of protease nexin-1 during mouse organogenesis and nervous system development. Development 119: 1119–1134. McCarthy, K. D., and de Vellis, J. (1980). Preparation of separate astroglial and oligodendroglial cell cultures from rat cerebral tissue. J. Cell Biol. 85: 890–902. McGrogan, M., Kennedy, J., Li, M. P., Hsu, C., Scott, R. W., Simonsen, C. C., and Baker, J. B. (1988). Molecular cloning and expression of two forms of human protease nexin 1. Biotechniques 6: 172–177. Meier, R., Spreyer, P., Ortmann, R., Harel, A., and Monard, D. (1989). Induction of glia-derived nexin after lesion of a peripheral nerve. Nature 342: 548–550. Monard, D., Solomon, F., Rentsch, M., and Gysin, R. (1973). Gliainduced morphological differentiation in neuroblastoma cells. Proc. Natl. Acad. Sci. USA 70: 1894–1897. Nitsch, C., and Schaefer, F. (1990). Calcium deposits develop in rat substantia nigra but not in striatum several weeks after local ibotenic acid injection. Brain Res. Bull. 25: 769–773. Nitsch, C., Scotti, A. L., Monard, D., Heim, C., and Sontag, K. H. (1993). The glia-derived protease nexin 1 persists for over 1 year in rat brain areas selectively lesioned by transient global ischaemia. Eur. J. Neurosci. 5: 292–297. Prendergast, G. C., Lawe, D., and Ziff, E. B. (1991). Association of Myn, the murine homolog of max, with c-Myc stimulates methylationsensitive DNA binding and ras cotransformation. Cell 65: 395–407. Reinhard, E., Meier, R., Halfter, W., Rovelli, G., and Monard, D. (1988). Detection of glia-derived nexin in the olfactory system of the rat. Neuron 1: 387–394. Reinhard, E., Suidan, H. S., Pavlik, A., and Monard, D. (1994). Glia-derived nexin/protease nexin-1 is expressed by a subset of neurons in the rat brain. J. Neurosci. Res. 37: 256–270. Reisman, D., and Rotter, V. (1993). The helix-loop-helix containing

37 transcription factor USF binds to and transactivates the promoter of the p53 tumor suppressor gene. Nucleic Acids Res. 21: 345–350. Riccio, A., Pedone, P. V., Lund, L. R., Olesen, T., Olsen, H. S., and Andreasen, P. A. (1992). Transforming growth factor beta 1-responsive element: Closely associated binding sites for USF and CCAATbinding transcription factor-nuclear factor I in the type 1 plasminogen activator inhibitor gene. Mol. Cell. Biol. 12: 1846–1855. Sambrook, J., Fritsch, E. F., and Maniatis, T. (1989). Molecular Cloning: A Laboratory Manual. Cold Spring Harbor Laboratory Press, Cold Spring Harbor. Sawadogo, M. (1988). Multiple forms of the human gene-specific transcription factor USF. J. Biol. Chem. 263: 11994–12001. Sawadogo, M., and Roeder, R. G. (1985). Interaction of a gene-specific transcription factor with the adenovirus major late promoter upstream of the TATA box region. Cell 43: 165–175. Sawadogo, M., Van Dyke, M. W., Gregor, P. D., and Roeder, R. G. (1988). Multiple forms of the human gene-specific transcription factor USF. J. Biol. Chem. 263: 11985–11993. Scotti, A. L., Monard, D., and Nitsch, C. (1994). Re-expression of glia-derived nexin/protease nexin 1 depends on mode of lesioninduction or terminal degeneration: Observations after excitotoxin or 6-hydroxydopamine lesions of rat substantia nigra. J. Neurosci. Res. 37: 155–168. Seal, S. N., Davis, D. L., and Burch, J. B. (1991). Mutational studies reveal a complex set of positive and negative control elements within the chicken vitellogenin II promoter. Mol. Cell. Biol. 11: 2704–2717. Sirito, M., Lin, Q., Maity, T., and Sawadogo, M. (1994). Ubiquitous expression of the 43- and 44-kDa forms of transcription factor USF in mammalian cells. Nucleic Acids Res. 22: 427–433. Sirito, M., Walker, S., Lin, Q., Kozlowski, M. T., Klein, W. H., and Sawadogo, M. (1992). Members of the USF family of helix-loop-helix proteins bind DNA as homo- as well as heterodimers. Gene Expr. 2: 231–240. Sommer, J., Gloor, S. M., Rovelli, G. F., Hofsteenge, J., Nick, H., Meier, R., and Monard, D. (1987). cDNA sequence coding for a rat glia-derived nexin and its homology to members of the serpin superfamily. Biochemistry 26: 6407–6410. Stone, S. R., Nick, H., Hofsteenge, J., and Monard, D. (1987). Glialderived neurite-promoting factor is a slow-binding inhibitor of trypsin, thrombin, and urokinase. Arch. Biochem. Biophys. 252: 237–244. Suidan, H. S., Stone, S. R., Hemmings, B. A., and Monard, D. (1992). Thrombin causes neurite retraction in neuronal cells through activation of cell surface receptors. Neuron 8: 363–375. Trouche, D., Grigoriev, M., Lenormand, J. L., Robin, P., Leibovitch, S. A., Sassone Corsi, P., and Harel Bellan, A. (1993). Repression of c-fos promoter by MyoD on muscle cell differentiation. Nature 363: 79–82. Watt, F., and Molloy, P. L. (1993). Specific cleavage of transcription factors by the thiol protease m-calpain. Nucleic Acids Res. 21: 5092–5100. Yutzey, K. E., and Konieczny, S. F. (1992). Different E-box regulatory sequences are functionally distinct when placed within the context of the troponin I enhancer. Nucleic Acids Res. 20: 5105–5113. Zurn, A. D., Nick, H., and Monard, D. (1988). A glia-derived nexin promotes neurite outgrowth in cultured chick sympathetic neurons. Dev. Neurosci. 10: 17–24. Received May 1, 1996 Accepted June 19, 1996