The Dual Role of Helix ± Loop ± Helix-Zipper Protein USF in Ribosomal RNA Gene Transcription in Vivo

The dual role of helix ± loop ± helix-zipper protein USF in ribosomal RNA gene transcription in vivo

Asish K Ghosh, Prasun K Datta and Samson T Jacob*

Department of Pharmacology and Molecular Biology, The Chicago Medical School, 3333 Green Bay Road, North Chicago, Illinois 60064, USA

We have previously demonstrated that the core poliovirus infection, in response to dierentiation, heat promoter of rat ribosomal RNA gene (rDNA) contains shock or inhibition of protein synthesis, and by an E-box-like sequence to which the core promoter glucocorticoid treatment of lymphosarcoma cells binding factor CPBF binds and that the 44 kDa subunit (Jacob, 1995). of this protein is immunologically related to USF1, the In general, rDNA transcription is species-speci®c helix ± loop ± helix-zipper DNA binding protein. Further, (Mishima et al., 1982; Grummt et al., 1982; Ishikawa et we showed that RNA polymerase I (pol I) transcription al., 1991) although a strict species speci®city can be in vitro is competed by oligonucleotides containing altered under certain conditions (Pape et al., 1990; USF-binding site, which suggested a key role for USF Ghosh et al., 1996). Three key cis-acting elements in rDNA transcription. To prove the potential role of namely core promoter, enhancers and terminators are USF in pol I transcription in vivo, USF1 and USF2 involved in the initiation, stimulation and termination homodimers and USF1/USF2 heterodimer were over- of rDNA transcription. Similarly, trans-acting factors expressed in CHO cells by transfection of the respective that speci®cally interact with the cis-acting elements are cDNAs. Co-transfection of a plasmid containing rDNA required for the initiation and/or regulation of rDNA followed by primer extension analysis showed that transcription. A complex called SL1 that consists of overexpression of USF1 and USF2 as homodimers TATA box-binding protein TBP and pol I-speci®c resulted in inhibition of rDNA transcription by as much TBP-associated proteins (TAFs) constitutes one of the as 85 ± 90% whereas overexpression of USF1/USF2 in essential trans-acting factors involved in the initiation the heterodimeric form activated transcription approxi- of rDNA transcription (Comai et al., 1992). Recent mately 3.5-fold. Transfection of mutant USF2 cDNA study has suggested the potential involvement of other that is devoid of the basic DNA-binding domain factors in the initiation of rDNA transcription (Joost et produced only minimal inhibition of rDNA transcription. al., 1994). Two other factors, E1BF/Ku and CPBF These data show that USF can modulate transcription (core promoter-binding factor), are involved in the of rRNA gene in vivo by functioning as a repressor basal or initiation of rDNA transcription (Zhang and (homodimer) or activator (heterodimer) of pol I Jacob, 1990; Ghosh et al., 1993; Ho and Jacob, 1993; transcription in vivo and suggest that inhibition of Ho et al., 1994; Liu and Jacob, 1994). Antibodies rDNA transcription may be responsible for the against the Ku protein can inhibit initiation of rDNA antiproliferative action of USF homodimers. transcription and this inhibition is signi®cantly restored

following addition of puri®ed E1BF/Ku protein (Ho Keywords: ribosomal RNA gene; upstream stimulatory et al., 1994). The factor CPBF is required for rDNA factor; repressor; activator; pol I transcription transcription in a reconstituted system (Liu and Jacob,

1994). Further study showed that CPBF and E1BF/Ku interact physically and functionally to promote rDNA transcription (Niu et al., 1995). It is not known

Introduction whether CPBF and E1BF/Ku directly interact with SL1 complex. It is also unclear how the various factors Ribosomal RNA (rRNA) is synthesized in the involved in the initiation of transcription confer species nucleolus by RNA polymerase I (pol I) as a large speci®city. precursor RNA (pre rRNA) which is processed to Recent study in our laboratory showed that mature rRNAs (28S, 18S and 5.8S) by a series of mammalian rRNA gene promoter contain an E-box- speci®c cleavage reactions (Paule, 1993; Jacob, 1995; like sequence (CACGcTG) to which the basic helix ± Moss and Stefanowsky, 1995). rRNA synthesis can be loop ± helix-zipper DNA binding protein USF binds regulated by a variety of physiological, pathological (Datta et al., 1995). Antibodies against USF1 cross- and nutritional conditions. The up-regulation of rRNA reacted with the 44 kDa polypeptide of rat CPBF. gene (rDNA) transcription is achieved by glucocorti- Further, oligonucleotide probes corresponding to coid in non-lymphoidal cells/tissues in response to USF-binding site and rDNA core promoter inhibited growth or cell proliferation and SV40-induced infec- both pol I and pol II transcriptions. Only tion. It is down-regulated by nutrient deprivation, oligonucleotides that contain the USF binding sequence competed eectively in transcription. This study revealed the functional relationship between Correspondence: ST Jacob CPBF and USF. *Present address: Medical Biochemistry, Ohio State University, 333 The present study was undertaken to determine Hamilton Hall, 1645 Neil Avenue, Columbus, OH 43210 The ®rst two authors contributed equally to this work whether USF can transactivate ribosomal RNA gene Received 22 July 1996; revised 2 October 1996; accepted 7 October promoter in vivo. Surprisingly, overexpression of 1996 USF1 and USF2 individually in cells inhibited pol I- rRNA gene regulation by USF AK Ghosh et al 590 mediated transcription whereas overexpression of kDa USF1/USF2 heterodimer activated the ribosomal

200.0 — L RNA gene promoter. This dual eect of USF on β-galactosidase RNA polymerase I-mediated transcription is speci®c, 97.4 — 69.0 — as overexpression of either subunits as homodimers or heterodimers did not aect a non E-box containing 46.0 — L USF2 promoter. L USF1 * M 1 2 3 4 5 Results Figure 2 Overexpression of USF1, USF2, DUSF2 homodimers and USF1/USF2 heterodimer in CHO cells. Whole cell extracts prepared from transfected CHO cells were subjected to 10% Overexpression of USF1, USF2 and DUSF2 in CHO SDS ± PAGE and transferred to nitrocellulose membrane. cells does not aect non E-box containing promoter Immunoblot analysis was done as described in Methods using antibodies against b-Galactosidase, USF1 and USF2. Lane M, Previous study showed that the 44 kDa subunit of the prestained molelcular weight standards (Amersham); lane 1, ribosomal RNA core promoter-binding factor CPBF is control extracts from pLTR and pSVb-gal DNA-transfected functionally and immunologically related to the basic cells; lane 2, pLTR, USF1 and pSVb-gal DNA-transfected cell extract; lane 3, pLTR USF2 and pSVb-gal DNA-transfected cell helix ± loop ± helix-zipper DNA binding protein USF1 extract; lane 4, pLTR DUSF2 (lacks the basic DNA binding (Datta et al., 1995). Oligonucleotides corresponding to domain) and pSVb-gal DNA DNA-transfected CHO cell extract, USF-binding site and rDNA core promoter inhibited and lane 5, USF1, USF2 and pSVb-gal DNA-co-transfected cell both pol I and pol II mediated transcription in vitro. extract. Arrows on the right indicate the position of 44 kDa (USF2) and 43 kDa (USF1) polypeptides and the 116 kDa b- This data suggested that USF trans-activated rDNA Galactosidase polypeptide. *denotes the endogenous cellular transcription by interacting with the USF-binding E- protein having similar immuno-reactivity with the anti-USF box-like sequence in the ribosomal RNA promoter antibodies in all transfected cell extracts element. To prove further the relevance of this ®nding under in vivo conditions, the 43 kDa USF1, 44 kDa USF2 and DUSF2 were overexpressed in CHO cells by devoid of the basic DNA-binding domain (see transfecting with the respective cDNA clones (see Materials and methods and Figure 1) was also Materials and methods and Figure 1 for plasmid overexpressed in the transfected CHO cells (lane 4). constructs). To monitor transfection eciency and to This study showed that the expressions of the demonstrate that overexpression of USF subunits does transfected cDNAs were signi®cantly greater than not in¯uence promoters lacking E-box sequence, a those of the endogenous USF genes (Figure 2, compare reporter construct pSVb-Galactosidase was used. To lane 1 with lanes 2 ± 5; also see the relatively lighter rule out the in¯uence of vector DNA on the expression bands of the endogenous 44 kDa USF2 and 43 kDa of reporter genes, pLTR vector DNA alone was USF1 in lanes 2 and 3, respectively). The immunoblot transfected. Forty-eight hours after transfection, whole analysis for the expression of b-galactosidase clearly cell extracts were made and subjected to Western blot demonstrates uniformity in the transfection assay and analysis using antibodies directed against b-Galactosi- that the overexpression of USF polypeptides either as dase, USF1 and USF2. The amounts of USF1 and homodimers or as heterodimers does not in¯uence the USF2 were indeed higher in the cells transfected with activity of the control promoter (SV40) and speci®cally USF1 and USF2 cDNAs separately than those a non E-box containing promoter. These data demon- transfected with the control (pLTR and pSVb-gal) strated that overexpression of USF polypeptides in plasmids (Figure 2, compare lanes 2 and 3 with lane 1). CHO cells does not aect the activity of a control Cells co-transfected with USF1 and USF2 cDNAs reporter plasmid (SV40-b-galactosidase) at the protein overexpressed both 43 kDa and 44 kDa polypeptides level. Subsequent series of experiments addressed the (Figure 2, lane 5). As an additional control for the in¯uence of the overexpression of USF subunits as functional studies described below, DUSF2 that is homodimers or as heterodimers on the ribosomal RNA gene promoter activity at the RNA level.

Overexpression of 44 kDa USF2 protein inhibits pol I transcription in vivo Homodimer as well as heterodimer of both USF subunits (44 kDa and 43 kDa) are known to bind the E-box-like sequences with equal anity whereas monomers are unable to bind DNA (Sawadogo and Roeder, 1985; Sirito et al., 1992). Further, translation of USF2 message of dierent lengths is known to form dimers in solution, which can bind to DNA (Sirito et Figure 1 (A ± E) Schematic representation of dierent plasmid al., 1992). First, we investigated the eect of USF2 constructs: (A) pEH2CAT, (B) pLTR, (C) pLTR/USF43 or overexpression on pol I transcription, CHO cells were USF1, (D) pSGU44 or USF2, (E) pSG44DBorDUSF2. : co-transfected with USF2 expression vector and EcoRI/HindIII fragment from pEH5.0; : pBluescript SK plasmid pEH2 CAT that contains rat rRNA gene vector; : bacterial CAT gene; : SV40 intron and promoter (see Figure 1 for the plasmid constructs). polyadenylation signal; : LTR promoter; : SV40 promoter; : Basic domain (B); : Helix ± loop ± helix Total RNA isolated from the transfected cells was (HLH) and leucine zipper (LZ) domains analysed by primer extension using a 20-mer CAT rRNA gene regulation by USF AK Ghosh et al 591 ’ ’ 3 ’ ’ 3 pEH2 CAT pLTR ∆USF2 USF2 USF1 G A T C 4 2 4 2 4 2 4µg pEH2 CAT pLTR USF1/USF2 GATC8µg4µg/4µg AGAGGTCC 5 ATAT * TATATCTCCAGG 3 TATATCTCCAGG ATATAGAGGTCC 5 ATATAGAGGTCC *

1 2 3 4 5 6 7 8 9 10 11

+15 +1 –1 3’ CCTGTCGCACAGTCGT 3 5’ GGACAGCGTGTCAGCATATATCTCCAGG 12 3 456 Figure 3 Primer extension analysis of total RNA isolated from 15 +1 –1 ’ CCTGTCGCACAGTCGT CHO cells co-transfected with pEH2CAT and dierent USF ’ GGACAGCGTGTCAGCA 3 + constructs. Lanes 1 ± 4 represent G, A, T and C sequencing 5 reactions of the plasmid pEH2CAT. Primer-extended products of Figure 4 Analysis of the 5' end of RNA from CHO cells RNA from CHO cells transfected with 4 mg of pLTR DNA transfected with rat rRNA promoter containing plasmid (control) and 10 mg of pEH2CAT (lane 5), 2 mgofDUSF2 (lacks pEH2CAT. Control pLTR vector DNA and both USF1 and the basic DNA binding domain) and 10 mg pEH2CAT (lane 6), USF2 were co-transfected with pEH2CAT. Lanes 1 ± 4 represent 4 mgofDUSF2 and 10 mg of pEH2CAT (lane 7), 2 mg of USF2 the dideoxynucleotide sequencing ladder, G, A, T and C and 10 mg pEH2CAT (lane 8), 4 mg of USF2 and 10 mg respectively. Lane 5, primer extended products from pLTR and pEH2CAT (lane 9), 2 mg of USF1 and 10 mg pEH2CAT (lane pEH2CAT co-transfected control cells; lane 6, USF1/USF2 and 10), 4 mg of USF1 and 10 mg pEH2CAT (lane 11). Arrow on the pEH2CAT co-transfected CHO cells. Arrow indicates the primer right indicates the +1 products from rat rDNA extended +1 products oligonucleotide and AMV reverse transcriptase (see synthesis is due to direct interaction between rRNA Materials and methods). Surprisingly, overexpression gene promoter and the USF2 homodimer, we of the 44 kDa USF2 protein in the cells transfected transfected the CHO cells with USF2 cDNA that with USF2 cDNA inhibited rRNA synthesis following lacks the basic DNA binding domain along with rat co-transfection with the rDNA plasmid whereas co- rRNA promoter containing plasmid. Overexpression transfection with the same amount of control plasmid of the mutant USF2 devoid of the DNA binding pLTR did not block rDNA transcription (Figure 3, domain had a slight inhibitory eect on rRNA compare lane 9 with lane 5). The inhibitory eect of synthesis when compared with the control level of overexpressed USF2 was dose-dependent (Figure 3, expression (Figure 3, compare lanes 6 and 7 with compare lane 8 with lane 9). When the relative lane 5). The slight inhibition of rRNA synthesis from transcriptional activity of rRNA gene promoter was the cloned rat rDNA in cells overexpressing mutant measured, the percent inhibition following transfection USF2 (DUSF2) may be due to sequestering of the with 2 mgand4mg of USF2 cDNA were approxi- endogenous 44 or 43 subunits by overexpressed mately 70% and 85% respectively. It was evident from DUSF2, resulting in unavailability of enough the sequence ladder of the template pEH2CAT that the endogenous 44/43 heterodimer to the initiation transcripts were initiated at the +1 site. complex. This data strongly suggests that the repressor activity of USF2 homodimer on rRNA synthesis is due to direct or speci®c interaction of Homodimer of 43 kDa (USF1) protein also inhibits rat rRNA gene promoter with USF2 homodimer rather rRNA gene transcription in vivo than protein-protein interaction. Next, we investigated the eect of USF1 overexpression on RNA pol I-mediated transcription. To address Overexpression of both USF1 and USF2 activates pol I this issue, USF1 cDNA was co-transfected with rat transcription in vivo rDNA containing plasmid pEH2 CAT into CHO cells and RNA synthesis was measured by primer extension We then investigated whether USF1/USF2 heterodimer analysis. As observed for USF2, overexpression of also functions as a repressor of rDNA transcription. USF1 also resulted in inhibition of rRNA synthesis in To address this issue, we co-transfected both USF1 vivo (Figure 3, lanes 10 and 11), as compared with the and USF2 expressing plasmids along with the rRNA control level of expression following transfection of gene promoter containing reporter gene into CHO rat rDNA plasmid with vector DNA (lane 5). The cells. Contrary to the eect of overexpressed homo- repressor activity of USF1 was also dose-dependent dimers transient increase in the level of USF1/USF2 (Figure 3, compare lane 10 with lane 11). As much as heterodimer led to stimulation of rRNA synthesis from 90% inhibition of rRNA synthesis was observed even the rat rRNA gene promoter in vivo (Figure 4, when 2 mg of USF1 cDNA was overexpressed. compare lane 6 with lane 5). Approximately 3.5-fold stimulation of rRNA synthesis was observed in response to overexpressed heterodimer relative to the Overexpression of 44 kDa USF2 protein lacking basic control level of rRNA synthesis in three independent DNA binding domain minimally aects pol I experiments. This result suggests that binding of USF1 transcription and USF2 to the E-box-like sequences of rat rRNA To con®rm further that the inhibitory eect of gene promoter acts as an activator only if they interact overexpressed 44 kDa USF2 protein on rRNA with the promoter element in the heterodimeric state. It rRNA gene regulation by USF AK Ghosh et al 592 is likely that stable initiation complex formation is DUSF2 does not alter pol I transcription (data not contingent upon interaction between the cis element shown). It is likely that the USF2 mutant dimerizes and the USF1/USF2 heterodimer. with USF1 to form an inactive USF complex that is incapable of binding to DNA. Consequently, the stimulatory eect of wild type USF1/USF2 hetero- Discussion dimer is not observed with wild type USF1/mutant USF2 heterodimer. Earlier study in our laboratory showed that the 44 kDa To our knowledge, this is the ®rst report that USF subunit of the core promoter-binding factor CPBF is functions as a repressor or activator of pol I immunologically related to the 43 kDa polypeptide of transcription in vivo depending upon the dimerization the basic helix ± loop ± helix-zipper DNA-binding pro- state of the constituent polypeptides. It should be tein USF1, and that the mammalian pol I core noted that expression of USF1 and USF2 separately promoter contains a sequence similar to the USF- results in homodimer formation whereas simultaneous binding E-box sequence initially discovered in adeno- expression of the two polypeptides can form hetero- virus major late promoter (Datta et al., 1995). Indeed, dimer (Sirito et al., 1994). Because each one of the the presence of E-box element is essential for the DNA- polypeptides is in considerable excess over the binding activity of USF. Accordingly, oligonucleotides endogenous levels, overexpression of each of the corresponding to rat rDNA core promoter or USF polypeptides individually must lead to higher level of inhibited both pol I and pol II transcription. The pol I homodimer over the heterodimer population in the promoter, therefore, belongs to a family of other E-box cells, while overexpression of both subunits theoreti- containing USF-activated cellular promoters that cally would result in 1:2:1 (USF1/USF1:USF1/ include those of alcohol dehydrogenase (Potter et al., USF2:USF2/USF2) dimer distribution as been ob- 1991), mouse metallothionein-I (Carthew et al., 1987; served in vitro (Sirito et al., 1994; Viollet et al., 1996) or Mueller et al., 1988), human heme oxygenase (Sato et in predominance of heterodimers. Most of USF1 and al., 1990), human insulin (Read et al., 1993) and human USF2 are known to exist as heterodimers in vivo while CD2 (Outram and Owen, 1994). In addition to the homodimers are underrepresented (Sirito et al., activation of alcohol dehydrogenase, human insulin 1994; Viollet et al., 1996). Co-transfection of USF1 and and CD2 genes, USF is also known to stimulate pol III USF2 into cells can, indeed mimic the in vivo pattern transcription of U6 small nuclear RNA gene (Li et al., and that the ratios of the USF1 and USF2 cDNAs 1994). USF thus appears to be a common transcription transfected also dictates the dimer composition (Viollet factor for all three RNA polymerases. et al., 1996). Based on these observations it is The present study has demonstrated a direct role reasonable to conclude that overexpression of indivi- for USF in pol I transcription in vivo. The most dual subunits results in homodimer formation and that noteworthy ®nding is that USF1 or USF2 homo- of both subunits results in heterodimer formation. dimer can function as a repressor of pol I There are several examples of transcriptional transcription in vivo and that heterodimerization of modulation by speci®c activators or repressors for pol USF1/USF2 is essential for the transactivation of II genes. There are, however, only a handful of pol I promoter. Unlike pol I transcription system, proteins that can act both as activators and USF1 and USF2 individually (homodimer) or in repressors. For example, the Drosophila Kruppel heterodimeric form can act as activator of several protein can activate or repress transcription of protein coding genes. Because members of the USF Drosophila Adh distal promoter activity depending family cannot interact with DNA as monomers upon their concentrations (Sauer and Jackle, 1993).

(Sirito et al., 1992), inhibition of pol I transcription Similarly, the pol I transcription factor E1BF/Ku following overexpression of USF1 or USF2 must be activates pol I transcription in vitro at low concentra- due to the eect of the homodimers. The rat rDNA tions and represses transcription at higher concentra- core promoter-protein complex is stable after heat tions (Ghosh et al., 1993; Ho and Jacob, 1993; Ho treatment and can be supershifted using anti USF1 et al., 1994). The tumor suppressor protein p53 can antibodies (data not shown) which suggested the function as activator or repressor of pol II mediated interacting protein with rat rDNA is indeed USF. transcription (Juven et al., 1993; Kley et al., 1992) and Therefore, the eect of overexpressed USF1, USF2 acts as a repressor of RNA polymerase I mediated and USF1/USF2 on rRNA synthesis in vivo is gene transcription in vivo (Ghosh and Jacob, unpub- probably due to speci®c binding of overexpressed lished results). The pol III transcription factor TFIIIB USF to E-box like sequences in rDNA. This is activated or repressed, depending upon its phos- conclusion is further substantiated by the observa- phorylated state (Gottesfeld et al., 1994). The tion that the non E-box containing SV40 promoter phosphorylated state of the retinoblastoma protein was unaected by overexpression of the USF pRB can determine its potential to activate or repress polypeptides. The slight inhibitory eect of over- transcription (Hinds, 1995). The tumor suppressor expressed dominant mutant DUSF2 on rRNA protein Rb also acts as a repressor of pol I synthesis may be due to sequestering of endogenous transcription (Cavanaugh et al., 1995). A recent study

44 or 43 kDa which causes low level of endogenous from our laboratory demonstrated that E1BF/Ku can 44/43 heterodimer available in the rDNA transcrip- be post-translationally modi®ed following serum tion initiation complex. Such sequestering eect of deprivation and the modi®ed form can function as a USF has also been suggested for the regulation of a repressor of pol I transcription (Niu and Jacob, 1994). bi-directional viral promoter (Meier et al., 1994) and USF may now be added to the list of proteins that can L-type pyruvate kinase gene promoter (Lefrancois- play a dual role in the modulation of pol I Martinez et al., 1995). Co-expression of USF1 and transcription in vivo. rRNA gene regulation by USF AK Ghosh et al 593 Finally, it would be of interest to know whether precipitation technique (for details, see Gorman et al., USF can modulate rDNA transcription under dierent 1982). For analysis of the expression of dierent cDNA physiological conditions. The DNA binding activity of encoded products at the protein level, the cells were USF is regulated in a growth-dependent manner while transfected with dierent concentrations (4 ± 8 mg) of the protein level remains the same during G1, G1/S pLTR (vector alone), 6 mgofpSV-b-gal (control), and 4 mg of USF1, USF2 and DUSF2. For analysis of the and S phase of growth cycle (Miltenberger et al., 1995). reporter gene expression at the RNA level, dierent This report suggested that post-translational modifica- concentrations (2 ± 4 mg) of USF1, DUSF2, USF2 and tion of USF could occur during transition of cells from pLTR (control) plasmid DNAs were used for co- quiescent to growing state. Whether such modi®cation transfection along with reporter plasmid pEH2CAT of USF is also manifested during the regulation of (10 mg). pLTR vector plasmid was used in control rDNA transcription at dierent stages of cell cycle transfection experiment along with pEH2CAT to eliminate remains to be seen. It is worthwhile to investigate the DNA concentration eect and the eect of vector DNA whether the alteration in rRNA synthesis under a on transfection eciency. variety of physiological, pathological and nutritional conditions is determined by the ratio of USF1 and RNA isolation and primer extension analysis USF2 homodimer concentration to the heterodimer Total RNA was isolated from transfected CHO cells 48 h concentration. The molecular mechanism by which the after transfection following the protocol of Xie and USF homodimers and heterodimers modulate pol I Rothblum (1991). The 5' end of RNA synthesized from transcription in vivo remains to be elucidated. Recently, transfected plasmids was analysed by primer extension of it has been demonstrated that expression of either total RNA (80 mg) using a 20-mer CAT oligo and AMV USF1 or USF2 has homodimer inhibits cellular Reverse transcriptase. Primer extended products were proliferation (Luo and Sawadogo, 1996). Because separated by electrophoresis on 7M urea-6% polyacryla- ribosome biogenesis is essential for cellular prolifera- mide gel (for details see Ghosh et al., 1993). tion, the inhibition of rDNA transcription in response to USF1 or USF2 homodimer may be one of the ways Western blot analysis of protein from transfected CHO cells to control cellular proliferation. Equal amount of protein extracts (85 mg) from transfected CHO cells were subjected to electrophoresis on 10% SDS- Materials and methods polyacrylamide gel, and proteins were transferred to nitrocellulose membrane. After blocking with 5% fat free Plasmid constructs milk in TBS buer (10 mM Tris-HCl, pH 8.0; 125 mM NaCl), the membrane was cut around the All plasmid constructs are schematically shown in Figure 1. 69 kDa protein marker horizontally and the upper half of pEH2CAT was constructed by inserting the rat rDNA the membrane was probed with anti-b-Galactosidase spanning from 74.5 kb to +124 bp (with respect to the antibodies (Promega) while the lower half of the +1 initiation site) in front of the bacterial chloramphenicol membrane was incubated with anti USF1 and anti USF2 acetyltransferase gene of pJFCAT1 (Fridovich-Keil et al., antibodies. The nitrocellulose membranes were incubated 1991) (see Figure 1A). The eucaryotic expression vector with appropriate anti-IgG alkaline phosphatase conjugate pLTR that contains the Moloney sarcoma virus long and washed three times in TBST buer (TBS plus 0.05% terminal repeat (LTR) and SV40 polyadenylation signal Tween 20) and the color was developed with BCIP/NBT was used as control plasmid for co-transfection experi- tablet (Sigma). ments (see Reisman and Rotter, 1993 and Figure 1B). pLTR/USF43 or USF1 was constructed by inserting human USF43 cDNA in the EcoRI site of pLTR vector (see Reisman and Rotter, 1993 and Figure 1C). pSGU44 or USF2 was constructed by inserting the end-®lled EcoRI ± NsiI fragment of USF44 cDNA (mouse) into BamHI site of pSG5 eucaryotic expression vector (see Lin et al., 1994 Acknowledgements and Figure 1D). pSGU44DBorDUSF2 was made by We thank Michele Sawadogo (University of Texas M.D. deleting the basic DNA-binding domain (aa 229 to aa 249) Anderson Cancer Center) for the generous gifts of USF2, of mouse USF2 inserted in pSG5 vector (see Meier et al., DUSF2 cDNAs, antibodies against USF1/USF2 and also 1994 and Figure 1E). pSV-b-Galactosidase (pSV-b-gal) was for critical comments on the manuscript, David Reisman obtained from Promega. (University of South Carolina) for the expression vectors pLTR and pLTR/USF43, Judith Fridovich-Keil (Emory University School of Medicine) for the plasmid JFCAT1, Cell culture and DNA transfection and Angela Simopoulos for secretarial assistance. This Chinese hamster ovary (CHO) cells were transfected with work was supported by a US Public Health Science Grant dierent plasmids using calcium phosphate-DNA co- (CA 31894) from the National Cancer Institute.

References

Carthew RW, Chodosh LA and Sharp PA. (1987). Genes Fridovich-Keil JL, Gudas JM, Bryan IB and Pardee AR. Dev., 1, 973 ± 980. (1991). BioTechniques, 11, 572 ± 579. Cavanaugh AH, Hempel WM, Taylo LJ, Rogalsky V, Ghosh AK, Ho CM and Jacob ST. (1993). Gene, 125, 217 ± Todorov G and Rothblum LI. (1995). Nature, 374, 177 ± 222. 180. Ghosh AK, Niu H-F and Jacob ST. (1996). Biochem. Comai L, Tanese N and Tjian R. (1992). Cell, 68, 965 ± 976. Biophys. Res. Commun., 225, 890 ± 895. Datta PK, Ghosh AK and Jacob ST. (1995). J. Biol. Chem., Gorman CM, Moat LF and Howard BH. (1982). Mol. Cell. 270, 8637 ± 8641. Biol., 2, 1044 ± 1051. rRNA gene regulation by USF AK Ghosh et al 594 Gottesfeld JM, Wolf VJ, Dang T, Forbes DJ and Hartl P. Moss T and Stefanowsky YY. (1995). Prog. Nucleic. Acids. (1994). Science, 263, 81 ± 84. Res. Mol. Biol., 50, 25 ± 66. Grummt I, Roth E and Paule MR. (1982). Nature, 269, 173 ± Mueller PR, Salser SJ and Wold B. (1988). Genes Dev., 2, 179. 412 ± 427. Hinds PW. (1995). Current Opinion in Genet. Dev., 5, 79 ± 83. Niu H-F and Jacob ST. (1994). Proc. Natl. Acad. Sci. USA, Ho CM and Jacob ST. (1993). Biochem. Biophys. Res. 91, 9101 ± 9105. Commun., 190, 747 ± 753 Niu H-F, Zhang J and Jacob ST. (1995). Gene Expr., 4, 111 ± Ho CM, Ghosh AK, Prabhakar BS and Jacob ST. (1994). 124. Proc. Natl. Acad. Sci. USA, 91, 762 ± 766. Outram SV and Owen MJ. (1994). J. Biol. Chem., 269, Ishikawa Y, Safrany G, Hisatake K, Tanaka N, Maeda Y, 26525 ± 26530. Kato H, Kominami R and Muramatsu M. (1991). J. Mol. Pape LK, Windle JJ and Sollner-Webb B. (1990). Genes Dev., Biol., 218, 55 ± 67. 4, 52 ± 62. Jacob ST. (1995). Biochem. J., 306, 617 ± 626. Paule MR. (1993). Gene Expr., 3, 1±9. Joost CBMZ, Beckmann H, Comai L and Tjian R. (1994). Potter JJ, Cheneval D, Dang CV, Resar LMS, Mezey E and Science, 266, 2015 ± 2018. Yang YW. (1991). J. Biol. Chem., 266, 15457 ± 15463. Juven T, Barak Y, Zauberman A, George DL and Oren M. Read ML, Clark AR and Docherty K. (1993). Biochem. J., (1993). Oncogene, 8, 3411 ± 3416. 295, 233 ± 237. Kley N, Chung RY, Fay S, Loeer JP and Seizinger BR. Reisman D and Rotter V. (1993). Nucleic Acids. Res., 21, (1992). Nucleic Acids. Res., 20, 4083 ± 4087. 345 ± 350. Lefrancois-Martinez A-M, Martinez A, Antoine B, Ray- Sato M, Ishizawa S, Yoshida T and Shibahara S. (1990). Eur. mondjean M and Kahn A. (1995). J. Biol. Chem., 270, J. Biochem., 188, 231 ± 237. 2640 ± 2643. Sauer F and Jackle H. (1993). Nature, 364, 454 ± 457. Li JRA, Parsons ST and Marzlu WF. (1994). Mol. Cell. Sawadogo M and Roeder RG. (1985). Cell, 43, 165 ± 175. Biol., 14, 2191 ± 2200. Sirito M, Lin Q, Maity T and Sawadogo M. (1994). Nucleic Lin Q, Luo X and Sawadogo M. (1994). J. Biol. Chem., 39, Acids. Res., 22, 427 ± 433. 23894 ± 23903. Sirito M, Walker S, Lin Q, Kozlowski MT, Klein WH and Liu Z and Jacob ST. (1994). J. Biol. Chem., 269, 16618 ± Sawadogo M. (1992). Gene Expr., 2, 231 ± 240. 16626. Viollet B, Lefrancois-Martinez A-M, Henrion A, Kahn A, Luo X and Sawadogo M. (1996). Proc. Natl. Acad. Sci. USA, Raymondjean M and Martinez A. (1996). J. Biol. Chem., 93, 1308 ± 1313. 271, 1405 ± 1415. Meier JL, Luo X, Sawadago M and Straus SE. (1994). Mol. Xie W-Q and Rothblum LI. (1991). BioTechniques, 11, 325 ± Cell. Biol., 14, 6896 ± 6906. 327. Miltenberger RJ, Sukow KA and Farnham PJ. (1995). Mol. Zhang J and Jacob ST. (1990). Mol. Cell. Biol., 10, 5177 ± Cell. Biol., 15, 2527 ± 2535. 5186. Mishima Y, Financsek I, Kominami R and Muramatsu M. (1982). Nucleic Acids. Res., 10, 6659 ± 6670.