Identification of a novel Krüppel-associated box domain protein, Krim-1, that interacts with c-Myc and inhibits its oncogenic activity.

We have used the Ras recruitment system to screen for proteins that interact with the N-terminally located transactivation domain of c-Myc. The Ras recruitment system is based on the activation of the mitogenic RAS signaling pathway in yeast by the mammalian GTPase Ha-Ras. This screen led to the identification of two novel nuclear proteins termed Krim-1A and Krim-1B that both contain an N-terminal KRAB box domain and 12 or 9 Krüppel C2H2 type zinc fingers at the C terminus, respectively. We found that sequences covering the Myc box II homology region are essential for the interaction with the Krim-1 proteins and that the second N-terminal zinc finger of Krim-1 is essential for Myc binding. Both Krim-1A and -B genes appear to be expressed ubiquitously with highest levels in spleen and lymph nodes. In particular, Krim-1B and, to a lesser extent, Krim-1A are able to decrease E-box-dependent transcriptional transactivation by c-Myc-Max complexes and also the ability of Myc to malignantly transform primary rat embryo fibroblasts, which is consistent with the functional repressive properties of their KRAB domains. The transcriptional corepressor Tif-1beta is a binding partner for Krim-1 and stabilizes the protein. Our findings suggest that Myc-mediated functions can be negatively regulated by Krim-1, potentially in a complex with Tif-1beta.

c-Myc is a member of a family of nuclear phosphoproteins that comprise c-, N-and L-Myc. Whereas c-Myc is found in almost all proliferating cells, expression of N-and L-Myc is more restricted to specific cell types (for a review, see Refs. 1 and 2). All three Myc family members have key roles in the regulation of cell growth, differentiation, and proliferation. The c-Myc gene for instance is induced by serum stimulation in cultured cells as a typical immediate early response gene, and the c-Myc protein exerts part of its function by driving cells into S-phase, whereas a block of Myc function can lead to arrest of cells in the G 1 phase of the cell cycle (reviewed in Refs. 1 and 2). In a large number of human cancers, Myc family genes are transcriptionally activated either by gene amplification or specific chromosomal alterations (reviewed in Refs. 3 and 4). Myc is able to malignantly transform cells in culture either alone or in cooperation with other activated oncogenes such as Ha-ras (5). Surprisingly, under growth factor deprivation, constitutive Myc expression can drive cells into apoptosis (programmed cell death) (6,7). It still remains controversial how the growth promoting and death promoting effects of Myc can be reconciled.
A first picture how Myc family proteins might regulate gene expression has emerged through a number of experiments in recent years. All Myc family members contain a transactivation domain (TAD) 1 at the amino terminus and a basic region directly followed by a basic helix loop helix-leucine zipper motif (b-HLH-LZip) at its carboxyl terminus (8 -10). The b-HLH-LZip domain is responsible for DNA binding and heterodimerization with partner proteins (8 -10). One of the known proteins that interact with Myc family members through this domain is the Max protein that also has a b-HLH-LZip domain (9 -12) (reviewed in Ref. 13). The Myc-Max complex specifically binds to E-box sequences with a preference for the sequence CACGTG (9,10). The TAD of Myc contains two regions that are highly conserved in sequence among Myc family members and are termed Myc box I and II. Both homology regions appear indispensable for an optimal transcriptional transactivation (14).
In addition to its activities as a transcriptional transactivator, high levels of Myc expression have been correlated with the down-regulation of a number of genes. The mechanisms for a direct transcriptional repression by Myc are less well understood but thought to be independent of E-box-mediated activation (15)(16)(17). Data obtained to date suggest that Myc-mediated repression is the result of an interference with proteins binding to the initiator element (INR), which is part of the core promoter (18 -21). In this model, Myc negatively interferes with proteins as YY1, TFII-I, or Miz-1 (18 -22), which transactivate target gene expression through INR binding. For instance, Miz-1-driven expression of INR-dependent reporter genes could be abrogated by Myc.
To be effective as a transcriptional transactivator, the Myc-Max complex must recruit proteins with the ability to modify chromatin in order to allow installment and progression of the basic transcription machinery (reviewed in Refs. 1 and 2). Indeed, both histone acetyltransferases and ATP-dependent chromatin remodeling enzymes (SWI/SNF) have been reported to associate with the Myc protein (23,24). It has been shown that TRRAP, a large protein that interacts with the N terminus of Myc, is part of a multiprotein complex with histone acetyltransferase activity (25)(26)(27)(28). Moreover, several other proteins are known today that interact with the Myc TAD and may be involved in the regulation of its activity: p107, Bin-1, and MM-1 (29 -31). Of particular interest in this respect is MM-1, which requires Myc box II sequences for binding to Myc. Most recently, it was demonstrated that MM-1 can bind to the transcriptional co-repressor Tif-1␤, which is known to recruit the chromatin-associated factors HP-1 and histone deacetylases. This suggested a mechanism for Myc-mediated transcriptional repression (29,31,32) that involves E-box-dependent gene regulation. In many cases, the isolation of Myc regulators has relied on yeast interaction cloning strategies (33). Although the conventional yeast two-hybrid system has been very powerful in the isolation of protein-protein interaction partners in a number of cases, its use with transcription factors is limited, because their transcriptional activation domains strongly interfere with the transcriptional selection mechanism of the system itself. In our present study, we have used a novel yeast interaction cloning strategy, the Ras recruitment system (RRS) (34 -38), to identify proteins that bind to the TAD of Myc. The RRS selection mechanism is independent of transcriptional regulation and relies on the direct activation of the mitogenic Ras pathway in yeast cells (37). In the RRS, the sequence that is used as a bait to isolate a partner protein is cloned as a fusion protein with the human Ha-Ras protein (37). The expression library that is to be screened with this "bait" encodes fusion proteins with the Src myristoylation signal, which will lead to attachment of these fusion proteins to the plasma membrane. If protein-protein interaction takes place, the bait-Ras fusion is recruited to the membrane and activates Ras-dependent signal transduction. This allows growth of a yeast mutant that is defective in this pathway due to a mutation in the endogenous Ras guanyl exchange factor Cdc25. Since no transactivation is needed to select for positive yeast clones, an endogenous transactivation property of the "bait" does not interfere with the selection process (37).
Here we have used the 262 N-terminal amino acids of c-Myc as a bait in the RRS, and we were able to isolate a novel KRAB box-containing zinc finger protein, Krim-1, that binds to the Myc TAD and can negatively influence Myc functions.

EXPERIMENTAL PROCEDURES
cDNA Library, Yeast Strains, Growth, and Complementation cdc25-2 yeast strain (a, ura3, lys2, leu2, trp1, cdc25-2, his3D200, ade101, GALϩ) was transfected with the indicated plasmids and grown in minimal medium supplemented with the appropriate amino acids and nucleotides at 25°C. To test for complementation, plates were replica-plated onto galactose minimal plates supplemented with amino acids and nucleotides lacking leucine and uracil and were grown at the nonpermissive temperature of 36°C. All used yeast plasmids were derived from the galactose-inducible Yes2 (Invitrogen) and the constitutive ADNS vector. DNA fragments of c-Myc, c-Jun, and c-Fos to be inserted in frame to either Ha-Ras or the Src myristoylation sequence were generated by PCR.
cDNA Library-Double-stranded cDNA was generated from poly(A) ϩ RNA of the GC rat pituitary cell line using random primer adapters and the ZAP-cDNA synthesis kit (Stratagene). The cDNA was directionally ligated with Yes2 plasmid containing the Src myristoylation sequences. The library represents 2 ϫ 10 6 primary transformants. To reduce the number of isolated Ras-containing clones from the library, the strain used for screening contained an expression plasmid with mammalian GAP (Ras GTPase-activating protein) (36).

Ras Recruitment Screening and Mutational Analysis
RRS Library Screening with c-Myc262 Bait-Ras recruitment screening was performed essentially as described (37). Briefly, for the construction of the Myc262-RRS bait plasmid the transactivation domain of c-Myc corresponding to amino acids 1-262 was amplified by PCR from a human c-Myc cDNA. This fragment was subcloned in frame into HindIII/SmaI-cut plasmid pADNS-Ras(mut), containing Ha-Ras(L61) cDNA lacking the CAAX-farnesylation signal. Yeast transformation and manipulation was performed as described (36,37). We used a cDNA expression library (34), constructed from mRNA of the GC cell line (rat pituitary tumor). To generate a bait strain, the Myc262 bait plasmid was cotransformed with the plasmid mGAP (36) into the yeast strain cdc25-2 and grown at 25°C on minimal medium lacking leucine and tryptophan. A starter culture from a single colony of this bait strain was grown in 100 ml of minimal medium, and then 2ϫ 200 ml of YPAD medium was inoculated to A 600 ϭ 0.4 and then grown at 25°C for 4 h. Then 60 g of library plasmid DNA were transformed into the bait strain and grown at 25°C on agar plates with minimal medium lacking leucine, tryptophan, and uracil for 3-4 days. For selection of clones with interacting protein pairs, cells were replica-plated to agar plates with galactose minimal medium (Ϫleu, Ϫtrp, Ϫura) and grown at 36°C for 6 days. These clones were streaked to glucose minimal medium (Ϫleu, Ϫtrp, Ϫura), incubated at 25°C for 3-4 days, and then tested for their dependence on expression from the galactose-inducible library plasmid. Colonies were replica-plated to agar plates containing galactose or glucose (both plates Ϫleu, Ϫtrp, Ϫura) and grown for 3-4 days at 36°C. Clones that showed glucose-repressible and galactose-inducible growth were used further for the isolation of the library plasmid. To test dependence on the bait, the isolated library plasmids were reintroduced into the cdc25-2 yeast strain in combinations with either the original screening bait or empty or a heterologous bait. One library clone was identified to confer growth in a Myc262 bait-specific manner at 36°C.
Ras Recruitment Screening with Krim-1 1-177 as a Bait-For RRS screening, a fragment with the N-terminal amino acids 1-177 of Krim-1 was amplified by PCR (primers: Krim bait-T (5Ј-GCATAAGCTTGCC-GCCATGGAGCCAGTGACCTTTG-3Ј) and Krim bait-B (5Ј-ATACCCG-GGATCGGAGGTGAAACTCCTAGATGCTTCG-3Ј); the Krim bait-B primer mutates an endogenous HindIII site silently). This fragment was inserted via HindIII/SmaI into the RRS bait vector pADNS-Ras-(mut) followed by introduction of this Krim-177-Ras bait plasmid and the mGAP expression plasmid into the yeast cdc25-2 strain. This bait strain was used for RRS screening of the GC cell cDNA expression library as described above.

RACE Amplification
First strand cDNA preparation-For preparation of first strand cDNA, 5 g of total RNA (from rat testis, bone marrow, or intestine) and 50 pmol of primer (for 5Ј-RACE, 5Ј-AAGGTCTCCTTCATCACGTC; for 3Ј-RACE dT17 adapter primer, 5Ј-GACTCGAGTCGACATCGA-T17) was mixed in 8.5 l of reverse transcription buffer, and heated to 70°C for 10 min. Then dithiothreitol, RNasin (20 units; Promega), and 200 units of reverse transcriptase (Superscript II; Invitrogen) was added in the recommended buffer. After incubation at 49°C (1 h), reverse transcriptase was heat-inactivated (70°C; 15 min) and incubated with 2 units of RNase H (Invitrogen) at 37°C for 20 min. This first strand cDNA preparation was purified on Qiaquick PCR purification spin columns (Qiagen, Hilden, Germany) and eluted in 50 l of elution buffer. Only for 5Ј-RACE, this first strand cDNA was dA-tailed using 1 mM dATP, 15 units of terminal deoxynucleotide transferase (Life Technologies) in a 75-l volume (37°C; 30 min) and then heat-inactivated (70°C; 10 min).
5Ј-RACE Amplification-The missing 5Ј region of the Krim-1 cDNA was obtained from dA-tailed cDNA by two-step amplification ("Synergy" Taq polymerase; Genecraft) using the common dT17 adapter primer and a nested pair of Krim-1-specific primers (step 1, 5Ј-AAGGTCTC-CTTCATCACGTC; step 2, 5Ј-GAATCCAACATAGTCCACTC). The PCR products were purified after separation by agarose gel electrophoresis, blunted, and subcloned. The 3Ј-RACE products were obtained using dT17 adapter-primed cDNA as template. After two-step amplification using a common dT17 adapter primer and a pair of nested Krim-1specific primers (step 1, 5Ј-CCTATTCCAGTGCTTGCTAT; step 2, 5Ј-CTTTCCCCAGATCCCTTCAC), the resulting PCR products were subcloned as described above. To generate full-length Krim-1A and -1B cDNAs, the insert of the Krim-1 library clone was assembled with the corresponding 3Ј-RACE fragments by ligation of partial HindIII-digested DNA fragments.

Expression Analysis by Real Time PCR
Total RNA was isolated from various rat tissues by homogenization of the corresponding tissues in isothiocyanate buffer and subsequent column purification (RNeasy; Qiagen, Hilden, Germany) according to the manufacturer's protocol. Immunoprecipitation COS-7 cells or HeLa cells were plated on 90-mm dishes and transiently transfected with 12 g of SR-HA-Krim and SP-c-Myc plasmids using 20 l of the lipofection reagent Rotifect™ (Rotilab) or with the calcium phosphate technique. Cells were refed with complete medium after 6 h and were harvested after another 18 h. Cells were washed with PBS and then scraped off the plate in 400 l of low salt lysis buffer (20 mM Hepes, pH 6.8, 5 mM KCl, 5 mM MgCl 2 , 0.5% Nonidet P-40, 0.1% sodium deoxycholate, and protease inhibitors). Then the cells were completely disrupted by Dounce homogenization (30 strokes), centrifuged for 5 min (microcentrifuge; 3600 rpm, corresponding to 1000 ϫ g). The cytosolic supernatant was discarded, and the remaining crude nuclear pellet was extracted with high salt lysis buffer (500 mM NaCl in low salt lysis buffer), and incubated for 10 min on ice. After adding an equal amount of low salt lysis buffer, the lysate was cleared by centrifugation at 13,000 rpm (microcentrifuge; 4°C for 15 min). For immunoprecipitation, 8 l of ␣-c-Myc polyclonal antibody (C-19; Santa Cruz) or 8 l of ␣-Cdk4 polyclonal antibody (H-22; Santa Cruz) and 15 l of protein A-agarose (50% slurry previously blocked in bovine serum albumin) was added to the lysate and rocked for 1 h at 4°C. Immune complexes were recovered by centrifugation (2500 rpm; microcentri-fuge; 4°C) and washed four times with 1.3 ml of medium salt lysis buffer (250 mM NaCl in low salt lysis buffer). Then recovered proteins were separated on 10% SDS-polyacrylamide gels and analyzed by Western blotting using monoclonal ␣-HA antibody (12CA5).

In Vitro Protein Binding Assay
For the preparation of affinity-purified glutathione S-transferase fusion proteins, 400-ml cultures of Escherichia coli carrying the plasmids pGST-Myc 1-262, pGST-Jun 1-223 (kind gift from M. Karin, University of California San Diego), and pGEX4T2 (Promega) were grown to an A 600 ϭ 0.5, induced by the addition of isopropyl-1-thio-␤-D-galactopyranoside to an end concentration of 0.2 mM, and grown for another 3 h. Cells were harvested by centrifugation (GSA, 4500 rpm, 4°C), and the cell pellet was resuspended in 10 ml of NETN buffer (20 mM Tris-Cl, pH 7.5, 100 mM NaCl, 1 mM EDTA, 0.5% Nonidet P-40, 2 mM dithiothreitol, and protease inhibitors). After the addition of 30 l of lysozyme (10 mg/ml), cells were incubated on ice for 45 min and sonicated (6 ϫ 15 s with 30-s incubation on ice), and the lysate was cleared by centrifugation (JA-20; 9500 rpm; 4°C). To the supernatant, 200 l of GSH-agarose beads (50% slurry; Sigma) was added and incubated on a rocker platform for at least 4 h. GSH-agarose beads with bound proteins were washed three times with NETN-buffer. Protein concentration on GSH-beads was normalized to 20 g of protein in 15 l of GSH-beads by the addition of plain GSH-slurry. For in vitro binding reactions, 15 l of GSH-agarose beads (with 20 g of bound protein) was mixed with 400 l of binding buffer (50 mM Tris-Cl, pH 8.0, 200 mM NaCl, 1 mM EDTA, 0.5% Nonidet P-40, 0.2 mg/ml bovine serum albumin), and 2.5 l of [ 35 S]methionine-labeled in vitro translated Krim-1 protein (2 g of YM-Y5A Library-Plasmid; TNT® coupled reticulocyte lysate system; Promega). The binding reaction was incubated on a rocker platform for 1.5 h (4°C) and washed five times with binding buffer. Bound proteins were analyzed by 12% SDS-polyacrylamide gel electrophoresis and autoradiography.

Mammalian Two-hybrid Assay
Expression vectors that code for fusion proteins between the DNA binding domain of Gal4 and the protein of interest (i.e. c-Myc or Krim-1) were cotransfected with vectors, allowing the expression of a putative partner protein or subdomains of this protein fused to a protein that is able to transactivate transcription (here the transactivation domain of the VP16 protein from herpes simplex virus) as well as reporter gene constructs into COS-7 cells. Cells were seeded at 50,000/well into a 12-well plate and transfected with a total amount of 1 g of DNA. The reporter gene construct contained a cassette with four Gal4 DNA binding sites 5Ј of a thymidine kinase minimal promoter driving the luciferase gene. After transfection, cells were washed with PBS and lysed in 100 l of 1% Triton X-100, 1 mM dithiothreitol. 10 l of the lysate was analyzed for luciferase activity as previously described (39 -41). -Fold induction was calculated by dividing the values obtained with the individual mutants by the value obtained with the Gal4-DBD construct alone. Mean values were calculated from at least three (mostly four or five) independent experiments with triplicate measurements each.

Subcellular Localization
Subcellular Fractionation-A series of C-terminal deletion clones of Krim-1 was generated, which contained three N-terminal fused hemagglutinin (HA) tags for immunodetection. Deletions were generated by PCR using the plasmid SK-HA-Krim-1A as template and a T7 promoter primer as a common 5Ј primer. The extent of the C-terminal deletion was specified by the used Krim-specific 3Ј-primers ("1-spacer" (aa 1-184), Y5-184BHSNot: 5Ј-TGTGCGGCCGCTAGGATCCGAAAGTTC-CCTCATAATCG-3Ј; "1-Zn1" (aa 1-214), Y5-214BH: 5Ј-ACGGGATCC-TTGAATGTGACTTCTCTCA-3Ј; "1-Zn2" (aa 1-242), Y5-242BH: 5Ј-A-GGGGATCCACCAGTGTGGATTCTCTCA-3Ј; "1-Zn3" (aa 1-270), Y5-270BH: 5Ј-CCCGGATCCTCCAGTGTGAGTCCTCTCA-3Ј; 1-Zn5Ј (aa 1-344), Y5-344BHSNot: 5Ј-ATCGCGGCCGCTAGGATCCGTATG-CAGAACATTTAAACG-3Ј). The deleted Krim-1 fragments were inserted as N-terminal HA-tagged fragments into a KpnI/BamHI-cut pcDNA3 vector with a stop codon 3Ј of the BamHI site. For the expression of the fusion proteins, the corresponding plasmid (13 g) was introduced into COS-7 cells (10-cm dish) by lipofection with 20 l of Rotifect reagent (Roth, Germany) according to the manufacturer's protocol. Cells were dislodged from the plates by short trypsin digestion and washed with phosphate-buffered saline. Cytoplasmic extracts were prepared by careful resuspension of the cells in 1000 l of CE buffer (20 mM HEPES buffer, pH 7.8, 10 mM KCl, 0.2 mM EDTA, 10% glycerol, 0.1% Nonidet P-40, 1 mM dithiothreitol, proteinase inhibitors), incubation on ice for 10 min, and centrifugation (microcentrifuge, 10,000 rpm, 20 s). The supernatant contained cytoplasmic extract. The pellet was resuspended in 1000 l of CE buffer and sedimented as above. This wash step was repeated. Nuclear extract was prepared by resuspending the washed pellet in 400 l of NE buffer (CE buffer with 500 mM NaCl), incubation for 10 min at 4°C and subsequent agitation on a roller platform (4°C) for another 10 min. The nuclear lysate was cleared by centrifugation (microcentrifuge, 14,000 rpm, 15 min, 4°C), and the pellet was discarded. An equal amount of the obtained extracts was separated by SDS-polyacrylamide gel electrophoresis and analyzed by Western blotting using ␣-HA monoclonal antibody. Cells were seeded on glass coverslips in 12-well plates, and after 12 h they were transfected by calcium phosphate precipitation with 500 ng of GFP fusion construct. After 1.5 days, the cells were rinsed twice with PBS, fixed for 10 min in methanol, and rinsed carefully twice with PBS. Then cells were stained for 20 min with PBS containing propidium iodide and RNase A (250 g/ml) and then rinsed three times with PBS. Coverslips with cells were sealed with ProLong Antifade (Molecular Probes, Inc., Eugene, OR) and analyzed by laser-scanning microscopy (LSM510; Zeiss).

RESULTS
The N Terminus of c-Myc Interacts with a Novel KRAB Boxcontaining Protein Termed Krim-1-To construct a "bait" plas-mid that is able to express Myc as a fusion with Ha-Ras, the N-terminal region (amino acids 1-262) of human c-Myc was amplified using PCR. This fragment was inserted into the yeast bait expression plasmid pADNS-Ras(mut) (37). The resulting construct was sequenced to check the integrity of the Myc insert and in frame insertion. To check the functionality of this vector and the production of a Myc-Ras fusion protein, the bait plasmid was introduced into yeast cells, and the expression of the fusion protein was detected by Western blotting using an antibody against the Ras protein (not shown). To perform a Ras recruitment screening, the bait plasmid as well as an mGAP expression plasmid (36) was introduced into the yeast mutant cdc25-2, which contains a temperature-sensitive allele of CDC25. The GEF CDC25 activates the endogenous yeast Ras pathway under permissive conditions (25°C); however, it can be inactivated by temperature shift to the restrictive temperature of 36°C. After introduction of DNA from the GC library (made from GC-1 rat pituitary cells) into this bait strain by transformation at the permissive temperature, the cells were plated on glucose-containing medium. For RRS selection, the resulting colonies were replica-plated onto galactose-contain-FIG. 1. Two novel zinc finger KRAB domain proteins were isolated in a screening with the N terminus of c-Myc using the RRS. a, RRS specificity tests with various bait constructs. Patch diagram of double transformed cdc25-2 yeast cells carrying either the positive library plasmid (Y5a) and one of the bait plasmids Myc262-5ЈRas, pADNS (empty bait plasmid), or Jun-leucine zipper-5ЈRas (Jun-LZ-Ras) or Brn-3a-5Јras (Brn-Ras). As a positive control, the c-Fos leucine zipper region cloned into the library plasmid to produce a fusion with a myristoylation signal sequence (Myr-Fos) and Jun-LZ-Ras were used. From each yeast transformation, three independent colonies were patched. Glucose, 25°C, all tested yeast clones show the same viability at the permissive temperature of 25°C. From this master plate, the galactose and glucose plates shown in columns 2 and 3 were replica-plated and incubated at 36°C. Galactose, 36°C, at the restricted temperature (36°C), complementation of the cdc25 phenotype occurs only with the positive control (Jun/Fos) and with the library plasmid (Y5a) only in combination with the Myc262-5ЈRas bait used for the screening. Glucose, 36°C, repression of the library plasmid promoter on glucose plates inhibits growth at the restrictive temperature for the positive control as well as for Myc262-5ЈRAS/Y5atransfected cells. b, the N-terminal amino acid sequence encoded by the library clone Y5a from the RRS screen shows a high degree of similarity to the KRAB-A domain consensus sequence. Shown is a schematic representation of the proteins encoded by the longest cDNA that could be isolated on the basis of the sequences in the Y5a clone. The arrow indicates where the coding sequences of the original library clone (Y5a) ended. c, protein sequence of Krim-1A. The shaded boxes delineate the KRAB domain and the 12 zinc finger domains.
ing plates (the Yes library plasmid is galactose-inducible) and shifted to the restrictive temperature (36°C). After 3-4 days, emerging colonies were picked and streaked on glucose plates for recovery. Dependence on expression from the library plasmid was then checked by replica plating of these clones to galactose and glucose plates at the restrictive temperature of 36°C. Glucose represses expression from the library plasmid and should abolish rescue of the temperature-sensitive CDC25-2 phenotype. Of ϳ3 ϫ 10 6 library clones, the c-Myc screening resulted in 570 initial growing clones, from which 32 showed library plasmid dependence on galactose/glucose medium. In these 32 clones, three types of colonies could still be expected: first, true positives carrying Myc-interacting proteins; second, proteins interacting with the Ha-Ras portion of the bait; and third, false positives that activate the yeast RAS pathway like overexpressed rat Sos GEF or small Ras-like proteins from the library, first or second site revertants in the yeast genome (36). To discriminate against these clone types, the library plasmids from these 32 clones were isolated and reintroduced into the mutant cdc25-2 strain along with either the original Myc bait plasmid, the empty ADNS-Ras(mut) plasmid (pADNS, empty bait vector), or a heterologous bait plasmid as, for instance, Ras fusions with c-Jun leucine zipper (Jun-LZ-Ras) or a Ras fusion with the N-terminal part of the POU factor Brn-3a (Brn-Ras) (Fig. 1a). As a positive control, a c-Jun-LZ-Ras bait and a library plasmid expressing the c-Fos leucine zipper domain (Myr-Fos) were used. Only 2 clones of 32 passed this bait specificity test. One clone (Y5a) contained DNA sequences from an as yet unidentified novel protein with a KRAB domain homology at the N terminus (42) and several C 2 H 2 zinc finger domains at the C-terminal half (Fig. 1b). We performed 5Ј-and 3Ј-RACE PCR reactions using rat thymus RNA to obtain the full-length cDNA sequence. RACE results showed that there are two cDNA isoforms of 2113 and 1861 bp in length. One, which we termed Krim-1A (for Krab box proteins interacting with Myc), encodes a KRAB box-containing protein 527 amino acids in length with 12 zinc fingers. The other one, Krim-1B, contained information for an almost identical protein with only nine zinc fingers, where the region between the middle of zinc finger 9 to the middle of zinc finger 12 was deleted (Fig. 1, b and  c). GenBank TM accession numbers are AY195874 for the rat sequence and AY195875 for the murine sequence.
Krim-1 Is a Ubiquitously Expressed Nuclear Protein-RNA from various rat tissues was extracted for quantitative PCR with primers specific for Krim-1A and GAPDH. The Krim-1 expression levels were calculated after normalization over the values obtained for GAPDH. Relative to the Krim-1 levels found in thymus (arbitrarily set to 1), all other organs tested express lower levels except spleen, which contained the highest level (Fig. 2a). Using a multiple tissue Northern blot (Stratagene) and a probe covering rat sequences from the spacer region of Krim-1 between the KRAB domains and the zinc finger domains, we detected two transcripts at around 2.4 and 2.7 kb (Fig. 2b). Both mRNA species were present in all tissues, but both showed highest levels in spleen (Fig. 2b, lane 3). A probe specific for Krim-1A covering sequences deleted in Krim-1B revealed that the major 2.4-kb RNA species is specific for the longer isoform of Krim-1 (Fig. 2b). A transcript specific for Krim-1B was not readily detected. The longest cDNAs that were obtained for Krim-1A and Krim-1B were 2113 and 1862 nt, respectively, suggesting that both still lack sequences presumably in the 3Ј-untranslated region. Interrogating the Celera data base and the NCBI-based expressed sequence tag libraries revealed the existence of a murine cDNA clone with the potential to encode a protein of 534 amino acids with over 94% sequence identity to the rat Krim-1A (Fig. 2c). The zinc finger region was entirely conserved, with no amino acid exchange, and the KRAB domain contained only one altered amino acid, indicating that this so far anonymous cDNA encodes the bona fide murine Krim-1 homologue (Fig. 2c). Although the rat Krim-1 cDNAs lacked a stop codon upstream of the AUG translation initiation triplet, the murine cDNA contained a stop codon 51 nt 5Ј of its AUG, suggesting that the 533-aa Krim-1A and the 451-aa Krim-1B molecules indeed represent the full-length proteins, which adds 6 amino acids to the N terminus of the originally identified sequence (see Fig. 1). According to the Celera data base, the murine Krim-1A gene is located on chromosome 19, and the coding region spreads over four exons, of which the second encodes the entire KRAB domain (Fig. 2d). In rats, the 5Ј-and 3Ј-untranslated regions are identical in Krim-1A and Krim-1B, suggesting that both mRNA molecules originate from the same genomic locus and very probably are generated by alternative splicing, although the splice site is located in an exon (Fig. 2d). The closest human homologue of Krim-1A that is known so far is the KRAB box containing protein Kox 13 with 41% sequence identity and an as yet unannotated sequence with 47% sequence identity on the amino acid level.
To test the subcellular localization of the Krim-1 proteins, the coding sequences were inserted into an expression vector that allowed the production as GFP fusion proteins under the control of a CMV promoter. Transient transfection into COS-7 cells demonstrated that both Krim-1A and Krim-1B reside in the nucleus (Fig. 3a). In the next step, we constructed expression vectors for the production of HA-tagged deletion mutants of Krim-1A and transfected them into NIH 3T3 cells. Cytoplasmic and nuclear extracts were prepared after transfection, separated by SDS-PAGE, and transferred to nitrocellulose membranes. Immunoblot analysis with an ␣-HA antibody demonstrated that only the full-length Krim-1A is entirely located in the nucleus (Fig. 3b, arrowheads). Whereas Krim-1B is mostly nuclear similar to Krim-1A, a small percentage of the Krim-1B protein can be found in the cytoplasm (Fig. 3b, arrowheads). In contrast, mutants that contain the KRAB box domain, the spacer region, and the first three zinc fingers are predominantly cytoplasmic. Only when regions containing zinc fingers 4 and 5 are present do the proteins show significant transport into the nucleus (Fig. 3b, arrowheads), suggesting that the C-terminal zinc finger domains are critical for nuclear localization.
The Krim-1 Protein Interacts with Sequences Covering the Homology Region Myc Box II-To find further support for the genetic evidence described above that suggests an interaction between Krim-1 and c-Myc, biochemical assays were performed. First, the sequences coding for the c-Myc N-terminal 262 amino acids were appended to the glutathione S-transferase (GST) coding sequence and expressed as a fusion protein that binds avidly to glutathione-agarose beads. Radioactively labeled Krim-1 proteins that span the region of the isolated library clone (aa 1-427), the full-length Krim-1 protein without KRAB box domain (aa 65-527), and two truncated Krim-1 mutants (aa 105-527 and aa 171-527) were generated by a coupled in vitro transcription/translation reaction (Fig. 4a). GST-Myc (aa 1-262) protein and as a control GST alone and GST-Jun (aa 1-223) containing the first 223 amino acids of c-Jun bound to glutathione-agarose beads were mixed with radiolabeled Krim-1A proteins and were washed extensively. Labeled proteins bound to immobilized GST fusion proteins were detected by SDS-polyacrylamide gel electrophoresis. Whereas radiolabeled Krim-1A 1-427 and Krim-1A 65-527 could be specifically precipitated with GST-Myc (262 amino acids), no signal was detected with GST alone or with GST-Jun (223 amino acids) (Fig. 4a). The truncated proteins Krim-1A 105-527 and 171-527 lacked specific interaction with Myc GST, very likely due to the predominance of the 12 zinc finger domains (Fig. 4a).
To show that Krim-1 proteins could also be purified from c-Myc complexes in mammalian cells, COS-7 cells were transiently transfected with expression vectors for HA-tagged Krim-1A or Krim-1B, along with c-Myc or the G 1 phase-specific kinase subunit Cdk4 as a control. Lysates from these cells were incubated with peptide-specific polyclonal antibodies directed against c-Myc or Cdk4. The resulting immune complexes were collected, extensively washed, and separated by SDS-PAGE. Immunoblot analyses with ␣-HA antibodies demonstrated that Krim-1A and Krim-1B are able to form a complex with c- Myc   FIG. 2. Krim-1 is expressed ubiquitously at different levels and is conserved in the murine genome. a, quantitative PCR with primers specific for Krim-1A. The data obtained for Krim-1A were normalized on the GAPDH expression levels, and the value for thymus RNA was arbitrarily set to 1. b, polyadenylated RNA on a multiple tissue Northern blot (Stratagene) was hybridized with a probe from Krim-1 containing sequences that encode the region between the KRAB domain and the zinc finger domains to ensure specificity for Krim-1 transcripts (upper panel) and with a probe containing the region missing in Krim-1B to reveal transcripts specific for Krim-1A (lower panel). Lane 1, heart; lane 2, brain; but not with Cdk4 (Fig. 4b). Next, we wished to more closely dissect the region in the N terminus of c-Myc that could be responsible for the interaction with Krim-1 proteins. We used the yeast RRS and generated a series of constructs in the pADNS bait vector for expression of fusion proteins between various Myc mutants (40) and the Ha-Ras sequence already used for the original bait (Fig. 4c). The Myc mutants were made on the basis of the original first 262 N-terminal amino acids and contained small deletions covering either Myc box I (aa 45-63), Myc box II (aa 128 -143), the region responsible for the interaction with the adenoviral protein E1A (aa 109 -126), or both Myc box II and the E1A-interacting region aa 104 -136 (Fig. 4c). In addition, we have also used the mutants Ala 58 / Ala 62 that delete putative phosphorylation sites at aa 58 and 62 and Asp 58 /Asp 62 , mimicking constitutively phosphorylated residues (Fig. 4c). These constructs were cotransfected with the Krim-1-containing, originally isolated Yes library plasmid (clone Y5a) into the yeast strain cdc25-2, and the cells were grown at the restrictive temperature (36°C). Deletion of the Myc box I (aa 45-63) and the E1A-binding region (aa 109 -126) did not have any effect on Krim-1 binding. However, mutants lacking sequences covering Myc box II and the N-terminal adjacent region did not interact with the Krim-1 protein (Fig.  4c), suggesting that the interaction of Krim-1 with Myc requires sequences of Myc box II.
The Transcriptional Corepressor Tif1␤ Stabilizes Krim-1 Expression-To search for proteins that interact with the KRAB box domain of Krim-1, we used the KRAB domain as a bait in a screen with the GC-library. Several bait-specific clones were obtained that contained the same insert and encoded the rat Tif-1␤/Kap-1 protein (data not shown). This protein contains multiple subdomains and has already been described as a cofactor with a role in transcriptional regulation that binds to Krab domains (32,45,46). We have obtained the FLAG-tagged full-length cDNA in a CMV-dependent expression vector (kind gift of W. Schaffner, University of Zü rich, Switzerland) and used it to transfect HeLa cells along with a c-Myc expression vector and vectors for HA-tagged Krim-1A and Krim-1B in order to confirm the interaction between Krim-1 proteins and Tif-1␤/Kap-1 by co-immunoprecipitation. We observed that the HA-Krim-1 proteins were expressed at significantly higher levels when Tif-1␤ was co-expressed (Fig. 5a). In addition, immunoprecipitations with ␣Ϫc-Myc antibodies (9E10) delivered more c-Myc-Krim-1 complexes in the presence of Tif-1␤ than in the absence of this protein (Fig. 5a). Both findings suggest a stabilizing function of Tif-1␤ for both Krim-1 proteins.
The Second N-terminal Zinc Finger of Krim-1A Is Essential for the Interaction with c-Myc-To quantify the binding of Krim-1A or Krim-1 mutants and to identify the region in Krim-1A that binds to the Myc N terminus, we performed a mammalian two-hybrid assay. For this assay, we generated expression constructs that allowed the production of a fusion protein between Myc and the transactivating domain of VP16. The cDNAs for Krim-1A and several Krim-1A mutants were expressed as fusions with a GAL4 DNA binding domain (GAL4-DBD). The interaction between Krim-1A or Krim-1A mutants and Myc was measured through the transcriptional activation of a luciferase reporter construct regulated by four GAL4-DBDbinding sites in the promoter (Fig. 5b). The correct expression of the VP16-Myc and the Gal4-Krim-1 fusion constructs was determined by Western blotting (data not shown). Clearly, the Krim-1A wild type protein and the mutant 65-360 are still able to interact with Myc (Fig. 5b), whereas all other mutants, including a mutant comprising aa 65-360 where specifically Immune precipitation was performed with ␣-c-Myc and ␣-Cdk4 antibodies. Krim-1 proteins were detected by Western blotting with ␣-HA antibodies and were contained in the ␣-c-Myc-precipitated complexes but not in ␣-Cdk4 precipitates. c, the indicated c-Myc mutants were inserted into the pADNS bait vector (pADNS-5ЈRas(mut)) and were expressed as fusions with Ha-Ras in yeast cells that contained the original Krim-1 clone Y5a. Three colonies obtained after transfection were patched for each mutant on galactose plates at the restrictive temperature of 36°C. As a control, the wild type c-Myc N terminus (Myc wt) was used as well as the empty bait vector and the empty library plasmid (YesM). The interaction with Krim-1 proteins requires the sequences covering Myc box II. zinc finger 2 was destroyed (Znf 265-360 ) by replacing the cysteines in positions 219 and 222 against arginines, failed to show such an interaction (Fig. 5b). This suggests that the zinc finger region and in particular an intact zinc finger 2 in Krim-1 is needed to bind the Myc TAD. Next, we transfected HeLa cells with expression vectors for FLAG-Tif-1␤ and HA-Krim-1A or the full-length HA-Krim-1A protein with the mutations in the zinc finger 2 (1A Zn2 ) or with an expression vector for the irrelevant HA-tagged Akt kinase as a control. Co-immune precipitation analyses with an ␣Ϫc-Myc antibody were developed with an ␣-HA antibody and showed that indeed the Krim-1A mutant with the destroyed zinc finger 2 as well as the irrelevant Akt kinase did not interact with the endogenous c-Myc protein (Fig.  5c, lanes 5 and 6). However, both Krim-1A and Krim-1B could be precipitated with antibodies against the endogenous c-Myc protein, without a previous transfection of a c-Myc expression vector (Fig. 5c, lanes 3 and 4). In addition, immune precipitations with an ␣ϪHA antibody and development with an ␣-FLAG antibody confirmed that both Krim-1 proteins can interact with Tif-1␤ (Fig. 5c).
Krim-1 Proteins Are Able to Negatively Regulate Myc Functions-It has been demonstrated before that KRAB box-containing proteins are transcriptional transrepressors (42). We wanted to test whether the KRAB box domain of Krim-1 has transrepressing activity as compared with, for example, the KRAB domain of the Kox-1 zinc finger protein. Therefore, the Krim-1 KRAB domain was expressed as a fusion with the DNA-binding domain of the doxycyclin-dependent Tet repressor protein (TetR-DBD). As expected, a TetR-DBD-Kox fusion protein was able to repress the transcriptional activity of a Tet FIG. 5. The second zinc finger in Krim-1A is essential for c-Myc binding. a, HeLa cells were transfected with expression constructs for HA-tagged Krim-1 proteins and FLAG-tagged Tif-1␤. Lysates were separated by SDS-PAGE, and Western blotting with ␣-HA antibodies and ␣-FLAG antibodies revealed the expression of Krim-1 proteins and of Tif-1␤, respectively. Notably, expression of Krim-1 proteins was higher in the presence of Tif-1␤ than in lysates without Tif-1␤. Immune precipitation with ␣-c-Myc antibodies were developed with an ␣-HA antibody and revealed higher levels of Krim-1 proteins when Tif-1␤ was coexpressed (lanes 5 and 6) than in the absence of Tif-1␤ (lanes 2 and 3). b, mammalian two-hybrid assays to test the interaction of Krim-1 and different Krim-1 mutants with c-Myc. COS-7 cells were transfected with a construct directing the expression of the c-Myc N-terminal 262 amino acids as fusion proteins with the VP16 transactivation domain (VP16-Myc). Co-transfected were expression constructs for fusion proteins between the Gal4 DNA binding domain (Gal4-DBD) and Krim-1A or different mutants of Krim-1A as well as a reporter gene construct containing a Gal4 binding site-dependent luciferase gene. The luminometrically measured enzymatic activities of the luciferase reporter gene are given as "-fold induction" compared with data obtained with Gal4 DNA binding alone. Average values with S.D. values of three independent measurements are given. c, HeLa cells were transfected with the indicated HA-tagged Krim-1 expression constructs and with a plasmid able to direct the expression of a HA-tagged Akt kinase as a control. Western blot with the ␣-HA antibodies demonstrates the expression of Krim-1A, Krim-1B, the Znf2 mutant of Krim 1A (1AZn2), and Akt. Immune precipitates with ␣-c-Myc antibodies were analyzed with ␣-HA antibodies; precipitates with ␣-HA antibodies were developed with ␣-FLAG antibodies.
operator-dependent CMV promoter element in the absence of antibiotic (Fig. 6a). To show that this effect was dependent on DNA binding, repression could be relieved by the addition of doxycyclin, which renders the TetR domain unable to bind to DNA (Fig. 6a). The TetR-Krim fusion showed a very similar activity, confirming that the Krim-1A or Krim-1B proteins contain bona fide KRAB domains with strong repressor activity, which is dependent on DNA binding (Fig. 6a). The repressor activity of both Kox-1 or Krim-1 KRAB domains was not affected by the presence of c-Myc (data not shown). Similarly, none of the Krim-1 proteins showed any effect on the transcription of the adenovirus major late promoter (17), the ferritin promoter (43), or the p15 ink4b promoter (21) that are all thought to be repressed by Myc through interaction with the INR element (data not shown). Next we tested whether Krim-1 could alter the transactivating function of c-Myc, and we transfected a Gal4 binding site-dependent luciferase reporter (H17-MX-Luc) construct together with a plasmid expressing a fusion of the Gal4 DNA binding domain and the c-Myc N terminus (TAD). The Gal4-Myc fusion was able to activate the reporter gene about 4 -5-fold, and coexpression of either Krim-1A or Krim-1B could repress this activation (Fig. 6b). Similar results were obtained when a luciferase reporter construct driven by the E-box-containing promoter of the prothymosin gene (39,44), which represents a bona fide Myc target gene promoter, was cotransfected together with plasmids expressing Myc or the Krim-1 proteins. (Fig. 6, c and d). Both Krim-1A and Krim-1B were able to repress E-box-dependent transcriptional activation of the prothymosin promoter in HeLa cells and in HepG2 cells (Fig. 6, c and d). Since HeLa cells contain very high endogenous levels of c-Myc, the E-box-dependent activation is dependent on limiting levels of Max protein (Fig. 6c). In HepG2 cells, this effect was also seen with Myc alone and without transfection of a Max expression construct. Coexpression of Krim-1A led to a repression of E-box-dependent transactivation by Myc to about 50% and of Krim-1B to even 10% in HeLa cells (Fig. 6c). In HepG2 cells, both proteins led to a repression of reporter gene activity to about 60% (Fig. 6d). The activity of FIG. 6. Krim-1 negatively regulates Myc functions. a, transient transfection of 3T3 cells with a luciferase reporter gene driven by a CMV enhancer/promoter with seven upstream Tet-operators (TetO7-CMV-luc) and CMV expression plasmids. CMV, empty expression plasmid; TetR, DNA-binding domain of the Tet repressor; TetR-Kox1 and TetR-Krim-1, fusion of DNA-binding domain of the Tet repressor with the KRAB domain of either Kox-1 or Krim-1 zinc finger proteins, respectively. The effect of the various fusion proteins on the TetO7-CMVluc reporter was measured, and the luciferase activity was plotted against the constructs transfected. In the presence of doxycyclin, the TetR DNA-binding domain is unable to bind DNA. The transcriptional repression of the KRAB box domains of Kox1 and Krim-1 is DNA-dependent, since the addition of doxycyclin released this repression. b, expression constructs for fusion proteins between the Gal4-DBD and the c-Myc TAD (Gal4-Myc), the Gal4-DBD vector itself, as well as HA-tagged Krim-1 expression constructs were transfected into NIH 3T3 cells along with a reporter gene construct containing four Gal4 binding sites and a minimal promoter-driven luciferase gene. Gal4-Myc fusion proteins were able to activate transcription of the reporter gene, and a repression of this activation is observed when Krim-1 proteins are present. c, HeLa cells were transfected with expression constructs for Krim-1 proteins, c-Myc and Max and the prothymosin reporter gene construct. In HeLa cells, expression of Max was sufficient to activate the prothymosin promoter, since high levels of endogenous c-Myc are expressed in these cells. Both Krim-1A and Krim-1B are able to repress E-box-dependent transactivation, although Krim-1B shows a higher efficiency. d, HepG2 cells were treated as in c. Here, the activation of the prothymosin reporter is seen when Myc alone is overexpressed. Similar to the results described in c, both Krim-1 proteins were able to repress c-Myc transactivation of the prothymosin reporter.
SV40 promoter-driven reporter gene constructs was unaffected by Krim-1 proteins, indicating that the observed effects on the prothymosin and Gal-binding site reporter construct are not due to a nonspecific inhibition of transcription (data not shown).
Another well established function of c-Myc is its ability to cooperate with activated Ha-Ras in the malignant transformation of rat embryo fibroblasts (REFs) (5). The cotransfection of Myc and Ha-Ras expression plasmids leads to the formation of foci of transformed cells. When constructs that drive the expression of HA-tagged Krim-1A or Krim-1B were cotransfected along with c-Myc and the oncogenically activated Ha-Ras mutant, we found that both Krim-1A and Krim-1B were able to repress focus formation (Fig. 7, a and b). Although both Krim-1 expression constructs were able to direct the production of almost identical levels of both proteins in primary REFs, the effect of Krim-1B on focus formation was more pronounced (Fig. 7, a and b). None of the foci that arose from Myc/Ras/Krim transfection and that were established as cell lines contained measurable Krim-1 protein, suggesting that foci do not form in the presence of Krim-1 proteins (not shown). In focus formation experiments using SV40 large T-antigen (SV40LT) and Ha-Ras, the coexpression of Krim-1 proteins did not show any effect (Fig.  7, a and b). Since the SV40 large T-antigen uses different pathways to transform primary cells than Myc, this finding suggested that the inhibitory effect of Krim-1B on focus formation is specific for the Myc function and does not affect the transformation of primary fibroblasts in general (e.g. due to toxic effects). In addition, when the Krim-1AZnf2 65-360 mutant that was no longer able to interact with Myc was used in cotransfections with Ha-Ras, a suppression of focus formation was markedly relieved in four of five independent experiments (Table I). DISCUSSION Myc family proteins control a number of biological processes that are important for cell proliferation, survival, and cell growth. Deregulation of c-Myc expression is strongly correlated with malignant transformation of a large number of cell lineages. Therefore, a tight and highly redundant regulation of c-Myc activity is critical to maintain a physiological status of a cell. c-Myc acts as a transcriptional transactivator of a presumably limited set of target genes through binding at upstream E-box elements and the activation of PolII-dependent transcription. The generally accepted view is that gene activation occurs via E-box elements and the recruitment of chromatinremodeling complexes. Whereas the activation of target genes by c-Myc through E-box elements in target gene promoters is supported by several independent lines of evidence, data that would reliably support a model for c-Myc-mediated transcriptional repression are more scarce but suggest mechanisms that are independent of E-box-mediated gene regulation and are mediated by initiator sequences of the basal promoter (see Introduction).
To investigate in more detail the mechanisms that are involved in c-Myc-mediated transactivation or transrepression, we have used a novel yeast interaction cloning system, the RRS, to identify proteins that bind to the TAD of c-Myc. We were able to isolate a novel 527-aa widely expressed nuclear protein, Krim-1, that contains a KRAB domain at the N terminus and 12 C 2 H 2 zinc finger domains at its carboxyl-terminal  7. Oncogenic activity of c-Myc in primary rat embryo fibroblasts is inhibited by Krim1 proteins. Expression constructs for c-Myc, Ha-Ras, SV-40 T-Ag, and for the HA-tagged versions of both Krim-1A and Krim-1B were transfected as indicated into primary rat embryo fibroblasts as described previously (41). Ha-ras is cotransfected in every experiment. After 10 -14 days, foci of transformed cells were counted. The table (a) shows the number of foci obtained in five independent experiments. b, the functionality of the Krim-1 and c-Myc expression constructs was controlled by immunoblotting of lysates of transiently transfected REFs with the indicated antibodies.
end. The protein exists in two isoforms: a longer Krim-1A and a shorter Krim-1B protein that only contains 9 instead of 12 zinc finger domains. Both Krim-1A and Krim-1B require sequences of the Myc box II homology region to be able to bind to c-Myc. The interaction between Krim-1 and c-Myc proteins was also supported by biochemical evidence. Krim-1-c-Myc complexes could be precipitated from mixtures of in vitro translated Krim-1 and GST-Myc fusion proteins but also from COS-7 or HeLa cells that overexpress both proteins or only Krim-1 proteins after transient transfection. In addition, we could demonstrate that both proteins interact in vivo in a mammalian two-hybrid assay. Our mutational analysis demonstrated that the Krim-1 protein requires the second zinc finger domain for this interaction; the KRAB box itself was found to be dispensable for c-Myc binding.
The KRAB domain is a conserved motif at the N terminus of a large number of zinc finger proteins and comprises a 75amino acid stretch that can be divided into the KRAB A domain, a 45-aa-long minimal repression module, and the KRAB B domain (42). The Krim-1 proteins contain the KRAB A domain only. KRAB box-containing zinc finger proteins have been shown to act as very potent transcriptional repressors that are dependent on DNA binding (32, 46 -52). Our experiment with TetR-Krim fusion proteins clearly demonstrated that Krim-1 is no exception to this rule and also bears this repressor activity. Since the zinc finger domains of KRAB box proteins can interact with a cognate DNA recognition sequence (51) and thus at least have the potential to act as upstream transcriptional regulators, the interaction between Krim-1 and c-Myc could suggest that c-Myc regulates the repressor activity of Krim-1. Whereas our experiments with TetR-Krim fusion proteins support a function of Krim-1 as a transcriptional repressor, we were unable to detect any effect of c-Myc on this activity of Krim-1. Therefore, a regulatory role of c-Myc on DNA-bound Krim-1 seems unlikely. A second possible model would implicate a negative regulatory function of Krim-1 on c-Myc activity given the potential of KRAB box proteins as transcriptional repressors. Our data point to the possibility that Krim-1 proteins could be involved in the regulation of E-box-dependent transactivation of c-Myc. This is supported by assays with the E-box-containing and c-Myc-dependent prothymosin reporter. We observed that Krim-1A and Krim-1B were able to repress c-Myc-mediated activation of both types of reporter gene in three different cell lines. Additional evidence for a role of Krim-1 proteins as repressors of c-Myc activity is provided by transfection experiments with primary REFs. The formation of malignantly transformed foci after cotransfection of constructs directing the expression of c-Myc and the activated human Ha-ras gene was found to be repressed severalfold in the presence of Krim-1A and Krim-1B. The loss of interaction mutant Krim-1AZnf2 65-360 was unable to exert this repression in four of five assays, providing additional support for the specificity of the Krim-1/Myc interaction. It has been shown that binding to Max is required for malignant transformation by c-Myc (53) and that target genes that are transcriptionally activated by c-Myc are themselves active, dominant oncogenes in the REF transformation assay, for instance genes encoding specific regulators of cell cycle progression (28) (reviewed in Refs. 1 and 2). Both findings argue that the E-box-dependent activation of target genes is critical for c-Myc in malignant transformation of fibroblasts. Therefore, our observation that Krim-1 proteins repress malignant transformation by c-Myc supports our view that Krim-1 proteins interfere with E-box-mediated transcriptional transactivation of c-Myc.
The E-box dependent transactivation of target genes by c-Myc requires the recruitment of enzymatic activities to the vicinity of the c-Myc binding site that allow chromatin modification, which in eukaryotic cells is executed by ATP-dependent nucleosome remodeling and by histone acetylation. It is known that c-Myc TAD binds to TRRAP, which is part of multiprotein complexes that also contain histone acetyltransferase activity (25)(26)(27)(28), suggesting that one mechanism of c-Myc-mediated activation of PolII-dependent transcription is the recruitment of histone acetyltransferases and the induction of histone acetylation. In this respect, it is interesting to note that, very similar to the c-Myc/Krim-1 interaction, the recruitment of TRRAP by c-Myc depends on the integrity of Myc box II sequences. This finding underscores the critical role of this homology domain in the process of transcriptional transregulation by c-Myc and may suggest that TRRAP-histone acetyltransferase complexes may compete with other multiprotein complexes for binding to the c-Myc TAD, possibly with Krim-1 or associated proteins.
We present evidence that the KRAB box of Krim-1 is dispensable for c-Myc binding and that a region in the Krim-1 protein that contains this domain can interact with the transcriptional co-repressor Tif-1␤/Kap-1. The interaction of KRAB box repressor domains with Tif-1␤ is supported by many experiments with other, similar KRAB box-containing zinc finger proteins reported in the literature (45,46,49,50,52). Tif-1␤ is a 97-kDa multisubunit phosphoprotein that interacts with the 45-amino acid minimal repression KRAB A box domain as a trimer through the N-terminal Ring-B box-coiled-coil domains, which have been suggested to function as a protein-protein interface (50). Several studies suggest that Tif-1␤ is able to function as a universal transcriptional co-repressor for KRAB box proteins (32,47,48,50,52). The carboxyl terminus of Tif-1␤ contains a plant homeo domain finger and Bromo domains, both of which can also function as transcriptional repressors when linked to heterologous DNA binding domains (50,54,55). Whereas it is widely accepted that Tif-1␤ is required for KRAB domainmediated repression, the exact mechanisms of how this is achieved remain to be clarified. Experimental evidence suggests that Tif-1␤ recruits the HP-1 proteins, which represent chromatin constituents associated with silencing of euchromatic genes (52,56,57). Recently, the methylation of lysine in histone H3 has emerged as one critical enzymatic activity in heterochromatin formation and gene silencing, which provides a signal to enhance the recruitment and binding of HP-1 proteins (58). In addition, Tif-1␤ can use its plant homeo domain and Bromo domains to link KRAB box zinc finger proteins to Mi-2␣ and other components of the nucleosome remodeling and deacetylase complex (55). By this interaction, Tif-1␤ is able to recruit not only HP-1 proteins but also histone deacetylases to KRAB box zinc finger proteins (55). This suggests that Tif-1␤mediated repression relies on multiple mechanisms involving histone modification. Given that these enzymatic activities can be activated or recruited by Tif-1␤, we hypothesize that pathways exists that link the TAD of c-Myc via the Krim-1 protein to the Tif-1␤ co-repressor complex by which both histone methyltransferases and histone deacetylases can be recruited to c-Myc-occupied sites on the chromatin. Although we can clearly demonstrate that c-Myc binds to Krim-1 and that Krim-1 can interact with Tif-1␤, further experiments are needed to show whether indeed a trimeric complex between c-Myc-Krim-1 and Tif-1␤ is formed in cells. This would support the possibility that Krim-1 can serve as a bridging molecule to recruit Tif-1␤ and the appended chromatin remodeling machinery consisting of histone deacetylases and HP-1 to c-Myc occupied sites to silence gene transcription.