Search

Aberrant transcriptional repression through chromatin remodelling and histone deacetylation has been postulated to represent a driving force underlying tumorigenesis because histone deacetylase inhibitors have been found to be effective in cancer treatment. However, the molecular mechanisms by which transcriptional derepression would be linked to tumour suppression are poorly understood. Here we identify the transcriptional repressor Pokemon (encoded by the Zbtb7 gene) as a critical factor in oncogenesis. Mouse embryonic fibroblasts lacking Zbtb7 are completely refractory to oncogene-mediated cellular transformation. Conversely, Pokemon overexpression leads to overt oncogenic transformation both in vitro and in vivo in transgenic mice. Pokemon can specifically repress the transcription of the tumour suppressor gene ARF through direct binding. We find that Pokemon is aberrantly overexpressed in human cancers and that its expression levels predict biological behaviour and clinical outcome. Pokemon's critical role in cellular transformation makes it an attractive target for therapeutic intervention.

THE mixed-lineage leukaemia gene (MLL/HRX/ALL-1) is disrupted by chromosomal translocation in human acute leukaemias that often display mixed lymphoid–myeloid phenotypes and present in infancy1–4. MLL possesses a highly conserved SET domain also found in Drosophila trithorax (trx) and Polycomb group (Pc-G) genes, which are known to regulate homeotic genes (HOM-C) in a positive or negative fashion, respectively5. Mll was targeted in mice by homologous recombination in embryonic stem (ES) cells to assess its role in pattern development. Mll heterozygous (+/−) mice had retarded growth, displayed haematopoietic abnormalities, and demonstrated bidirectional homeotic transformations of the axial skeleton as well as sternal malformations. Mll deficiency (−/−) was embryonic lethal. Anterior boundaries of Hoxa-7 and Hoxc-9 expression were shifted posteriorly in Mll +/− embryos, but their expression was abolished in Mll −/− embryos. Thus Mll is required for proper segment identity in mammals, displays haplo-insufficiency, and positively regulates Hox gene expression.

Chromosomal abnormalities in tumours were recognized at the end of the last century but their significance has only recently become clear. Distinct translocations in leukaemias and in solid tumours lead to the activation of proto-oncogene products or, more commonly, creation of tumour-specific fusion proteins. The proteins in both categories are often transcription factors and thus disruption of transcriptional control plays a major role in the aetiology of cancer. Fusion proteins formed after chromosomal translocations are common in a range of tumour types; these are unique tumour antigens and are therefore potential targets for therapy design.

A KEY event in the response of cells to proliferative signals is the rapid, transient induction of the c-fos proto-oncogene, which is mediated through the serum response element (SRE) in the fos promoter1–4. Genomic footprinting5,6 and transfection experi-ments7–9 suggest that this activation occurs through a ternary complex that includes the serum response factor (SRF)10 and the ternary complex factor p62 (ref. 7). Interaction of p62TCF with the SRF-SRE binary complex requires a CAGGA tract immediately upstream of the SRE (ref. 7). Proteins of the ets proto-oncogene family bind to similar sequences11–13and we have found that a member of this family, Elk-1 (ref. 14), forms SRF-dependent ternary complexes with the SRE. Elk-1 and p62TCF have the same DNA sequence requirements and antibodies against Elk-1 block the binding of both proteins. Furthermore, we show that like p62TCF, Elk-1 forms complexes with the yeast SRF-homologue MCM1 but not with yeast ARG80 (ref. 15). But ARG80 mutants that convey interaction with p62TCF can also form complexes with Elk-1. The similarity, or even identity, between Elk-1 and p62TCF suggests a novel regulatory role for Ets proteins that is effected through interaction with other proteins, such as SRF. Furthermore, the possible involvement of an Ets protein in the control of c-fos has interesting implications for proto-oncogene cooperation in cellular growth control16.

The majority (approximately 75%) of infant acute leukaemias have a reciprocal translocation between chromosome 11q23 and one of several partner chromosomes. The gene at 11q23 (named MLL, ALL-1, HRX or HTRX-1; refs 2-6) has been cloned and shares homology with the Drosophila developmental gene trithorax. Rearrangements of this gene (called HRX here) occur in introns and cluster in a region of approximately 10 kb; individual patients have different breakpoints. Here we describe three pairs of infant twins with concordant leukaemia who each share unique (clonal) but non-constitutive HRX rearrangements in their leukaemic cells, providing evidence that the leukaemogenic event originates in utero and unequivocal support for the intra-placental 'metastasis' hypothesis for leukaemia concordance in twins.

Genomic instability promotes tumorigenesis and can occur through various mechanisms, including defective segregation of chromosomes or inactivation of DNA mismatch repair. Although B-cell lymphomas are associated with chromosomal translocations that deregulate oncogene expression, a mechanism for genome-wide instability during lymphomagenesis has not been described. During B-cell development, the immunoglobulin variable (V) region genes are subject to somatic hypermutation in germinal-centre B cells. Here we report that an aberrant hypermutation activity targets multiple loci, including the proto-oncogenes PIM1, MYC, RhoH/TTF (ARHH) and PAX5, in more than 50% of diffuse large-cell lymphomas (DLCLs), which are tumours derived from germinal centres. Mutations are distributed in the 5' untranslated or coding sequences, are independent of chromosomal translocations, and share features typical of V-region-associated somatic hypermutation. In contrast to mutations in V regions, however, these mutations are not detectable in normal germinal-centre B cells or in other germinal-centre-derived lymphomas, suggesting a DLCL-associated malfunction of somatic hypermutation. Intriguingly, the four hypermutable genes are susceptible to chromosomal translocations in the same region, consistent with a role for hypermutation in generating translocations by DNA double-strand breaks. By mutating multiple genes, and possibly by favouring chromosomal translocations, aberrant hypermutation may represent the major contributor to lymphomagenesis.

THE bcl-2 proto-oncogene is activated by translocation in a variety of B-lymphoid tumours and synergizes with the c-myc oncogene in tumour progression1. The mechanism of synergy is unclear but bcl-2 expression inhibits apoptosis2–6, a property presumably pertinent to its proto-oncogenic mode of action4. We have shown that the c-myc gene is a potent inducer of apoptosis, in addition to its established role in mitogenesis7. Here we show that expression of the bcl-2 protein, Bcl-2, specifically abrogates c-mvoinduced apoptosis without affecting the c-myc mitogenic function. This provides a novel mechanism for oncogene cooperation, of potential importance both in carcinogenesis and in the evolution of drug resistance in tumours.

The NF-kappa B subunits p50 and p65 and the product of the rel proto-oncogene are members of a growing class of transcription factors with a unique DNA-binding and dimerization domain. Nuclear transfer of each of these factors is controlled by cytoplasmic inhibitors, and regulated by specific stimuli. The inhibitors I kappa B-alpha and -beta and pp40 recognize either p65 or the c-rel protein. We show here that the proto-oncogene bcl-3, believed to be involved in certain human B-cell leukaemias, encodes a protein that functions as an I kappa B-like molecule for native NF-kappa B but is specific for the p50 subunit. The ankyrin repeat domain of the bcl-3 product is shown to mediate complex formation with NF-kappa B dimers by contracting the conserved dimerization domain of NF-kappa B.

THE int-2 proto-oncogene encodes several products related to the fibroblast growth factor (FGF) family1,2. FGFs have been associated with mesoderm induction in the amphibian embryo2,3 and int-2 has a distinct pattern of expression throughout development in vertebrates4,5. But evidence for a function of int-2 in embryo-genesis has been lacking. In the mouse embryo, int-2 transcripts have been detected in the rhombencephalon at a developmental stage where classical experiments showed that the induction of the inner ear occurs6,7. This raises the possibility that int-2 may constitute a signal for the induction of the otic vesicle, the primor-dium of the inner ear. We provide direct evidence for this view by showing that (1) the formation of the otic vesicle is inhibited by antisense oligonucleotides targeted to the secreted form of int-2, and by antibodies against int-2 oncoproteins, and (2) basic FGF (bFGF) can mimic the inductive signal in the absence of the rhombencephalon.

THE bmi-1 proto-oncogene can be activated by Moloney murine leukaemia proviral insertions in Eµ-myc transgenic mice1,2. It encodes a highly conserved nuclear protein of 324 amino acids which belongs to a family of proteins containing a putative new zinc-finger. Another closely related member of this family is the mouse protein Mel-18 (ref. 3). Here we report on the cloning and characterization of a homologous gene (D-bmi) from Drosophila melanogaster. Our analysis indicates that distinct domains of the mouse Bmi-1 protein, including the putative zinc-finger motif, are highly conserved within the much larger D-Bmi protein. Chromosomal localization and sequence comparison reveal that D-bmi is identical to Posterior Sex Combs (Psc)4-6 and indicate that the conserved domains between mouse bmi and Psc are also conserved within Suppressor-2 of Zeste (Su(z)2)6-8

Specific genes can be inactivated or mutated in the mouse germ line. The phenotypic consequences of the mutation can provide pivotal information on the function of the gene in development and maintenance of the mammalian organism. The procedure entails homologous recombination in embryonic stem cells, which, on fusion to recipient blastocysts, give rise to chimaeric mice that can transmit the mutant gene to their offspring. Inbreeding can then yield mice carrying the mutation in both alleles allowing the phenotypic analysis of recessive mutations. In addition to mice lacking a particular gene function, cell lines carrying null alleles of normally expressed genes can be instrumental in assessing the function of the gene. These cell lines can either be obtained from homozygous animals or, should the mutation be lethal early in embryonic development, be generated by consecutive inactivation of both alleles by homologous recombination in cultured cells. Here we illustrate the feasibility of this latter approach by the efficient consecutive inactivation of both alleles of the pim-1 proto-oncogene in embryonic stem cells.

ALTHOUGH most proto-oncogenes such as c-myc are involved in cell proliferation, being expressed in a wide range of tissues as well as in progenitors of transformed cells1, others may normally function in cellular differentiation2. We now report on a gene on human chromosome 11, at the junction of the T-cell tumour-associated chromosomal translocation t(11;14) (p15;q11) and known as the 11p15 gene (ref. 3) or Ttg (ref. 4), which is believed to be involved in the pathogenesis of the tumour. It has two transcriptional promoters (both retained by the translocated allele5) and is expressed in tumour cells with neuro-endocrine properties, suggesting that normal expression may occur in nerve cells. Using fusion constructs of one 11p15 promoter and lacZ in transgenic mice, we found that the gene is expressed in a segment-specific manner in rhombomeres6–13 of the developing mouse hind-brain. During subsequent development, the gene is more widely expressed, again in precisely defined regional patterns, but in post-mitotic neurons confined to the central nervous system. Thus, this presumptive T-cell oncogene is both developmentally regulated and segmentally restricted in a tissue different from that in which the original tumour arose.

Fragments of foreign antigens associated with class I molecules of the major histocompatibility complex (MHC) are presented at the cell surface to elicit an immune response. This presentation requires the coordinated expression of several genes contained in the MHC1,2,3,4,5, including those encoding the MHC class I heavy chain, the proteins LMP-2 and LMP-7, which are involved in the proteasomal degradation of cytosolic antigens into peptide fragments that are destined for association with MHC class I molecules, and TAP-1 and TAP-2, which transport these fragments across the membrane of the endoplasmic reticulum at the start of their journey to the cell surface. In many virus-transformed cell lines6,7 and spontaneous tumours8,9,10, these genes are simultaneously repressed. However, the key factor(s) that are essential for their expression and repression have not been identified. Here we report that the proto-oncogene product PML induces expression of LMP-2, LMP-7, TAP-1 and TAP-2 in an MHC-class I-negative, recurrent tumour, leading to the re-expression of cell-surface MHC in tumours and to rejection of the tumours. PML also regulates MHC expression in untransformed fibroblasts. We conclude that malfunction of PML may enable a tumour to evade the immune defence of its host.

HIRSCHSPRUNG'S disease is a genetic disorder of neural crest development affecting 1 in 5,000 births. It is characterized by the absence of intramural ganglion cells in the hindgut, which often results in partial to complete intestinal obstruction during the first years of life. An autosomal dominant gene causing this disease was recently mapped to chromosome 10q11.2 (refs 1, 2), using an interstitial deletion of this region isolated in a cell hybrid3,4. It was subsequently localized to a 250-kilobase interval which contains the RET proto-oncogene5. Using flanking intronic sequences as primers6 to amplify 12 of the 20 exons of RET from genomic DNA of 27 Hirschsprung's disease patients, we have now identified four mutations (one frameshift and three missense) that totally disrupt or partially change the structure of the tyrosine kinase domain of the RET protein (Ret). Mutations in the extracellular cysteine-rich domain of Ret have been identified previously7,8 in patients with multiple endocrine neoplasia type 2A, and a targeted mutation in the tyrosine kinase domain of the same gene produces intestinal aganglionosis and kidney agenesis in homozygous transgenic mice9. Our results support the hypothesis that RET, in addition to its potential role in tumorigenesis, plays a critical role in the embryogenesis of the mammalian enteric nervous system.

Multiple endocrine neoplasia type 2A (MEN 2A) is a dominantly inherited cancer syndrome that affects tissues derived from neural ectoderm. It is characterized by medullary thyroid carcinoma (MTC) and phaeochromocytoma. The MEN2A gene has recently been localized by a combination of genetic and physical mapping techniques to a 480-kilobase region in chromosome 10q11.2 (refs 2,3). The DNA segment encompasses the RET proto-oncogene, a receptor tyrosine kinase gene expressed in MTC and phaeochromocytoma and at lower levels in normal human thyroid. This suggested RET as a candidate for the MEN2A gene. We have identified missense mutations of the RET proto-oncogene in 20 of 23 apparently distinct MEN 2A families, but not in 23 normal controls. Further, 19 of these 20 mutations affect the same conserved cysteine residue at the boundary of the RET extracellular and transmembrane domains.

WHEN the mammalian proto-oncogene bcl-2 is overexpressed it can protect various types of cells both from normal and from experimentally induced apoptosis1–6 A but the molecular mechanisms involved are unknown. Although the Bcl-2 protein is membrane-associated7–10, its subcellular location is controversial: two studies have suggested that it is mainly associated with the nuclear envelope and endoplasmic reticulum8,10, whereas another study has suggested that it is mainly located in the inner mitochondrial membrane9. The latter study has suggested that Bcl-2 might protect cells from apoptosis by altering mitochondrial function and that mitochondria may be involved in apoptosis9,11. Here we report that human mutant cell lines that lack mitochondrial DNA (mtDNA), and therefore do not have a functional respiratory chain, can still be induced to die by apoptosis, and that they can be protected from apoptosis by the overexpression of bcl-2, suggesting that neither apoptosis nor the protective effect of bcl-2 depends on mitochondrial respiration. We also show that the Bcl-2 protein in overexpressing cells is associated with the nuclear envelope and endoplasmic reticulum, as well as with mitochondria.

SYNTHETIC antisense oligonucleotides have been used to dissect gene function in vitro. Technical difficulties prevented the use of this approach for investigating the effect of gene products in vivo. Here we report the use of local delivery of antisense c-myb oligonu-cleotide to suppress intimal accumulation of rat carotid arterial smooth muscle cells. Our results suggest that antisense oligonucleotides can be used to define the in vivo biological role of specific macromolecules in the blood vessel wall and could potentially serve as a new class of therapeutic agents for cardiovascular disorders.

Ha-Ras augments c-Jun-mediated transactivation by potentiating the activity of the c-Jun activation domain. Ha-Ras also causes a corresponding increase in phosphorylation of specific sites in that part of the c-Jun protein. A Ha-Ras-induced protein kinase cascade resulting in hyperphosphorylation of the c-Jun activation domain could explain how these oncoproteins cooperate to transform rat embryo fibroblasts.

Nerve growth factor (NGF) interacts with two different low-affinity receptors that can be distinguished by affinity crosslinking. Reconstitution experiments by membrane fusion and transient transfection into heterologous cells indicate that high-affinity NGF binding requires coexpression and binding to both the low-affinity NGF receptor and the tyrosine kinase trk gene product. These studies reveal a new growth factor receptor-mediated mechanism of cellular differentiation involving trk and the low-affinity NGF receptor.

Analysis of complementary DNA for human erythroid ankyrin indicates that the mature protein contains 1,880 amino acids comprising an N-terminal domain binding integral membrane proteins and tubulin, a central domain binding spectrin and vimentin, and an acidic C-terminal 'regulatory' domain containing an alternatively spliced sequence missing from ankyrin variant 2.2. The N-terminal domain is almost entirely composed of 22 tandem 33-amino-acid repeats. Similar repeats are found in yeast and invertebrate proteins involved in cell-cycle control and tissue differentiation.

In vertebrates, mature eggs are arrested at the second meiotic metaphase by the cytostatic factor (CSF), now known to be the c-mos proto-oncogene product (Mos). Fertilization or egg activation triggers a transient increase in the cytoplasmic free calcium and releases the meiotic arrest by inactivating maturation/mitosis-promoting factor (MPF). CSF or Mos, which is also inactivated by the calcium transient, seems to stabilize MPF in mature eggs and CSF-injected embryos. Thus, it was assumed that CSF inactivation is the primary cause of MPF inactivation on meiotic release. We have directly compared the degradation kinetics of CSF (Mos) and MPF during meiotic release, using the same batch of Xenopus eggs. We report here that, at the molecular level, cyclin subunits of MPF are degraded before Mos is degraded and, at the physiological level, that MPF activity is inactivated before CSF activity during activation of Xenopus eggs. These results, in conjunction with circumstantial evidence, support the novel view that a calcium transient on fertilization induces a CSF-independent pathway for MPF inactivation, whereas CSF inactivation during meiotic release serves only to allow the fertilized egg to enter mitosis.

There has been much interest in the biochemical and biophysical processes that couple extracellular signals to alterations in gene expression. While many early events associated with the treatment of cells with growth factors have been described (for example, ion flux and protein phosphorylation), it has proved difficult to establish biochemical links to gene expression. Recently, the study of such genomic control signals has been facilitated by the demonstration that the c-fos proto-oncogene is rapidly and transiently induced by treatment of several cell types with polypeptide growth factors and other growth modulating substances. In one particular system it has been shown that nerve growth factor (NGF) causes a transient induction of c-fos in the phaeochromocytoma cell line PC12, within 15 min. Furthermore, the magnitude of this induction can be modulated with pharmacological agents such as peripheral-type benzodiazepines (BZDs). Thus, the study of c-fos expression in PC12 cells could yield valuable clues to the coupling mechanisms linking cell surface activation to genomic events. Here we demonstrate that c-fos is induced in PC12 cells either by receptor-ligand interaction or by agents or conditions that effect voltage-dependent calcium channels.

The induction of tumours by retroviruses lacking transduced oncogenes can involve the transcriptional or functional activation of cellular proto-oncogenes by an integrated provirus. Thus, the two cellular genes int-1 and int-2, identified as common targets for activation by mouse mammary tumour virus (MMTV), may constitute previously unrecognized oncogenes. In tumours, proviral insertion at these loci leads to expression of messenger RNAs which are undetectable in normal mammary glands. Here we report that in a survey of the two transcriptional activity and structural integrity of the two int loci in 30 BR6 mouse mammary tumours, around 50% of the tumours expressed both of these genes, in ostensibly monoclonal cell populations. Our data suggest that int-1 and int-2 may act cooperatively in the genesis of mammary carcinomas. However, because three tumours (10%) involved neither gene, and because in five cases activation occurred in the apparent absence of an adjacent provirus, it is clear that other loci and mechanisms contribute to tumorigenesis.

The ras proto–oncogene1, found in all eukaryotes so far examined2–9, encodes a protein with guanine nucleotide-binding and GTPase activity10–13. Gene disruption experiments in yeast indicate that ras is essential for cell growth14. Anti-sense mutagenesis approaches suggest that this is also true for Dictyostelium15. Most mutations causing an amino-acid substitution for Gly 12 result in decreased GTPase activity and produce a transforming phenotype16–19. In yeast, a Gly 19 → Val 19 missense mutation (Gly 19 is similar to Gly 12 in mammalian and Dictyostelium ras proteins) causes a series of dominant phenotypes, including elevated adenylate cyclase activity20. In mammalian cells there is no evidence that ras activates adenylate cyclase activity21. D. discoideum contains a single ras gene (Dd-ras) that encodes a protein very similar to the mammalian ras protein4 and identical to c-ras at the potentially transforming positions. Dd-ras is expressed in vegetative cells and later in development in prestalk cells whereas ras protein is found in vegetative and developing cells. In the migrating pseudoplasmodium, ras protein is found in prestalk but not prespore cells, suggesting it is involved in the function and/or differentiation of the anteriorly localized prestalk cells. In this report we examine the effects of expression of a Dd-ras gene carrying a Gly 12 → Thr 12 missense mutation.

The proto-oncogene c-myc is the cellular homologue of the transforming sequence carried by the avian myelocytomastosis virus MC29. A growing body of evidence implicates structural and functional alterations in and around proto-oncogenes such as c-myc in tumorogenesis. Here we report that comparison of the structure of myc from a ductal adenocarcinoma of the breast and from normal breast tissue of the same patient (Sc) revealed a tumour-specific rearrangement of one myc locus and amplification of the other myc locus. (For myc reviews see refs 1-4; for myc involvement in breast neoplasia see refs 5-7.) Within the second intron of the rearranged locus was a non-myc sequence with nearly complete homology to a long interspersed repetitive element (a LINE-1 sequence or L1). In this case, the L1 sequence has functioned as a mobile genetic element to produce a somatic mutation.

Many receptors, in response to ligand activation, trigger inositol phospholipid breakdown, which leads to rapid intracellular responses. The sustained activation of this pathway is believed to be at least one of the factors involved in the stimulation of cell growth and there has been much speculation that certain oncogenes use this pathway to effect uncontrolled cellular proliferation. It has been suggested, by analogy with the receptor-mediated control of adenylate cyclase, that the receptor stimulation of inositol phospholipid metabolism is mediated through a guanine nucleotide regulatory protein (G-protein) called Gp (or Np). Although such a species has not been identified, there is now strong experimental evidence that this process is mediated by a G-protein distinct from the stimulatory and inhibitory G-proteins (Gs and Gi, respectively). The ras genes code for a plasma membrane protein, p21, whose only known biochemical property is a high-affinity GTPase activity. We show here that the expression of normal p21N-ras in NIH 3T3 fibroblasts leads to the coupling of certain growth factor receptors to stimulated inositol phosphate production. We propose that the N-ras proto-oncogene encodes a protein which couples the receptors for certain growth factors to the stimulation of phospholipase C. Thus, N-ras p21 may be the putative Gp or a functionally related protein.

Recent studies have shown that normal cellular sequences on chromosome 13 are lost during the development of retinoblastomas and that sequences on chromosome 11 are similarly lost during the development of Wilms' kidney tumours and embryonal tumours. Cells from these tumors have been found to contain either the paternal or maternal copies of loci on the affected chromosome, but not both. Thus, the somatic loss of heterozygosity for sequences on chromosome 13 or 11 is hypothesized to result in homozygosity for a recessive mutant allele on these chromosomes, and in this way the chromosomal loss may contribute to the development of these tumours. We sought to investigate whether similar losses of heterozygosity for chromosome 11 sequences occurred in a common adult tumour. We chose to analyse bladder cancers, since such cancers are common in the adult population and are derived from urogenital tissue, as are Wilms' tumours. We examined constitutional and tumour genotypes at loci on the short arm of chromosome 11 (11p) in 12 patients with transitional cell carcinomas. In five tumours, we observed the somatic loss of genes on 11p resulting in homozygosity or hemizygosity of the non-deleted alleles in the tumour cells. Our results show that the frequency of loss of 11p sequences in bladder cancer approaches that seen in Wilms' tumour (42% compared with 55%), and suggest that recessive genetic changes involving sequences on 11p may contribute to the development of bladder neoplasms.

In several cell types, messenger RNA levels of the nuclear proto-oncogene c-myb vary as a function of cellular proliferation ; a transient increase in c-myb steady-state mRNA, mediated by post-transcriptional mechanisms, occurs during cell-cycle progression. In contrast, both quiescent and proliferating immature thymocytes contain exceptionally high levels of c-myb mRNA as a consequence of increased c-myb transcription.

Proto-oncogenes encode proteins with three main sites of action: the cell-surface membrane, the cytoplasm and the nucleus. Although the exact biochemical function of most proto-oncogene products is not understood, several of them are known to be involved in signal transduction. A role in gene regulation through DNA binding has been suggested for a recently isolated member of the group of oncogenes acting at the nucleus, v-jun. The C-terminus of the putative v-jun-encoded protein is similar in sequence to the C-terminus of the yeast transcriptional activator GCN4 (refs 8, 9), which forms its minimal DNA-binding domain. GCN4 binds to specific sites whose consensus sequence is highly similar to the recognition sequence of the mammalian transcriptional activator AP-1 (refs 12, 13). Like GCN4, AP-1 binds to promoter elements of specific genes and activates their transcription. Because of the similarity between the recognition sites for GCN4 and AP-1, we examined the possibility that AP-1 could be the product of the c-jun proto-oncogene. The experimental results reported here indicate that the JUN oncoprotein is a sequence-specific transcriptional activator similar to AP-1.