Cell, Vol. 64, 303-312,
January
25, 1991, Copyright
0 1991 by Cell Press
Oncogenic Conversion by Regulatory Changes Transcription Factors Benjamin Cell
Review in
Lewin
Revealed by their presence in transforming viruses or by the ability to transform 3T3 cells in transfection assays, some 100 oncogenes have been identified. The functions of the corresponding proto-oncogenes from which they are derived fall into several groups, ranging from those that are presumed to initiate oncogenic cascades (such as cell surface receptors engaged in signal transduction) to those that may directly change patterns of gene expression (corresponding to transcription factors). The latter class are especially interesting since they offer the prospect of characterizing the target genes that are controlled by the proto-oncogene; among these loci may be genes whose expression is altered by the aberrant function of the oncogene. One attractive paradigm for transformation is that these genes may be responsible for the set of epigenetic changes that convert a normal cell into a tumorigenic cell. Here I consider in detail the properties of the products of four such gene families, fel, jun, fos, and erbA, identified by v-one genes in retroviruses and by counterpart cellular c-one genes. In the cases of Rel, Jun, and ErbA proteins, there are differences in transcriptional activity between the c-One and v-One proteins that may be related to transforming capacity. Other oncogenes whose products function in the nucleus include v-myb and v-myc, but the characterization of their functions in regulating transcription is less advanced. Myb proteins activate transcription, but the relationship between the functions of c-Myb and v-Myb is not clear. Myc proteins bind to DNA, but have yet to be shown directly to activate individual genes. Although DNA-transforming viruses also code for oncoproteins that affect transcription, including most notably the polyoma T antigens and adenovirus ElA, these proteins have pleiotropic effects, and in none of these cases has the ability to influence transcription been correlated with transforming activity. T antigens and ElA do not themselves function as transcription factors and may therefore influence transcription indirectly, possibly involving functions different from those required for transformation. In this review, I shall address the relationship between the retroviral-transforming v-one genes and the c-one genes coding for transcription factors to ask whether changes in the control of transcription may account for the generation of the oncogenic phenotype. As with other classes of oncogenes, we may enquire whether the expression patterns of the cellular proto-oncogenes offer any insights into the specificity of their oncogenic derivatives. As oncogenes carried by retroviruses, the v-one genes are defined as dominant oncogenes. Their actions may in principle be quantitative or qualitative, and those that af-
feet transcription might either increase or decrease expression of particular genes. By virtue of increased expression or activity they could turn up transcription of genes whose products can be tolerated only in small amounts. Failure to respond to normal regulation of activity by other cellular factors also might lead to increased gene expression. A less likely possibility is the acquisition of specificity for new target promoters. Alternatively, if the oncoproteins are defective in the ability to activate transcription, they might function as dominant negative suppressors of the cellular transcription factors. The first steps toward distinguishing these possibilities lie with determining which functions are altered in v-One as opposed to c-One proteins: is DNA-binding altered either quantitatively or qualitatively; is the ability to activate transcription altered? Recent work provides the first answers. The rel Gene Family Includes a Morphogen and Transcription Factor The oncogene v-rel was identified as the transforming function of the avian (turkey) reticuloendotheliosis virus. The retrovirus is highly oncogenic in chickens, where it causes B cell lymphomas. Figure 1 summarizes the relationship between the transforming and cellular genes; v-rel is a truncated version of c-rel, lacking the ~100 C-terminal amino acids, and has a small number of point mutations in the remaining sequence (Wilhelmsen et al., 1984). (Of course, v-rel and the other v-one genes discussed here are synthesized as part of retroviral polyproteins in which viral sequences-usually derived from gag and envgenes-flank the transforming sequence; the figures here compare only the v-one region with the c-one gene. It is probably usually the case that the flanking retroviral sequences do not directly contribute to oncogenicity, but they may not always be irrelevant.) The fel gene turns out to be part of a family with interesting members whose properties have profound implications for the regulation and function of c-rel and we/. Two classes of cellular genes that belong to this family are summarized in Figure 1: 9 The dorsal gene of Drosophila melanogaster is involved in establishing the dorsal-ventral axis. Coding for a protein of 678 amino acids, the region between amino acids 46 and 295 is closely related to c-n?/, with -80% similarity allowing for conservative substitutions (Steward, 1987). l
The transcription factor NF-KB was originally discovered as a regulator of transcription via an enhancer associated with the immunoglobulin K genes; the KB target site has since been found in many other enhancers and promoters. Excepting the N-terminal and C-terminal regions, the DNA-binding subunit of NF-KB is ~60% similar to c-re/(Kieran et al., 1990; Ghosh et al., 1990).
Cell 304
IKB NFKB 65K 50K Activation 1
I
:::.:.. .......~. ,:,:.:.:.:. ,.:.:.. .... .:.:.:.:. ..:.:.:.. +
8
V-Rel :~ii:2$3it!riii 80% .~.A..1..~11,.1..>..1.. ..:........
si m i lo rityi$;$z-zii, ... . . . .-....:./ :.
6761
dorsalr
DNA-binding
& dimerizotion
IKB association?
Activation Cytoplasm
Figure 1. v-rel Is a Truncated Form of c-ml That Lacks Region and Has ~4% Point Substitutions The Rel proteins are related the morphogen dorsal.
to the transcription
factor
the C-Terminal NF-KB and to
The immediate implication of these homologies is that the members of this family function as regulators of transcription. This is consistent with knowledge about dorsal, which is a maternal effect gene that probably regulates embryonic dorsal-ventral polarity directly. The homology with NF-KB suggests that it does so by controlling the response of a set of target genes whose products are involved in creating this axis. Indeed, dorsal protein binds to the promoters of target genes (see below). These comparisons suggest that the basis for v-refs properties may lie in a distortion of the properties of c-rel, which may be one of a number of proteins that either constitute NF-KB activity or provide similar activities. Dorsal and NF-KB share the unusual property that their movement from cytoplasm to nucleus is a regulatory event, Features in the v-Rel and c-Rel proteins that are responsible for their localization in either nucleus or cytoplasm have been identified, but any relationship to the oncogenic properties of v-Rel is unclear at present. Nuclear localization is crucial to dorsal function, since dorsal protein establishes the dorsal-ventral axis by dint of forming a nucleocytoplasmic gradient (Rushlow et al., 1989; Steward, 1989; Roth et al., 1989). Proceeding from the dorsal to the ventral side of the embryo, an increasing amount of the dorsal protein is found in the nucleus as opposed to the cytoplasm. Since the total amount of dorsal protein is initially more or less constant across the embryo, it appears to be the nuclear gradient that is responsible for dorsoventral morphogenesis. The region of homology between dorsal and rel includes a sequence that may be required for nuclear localization, but the sequence that regulates dorsal protein movement may be different; removal of the C-terminus of dorsal causes the protein to enter the nucleus in tissue culture cells, which suggests that this (unique) sequence may function as a cytoplasmic retention signal. This may be a common feature of Rel proteins, since deletion of C-terminal sequences from chick c-Rel allows the protein to enter the nucleus in fibroblasts.
Figure 2. NF-KB Contains Subunits of 50 kd and 65 kd; the 65 kd Subunit Is Bound by the Inhibitor IKB Activation causes phosphorylation of IICE, which then dissociates from NF-KB. Free NF-KB migrates to the nucleus, where it binds to KB sites at promoters or enhancers of target genes. Note that NF-KB is shown for simplicity as a dimer, but may function as a tetramer.
The transcription factor NF-KB is also regulated by nucleocytoplasmic transport. NF-KB is an inducible factor, but the induction does not require synthesis of protein. In uninduced cells, NF-KB is present in the cytosol, where it is bound to an inhibitor protein, IKB. Figure 2 shows that upon induction by any one of a variety of agents (including certain cytokines), IKB becomes phosphorylated and as a result releases NF-KB, which is then transported to the nucleus (Baeuerle and Baltimore, 1988; Ghosh and Baltimore, 1990). NF-KS activity resides in a heteromeric protein that contains two types of subunit, ~50 and ~65. The relationship of p50 to Rel has been established by sequence (as summarized in Figure 1); ~65, which has yet to be fully sequenced, may also be related to Rel. DNAbinding activity is certainly possessed by ~50; the DNAbinding activity of ~65 is not quite clear. IKB binds to ~65, and thereby holds the intact (heteromeric) factor in the cytoplasm. It remains to be seen whether the mechanism for controlling nucleocytoplasmic localization of dorsal and NFKB is the same. The product of the Drosophila gene cactus may be involved in controlling localization of dorsal, and one of the most interesting questions of the moment is therefore whether cactus will be related to the gene coding for IKB. The C-terminal sequences that are required for cytoplasmic retention of both dorsal and c-Rel are deleted from v-Rel. Since v-Rel is located in the nucleus in chick embryo fibroblasts, it was originally thought that differences in localization might explain the oncogenicity of v-Rel compared with c-Rel, but later it appeared that this is not the case. Addition of the C-terminal sequences of c-Rel to v-Rel results in its retention in the cytoplasm, but it remains transforming (Gilmore and Temin, 1988; Hannink and Temin, 1989). Furthermore, v-Rel is largely cytoplasmic in spleen cells, suggesting that there may be more than one mechanism for determining the localiza-
Review: 305
Oncogenic
Transcription
Factors
tion of Rel proteins. The ability of v-Rel to transform in either cytoplasmic or nuclear state remains a puzzle, but raises the possibility that the oncoprotein has more than one mode of action. What regions of rel are required for transformation? Analysis remains at a primitive stage, but generally it seems that the N-terminal half of the v-Rel protein is important. This region is required for DNA-binding and for dimerization, but its role in activating transcription is unclear. Experiments in which Rel proteins are fused to yeast DNA-binding sequences and then tested for transcriptional activation agree that the C-terminal region of c-Rel (which is missing from v-Rel) is a potent activator; but there is disagreement on whether the N-terminal region is itself an additional activator or instead represses the C-terminal activator (Kamens et al., 1990; Bull et al., 1990). At least four proteins related immunologically to v-Rel are found in the cytoplasm of resting T cells, but enter the nucleus following activation. All of these proteins bind to DNA, and have been described by the size of adducts cross-linked to DNA as ~50, ~55, ~75, and p85 (Ballard et al., 1990). ~50 is a product of ~55, which corresponds to the small (50 kd) subunit of NF-ICB. In addition to its ability to form an oligomer with the 65 kd (non-DNA-binding) subunit of NF-KB, ~55 (NF-KB 50 kd) can form an oligomer with ~75 that binds to DNA. The largest of the proteins, ~85, appears to correspond to c-Rel. It is possible therefore that NF-KB is one of several transcription factors that can be formed by combinatorial associations between various subunits, and the functional consequences of this situation remain to be explored. Other open questions concern the nature of differences that may exist between individual members of the family, and to what degree the pleiotropic effects attributed to NFKB may reside in different family members as opposed to a single transcription factor. jun and fos Gene Products Contribute to AP-1 and Activate Transcription The transcription factor AP-1 recognizes a short sequence (TGACTCA) originally identified in the enhancer of SV40, but since found in promoters and enhancers of cellular genes. It is the nuclear factor required to mediate transcription induced by phorbol ester tumor promoters (such as TPA), and an AP-1 binding site confers TPA inducibility upon a target gene (Angel et al., 1987; Lee et al., 1987). Preparations of the factor include a variety of polypeptides, amongst which are the products of the genes c-jun and &OS. Thev-jun gene is carried by avian sarcomavirus 17 (ASV-17) which induces fibrosarcomas in chickens, and can transform chick embryo fibroblasts. The v-fos gene provides the transforming activity of FBJ murine osteosarcoma virus (FBJ-MuSV), which induces chondrosarcomas. (v-fos is also found in a chicken sarcoma virus.) Some original confusion concerning the active components of AP-1 preparations is explained by the fact that both c-Jun and c-Fos proteins are present and contribute to the activity. The subunits can form c-Jun homodimers as well as c-Jun-c-Fos heterodimers. Jun was identified as a component of AP-1 by the
demonstration that antibodies to v-Jun cross-react with AP-1; this was followed by cloning of the c-jun gene and the demonstration that its product is present in AP-1 preparations (Bohmann et al., 1987). Figure 3 shows that v-Jun is derived from c-Jun by the deletion of a sequence of 27 amino acids called the 5 region as well as by individual point mutations in the remaining region. In addition, the v-Jun mRNA lacks an extensive 3’ nontranslated region that is present in c-Jun transcripts. c-Jun can be converted into a transforming species by introducing some of the features characteristic of v-Jun (Bos et al., 1990). Removal of the 6 region increases the ability to transform chick embryo fibroblasts by an order of magnitude; deletion of the 3’ nontranslated region also increases the number of foci (possibly because this region contains a sequence that destabilizes the mRNA and therefore reduces the quantity of protein product). The point mutations do not seem to be involved in oncogenic capacity. Jun proteins have three regions toward the C-terminus that are involved in the ability to activate transcription. A leucine zipper provides a dimerization motif; since dimer formation is essential for binding to DNA, loss of this function leads indirectly to inability to bind DNA. An adjacent basic region (located in effect at the foot of the zipper) is required directly for DNA binding, and close by is a region (A2) rich in proline residues that is required to activate transcription. These three regions are sufficient for activity in HeLa cells. Both c-Jun and v-Jun proteins rely upon their zippers to form dimers that activate transcription at AP-1 target sites (Bohmann and Tjian, 1989). Although c-jun is the closest homolog to v-jun in human genomic DNA, the iun gene family contains other members, jun6 and ]unD, which possess indistinguishable abilities to form homodimers amongst themselves, heterodimers in any pairwise combination, and heterodimers with Fos; all these forms can bind to DNA at the AP-1 site (Nakabeppu et al., 1988). However, homodimers formed with JunB do not activate transcription at promoters that contain individual AP-1 sites, although they can do so when multimeric response elements are present (Chiu et al., 1989). Correspondingly, in cotransfection experiments, JunB antagonizes the transforming capacity of c-Jun (which has transforming abilities when overexpressed, although it is much less ef0
40
67
340
c-Jun
9 ..,,Ji
3
substitutions
‘.j v-Jun c Al
> Activation
A2 Activation (Pro rich)
000 J
DNA-binding (Basic) Dimerizotion (Leu zipper)
Figure 3. v-jun Is Related to c-jun by Deletion of the 27 Amino Acid S Region and by a Very Small Number (
January
25, 1991, Copyright
0 1991 by Cell Press
Oncogenic Conversion by Regulatory Changes Transcription Factors Benjamin Cell
Review in
Lewin
Revealed by their presence in transforming viruses or by the ability to transform 3T3 cells in transfection assays, some 100 oncogenes have been identified. The functions of the corresponding proto-oncogenes from which they are derived fall into several groups, ranging from those that are presumed to initiate oncogenic cascades (such as cell surface receptors engaged in signal transduction) to those that may directly change patterns of gene expression (corresponding to transcription factors). The latter class are especially interesting since they offer the prospect of characterizing the target genes that are controlled by the proto-oncogene; among these loci may be genes whose expression is altered by the aberrant function of the oncogene. One attractive paradigm for transformation is that these genes may be responsible for the set of epigenetic changes that convert a normal cell into a tumorigenic cell. Here I consider in detail the properties of the products of four such gene families, fel, jun, fos, and erbA, identified by v-one genes in retroviruses and by counterpart cellular c-one genes. In the cases of Rel, Jun, and ErbA proteins, there are differences in transcriptional activity between the c-One and v-One proteins that may be related to transforming capacity. Other oncogenes whose products function in the nucleus include v-myb and v-myc, but the characterization of their functions in regulating transcription is less advanced. Myb proteins activate transcription, but the relationship between the functions of c-Myb and v-Myb is not clear. Myc proteins bind to DNA, but have yet to be shown directly to activate individual genes. Although DNA-transforming viruses also code for oncoproteins that affect transcription, including most notably the polyoma T antigens and adenovirus ElA, these proteins have pleiotropic effects, and in none of these cases has the ability to influence transcription been correlated with transforming activity. T antigens and ElA do not themselves function as transcription factors and may therefore influence transcription indirectly, possibly involving functions different from those required for transformation. In this review, I shall address the relationship between the retroviral-transforming v-one genes and the c-one genes coding for transcription factors to ask whether changes in the control of transcription may account for the generation of the oncogenic phenotype. As with other classes of oncogenes, we may enquire whether the expression patterns of the cellular proto-oncogenes offer any insights into the specificity of their oncogenic derivatives. As oncogenes carried by retroviruses, the v-one genes are defined as dominant oncogenes. Their actions may in principle be quantitative or qualitative, and those that af-
feet transcription might either increase or decrease expression of particular genes. By virtue of increased expression or activity they could turn up transcription of genes whose products can be tolerated only in small amounts. Failure to respond to normal regulation of activity by other cellular factors also might lead to increased gene expression. A less likely possibility is the acquisition of specificity for new target promoters. Alternatively, if the oncoproteins are defective in the ability to activate transcription, they might function as dominant negative suppressors of the cellular transcription factors. The first steps toward distinguishing these possibilities lie with determining which functions are altered in v-One as opposed to c-One proteins: is DNA-binding altered either quantitatively or qualitatively; is the ability to activate transcription altered? Recent work provides the first answers. The rel Gene Family Includes a Morphogen and Transcription Factor The oncogene v-rel was identified as the transforming function of the avian (turkey) reticuloendotheliosis virus. The retrovirus is highly oncogenic in chickens, where it causes B cell lymphomas. Figure 1 summarizes the relationship between the transforming and cellular genes; v-rel is a truncated version of c-rel, lacking the ~100 C-terminal amino acids, and has a small number of point mutations in the remaining sequence (Wilhelmsen et al., 1984). (Of course, v-rel and the other v-one genes discussed here are synthesized as part of retroviral polyproteins in which viral sequences-usually derived from gag and envgenes-flank the transforming sequence; the figures here compare only the v-one region with the c-one gene. It is probably usually the case that the flanking retroviral sequences do not directly contribute to oncogenicity, but they may not always be irrelevant.) The fel gene turns out to be part of a family with interesting members whose properties have profound implications for the regulation and function of c-rel and we/. Two classes of cellular genes that belong to this family are summarized in Figure 1: 9 The dorsal gene of Drosophila melanogaster is involved in establishing the dorsal-ventral axis. Coding for a protein of 678 amino acids, the region between amino acids 46 and 295 is closely related to c-n?/, with -80% similarity allowing for conservative substitutions (Steward, 1987). l
The transcription factor NF-KB was originally discovered as a regulator of transcription via an enhancer associated with the immunoglobulin K genes; the KB target site has since been found in many other enhancers and promoters. Excepting the N-terminal and C-terminal regions, the DNA-binding subunit of NF-KB is ~60% similar to c-re/(Kieran et al., 1990; Ghosh et al., 1990).
Cell 304
IKB NFKB 65K 50K Activation 1
I
:::.:.. .......~. ,:,:.:.:.:. ,.:.:.. .... .:.:.:.:. ..:.:.:.. +
8
V-Rel :~ii:2$3it!riii 80% .~.A..1..~11,.1..>..1.. ..:........
si m i lo rityi$;$z-zii, ... . . . .-....:./ :.
6761
dorsalr
DNA-binding
& dimerizotion
IKB association?
Activation Cytoplasm
Figure 1. v-rel Is a Truncated Form of c-ml That Lacks Region and Has ~4% Point Substitutions The Rel proteins are related the morphogen dorsal.
to the transcription
factor
the C-Terminal NF-KB and to
The immediate implication of these homologies is that the members of this family function as regulators of transcription. This is consistent with knowledge about dorsal, which is a maternal effect gene that probably regulates embryonic dorsal-ventral polarity directly. The homology with NF-KB suggests that it does so by controlling the response of a set of target genes whose products are involved in creating this axis. Indeed, dorsal protein binds to the promoters of target genes (see below). These comparisons suggest that the basis for v-refs properties may lie in a distortion of the properties of c-rel, which may be one of a number of proteins that either constitute NF-KB activity or provide similar activities. Dorsal and NF-KB share the unusual property that their movement from cytoplasm to nucleus is a regulatory event, Features in the v-Rel and c-Rel proteins that are responsible for their localization in either nucleus or cytoplasm have been identified, but any relationship to the oncogenic properties of v-Rel is unclear at present. Nuclear localization is crucial to dorsal function, since dorsal protein establishes the dorsal-ventral axis by dint of forming a nucleocytoplasmic gradient (Rushlow et al., 1989; Steward, 1989; Roth et al., 1989). Proceeding from the dorsal to the ventral side of the embryo, an increasing amount of the dorsal protein is found in the nucleus as opposed to the cytoplasm. Since the total amount of dorsal protein is initially more or less constant across the embryo, it appears to be the nuclear gradient that is responsible for dorsoventral morphogenesis. The region of homology between dorsal and rel includes a sequence that may be required for nuclear localization, but the sequence that regulates dorsal protein movement may be different; removal of the C-terminus of dorsal causes the protein to enter the nucleus in tissue culture cells, which suggests that this (unique) sequence may function as a cytoplasmic retention signal. This may be a common feature of Rel proteins, since deletion of C-terminal sequences from chick c-Rel allows the protein to enter the nucleus in fibroblasts.
Figure 2. NF-KB Contains Subunits of 50 kd and 65 kd; the 65 kd Subunit Is Bound by the Inhibitor IKB Activation causes phosphorylation of IICE, which then dissociates from NF-KB. Free NF-KB migrates to the nucleus, where it binds to KB sites at promoters or enhancers of target genes. Note that NF-KB is shown for simplicity as a dimer, but may function as a tetramer.
The transcription factor NF-KB is also regulated by nucleocytoplasmic transport. NF-KB is an inducible factor, but the induction does not require synthesis of protein. In uninduced cells, NF-KB is present in the cytosol, where it is bound to an inhibitor protein, IKB. Figure 2 shows that upon induction by any one of a variety of agents (including certain cytokines), IKB becomes phosphorylated and as a result releases NF-KB, which is then transported to the nucleus (Baeuerle and Baltimore, 1988; Ghosh and Baltimore, 1990). NF-KS activity resides in a heteromeric protein that contains two types of subunit, ~50 and ~65. The relationship of p50 to Rel has been established by sequence (as summarized in Figure 1); ~65, which has yet to be fully sequenced, may also be related to Rel. DNAbinding activity is certainly possessed by ~50; the DNAbinding activity of ~65 is not quite clear. IKB binds to ~65, and thereby holds the intact (heteromeric) factor in the cytoplasm. It remains to be seen whether the mechanism for controlling nucleocytoplasmic localization of dorsal and NFKB is the same. The product of the Drosophila gene cactus may be involved in controlling localization of dorsal, and one of the most interesting questions of the moment is therefore whether cactus will be related to the gene coding for IKB. The C-terminal sequences that are required for cytoplasmic retention of both dorsal and c-Rel are deleted from v-Rel. Since v-Rel is located in the nucleus in chick embryo fibroblasts, it was originally thought that differences in localization might explain the oncogenicity of v-Rel compared with c-Rel, but later it appeared that this is not the case. Addition of the C-terminal sequences of c-Rel to v-Rel results in its retention in the cytoplasm, but it remains transforming (Gilmore and Temin, 1988; Hannink and Temin, 1989). Furthermore, v-Rel is largely cytoplasmic in spleen cells, suggesting that there may be more than one mechanism for determining the localiza-
Review: 305
Oncogenic
Transcription
Factors
tion of Rel proteins. The ability of v-Rel to transform in either cytoplasmic or nuclear state remains a puzzle, but raises the possibility that the oncoprotein has more than one mode of action. What regions of rel are required for transformation? Analysis remains at a primitive stage, but generally it seems that the N-terminal half of the v-Rel protein is important. This region is required for DNA-binding and for dimerization, but its role in activating transcription is unclear. Experiments in which Rel proteins are fused to yeast DNA-binding sequences and then tested for transcriptional activation agree that the C-terminal region of c-Rel (which is missing from v-Rel) is a potent activator; but there is disagreement on whether the N-terminal region is itself an additional activator or instead represses the C-terminal activator (Kamens et al., 1990; Bull et al., 1990). At least four proteins related immunologically to v-Rel are found in the cytoplasm of resting T cells, but enter the nucleus following activation. All of these proteins bind to DNA, and have been described by the size of adducts cross-linked to DNA as ~50, ~55, ~75, and p85 (Ballard et al., 1990). ~50 is a product of ~55, which corresponds to the small (50 kd) subunit of NF-ICB. In addition to its ability to form an oligomer with the 65 kd (non-DNA-binding) subunit of NF-KB, ~55 (NF-KB 50 kd) can form an oligomer with ~75 that binds to DNA. The largest of the proteins, ~85, appears to correspond to c-Rel. It is possible therefore that NF-KB is one of several transcription factors that can be formed by combinatorial associations between various subunits, and the functional consequences of this situation remain to be explored. Other open questions concern the nature of differences that may exist between individual members of the family, and to what degree the pleiotropic effects attributed to NFKB may reside in different family members as opposed to a single transcription factor. jun and fos Gene Products Contribute to AP-1 and Activate Transcription The transcription factor AP-1 recognizes a short sequence (TGACTCA) originally identified in the enhancer of SV40, but since found in promoters and enhancers of cellular genes. It is the nuclear factor required to mediate transcription induced by phorbol ester tumor promoters (such as TPA), and an AP-1 binding site confers TPA inducibility upon a target gene (Angel et al., 1987; Lee et al., 1987). Preparations of the factor include a variety of polypeptides, amongst which are the products of the genes c-jun and &OS. Thev-jun gene is carried by avian sarcomavirus 17 (ASV-17) which induces fibrosarcomas in chickens, and can transform chick embryo fibroblasts. The v-fos gene provides the transforming activity of FBJ murine osteosarcoma virus (FBJ-MuSV), which induces chondrosarcomas. (v-fos is also found in a chicken sarcoma virus.) Some original confusion concerning the active components of AP-1 preparations is explained by the fact that both c-Jun and c-Fos proteins are present and contribute to the activity. The subunits can form c-Jun homodimers as well as c-Jun-c-Fos heterodimers. Jun was identified as a component of AP-1 by the
demonstration that antibodies to v-Jun cross-react with AP-1; this was followed by cloning of the c-jun gene and the demonstration that its product is present in AP-1 preparations (Bohmann et al., 1987). Figure 3 shows that v-Jun is derived from c-Jun by the deletion of a sequence of 27 amino acids called the 5 region as well as by individual point mutations in the remaining region. In addition, the v-Jun mRNA lacks an extensive 3’ nontranslated region that is present in c-Jun transcripts. c-Jun can be converted into a transforming species by introducing some of the features characteristic of v-Jun (Bos et al., 1990). Removal of the 6 region increases the ability to transform chick embryo fibroblasts by an order of magnitude; deletion of the 3’ nontranslated region also increases the number of foci (possibly because this region contains a sequence that destabilizes the mRNA and therefore reduces the quantity of protein product). The point mutations do not seem to be involved in oncogenic capacity. Jun proteins have three regions toward the C-terminus that are involved in the ability to activate transcription. A leucine zipper provides a dimerization motif; since dimer formation is essential for binding to DNA, loss of this function leads indirectly to inability to bind DNA. An adjacent basic region (located in effect at the foot of the zipper) is required directly for DNA binding, and close by is a region (A2) rich in proline residues that is required to activate transcription. These three regions are sufficient for activity in HeLa cells. Both c-Jun and v-Jun proteins rely upon their zippers to form dimers that activate transcription at AP-1 target sites (Bohmann and Tjian, 1989). Although c-jun is the closest homolog to v-jun in human genomic DNA, the iun gene family contains other members, jun6 and ]unD, which possess indistinguishable abilities to form homodimers amongst themselves, heterodimers in any pairwise combination, and heterodimers with Fos; all these forms can bind to DNA at the AP-1 site (Nakabeppu et al., 1988). However, homodimers formed with JunB do not activate transcription at promoters that contain individual AP-1 sites, although they can do so when multimeric response elements are present (Chiu et al., 1989). Correspondingly, in cotransfection experiments, JunB antagonizes the transforming capacity of c-Jun (which has transforming abilities when overexpressed, although it is much less ef0
40
67
340
c-Jun
9 ..,,Ji
3
substitutions
‘.j v-Jun c Al
> Activation
A2 Activation (Pro rich)
000 J
DNA-binding (Basic) Dimerizotion (Leu zipper)
Figure 3. v-jun Is Related to c-jun by Deletion of the 27 Amino Acid S Region and by a Very Small Number (
