| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Cancer Research | Clinical Cancer Research |
| Cancer Epidemiology Biomarkers & Prevention | Molecular Cancer Therapeutics |
| Molecular Cancer Research | Cell Growth & Differentiation |
Department of Biochemistry, Case Western Reserve University, Cleveland Ohio 44106 [R. N., G. Z., E. S.]; Department of Molecular Genetics, Biochemistry and Microbiology, University of Cincinnati College of Medicine, Cincinnati, Ohio 45267 [R. N.]; National Cancer Institute-Frederick Cancer Research and Development Center, Frederick, Maryland 21702 [P. S.]; and Department of Animal Science, University of Minnesota, St Paul, Minnesota 55108 [D. N. F.]
 |
Abstract |
|---|
 TOP Abstract IntroductionResultsDiscussionMaterials and MethodsReferences |
|---|
|
Introduction |
|---|
|
TOPAbstractIntroductionResultsDiscussionMaterials and MethodsReferences |
|---|
The truncated v-Ski protein is missing a high-affinity dimerization domain that consists of a series of five tandem repeats and a leucine zipper (11, 12, 13) . This domain mediates Ski:Ski homodimerization and Ski heterodimerization with SnoN (14) . This domain is not required for transformation, but the more potent transforming activity of c-Ski relative to v-Ski suggests that dimerization plays a role in Ski-induced transformation (7) . v-Ski contains a lower-affinity dimerization domain that is sufficient for transformation if the protein is expressed at high levels (14) .
A mutational analysis has demonstrated that the NH2-terminal 304-amino acid v-Ski are both necessary and sufficient for Ski induced transformation (15)
. Limited protease digestion of Ski proteins has shown that this domain is compact and globular, which suggests that it forms a single functional and structural unit (15)
. This region contains potentially important motifs typical of transcription factors, including a proline-rich region and several predicted 
helical segments including a possible helix-loop-helix. In addition, there are several groupings of histidine and cysteine residues that are highly conserved between Ski family members, and could form zinc finger-like structures (6)
. The contribution of these structural motifs to Ski’s transforming and myogenic functions has been assessed by analyzing a series of deletion and insertion mutations in the transforming domain of v-Ski (15)
. This analysis shows that most deletions within this domain resulted in transformation-defective Ski proteins, although small deletions within the first 78 amino acids of the protein, which includes the proline-rich domain, did not have an effect. The region of the protein between amino acids 128 and 245 was least tolerant of any insertion or deletion mutations.
Ski activates transcription from a variety of transcriptional regulatory elements, including the myosin light chain 1/3 (MLC1/3) and the muscle creatine kinase enhancers, and a number of viral enhancer elements (1 , 2) . Experimental evidence suggests that this transcriptional activation is mediated by protein:protein interaction rather than direct DNA binding. For example, stimulation of the MLC1/3 promoter by Ski has been shown to be dependent on the muscle specific helix-loop-helix protein, MyoD (1) . In addition, Ski specifically interacts with nuclear factor I (NFI) family proteins and increases their transcriptional activation of promoters with upstream NFI binding sites (3) . Despite these demonstrations of coactivation of transcription with a number of different transcription factors, Ski does not seem to contain an independently functioning transcription activation domain.7
We have shown that c-Ski and v-Ski can bind to the consensus element, GTCTAGAC, in association with other, as yet unidentified, proteins (16) . Binding is cooperative and apparently requires dimerization of Ski proteins because c-Ski binds this sequence much more efficiently than v-Ski (16) . This binding sequence is unique to Ski and Sno proteins. Moreover, unlike activation, this repression activity is an intrinsic property of Ski and SnoN proteins and is not limited to interaction with this binding site. These proteins can repress transcription when fused to a heterologous DBD (16 , 17) . In addition, Dahl et al. (4) have shown that Ski can also repress transcription of a retinoic acid-responsive reporter through interaction with the retinoic acid receptor.
The present studies were undertaken with two purposes: (a) to map the regions of the Ski protein that are required for GTCT binding and for transcriptional repression; and (b) to determine whether transforming activity is dependent on the ability to bind to the GTCT binding site and/or to repress transcription. To accomplish these goals, we have analyzed a set of Ski mutants for the ability to bind the GTCTAGAC binding site, repress transcription, induce transformation, and promote myogenesis. Because binding of the GTCT element by v-Ski is very weak, we thought that comparisons of the effects of different mutations on binding would be difficult and easily compromised by small differences in protein expression. Binding by c-Ski is quite high affinity and much less subject to differences in protein expression as evidenced by the fact that the endogenous protein is positive for binding in mobility shift assays, although it is expressed at very low levels (16) . For this reason, we first transferred mutations that had previously been analyzed in the context of v-Ski into c-Ski. Because the addition of c-Ski’s high affinity dimerization domain might alter the biological activities of some mutants, we assayed the full-length forms for transformation and muscle differentiation.
Our results show that the addition of the c-Ski dimerization domain partially complements the defects in transformation or myogenesis of some v-Ski mutants. We found that mutants that are defective in transformation are consistently attenuated in their ability to bind to the GTCT element and to repress transcription. We have mapped a potent transcriptional repression domain within the NH2 terminal quarter of the Ski protein that overlaps the region required for binding to the GTCT element. Both the GTCT binding and the repression activities map within the minimum transforming domain of the Ski protein. This study confirms the importance of dimerization in Ski function, identifies the DNA binding and repression domains within the Ski protein, and demonstrates the likely relevance of these activities to the process of Ski-induced transformation and myogenesis.
|
Results |
|---|
|
TOPAbstractIntroductionResultsDiscussionMaterials and MethodsReferences |
|---|
, and examples of morphology and muscle differentiation are shown in Fig. 2
. Soft agar cloning efficiencies for v-Ski and c-Ski are the same, except that c-Ski-transformed CEFs produce larger, more rapidly growing colonies (7)
.
|
|
1 and v-Ski3) were found to be completely defective in all of the biological assays (15)
. However, as mutations in c-Ski, both c-Ski1 and c-Ski3 have partial activity. c-Ski1 generates colonies in soft agar, albeit at reduced efficiency (50% relative to wild-type c-Ski) and induces partial myogenesis. c-Ski3 has no transforming activity but does induce partial myogenesis. c-Ski2 could not be assessed because the protein is not expressed at detectable levels. These results confirm those of the previous study (15)
by showing that the residues deleted in c-Ski1 (17–120) and c-Ski3 (230–324) are important for full biological activity of Ski. It is evident from the increased activity of these two c-Ski mutants relative to that of their v-Ski forms that the effect of these mutations can be partially suppressed by the addition of the c-Ski dimerization domain.
We next analyzed a series of five smaller deletions that were generated to provide a finer map of important functional elements. The results of biological assays with these c-Ski mutants are shown in Fig. 1
. Four of these mutations have the same effect on the activity of c-Ski as they did with v-Ski. Two deletions near Ski’s NH2 terminus—AH1, which removes a predicted -helical segment, and Pro, which removes a 23 amino acid proline-rich region—have no effect on transforming and myogenic activity. The morphology of CEFs transformed by Pro (Fig. 2C)
is identical to that of AH1 and indistinguishable from that induced by c-Ski. The other two deletions, AH4 and Z3/4, are within the COOH-terminal one-third of the transformation domain, and result in c-Ski proteins that are completely transformation- and myogenesis-defective (Fig. 1
and Fig. 2D)
. AH4 removes part of a predicted helix, and Z3/4 removes a segment containing a possible zinc finger motif (15)
. The fifth mutation, AH2, deletes a predicted -helical segment. In the context of v-Ski, the AH2 deletion severely impairs morphological transformation and soft agar cloning (32% relative to wild-type v-Ski), and gives a slightly reduced level of myotube formation (15)
. The c-Ski form of AH2 is much less impaired, giving wild-type soft agar cloning and muscle differentiation but significantly less dramatic morphological transformation (Fig. 2B)
. Thus, AH2 provides another example of a mutation that is partially suppressed by the addition of the c-Ski dimerization domain.
Three new c-Ski mutants were analyzed for biological activity (Fig. 1)
. These mutants were produced by the substitution of serine for three highly conserved cysteine residues (136, 209, and 228) and are referred to as MT1, MT2 and MT3, respectively. The MT1 and MT2 mutants are both wild-type for morphology, soft agar cloning, and myogenesis (Fig. 1
and Fig. 2E)
. On the other hand, MT3 is completely defective for both transformation and myogenesis. The MT3 substitution is overlapped by the Z3/4 deletion and the defective nature of both of these mutants suggests that this grouping of cysteines and histidines is functionally important. The results with these cysteine substitution mutants together with the small deletion mutants described above indicate that the most important determinants of Ski’s transforming and myogenic activities lie in the COOH-terminal half of the transformation domain.
The relative level of expression of all of the mutant c-Ski proteins was determined by Western blotting nuclear extracts of infected CEFs (Fig. 3, A and B)
. The anti-Ski monoclonal antibodies, G8 and M6, were used to probe the blot in Fig. 3A
. The c-Ski1 and c-SkiPro proteins are missing the epitopes for these monoclonal antibodies, but these proteins are detected (Fig. 3B)
with the polyclonal antibody 32360 that reacts with the COOH-terminal region of the Ski protein (19)
. The v-Ski, MT3, and c-Ski1 proteins are expressed at reduced levels relative to other mutant proteins (Fig. 3A
, Lanes 2 and 6, and Fig. 3B
, Lane 4). Previous results (7)
showed that c-Ski is fully active when expressed at about 20% of the level obtained with the RCASBP virus used in these studies. Thus the low levels of expression of the c-Ski1 and MT3 proteins is unlikely to account for their reduced level of activity. Although the v-Ski protein seems to be expressed at low levels in this particular extract, levels more comparable to that of c-Ski are observed with other extracts (data not shown).
|
. The probe contains two tandem copies of the GTCTAGAC element (GTCT/2) to take advantage of the strong cooperative binding of Ski complexes to adjacent copies of the GTCT site (16)
. As shown previously (16)
, c-Ski forms two complexes with the GTCT/2 probe that are supershifted by the anti-Ski monoclonal antibody, M6 (Fig. 4A
, Lanes 1 and 2). The faster migrating complex (complex 1) results from the binding by a Ski-containing multiprotein complex to only one of the two 8-bp GTCT elements, whereas, the slower migrating complex (complex 2) results from binding to both copies (16)
. Endogenous Ski forms a single complex that is supershifted by M6 and migrates slightly slower than complex 1, which is formed by exogenous Ski (Fig. 3A
, Lanes 19 and 20).
|
, Lanes 3–6). The transformation defective cysteine substitution mutant, MT3, does not bind to the GTCT/2 probe at all (Fig. 4A
, Lanes 7 and 8). Similarly, the fully transformation defective mutants c-SkiAH4, c-SkiZ3/4 and c-SkiCT (lacks the entire transformation domain) are negative for DNA binding (Fig. 4, A
, Lanes 15–18, and B
, Lanes 11 and 12). Some mutations that only slightly attenuate (c-SkiAH2) or have no effect (c-SkiAH1, c-SkiPro) on c-Ski’s transforming activity result in binding activities that are somewhat less than that of c-Ski (Fig. 4A
, Lanes 13 and 14, and B
, Lanes 3–8). The correlation between DNA binding and transformation may extend to the nature of the complexes formed with the GTCT element in addition to the amount of binding. For example, the c-SkiAH2 mutant, which is partially defective for transformation, shows not only decreased binding to the GTCT/2 probe but also favors formation of complex 1 over complex 2, in contrast to wild-type c-Ski (Fig. 4, A
, Lanes 13 and 14, and B
, Lanes 7 and 8). Because the formation of complex 2 involves cooperative interactions, this result suggests that the AH2 mutation impairs the domain in c-Ski that participates in these interactions. Other mutants that are wild-type for transforming and myogenic activity, c-SkiAH1, c-SkiPro, and c-Ski4, show almost exclusively complex 2 formation. All of the complexes formed by the mutant Ski proteins can be supershifted by the addition of the anti-Ski mAb, M6, with the exception of the complex formed by c-SkiPro. As mentioned above, the Pro deletion removes the epitopes for the G8 and M6 monoclonal antibodies; therefore, the failure of a supershift in this case serves as a specificity control for the other supershifts and confirms the identity of this mutant (Fig. 3B
, Lanes 5 and 6).
Observed failures to bind DNA cannot be attributed to differences in mutant protein expression. With the exception of c-Ski1 and MT3, all other forms, whether positive or negative for GTCT binding, are expressed at roughly equivalent levels (Fig. 3)
. Moreover, complex formation is detected with endogenous c-Ski (Fig. 4A
, Lanes 19 and 20), although it is expressed at such low levels that it is not visible at the exposure of the Western blot shown in Fig. 3
(Lanes 2, 11, and 5). This result suggests that, even for c-Ski1 and MT3, reduced expression is unlikely to account for the complete lack of binding to the GTCT element. This being the case, the two mutants with large deletions in the transforming domain, c-Ski1 and c-Ski3, are the only exceptions to the correlation between biological activity and DNA binding. Although partially active in transformation and/or myogenesis, neither of these proteins exhibits detectable binding to the GTCT binding site in this assay (Fig. 4A
, Lanes 9–12). The single faint complex observed with c-Ski3 migrates slightly slower than the exogenous complex 1 and is most likely the endogenous Ski complex. It is possible, in light of the weak binding observed for v-Ski (16)
, that these proteins bind to GTCT, but their interaction is too unstable to withstand EMSA conditions.
Transformation-defective c-Ski Mutants Are Defective in Transcriptional Repression.
We next sought to determine whether the mutants described above repress transcription as had been shown for both v-Ski and c-Ski (16)
. The mutants were cloned into the transient expression vector RSVPL, and cotransfected into UMNSAHDF1#1 CEFs along with an equal amount of the GTCT2X2tkCAT reporter (16)
. As shown in Fig. 5
, none of the Ski proteins tested is completely defective in transcriptional repression, including several forms (MT3, MT1/3, c-SkiAH4, c-SkiZ3/4, c-Ski3, and c-SkiCT) that are negative for binding to the GTCT/2 element by EMSA (Fig. 4, A
and B). However, these forms exhibit only 4–12% of the activity of c-Ski, and previous work had shown that c-Ski represses transcription by about this amount from the same reporter but lacking the upstream GTCT element (16)
. It is possible that the residual repression activity of these mutant forms is due to this binding-site-independent mechanism that does not correlate with Ski’s transforming activity.
|
, Fig. 4
, Lanes 3–6, and Fig. 1)
. Group II, with 20–50% of c-Ski’s repression activity, includes all of the other forms that are positive for binding to GTCT, although some do so less efficiently than c-Ski. This group includes v-Ski and c-SkiAH2, which are slightly impaired in transforming ability and bind the GTCT element about 5–10 fold less efficiently than c-Ski (Fig. 5
, Fig. 4A
, Lanes 13 and 14, and Fig. 1
; Ref. 16
). The other mutants in this intermediate group, c-SkiAH1, c-SkiPro, and c-Ski4, exhibit wild-type transforming activity but vary in DNA binding and repression activities. c-SkiAH1 shows a 3–5 fold reduction in GTCT binding, whereas, c-Ski4 shows wild-type binding activity; yet these mutants repress expression from GTCT2X2tkCAT by similar amounts. c-SkiPro, and c-Ski4 both show wild-type DNA binding but differ by a factor of two in repression activity. c-Ski1 is the only mutant that does not fall into any of the above categories because, despite its partial transforming and myogenic activity, it exhibits no detectable DNA binding activity. However, this mutant represses GTCT2X2tkCAT by approximately 17% of wild-type, which is greater than the fully defective mutants and is in line with its partial biological activity. Clearly, transformation, GTCT binding, and repression—as measured in these studies—cannot be described by a simple linear relationship. However, it is likely that the minor inconsistencies reflect differences in the sensitivities of the assays used. The general trend of these results suggests that GTCT-dependent repression plays a role in Ski-induced transformation.
A Repression Domain Is Contained within the Minimum Transforming Domain of Ski.
We had previously found that Ski contains a repression domain that can function independently of GTCT binding when fused to the Gal4 DBD (16)
. This approach affords us the opportunity to map the repression domain with respect to Ski’s transformation domain. To accomplish this, we have fused various segments of Ski to the Gal4 DBD and assayed the fusion proteins for the ability to repress the G5tkLuc reporter.
As shown previously, full-length Ski fused to the Gal4 DBD decreases expression from G5tkLuc to 6% of the control value (Fig. 6A)
. The NH2 terminal 452 amino acids, which are roughly equivalent to v-Ski, yields a similar level of repression of G5tkLuc. An additional COOH-terminal truncation in Gal-Ski (1–325) results in enhanced repression activity, reducing expression to 1% of the control. The COOH-terminal sequence on its own, present in Gal-Ski (510–750) and Gal-Ski (325–750), gives minimal repression of the G5tkLuc reporter (33–50% of the control value). For this reason, we restricted further mapping of the repression domain to the NH2 terminal half of the Ski protein.
|
). An NH2-terminal deletion of residues 1–76 in (Gal4-Ski77–214) reduces repression activity by a factor of 14 compared to that of Gal4-Ski1–214. Deletion of either the NH2-terminal half (residues1–45) or the COOH-terminal proline-rich half (residues 53–76) of this 76-residue segment restores one-half of the lost repression activity. Therefore, the entire first 214 amino acids of Ski contribute to the repression activity and constitute the optimal repression domain, although a subdomain within this region from residues 77–189 seems to constitute an active core of this domain.
To insure that the Gal fusion proteins expressed from these constructs were the appropriate size, we analyzed the proteins by Western blotting lysates of transiently transfected CEFs with an anti-Gal4 monoclonal antibody. As previously observed with similar fusions of SnoN (17)
, we found that the expression of many of the fusion proteins is undetectable in CEFs. To surmount this problem, we repeated the Western analysis using lysates from transiently transfected Cos cells. As seen in Fig. 6B
, all of the proteins are of the predicted size, but there are large differences in the levels of expression among the different forms. If these differences reflect the expression levels in CEFs, they could distort our estimates of the relative repression activity of the proteins. In particular, Gal-Ski (1–425) and Gal-Ski (77–214) are expressed at much lower levels than the other proteins. If we normalize for the differences in expression, these fusion proteins would have only 2-fold less repression activity than Gal-Ski (1–214). This adjustment does not affect our conclusions but would strengthen the suggestion that the core repression domain is contained in the residue 77–189 segment.
We have previously mapped a repression domain to the NH2 terminal 254 amino acids of the chicken Ski homologue, SnoN (17)
. Because SnoN has a 75-amino acid extension at its NH2 terminus that is not conserved in Ski, the optimal Sno repression domain is approximately equivalent to Gal4-Ski (1–189; Fig. 6A
). To compare the repression activities of these two homologous domains directly, we have included the Gal4-Sno (1–254) in the same experiment with the Gal4-Ski fusions. We find that Gal4-Sno (1–254) represses transcription to about 2% of the control value, which is approximately the same as Gal4-Ski (1–189).
Separation of Repression and DNA Binding by Analysis of c-ski Mutants as Gal4 Fusion Proteins.
In the studies of the repression activity of the c-Ski mutants described above, we encountered several mutants whose repression activity could not be directly assessed with the GTCT reporter because the proteins fail to bind the GTCT element. Three of these (MT3, c-SkiH4, and c-SkiZ3/4) are of particular interest because the mutated region in each lies downstream of the repression domain mapped using Gal4 DBD fusions. We were interested in knowing whether these mutations would have an indirect effect on the repression domain, or whether they are defective only in GTCT binding. To answer this question, these c-Ski mutants were fused to the Gal4 DBD and tested for their ability to repress the G5tkLuc reporter (Fig. 7A)
. We find that these three mutants, freed of the requirement for DNA binding, repress transcription at least as well as wild-type c-Ski. A Western analysis shows that the expressed proteins are the predicted sizes of the full-length forms (Fig. 7B)
. Because these mutants are defective in DNA binding but wild-type in repression activity, these results locate the DBD downstream of the repression domain, which suggests that the two together constitute the minimum transforming domain of Ski. The Gal4-SkiZ3/4 fusion is almost three times more potent a repressor than the other proteins, which suggests that the segment of the DBD deleted in this mutant may interfere with the repression of Gal4 DBD fusion proteins. This idea is consistent with the results presented in Fig. 6
showing that Gal-Ski (1–214) is about three times more potent a repressor than Gal-Ski (1–325). The region that differentiates these two forms (residues 215–325) includes the region deleted in SkiZ3/4 (residues 229–245).
|
|
Discussion |
|---|
|
TOPAbstractIntroductionResultsDiscussionMaterials and MethodsReferences |
|---|
. Further deletion at either the NH2 or COOH terminus of this optimal Ski repression domain results in a substantial decrease in repression activity.
|
helix containing a critical glutamine (23)
. There is an alanine-rich potentially -helical segment near the NH2 terminus of Ski that contains a glutamine residue (residues 34–45), and this region is followed by a proline-rich/hydrophobic region (residues 53–76). Deletion of either one of these elements from c-Ski in mutants AH1 and Pro does not destroy repression activity but decreases it 2- to 4-fold, and the deletions do not effect DNA binding. A second glutamine-containing predicted -helical region is found within the repression domain at residues 137–150. Deletion of this segment from c-Ski (AH2) decreases repression by 3-fold, but some of this may be caused by a decrease in DNA binding. Finally, the COOH-terminal end of the repression domain contains a predicted helix that is highly charged and acidic (residues 190–204). Deletion of this region (AH4) destroys GTCT binding and consequently eliminates repression via this element by c-Ski. However, its deletion from the Gal4 fusion proteins has no effect on repression in the context of full-length Ski or causes only a 3-fold reduction relative to the Gal4Ski1–214 protein. From these results, we conclude that the AH1, Pro, and AH2 domains may contribute to repression activity, but they do not seem to be essential. The role of AH4 is less clear but seems to be more important for DNA binding than for repression per se.
Zinc fingers have been shown to function in protein:protein interaction and DNA binding, but it is not clear what role these structures play in repression domains (24, 25, 26, 27, 28)
. Although Ski has not been shown to contain any zinc-binding domains, the presence of a number of highly conserved cysteines and histidines in the Ski/Sno transformation domain (Fig. 8)
has suggested that some of these residues may participate in zinc finger-like structures (6)
. The MT1 and MT2 mutations, which substitute serines for cysteines 136 and 209, respectively, have no effect on repression activity. An additional mutant with two substitutions involving cysteine and histidine residues (H130Y and C133S) was shown to be wild-type for transformation and myogenesis (15)
. Therefore, if a metal binding domain does form in this region of the protein, it is unlikely to be required for repression or transformation.
Many transferable repression domains correspond to relatively short sequences that function as protein:protein interaction domains. For example, a 35-amino acid mSin3 Interaction Domain (SID) in Mad family proteins interacts with the transcriptional corepressor mSin3; and a 4-amino acid WRPW motif, found in hairy-related repressor proteins, interacts with the transcriptional corepressor, Groucho (29, 30, 31, 32) . The Ski repression domain does not contain relatives of either of these motifs. Transcriptional repressors of the steroid/thyroid hormone receptor superfamily contain extended repression domains composed of multiple regions that interact with the basal transcription machinery as well as with the corepressors N-CoR and SMRT (33, 34, 35, 36, 37, 38) . The region that we have defined as Ski’s optimal repression domain is quite large (214 amino acids), and small deletions within this region reduce but do not eliminate repression activity. It may be that most of Ski’s repression activity is mediated by a short sequence that we have not identified yet, or, like the hormone receptors, Ski’s repression domain may actually contain multiple repression modules that are responsible for different protein:protein interactions. In the latter case, elimination of one such interaction may not be sufficient to abrogate repression activity. In support of this notion, Ski has been found to bind the TFIID subunit TAFII110 (data not shown) whose binding site in SnoN is within the NH2 terminal half of the related repression domain (17) .
Location of Ski’s GTCT Binding Domain.
The AH4 (190–204), Z3/4 (229–245), 3 (230–324), and MT3(C228S) mutations all eliminate binding to the GTCT element and are located within the transformation domain but COOH terminal to the core repression domain (Fig. 8)
. These proteins retain repression activity when fused to Gal4, which makes it unlikely that the mutations cause an overall disruption of protein structure and which suggests that this region is crucial for GTCT binding by Ski. The AH4 deletion removes a portion of a potential amphipathic -helix, whereas the remaining three mutations all affect a grouping of conserved histidines and cysteines. Of the three cysteine substitution mutations included in this study, MT3 is the only one that impairs Ski DNA binding. This substitution involves one of the cysteines of a possible cys2-his2 zinc finger that is deleted in the Z3/4 mutation. The fact that both mutations eliminate GTCT binding provides support for the functional importance of this element in DNA binding by Ski. The 3 deletion removes a basic region located at the end of the transformation domain. However, because the 3 deletion overlaps with the Z3/4 region, we were not able to determine whether the basic region by itself plays a role in GTCT binding.
Ski binds to the GTCT site as part of a complex with several unidentified cellular proteins (16) . As part of this complex, Ski can be UV cross-linked to the GTCT binding site, which suggests that a region of the Ski protein is in very close contact with the DNA. However, because Ski is unable to bind the GTCT element on its own, we do not know how much this direct contact contributes to Ski’s interaction with this binding site. Therefore, it is likely that the GTCT binding region that we have defined functions as both a protein:protein interaction domain and a direct DBD. Ski may bind DNA like the viral transactivator VP16 whose interaction with the TAATGARAT motif is dependent on interactions with the cellular proteins Oct-1 and HCF as well as with the binding site itself (39) . Now that we have identified a region within the Ski that is critical for binding to the GTCT site, isolation of proteins that interact with this region of the Ski protein should lead to the identification of Ski-GTCT cobinding proteins. This will then allow us to distinguish between protein:protein and protein:DNA interactions that are involved in Ski’s binding to the GTCT element.
Using a tandemly repeated binding site probe (GTCT/2) in EMSAs, we identify two Ski-containing complexes. The slower migrating complex (Ski2) is produced by cooperative binding to both copies of the GTCTAGAC sequence and the faster migrating complex (Ski1) results from the binding to only one copy of this element (16)
. In the present studies, the c-SkiAH1, c-SkiPro, and c-Ski4 mutants all form exclusively the Ski2 complex, which suggests that binding of these mutant proteins to the binding site is highly dependent on cooperative interactions. On the other hand, c-SkiAH2 forms predominantly the Ski1 complex. This result indicates a reduced level of cooperativity between c-SkiAH2 complexes bound to adjacent GTCT elements and suggests that a region defined by this deletion plays a role in cooperative binding of Ski to repeated GTCT elements.
Dimerization Is Important for Transformation and DNA Binding.
Although the first 304 amino acids of Ski have been shown to be sufficient for transformation, this domain by itself shows reduced morphological transformation and soft agar growth relative to v-Ski. The addition of a nuclear localization signal to this domain results in transforming ability identical to that of v-Ski but still less than that of c-Ski (7
, 15)
. We have suggested that the remaining difference is due to the absence in v-Ski of c-Ski’s high affinity dimerization domain (14
, 15)
. A similar conclusion is reached by comparing the influence of identical mutations in v-ski and c-ski on their transforming activity. None of the mutants is less active in the context of c-Ski and three (1, 3, and AH2) show enhanced transforming activity compared to their v-Ski forms. The results suggest that the added dimerization domain partially complements defective Ski:Ski interactions caused by these mutations.
The activity of the c-Ski4 protein provides a compelling argument for the importance of dimerization in the DNA binding and transforming activities of Ski. c-Ski4 consists essentially of the transformation domain (residues 1–304) and the neighboring nuclear localization signal (residues 305–325) fused to the COOH-terminal dimerization domain of c-Ski (residues 431–750). c-Ski4 transforms CEFs and binds the GTCT element with an efficiency equivalent to c-Ski. This is not caused by the ability of the dimerization domain to perform these functions independently. The c-SkiCT protein (residues 326–750) consists of the entire c-Ski region downstream of the transforming domain, and it does not bind the GTCT element and has no transforming activity. These results suggest that Ski dimerization is the only function supplied by the COOH-terminal segment and demonstrate the importance of dimerization for both DNA binding and transformation. This conclusion is consistent with the results of a recent study (26)
that showed that the dimerization domain is essential for transcriptional activation of the myogenin promoter/enhancer by Ski.
Relationship between DNA Binding, Repression, and Transformation.
The domains responsible for the GTCT binding and repression activities of Ski are located within the minimal domain that is required for the induction of transformation and myogenesis (Fig. 8)
. Furthermore, all of the mutations in Ski that drastically reduce its transforming and myogenic activity also eliminate its ability to bind to the GTCT binding site. Mutations that have little or no effect on DNA binding also leave Ski’s biological activities intact. The correlation between repression activity and transforming/myogenic activity is not perfect but is quite convincing. All five of the mutations that completely eliminate Ski’s transforming activity reduce its repression activity to 10% or less of the wild-type level. However, four mutations that have little effect on transformation and myogenesis by Ski diminish its repression activity to about 25% of wild-type. This class includes c-Ski4, which has intact transforming and repression domains suggesting that this level of repression is above a critical threshold that is required for transformation by Ski. These results strongly suggest that repression of transcription by binding to the GTCT element underlies Ski’s ability to induce cellular transformation and muscle differentiation in avian fibroblasts.
Several oncogenes have been shown to function as transcriptional repressors. For Qin, v-erbA, and EVI-1 their repression domains are required for oncogenic activity (40, 41, 42) . In the case of Qin, the use of chimeric proteins has demonstrated that transformation results from transcriptional repression of genes targeted by Qin’s DBD (43) . The cellular mechanisms of transformation by transcriptional repressors are likely to be as varied as those of activators. The Gfi-1 oncogene is a transcriptional repressor that renders T cells growth factor-independent by repressing expression of the pro-apoptotic gene bax (25 , 44 , 45) . An oncogenic mechanism used by v-erbA involves blocking differentiation of erythroid progenitors by down-regulating genes that are necessary for progression toward a terminally differentiated state (46, 47) . ski has also been shown to participate in transformation of erythroid progenitors, and its action seems to involve repression of hormone-regulated genes (4 , 48) .
Ski can either activate or repress transcription depending on promoter and cellular context, so the relationship between its transcriptional and biological properties is likely to be complex. In this way, it may be analogous to myc, which provides a precedent for repression-linked transforming activity by an oncogene known to function as a transcriptional activator (49) . As in our previous study of v-Ski, the mutants analyzed in this study show cosegregation between the biological activities of transformation and muscle differentiation (15) . This is a surprising result given the opposite nature of these two processes. This could result from Ski’s ability to repress transcription of a gene under one set of conditions and activate it under another. Support for this hypothesis is provided by studies involving expression of muscle-specific reporters that show a switch from repression to activation by Ski, depending on cellular context (1 , 50) . Ski seems to contribute to neural and muscular development in the mouse by regulating both cellular growth and differentiation (10) . It is possible that a transition between repression and activation of gene expression by Ski (or vice versa) may help to regulate the delicate balance between these processes. Because overexpression of c-Ski is the only requirement for inducing aberrant growth and the resulting transformation (7) , it may be that the level of expression determines whether activation or repression is the dominant activity. In the present study, we have shown a strong correlation between repression by Ski through the GTCT binding site and the biological activities of transformation and myogenesis; additional investigation will be required to determine what role activation plays in these processes.
|
Materials and Methods |
|---|
|
TOPAbstractIntroductionResultsDiscussionMaterials and MethodsReferences |
|---|
EMSA.
The radiolabeled GTCT/2 probe (5 x 107 cpm/pmol) was prepared by PCR amplification in the presence of [32P]dATP (3000 Ci/mmol) according to the method of Mertz and Rashtchian (52)
as described previously (16)
. The sequence of the GTCT/2 probe, with primer sequences underlined, is as follows: GGCGGATCCACCTACACGTAGTCTAGACGTCTAGACAATGTGCACTGCAGTGGC.The synthesized probe was purified by electrophoresis on a 7% polyacryamide and elution of the GTCT/2 band by soaking overnight in 150 µl of gel shift buffer [25 mM HEPES (pH = 7.5), 100 mM NaCl, 0.2 mM EDTA, and 0.1% NP40]. About 2 x 105 cpm of probe was added to each 20-µl binding reaction containing 8 µg of nuclear extract, 500 ng of (poly)dIdC, and 500 ng of RsaI digested bovine DNA, in gel shift buffer with 10% glycerol, 2 mM DTT, 0.5 mM phenylmethanesulfonyl fluoride (PMSF), and 0.5 mM Pefabloc (Boehringer Mannheim). After a 20-min incubation at room temperature, reactions were analyzed by electrophoresis on a 4% polyacrylamide (60:1) gel as described previously (16)
. For antibody supershifts, the anti-Ski monoclonal antibody, M6 (0.5 µl of 2 µg/ml protein A purified) was added to the binding reactions after incubation for 15 min with probe, and the incubation was continued for an additional 10 min.
Plasmid Construction.
Construction of deletions in v-ski has been described previously (15)
. The c-ski forms of the mutants—c-ski1, c-ski3, c-ski4, c-skiAH1, c-skiPro, c-skiAH2, c-skiAH4, and c-skiZ3/4—were made by replacing a segment of c-ski cDNA (clone FB29; 53)
in the adaptor plasmid ClaI2Nco (54)
with the corresponding fragment from the cognate v-ski deletion mutant. The cysteine-to-serine missense mutations MT1, MT2, and MT3 mutants were made by standard oligonucleotide-directed mutagenesis. The mutants c-skiMT1, c-skiMT2, c-skiMT3, c-ski1, c-ski3, c-skiAH2, c-skiAH4, c-skiZ3/4, and c-skiCT were cloned into the retroviral vector RCASBP as ClaI fragments.
Construction of RSVc-ski and RSVv-ski in the RSVPL expression plasmid has been described previously (16)
. Mutants were transferred from the RCAS retroviral vector (55)
or the ClaI2Nco shuttle vector into RSVPL (16)
for transient expression in reporter assays. The mutants c-skiMT1, c-skiMT2, c-skiMT3, c-ski1, c-ski3, c-skiAH2, c-skiAH4, c-skiZ3/4, and c-skiCT were cloned as ClaI to SalI fragments into likewise digested RSVPL. The mutants c-ski4, c-skiAH1, and c-skiPro were cloned into RSVEEPL by replacing sno in RSVEEPLsno (17)
with NcoI to ClaI fragments from the appropriate ClaI2Nco clones. To generate proviruses, these mutants c-skiPro were excised from RSVEEPL as BsiWI to XbaI fragments and directionally cloned into a modified version of RCASBP, called RCASBPXS.
To generate fusions with the Gal4 DBD, RSVPL or RSVEEPL plasmids containing c-ski, c-skiMT1, c-skiMT2, c-skiMT3, c-skiAH1, c-skiPro, c-skiAH2, c-skiAH4, c-skiZ3/4, c-ski1, c-ski4, and v-ski were digested with NcoI, filled in with Klenow polymerase, digested with XbaI, and cloned into SmaI-XbaI-digested pSG424 (56)
. pSG424Ski1–452 was made by digesting SG424c-Ski with DraIII and XbaI, filling in, and religating. SG424Ski1–325 and SG424Ski1–214 were made by the same strategy, except the upstream enzyme sites were BamHI and Acc65 I, respectively. SG424Ski46–214 was made by PCR amplification of a segment of c-skiAH1L with the primers: GCTGGATCCCTGCTAGCAAGAAAG and CAGGGATTTGCTAGCGCATCGGATGCAGGCT. The amplified fragment was digested with BamHI and KpnI and cloned into BamHI to KpnI digested pSG424. pSG424ProSki1–214 and pSG4241Ski1–214 were made by digesting pSG424ProLSki and pSG4241Ski, respectively, with Acc65 I and XbaI, filling in, and religating. pSG424Ski78–214 was made by digesting pSG424Ski46–214 and pSG424ProSki1–214 with NheI and PvuI, and recombining such that the ski sequence between the two NheI sites was eliminated. pSG424Ski1–189 was made by digesting pSG424AH4Ski with NheI and XbaI and religating the compatable ends to delete the intervening sequence. The construction of pSG424Sno1–254 has been described previously (17)
.
Construction of the reporter constructs GTCT2X2tkCAT and G5tkLuciferase has been described previously (16) .
Reporter Gene Assays.
The UMN-SAH/DF#1 chicken fibroblast cell line was used for all of the reporter gene assays. These cells were seeded at a density of 2.5 x 105 cells per 35-mm plate in DMEM with 10% fetal bovine serum and cultured overnight before transfection. Triplicate or duplicate plates were cotransfected with the indicated chloramphenicol acetyltransferase (CAT) or luciferase reporter (600 ng) and the amount of ski expression plasmid indicated in the figures or figure legends. The total amount of DNA per 35-mm well was kept constant at 1.2 µg by the addition of the appropriate amount of empty expression vector DNA (RSVPL). Transfection with DOTAP reagent (Boehringer), harvesting of cells, and assays for CAT or luciferase activity were carried out as described previously (16)
. The protein content of each lysate was determined by the Bio-Rad protein assay, and these values were used to normalize the CAT and luciferase activity data. Normalization to protein concentration did not significantly change the results of the CAT or luciferase assays. We do not use an internal control for transfection efficiency because our earlier work had shown that Ski activates expression of all of the commonly used control plasmids (2)
.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
1 Supported in part by the National Cancer Institute, USPHS, through Grant CA-43600 (to E. S.) and under a contract to A.B.L. (to P. S.) and in part by Fort Dodge Animal Health, Fort Dodge, IA (to D. N. F.). R. N. was supported, in part, by a fellowship from the Albert J. Ryan Foundation. 
2 Present address: Department of Molecular Biology and Oncology, University of Texas Southwestern Medical Center, 6000 Harry Hines Boulevard, Dallas, TX 75235-9148.
3 Present address: AviGenics, Inc., 220 Riverbend Road, Athens, GA 30602.
4 To whom requests for reprints should be addressed, at 10900 Euclid Avenue, Cleveland, OH 44106-4935. Phone: (216) 368-3353; Fax: (216) 368-3419; E-mail: exs44{at}po.cwru.edu
5 P. Tarapore, G. Zheng, and E. Stavnezer. Ski and NFI are conditional partners in transactivation, submitted for publication.
6 The abbreviations used are: QEF, quail embryo fibroblast; CEF, chicken embryo fibroblast; EMSA, electrophoretic mobility shift assay; DBD, DNA binding domain.
7 R. Nicol and E. Stavnezer, unpublished results.
Received for publication 12/16/98. Accepted for publication 2/ 1/99.
|
References |
|---|
|
TOPAbstractIntroductionResultsDiscussionMaterials and MethodsReferences |
|---|
-helical domain. J. Biol. Chem., 272: 31855-31864, 1997.This article has been cited by other articles:
 |
|
 |
|
  N. Ueki and M. J. Hayman Direct Interaction of Ski with Either Smad3 or Smad4 Is Necessary and Sufficient for Ski-mediated Repression of Transforming Growth Factor-{beta} Signaling J. Biol. Chem., August 29, 2003; 278(35): 32489 - 32492. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Akiyoshi, H. Inoue, J.-i. Hanai, K. Kusanagi, N. Nemoto, K. Miyazono, and M. Kawabata c-Ski Acts as a Transcriptional Co-repressor in Transforming Growth Factor-beta Signaling through Interaction with Smads J. Biol. Chem., December 3, 1999; 274(49): 35269 - 35277. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. C. Mertani, T. Zhu, E. L. K. Goh, K.-O. Lee, G. Morel, and P. E. Lobie Autocrine Human Growth Hormone (hGH) Regulation of Human Mammary Carcinoma Cell Gene Expression. IDENTIFICATION OF CHOP AS A MEDIATOR OF hGH-STIMULATED HUMAN MAMMARY CARCINOMA CELL SURVIVAL J. Biol. Chem., June 8, 2001; 276(24): 21464 - 21475. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 W. Xu, K. Angelis, D. Danielpour, M. M. Haddad, O. Bischof, J. Campisi, E. Stavnezer, and E. E. Medrano Ski acts as a co-repressor with Smad2 and Smad3 to regulate the response to type beta transforming growth factor PNAS, May 23, 2000; 97(11): 5924 - 5929. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Cancer Research | Clinical Cancer Research |
| Cancer Epidemiology Biomarkers & Prevention | Molecular Cancer Therapeutics |
| Molecular Cancer Research | Cell Growth & Differentiation |
the minimum transformation domain (amino acids 1–304) of the Ski protein 
, 13 cysteines that are conserved between Ski and Sno proteins from multiple species; arrowheads, MT1, MT2, and MT3 cysteine-to-serine substitutions; heavy lines below the transformation domain, location of the repression domain and GTCT binding domain; blocks above the transformation domain, locations of deletions.