Transcriptional Regulation of the Mouse Ferritin H Gene

We previously identified a major enhancer of the mouse ferritin H gene (FER-1) that is central to repression of the ferritin H gene by the adenovirus E1A oncogene (Tsuji, Y., Akebi, N., Lam, T. K., Nakabeppu, Y., Torti, S. V., and Torti, F. M. (1995) Mol. Cell. Biol. 15, 5152–5164). To dissect the molecular mechanism of transcriptional regulation of ferritin H, E1A mutants were tested for their ability to repress FER-1 enhancer activity using cotransfection with ferritin H-chloramphenicol acetyltransferase (CAT) reporter constructs. Here we report that p300/CBP transcriptional adaptor proteins are involved in the regulation of ferritin H transcription through the FER-1 enhancer element. Thus, E1A mutants that failed to bind p300/CBP lost the ability to repress FER-1, whereas mutants of E1A that abrogated its interaction with Rb, p107, or p130 were fully functional in transcriptional repression. Transfection with E1A did not affect endogenous p300/CBP levels, suggesting that repression of FER-1 by E1A is not due to repression of p300/CBP synthesis, but to E1A and p300/CBP interaction. In addition, we have demonstrated that transfection of a p300 expression plasmid significantly activated ferritin H-CAT containing the FER-1 enhancer, but had a marginal effect on ferritin H-CAT with FER-1 deleted. Furthermore, both wild-type p300 and a p300 mutant that failed to bind E1A but retained an adaptor function restored FER-1 enhancer activity repressed by E1A. Sodium butyrate, an inhibitor of histone deacetylase, mimicked p300/CBP function in activation of ferritin H-CAT and elevation of endogenous ferritin H mRNA, suggesting that the histone acetyltransferase activity of p300/CBP or its associated proteins may contribute to the activation of ferritin H transcription. Recruitment of these broadly active transcriptional adaptor proteins for ferritin H synthesis may represent an important mechanism by which changes in iron metabolism are coordinated with other cellular responses mediated by p300/CBP.

Ferritin is the major cellular iron-binding protein in eucaryotic cells. It is composed of 24 subunits of two types (termed ferritin H and ferritin L) that are assembled in various ratios in different tissue and disease states, including inflammation and cancer (reviewed in Refs. 1 and 2). Iron is a potent regulator of ferritin synthesis. In the past decade, a number of elegant studies have shown that iron coordinately regulates both the H and L subunits of ferritin at a post-transcriptional level through the interaction of an iron-responsive element with its binding proteins (reviewed in Ref. 3).
We and others have shown that ferritin is also subject to transcriptional regulation. Transcription of the ferritin H gene is preferentially modulated by cytokines (4,5), hormones (6), oncogenes (7,8), and some inducers of cell differentiation (9). The preferential alteration of ferritin H synthesis in response to these stimuli may be an adaptive cellular response since ferritin H appears to be responsible for more rapid iron uptake than ferritin L (10) and may therefore protect cells from oxidative stress mediated by iron. In fact, it has been reported that oxidative stress stimulates ferritin H and L synthesis at both transcriptional and post-transcriptional levels (11). Some studies have demonstrated that ferritin synthesis is elevated in tumor tissues compared with their normal counterparts (12)(13)(14)(15), and elevation of serum ferritin has also been reported in various cancer patients (16), although little is known about the molecular mechanism and biological meaning of the increase in ferritin associated with malignancy. Recent studies have suggested that ferritin H may function as an autocrine growth factor for some tumors (17,18). An appreciation of mechanisms regulating ferritin synthesis is critical to understanding these responses.
We have previously reported that the adenovirus E1A oncogene selectively represses transcription of the ferritin H gene in mouse NIH3T3 fibroblasts (8). Reduced ferritin transcription in E1A-expressing cells may contribute to their enhanced sensitivity to tumor necrosis factor cytotoxicity (19). Repression of ferritin H by E1A is also consonant with numerous reports showing that E1A can repress markers of cell differentiation (20).
Extensive studies on the mechanism of action of E1A have revealed that many of its pleiotropic effects on gene transcription are mediated by interaction with important and broadly active intracellular regulators. These include members of the Rb 1 tumor suppressor family (p105Rb, p107, and p130) as well as p300 and CBP (CREB-binding protein). p300 and CBP are functionally related transcriptional coactivators with intrinsic histone acetylase activity that link key transcription factors (including nuclear hormone receptors, MyoD, Fos, Jun, CREB, and p53) to the basal transcriptional machinery (reviewed in * This work was supported by Grant DK-42412 from the National Institutes of Health. Phosphoimaging analysis was performed in a facility supported by Grant CA12197 from the National Institutes of Health and by Grant 9510-IDG-1006 from the North Carolina Biotechnology Center. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  Ref. 21). It has been demonstrated that p300 plays a role in cell cycle control, inhibiting exit from the G 0 /G 1 phases of the cell cycle and stimulating differentiation (22)(23)(24)(25). p300/CBP thus functions as a nodal point in mediating cellular responses to a broad range of intrinsic and extrinsic stimuli (reviewed in Ref. 21).
We have identified an E1A-responsive element (termed FER-1) in the 5Ј-flanking region of the mouse ferritin H gene (26). FER-1 serves as a basal enhancer of the mouse ferritin H gene. FER-1 comprises two elements, an AP1-like sequence followed by a dyad symmetry element, both of which are essential for maximal FER-1 enhancer activity (26). Very recently, we identified transcription factors that directly bind to the FER-1 element: JunD, FosB, and ATF1 (bound to the AP1-like sequences) and Sp1 and Sp3 (bound to the C-rich sequences in the dyad) (27). Our discovery of the regulation of ferritin H by E1A allowed us to further dissect and explore the relationship between binding of these transcription factors and functional modulation of ferritin H synthesis. To study the possible involvement of E1A-associated proteins in the regulation of ferritin H transcription via FER-1, we utilized E1A mutant proteins defective in their ability to associate with selected E1A target proteins. Here we demonstrate that 1) the functional region of E1A containing the p300/CBP-binding domain contributes to the repression of endogenous ferritin H mRNA; 2) E1A point mutants defective in the p300/CBP-binding domain fail to repress ferritin H transcription; 3) p300 is able to restore ferritin H enhancer activity repressed by E1A; 4) in the absence of E1A, p300 and CBP are able to activate FER-1 enhancer activity; and 5) sodium butyrate, a histone deacetylase inhibitor, is able to mimic the p300/CBP effect in terms of the activation of endogenous ferritin H mRNA synthesis as well as ferritin H-CAT. These results suggest that p300/CBP is an additional component that regulates the transcription of the mouse ferritin H gene via the FER-1 element.
Plasmids-All CAT reporter constructs used in this study have been described previously (26). Briefly, expression of CAT in pBluescript KS(Ϫ)Ϫ4.8kbFHCAT is driven by the 5Ј-flanking region of the mouse ferritin H gene from nucleotides Ϫ4819 to ϩ24, containing the FER-1 enhancer element. pBluescript KS(Ϫ)Ϫ4kbAP1FHCAT transcription is driven by nucleotides Ϫ4060 to ϩ24 of the mouse ferritin H gene, containing an AP1(NF-E2) site, but not FER-1. pBluescript KS(Ϫ)Ϫ4kbAP1ϩ␦FHCAT transcription is driven by nucleotides Ϫ4109 to ϩ24 of the mouse ferritin H gene, containing FER-1. The human p300 expression plasmids CMV␤p300 and CMV␤p300del30 (deletion of amino acids 1737-1809 in the E1A-binding region of p300) were kindly provided by R. Eckner (University of Zurich, Zurich, Switzerland) (28). The mouse CBP expression plasmid pRc/RSV-mCBP-HA was a generous gift from R. Goodman (Vollum Institute, Portland, OR) (29). The E1A plasmids used in this study have been described elsewhere (24,30).
DNA Transfection and CAT Assay-Transient DNA transfection into NIH3T3 cells was performed by the calcium phosphate precipitation method as described previously (19) with the following minor modifications. Cells were plated in duplicate at a density of 3.5 ϫ 10 5 /60-mm dish containing 4 ml of the culture medium. After incubation for 16 -24 h, 400 l of calcium phosphate/DNA precipitate solution containing 5 g of a CAT reporter plasmid plus a total 5 g of p12SE1A, the human p300 plasmid, or the mouse CBP plasmid was added to the culture and incubated at 37°C in 5% CO 2 for 16 -24 h. Cells were then washed twice with phosphate-buffered saline, followed by further incubation for 16 -24 h in culture medium. CAT assays were carried out as described previously (8). Sodium butyrate was purchased from Sigma.
Western and Northern Blots-Expression of endogenous p300/CBP and various E1A mutants transiently transfected into NIH3T3 cells was estimated by Western blotting as described previously (31) using the same cell extract as used in measuring CAT activity. An anti-p300 monoclonal antibody (NM11) that reacts with both p300 and CBP (32) was used to detect p300/CBP proteins. The anti-E1A antibody M58 was a generous gift from E. Harlow (Harvard Medical School). Ferritin mRNA was analyzed by Northern blotting as described previously (8).

Repression of Ferritin H by E1A
Requires an Intact p300binding Domain-We previously demonstrated that the E1A oncogene preferentially represses the H subunit of the ferritin gene in NIH3T3 cells (8). To assess the involvement of functional domains of E1A in ferritin H repression, we measured the mRNA level of ferritin in NIH3T3 cells that were stably transfected with intact E1A or E1A mutants in which the E1A region containing conserved region 1 (⌬23-107) or conserved regions 1 and 2 (⌬23-150) was deleted. Mutants of E1A containing conserved region 1 deletions are impaired in binding to p300/CBP; conserved region 2 deletions interfere with binding to the pocket proteins p105Rb, p107, and p130 (reviewed in Ref. 33). Total RNA isolated from control vector (Hyg and HygSR␣)or E1A (intact E1A, ⌬23-107, and ⌬23-150)-transfected cells was analyzed by Northern blotting after successive hybridization with cDNA probes for ferritin H, ferritin L, and ␤-actin. As we reported previously, expression of intact E1A repressed the mRNA for the H subunit of ferritin in the absence of any effect on the mRNA for the L subunit of ferritin or ␤-actin (Fig. 1). In contrast, deletion of the E1A region containing conserved region 1 (⌬23-107, two independent cell pools (A and B)) or conserved regions 1 and 2 (⌬23-150) completely impaired the ability of E1A to repress ferritin H. The expression level of these E1A mutant proteins was higher than that of intact E1A protein (data not shown) (19). These results suggest that the functional region of E1A containing the p300-binding domain and conserved region 1 contributes to the repression of ferritin H.
To further investigate the domain of E1A required for transcriptional repression of the ferritin H gene, a series of E1A mutants defective in association with either p300 or the Rb family proteins (Rb, p107, or p130) were cotransfected with Ϫ4.8kbFHCAT, our largest ferritin H 5Ј-fragment fused to the CAT gene, into NIH3T3 cells ( Fig. 2A). The E1A mutants that retained the ability to associate with all these cellular proteins (DA21, LP49, EV55, and ⌬81-120) repressed ferritin H enhancer activity as efficiently as wild-type E1A (12Swt). Comparable repression of the ferritin H enhancer was seen by E1A mutants blocked in the ability to interact with Rb and p130 (YH47) or Rb, p107, and p130 (YH47/928). In contrast, mutations that interfered with p300 interaction, but not with the Rb family proteins (⌬2-36, ⌬15-35, and RG2), significantly reduced the ability of E1A to repress ferritin H. Loss of ferritin H repression was similarly observed following cotransfection of Ϫ4.8kbFHCAT with E1A mutants containing two point mutations (RG2/928) that abrogate the interaction of E1A with both p300 and Rb family members. Similar expression levels of all E1A proteins were detected by Western blotting (Fig. 2B). Endogenous p300 levels were not altered by E1A transfection (Fig. 2B), suggesting that repression of the ferritin H enhancer by E1A is not due to repression of p300 synthesis by E1A, but rather is dependent on the interaction of E1A with p300.
p300/CBP Activates Ferritin H Transcription-p300 was initially identified as an E1A-associated protein (34,35). Subsequent studies have demonstrated that p300 has a structural homology to CBP (36) as well as a functional similarity to CBP (37, 38) as a coactivator of transcription factors such as CREB and AP1 (39,40). We previously identified an element (termed FER-1) as responsible for transcriptional repression of the mouse ferritin H gene by E1A (26). FER-1 is a 37-base pair composite element that binds AP1 and Sp1 family members and functions as a basal enhancer of the mouse ferritin H gene through the cooperative action of these two families of transcription factors (27). These results and the data obtained in Fig. 2, taken together with evidence for the interaction between p300/CBP and some members of the AP1 family (41,42), prompted us to test whether or not p300 and CBP contribute to ferritin H enhancer activity through FER-1. To test this possibility, we first carried out cotransfection experiments using a p300 or CBP expression plasmid and Ϫ4.8kbFHCAT. As shown in Fig. 3, both p300 and CBP activated ferritin H transcription in a dose-dependent manner. Activation was also seen with p300del30, a mutant of p300 that retains transcriptional adaptor activity (28). These functional assays thus demonstrate that p300/CBP activates the ferritin H promoter.
p300-dependent Activation of Ferritin H Is Mediated by the FER-1 Element-To assess the involvement of the FER-1 element in ferritin H enhancer activation by p300, we then performed cotransfection of a p300 expression plasmid with ferritin H-CAT that lacks FER-1 (Ϫ4kbAP1) or contains FER-1 (Ϫ4kbAP1ϩ␦). p300 by itself significantly activated FER-1(ϩ)FHCAT in a dose-dependent manner (Fig. 4). In contrast, only a weak activation of CAT by p300 was observed in FER-1(Ϫ)FHCAT (this weak activation may be due to the existence of another potential p300-targeted sequence on the mouse ferritin H gene; see "Discussion"). These results suggest that the p300 adaptor protein activates the ferritin H enhancer via the FER-1 element.
E1A Represses Ferritin H by Targeting p300 -We then asked whether E1A repressed ferritin H transcription by targeting p300. To this end, a FER-1-containing ferritin H-CAT plasmid was cotransfected with an E1A expression plasmid together with either wild-type p300 or a p300 deletion mutant (p300del30) that is defective for interaction with E1A, but is active as an adaptor (28). CAT activity driven by the FER-1containing ferritin H promoter was ϳ60% repressed by 12SE1A (Fig. 5). Cotransfection of wild-type p300 resulted in ϳ80% restoration of E1A-mediated repression of CAT activity, and FIG. 2. E1A mutants defective in p300 binding lose the ability to repress ferritin H transcription. A, 5 g of pUC18 or 5 g of each E1A plasmid (except 12Swt, YH47, or YH47/928, for which 3 g of each E1A plasmid plus 2 g of pUC18 were utilized to attain expression levels comparable to those exhibited by other E1A mutants) was cotransfected with 5 g of pBluescriptϪ4.8kbFHCAT into NIH3T3 cells. After 36 -48 h, CAT activity was assayed, and the radioactive spots on thin-layer chromatography plates were quantitated by a phosphoimage analyzer. Percent acetylation of chloramphenicol in extracts from cells transfected with pUC18 was defined as 100%. DNA transfection was carried out in duplicate in each experiment, and the results of five independent experiments and standard errors are shown. The abilities of E1A mutants to interact with p300, p105Rb, p107, and p130 are shown at the top (ϩ, interaction, Ϫ, no interaction) according to previous studies (24,30). B, levels of endogenous p300/CBP (top) and transfected E1A proteins (bottom) were simultaneously measured by Western blotting using anti-p300/CBP antibody NM11 and anti-E1A antibody M58, respectively. E1A mutant ⌬81-120 protein was not detected by anti-E1A antibody M58 since it lacks the M58 epitope, but an expression level similar to that of 12 S wild-type E1A was confirmed by Western blotting using anti-E1A antibody M73 (data not shown). p300del30 completely restored the CAT activity repressed by E1A. A control vector (pCMV) had no effect (Fig. 5). The fact that p300del30 efficiently antagonized E1A-mediated ferritin H repression (p300del30 cannot associate with E1A, but retains a transcriptional adaptor function) suggests that the CAT activity restored by p300 protein is not simply due to sequestration of E1A, but rather to transcriptional activation of the ferritin H enhancer by p300.
Inhibitor of Histone Deacetylase Augments Ferritin H Transcription-It has recently been demonstrated that p300 and CBP both bind to the histone acetylase PCAF (49) and possess intrinsic histone acetyltransferase activity (43). To test the importance of histone acetylation in transcriptional activation of the ferritin H gene, sodium butyrate, a histone deacetylase inhibitor, was employed to investigate whether sodium butyrate can mimic p300/CBP function in ferritin H transcription.
NIH3T3 cells were transfected with a CAT reporter plasmid driven by the ferritin H promoter/enhancer, followed by treatment with a different concentration of sodium butyrate. As shown in Fig. 6, sodium butyrate increased CAT activity up to 2-fold in a dose-dependent fashion, an increase comparable to the 2-3-fold activation of ferritin H-CAT by p300 or CBP (Fig.  3). We also tested the effect of sodium butyrate on endogenous ferritin H mRNA expression by Northern blotting. These experiments demonstrated that ferritin H mRNA was induced 2.4-fold by 2 mM sodium butyrate and 3.0-fold by 10 mM sodium butyrate after 24 h (Fig. 6). These results indicate that conditions that block histone deacetylation activate ferritin H transcription and suggest that the histone acetyltransferase activity of p300/CBP or its associated proteins may contribute to its Ϫ4kbAP1ϩ␦ (ϩFER-1) was cotransfected with 5 g of CMV vector (z), 4 g of CMV vector plus 1 g of CMVp300 (^), or 5 g of CMVp300 (f) into NIH3T3 cells. After 36 -48 h, CAT activity was measured, and the radioactive spots on thin-layer chromatography plates were quantitated by a phosphoimage analyzer. Percent acetylation of chloramphenicol in extracts from cells transfected with 5 g of CMV vector (z) and 5 g of Ϫ4kbAP1ϩ␦ (ϩFER-1) was defined as 100%. DNA transfection was carried out in duplicate in each experiment, and the results of six independent experiments and standard errors are shown.
FIG. 5. p300 restores E1A-mediated ferritin H repression. 5 g of Ϫ4.8kbFHCAT was cotransfected with 5 g of pUC18, 3 g of pUC18 plus 2 g of p12SE1A, 3 g of CMV vector plus 2 g of p12SE1A, 3 g of CMVp300 plus 2 g of p12SE1A, or 3 g of CMVp300del30 plus 2 g of p12SE1A into NIH3T3 cells. After 36 -48 h, CAT activity was measured, and the radioactive spots on thin-layer chromatography plates were quantitated by a phosphoimage analyzer. Percent acetylation of chloramphenicol in extracts from cells transfected with 5 g of pUC18 was defined as 100%. DNA transfection was carried out in duplicate in each experiment, and the results of four independent experiments and standard errors are shown. activation of the FER-1 enhancer and elevation of ferritin H transcription. DISCUSSION We previously observed that E1A preferentially represses transcription of the ferritin H gene in mouse NIH3T3 fibroblasts, resulting in an alteration of the H/L subunit ratio of ferritin (8). To understand the molecular mechanisms of transcriptional regulation of the ferritin H gene and its repression by the E1A oncogene, we used deletion and mutational analyses to investigate an E1A-responsive element in the mouse ferritin H gene. These studies defined a 37-base pair element (termed FER-1) as both the E1A-responsive element and a basal enhancer of the mouse ferritin H gene (26). FER-1, located 4.1 kilobases upstream from the transcription initiation site, comprises an AP1-like element followed by an element with dyad symmetry (26). Our recent studies have identified the nuclear factors bound to these two elements of FER-1: the AP1-like element has the ability to bind JunD, FosB, and ATF1, and the C-rich sequences in the dyad symmetry can bind Sp1 and Sp3 (27). Using recombinant ATF1 and Sp1, we have demonstrated that these transcription factors bind simultaneously to each element of FER-1 (27), regulating ferritin H transcription by binding directly to FER-1 sequences.
In this study, we have identified p300/CBP as an additional component required for FER-1 enhancer activity. Thus, E1A mutants that did not have a p300-binding region including conserved region 1 failed to repress the endogenous ferritin H gene (Fig. 1). Furthermore, in CAT reporter assays, E1A point mutants lacking the ability to bind p300 failed to repress FER-1, whereas E1A mutants defective in interaction with members of the Rb family retained the ability to repress FER-1 ( Fig. 2A). Since endogenous p300/CBP levels were not affected by E1A expression (Fig. 2B), the mechanism of E1A-mediated repression of FER-1 enhancer activity is not due to downregulation of p300/CBP synthesis by E1A, but to E1A and p300/CBP interaction.
Overall, the transcriptional activity of the ferritin H promoter depends in a complex way on the AP1, Sp1, and p300/ CBP transcription factor families. For example, although the Sp1/Sp3-binding component of FER-1 does not by itself confer enhancer activity, it is required for the enhancer activity of the AP1-like element of FER-1 to be expressed (26). In contrast, cotransfection experiments have demonstrated that FosB and JunD directly activate ferritin H transcription. Thus, cotransfection of FosB, JunD, and ferritin H-CAT in F9 cells (which do not contain appreciable endogenous levels of AP1 (44)) led to an ϳ6-fold enhancement of ferritin H-CAT activity (27). As described here, p300/CBP also plays a role in ferritin H transcription, augmenting ferritin H-CAT activity 2-3-fold in NIH3T3 cells (Fig. 3) and BNL CL.2 cells. 2 This effect of p300/CBP on ferritin H transcription is within the range reported for the effects of p300/CBP on other enhancers (2-3-fold) (23,45,46). Furthermore, indirect evidence suggests that the phosphorylation status of p300 may influence its ability to function as a coactivator in ferritin H transcription. Thus, in undifferentiated F9 cells, p300 is poorly phosphorylated and possesses weak coactivator activity; however, both phosphorylation and coactivator activity increase following differentiation of F9 cells (47,48). We observed that cotransfection of p300 and ferritin H-CAT into F9 cells resulted in a minimal augmentation of ferritin H-CAT activity over that seen with AP1 alone, 2 despite the demonstrated functionality of the FER-1 element in F9 cells (27). This suggests that in cellular contexts in which p300 is poorly phosphorylated, the effect of p300 on ferritin H transcription may be blunted. It will be interesting to test this hypothesis by assessing the relationship between p300 phosphorylation and FER-1 enhancer activity during the differentiation of F9 cells. Taken together, these results suggest that the activity and abundance of Sp1, AP1/ATF1, and p300/CBP transcription factor families collectively determine the activity of the FER-1 element of ferritin H in a complex and cell typedependent fashion.
What is the role of p300/CBP in the FER-1 complex? Recent work has suggested that p300 is more than a simple adaptor between DNA-binding proteins and transcription initiation factors. Thus, p300/CBP was shown to possess intrinsic histone acetyltransferase activity (43) as well as to associate with PCAF, a protein that also functions in histone acetylation (49). Since transcriptionally active chromatin is hyperacetylated (reviewed in Ref. 50), p300/CBP-mediated histone acetylation may function to facilitate access of transcription factors to the target sequences of DNA normally compacted in chromatin. Our experiments provide two lines of evidence that are consistent with a role for p300/CBP-mediated histone acetylation in FER-1 function. The first derives from the observation that E1A competes with PCAF for p300 interaction since they share an overlapping binding domain on the p300 protein (E1A binds amino acids 1572-1818 of p300 (28), and PCAF binds amino acids 1801-1851 of p300 (49)). In our experiments, a p300 mutant lacking the E1A-binding domain (p300del30, deletion of amino acids 1737-1809 of p300) was able not only to activate ferritin H-CAT (Fig. 3), but to restore E1A-mediated repression of ferritin H-CAT (Fig. 5). We also examined the behavior of another p300 mutant, p300del33. This mutant contains a deletion encompassing 70% of the PCAF-binding region (amino acids 1737-1836) and is predicted to have little or no interaction with PCAF (49). In contrast to p300del30, p300del33 was unable to substantially augment ferritin H-CAT expression, 2 as was observed in the study of functional collaboration of p300 and C/EBP␤ (CCAAT-box enhancer binding protein ␤) (51). These results suggest that activation of ferritin H by p300/CBP may require histone acetylation and, in particular, association p300/CBP with PCAF. The second line of evidence supporting the idea that histone acetylation plays an important role in ferritin H transcription is the observation that the histone deacetylase inhibitor sodium butyrate can mimic p300/CBP as an activator of the endogenous ferritin H gene as well as ferritin H-CAT (Fig. 6). These results are consistent with previous observations (52).
Based on the results obtained in this study, we suggest the model depicted in Fig. 7 for transcriptional regulation of the mouse ferritin H gene. We speculate that p300/CBP-mediated transcriptional activation of the ferritin H gene may be achieved, at least in part, by tethering the FER-1-binding complex to the basal transcription regulatory region (Fig. 7). This idea is supported by two observations. First, p300/CBP can associate with AP1/CREB family members including ATF1 (reviewed in Ref. 21), which were identified as components of the FER-1-binding complex; second, p300 and CBP were found to be components of the TATA-binding protein complexes (32). Interactions at FER-1 may thus serve to target p300/CBP and its associated histone acetylase activity to the ferritin H promoter. Taken together with our prior results demonstrating a reduction in factors bound to the AP1-like element of FER-1 in the presence of E1A (26), this model suggests at least two mechanisms of E1A-mediated transcriptional repression of ferritin H. One is that E1A reduces the ability of factors to bind to the AP1-like element (26), as has been reported for repression of the AP1 site in the collagenase gene by E1A (53). The other is that E1A removes p300 or CBP from the protein complexes of FER-1 and the TATA box. Both mechanisms may collaborate to decrease ferritin H transcription.
Our results suggest that p300 plays an important role in ferritin H transcriptional activity through its interaction with FER-1. While this work was in progress, another potential target of p300/CBP in the ferritin H gene was identified. Bevilacqua et al. (54) reported the presence of a cyclic AMP-responsive element (B-site, Ϫ62 to Ϫ45 nucleotides upstream from the transcription initiation site) in the human ferritin H gene. They suggested that cAMP-mediated induction of human ferritin H transcription is inhibited by E1A through sequestration of p300/CBP. The B-site does not contain a canonical cAMPresponsive element sequence and was reported to bind a pro-tein complex containing a 120-kDa protein and p300 (54). This element overlaps an NF-Y-binding region containing a CCAAT box in inverse orientation, which is important in ferritin H induction during monocyte to macrophage differentiation (55). Comparison of mouse and human proximal 5Ј-flanking sequences reveals 78% conservation in the B-site and complete conservation of the inverse CCAAT box (for mouse ferritin H DNA 5Ј-sequences (56)). Although this region does not function as a strong enhancer in the mouse ferritin H gene, 2 it may have weak enhancer activity, and FER-1(Ϫ)FHCAT used in Fig. 4, which contains the B-site/CCAAT box, may have been slightly activated by p300. Thus, p300 may influence multiple regulatory elements in the ferritin H gene. However, our results indicate that p300/CBP activates transcription of the mouse ferritin H gene primarily through the FER-1 enhancer element. These results also suggest that similar mechanisms of E1Amediated transcriptional repression of ferritin H are operative in both mouse and human genes, despite differences in functional roles and binding complexes between the FER-1 element and the B-site.
What is the biological implication of the involvement of p300/ CBP in the FER-1 enhancer activity of the ferritin H gene? One possibility is that p300/CBP may, at least in part, play a role in determining tissue-specific expression of ferritin H. FER-1 lies within the 180-nucleotide SacI-BglII fragment identified as responsible for the up-regulation of ferritin H in erythroid cells induced to differentiate with N,NЈ-hexamethylene-bis-acetamide (9). It is unclear whether the FER-1-binding complex (containing AP1 and Sp1 family members as well as p300/CBP) is involved in N,NЈ-hexamethylene-bis-acetamide-dependent induction of ferritin H in erythroid cells. However, it has been reported that cooperative interaction between AP1 and Sp1 family members may underlie myeloid-specific expression of the leukocyte integrin gene CD11c (57), suggesting that tissuespecific alterations in the constellation of proteins binding to FER-1 may similarly contribute to tissue-specific regulation of ferritin H in other cell types, including erythroid cells. In conjunction with transcription factors directly bound to FER-1, p300/CBP may also modulate ferritin H transcription during cell immortalization. It will be important to dissect the potentially changing composition of the FER-1-binding complex during these processes.
In summary, we have demonstrated the involvement of p300/ CBP in the basal enhancer activity of the mouse ferritin H gene via the FER-1 element, indicating that at least three families of factors (AP1/ATF, Sp1/Sp3, and p300/CBP) regulate ferritin H transcription through the FER-1 element. Environmental or intracellular events that target each of these, individually or collectively, may be expected to modulate ferritin H transcription. These molecular observations may help to explain the response of ferritin H to diverse cellular signals. FIG. 7. Model of FER-1-mediated transcriptional regulation of the mouse ferritin H gene. FER-1 is composed of an Sp1-like and an AP1-like element, to which Sp1/Sp3 and FosB/JunD/ ATF1 bind. p300/CBP may tether this complex to the basal enhancer complex (TBC), and its histone acetyltransferase activity may also be involved in the activation of FER-1. E1A reduces the binding of factors to the AP1-like element of FER-1. E1A interacts with p300/CBP, a coactivator of FER-1, inhibiting the adaptor function of p300/CBP and leading to repression of ferritin H transcription. kb, kilobases.