Advertisement

Transcription factor‐dependent regulation of CBP and P/CAF histone acetyltransferase activity

Evi Soutoglou, Benoit Viollet, Martine Vaxillaire, Moshe Yaniv, Marco Pontoglio, Iannis Talianidis

Author Affiliations

  1. Evi Soutoglou1,
  2. Benoit Viollet2,
  3. Martine Vaxillaire2,3,
  4. Moshe Yaniv2,
  5. Marco Pontoglio2 and
  6. Iannis Talianidis*,1
  1. 1 Institute of Molecular Biology and Biotechnology, Foundation for Research and Technology Hellas, PO Box 1527, 711 10, Heraklion, Crete, Greece
  2. 2 Unite des Virus Oncogenes, CNRS, Institut Pasteur, 25 rue du Dr Roux, 75724, Paris, Cedex 15, France
  3. 3 Present address: CNRS EP‐10, Institut Pasteur de Lille, Lille, France
  1. *Corresponding author. E-mail: talianid{at}imbb.forth.gr
  1. E.Soutoglou and B.Viollet contributed equally to this work

View Full Text

Abstract

CREB‐binding protein (CBP) and CBP‐associated factor (P/CAF) are coactivators possessing an intrinsic histone acetyltransferase (HAT) activity. They are positioned at promoter regions via association with sequence‐specific DNA‐binding factors and stimulate transcription in a gene‐specific manner. The current view suggests that coactivator function depends mainly on the strength and specificity of transcription factor–coactivator interactions. Here we show that two dominant‐negative mutants of hepatocyte nuclear factor‐1α (HNF‐1α), P447L and P519L, occurring in maturity onset diabetes of the young (MODY3) patients, exhibit paradoxically stronger interactions than the wild‐type protein with either CBP or P/CAF. However, CBP and P/CAF recruited by these mutants lack HAT activity. In contrast, wild‐type HNF‐1α and other transcription factors, such as Sp1 or HNF‐4, stimulated the HAT activity of CBP. The results suggest a more dynamic role for DNA‐binding proteins in the transcription process than was considered previously. They are not only required for the recruitment of coactivators to the promoter but they may also modulate their enzymatic activity.

Introduction

Acetylation and deacetylation of core histones by histone acetyltransferases (HATs) and histone deacetylases (HDACs) is a tightly regulated process that plays a key role in transcriptional regulation (Grunstein, 1997; Struhl, 1998). Coactivator complexes possessing acetyltransferase activities are positioned on the promoter via physical association with sequence‐specific DNA‐binding factors, and stimulate transcription by acting as bridging proteins between activators and the general transcription factors and by leading to localized hyperacetylation of the neighboring chromatin (Xu et al., 1999). The current view suggests that the main function of transcription factors is to provide an initial platform for the selective recruitment to and further assembly of multicomponent coactivator complexes on the promoter. Coactivators themselves have a low inherent transcriptional activity, which is highly stimulated upon their interaction with transcription factors (Puigserver et al., 1999). This switch in the transcriptional activity of coactivators has been widely observed and, at least in the case of PGC‐1, it has been shown that the mechanism involves transcription factor‐mediated increased recruitment of additional coactivator proteins to form a highly active coactivator complex (Puigserver et al., 1999). The potential modulation of the enzymatic activities of coactivators as a result of their differential association with transcription factors has not been addressed before.

To investigate this issue, we studied the functional interactions of coactivator proteins with point mutant forms of hepatocyte nuclear factor‐1α (HNF‐1α) naturally occurring in maturity onset diabetes of the young type 3 (MODY3). HNF‐1α originally was characterized as a liver‐specific transcription factor (Tronche and Yaniv, 1992) playing a central role in the coordination of the complex cross‐regulatory network that determines the hepatic phenotype (Tian and Schibler, 1991; Kuo et al., 1992; Ktistaki and Talianidis, 1997). The discovery that heterozygous HNF‐1α mutations are associated with MODY3 has established the important role of this transcription factor in pancreatic β‐cell function (Yamagata et al., 1996; Frayling et al., 1997; Glucksmann et al., 1997; Hansen et al., 1997; Vaxillaire et al., 1997). Recent studies have demonstrated a clear link between functional defects in HNF‐1α‐dependent transactivation and MODY3 mutations (Vaxillaire et al., 1999). These defects include reduced protein stability, DNA binding or transactivation potential. In addition, some mutant proteins (e.g. P447L or P519L) exhibited a dominant‐negative effect in transient co‐transfection assays (Vaxillaire et al., 1999).

HNF‐1α interacts with multiple coactivator proteins, and the synergistic action of the CREB‐binding protein (CBP)/p300 and CBP‐associated factor (P/CAF) acetyltransferases is required for high level target gene activation in the in vivo chromatin context (Soutoglou et al., 2000a). The requirement for the simultaneous action of CBP and P/CAF on HNF‐1α‐dependent transcription (Soutoglou et al., 2000a) raised the possibility that the reduced transactivation potential of several MODY3 mutants may be due to reduced interactions of these proteins with either one of the two coactivators. Alternatively, it could be related to differential recruitment of corepressor complexes to the promoter.

In the present study, we show that two naturally occurring HNF‐1α mutants (P447L and P519L) unexpectedly bound CBP and P/CAF more avidly than wild‐type HNF‐1α. However, the bound coactivators lacked HAT activity both in vitro and in vivo. In contrast, wild‐type HNF‐1α, Sp1 or HNF‐4 was able to stimulate the HAT activity of CBP.

Results

CBP and P/CAF interact strongly with P447L and P519L MODY3/HNF‐1α mutants, but fail to coactivate them

HNF‐1α mutations are associated with a familial form of type 2 diabetes mellitus, also known as MODY3. Two of these mutations involving single amino acid substitutions in the transactivation domain (P447L or P519L) abolish transcriptional activation and generate dominant‐negative acting proteins (Vaxillaire et al., 1999 and data not shown). Promoter activation by HNF‐1α requires the synergistic action of both CBP and P/CAF acetyltransferases (Soutoglou et al., 2000a). This raised the possibility that the transcriptional activation deficiency and the dominant‐negative nature of P447L and P519L mutants could be related to a defective interaction with these coactivators.

To test this hypothesis, we first monitored whether addition of excess CBP or P/CAF could restore the transcriptional activity of the P519L and P447L mutants in an NIH 3T3 cell line containing a genome‐inte grated 3×AlbPE‐AdML‐CAT HNF‐1‐responsive reporter (Soutoglou et al., 2000a). Overexpression of CBP, P/CAF or both failed to induce transactivation by P519L and P447L (Figure 1A), suggesting that these mutants may not be able to recruit or functionally interact with the coactivator proteins. To test this, we performed co‐immunoprecipitation assays using lysates from Cos‐1 cells transfected with the individual mutants together with CBP or Flag epitope‐tagged P/CAF. Surprisingly, both P519L and P447L were bound much more avidly by either CBP (Figure 1B) or P/CAF (Figure 1C) than was wild‐type HNF‐1α. These results suggested that even though both mutants were able to interact with the coactivators, this interaction was non‐productive.

Figure 1.

Functional analysis of CBP and P/CAF interactions with HNF‐1α mutants. (A) NIH 3T3 cells with stably integrated 3×AlbPE AdML‐CAT were transfected with 0.5 μg of the indicated CMV‐HNF‐1α mutant plasmids, together with 2 μg of CMV‐CBP or CMV‐P/CAF expression vectors. The bars represent normalized CAT activities and standard errors from four independent experiments, expressed as fold induction above basal levels. Note the different scale of the ordinates. (B and C) Extracts from Cos‐1 cells transfected with the indicated plasmids were immunoprecipitated (IP) with α‐CBP or α‐Flag antibodies, and the presence of HNF‐1α in the immunoprecipitates was assessed in western blots by α‐HNF‐1α antibody. Input lanes represent 15% of the extracts used for immunoprecipitations.

The two HNF‐1α mutations affect proline residues that are located in conserved blocks of amino acids localized in the C‐terminal transactivation domain. When this domain of the protein was fused to a Gal4 DNA‐binding domain, it activated transcription of a Gal4 reporter (Figure 2A). The P447L and P519L mutations, in the same Gal4 fusion context, lacked transactivation potential and behaved as dominant‐negative effectors (Figure 2A and B). We have shown previously that CBP and P/CAF interact with the N‐ and C‐terminal part of HNF‐1α, respectively (Soutoglou et al., 2000a). To evaluate the interactions of CBP and P/CAF with the mutated C‐terminal domains of HNF‐1α, we performed in vitro pull‐down experiments. When the full‐length HNF‐1α proteins were tested, both CBP and P/CAF showed stronger interactions with P447L and P519L compared with the wild‐type protein (Figure 2C). As expected, we observed increased in vitro interaction of P/CAF with the P447L and P519L C‐terminal domains (Figure 2C). On the other hand, CBP did not interact directly with either wild‐type or mutant C‐terminal domains (Figure 2C), suggesting that the increased binding observed with the full‐length P447L and P519L is due to a potential altered conformation of the proteins caused by the mutations that affects CBP interaction with the N‐terminus of HNF‐1α. The existence of such a conformational change is supported by the partial protease digestion pattern of wild‐type and mutant HNF‐1α proteins. Upon partial V8 protease digestion, all three proteins showed similar sensitivity to this enzyme; however, the proteolytic pattern of P447L and P519L differed from that of wild‐type HNF‐1α (Figure 2D).

Figure 2.

HNF‐1α C‐term P447L and P519L Gal4 fusion proteins recapitulate HNF‐1α P447L and P519L full‐length‐mutant properties. (A) C33 cells were co‐transfected with 1 μg of 5×Gal4 E1b‐luc and 0.2 μg of Gal4‐wt, Gal4‐P447L or Gal4‐P519L mutant expression vectors as indicated. (B) C33 cells were co‐transfected with 1 μg of 5×Gal4 E1b‐luc and 0.2 μg of Gal4‐wt together with increasing amounts (0.4, 0.8 and 1.6 μg) of either Gal4, or Gal4 P447L or Gal4‐P519L expression vectors as indicated. Bars represent normalized luciferase activity and standard errors from at least four independent experiments. The data are expressed as fold activation above basal levels. (C) In vitro pull‐down experiments were performed with 35S‐labeled, in vitro translated CBP and P/CAF probes. Full‐length HNF‐1α and its mutant derivatives were expressed as His6‐tagged proteins and immobilized on Talon resin. As a control (Contr.), a column containing the unrelated HrpE protein was used. The C‐terminal parts of HNF‐1α and of its mutant derivatives were expressed as GST fusion proteins. Input lanes represent 15% of the probe used for the interactions. (D) 35S‐labeled, in vitro translated full‐length HNF‐1α and its mutant derivatives were digested with the indicated amounts of V8 protease and analyzed by SDS–PAGE. Arrows indicate protease‐induced bands detectable only with the mutant proteins.

The dominant‐negative effect of MODY3/HNF‐1α mutants is not due to a preferential recruitment of corepressors

The fact that P447L and P519L mutants exhibited reduced transactivations, in spite of their stronger interactions with coactivators, is in sharp contrast to the currently prevailing view, according to which the strength of interaction between transcription factors and coactivators is a major determinant of transcription activation potential. To try to solve this paradox, we have investigated the possible involvement of corepressors in conferring the loss of activation potential to P519L and P447L mutants. To analyze whether such an interaction could occur in vivo, we performed co‐immunoprecipitation experiments using extracts of Cos‐1 cells co‐transfected with cytomegalovirus (CMV)‐NCoR, CMV‐HDAC‐1 and CMV‐wtHNF‐ 1α plasmids or CMV‐HNF‐1α mutant derivatives. Wild‐type HNF‐1α interacted efficiently with NCoR and HDAC‐1 in this assay (Figure 3A). However, glutathione S‐transferase (GST) pull‐down experiments have shown that only NCoR is capable of interacting with the N‐terminal part of HNF‐1α, suggesting that HDAC‐1 interacted with HNF‐1α via NcoR (data not shown). Interestingly, no significant increase in the amounts of either NCoR or HDAC‐1 was detected in the immunoprecipitates when P519L or P447L was examined (Figure 3A). To investigate the functional significance of the corepressor–HNF‐1α association, we treated the cells with trichostatin A (TSA), a specific inhibitor of HDACs (Yoshida et al., 1995), and analyzed the formation of the corepressor complex on HNF‐1α proteins, as well as the effect of this treatment on HNF‐1α‐mediated transcription. After a 12 h treatment of the cells with TSA, we failed to detect any co‐immunoprecipitating NCoR or HDAC‐1 with wild‐type or mutant HNF‐1 (Figure 3A), indicating that allosteric inhibition of HDAC‐1 disrupts the corepressor complex formation in vivo, on both wild‐type HNF‐1 and its mutant forms. In functional assays, TSA treatment resulted in a dramatic increase in wild‐type HNF‐1α‐mediated transcription, to a level that could not be increased further by overexpression of CBP and P/CAF (Figure 3B). Interestingly, HNF‐1α P447L and HNF‐1α P519L remained inactive even after TSA treatment (Figure 3B). This suggests that disruption of the corepressor complex formed on these mutants does not lead to activation and, therefore, could not account for their functional inactivity.

Figure 3.

In vivo association of NCoR and HDAC‐1 with HNF‐1α mutants. (A) Cos‐1 cells were transfected with the indicated plasmids and incubated with or without 1 μM TSA for 12 h before harvest. Nuclear extracts were prepared and immunoprecipitated with α‐HNF‐1α antibody. The precipitated proteins were analyzed in western blots using α‐NCoR or α‐HDAC‐1 antibodies. (B) The effect of TSA treatment on HNF‐1‐mediated transcription. NIH‐3T3 cells with stably integrated 3×AlbPE AdML‐CAT were transfected by 0.5 μg of wild‐type CMV‐HNF‐1α, CMV‐HNF‐1α P519L or CMV‐HNF‐1α P447L, together with 2 μg of CMV‐CBP, CMV‐P/CAF or CMV‐CBP + CMVP/CAF expression vectors. Where indicated, the cells were treated with 1 μM TSA for 12 h, before harvesting. The bars represent normalized CAT activities and standard errors from three independent experiments, expressed as fold inductions above basal levels.

In order to evaluate further the existence of any possible trans‐repressor effect of the mutant HNF‐1α proteins on transcriptional activation, we performed transient transfection using the proximal regulatory regions of the L‐PK promoter gene containing both HNF‐1‐and HNF‐4‐binding sites (Figure 4A). In C33 cells, the L‐PK −150 bp promoter fragment was inactive because of the lack of any endogenous HNF‐1 and HNF‐4 proteins. The −150PK/luc contruct was activated up to 10‐fold by co‐transfection with an HNF‐4 expression vector. Addition of the wild‐type HNF‐1α expression vector resulted in an additional 6‐fold activation. In the presence of the HNF‐4 expression vector, co‐transfection with either HNF‐1α P447L or P519L mutant expression vectors did not significantly change the HNF‐4‐dependent transcription (Figure 4B). Activation levels were comparable to that obtained with the HNF‐4 expression vector alone. These experiments demonstrate that HNF‐1α P447L and P519L mutants do not act as trans‐repressors on another transcription factor.

Figure 4.

Trans‐acting effect of HNF‐1α P447L and P519L mutants on the transcriptional activity of the L‐PK gene promoter. (A) Schematic representation of the −150PK/luc construct corresponding to the proximal regulatory regions of the L‐PK gene promoter. The −150PK/luc construct consists of HNF‐1‐ and HNF‐4‐binding sites in front of the luciferase reporter gene. (B) C33 cells were co‐transfected with 1 μg of −150PK/luc, 1 μg of CMV HNF‐4 and 1 μg of CMV HNF‐1, CMV HNF‐1 P447L or CMV HNF‐1 P519L expression vectors as indicated. Bars represent normalized luciferase activity and standard error from at least three independent experiments. The data are expressed as fold activation above basal level. Statistical significance was determined by a two‐tailed Student's t‐test.

MODY3/HNF‐1α mutants can be part of an HNF‐1α–CBP–P/CAF trimeric complex whose HAT activity is drastically reduced

Since maximal HNF‐1α‐dependent activation requires the simultaneous assembly of both CBP and P/CAF on the molecule (Soutoglou et al., 2000a), we compared the ability of wild‐type and mutant HNF‐1α proteins to form HNF‐1α–CBP–P/CAF trimeric complexes in vivo. Extracts from transfected Cos‐1 cells were first immunoprecipitated with α‐CBP antibody to eliminate HNF‐1α–P/CAF dimeric complexes. The precipitated proteins were eluted with excess amounts of CBP peptide and immunoprecipitated with α‐Flag antibody to get rid of HNF‐1α–CBP dimeric complexes. The amounts of HNF‐1α proteins in a trimeric complex were estimated by western blot analysis of the proteins surviving the second immunoprecipitation. No major differences between wild‐type and mutant HNF‐1α proteins were observed with respect to their capacity to form such trimeric complexes (Figure 5A), suggesting that the increased amounts of CBP and P/CAF associated with HNF‐1α mutants may reflect functionally less active HNF‐1α–CBP and HNF‐1α–P/CAF dimers. On the other hand, the substantial amounts of P519L and P447L detected in the trimeric complex argue against the possibility that their loss of function is due to the failure to form such a complex.

Figure 5.

Analysis of the in vivo formation of the HNF‐1α–CBP–P/CAF trimeric complex and the HAT activities of CBP and P/CAF associated with HNF‐1α proteins. (A) Extracts from Cos‐1 cells transfected with the indicated plasmids were first immunoprecipitated with α‐CBP antibody. The immune complexes were eluted from the protein A–Sepharose column by incubating the beads with 0.1 mg/ml CBP peptide (Santa Cruz Biotechnologies). The eluates were subjected to a second immunoprecipitation with α‐Flag antibody. HNF‐1α proteins in the HNF‐1α–CBP–P/CAF trimeric complex form were detected in the resulting precipitates by western blot analysis using α‐HNF‐1α antibody. The input panel represents 5% of the amounts of extracts used for the first immunoprecipitation. (B) Extracts from Cos‐1 cells transfected by the indicated plasmids were immunoprecipitated with α‐HNF‐1α antibody. The amounts of CBP and P/CAF proteins in the immunoprecipitates were estimated by western blot analysis (upper panels), and their corresponding HAT activities were assessed by IP‐HAT assay (lower panel).

We next examined whether the HAT activities of CBP or P/CAF were altered as a result of their in vivo association with mutant HNF‐1α proteins. To this end, HNF‐1α was immunoprecipitated from Cos‐1 cells overexpressing wild‐type HNF‐1α or its mutant forms together with CBP or P/CAF, and the co‐precipitated HAT activity was measured using histones as substrates. As expected, immunoprecipitates from P519L‐ and P447L‐overexpressing extracts contained much higher amounts of either CBP or P/CAF compared with those co‐precipitating with wild‐type HNF‐1α (Figure 5B). However, the HAT activities of the coactivators co‐precipitated with either P519L or P447L mutants were barely detectable (Figure 5B).

Wild‐type HNF‐1α, but not MODY3 mutants, stimulates the HAT activity of CBP in vitro

To exclude the involvement of any other potential effector that may have co‐precipitated with HNF‐1 and CBP or P/CAF, we performed in vitro HAT assays with purified recombinant proteins. When excess amounts of full‐length P447L and P519L recombinant proteins were incubated with either CBP or P/CAF, we observed a strong reduction in their HAT activity, compared with the activity of those incubated with wild‐type HNF‐1α (Figure 6A and B). The HAT activity of P/CAF but not that of CBP was reduced when the coactivators were incubated with the mutant recombinant proteins containing only the C‐terminal part of HNF‐1α (Figure 6A). Since CBP does not interact with this part of the protein (Figure 2C), the above results suggest that the observed inhibition depends directly on protein–protein interactions. Furthermore, we noticed increased CBP‐mediated histone acetylation in samples that were incubated with the full‐length wild‐type HNF‐1 compared with those incubated with the non‐interacting GST fusion proteins (compare lane 1 with lane 4, Figure 6A). This suggested that the HAT activity of CBP could be up‐regulated by the physical interaction with a transcription factor. To corroborate this assumption further, we tested two other transcription factors, HNF‐4 and Sp‐1, which are known to interact with CBP (Kundu et al., 2000; Soutoglou et al., 2000b). Both proteins increased the HAT activity of CBP at different levels (Figure 6A).

Figure 6.

Analysis of the effects of HNF‐1α and its mutant derivatives on the HAT activity and conformation of CBP and P/CAF, in vitro. A 60 ng aliquot of baculovirus‐expressed CBP (A) or P/CAF (B) was incubated with 1 μg of the indicated recombinant proteins in 1× HAT buffer, for 1 h at 4°C. After the addition of core histones and [3H]acetyl‐CoA, the samples were incubated further at 30°C for 1 h and analyzed by SDS–PAGE. (C and D) CBP and P/CAF were translated in vitro in a 10 times scaled‐up reaction and allowed to interact with Talon resin‐bound full‐length wild‐type HNF‐1α or P447L mutant. After extensive washing, proteins remaining in the beads were digested with the indicated amounts of V8 protease. Dots on the right indicate protease‐induced bands.

The above results provide examples of regulation of the histone acetylase function of coactivators by virtue of their differential association with DNA‐binding transcription factors.

The molecular basis for the transcription factor‐mediated activation–inactivation of the enzymatic activity of CBP and P/CAF may involve differential masking of their HAT domains by the altered conformation of the interacting partner. Alternatively, CBP and P/CAF themselves may undergo a conformational change. To test this, we have compared the protease cleavage pattern of CBP and P/CAF bound to full‐length wild‐type HNF‐1α and P447L. After limited V8 protease digestion, we observed that both coactivators were more resistant to the enzyme when bound to P447L than to the wild‐type protein (Figure 6C and D), suggesting that the altered interactions induce a conformational change on the coactivators themselves.

MODY3/HNF‐1α mutants recruit CBP and P/CAF in vivo, but fail to induce histone hyperacetylation

To demonstrate the in vivo relevance of the above findings in the chromatin context of intact cells, we employed chromatin immunoprecipitation (Chip) assays using NIH 3T3 cells harboring genome‐integrated copies of the HNF‐1α‐dependent reporter. The results reported in Figure 7A show that wild‐type HNF‐1α and its mutant derivatives can bind to the promoter embedded in chromatin and specifically recruit both CBP and P/CAF. Between 2‐ and 3‐fold increased amounts of promoter DNA were detected in CBP and P/CAF immunoprecipitates of P519L‐ and P447L‐transfected cells compared with those of wild‐type HNF‐1α‐transfected cells. This suggests that the increased strength of associations observed in biochemical assays could lead to the increased recruitment of coactivators to the promoter. On the other hand, no local hyperacetylation of histones H3 and H4 could be detected in cells transfected with mutant HNF‐1α expression vectors, an effect that was clearly seen in wild‐type HNF‐1α‐transfected cells (Figure 7B). Interestingly, in these latter samples, H4 hyperacetylation was much more pronounced than that of H3, suggesting that selective acetylation of core histone subtypes may be of functional importance. Since targeted histone acetylation is a crucial event in activator‐dependent transcription initiation, we conclude that the lack of activation phenotype of the mutant HNF‐1α proteins is due primarily to their inability to switch on the HAT activity of the recruited coactivators.

Figure 7.

Analysis of the HNF‐1α‐dependent recruitment of CBP and P/CAF to chromatin template in intact cells and the acetylation state of the neighboring nucleosomes. Soluble chromatin from formaldehyde‐fixed NIH 3T3 cells transfected with the indicated plasmids was prepared and immunoprecipitated with (A) α‐HNF‐1α, α‐CBP or α‐P/CAF, (B) anti‐acetyl H3 or anti‐acetyl H4 antibodies. The radioactivities of the 271 bp PCR products were quantitated by phosphoimage analysis. The data are expressed as percentages of the normalized radioactivities obtained with wild‐type HNF‐1α‐transfected samples in (A) and those obtained with vector transfected samples in (B). The corresponding 100% values for the numbers with an asterisk are the HNF‐1α immunoprecipitated from HNF‐1α‐transfected cells in (A) and the α‐acetyl H3 immunoprecipitated from vector‐transfected cells in (B). Errors correspond to deviations obtained from at least two independent experiments. Since the basal levels of acetylated H3 and H4 co‐precipitating DNAs were also observed in non‐transfected cells (data not shown), taking into account the efficiency of transfection (∼5%), the percentage increases observed in wild‐type HNF‐1α‐transfected samples are clearly underestimates.

Discussion

The recent discovery that CBP/p300 and P/CAF possess intrinsic HAT activities (Bannister and Kouzarides, 1996; Ogryzko et al., 1996; Yang et al., 1996) provided new ground to our understanding of the complex molecular mechanisms involved in gene regulation. Histone acetylation by coactivators recruited to the gene regulatory regions leads to a localized remodeling of chromatin, which increases the accessibility of transcription factors to nucleosomal DNA and correlates with transcriptional activation (Grunstein, 1997; Struhl, 1998; Kundu et al., 2000). Understanding the potential pathways that may modulate HAT enzymatic activity is of great importance, since nucleosome acetylation is a crucial step in the process of transcription initiation. To date, two types of mechanism that may affect the HAT activity of coactivators are known. The first is phosphorylation of CBP by cyclin E–Cdk2, which increases its HAT activity (Ait‐Si‐Ali et al., 1998), while the second type involves association with other cofactors, such as E1A or Twist, which may modulate the acetyltransferase activity of coactivators by masking their catalytic domain (Chakravarti et al., 1999; Hamamori et al., 1999).

The main finding of this work is that a transcription factor itself can greatly influence the HAT activity of associated coactivator proteins. We have been driven to this conclusion by investigating the molecular mechanism responsible for the lack of transcriptional activation potential of two naturally occurring mutants of HNF‐1α. We have previously shown that HNF‐1α‐mediated transcription requires the synergistic action of CBP and P/CAF (Soutoglou et al., 2000a). In addition, the HAT activity of both coactivators is important for HNF‐1α‐dependent activation in the in vivo chromatin context (Soutoglou et al., 2000a). On the other hand, as demonstrated in this study, HNF‐1α may also be subject to negative regulation by an associated corepressor complex containing NCoR and HDAC‐1. This possibility is corroborated by the dramatic increase in HNF‐1α‐dependent transcription upon inhibition of HDACs by TSA, as well as the in vitro and in vivo interactions between HNF‐1α, NCoR and HDAC‐1. The observations that TSA treatment disrupts this complex and that the extent of transcriptional activation by this drug could not be induced further by CBP and P/CAF overexpression indicate a potential dynamic exchange between the corepressors and coactivators possibly involved in HNF‐1α‐mediated gene activation. Two lines of evidence suggest that the lack of activation by the inactive HNF‐1α mutants, P519L and P447L, is not due to potential defects in such corepressor–coactivator exchange. First, TSA treatment did not increase transactivation by these mutants, despite dissociating the corepressor complex. Secondly, both mutants interacted with and recruited CBP and P/CAF to a genome‐integrated promoter much more efficiently than wild‐type HNF‐1α. This latter observation is rather intriguing, since the increased recruitment of a coactivator to the promoter is expected to lead to increased transcription rates (Martinez‐Balbas et al., 1998; Xu et al., 1999; Kundu et al., 2000).

This indicates that the efficient recruitment of coactivators into the promoter is not sufficient to induce transcription. Thus, coactivators should not be viewed as constitutively active proteins requiring only proper positioning on the genome, but rather may acquire an active configuration as a result of their interaction with DNA‐binding proteins. In agreement with this speculation is the finding that the docking of PGC‐1 coactivator to PPARγ greatly increases its transcriptional activity (Puigserver et al., 1999). The suggested mechanism involves a protein–protein interaction‐dependent conformational change of PGC‐1 that results in a state permissive for capturing CBP and Src‐1 to generate an active coactivator complex (Puigserver et al., 1999). The above case represents an example of transcription factor‐dependent regulation of coactivator function via the recruitment of additional coactivator proteins. The present study reveals a different type of regulation that involves allosteric modulation of the HAT activity of coactivators as a result of their altered interactions with a DNA‐binding protein. Both CBP and P/CAF acetylate histones efficiently when complexed with wild‐type HNF‐1α. On the other hand, their ability to acetylate histones is completely abolished when they are in a complex with the P519L or P447L mutated forms of HNF‐1α. This could be due to the altered conformation adopted by the mutants, which may bury the catalytic site of the coactivators in such a way that they are unable to recognize histone substrates. On the other hand, both CBP and P/CAF acquire an apparently different conformation when complexed with wild‐type or mutant HNF‐1 molecules, suggesting that HNF‐1 proteins may actively control the conformation required for their HAT activity and consequently their transcriptional activation potential. These two possibilities may not necessarily be mutually exclusive, as the interaction‐induced conformational change in CBP and P/CAF may lead to differential exposure of their catalytic surface to histones.

As demonstrated here, the transcription factor‐mediated switch on and off of the HAT activity of coactivators is not limited to HNF‐1α. Differential up‐regulation of the in vitro HAT activity of CBP by Sp‐1, HNF‐4 and HNF‐1α proteins suggests that this type of regulation may also function with other DNA‐binding factors. Moreover, it has been widely observed that distinct classes of transcription factors have selective requirements for the HAT functions of coactivators. For example, the HAT activity of P/CAF but not that of CBP appears to be important for nuclear hormone receptor‐, MyoD‐ or NF‐κB‐dependent activation (Puri et al., 1997; Korzus et al., 1998; Kurokawa et al., 1998; Sheppard et al., 1999), whereas the HAT activity of CBP is required for CREB or STAT‐1 function (Korzus et al., 1998; Kurokawa et al., 1998). Differential modulation of the individual HAT activities of the coactivator complexes via interactions with the DNA‐binding factors may well contribute to the observed selective requirements, suggesting that this type of regulation could be of general importance.

Materials and methods

Plasmids

The generation of human HNF‐1α mutants by site‐directed mutagenesis has been described (Vaxillaire et al., 1999). The coding regions of P519L and P447L have been amplified from the corresponding pRSV plasmids and subcloned into the EcoRI–XbaI sites of the pCMVmyc vector. To generate histidine‐tagged HNF‐1 bacterial expression plasmids, the above inserts were subcloned into pRSET‐C vector (Invitrogen). To generate pCMVmyc‐HDAC‐1 and pCMVFlag‐HDAC‐1, the coding region of the human HDAC‐1 cDNA from pBs‐HDAC‐1 (Taunton et al., 1996) (a gift from C.Hassig) was subcloned into the EcoRI–XbaI or EcoRI–NheI sites of pCMVmyc or pCMVFlag (Soutoglou et al., 2000a,b) vectors. Gal4–HNF‐1α (370–631) and GST–HNF‐1α (370–631) containing the wild‐type and P447L and P519L mutations were obtained by insertion of the BamHI–EcoRI fragments of the corresponding pRSV plasmids into the same sites of pSG424 (Sadowski et al., 1989) and pGEX‐2TK (Amersham Pharmacia Biotech), respectively. The reporter plasmid 5×Gal4 E1b‐luc (a gift from C.Muchardt and Q.K.Tran) has been constructed by replacing the CAT reporter gene by the luc reporter gene in the pG5E1bCAT construct (Martin et al., 1990). All constructs have been verified by sequencing by automatic laser fluorescence sequencer ABI. pCMV HNF‐4 and −150PK/luc constructs have been described previously (Kennedy et al., 1997; Viollet et al., 1997). pCDNA‐huNCoR (Wang et al., 1998), pCMXFlag‐PCAF (Yang et al., 1996) and pCDNA3‐CBP (Munshi et al., 1998) were gifts from Drs J.Liu, Y.Nakatani and D.Thanos, respectively.

Cell culture and protein–protein interactions

The NIH 3T3 cell line containing genome‐integrated copies of the 3×AlbPE‐CAT (Ktistaki and Talianidis, 1997) reporter has been described previously (Soutoglou et al., 2000a). Cell culture conditions, transfections, cellular extract preparations, reporter activations and immunoprecipitations were performed as described (Ktistaki and Talianidis, 1997; Soutoglou et al., 2000a). The antibodies used in this study were CBP (A22), P/CAF (H‐369), HNF‐1α (C20), NCoR (C20), HDAC‐1 (H11) (Santa‐Cruz Biotechnologies), Flag M2 (Sigma), anti‐acetyl H3 and anti‐acetyl H4 (Upstate Biotechnology).

Recombinant HNF‐1 proteins, expressed as either GST or His6 fusions, were purified on GST–Sepharose (Amersham‐Pharmacia) or Talon metal affinity resin (Clontech), respectively. GST pull‐down and Talon‐based pull‐down assays were performed as described (Soutoglou et al., 2000a,b). Baculovirus expression vectors for CBP and P/CAF were obtained from D.Thanos and P.Chambon, respectively. Recombinant CBP and P/CAF were expressed in Sf9 cells and purified on Talon metal affinity resin (Clontech), according to the manufacturer's instructions.

V8 protease (Sigma) digestions of either 35S‐labeled in vitro translated HNF‐1 proteins or 35S‐labeled CBP and P/CAF proteins absorbed on resin‐bound HNF‐1 proteins were performed in a buffer containing 25 mM HEPES pH 7.9, 40 mM KCl and 10% glycerol. Incubations were performed at 25°C for 5 min and stopped by the addition of SDS loading buffer.

Immunoprecipitation‐HAT assay

For IP‐HAT assays, the cells were treated with 1 μM TSA for 12 h and processed for immunoprecipitation using HNF‐1α antibody as described (Soutoglou et al., 2000b), except that all buffers contained 0.2 mM sodium butyrate to ensure that no traces of active HDACs are present in the precipitates. After washes, half of the beads were used for western blot analysis and the other half were incubated at 30°C for 1 h with 10 μg of core histones (Roche), 0.25 μCi of [3H]acetyl‐CoA in a buffer containing 50 mM HEPES pH 8.0, 10% glycerol, 1 mM dithiothreitol (DTT), 0.1 mM EDTA and 0.2 mM sodium butyrate. Reaction products were analyzed by 12% SDS–PAGE.

Chromatin immunoprecipitations

This assay was performed as described (Orlando et al., 1997), with several modifications. Briefly, transfected cells were treated with 1% formaldehyde for 10 min at room temperature. Cross‐linking was stopped by the addition of glycine to a final concentration of 0.125 M. The cells were washed with cold phosphate‐buffered saline (PBS) and swollen in ice for 10 min in 25 mM HEPES pH 7.8, 1.5 mM MgCl2, 10 mM KCl, 0.1% NP‐40, 1 mM DTT and protease inhibitor cocktail (Roche). Following dounce homogenization (10 strokes, pestle A), the nuclei were collected and resuspended in 'sonication buffer’ containing 50 mM HEPES pH 7.9, 140 mM NaCl, 1 mM EDTA, 1% Triton X‐100, 0.1% sodium deoxycholate, 0.1% SDS and protease inhibitors, and sonicated on ice to an average length of 200–1000 bp. The samples were centrifuged at 14 000 r.p.m. and pre‐cleared with protein G–Sepharose in the presence of 2 μg of sonicated λDNA and 1 mg/ml bovine serum albumin (BSA). Twenty‐five A260 units of the pre‐cleared chromatin were immunoprecipitated with 5 μl of antibodies and the immune complexes were collected by adsorption to protein G–Sepharose. The beads were washed twice with 'sonication buffer’, twice with sonication buffer containing 500 mM NaCl, twice with 20 mM Tris pH 8.0, 1 mM EDTA, 250 mM LiCl, 0.5% NP‐40, 0.5% sodium deoxycholate, and twice with TE buffer. Immune complexes were eluted with 50 mM Tris pH 8.0, 1 mM EDTA and 1% SDS at 65°C for 10 min, adjusted to 200 mM NaCl and incubated at 65°C for 5 h to reverse the cross‐links. After successive treatments with 10 μg/ml RNase A and 20 μg/ml proteinase K, the samples were extracted with phenol–chloroform and precipitated with ethanol. One‐tenth of the immunoprecipitated DNA and input DNA (from extracts corresponding to 0.025 A260 units) were analyzed by PCR using primers from sequences surrounding the chimeric promoter (5′‐CGCCAGGGTTTTCCCAGTCACGAC and 5′‐CTCCATTTTAGCTTCCTTAGCTCCTG). Amplifications (25 cycles) were performed in the presence of 10 μCi of [α‐32P]dCTP and the products were analyzed in 4% polyacrylamide gels. To ensure that the amounts of PCR products accurately reflected the amounts of template DNA, control PCRs (20, 25 and 30 cycles) were performed with decreasing amounts of templates (not shown). The intensities of the bands were quantitated by phosphoimage analysis and their ratios to the inputs were calculated. These values were then used for comparisons between the different samples.

Acknowledgements

We thank P.Chambon, C.Hassig, J.Liu, Y.Nakatani and D.Thanos for the indicated plasmids, E.Ktistaki for advice with the chromatin immunoprecipitation assay, A.Doyen for technical assistance, and P.Hatzis, J.Papamatheakis and G.Thireos for critical reading of the manuscript. This work was supported by the Greek General Secretariat of Science and Technology, the European Union (HPRN‐CT‐2000‐00087) and by ARC and LNFCC.

References

View Abstract