Intracellular interleukin-1alpha functionally interacts with histone acetyltransferase complexes.

Interleukin-1alpha (IL-1alpha) is an inflammatory cytokine acting extracellularly via membrane receptors. Interestingly, a significant portion of synthesized IL-1alpha is not secreted; instead, it is actively translocated into the cell nucleus. IL-1alpha was indeed shown to be involved in certain intracellular processes, such as control of proliferation, apoptosis, or migration, however, the mechanisms of such actions are not known. Here we show that intracellular IL-1alpha fused to the Gal4p DNA-binding domain (Gal4BD) possesses strong transactivation potential that can be boosted by overexpression of the transcriptional coactivator p300. We demonstrate that the IL-1alpha precursor interacts via its N-terminal peptide (IL-1NTP) with histone acetyltransferases p300, PCAF, Gcn5 and with the adaptor component Ada3, and that it integrates into the PCAF.p300 complex in a non-destructive manner. In analogy with known acidic coactivators, yeast strains expressing Gal4BD/IL-1NTP display a toxic phenotype that can be relieved by depletion of various components of the SAGA complex. Our data provide the first solid evidence for the nuclear target of the IL-1alpha precursor and suggest its novel function in transcriptional control.

The constantly growing interleukin-1 (IL-1) 1 family currently consists of eight different protein ligands, and as much as 10 surface or soluble receptors (1,2). The IL-1 system represents an evolutionary conserved signaling mechanism homologous to the Drosophila Toll pathway, being a central mediator of the host innate defense mechanisms (3,4). IL-1 proteins are expressed by most of the cell types, and they act as the first alarming signals inducing release of the full set of proinflammatory molecules such as prostaglandins, tumor necrosis factor ␣, IL-6, chemokines, proteins of the acute phase, and IL-1 per se (5).
IL-1␣ is translated as a 33-kDa precursor, which is proteolytically processed into mature IL-1␣ (IL-1MAT) and the IL-1␣ N-terminal peptide (IL-1NTP) (6). For many years, the striking IL-1NTP sequence homology among different species fostered speculations about its putative biological function. The nuclear localization signal (NLS), present in the IL-1NTP molecule, was shown to be essential for the biological activity of intracellular IL-1␣ (7,8). Indeed, overexpressed intracellular IL-1␣ precursor (pre-IL-1␣), but not IL-1MAT, was able to inhibit cell growth and to induce the expression of the plasminogen activator inhibitor-1 and collagenase genes (9,10). Pre-IL-1␣ was shown to stimulate proliferation of smooth muscle cells (11) and to regulate the migration and the lifespan of endothelial cells (12,13). Thus, various effects of pre-IL-1␣ are found to be dependent on the presence of the acidic IL-1NTP fragment with the predicted pI 5.1. Stevenson et al. (14) have shown that IL-1NTP overexpressed in mouse 3T3 fibroblasts behaves as a nuclear oncoprotein and induces their transformation (14). IL-1NTP was also found to be specifically associated with the novel protein HAX-1 that has been implicated in modulating apoptosis and cytoskeletal functions (15). Recently, a strong proapoptotic effect of ectopically expressed IL-1NTP in numerous human tumor cell lines has been reported (16). Taken together, several lines of evidence point to a role of IL-1NTP in certain yet unknown nuclear processes.
Initiation of transcription in eukaryotes is mediated by recruitment of different multiprotein complexes to the promoter sequence involving the RNA polymerase II complex, general transcription factors and transcriptional cofactors, histone acetytransferase (HAT) complexes, and sequence-specific transcription factors (17,18). The sequence-specific transcription factors are often composed of separate DNA-binding and activation domains (19,20). Depending on the amino acid composition, three types of activation domains are recognized: glutamine-, proline-, and acidic-rich domains. The transcription factors containing the acidic-rich activation domain, such as yeast Gal4p, viral proteins VP16 and E1A, and human c-Jun and NF-B RelA(p65), are evolutionary well conserved, and are able to activate transcription in Saccharomyces cerevisiae as well as in mammalian cells (21)(22)(23)(24)(25). Moreover, the acidic transcription activators interact with different types of HATs and adaptor proteins, and their overexpression in yeast cells is often associated with growth inhibitory phenotypes (26,27).
The eukaryotic chromatin structure with DNA wrapped around histone proteins forming a repeating array of nucleosomes leaves promoter DNA sequences inaccessible to the transcriptional machinery (28). Therefore, the chromatin structure must be remodeled prior to gene targeting (17). Up to date, a number of different factors with chromatin remodeling activities have been identified (29). The fact that histone hyperacety-* 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 U.S.C. Section 1734 solely to indicate this fact. lation is associated with the transcriptionally active chromatin suggests that histone acetylation is one of the chromatin remodeling mechanisms important for transcription initiation (30). The finding that HAT activities are intrinsic to the transcriptional coactivators p300/CBP, PCAF, and Gcn5, interacting with a number of sequence-specific transcription factors, confirmed the functional link between histone acetylation and transcriptional control (31). However, HATs cannot acetylate histones in the chromatin context per se, they can do so only in context of multiprotein HAT complexes where they interact with adaptor proteins facilitating the accessibility of the histone tails to HAT catalytic domains. Two distinct HAT complexes have been characterized so far in yeast, SAGA and ADA (32,33). Mass spectrometric analysis has revealed that the human PCAF complex involves, besides transcriptional coactivator CBP/p300 and others, human counterparts of yeast HAT cofactors Ada2, Ada3, and Spt3 (34). This mega-dalton complex is composed of more than 14 different subunits that seem to be organized in the three functionally divergent modules responsible: (i) for HAT activity, (ii) for interaction with the sequencespecific transcription factors, and (iii) for association with the preinitiation complex containing TATA-binding protein-associated factors and RNA polymerase II (35). Therefore, the HAT complex represents a physical bridge between distant DNAbound activators and the transcriptional machinery at the core promoter (36).
In this study we have investigated functions of the intracellular forms of IL-1␣. We have found that when fused to Gal4BD, IL-1NTP acts as an activator of transcription in both yeast and mammalian cells. The transcriptional activation requires an intact SAGA complex in yeast and the transcriptional coactivator p300 in mammalian cells. We also demonstrate that IL-1NTP physically interacts with mammalian histone acetyltransferases Gcn5, PCAF, and p300, and with adaptor protein Ada3 in vitro. In coimmunoprecipitation experiments we show in detail the specific interactions between intracellular forms of IL-1␣ and different histone acetyltransferases in vivo.
The E1A and HA-tagged human p300 expression plasmids were from A. Hecht (40). Plasmids encoding HA-tagged human p300 mutants were kindly provided by T. Fujita (41). Plasmids expressing N-terminal FLAG-tagged mouse PCAF and GCN5 were a gift from T. Honjo (42) and the expression vector encoding HA-tagged human PCAF was kindly supplied by M. Ott (43). R. Brachmann provided us with the vector expressing the N-terminal FLAG-tagged human Ada3 (44), which was also used as a template for amplification and subcloning of Ada3 into pCMV-HA (Clontech) vector expressing HA-tagged human Ada3 protein. All amplifications were done using Pfu DNA polymerase (Stratagene) and products were verified by direct sequencing.
Cell Culture and Luciferase Assay-HEK293 cells from the American Type Culture Collection were grown in high-glucose Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. Twentyfour hours before transfection, cells were plated in 24-well plates so that they were 80% confluent the next day. The cells were cotransfected with 0.6 g of total amount of expression plasmid DNA mixture and pFRLuc Gal4 reporter construct (Stratagene) per well using the Superfect transfection reagent (Qiagen) according to the manufacturer's instructions. Empty vectors were used to adjust the amounts of DNA in each experiment. After 48 h cells were harvested and lysates were assayed for luciferase activity using the dual luciferase reporter assay (Promega) and the DLR 96-well plate luminometer (STRATEC Biotechnology Systems AG, Birkenfeld, Germany). Alternatively, the measured luciferase activities were normalized according to total protein concentrations. All experiments were performed in triplicates.
Preparation of GST Fusion Proteins and Pull-down Assay-Purifications of recombinant proteins as well as pull-down assays were described previously (45). Briefly, GST fusion proteins (0.5 g) bound to glutathione-agarose beads were incubated with 300 l of whole cell lysates (see next chapter) from HA-tagged p300, PCAF, and Ada3 or FLAG-Gcn5-transfected HEK293 cells in 700 l of immunoprecipitated (IP) lysis buffer at 4°C for 4 h. After three washes (10 min at 4°C) with IP lysis buffer, the proteins bound to the beads were resolved by Tricine-SDS-PAGE (8% gel), transferred onto nitrocellulose membranes (Bio-Rad), and probed with anti-FLAG (M2; Sigma), anti-HA (sc-805), or anti-RelB (sc-226; Santa Cruz Biotechnology) antibodies.
Transient Transfection and Immunoprecipitation-HEK293 cells were transfected in 10-cm Petri dishes using 10 g of total plasmid DNA as described before. Forty-eight hours later the cells were washed with phosphate-buffered saline and lysed in IP lysis buffer (50 mM Tris-HCl, pH 7.6, 150 mM NaCl, 1.5 mM MgCl 2 , 1 mM EDTA, 0.5% Nonidet P-40, 10% glycerol, Protease and Phosphatase Inhibitor Cocktails; Sigma) for 30 min on ice. After centrifugation, the supernatants were stored and the sediments containing nuclei and cell debris were extracted with buffer B (20 mM HEPES, pH 7.9, 420 mM NaCl, 1.5 mM MgCl 2 , 1 mM EDTA, 1 mM dithiothreitol, protease and phosphatase inhibitor mixtures; Sigma) for 15 min on ice. The supernatants from both extractions were mixed. The resulting whole cell extracts were precleared with 50 l of protein G-agarose for 1 h at 4°C. After centrifugation, the cleared lysates were incubated with 50 l of protein G-agarose and 1 g of anti-FLAG (Santa Cruz Biotechnology) antibody in 700 l of IP lysis buffer overnight at 4°C. The beads were then washed three times with IP lysis buffer and subjected to Western blot detection of bound material as described before.
␤-Galactosidase Assay-Yeast strains harboring UAS GAL1 -lacZ reporter constructs were grown to late logarithmic phase, harvested by centrifugation, and resuspended in 200 l of breaking buffer (100 mM potassium phosphate, pH 7.8, 0.2% Triton X-100, 0.5 mM dithiothreitol, and protease inhibitor mixture; Sigma). Cell extracts were prepared by vortexing with 0.5-mm glass beads four times for 30 s. Activities were determined from at least three independent clones using the luminescent ␤-galactosidase detection kit Galacto-Light (Tropix, Bedford, MA).
The measured activities were normalized according to optical densities of the used yeast cultures.

The Gal4BD/IL-1NTP Fusion Protein Transactivates the Gal4 Promoter and Confers Growth Inhibition in Yeast-Ex-
periments with the yeast two-hybrid system aiming to identify proteins interacting with IL-1NTP surprisingly revealed that expression of either mouse or human Gal4BD/IL-1NTP fusion protein enabled the growth of the reporter yeast strain on selection drop-out media lacking adenine, histidine and tryptophan (SD-AHT). Fig. 1A shows that the yeast strains expressing the mouse or human Gal4BD/IL-1NTP fusion proteins alone grow by a rate similar to that of the positive control strain expressing the interacting fusion proteins Gal4BD/p53 and Gal4AD/SV40 T large antigen, and when incubated with 5-bromo-4-chloro-3-indolyl ␤-D-galactopyranoside (X-␣-gal) substrate, they turn blue (data not shown) indicating the presence of ␤-galactosidase activity. In contrast, strains expressing only Gal4BD, or mouse or human Gal4BD/IL-1MAT were unable to grow on SD-AHT even after 5 days of cultivation at 30°C, and even when inoculated at high densities. Thus, in accordance with their high degree of homology, both mouse and human Gal4BD/IL-1NTP chimeras effectively transactivate the GAL1 UAS -GAL1 TATA -HIS3, GAL2 UAS -GAL2 TATA -ADE2, and MEL1 UAS -MEL1 TATA -lacZ reporters harbored by the AH109 strain.
Studies in the eukaryote S. cerevisiae have shown that expression of acidic transactivation proteins fused to Gal4BD can inhibit yeast growth (24,25,27). To test the effect of IL-1NTP on yeast growth, and to further investigate the mechanism of IL-1NTP-mediated activation, strain YMH171 and the indicated isogenic derivatives were transformed with the mouse IL-1NTP/pAS2-1 vector expressing higher levels of Gal4BD/IL-1NTP. As shown in Fig. 1B, strain YMH171 (wt) expressing Gal4BD/IL-1NTP grows markedly slower than the strain expressing Gal4BD only (control). However, the Gal4BD/IL-1NTP-induced growth inhibition was strictly dependent on the presence of the intact SAGA transcriptional activation complex. The growth inhibitory effect of IL-1NTP was completely relieved in yeast strains lacking one of the SAGA complex components: Gcn5, Ada2, Ada3, and Spt7. Surprisingly, deletion of the AHC1 gene product, a unique structural subunit of the yeast ADA transcriptional activation complex, did not relieve the Gal4BD/IL-1NTP-mediated growth inhibition. These data suggested specific interference of Gal4BD/IL-1NTP with certain components of the yeast SAGA complex as it has been described for the acidic coactivators Gal4p, E1A, or VP16 (26,27,46). Similar results were obtained when identical yeast strains harboring the UAS GAL1 -TATA GAL1 -lacZ reporter and expressing Gal4BD, mouse Gal4BD/IL-1NTP, or the strong Gal4BD/ VP16 activator were assayed for their respective ␤-galactosidase activities (Fig. 2B). The IL-1NTP-mediated ␤-galactosidase activities reached about 25% of activities stimulated by Gal4BD/VP16 in wild type strain YMH171 (Fig. 2B, wt). The gcn5⌬, ada2⌬, and ada3⌬ deletions severely reduced the activities of both IL-1NTP and VP16, an effect more pronounced for IL-1NTP than for VP16 (5-versus 4-fold reduction, respectively). The most severe reduction (20-fold for both IL-1NTP and VP16) was observed in the strain defective in expression of Spt7, the critical structural subunit of the SAGA complex. In agreement with previous results, ahc1⌬ deletion had no impact on activities mediated by both VP16 (37) and IL-1NTP, suggesting that IL-1NTP acts via interaction with the SAGA, but not with the ADA transcriptional complex. Furthermore, the point mutation in the TATA core promoter element (UAS GAL1 -TGTA GAL1 -lacZ) abolished the transactivation potentials of both IL-1NTP and VP16 proteins ( Fig. 2A), indicating that the IL-1NTP-mediated activation occurs specifically through a common RNA polymerase II activation mechanism.
Characterization of Domains in the IL-1NTP Molecule Responsible for the Transactivation Potential-Besides the well characterized NLS domain (aa 79 -86) (7), our structure-prediction analyses identified two additional highly conserved acidic ␣-helixes located at the N and C termini of the IL-1NTP molecule (aa 7-19 and 98 -108, respectively). In a search for the structural motifs mediating the IL-1NTP transactivation potential, a set of deletion mutants fused to Gal4BD were prepared (Fig. 3A) and tested for their activities in the yeast reporter strain. The yeast strain AH109 was transformed with plasmid constructs encoding different Gal4BD/IL-1NTP deletion mutants. As shown in Fig. 3B, deletion of the NLS domain did not affect the transactivation potential of the Gal4BD/IL-1NTP, whereas deletions of either N-or C-terminal acidic helixes markedly slowed the growth rate. Moreover, the combination of both N-and C-terminal deletions had deteriorating FIG. 1. The Gal4BD/IL-1NTP transactivates the Gal4 promoter and confers growth inhibition in yeast. A, yeast strains AH109 expressing Gal4BD/p53ϩGal4AD/SV40 T-Ag (p53ϩT), Gal4BD (Con), mouse Gal4BD/IL-1NTP (mIL-1NTP), human Gal4BD/IL-1NTP (hIL-1NTP), mouse Gal4BD/IL-1␣ mature protein (mIL-1mat) or human Gal4BD/IL-1␣ mature protein (hIL-1mat) were serially diluted (factor 10) and dropped onto synthetic selection medium (SD-AHT). The plates were incubated for 5 days at 30°C and recorded. B, the SAGA transcriptional activation complex is required for growth inhibition by IL-1NTP. The wild type (wt, YMH171) or the indicated mutant strains were transformed with vector expressing Gal4BD alone (Control) or with vector expressing the mouse Gal4BD/IL-1NTP fusion protein (mIL-1NTP). The plates were incubated as before.
impact on the growth potential on selection media. Analogously, the ␤-galactosidase activities in the Gal4BD/IL-1NTP expressing AH109 yeast strain were comparable with those in the positive control strain (expressing Gal4BD/p53 and Gal4AD/T-Ag together), and deletions of either one or both terminal helixes in the IL-1NTP molecule dramatically reduced ␤-galactosidase activities in the respective strains. Interestingly, the internal deletion mutants lacking the NLS domain (ATSSN and TSETS) showed enhanced activities that were as much as 2-fold higher than activities observed in the positive control strain (Fig. 3C). The expression levels of the corresponding Gal4BD fusion proteins in strains tested for growth on drop-out medium (Fig. 3B) and for ␤-galactosidase activities (Fig. 3C) are shown in Fig. 3D. Lower expression levels of Gal4BD/ATSSN and Gal4BD/TSETS reflect their higher transcriptional activities (Fig. 3C) and higher toxicities (data not shown), and further support the idea about the functional importance of both the IL-1NTP terminal helixes. These data indicate that the NLS domain is not required for the transactivation potential of Gal4BD/IL-1NTP, and that its deletion results even in enhanced transcriptional activity of the chimera.
Gal4BD/IL-1NTP Transactivates the Gal4 Promoter in Mammalian Cells via Functional Cooperation and Physical Association with the Transcriptional Coactivator p300 -Our data from the yeast model are consistent with those of other acidic transcriptional coactivators and support our hypothesis about a novel nuclear function of IL-1␣. To determine the physiological role of the intracellular IL-1␣, we further tested its transactivation potential as well as its ability to cooperate with different histone acetyltransferases in mammalian cells. HEK293 cells cotransfected with a reporter plasmid and plasmid constructs encoding the indicated Gal4BD fusion proteins, and/or the indicated cofactors were assayed for luciferase activities. As shown in Fig. 4A, the expression of Gal4BD/IL-1NTP alone increased the luciferase activity in a dose-dependent manner. However, when Gal4BD/IL-1NTP was expressed together with p300, the luciferase activity reached 55% of the value stimulated by the Gal4BD/p65 NF-B activation domain fusion protein. Furthermore, it could be antagonized by coexpression of the adenoviral protein E1A that binds specifically p300, thereby demonstrating the specificity of cooperation between p300 and IL-1NTP.
In analogy to experiments with yeast, we wanted to map critical domains in the IL-1NTP molecule responsible for functional cooperation with the transcriptional coactivator p300. When the Gal4BD/IL-1NTP deletion mutants were used for cotransfections of the HEK293 cells together with p300, the resulting luciferase activities (Fig. 4B) were in analogy to the ␤-galactosidase activities observed in yeast (Fig. 3C). Both the NLS-deleted mutants (ATSSN and TSETS) showed markedly enhanced activities when cotransfected alone or together with p300 (Fig. 4B). One or double IL-1NTP terminal helix deletions showed dramatic reduction of the reporter activation, suggesting that both terminal helixes are required for an efficient cooperation with p300. The residual activity of IL-1NTP terminal mutants indicates that it might cooperate also with some other transcriptional coactivators. These data indicate that IL-1NTP uses an evolutionary conserved mechanism for the interaction with the transcription activation machinery.
To test the dependence of the Gal4BD/IL-1NTP transactivation potential on known functional domains of p300, cells cotransfected with a reporter plasmid, Gal4BD/IL-1NTP expression plasmid, and with the set of constructs encoding the indicated p300 mutants were assayed for their luciferase activities. As illustrated in Fig. 5A, the p300 deletion mutants lacking N-terminal domains between amino acids 142 and 1241 were fully functional in transactivation of the Gal4 promoter (columns 3 and 4), as well as the 1-1946 mutant lacking the C-terminal part including the Gln-rich domain (column 6), which was shown to be an interface mediating the association with several transcription factors including IRF-3 (41). Also the functional deletion of the CH3 domain in the ⌬143-957⌬Zn1/2 mutant did not significantly abrogate the cooperation (column FIG. 2. The SAGA transcriptional activation complex and the TATA promoter element are required for the transactivation by IL-1NTP. A, yeast strains harboring reporter plasmids carrying either the TATA core promoter element (TA/G1Z) or its TGTA derivative (TG/G1Z) were transformed with the vectors expressing Gal4BD (Con), mouse Gal4BD/IL-1NTP (IL-1NTP), or strong transcriptional activator Gal4BD/VP16 (VP16) and assayed for ␤-galactosidase activities. B, the wild type (wt, YMH171) or the indicated mutant strains carrying the UAS GAL1 -TATA GAL1 -lacZ reporter were transformed with vector expressing only Gal4BD (Con), with vectors expressing mouse Gal4BD/ IL-1NTP (IL-1NTP), or Gal4BD/VP16 (VP16) fusion proteins. ␤-Galactosidase activities were measured in triplicate from three independent transformants as described under "Experimental Procedures." 11). Interestingly, even the ⌬194 -1572 mutant (column 5) lacking a large portion of the molecule including the HAT domain was able to effectively cooperate with Gal4BD/IL-1NTP. However, two different mutants containing enzymatically inactive HAT domains did not enhance the transactivation potential of Gal4BD/IL-1NTP (columns 9 and 10). The significance of the HAT domain is further stressed by the data obtained with the minimal (six) amino acid mutant ⌬143-957MutAT2 with crippled HAT activity (47), whose counterpart with the wild type HAT domain was highly active (Fig. 5A,  columns 3 versus 10). The enhanced activity of the ⌬143-957 mutant over full-length p300 was probably due to a higher expression level (Fig. 5B). To our surprise, the 1235-2412 mutant containing functional HAT as well as the Gln-rich domains was not active until the C-terminal part of the molecule was removed, which led to an extremely high activity (Fig. 5A,  columns 7 versus 8); yet, the expression levels of both chimeras were similar (Fig. 5B, lanes 7 and 8).
We further addressed the question whether IL-1NTP can physically interact with the transcriptional coactivator p300. Therefore, we performed a set of GST pull-down experiments where beads with immobilized recombinant GST/IL-1NTP protein were incubated with input cell lysates expressing the indicated HA-tagged p300 and its mutants used in the luciferase assay (Fig. 5C, upper panel). As expected, the GST/IL-1NTP chimera bound full-length p300 as well as all the mutants that enhanced transactivation of the Gal4 promoter (Fig.  5C, center panel, lanes 2-5, 8, and 11). It also bound both mutants with inactive HAT domains, suggesting that HAT activity is not required for association with IL-1NTP (lanes 9 and 10). However, a strong interaction was also detected for 1235-2412 that did not cooperate in the luciferase assay (lane 7), whereas no interaction could be detected for 1-1946 (lane 6) that efficiently cooperated with the IL-1NTP in luciferase assay. The Gln-rich domain is therefore required for the in vitro interaction with IL-1NTP, as was reported for IRF-3 (41). The seeming discrepancies between the data from luciferase assays and pull-down experiments are probably because of the complexity of molecular interactions of the components of the p300⅐PCAF complex. These data also point toward a redundancy of histone acetyltransferases in the preinitiation complex and suggest a cooperation of Gal4BD/IL-1NTP with other additional cofactors.
The Intracellular IL-1␣ Cooperates and Physically Interacts with Different Subunits of the p300⅐PCAF Complex-To assess a possible contribution of other histone acetyltransferases ho-

FIG. 3. The transactivation potential of IL-1NTP is mediated by both N-and C-terminal acidic helixes.
A, scheme of mouse IL-1NTP deletion mutants illustrating the positions of NLS (aa 79 -86), and N-(aa 7-19) and C-terminal (aa 98 -108) acidic ␣-helixes. B, yeast strain AH109 was transformed with constructs encoding for Gal4BD fusion proteins with the indicated IL-1NTP deletion mutants; the drop test was performed as described in the legend to Fig. 1. C, the same strains were assayed for ␤-galactosidase activities as described before. D, Western blot showing the expression levels of Gal4BD fusion proteins in the yeast strains used in B.
mologous to the yeast SAGA components, we studied the effects of PCAF and Gcn5. As shown in Fig. 6A, cotransfection of Gal4BD/IL-1NTP together with mouse acetyltransferase Gcn5 expression plasmid resulted in a significant increase over the Gal4BD/IL-1NTP induced activity, whereas the combination with mouse PCAF acetyltransferase had no stimulatory effect. Similar results were obtained in mouse 3T3 fibroblasts (data not shown). To further unravel the molecular mechanism responsible for the nuclear function of IL-1␣, we have investigated physical interactions between IL-1NTP and various components of the PCAF complex. HEK293 cells were transfected with expression plasmids encoding FLAG-tagged Gcn5, and HA-tagged PCAF, Ada3, and p300. Corresponding cell lysates were incubated with an immobilized recombinant control (GST), or with the mouse and human GST/IL-1NTP or the mature IL-1␣ (GST/IL-1MAT) protein chimeras. As demonstrated in Fig. 6B, none of the proteins tested showed any detectable interaction with the GST control or with GST/IL-1MAT, but considerable amounts of all the transcriptional cofactors tested were associated with mouse and human GST/IL-1NTP. Interestingly, the weakest binding with GST/IL-1NTP proteins was detected in the case of p300, which was found to be a critical cofactor indispensable for transcription activity of the Gal4BD/IL-1NTP chimera. Both GST/IL-1NTP proteins interacted with equal affinities with Gcn5, PCAF, and Ada3. However, because in the cell lysates also endogenous histone acetyltransferases are present, we cannot distinguish between direct interaction of GST/IL-1NTP with p300, PCAF, and Ada3, and indirect interaction mediated through binding with Gcn5, which could associate with the mentioned cofactors (48). As a negative control, an identical pull-down experiment was performed and probed with anti-RelB antibody. In this case, no specific interaction between GST/IL-1␣ proteins and the RelB transcription factor could be detected.
The Intracellular IL-1␣ Associates with Subunits of the p300⅐PCAF Complex in Vivo-Previous experiments indicated that intracellular IL-1␣ could be involved in some aspects of transcriptional control, participating probably in a formation and/or a function of the transcriptional initiation complex, such as SAGA or PCAF. To confirm these results in vivo with native IL-1␣ proteins, we performed a series of additional coimmunoprecipitation experiments. Lysates from HEK293 cells coexpressing FLAG-tagged mouse or human IL-1␣ proteins, including IL-1NTP, IL-1MAT, or pre-IL-1␣, together with HA-tagged cofactors p300, PCAF, or Ada3, were immunoprecipitated with an anti-FLAG antibody. Consistent with the data from the pull-down assays, all the transcriptional cofactors tested interacted exclusively with forms containing the N-terminal domain, i.e. pre-IL-1␣ and IL-1NTP, no matter whether mouse or human (Fig. 7A). No specific interaction was observed in samples transfected with mouse or human IL-1MAT. These data provide clear evidence that nuclearly localized IL-1␣ associates via its N-terminal domain with the SAGA/PCAF transcriptional activation complex and suggest its participation in transcriptional control.
To confirm that the association interface of the IL-1NTP molecule consisting of two terminal ␣-helixes mediates the critical molecular interactions also in vivo, corresponding mutants of human IL-1NTP lacking the N-terminal helix (EDSSS), the C-terminal helix (ITDDD), or the NLS domain (VVATN) were tested for association with p300, PCAF, and Ada3. As shown in Fig. 7B, removal of either the N-or Cterminal helixes virtually abolished the association of IL-1NTP with p300, whereas the ⌬NLS mutant associated with p300 in a similar manner as full-length IL-1NTP. Interestingly, all the mutants tested associated efficiently with PCAF as well as with Ada3, indicating that IL-1NTP interacts through terminal helixes exclusively with p300. The interaction with PCAF and Ada3 is mediated therefore via different domains of IL-1NTP. The slightly weaker interaction of EDSSS and ITDDD with PCAF is probably because of lower expression levels (Fig. 7B,  bottom panel).
IL-1␣ Integrates Non-destructively into the p300⅐PCAF Complex-To clarify whether IL-1␣ is an integral constituent of the complex, or whether it competes with p300 or PCAF for the same binding site displacing them from the complex we performed further coimmunoprecipitation experiments. HEK293 cells were cotransfected with vectors encoding FLAG-tagged mouse pre-IL-1␣ together with either p300 or PCAF expression vectors, or with all the vectors together. Whereas pre-IL-1␣ interacted separately with p300 or PCAF (Fig. 8, lanes 8 and  10), combined coexpression of all the factors did not result in any displacement of either p300 or PCAF, as both proteins FIG. 4. Gal4BD/IL-1NTP transactivates the Gal4 promoter also in mammalian cells via cooperation with transcriptional coactivator p300. A, HEK293 cells were cotransfected with Gal4-inducible reporter plasmid together with expression plasmids encoding mouse Gal4BD/IL-1NTP, human p300, viral p300 inhibitor E1A, and/or Gal4BD/RelA (RelA/p65), and assayed for luciferase activities. B, the IL-1NTP deletion mutants lacking the terminal acidic helixes are unable to efficiently cooperate with p300. Empty vector (Vec) or the indicated mutants (as illustrated in Fig. 3A) fused to the Gal4BD were expressed with (p300) or without (Con) p300 transcriptional cofactor and luciferase activities were measured as described under "Experimental Procedures." Error bars represent S.E. from three independent experiments.
FIG. 5. The Gln-rich domain of p300 is required for the association with IL-1NTP but not for the transactivation of the Gal4 promoter. A, mapping of the p300 domains required for the cooperation with IL-1NTP. Functional domains of wild type p300 (top) and its mutants are illustrated on the left. HEK293 cells cotransfected with the Gal4 reporter plasmid together with expression plasmids encoding for mouse Gal4BD/IL-1NTP and the indicated p300 mutants were assayed for luciferase activities (right) as before. B, Western blot showing expression levels of HA-tagged p300 and its mutants in cell lysates used for luciferase assays (A). C, in vitro interactions of IL-1NTP with p300 and its mutants. Whole cell lysates from HEK293 cells transfected with empty vector (Con) or with vectors expressing the indicated HA-tagged p300 proteins were incubated with glutathione beads containing mouse recombinant GST/IL-1NTP (GST/mIL-1NTP) protein. Bound material was immunoblotted with anti-HA antibodies (GST pull-down). For the expression controls 5% of the input lysates from transfected cells were loaded onto the gel and immunoblotted (Input). An asterisk (*) indicates a nonspecific band. The bottom panel shows an immunoblot of pull-down samples using anti-GST antibody, demonstrating equal loading of the recombinant proteins used.

Nuclear Function of Intracellular IL-1␣ 4023
were detectable in a single complex (lane 12). In this context it is of interest that viral interferon regulatory factor-1, which inhibits p300, actively displaced the interaction of p300 with ectopically expressed PCAF in a similar experiment (49). Indeed, our data shows that the IL-1␣ precursor integrates into the p300⅐PCAF histone acetyltransferase complex in a nondestructive manner.

DISCUSSION
Our data provide the first solid evidence that the evolutionary conserved IL-1NTP is not only a by-product of proteolytic maturation of IL-1␣, but might indeed convey long sought nuclear functions of intracellular forms of IL-1␣. Here we show that IL-1NTP, when fused to the Gal4 DNA binding domain, behaves as a potent activator of transcription in mammalian cells as well as in yeast. Such behavior has been frequently reported for numerous prooncogenic viral factors, and is therefore in the case of IL-1␣ somewhat surprising. There are basically two possible explanations for this finding: first, IL-1NTP may coincidentally carry sequence similarities resembling common transactivation domains, or second, the transactivation potential is either part or consequence of a yet unknown nuclear function of IL-1NTP. The first possibility is highly unlikely because no sequence similarities could be detected. Although the subversion of the transcriptional complex by numerous viral factors is well known, no innate protection mechanism against these factors has been described so far. Taking into account that the IL-1␣ is rapidly synthesized and actively translocated to the nucleus upon different proinflammatory stimuli, a novel function linked with the transcriptional control and/or protection against viral factors attacking the p300 transcriptional coactivator is feasible. From this point of view the second option is clearly more appealing. We hypothesize that rather than regulating transcription by binding to DNA or sequence-specific transcription factors, intracellular IL-1␣ acting via its NTP domain could represent evolutionary conserved endogenous protection against viral factors inhibiting transcription of antiviral and proinflammatory genes. Such cellular protectors should be swiftly synthesized in response to various inflammatory stimuli and translocated immediately into the nucleus to neutralize viral factors. Indeed, IL-1␣ appears to be a molecule exhibiting all these features.
Previous studies with the strong herpes simplex virus transcriptional activator VP16 have shown that the Gal4BD/VP16 chimera inhibit the growth of transformed yeast strains, but fail to do so in strains with defective expression of components of the SAGA transcriptional activation complex (37). Overexpression of Gal4BD/IL-1NTP in yeast yielded a similar growth inhibitory effect that was also dependent on the presence of an intact SAGA complex. Deletion of adaptor (Ada2, Ada3), enzymatic (Gcn5), or structural (Spt7) subunits attenuated or even abolished the IL-1NTP-dependent growth inhibition, whereas deletion of the critical structural subunit (Ahc1) of the distinct ADA transcriptional activation complex had no effect. The same results were obtained when the transactivation efficiency of IL-1NTP was tested in mutated strains. These data clearly demonstrate that IL-1NTP specifically interferes with and requires the SAGA complex for its function.
To test the transactivation potential of Gal4BD/IL-1NTP in mammalian cells, similar experiments were performed by transient transfections of HEK293 cells with Gal4-responsive luciferase reporter constructs. As expected on the basis of high homologies in transcriptional complexes between yeast and mammals, the Gal4BD/IL-1NTP chimera showed a high transactivation potential in mammalian cells as well. Moreover, cotransfection of transcriptional coactivator p300 dramatically enhanced the effect of Gal4BD/IL-1NTP on the reporter activ- ity. Because the yeast counterpart of mammalian p300 is not known yet, we tested the effects of mouse PCAF and Gcn5, whose yeast homologues are known constituents of the yeast SAGA complex. However, whereas Gcn5 was able to cooperate with Gal4BD/IL-1NTP-induced reporter transactivation, the PCAF acetyltransferase slightly inhibited the basal transactivation mediated by Gal4BD/IL-1NTP itself. Because PCAF can associate with IL-1NTP both in vitro and in vivo, this effect might be mediated by a direct acetylation of IL-1NTP, as was recently reported for IRF-7 (50).
Modeling in silico followed by an extensive mutation analysis allowed the functional identification of acidic terminal ␣-helixes both of which are required for the function of Gal4BD/IL-1NTP. Interestingly, these terminal regions represent the most conserved parts of the molecule, suggesting that they form domains mediating the biological function. The deletion of the third conserved domain known as NLS, which is believed to be responsible for the nuclear localization of the molecules containing IL-1NTP (7, 51), had no obvious inhibitory effect on the promoter transactivation. However, for the effects observed, the NLS domain of IL-1NTP might not be required, as another NLS domain is present in the Gal4BD part of the fusion protein. Interestingly, in all experiments both NLS deletion mutants of IL-1NTP (ATSSN and TSETS) showed significantly higher activities, which were more pronounced for the larger deletion part (TSETS). These data indicate that the NLS domain does not mediate the interaction with the transcriptional initiation complex. Furthermore, it might be possible that bringing the acidic terminal helixes into a bilateral proximity might change the protein conformation and thereby facilitate the interaction with the SAGA⅐PCAF transcriptional activation complex.
To provide deeper insights into the functional cooperation between IL-1NTP and different subunits of the SAGA complex, interactions with certain HATs using immobilized recombinant GST fusion IL-1␣ proteins were investigated. To demonstrate an evolutionary conserved homology among IL-1␣ proteins, especially in their N-terminal domains, we used both mouse and human IL-1NTPs. Their mature counterparts (IL-1MAT) were used to distinguish which part of the protein is responsible for the binding. As expected from the luciferase assay data, showing that IL-1NTP cooperated most efficiently with the transcriptional coactivator p300, GST/IL-1NTP fusion protein bound specifically to ectopically expressed p300 in vitro. Surprisingly, GST/IL-1NTP bound with high affinities to histone acetyltransferases PCAF and Gcn5, although no significant cooperation with PCAF and only a partial one with Gcn5 were observed in the transactivation experiments based on the Gal4 promoter system in HEK293 cells. Interestingly, GST/IL-1NTP interacted also with the human Ada3 protein, which is an important adaptor component of the SAGA⅐PCAF complex. This redundancy in interactions with multiple HATs and Ada3 could be explained by the existence of a common interface in these enzymes or by the possibility that also endogenous HATs FIG. 8. IL-1␣ precursor protein integrates non-destructively into the p300⅐PCAF protein complex. Whole cell lysates from HEK293 cells cotransfected with plasmids expressing HA-tagged p300, and/or PCAF, together with FLAG-tagged IL-1␣ mouse precursor (IL-1) were IP with anti-FLAG monoclonal antibody as before. Bound material was immunoblotted with anti-HA antibodies. For expression controls, 5% of the input lysates were loaded onto the gel. The corresponding bands of p300 and PCAF are marked with arrows; an asterisk (*) indicates a nonspecific band.

FIG. 7.
In vivo interactions of intracellular IL-1␣ with subunits of the histone acetyltransferase complexes are mediated by both terminal helixes of IL-1NTP. A, mouse and human IL-1NTP and IL-1␣ precursor proteins specifically associate with the histone acetyltransferases p300 and PCAF, and with the transcriptional adaptor Ada3. Whole cell lysates from HEK293 cells cotransfected with plasmids expressing HA-tagged p300, PCAF, or Ada3 together with empty vector (Vec) or with the indicated FLAG-tagged IL-1␣ proteins were immunoprecipitated with anti-FLAG monoclonal antibody, and bound material was immunoblotted with anti-HA antibody. For expression controls, 5% of the input lysates (Input) from transfected (Trans) and non-transfected (Con) cells were loaded onto the gel. The immunoblot with anti-FLAG antibody shows expression levels and corresponding mobilities of IL-1␣ proteins used. An asterisk (*) indicates a nonspecific band. The blots are representatives of at least three independent experiments. B, both terminal ␣-helixes of IL-1NTP are required for the association with p300, but not for the association with PCAF and Ada3. Whole cell lysates from HEK293 cells cotransfected with plasmids expressing HA-tagged p300, PCAF, or Ada3 together with empty vector (Vec), FLAG-tagged wild type human IL-1NTP, ⌬NLS mutant (VVATN), N-terminal helix deletion mutant (EDSSS), or C-terminal helix deletion mutant (ITDDD) were immunoprecipitated with anti-FLAG monoclonal antibody and bound material was immunoblotted with anti-HA antibody. For expression controls, 5% of the input lysates (Input) from transfected (Trans) and non-transfected (Con) cells were loaded onto the gel. The immunoblot with anti-FLAG antibody shows expression levels and corresponding mobilities of the IL-1␣ proteins used. An asterisk (*) indicates a nonspecific band. and/or adaptor proteins are bound in the complex and so detected together.
The experiments with the p300 mutants revealed that IL-1NTP binds to p300 through the Gln-rich domain as was reported for IRF-3 (41). Although the functional cooperation of these mutants in the luciferase assays suggests intricate functional interactions in the transcription complex, such activity was strictly dependent on the presence of a functional HAT domain. This is most evident for the molecule lacking the large N-terminal part of p300 (⌬143-957) that proved to be very active, but which completely lost its activity upon blocking its HAT activity by mutation of six amino acids in the HAT domain (⌬143-957MutAT2) (47). Interestingly, removal of the HAT domain (⌬194 -1572) restored the transcriptional activity, suggesting a functional replacement by some other HAT protein.
The most surprising observation is the strong activity of the p300 1235-2221 protein that differs from the inactive mutant 1235-2412 only by the lack of the C-terminal domain. We propose that this domain either binds an inhibitory component or that it leads to a misfolding of the mutant thereby blocking the binding to the transcriptional complex.
To confirm interactions of the intracellular forms of IL-1␣ in their native state, we performed coimmunoprecipitation experiments with transiently transfected cells. The results obtained in these experiments were completely in line with the data from pull-down experiments, further confirming associations of both human and mouse intracellular forms of IL-1␣ specifically with p300 and PCAF, but also with the adaptor protein Ada3. Moreover, using mutants with deleted terminal helixes we confirmed that they act as an interface for the association specifically with p300 in vivo, whereas they are not required for the interaction with PCAF or Ada3. This means that IL-1NTP binds through multiple domains to different subunits of the p300⅐PCAF complex, and therefore, it can integrate non-destructively into the p300⅐PCAF complex without affecting the molecular interactions inside the native complex.
Taken together, the data published in this paper show that intracellular IL-1␣ associates via IL-1NTP with different HATs in vitro, and functionally cooperates with the SAGA transcriptional complex in vivo, as demonstrated in yeast. Furthermore, IL-1NTP fused to Gal4BD transactivates the Gal4 promoter also in mammalian cells and functionally cooperates with the transcriptional cofactor p300 known as the limiting factor of the transcriptional complex. Both intracellular forms of IL-1␣, IL-1NTP, and the IL-1␣ precursor, associate in their native states with histone acetyltransferases p300 and PCAF, and also with the transcriptional adaptor protein Ada3 in vivo. Moreover, IL-1␣ seems to be an integral part of the p300⅐PCAF complex, as its overexpression does not lead to disintegration of the complex.
Our findings provide the first solid evidence for an evolutionary conserved novel function of IL-1␣ in transcriptional control, specifically by its functional cooperation with the SAGA⅐PCAF acetyltransferase complex. This interaction sheds new light on the functional role of a well known cytokine, implying its novel role in intracellular defense mechanisms and opening a new area of cytokine research.