Journal of Molecular Biology
Volume 384, Issue 2, 12 December 2008, Pages 335-348
Journal home page for Journal of Molecular Biology

Assembling the Human IFN-β Enhanceosome in Solution

https://doi.org/10.1016/j.jmb.2008.09.015Get rights and content

Abstract

Assembly of interferon-β enhanceosome from its individual protein components and of enhancer DNA has been studied in solution using a combination of fluorescence anisotropy, microcalorimetry, and CD titration. It was shown that the enhancer binds only one full-length phosphomimetic IRF-3 dimer at the PRDIII–PRDI sites, and this binding does not exhibit cooperativity with binding of the ATF-2/c-Jun bZIP (leucine zipper dimer with basic DNA recognition segments) heterodimer at the PRDIV site. The orientation of the bZIP pair is, therefore, not determined by the presence of the IRF-3 dimer, but is predetermined by the asymmetry of the PRDIV site. In contrast, bound IRF-3 dimer interacts strongly with the NF-κB (p50/p65) heterodimer bound at the neighboring PRDII site. The orientation of bound NF-κB is also predetermined by the asymmetry of the PRDII site and is the opposite of that found in the crystal structure. The HMG-I/Y protein, proposed as orchestrating enhanceosome assembly, interacts specifically with the PRDII site of the interferon-β enhancer by inserting its DNA-binding segments (AT hooks) into the minor groove, resulting in a significant increase in NF-κB binding affinity for the major groove of this site.

Introduction

An essential step in the formation of human interferon-β (IFN-β) enhanceosome, which regulates IFN-β synthesis, involves interferon regulatory factors (IRFs), which are activated upon virus infection. Activation of IRFs is triggered by phosphorylation of certain Ser/Thr residues in the C-terminal domain (CTD) of this two-domain protein and results in its dimerization and binding to the IFN-β enhancer. This is followed by binding of the other two main components of the enhanceosome: the NF-κB (p50/p65) and bZIP ATF-2/c-Jun heterodimers.

Although the mechanism of IRF activation and binding to the enhancer has been extensively studied by various groups, there are still many questions to be answered. In particular, it is unclear exactly which of the Ser residues in the CTD need to be phosphorylated for dimerization of the IRFs and their binding to the IFN-β enhancer. How does the change in the CTD switch on the ability of the DNA-binding N-terminal domain (NTD) to bind to DNA? How many dimeric IRFs bind to the DNA of the IFN-β enhancer? Does the binding of IRF to the IFN-β enhancer induce cooperative assembly of other components of the enhanceosome?

Among the currently known IRFs, most attention has focused on IRF-3, since this protein seems to be the key positive regulator of IFN-β expression.1, 2, 3, 4 According to Lin et al., who have mimicked the phosphorylation of Ser/Thr by their replacement with the charge-bearing Asp residue, only the substitution of five Ser/Thr residues in the 5ST cluster—the so-called 5D mutants (S396D, S398D, S402D, T404D, and S405D), but not mutants S385D and S386D—can generate the DNA-binding form of IRF-3.5, 6 However, according to Panne et al., phosphorylation of the residues in the 5ST cluster serves only to facilitate the phosphorylation of residues S385 and S386, which are specifically required for IRF-3 dimerization.7 According to Mori et al., viral infection specifically induces phosphorylation of S386 and dimerization of IRF-3.8 These authors concluded that the 5D mutant is active when produced in human 293T cells, is much less active when produced in mouse L929 cells, and is produced as an inactive monomer in Escherichia coli. Since E. coli lacks the capability for posttranslational modification, it was concluded that the active 5D mutant, designed to mimic the phosphorylation of Ser/Thr residues other than S385 and S386, is secondarily phosphorylated at S386, presumably by an irrelevant kinase.

Based on the crystal structure of the CTD, it was proposed that phosphorylation of Ser/Thr residues in the CTD results in a structural rearrangement, exposing the hydrophobic surfaces in this domain that are responsible for its dimerization and thereby unmasking the N-terminal DNA-binding domain to enable binding of an IRF-3 dimer to the enhancer.9

The IFN-β enhancer has been subdivided into four positive regulatory domains (PRDs): PRDIV, PRDIII, PRDI, and PRDII (Fig. 1).11, 12, 13, 14, 15, 16 Earlier, it was assumed that IRF-3 binds to two consensus binding sites (AANNGAAA) at PRDIII and PRDI.1, 17, 18 However, the crystal structure of the complex of this enhancer DNA element with the isolated NTDs of IRF-3 showed association with four NTDs.19, 20, 21, 22 It was therefore concluded that, in addition to two consensus sites, this DNA also has two nonconsensus sites for binding IRFs. These two pairs of sites are roughly on opposite faces of the DNA duplex (Fig. 2). According to this crystal structure, the numbers of bonds between the NTD of IRF-3 with the consensus site and the NTD of IRF-3 with the nonconsensus site do not differ greatly,20, 21 suggesting rather similar affinities of the NTD for these two types of site.

Earlier observations had indicated that binding of the IRFs to the IFN-β enhancer is affected by the presence of the bZIP pair ATF-2/c-Jun at the neighboring binding site (PRDIV) and, conversely, the presence of IRF affects binding of the bZIP pair.19, 20, 22 It has also been suggested that the IRF-3 cooperates with NF-κB in binding to the enhancer.20 Thus, it was inferred that the IRFs, ATF-2/c-Jun, and NF-κB dimers altogether form the combinatorial code regulating the IFN-β gene.20, 21

The hypothesis that multiple binding sites within the IFN-β enhancer create a combinatorial code is attractive but raises a number of questions: namely, how specific is binding of the IRFs to the nonconsensus sequences of this DNA (i.e., how many full-length dimeric IRFs could be bound to this DNA upon enhanceosome assembly in the cell?). If this DNA binds two full-length IRF-3 dimers, they would be expected to project in opposite directions (Fig. 2), and this should have principal importance for the structure of the enhanceosome. It was also unclear how the binding of IRF and ATF-2/c-Jun might cooperate when, according to the crystal structure, these proteins are not in direct contact when bound to the IFN-β DNA. This was explained by the assumption that IRF might induce some change in the DNA structure, propagating into the neighboring PRDIV site that binds ATF-2/c-Jun.19, 20

It should be noted that the cooperativity of enhanceosome assembly was discussed earlier, and it was concluded that it is orchestrated by the presence of the HMG-I/Y protein.10, 23, 24, 25, 26 This protein is completely unfolded in solution and binds specifically to DNA with its three segments—the AT hooks, which penetrate deep into the minor groove at AT-rich sites.27, 28 It was suggested that the IFN-β enhancer has two pairs of HMG-I/Y binding sites: one flanking the PRDIV site (nucleotides − 105 to − 98, and nucleotides − 91 to − 83), and the other at the PRDII site (nucleotides − 63 to − 54, and nucleotides − 50 to − 42) (Fig. 1).10 However, recent publications do not mention HMG-I/Y either as a component of the enhanceosome or as a means responsible for the cooperation of all of its components.20, 21 Exclusion of HMG-I/Y from considerations of enhanceosome assembly was perhaps the consequence of failure to crystallize components of the complex that contained this protein.20, 29 Thus, the role of HMG-I/Y in the assembly of the enhanceosome remains an outstanding problem.

We have previously shown that, despite some doubts,8 the phosphomimetic mutant 5D of IRF-3 (S396D, S398D, S402D, T404D, and S405D) produced in E. coli dimerizes and, as a simple dimer, binds strongly to the 26-bp fragment of the IFN-β enhancer containing the PRDIII–PRDI binding sites.30 It was also shown that dimerization of this protein is the sole reason for the enhanced ability of activated IRF to bind its cognate DNA. Thus, contrary to predictions,9 the increased DNA-binding ability of activated IRF-3 (which is denoted below as IRF3) is not due to unmasking of the N-terminal DNA-binding domain. To address the other above questions, we have now carried out fluorescence anisotropy, isothermal calorimetry and CD titrations of various fragments of the IFN-β enhancer with all the other isolated components of the enhanceosome, namely, the phosphomimetic IRF-3 dimer, the bZIP dimers of ATF-2 and c-Jun, the NF-κB dimers of p50 and p65, and HMG-I (the original name for the architectural transcription factor HMGA1a).

Section snippets

Binding of IRF3⁎ dimer to the IFN-β enhancer

One of the most important problems in enhanceosome assembly is determining whether there are one or two activated IRF-3 dimers bound to the enhancer DNA in solution. In other words, does the IRF-3 dimer bind only to the pair of consensus sites (PRDIII–PRDI) or does it bind also to the recently proposed pair of nonconsensus sites (Fig. 2)? Since the consensus and nonconsensus sites are on opposite faces of the DNA, resolution of this dilemma is critical for understanding the enhanceosome

Conclusions

The titration studies of IFN-β enhanceosome assembly in solution have shown that:

  • (a)

    The IFN-β enhancer binds only one full-length IRF3 dimer [the phosphomimetic IRF-3(5D) dimer] specifically and with very high affinity (Kd = 6 nM) at the PRDIII–PRDI consensus binding sites. Binding of subsequent dimeric IRF3 to this enhancer proceeds with an affinity at least 3 orders of magnitude lower and cannot therefore be regarded as specific.

  • (b)

    The ATF2/cJun heterodimer binds to its cognate site at PRDIV more

IRF-3⁎, the phosphomimetic mutant (S396D, S398D, S402D, T404, and S405) of IRF-3

An expression plasmid encoding a full-length glutathione S-transferase–IRF-3 fusion was kindly provided by Dr. J. Hiscott. Modified versions with substitutions S396D, S398D, S402D, T404D, and S405D (IRF-3(5D)) were made via mutagenesis in our laboratory. The isolation and purification of this protein were described in our previous work.30 An expression plasmid (pMal c2x) encoding the full-length IRF-3(5D) fused to the maltose-binding protein (MBP) was also generated, and this protein was

Acknowledgements

The authors thank Dr. J. Hiscott for the expression plasmid encoding a full-length glutathione S-transferase–IRF-3 fusion, Dr. C. Vinson for the expression plasmids pATF-2 and pc-Jun encoding the bZIP parts of ATF-2 and c-Jun, and Dr. G. Manfioletti for the expression plasmid pAR3038h of HMG-I. We acknowledge the contributions of Victoria Hargreaves and Dr. Elena Makeyeva at the initial phases of this project. We are also grateful to Dr. Crane-Robinson, Dr. Remeta, and Dr. Minetti for critical

References (42)

  • KimT.K. et al.

    The mechanism of transcriptional synergy of an in vitro assembled interferon-beta enhanceosome

    Mol. Cell

    (1997)
  • DraganA.I. et al.

    The energetics of specific binding of AT-hooks from HMGA1 to target DNA

    J. Mol. Biol.

    (2003)
  • OlmstedM.C. et al.

    Grand canonical Monte Carlo molecular and thermodynamic predictions of ion effects on binding of an oligocation (L8 +) to the center of DNA oligomers

    Biophys. J.

    (1995)
  • DraganA.I. et al.

    DNA binding of a non-sequence-specific HMG-D protein is entropy driven with a substantial non-electrostatic contribution

    J. Mol. Biol.

    (2003)
  • DraganA.I. et al.

    Thermodynamic signature of GCN4-bZIP binding to DNA indicates the role of water in discriminating between the AP-1 and ATF/CREB sites

    J. Mol. Biol.

    (2004)
  • EscalanteC.R. et al.

    Structure of NF-κB p50/p65 heterodimer bound to the PRDII DNA element from the interferon-β promoter

    Structure

    (2002)
  • PrivalovP.L. et al.

    What drives proteins into the major and minor grooves of DNA?

    J. Mol. Biol.

    (2007)
  • HiscottJ. et al.

    Triggering the interaction response: the role of IRF3 transcription factor

    Interferon Cytokine Res.

    (1999)
  • YoneyamaM. et al.

    Direct triggering of the type I interferon system by virus infection: activation of a transcription factor complex containing IRF3 and CBP/p300

    EMBO J.

    (1998)
  • LinR. et al.

    Virus-dependent phosphorylation of the IRF-3 transcription factor regulates nuclear translocation, transactivation potential, and proteasome-mediated degradation

    Mol. Cell. Biol.

    (1998)
  • LinR. et al.

    Structural and functional analysis of interferon regulatory factor 3: localization of the transactivation and autoinhibitory domains

    Mol. Cell. Biol.

    (1999)
  • Cited by (0)

    View full text