Nature Structural Biology
10, 913 - 921 (2003)
Published online: 12 October 2003; | doi:10.1038/nsb1002
Crystal structure of IRF-3 reveals mechanism of autoinhibition and virus-induced phosphoactivationBin Y Qin1, 4, Cheng Liu2, 4, Suvana S Lam1, Hema Srinath1, Rachel Delston2, John J Correia3, Rik Derynck2
& Kai Lin11 Department of Biochemistry and Molecular Pharmacology, University of Massachusetts Medical School, Worcester, Massachusetts 01605, USA. 2 Departments of Growth and Development, and Anatomy, Programs in Cell Biology and Developmental Biology, University of California at San Francisco, San Francisco, California 94143-0640, USA. 3 Department of Biochemistry, University of Mississippi Medical Center, Jackson, Mississippi 39216, USA. 4 These authors contributed equally to this work.
Correspondence should be addressed to Rik Derynck derynck@itsa.ucsf.edu or Kai Lin kai.lin@umassmed.eduIRF-3, a member of the interferon regulatory factor (IRF) family of transcription factors, functions as a molecular switch for antiviral activity. IRF-3 uses an autoinhibitory mechanism to suppress its transactivation potential in uninfected cells, and virus infection induces phosphorylation and activation of IRF-3 to initiate the antiviral responses. The crystal structure of the IRF-3 transactivation domain reveals a unique autoinhibitory mechanism, whereby the IRF association domain and the flanking autoinhibitory elements condense to form a hydrophobic core. The structure suggests that phosphorylation reorganizes the autoinhibitory elements, leading to unmasking of a hydrophobic active site and realignment of the DNA binding domain for transcriptional activation. IRF-3 exhibits marked structural and surface electrostatic potential similarity to the MH2 domain of the Smad protein family and the FHA domain, suggesting a common molecular mechanism of action among this superfamily of signaling mediators.Interferon regulatory factors (IRFs) are structurally related transcription factors that have important roles in host defense against viral infection, development of the immune system and apoptosis1,
2,
3. IRF family members are critically involved in the transcriptional induction of type I interferons (IFNs), and regulate the expression of IFN-responsive genes.
All mammalian IRFs share a conserved N-terminal DNA-binding domain with characteristic five-tryptophan repeats4,
5. The C-terminal portions of IRF-3 to IRF-9 contain a conserved IRF association domain (IAD) that mediates homo- and heteromeric interactions among IRFs and interacts with other transcriptional comodulators. The transactivation functions of the IAD of IRF-3, IRF-4, IRF-5 and IRF-7 are suppressed by autoinhibitory structures6,
7,
8,
9,
10.
The closely related IRF-3 and IRF-7 function as molecular scouts for viral infection11,
12,
13,
14. IRF-3 is constitutively expressed and localized in the cytoplasm as an inactive monomer maintained by autoinhibitory domains that flank the IAD6. Upon viral infection, IRF-3 is phosphorylated by the virus-activated kinase, which contains the I B kinase (IKK)-related kinases IKK and TANK-binding kinase 1 (TBK1)15,
16,
17,
18,
19. Phosphorylation converts IRF-3 into an active oligomer, which enters the nucleus and activates expression of IFN- / . IRF-7 is similarly activated, yet displays a basal activity in the absence of virus infection7. Distinctively from IRF-3, the expression of IRF-7 is induced upon viral infection following receptor activation by IFN-/7,
20,
21,
22.
Seven phosphorylation sites within the C-terminal autoinhibitory sequence of IRF-3 have been implicated in virus-induced activation. Mutation of Ser385 or Ser386 rendered IRF-3 incapable of activation15. The corresponding residues in IRF-7 are also important for IRF-7 activation7,
23. Two serine-threonine clusters in IRF-3 downstream of the Ser385-Ser386 cluster are also targets of phosphorylation. Mutations of the Ser396-Ser398 cluster or the Ser402-Thr404-Ser405 cluster to alanine render IRF-3 incapable of undergoing virus-induced phosphorylation and activation16. Conversely, substitution of these two clusters with phosphomimetic aspartate residues generates a constitutively active mutant that can enter the nucleus and transactivate a reporter gene without viral infection16. IRF-7 contains serine residues in the corresponding region and substitution of these with aspartate activates IRF-7, similarly to IRF-3 (ref. 7).
It has been proposed that the IAD may have a structural fold similar to the MH2 domain of Smad proteins, which act as transforming growth factor (TGF-) signaling mediators and transcription factors24. In response to TGF-, the TGF- type I receptor kinase phosphorylates and thereby activates the receptor-activated Smads (R-Smads)25, resulting in their conversion from a basal monomer state into the active trimer state, and a subunit exchange with Smad4 to form a heterotrimer26,
27,
28. The heteromeric Smad complexes enter the nucleus and form complexes with DNA sequence−specific transcription factors to regulate transcription of target genes. The IAD and MH2 domain share many mechanistic similarities. Both domains have transactivation functions in gene expression, are activated by phosphorylation and mediate protein-protein interactions. Phosphorylation of the R-Smads occurs near the C terminus at locations similar to those of phosphorylation in IRF-3 and IRF-7.
The forkhead-associated (FHA) domain shares the same protein fold as the MH2 domain29,
30. The FHA domains were first identified in forkhead transcription factors and have since been found in many signaling proteins. The FHA and MH2 domains share topologically related basic and hydrophobic surface features with important functional roles.
Here, we report the crystal structure of the IRF-3 transactivation domain, which includes the IAD as well as the flanking autoinhibitory structures. The structure reveals the molecular basis of autoinhibition and lends insights into the mechanism of virus-induced phosphoactivation. The structure and surface electrostatic properties of IAD are reminiscent of the MH2 and FHA domains, suggesting that the three domains may use these surface features for a common molecular mechanism of action.
Results
Overall structure
The asymmetric unit of the crystal contains two molecules of the IRF-3 transactivation domain (Fig. 1a). The two molecules are essentially identical, with a r.m.s. deviation of 0.6 Å for the C trace, and contain a buried surface area of  300 Å2. The dimeric arrangement in the asymmetric unit is unlikely to be physiological, as sedimentation analysis of the protein reveals a monomeric state with little propensity to oligomerize (Fig. 1b and see Supplementary Figs. 1 and 2 online). The structure of the IRF-3 C-terminal transactivation domain contains a central MH2 domain fold of the Smad protein family, flanked by novel structural extensions at the N and C termini (compare Fig. 1c and d). The IAD of IRF-3 matches the MH2 domain fold, consisting of a -sandwich core capped by helices and loops. The center -sandwich core of the IAD domain is conserved in the MH2 domain, whereas the connecting secondary structures exhibit some variations. The sequence corresponding to the L3 loop in the IAD, which connects the 7 and 8 strands, forms a helix (H2) in the MH2 domain. Also, the sequence corresponding to the L4 loop in the IAD, which connects the H3 helix and the 10 strand, forms a helix (H4) in the MH2 domain. Based on sequence alignment of the human IRF protein family, the overall IAD fold observed in IRF-3 is conserved in IRF-4 to IRF-9 (Fig. 2). Structures flanking the IAD correspond to the autoinhibitory structures of IRF-3. The N-terminal region forms a helix (H1). The C-terminal region forms a two-stranded -sheet (12 and 13) interconnected by a loop (L6), followed by a helix (H5) and a C-terminal extension. Notably, these N- and C-terminal autoinhibitory structures interact with each other as well as with the IAD to form a compact structure. The proposed phosphorylation sites are located within the C-terminal autoinhibitory structures, on the H4-12 turn, L6 loop, 13 strand and 13-H5 turn. Sequence alignment reveals that the N- and C-terminal sequences flanking the IAD are diverse among the IRFs.
 | | |
Structural basis of autoinhibition
The N- and C-terminal autoinhibitory segments of IRF-3 interact with each other and together cover a hydrophobic surface of 1,200 Å2 on the H3 and H4 helices of the IAD (Figs. 1c and 3a). The structure suggests a synergistic effect of the N- and C-terminal sequences in autoinhibition as they form an interdigitating interaction, with the N-terminal H1 helix inserting into the space between the C-terminal 12-L6-13 structure and H5 helix. Together, the N- and C-terminal autoinhibitory sequences stabilize the hydrophobic surface on the H3 and H4 helices of the IAD by providing a complementary hydrophobic surface for interaction.
| | Figure 3. Structural basis of IRF-3 regulation by autoinhibition and virus-induced phosphoactivation. | | |  | (a) Autoinhibition is mediated by hydrophobic contacts between the H3 and H4 helices of IAD, and the autoinhibitory structures. The interaction surfaces on the IAD (cyan) and autoinhibitory structures (purple) are shown side-by-side by separating the two surfaces of the autoinhibited IRF-3, so that the hydrophobic side chains (gray) on each surface can be seen. (b) Effects of mutations in the autoinhibitory elements on IRF-3 homomeric interaction (top) and transcriptional activity (bottom). Top: 293T cells expressing Myc-tagged and HA-tagged, wild-type or mutant IRF-3, as indicated, were subjected to immunoprecipitation (IP) with anti-Myc, followed by immunoblotting (IB) with anti-HA to detect the coprecipitated IRF-3, indicative of homomeric interaction. Cells were infected with Sendai virus (Sv) (+) or left uninfected (-). Bottom: HeLa cells were transfected with the IRF-3 reporter plasmid p31x3-Luc, along with expression plasmids for wild-type or mutant IRF-3, as indicated. Cells were mock treated (-) or infected with Sendai virus (Sv). Normalized luciferase activities with standard deviations are shown relative to those of control transfected cells in the presence of virus. The IRF-3 mutations are as follows: H1M, Leu192, Leu195 and Leu196 replaced by arginine; L6M, Val391, Leu393 and Ile395 replaced by arginine; H5M, Tyr408, Leu412, Leu415 and Val416 replaced by arginine; CM, Met419 and Phe421 replaced by arginine; N394, protein terminated after residue 394; N388, protein terminated after residue 388. (c) Surface representation of the phosphorylation site clusters on IRF-3. The IAD is colored cyan. The N- and C-terminal autoinhibitory structures are purple. The surface of the segment containing the phosphorylation sites (residues 385−406) is shown in semi-transparent mode. The three clusters of phosphorylation site are circled. Acidic residues near the phosphorylation sites are colored red. (d) Stereo view of the Fo - Fc simulated annealing omit map of the segment containing the phosphorylation sites (residues 385−406). The three clusters of phosphorylation site are circled as in Figure 3c.
 Full Figure and legend (126K) |
| |
Hydrophobic residues on the IAD that are buried by the presence of the autoinhibitory structures include Leu322, Pro324, Ile326, Val327, Leu329 and Ile330 in the H3 helix, as well as Cys371, Ala374, Leu375, M378 and Ala379 in the H4 helix (Fig. 3a, left). In the absence of the autoinhibitory segments, these residues form a continuous hydrophobic surface that may be involved in protein-protein interaction in the active form of IRF-3. Leu329 in the H3 helix is conserved in all IADs, and mutation of the corresponding leucine in IRF-4, IRF-8 and IRF-9 impairs signaling31,
32. Notably, the corresponding residue in the MH2 domain of the R-Smads is also a conserved leucine. Sequence analysis suggests that the corresponding surfaces in IRF-4 to IRF-9 are also hydrophobic, although in general they contain fewer hydrophobic residues than IRF-3, and the residues are not always identical (Fig. 2).
The hydrophobic residues within the N- and C-terminal autoinhibitory segments form an interacting surface that complements the hydrophobic surface on the H3 and H4 helices of the IAD (Fig. 3a, right). The hydrophobic surface in the autoinhibitory structure is primarily provided by Leu192, Leu195 and Leu196 in the H1 helix, Val391, Leu393 and Ile395 in the L6 loop, Tyr408, Leu412, Leu415 and Val416 in the H5 helix, and Met419 and Phe421 in the C-terminal extension.
To address the functional consequences of disrupting the autoinhibitory contacts, we generated IRF-3 mutants in which the hydrophobic amino acids in the H1 or H5 helices, the L6 loop or the C-terminal tail were replaced by arginines. Whereas wild-type IRF-3 demonstrated the expected virus-dependent homomeric assembly, all mutants exhibited constitutive homomeric interaction, except for the one mutated in the L6 loop, which showed a low level of interaction that was not enhanced in response to virus infection (Fig. 3b, top). In addition, deletion of the C-terminal autoinhibitory sequence from residues 388 or 394 to the C terminus also conferred constitutive homomeric interaction. These results provide evidence that these hydrophobic residues in the autoinhibitory structures synergize to maintain IRF-3 in a monomeric state, and suggest a conformational transition upon activation to stabilize the oligomeric form. The low level of homomeric interaction of the L6 loop mutant suggests that this mutation also interferes with subunit oligomerization. We also tested the abilities of these mutants to activate transcription from three tandem PRDIII−PRDI sequences that are derived from the IFN- promoter and known to bind IRF-3 and IRF-7 (ref. 33). Notably, with the exception of the N-terminal H1 helix mutant, all C-terminal autoinhibitory domain mutants exhibited a substantially decreased transactivation activity from this reporter (Fig. 3b, bottom). These findings strongly suggest a transactivation role for the C-terminal hydrophobic residues, in addition to their autoinhibitory function, presumably through novel protein-protein interactions in the active state. The inducible transactivation function of the H1M mutant, which shows constitutive homo-oligomerization, is likely to reflect the role of virus-induced phosphorylation in transcriptional activation. As with wild-type IRF-3, and by analogy with the R-Smads, this phosphorylation may have a role in stabilizing the IRF complex and in recruiting the coactivators that are required for transcription.
With the exception of IRF-7, the same autoinhibitory structures observed for IRF-3 may not exist in other IRFs, as their sequences flanking the central IAD domain are diverse (Fig. 2). Thus, the IRF-4, IRF-5, IRF-6, IRF-8 and IRF-9 sequences do not reveal a corresponding N-terminal autoinhibitory helix to that in IRF-3. Also, these IRFs contain only a subset of the hydrophobic residues present in the IRF-3 C-terminal autoinhibitory segment (12-L6-13-H5). In the extreme case of IRF-9, this C-terminal autoinhibitory sequence is completely absent. It is therefore possible that the diverse N- and C-terminal sequences flanking the IAD in the different IRFs confer differential regulation of the transactivation function. Alternatively, different autoinhibitory mechanisms may exist for different IRFs. For example, IRF-4 and IRF-5 become more activated upon deletion of the C-terminal sequence, suggesting the presence of an autoinhibitory mechanism9,
10. In contrast to these other IRFs, IRF-7 and IRF-3 show a similar pattern of hydrophobic sequences flanking the IAD, suggesting that IRF-7 may use a similar autoinhibitory structure and mechanism to that of IRF-3. The lower number of hydrophobic residues in the IAD of IRF-7 may account for a higher basal activity of IRF-7 than that of IRF-3 (ref. 7), which is further enhanced upon virus infection.
Structure of the phosphorylation sites
The location and environment of the phosphorylation sites in the IRF-3 crystal structure provide insight into how IRF-3 is activated. Seven serine and threonine residues within the C-terminal autoinhibitory segment have been shown to be potential sites of phosphorylation (Figs. 1c and 2). Most of these sites are partially buried by hydrophobic residues and located adjacent to acidic residues, suggesting that phosphorylation will perturb autoinhibitory interactions owing to charge repulsion. This subsequently leads to activation (Fig. 3c,d). Based on their locations in the structure, the phosphorylation sites can be grouped into three clusters. Cluster 1 contains Ser385 and Ser386, which are located in the H4-12 turn connecting the IAD to the C-terminal autoinhibitory segment. The side chain of Ser385 is partially buried by the side chain of Leu387. As Ser385 is located at the hinge point between the IAD and the C-terminal autoinhibitory structure, phosphorylation of Ser385 is expected to trigger a local structural destabilization that could have a long-range impact on the overall stability of the autoinhibitory structures. The side chain of Ser386 is well exposed and its phosphorylation should have less effect on the conformation of the protein. Cluster 2 contains Ser396 and Ser398, which are located within the L6 loop. The side chain of Ser396 potentially forms a hydrogen bond with the side chain of Glu200, and is within van der Waals contact of the side chain of Asn397. The side chain of Ser398 may form a hydrogen bond with the side chain of Arg194, and is within van der Waals contact of the side chains of Phe204 and Tyr260. As this part of the L6 loop is involved in the interaction with the N-terminal autoinhibitory helix H1, phosphorylation of Ser396 and Ser398 likely destabilizes the interactions between the N- and C-terminal autoinhibitory structures. Ser396 has recently been identified as the minimal phosphoacceptor residue required for the in vivo activation of IRF-3 in response to viral infection34. Cluster 3 contains Ser402, Thr404 and Ser405. Ser402 is located in the 13 strand, potentially forms a hydrogen bond with the side chain of Glu388 and is within van der Waals contact of the side chain of Leu401. Thr404 and Ser405 are located on the 13-H5 turn. The side chain of Thr404 is within van der Waals contact of Gln407, Asp406 and Glu388, whereas the side chain of Ser405 is within van der Waals contact of Lys409, Asp406 and Tyr408. As 13 and the 13-H5 turn form part of the hydrophobic core that stabilize the autoinhibitory structure, phosphorylation of the cluster 3 residues likely causes unfolding of the autoinhibitory structure. Overall, our analysis suggests that phosphorylation will perturb the hydrophobic core between the IAD and the autoinhibitory structures, leading to structural rearrangement of the autoinhibitory structures and unmasking of the hydrophobic surface on the H3 and H4 helices.
Although the structure suggests how phosphorylation can disrupt the autoinhibitory mechanism, it raises a question as to how multiphosphorylation occurs, especially with respect to the accessibility of the phosphorylation sites to the recently identified virus-activated kinase subunits IKK and TBK1 (refs. 18,19). Whereas Ser385 and Ser386 are located on a solvent-accessible turn, the phosphoacceptor residues in clusters 2 and 3 are not easily accessible, as they are packed in between the autoinhibitory structures and the IAD. Some dynamic movement of the autoinhibitory structures in solution may provide access of the phosphoacceptor residues to the kinase. IRF-3 could also be stabilized in a conformation favoring phosphorylation through interactions with IKK or TBK1 or some components of the virus, such as the nucleocapsid protein11. In addition, phosphorylation may occur sequentially, where phosphorylation of the first, more accessible residue leads to exposure of additional phosphorylation sites. For example, phosphorylation of Ser385 and/or Ser386 in cluster 1 may partially unfold the autoinhibitory structure, facilitating phosphorylation of cluster 2 and cluster 3 residues. This mechanism is consistent with the previous observation that phosphorylation of the cluster 2 and cluster 3 residues in IRF-3 promotes DNA binding and transactivation, but the effect is dependent on the presence of Ser385 and Ser386 (ref. 35), presumably through phosphorylation of these two residues.
Surface electrostatic property of IRF-3
Analysis of the surface electrostatic potential of the IRF-3 C-terminal transactivation domain reveals two oppositely charged surfaces that may be involved in IRF-3 activation. An acidic surface is associated with the C-terminal autoinhibitory segment (residues 385−427), which is rich in glutamate and aspartate residues and has a calculated pKa of 4.0. In the autoinhibited structure, all the glutamate and aspartate side chains in that segment are solvent exposed, creating an acidic surface (Fig. 4a, bottom far left). As the phosphorylation sites are also located within this segment, phosphorylation is expected to further increase the acidity of the C-terminal tail. This observation raises the question of how IRF-3 manages the molecular transformation of this highly acidic structure upon phosphorylation-induced unfolding of the C-terminal tail. This acidic characteristic of the C-terminal sequence is shared by IRF-7 (pKa of 3.2 over the corresponding C-terminal segment), which is also activated by phosphorylation of residues in the C-terminal region. The corresponding C-terminal regions of other IRFs are less acidic, with pKa values ranging from 4.8 in IRF-8 to 9.4 in IRF-6.
| | |
Upon phosphorylation, the acidic C-terminal tail of IRF-3/7 may be stabilized through interactions with basic residues. A basic surface is identified on IRF-3 on the opposite end of the -sandwich core relative to the autoinhibitory structures (Fig. 4a, top left two panels). The basic surface is formed by Arg211 and Arg213 in the 1-2 turn, Arg255, Arg262 and His263 in the H2 helix, Arg285, His288 and His290 in the 6-7 turn, Lys313 in the 8-9 turn, and Lys360 and Arg361 in the L5 loop (Fig. 4b). Notably, the corresponding surfaces in the MH2 and FHA domains are also basic and have been shown to be functionally important. In the MH2 domain of the R-Smads, one group of basic residues has been implicated to serve as a docking site for the phosphorylated juxtamembrane segment of the type I receptor, a mechanism critical for specific recruitment and phosphorylation of the R-Smad by the receptor kinase (Fig. 4a, top third panel)36. These residues correspond to Asn240, Gln241, Arg287 and His288 of Smad3, of which Arg287 and His288 can be structurally aligned to Arg262 and His263 in IRF-3 (Fig. 4b). Adjacent basic residues have also been shown to participate in Smad-Smad interaction by serving as a docking site for the phosphorylated C-terminal sequence of another Smad molecule27,
28,
37. These residues are Lys332, Lys377 and Arg385 in Smad3, structurally corresponding to Lys313, Lys360 and R361, respectively, in IRF3. The corresponding surface in the FHA domain is also basic and has been demonstrated to bind a phosphorylated peptide in mediating protein-protein interactions (Fig. 4a, top far right). Thus, despite their low sequence homology, these three structurally related domains have maintained a basic electrostatic property over this surface, which in the MH2 and FHA domains has been shown to mediate phosphopeptide binding. These findings suggest that the corresponding basic surface on IRF-3 may have a similar role, potentially in stabilizing the acidic C-terminal tail in the active conformation.
To assess the roles of these basic residues in IRF-3 activation, we generated mutants in which the IRF-3 basic residues, mentioned above, were replaced by alanines. Consistent with the structural analysis, all mutants, except for the K313A mutant, exhibited a defect in virus-induced homomeric interaction (Fig. 5a, top). They all showed a low or intermediate level of constitutive complex formation that was not enhanced upon virus infection. In contrast, the low-level homomeric interaction of the K313A mutant was enhanced by virus to a level similar to that of wild-type IRF-3. In agreement with these interaction results, these mutants, again with the exception of the K313A mutant, showed impaired virus-induced transactivation at the PRDIII−PRDI sequence (Fig. 5a, bottom), thus correlating impaired virus-induced transactivation with impaired homomerization. Together, these data suggest a role of these basic residues, except for Lys313, in homomeric interaction and transcriptional activation of IRF-3, potentially by binding to phosphorylated residues in another IRF-3 (or IRF-7) molecule or another transcription cofactor.
| | |
Buried underneath the acidic C-terminal autoinhibitory structure is the putative active site of the IAD. Upon phosphorylation-induced conformational change of the autoinhibitory structures, the hydrophobic side chains on H3 and H4 helices will likely become exposed (Fig. 4a, bottom left two panels). The corresponding surface in the MH2 domain of the R-Smads, formed by the H3 and H5 helices, is also hydrophobic and has important roles in TGF- signaling. In Smad2 and Smad3, this hydrophobic surface interacts with SARA, a membrane-anchored protein that recruits Smad2 and Smad3 to the receptor for phosphorylation (Fig. 4a, bottom third panel)38,
39,
40. This hydrophobic surface also interacts with c-Ski, which functions as a Smad2/3 transcriptional corepressor40,
41. In the FHA domain of Chk2, a kinase critical for the activation of DNA damage response, an I157T mutation was identified in Li-Fraumeni syndrome, a highly penetrant familial cancer phenotype42. This mutation occurs on a hydrophobic surface of Chk2 corresponding to part of the SARA binding surface in Smad2/3 (Fig. 4a, bottom right). The analysis suggests that the IAD may use the hydrophobic surface that is topologically related to those in the MH2 and FHA domains for functional interactions once activated by phosphorylation. Consistently, mutation of the hydrophobic residues on the H3 or H4 helices of the IAD to alanines results in a substantial decrease in the transactivation activity (Fig. 5b, bottom). These mutants homo-oligomerize in the absence of virus infection, presumably owing to disruption of the autoinhibitory hydrophobic core that is essential for maintaining IRF-3 in the monomeric state (Fig. 5b, top).
Discussion
Although further structural investigation is required to elucidate the mechanism of phosphorylation-induced IRF-3 oligomerization, the analysis above suggests a possible activation model using the basic and hydrophobic surfaces (Fig. 6). Phosphorylation triggers a concerted rearrangement of the autoinhibitory structures, exposing the hydrophobic surface on the H3 and H4 helices. The structural rearrangement is also likely to reposition the N-terminal DNA-binding domain and unmask the DNA-binding surface for promoter interaction. The exposed hydrophobic surface on the H3 and H4 helices and the autoinhibitory elements are likely to be involved in transactivation, presumably through interaction with the coactivator CBP/p300 and/or other transcriptional partners15,
43. We propose that the phosphorylated and acidic C-terminal tails are stabilized in an oligomeric arrangement, presumably a dimer, through interactions with the basic surface of the neighboring IRF-3 or IRF-7 molecule, which can both homo- and hetero-oligomerize to activate transcription. In addition to the interactions mediated by the phosphorylated tail, the subunit-to-subunit contact likely involves a novel interface common to IRF-3 and IRF-7, and is facilitated by the phosphorylation-induced conformational change.
| | |
The proposed model of IRF-3 activation is reminiscent of the activation mechanism in R-Smads, in which the phosphorylated C-terminal tail stabilizes a homotrimeric complex by bridging the basic surface of the neighboring subunit28,
37. However, IRF-3 is unlikely to form a Smad-like trimer, as modeling of IRF-3 to the trimeric scaffold of Smad1 reveals a marked clashing between the L1 loop and H4 helix of adjacent subunits, owing to a different conformation of the L1 loop when compared with the corresponding loop in R-Smads (data not shown). The trimer interface residues conserved in the Smads are not conserved in the IRF family members. In a related study of IRF-3 also published in this issue of Nature Structural Biology, Takahasi et al. suggest that a dimeric arrangement of the subunit, similar to the dimer observed in the asymmetric unit of our crystal (Fig. 1a), represents the active conformation of IRF-3 (ref. 44). Additional data will be required to test the validity of the model. Nevertheless, it is worth noting that the model does not take into consideration the role of autoinhibition and the cluster 2 and cluster 3 phosphorylation sites.
In conclusion, the crystal structure of the IRF-3 transactivation domain reveals the molecular mechanism of autoinhibition and offers insights into the mechanism of phosphoactivation. The observation that the IAD of IRF-3 has functional surfaces with similar electrostatic features as the MH2 domain of Smads and the FHA domain suggests a common molecular mechanism of action among this superfamily of signaling mediators. Explaining how these proteins tailor these common features for their specific mechanisms of action may lead to design of small molecules with specific therapeutic potential.
Methods
Expression, purification and crystallization of IRF-3 transactivation domain.
The cDNA fragment encoding the polypeptide sequence of residues 173−427 of human IRF-3 was generated by PCR and subcloned into the pGEX-4T-2 vector (Amersham). The GST-fused IRF-3 was expressed in Escherichia coli, extracted using glutathione-Sepharose and cleaved by thrombin to release IRF-3. IRF-3 was further purified on a DEAE column equilibrated with 20 mM Tris, pH 7.4, 10 mM NaCl, 0.1 mM EDTA and 1 mM DTT, and eluted by a NaCl gradient from 50 mM to 250 mM. The purified IRF-3 was concentrated to 30 mg ml-1. Crystals were obtained by the hanging-drop vapor diffusion technique with well solution containing 250 mM ammonium phosphate and 100 mM bis-Tris buffer, pH 6.0.
Structure determination.
The structure of IRF-3 was determined by the multiple isomorphous replacement (MIR) technique (Table 1). Native and heavy atom soaked crystals of IRF-3 were transferred to cryosolvents consisting of 25% (v/v) glycerol and 75% of the well solution, and were flash frozen in liquid nitrogen. The data were obtained from beam line 5.0.1 of the Advanced Light Source at Berkeley Lab (Berkeley, California, USA). All crystals grew in space group P4212. Data were integrated and reduced using DENZO and SCALEPACK45. Heavy atom refinement, phase calculation, and solvent modification were carried out using CNS46. Model building was done using O47. Structural refinement was carried out using the conjugated gradient and simulated annealing protocols in CNS. Both IRF-3 molecules in the asymmetric unit contain residues 189−427 in the final model. The N-terminal residues 173−188 were disordered. Ramachandran analysis indicates that 92% and 8% are in the core and allowed regions, respectively.
| | |
Analytical ultracentrifugation.
Sedimentation equilibrium and velocity experiments were carried out as a function of loading concentration, at 19.7 °C, in a buffer containing 20 mM HEPES, 0.1 mM EDTA and 100 mM NaCl at pH 7.3. Both techniques revealed concentration independence and the presence of predominantly monomeric IRF-3 with a small amount of irreversible dimer (4−7%) (see Supplementary Figs. 1 and 2 online for details). To demonstrate this, one data set was plotted as ln C vs. r2 / 2 and compared to the expected line for a monomer and a dimer. The IRF-3 transactivation domain is clearly a monomer under these conditions and over this concentration range.
IRF-3 expression plasmids.
Wild-type IRF-3 was expressed as an N-terminally Flag-tagged or HA-tagged protein from the CMV promoter in pRK5 expression plasmids. All point mutants and deletion mutants were generated using the QuikChange site-directed mutagenesis kit (Stratagene) and verified by DNA sequencing. The mutations in IRF-3 are mentioned in the figure legends. The p31x3-Luc luciferase reporter plasmid containing three copies of the PRDIII−PRDI sequence of the human IFN- promoter was prepared by inserting oligonucleotides corresponding to the PRDIII−PRDI sequences into the HindIII site of the minimal TATA-box promoter, pTA-Luc48.
Cell culture, transient transfections and luciferase reporter assays.
293T cells and HeLa cells were maintained in DMEM supplemented with 15% (v/v) FBS, 100 IU ml-1 ampicillin, and 100  g ml-1 streptomycin. HeLa cells were transiently transfected in 12-well tissue culture plates using Fugene 6 according to manufacturer's instructions (Roche). For each transfection, 0.25 g of the luciferase reporter plasmid, 10 ng of the -galactosidase expression plasmid (pRK5--Gal) and 50 ng wild-type or mutant IRF-3 expression plasmid, as shown in the figures, were used. When indicated, infection with Sendai viruses (100 HA units ml-1) was done at 24 h after transfection for 20 h. Luciferase assays were carried out as described49.
Immunoprecipitation and western blot analyses.
293T cells were transfected with expression plasmids encoding HA-tagged and Myc-tagged wild-type or mutated IRF-3 as indicated in the figures, using Lipofectamine Plus (Invitrogen). For virus infection, transfected cells were infected with Sendai virus (80 HA units ml-1) for 12 h at 24 h after transfection. Whole-cell lysates were prepared in MLB buffer (20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% (v/v) NP-40, and Complete protease inhibitors (Roche)), and immunoprecipitated with 1 g anti-Myc 9E10 (Covance) and protein A-Sepharose 4B (Amersham-Pharmacia). Immunoprecipitated proteins were separated on SDS-PAGE. Western blotting was carried out using anti-HA (Covance). Bands were visualized using ECL reagents (Amersham-Pharmacia).
Coordinates.
Coordinates and structure factors have been deposited in the Protein Data Bank (accession code 1QWT).
Note: Supplementary information is available on the Nature Structural Biology website.
Received 11 June 2003; Accepted 15 September 2003; Published online: 12 October 2003.
REFERENCES
- Mamane, Y. et al. Interferon regulatory factors: the next generation. Gene 237, 1-14 (1999). | Article | PubMed | ISI | ChemPort |
- Barnes, B., Lubyova, B. & Pitha, P.M. On the role of IRF in host defense. J. Interferon Cytokine Res. 22, 59-71 (2002). | Article | PubMed | ISI | ChemPort |
- Taniguchi, T., Ogasawara, K., Takaoka, A. & Tanaka, N. IRF family of transcription factors as regulators of host defense. Annu. Rev. Immunol. 19, 623-655 (2001). | Article | PubMed | ISI | ChemPort |
- Fujii, Y. et al. Crystal structure of an IRF-DNA complex reveals novel DNA recognition and cooperative binding to a tandem repeat of core sequences. EMBO J. 18, 5028-5041 (1999). | Article | PubMed | ISI | ChemPort |
- Escalante, C.R., Yie, J., Thanos, D. & Aggarwal, A.K. Structure of IRF-1 with bound DNA reveals determinants of interferon regulation. Nature 391, 103-106 (1998). | Article | PubMed | ISI | ChemPort |
- Lin, R., Mamane, Y. & Hiscott, J. Structural and functional analysis of interferon regulatory factor 3: localization of the transactivation and autoinhibitory domains. Mol. Cell Biol. 19, 2465-2474 (1999). | PubMed | ISI | ChemPort |
- Lin, R., Mamane, Y. & Hiscott, J. Multiple regulatory domains control IRF-7 activity in response to virus infection. J. Biol. Chem. 275, 34320-34327 (2000). | Article | PubMed | ISI | ChemPort |
- Au, W.C., Yeow, W. & Pitha, P.M. Analysis of functional domains of interferon regulatory factor 7 and its association with IRF-3. Virology 280, 273-282 (2001). | Article | PubMed | ISI | ChemPort |
- Brass, A.L., Kehrli, E., Eisenbeis, C.F., Storb, U. & Singh, H. Pip, a lymphoid-restricted IRF, contains a regulatory domain that is important for autoinhibition and ternary complex formation with the Ets factor PU.1. Genes Dev. 10, 2335-2347 (1996). | PubMed | ISI | ChemPort |
- Barnes, B.J., Kellum, M., Field, A.E. & Pitha, P.M. Multiple regulatory domains of IRF-5 control activation, cellular localization, and induction of chemokines that mediate recruitment of T lymphocytes. Mol. Cell. Biol. 22, 5721-5740 (2002). | Article | PubMed | ISI | ChemPort |
- tenOever, B.R., Servant, M.J., Grandvaux, N., Lin, R. & Hiscott, J. Recognition of the measles virus nucleocapsid as a mechanism of IRF-3 activation. J. Virol. 76, 3659-3669 (2002). | Article | PubMed | ISI | ChemPort |
- Iwamura, T. et al. Induction of IRF-3/-7 kinase and NF-B in response to double-stranded RNA and virus infection: common and unique pathways. Genes Cells 6, 375-388 (2001). | Article | PubMed | ISI | ChemPort |
- Sato, M. et al. Distinct and essential roles of transcription factors IRF-3 and IRF-7 in response to viruses for IFN-/ gene induction. Immunity 13, 539-548 (2000). | PubMed | ISI | ChemPort |
- Wathelet, M.G. et al. Virus infection induces the assembly of coordinately activated transcription factors on the IFN- enhancer in vivo. Mol. Cell 1, 507-518 (1998). | PubMed | ISI | ChemPort |
- Yoneyama, M. et al. Direct triggering of the type I interferon system by virus infection: activation of a transcription factor complex containing IRF-3 and CBP/p300. EMBO J. 17, 1087-1095 (1998). | Article | PubMed | ISI | ChemPort |
- Lin, R., Heylbroeck, C., Pitha, P.M. & Hiscott, J. Virus-dependent phosphorylation of the IRF-3 transcription factor regulates nuclear translocation, transactivation potential, and proteasome-mediated degradation. Mol. Cell Biol. 18, 2986-2996 (1998). | PubMed | ISI | ChemPort |
- Servant, M.J. et al. Identification of distinct signaling pathways leading to the phosphorylation of interferon regulatory factor 3. J. Biol. Chem. 276, 355-363 (2001). | Article | PubMed | ISI | ChemPort |
- Sharma, S. et al. Triggering the interferon antiviral response through an IKK-related pathway. Science 300, 1148-1151 (2003). | Article | PubMed | ISI | ChemPort |
- Fitzgerald, K.A. et al. IKK and TBK1 are essential components of the IRF3 signaling pathway. Nat. Immunol. 4, 491-496 (2003). | Article | PubMed | ISI | ChemPort |
- Au, W.C., Moore, P.A., LaFleur, D.W., Tombal, B. & Pitha, P.M. Characterization of the interferon regulatory factor-7 and its potential role in the transcription activation of interferon A genes. J. Biol. Chem. 273, 29210-29217 (1998). | Article | PubMed | ISI | ChemPort |
- Zhang, L. & Pagano, J.S. Structure and function of IRF-7. J. Interferon Cytokine Res. 1, 95-101 (2002). | Article |
- Levy, D.E., Marie, I., Smith, E. & Prakash, A. Enhancement and diversification of IFN induction by IRF-7-mediated positive feedback. J. Interferon Cytokine Res. 1, 87-93 (2002). | Article |
- Marie, I., Durbin, J.E. & Levy, D.E. Differential viral induction of distinct interferon- genes by positive feedback through interferon regulatory factor-7. EMBO J. 17, 6660-6669 (1998). | Article | PubMed | ISI | ChemPort |
- Eroshkin, A. & Mushegian, A. Conserved transactivation domain shared by interferon regulatory factors and Smad morphogens. J. Mol. Med. 77, 403-405 (1999). | Article | PubMed | ISI | ChemPort |
- Zhang, Y., Feng, X.H., Wu, R.Y. & Derynck, R. Receptor-associated Mad homologues synergize as effectors of the TGF- response. Nature 383, 168-172 (1996). | Article | PubMed | ISI | ChemPort |
- Kawabata, M., Inoue, H., Hanyu, A., Imamura, T. & Miyazono, K. Smad proteins exist as monomers in vivo and undergo homo- and hetero-oligomerization upon activation by serine/threonine kinase receptors. EMBO J. 17, 4056-4065 (1998). | Article | PubMed | ISI | ChemPort |
- Chacko, B.M. et al. The L3 loop and C-terminal phosphorylation jointly define Smad protein trimerization. Nat. Struct. Biol. 8, 248-253 (2001). | Article | PubMed | ISI | ChemPort |
- Qin, B.Y. et al. Structural basis of Smad1 activation by receptor kinase phosphorylation. Mol. Cell 8, 1303-1312 (2001). | PubMed | ISI | ChemPort |
- Durocher, D. et al. The molecular basis of FHA domain:phosphopeptide binding specificity and implications for phospho-dependent signaling mechanisms. Mol. Cell 6, 1169-1182 (2000). | PubMed |
|
