Regulation of Interferon Production by RIG-I and LGP2: A Lesson in Self-Control

See allHide authors and affiliations

Science's STKE  01 May 2007:
Vol. 2007, Issue 384, pp. pe20
DOI: 10.1126/stke.3842007pe20


The cytoplasmic CARD-containing DExD/H box RNA helicases RIG-I and MDA5 act as sensors of viral infections through recognition of viral double-stranded (ds) RNAs. They both associate with the mitochondrial adaptor IPS-1 (also referred to as MAVS, VISA, and CARDIF) through homotypic CARD-CARD interactions. IPS-1, in turn, triggers signaling pathways, including activation of the protein kinases TBK1 and IKKε, responsible for the phosphorylation of IRF3, a key transcription factor involved in interferon (IFN) synthesis, one essential element of the innate immune response. RIG-I remains in an autoinhibited state in the absence of dsRNA, through an internal repressor domain (RD) that binds within both its CARD and its RNA helicase domains and therefore acts in cis to control its multimerization and interaction with IPS-1. Ectopic expression of the RD prevents signaling and increases cell permissiveness to viruses, including hepatitis C virus. LGP2, which is another DExD/H RNA helicase of the RIG-I and MDA5 family and which is devoid of CARD domain, negatively controls IFN induction at different levels: by sequestering dsRNA, by blocking RIG-I’s multimerization in trans through a domain analogous to the RIG-I RD, and by competing with the protein kinase IKKε for a common interaction site on IPS-1. The ability of RIG-I and LGP2 to exert such a feedback control at the earliest steps of IFN synthesis allows the cells to exert a tight regulation of the induction of the innate immune response.

As 2007 celebrates the 50th anniversary of the discovery of interferon (IFN), the molecular mechanisms leading to its transcriptional activation and secretion in response to microbial infection are now being discovered at an impressive pace, revealing, at the same time, an impressive complexity.

IFN is a cytokine, the biosynthesis and secretion of which are increased in cells in response to several stress situations, including viral infection. The rapid synthesis of IFN, its release, and its interaction with different cell types allows the subsequent induction of hundreds of genes involved in antiviral, antiproliferative, and apoptosis-stimulating activities. Together with IFN, the synthesis and secretion of several other cytokines is also stimulated, and all of these factors contribute to the activation of adaptive immunity, therefore allowing the organism to mount an efficient immune response.

The cytoplasmic proteins RIG-I (retinoic acid induced gene I) (1) and MDA5 (melanoma differentiation associated gene 5) (2) are first in line to recognize the invading viruses through the viral RNAs and to initiate IFN transcription and secretion (3, 4). They both contain two caspase-recruitment domains (CARDs) and a DExD/H-box helicase domain (5). RIG-I activates upon recognition of uncapped 5′-triphosphate RNA, which is generated in infected cells by viruses such as influenza, Sendaï virus, vesicular stomatitis virus (VSV), rabies virus, and viruses of the Flaviviridae family, including hepatitis C virus (HCV). In contrast, activation of MDA5, which remains to be more clearly defined, occurs upon recognition of viruses with protected 5′ RNA ends, such as in picornaviruses (6, 7). Through their N-terminal CARD domains, both RIG-I and MDA5, through CARD-CARD homotypic interactions, activate the same downstream partner, a mitochondria-bound protein: IPS-1/VISA/MAVS/CARDIF (Fig. 1). [RIG-I’s mitochondria-bound adaptor has been given four different names according to the various groups who identified it: MAVS, mitochondrial antiviral signaling (8); IPS-1, interferon-β promoter stimulator 1 (9); VISA, virus-induced signaling adaptor (10); CARDIF, CARD adaptor inducing IFN-β (11).] This partner, here referred to as IPS-1, in turn recruits an array of protein kinases through other adaptors, leading to activation of transcription factors required for stimulation of genes encoding IFNs.

Fig. 1.

The RIG-I-mediated pathway of IFN production. In the absence of appropriate stimulation, the RNA helicase RIG-I remains locked in an autorepressed conformation through binding of its extreme C-terminal 190-amino-acid region, referred to as repression domain (RD), to its CARD-containing N terminus and an internal domain located in a linker region between the RNA helicase domains III and IV (cis repression) (A). Upon viral infection (B), viruses presenting a free triphosphate structure at the 5′ end of their RNAs activate RIG-I by binding to its RNA helicase domain (C). This provokes a change in the RIG-I conformation (D) and leads to RIG-I dimerization, allowing it to interact with the mitochondria-bound IPS-1 protein through CARD-CARD homotypic interactions (E). In turn, IPS-1 recruits its downstream partners required for activation of the NF-κB and IRF3 pathways leading to transcription of genes encoding IFN and of the innate immune response (F).

LGP2 is another DExD/H RNA helicase that has domains homologous to RIG-I and MDA5 (12). However, because it lacks the CARD domain, LGP2 is not able to stimulate IFN transcription and plays a negative regulatory role in the RNA helicase pathway through its ability to compete with RIG-I and MDA5 for binding RNAs (5, 12) (Fig. 2).

Fig. 2.

IFN stimulates the expression of hundreds of genes, including those encoding RIG-I and LGP2. Accumulation of LGP2 in the cytosol allows this molecule to prevent further stimulation of transcription of genes encoding IFNs through three different mechanisms: (A) inhibition of the multimerization of RIG-I and its engagement with IPS-1 through an interaction between LGP2 C-terminal domain that is analogous to that of RIG-I (RD-like) (trans repression), (B) competition with RIG-I for binding to the viral RNA, and (C) prevention of access to the IPS-1 downstream partners required for IFN production by binding to the C-terminal domain of IPS-1.

Saito et al. show that the presence of a repressor domain (RD) at the C terminus of RIG-I inhibits RIG-I activation in cis. Interestingly, they show that the presence of an analogous RD at the C terminus of LGP2 allows this protein to inhibit RIG-I activation in trans. This highlights the importance of LGP2 in controlling RIG-I–mediated induction of IFN transcription and secretion (13).

These authors first observed that association of RIG-I with double-stranded RNA (dsRNA) and adenosine triphosphate (ATP), in vitro, conferred limited resistance to trypsin digestion and resulted in the generation of a trypsin-resistant 30-Kd polypeptide. This prompted them to characterize the importance of RIG-I’s conformation for its subsequent interaction with IPS-1. A 190-amino-acid domain located at the extreme C terminus of RIG-I, which they termed the "repressor domain (RD)", exerted a negative effect on RIG-I’s function. Interestingly, RD binds to two different RIG-I domains: the CARD-containing N terminus and an internal domain located in a linker region between the RNA helicase domains III and IV. How does this binding affect RIG-I’s activity? The authors showed that RIG-I dimerizes and that this self-association does not occur through CARD-CARD homotypic interaction, but rather involves both its CARD and C-terminal regions. Furthermore, the RIG-I/RIG-I multimerization step seems to be required before interaction of RIG-I with IPS-1, because the tandem CARDs of RIG-I, which constitutively induces IFN-β promoter activity in wild-type mouse embryonic fibroblasts (MEFs), fails to do so in RIG-I-null MEFs. One hypothesis is that one tandem CARD of RIG-I is binding to an endogenous RIG-I and acquires an activation signal to recruit IPS-1. Strikingly, RD prevented the self-association of RIG-I and prevented the interaction of RIG-I with IPS-1.

Because the RNA helicase domain of RIG-I has sequence similarity with that of MDA5 and LGP2 [35% and 31%, respectively (5)], Saito et al. examined the function of domains with homology to the RIG-I’s RD from these two other RNA helicases. Similarly to the RIG-I RD, the C-terminal domain of LGP2 prevented stimulation of IFN transcription and secretion, whereas the C terminus of MDA5 was inactive in affecting IFN production. In contrast to RIG-I, overexpression of MDA5 in the absence of an RNA activator was found to stimulate IFN production, in accord with previous data (3). It would be of interest to examine whether replacing the C terminus of MDA5 with that of RIG-I would mimic the cis repression reported for RIG-I.

Huh7 cells lines engineered to constitutively express the RIG-I RD (Huh7-RD) did not produce IFN-β in response to Sendaï virus (13). Interestingly, these cells are permissive to HCV infection, in contrast with the poorly permissive parental cells. Previously, only the Huh7.5 clone (14), which expresses a nonfunctional RIG-I protein because of a single mutation in its first CARD domain, efficiently supported HCV infection, at least in models of infection with recombinant HCV particles (1518). Although the viral titers and the in vivo expression of the HCV proteins were lower in Huh7-RD than in Huh7.5, these data nevertheless confirm that RIG-I represents a major threat for HCV. Indeed, HCV viral RNAs are activators of RIG-I (13, 19). The Huh7-RD cells proved also to be more permissive to infection by VSV. The data of Saito et al., therefore, demonstrate the importance of the C terminus of RIG-I in the control of its activation and may lead to new therapeutic applications.

Both the C terminus of RIG-I and of LGP2 conferred the same negative regulation to the pathway leading to IFN synthesis and secretion, either in cis (the RD of RIG-I) or in trans (LGP2). Does this mode of cis and trans regulation occur for other CARD-containing proteins? Saito et al. propose Nod1 and Apaf-1 as two examples. Both these proteins use N-terminal CARD domains to mediate protein association. Nod1 interacts with RICK, a CARD domain-containing kinase, to activate the transcription factor NF-κB (nuclear factor κB) through recruitment of IKKγ (inhibitor of κB kinase γ, also known as NEMO). Apaf-1 interacts with caspase-9 to activate the downstream caspase-3, which contributes to the progression of apoptosis. The C-terminal leucine-rich repeats (LRR) of Nod1 can inhibit the Nod1-mediated activation of NF-κB. However, this mode of inhibition may not occur through a direct interaction with the Nod1 CARD domain (20). The situation with Apaf-1 is more similar to that of RIG-I: The C-terminal WD-40 repeat domain associates in cis with the N-terminal CARD domain (21). Interestingly, caspase-9 and Apaf-1 form a ternary complex with the antiapoptotic protein Bcl-XL (22), and indeed some regions of homology exist between Bcl-XL and the WD-40 repeat–containing domain of Apaf-1 (ClustalW; This may represent another example of cis and trans regulation by similar domains in CARD-containing proteins. Although not a CARD-containing protein, PKR (protein kinase RNA dependent) can also be cited as an example of cis and trans regulation by similar domains. This protein kinase, which is encoded by a gene stimulated by IFN, activates in the presence of dsRNA to regulate the expression of a number of genes. In the absence of dsRNA, its catalytic domain remains in a locked conformation by close association with the most proximal of its two dsRNA-binding domains (DRBD) present at its N terminus (negative cis regulation). Upon binding to dsRNA, PKR changes its conformation, which unmasks its catalytic domain [reviewed in (23)]. Negative trans regulation of PKR is possible through a DRBD/DRBD interaction with TRBP (TAR RNA-binding protein), which is another protein of the family of dsRNA-binding proteins and which also contains two DRBDs (24, 25).

The role of LGP2 in controlling IFN induction was also reported by Komuro and Horvath (26). LGP2 associated with the C terminus of IPS-1 in a region that is also necessary for the recruitment of the IRF3-phosphorylating kinase IKKε (11, 27). In this respect, LGP2 plays a role downstream of RIG-I and therefore gives the cells another opportunity to slow down IFN induction. Therefore, both the Saito et al. and the Komuro and Horvath reports illuminate the importance of LGP2 as a key regulator of the pathway leading to IFN synthesis and secretion, because LGP2 can control the activity of both RIG-I and its mitochondrial downstream partner.

How does the regulation of IFN synthesis and secretion by RIG-I, LGP2, and IPS-1 take place? A model (Fig. 1) can be proposed based on these two reports. Under normal physiological conditions, RIG-I is locked in an inactive conformation through the interaction of its RD with both its RNA helicase and CARD domains. Upon viral infection, RIG-I binds to its viral cognate RNA and undergoes a change of conformation. Its first CARD domain then interacts with the CARD domain of IPS-1. The latter changes its conformation and recruits, at its C terminus, the NF-κB–activating IKKαβ complex and the IRF3-phosphorylating kinases TBK1 and IKKε, through a series of adaptors involving proteins from the TRAF and TANK families, which lead to synthesis and secretion of IFN through an amplification loop (28). Once secreted into the intercellular medium, IFN, in turn, rapidly stimulates the synthesis of hundreds of genes, among which are RIG-I and LGP2 (and also MDA5). Therefore, these RNA helicases accumulate in the cytosol at a time when the virus is still replicating in the cell. Accumulation of RIG-I allows continuous IFN production through its ability to bind the viral RNAs and to interact with IPS-1. Accumulation of LGP2 inhibits both the interaction of RIG-I with IPS-1 and the recruitment to IPS-1 of the complex containing the kinases responsible for stimulation of IFN production. One possible scenario for this inhibition is that LGP2 binds to one of the RIG-I CARD domains using its analogous RD domain and, thus, prevents the multimerization process and the formation of an active complex between RIG-I and IPS-1.

As shown now by Komuro and Horvath, LGP2 also binds a domain at the C terminus of IPS-1 and prevents recruitment of the IRF3-phosphorylating kinase IKKε (26). In accord with the ability of LGP2 to bind both RIG-I (13) and IPS-1, Komuro and Hovarth showed that the three partners can indeed coprecipitate (26). In such a situation, LGP2 would be able to control IFN production at two different stages in the formation of the complex made by RIG and IPS-1.

Negative control of IFN production is an essential physiological process. Once the antiviral and antiproliferative programs or activation of innate immune cells are launched, it is essential for the organism to prevent excess IFN production. For instance, negative regulation of IFN production occurs through the action of specialized proteins, such as SOCS [suppressor of cytokines signaling (29)] or the NF-κB inhibitor A20 (30). In addition to these is LGP2, which negatively regulates two early events in IFN production. Although IFN is a key player in restricting the propagation of invading pathogens, the existence of multiple mechanisms to limit IFN production confirms the need for cells to control the deleterious events that can also be provoked by IFN.

The C terminus of IPS-1 plays an important role in the stimulation of IFN transcription as it participates in the recruitment of the downstream kinases, such as the NF-κB–activating IKK complex and the IRF3-phosphorylating kinase IKKε. One intriguing aspect is that this C-terminal region also contains a cleavage site for the HCV NS3/4A protease (11). Therefore, the same domain of IPS-1 can be targeted by a cellular protein (LGP2) or a viral protein (NS3/4A) with the same outcome, inhibition of IFN production. In view of the privileged association of IPS-1 with the mitochondria, it would be interesting to determine whether it is involved in other physiological processes besides the induction of IFN and innate immunity. The story is still unfolding.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
View Abstract

Stay Connected to Science Signaling

Navigate This Article