Here, we show that HAV disrupts TLR3 signaling by targeting the essential adaptor protein TRIF for degradation by the 3CD protease-polymerase processing intermediate. The ability of poly-I:C to stimulate the IFN-β promoter or induce the expression of ISGs when added to media was markedly attenuated in HAV-infected Huh7 hepatoma cells in which TLR3 expression had been reconstituted by retroviral gene transduction (). This disruption of TLR3 signaling was associated with a loss of detectable TRIF (), and could be recapitulated by ectopic expression of 3ABCD or 3CD in both Huh7-TLR3 cells and HeLa cells which possess an endogenous TLR3 signaling pathway (). The loss of TRIF expression was linked to the cysteine protease activity residing within the 3C sequence of 3CD, which we demonstrate cleaves TRIF sequentially at two noncanonical 3Cpro cleavage sites (). Additional studies suggested that this is due to the ability of the 3D sequence in 3CD to alter the substrate specificity of the protease such that it better accommodates the acidic residues present at the P4 position of cleavage sites in TRIF (). These observations add to our understanding of the pathogenesis of HAV, a significant human pathogen that has received scant attention in recent years.
In previous work, we demonstrated that HAV also antagonizes the induction of IFN responses by the cytosolic RLR pattern recognition receptors, RIG-I and MDA-5, by inducing proteolysis of the adaptor protein MAVS 
(). As we report here with poly-I:C-induced TLR3 signaling, we found that ectopically expressed 3ABCD was capable of disrupting Sendai virus-induced RIG-I signaling. 3ABCD results from secondary processing of the HAV polyprotein at the P2-P3 junction, and is itself subject to further processing via two distinct pathways, one leading to production of 3ABC and the other to 3CD (). Both intermediates contain the catalytically active 3Cpro
cysteine protease domain, but they have distinct cellular localization and substrate specificities (). 3ABC, due to the presence of a mitochondrial targeting transmembrane domain in 3A, localizes to the mitochondrial membrane where it cleaves MAVS (). In contrast, 3CD appears to be localized primarily to the perinuclear ER, and its ability to cleave TRIF is dependent upon its unique substrate specificity rather than its intracellular localization ().
Interferon-activating pathways disrupted during HAV infection by 3Cpro precursor-mediated proteolysis of signaling adaptor proteins.
Our results reveal an unexpected role of the 3D sequence in modulating the substrate specificity of 3CD. 3Cpro
cleavage sites within the HAV polyprotein, as well as MAVS, contain a hydrophobic amino acid (Leu, Ile, or Val) at the P4 position 
that fits into the hydrophobic S4 binding pocket within the crystal structure of 3Cpro 
. In contrast, both 3CD cleavage sites within TRIF contain an acidic amino acid residue (Asp-190 and Glu-551) at the P4 position (), and therefore do not conform to the canonical cleavage sequence. A previous study showed that a peptide substrate with a Glu substitution (underlined) at the P4 position, Ac-EE
SFS-NH2, which is similar to the TRIF cleavage site EQSQ554
, was not cleaved by 3Cpro 
. Our data suggest that 3CD possesses a unique substrate specificity that allows it to recognize and hydrolyze cleavage sites within TRIF that are otherwise relatively resistant to 3Cpro
. In support of this notion, we showed that changing the non-canonical cleavage site at Gln-190 of TRIF to a canonical 3Cpro
cleavage sequence resulted in efficient 3Cpro
proteolysis and a reversal of the order of cleavage at the two sites in TRIF (.5D). A similar change at the Gln-554 site did not make it fully permissive for 3Cpro
cleavage, however, suggesting that there are other differences in the substrate specificities of 3Cpro
and 3CD. Our data indicate that the change in substrate specificity of 3CD is conferred in cis
by the 3D sequence (), although the structural basis for this remains to be determined.
In addition to their differentiated roles in evading innate immune responses, 3ABC and 3CD are likely to have specialized roles in the viral life cycle. 3ABC is a stable intermediate that is important in processing of the P1-2A segment of the polyprotein required for assembly of the viral capsid 
. 3CD, based on studies with other picornaviruses, may play a role in the uridylyation of the protein primer of RNA synthesis, 3B (VPg) 
. The multiple functions of these viral proteins reflect a strategy used by picornaviruses to create processing intermediates that are functionally distinct from their mature products 
The dual targeting of RLR and TLR3 signaling by HAV 3ABCD processing intermediates is reminiscent of the HCV NS3/4A protease, which disrupts both RIG-I and TLR3 signaling pathways by proteolytically cleaving the same signaling adaptor proteins, MAVS and TRIF, respectively 
and NS3/4A are both chymotrypsin-like proteases with double β-barrel folds 
, but they are not closely related phylogenetically. The HAV 3Cpro
protease has a cysteine nucleophile in its active site, while NS3/4A has a serine. These viral proteases have very different substrate specificities, and they cleave MAVS and TRIF at distinctly different sites [7,8,9,12, and ]. The fact that both of these hepatotropic viruses express proteases targeting these two critical adaptor molecules is thus a remarkable example of convergent evolution. It also speaks strongly to the importance of these signaling pathways in the control of RNA viruses in the liver. However, since HAV infection is always successfully controlled by the host (except in rare cases of fulminant disease), these data indicate that the disruption of RLR and TLR3-mediated antiviral defenses is not sufficient for a virus to establish the longterm persistence that typifies most HCV infections. HCV must possess additional immune evasion strategies to account for its unique capacity to establish chronic infections.
We demonstrated a minimal gain of permissiveness for HAV replication in hepatocyte-derived cells in which TLR3 or TRIF expression was depleted (), and a reduction in viral antigen expression in hepatoma cells with active TLR3 signaling (). However, these effects were modest, potentially reflecting very efficient control of TLR3 signaling by 3CD in infected cells such that TLR3 has little impact on viral replication. Alternatively, it may be that the primary advantage gained by HAV in antagonizing TLR3 signaling is impaired production of proinflammatory cytokines and reduced inflammation associated with the infection. TLR3 signaling is critically important to murine host defense against coxsackievirus B, another picornavirus 
, and it is plausible that the disruption of TLR3 signaling has significance beyond impairing the type I IFN response.
The subversion of both RLR and TLR3 signaling likely contributes to the relatively lengthy, clinically silent incubation period that precedes acute liver injury in hepatitis A. This period is characterized by robust viral replication within the liver and shedding of virus in feces, which reaches a maximum at the onset of hepatic inflammation 
. The absence of a type I IFN response in acute infectious hepatitis was hinted at in clinical studies done almost 40 years ago 
. Consistent with this, we recently documented a paucity of type I IFN-dependent ISG expression (e.g.
, IFIT-1, ISG15) within the liver of HAV-infected chimpanzees during the first weeks of infection despite high viral RNA copy numbers 
. The cleavage of MAVS and TRIF by 3ABC and 3CD, respectively, provides a partial mechanistic explanation for this. By dealing a double blow to two major cellular antiviral response pathways, HAV appears able to block somatic cell expression of IFN-α/β, thus facilitating its replication. Yet to be explained is how it evades recognition by plasmacytoid dendritic cells (pDCs), which may play a significant role in sensing HCV infection in the liver and generating the strong intrahepatic ISG responses that are often observed in acute and chronic hepatitis C