Figure 3 of this Letter contains an inadvertently duplicated panel: the PBS 30 panel is identical to the aGalCer panel (top right). The corrected panels are shown here. Our results and conclusions are unaffected by this oversight. CORRIGENDUM

In humans, ten Toll-like receptor (TLR) paralogues sense molecular components of microbes, initiating the production of cytokine mediators that create the inflammatory response. Using N-ethyl-N-nitrosourea, we induced a germline mutation called Lps2, which abolishes cytokine responses to double-stranded RNA and severely impairs responses to the endotoxin lipopolysaccharide (LPS), indicating that TLR3 and TLR4 might share a specific, proximal transducer. Here we identify the Lps2 mutation: a distal frameshift error in a Toll/interleukin-1 receptor/resistance (TIR) adaptor protein known as Trif or Ticam-1. Trif(Lps2) homozygotes are markedly resistant to the toxic effects of LPS, and are hypersusceptible to mouse cytomegalovirus, failing to produce type I interferons when infected. Compound homozygosity for mutations at Trif and MyD88 (a cytoplasmic TIR-domain-containing adaptor protein) loci ablates all responses to LPS, indicating that only two signalling pathways emanate from the LPS receptor. However, a Trif-independent cell population is detectable when Trif(Lps2) mutant macrophages are stimulated with LPS. This reveals that an alternative MyD88-dependent 'adaptor X' pathway is present in some, but not all, macrophages, and implies afferent immune specialization.

Several subsets of dendritic cells have been shown to produce type I IFN in response to viral infections, thereby assisting the natural killer cell-dependent response that eliminates the pathogen. Type I IFN production can be induced both by unmethylated CpG-oligodeoxynucleotide and by double-stranded RNA. Here, we describe a codominant CpG-ODN unresponsive phenotype that results from an N-ethyl-N-nitrosourea-induced missense mutation in the Tlr9 gene (Tlr9 CpG1 ). Mice homozygous for the Tlr9 CpG1 allele are highly susceptible to mouse cytomegalovirus infection and show impaired infection-induced secretion of IFN-␣͞␤ and natural killer cell activation. We also demonstrate that both the Toll-like receptor (TLR) 9 3 MyD88 and TLR3 3 Trif signaling pathways are activated in vivo on viral inoculation, and that each pathway contributes to innate defense against systemic viral infection. Whereas both pathways lead to type I IFN production, neither pathway offers full protection against mouse cytomegalovirus infection in the absence of the other. The Tlr9 CpG1 mutation alters a leucine-rich repeat motif and lies within a receptor domain that is conserved within the evolutionary cluster encompassing TLRs 7, 8, and 9. In other TLRs, including three mouse-specific TLRs described in this paper, the affected region is not represented. The phenotypic effect of the Tlr9 CpG1 allele thus points to a critical role for TLR9 in viral sensing and identifies a vulnerable amino acid within the ectodomain of three TLR proteins, essential for a ligand response.