The innate immune response to infection is a rapid and highly potent process1 that often involves TLRs. TLRs are an important family of innate sensors, recognizing diverse microbial products and launching signaling pathways that ultimately lead to the clearance of the pathogen from the host and the establishment of a memory response in anticipation of any subsequent attack1. Multiple mechanisms have been found in microbes, notably viruses, that avoid or subvert TLR activation2. How bacteria achieve this is less well understood.

Cirl et al.3 begin to fill in the gaps, introducing an intriguing mechanism used by certain bacteria to prevent TLR signaling. They find that strains of E. coli that infect the kidney, and several strains of Brucella, secrete proteins that are internalized into macrophages and disable TLR signaling by targeting MyD88, the key TLR signaling adaptor. This strategy emphasizes the importance of TLRs for innate immunity in humans and provides insights into bacterial virulence and infectious diseases that might point to new therapies.

The inspiration for the study of Cirl et al.3 was the previous observation that vaccinia virus encodes a protein termed A46R, which interferes with TLR signaling by sequestering MyD88 (refs. 4,5). A46R was shown to contain a domain homologous to the Toll–interleukin-1 receptor (TIR) domain—a domain found in MyD88 and the cytosolic face of each TLR6. The interaction of the TLR and MyD88 TIR domains initiates signal transduction, and the TIR domain of A46R seems to interfere with this by interacting with MyD88. There has also been a report of a protein termed TIR-like protein A in Salmonella enteritica serovar Enteritidis that also impairs TLR- and MyD88-mediated signaling and, in doing so, promotes intracellular bacterial accumulation7.

Given these observations, Cirl et al.3 screened bacterial genomes for proteins similar to the Salmonella protein. They found such a protein, dubbed TcpB in Brucella melitensis, that causes brucellosis (characterized by fever and muscle pain) and another, TcpC, in the uropathogenic E. coli strain CFT073. Structural modeling confirmed the similarity of the proteins' TIR domain to the mammalian TIR domain.

To test the function of TcpC, they deleted it from E. coli CFT073 and then used this strain to infect a macrophage cell line and a uroepithelial cell line. Infection with the mutant resulted in a stronger induction of the proinflammatory cytokines tumor necrosis factor (TNF) and interleukin-6. TcpC was also shown to facilitate the intracellular survival of E. coli CFT073.

The authors also transfected cells with genes encoding TcpC or TcpB3. These genes blocked activation of the key inflammatory transcription factor NF-κB by the Gram-negative bacterial product lipopolysaccharide (LPS), which acts via TLR4. Neither Tcp protein affected the NF-κB signal when activated by receptors that do not signal via MyD88, pointing to the proteins' specificity. Importantly, TcpC was shown to bind MyD88 directly. All of these results clearly indicate that certain strains of E. coli and Brucella have TIR domain–containing proteins that target MyD88 to limit innate immunity.

These molecular observations seem to be relevant for virulence. To examine this question, the authors used a mouse model of acute kidney infection. In mice infected with wild-type E. coli CFT073, a much higher bacterial burden and a greater likelihood of kidney abscesses were observed relative to mice infected with the mutant strain lacking TcpC. Most interestingly, E. coli strains were isolated from the urine of people with kidney infections (acute pyelonephritis), bladder infections (acute cystitis) or asymptomatic bacteriuria, which represent decreasingly severe forms of human urinary tract infections (UTIs). Forty percent of acute pyelonephritis isolates contained bacteria with TcpC, with these bacteria being less common in cystitis (21%) and asymptomatic bacteriuria (16%). These results suggested that TcpC is associated with enhanced virulence, given its association with the clinical severity of UTIs in humans.

How might these extracellular bacteria get all of this to work inside the host cell? The authors addressed the issue of TcpC secretion3. Low pH led to the release of TcpC from E. coli CFT073. It was also taken up by macrophages and shown to impair TNF induction by various TLR ligands. Finally, a drug that can inhibit the efflux pump in E. coli, called phenylalanine-arginine-β-naphtylamide (PAβN), blocked the ability of E. coli CFT073 to inhibit TNF production from macrophages and prevented the secretion of TcpC.

The findings suggest that TcpC increases the severity of UTIs in humans, which is consistent with the important role of TLR4 in host defense in the urinary tract8. With TcpC, the bacteria can get a foothold in the host, replicate, and cause tissue injury and disease—providing the first clear evidence that human bacterial pathogens target TLR signaling in order to survive and spread (Fig. 1).

Figure 1: Bacterial interference with TLR signaling.
figure 1

Kim Caesar

(a) Gram-negative bacteria such as E. coli secrete LPS, which is sensed by TLR4 on macrophages. Signaling is then initiated from the plasma membrane by the adaptor proteins Mal and MyD88 (ref. 12), leading to induction of proteins important in host defense (such as TNF) and ultimately to the clearance of the pathogen. (b) Cirl et al.3 find that in the uropathogenic strain of E. coli CFT073, a protein termed TcpC is secreted and taken up by cells, where it targets MyD88 and prevents signaling. TcpC limits the host defense response, allowing the pathogen to get a foothold, spread and cause pathology. Targeting this process may be useful therapeutically.

A number of issues arise from this study. In viral infections, it's easy to envisage how the decoy protein gets to its target, because the protein will be made in the cytosol of the infected cell. For bacteria such as E. coli that are extracellular, at least initially, the decoy must be secreted and then taken up by the target cell. The mechanism of this uptake is still not clear, but may involve the decoy's interaction with lipid rafts.

How widely used is this mechanism in pathogenic bacteria? Cirl et al.3 report that several other human pathogens contain proteins similar to TcpB and TcpC, notably Brucella suis, Brucella abortus and the Staphylococcus aureus strain MSSA476. It is therefore possible that these bacteria, and perhaps others yet to be described, also avoid innate responses by decreasing the TLR response.

Finally, how useful might these findings be for developing new treatments for bacterial infections—or for inflammatory diseases involving TLRs or interleukin-1 (which also signals via MyD88)9,10? For bacterial infections, as explored by the authors, agents such as PAβN that prevent secretion of TcpC could be useful adjuncts to antibiotics. Moreover, bacteria deficient in Tcps may have prospects as new vaccine candidates, as activation of TLRs clearly has potent adjuvant activity11. And as for the inflammatory diseases, Tcps might provide leads for the development of selective inhibitors of MyD88, which could have utility as anti-inflammatory agents.