Innate immune responses yield tissue-specific bottlenecks that scale with pathogen dose

Abstract

To cause infection, pathogens must overcome bottlenecks imposed by the host immune system. These bottlenecks restrict the inoculum and largely determine whether pathogen exposure results in disease. Infection bottlenecks therefore quantify the effectiveness of immune barriers. Here, using a model of Escherichia coli systemic infection, we identify bottlenecks that tighten or widen with higher inoculum sizes, revealing that the efficacy of innate immune responses can increase or decrease with pathogen dose. We term this concept "dose scaling". During E. coli systemic infection, dose scaling is tissue specific, dependent on the LPS receptor TLR4, and can be recapitulated by mimicking high doses with killed bacteria. Scaling is therefore due to sensing of pathogen molecules rather than interactions between the host and live bacteria. We propose that dose scaling quantitatively links innate immunity with infection bottlenecks and is a valuable framework for understanding how the inoculum size governs the outcome of pathogen exposure.

Figure 1.. Bottleneck-dose response analysis for E. coli systemic infection.

A-B) CFU (A) and FP (Ns) (B) are shown for the liver as a function of dose. Each point represents an animal and the blue box indicates animals that developed abscesses. C-F are identical to A-B but represent spleen and lung CFU (C and E) and FP (D and F). Best fit lines from linear regression in FP plots are shown with 95% confidence bands. G) Dose-FP plots for different scaling patterns. Dotted lines represent a 0% bottleneck. With slopes greater than one, fractionally more bacteria survive to become founders at higher doses, suggesting that innate immune responses are less effective at higher inoculum sizes, and therefore “negatively scale” with dose. With slopes less than one, the immune response is “positively scaled”, since fractionally fewer bacteria survive (more are killed) at higher doses. With a slope equal to 1, a fixed fraction of the inoculum survives host bottlenecks.

Figure 2.. Tissue and TLR4 dependent responses to spike-in of killed bacteria

A) Barcoded live bacteria or the same quantity of live cells plus 40-fold excess of formalin-fixed bacteria were IV inoculated into TLR4^Het or TLR4^KO littermates. B-C) CFU (B) and FP (Ns) (C) of the liver are shown (lines represent medians). The blue box indicates animals that developed abscesses. D-G are identical to B-C but represent spleen and lung CFU (D and F) and FP (E and G). H) Curves from Figure 1 (left to right: liver, spleen, and lung) are schematized for reference