Does the canonical model give a complete picture of the innate immune response to bacterial invasion?

NIAID researchers found that S. aureus toxins directly attract leukocytes from the bloodstream after invasion. This precedes the canonical response by several hours and is dependent on the transcription factor EGR1, which the researchers found has a significant impact on S. aureus skin infection.

Like Comment
Read the paper

Our group at NIAID is interested in host-pathogen interaction, focusing on staphylococci. In 2008, we moved from Rocky Mountain Laboratories in Hamilton, MT to the NIH’s main campus in Bethesda. Kevin Rigby, a postdoc in my lab, had been working on phenol-soluble modulins (PSMs), a class of staphylococcal peptide toxins we had just discovered in Staphylococcus aureus some years earlier. Kevin was specifically interested in the degree and type of gene expression changes that PSMs stimulate in neutrophils, and staying back in Montana, finished this work in the lab of Dr. DeLeo in Hamilton. However, with the move of my group to Bethesda, the project was not continued until four years later, when Thuan Nguyen joined the lab first as a postbac IRTA and then continued his studies as a PhD student in a Graduate Partnership Program with Georgetown University.

Thuan was supported by Dr. Gordon Cheung, a Staff Scientist in my group. Together with Dr. Olena Kamenyeva, a Staff Scientist from NIAID’s Biological Imaging Section, Thuan and Gordon made an astounding discovery when, using intravital imaging of S. aureus skin infection in mice, they realized that leukocyte influx to the site of infection was delayed by several hours when mice were infected with a PSM-negative as compared to an isogenic S. aureus wild-type strain. We soon realized the importance of this finding, which meant that secreted toxins stimulate a response that considerably precedes the stimulation of leukocyte influx via interaction of conserved bacterial surface structures with resident skin cells  - the canonical model of the innate immune response to bacterial invasion.


The figure shows our initial data obtained with two-photon intravital imaging in mice that were infected with S. aureus strain LAC (USA300), its isogenic PSM-deficient mutant, or a PBS control, and received bone marrow-derived cells by adoptive transfer from transgenic mice that constitutively express red fluorescent protein. The 3-h time point of a movie taken from 1 to 6 hours post infection is shown. Red, leukocytes. Green circle, area of bacterial infection. The entire movie is at

Later, for the data shown in the paper, we switched to confocal microscopy using a mouse strain that produces green fluorescent neutrophils due to a GFP protein encoded downstream of the lysozyme M promoter. This method allowed for more repeats and also did not lead to early saturation of the fluorescence signal.

Thuan, Gordon, and Olena then optimized the intensely laborious intravital imaging model, allowing for higher throughput and statistical evaluation. Furthermore, we analyzed Kevin’s earlier findings, which we soon realized pointed to a key role of the transcription factor EGR1 in the neutrophil response to PSMs. Using mice deficient in EGR1 and mice deficient in the PSM receptor FPR2, the team established the pivotal importance of the PSM-FPR2-EGR1 axis in early leukocyte attraction to the infection site.

At that point, we realized that a central question left to be answered was whether PSMs secreted by S. aureus attract leukocytes from the bloodstream directly or via stimulation of resident skin cells. To answer that question experimentally was extremely difficult, particularly as PSMs cannot be measured in the blood or tissue at the low concentrations that are pro-inflammatory. With the help of Daniel Barber from NIAID’s Laboratory of Parasitic Diseases, we designed an adoptive transfer experiment that we believed would help decide whether resident cells are involved, based on the dependence of the effect on FPR2 and EGR1.

We all anticipated that the experiment would turn out to show at least a partial involvement of resident cells, because the alternative almost seemed too “revolutionary”. The experiment was based on adoptively transferring 1:1:1 mixtures of bone marrow-derived leukocytes from wild-type, EGR1-/- and FPR2-/- mice (leukocyte preparations from each donor were labeled with different fluorescent dyes), into recipient mice of those three different genotypes and monitoring influx to the infection site. Quite surprisingly, the results clearly ruled out an involvement of resident cells, because they showed importance of the genotype of the transferred cells, but not of that of the recipient mice, and thus indicated a mechanism of direct leukocyte attraction by the secreted PSM molecules.


An example picture of the adoptive transfer experiment illustrating the increased influx of wildtype leukocytes (magenta) as compared to leukocytes from EGR1-/- (blue) and FPR2-/- (green) mice to the infection site (here in a wildtype recipient mouse). Cyan, bacteria. The picture is taken from a movie spanning 2 hours (6 to 8 hours post infection) available at

Finally, showing that the mechanism we discovered contributes to the control of S. aureus infection in the long run was also not easy. This is because PSMs not only serve as signals for the host to launch a swift counterattack to S. aureus, as we demonstrated in this study, but they are also cytolytic toxins that strongly promote skin infection – the latter likely being one of the reasons the host developed ways to recognize their presence. With the analysis of a PSM-deficient mutant thus not being informative in answering the question about the importance of the newly discovered mechanism for infection control, we examined whether EGR1, as a necessary component of PSM-triggered leukocyte influx (but not PSM-mediated cytolysis), has an impact on skin infection. Mice deficient in EGR1 developed much larger abscesses than wild-type mice, indicating a strong contribution of EGR1 to the control of S. aureus infection. While we cannot rule out that EGR1 impacts S. aureus infection also via other pathways, additional results achieved with mice that were infected with a PSM-deficient mutant S. aureus suggested at least a considerable contribution of PSM stimulation.

In conclusion, our study shows that the canonical model of the innate immune response to bacterial invasion cannot account completely for the events that contribute to the recognition of invading S. aureus, for which we demonstrated an additional, pathogen-specific, fast high-alarm mechanism. With PSMs also being produced by other staphylococcal species, the breach of other staphylococci, such as S. epidermidis, through the skin presumably triggers a similar response. However, that response is likely reduced, due to the lower production of strongly pro-inflammatory PSMs in those species; and it is tempting to speculate that the PSM-recognizing receptor has evolved to distinguish strongly pathogenic from less harmful staphylococci. Whether there are similar fast-response mechanisms based on the recognition of toxins produced by other potentially harmful skin bacteria is currently not known.








Michael Otto

Senior Investigator, NIAID/NIH