Specificity and Heterogeneity of Trained Immunity in Hematopoietic Stem and Progenitor Cells

When innate immune cells and hematopoietic stem and progenitor cells (HSPCs) are exposed to pathogenic agents, they develop heightened responses to subsequent infection through a process called trained immunity. After exposure to a pathogen, bone marrow derived macrophages (BMDMs) generated from trained HSPCs are capable of enhanced pathogen clearance and cytokine production, while exhibiting persistent metabolic rewiring. Because some models of HSPC trained immunity are dependent on interferon gamma (IFNγ) signaling, and our lab has described extensive changes in HSC self-renewal and differentiation upon IFNγ exposure, we hypothesized that persistent IFNγ signaling induced by chronic infection results in reprogramming of HSPCs, causing improved non-specific immunity. To test our hypothesis, we generated a chimeric mouse model of trained immunity by transplanting control or M. avium-exposed HSPCs into naïve recipient mice. Mice that received M. avium trained HSPCs had decreased bacterial load, less splenomegaly, and fewer granulomas upon subsequent M. avium infection, indicating improved immunity. Furthermore, BMDMs generated from mice trained with a single dose of recombinant IFNγ (rIFNγ) exhibited increased pathogen clearance and metabolic profiles ex vivo. To test if rIFNγ training was sufficient to induce HSPC trained immunity in vivo, we isolated HSPCs from rIFNγ-exposed mice and challenged the recipients 4 months later. These in vivo experiments demonstrated that rIFNγ training was insufficient to induce substantial protection against subsequent M. avium challenge, but still induced BMDM metabolic rewiring four months post HSPC training. Collectively, our studies indicate that there are degrees of training that occur upon IFNγ exposure, likely related to the concentration and duration of the primary stimulus.

To assess the specificity of cross protection of HSPC trained immunity, we utilized our chimeric mouse model and tested two different training and infection pathogens: M. avium and influenza. When we challenged M. avium-trained HSPC recipients with influenza, we found that although there was mildly decreased lung histopathology and increased production of IFNγ and TNFα, mice succumbed to infection like untrained controls. When we swapped the order of pathogens, we observed that mice receiving influenza-trained HSPCs produced BMDMs with increased killing capability and systemically higher IL-6 and RANTES levels, but these features were insufficient to significantly reduce bacterial CFU counts upon M. avium challenge. These transplant experiments indicate that trained immunity encoded in HSPCs is pathogen specific.

To dissect the mechanism of M. avium-induced trained immunity in HSPCs, we performed RNAseq analysis on M. avium-trained HSPCs post-transplant and cross referenced it with RNAseq and WGBS data on primary M. avium-exposed HSPCs. These studies showed consistent differences in cellular signaling, metabolism, immunity, and antigen processing and presentation in the trained HSPCs, indicating that epigenetic and transcriptional reprogramming induced by M. avium exposure is durable following transplant and secondary challenge.

To ascertain whether transcriptional changes are homogeneous throughout the HSC compartment, we completed scRNA-seq on naïve and M. avium-exposed hematopoietic cells. We found that genes upregulated upon M. avium exposure in HSCs, including Batf2 and Cxcl9, were induced in a subset of HSPCs, indicating that there is a heterogeneous response to training within the HSPC pool. Strikingly, the trained immunity signature was maintained in neutrophils and macrophages but lost in mature B cells, indicating specific propagation of genetic signatures induced by training in certain lineages. Finally, we found an emerging population of HSCs with B cell gene signatures upon M. avium exposure. Emergence of this HSPC subpopulation may suggest the development of a cell that acts as a direct intermediate between HSC and B cells following training.

Our work shows that trained immunity induced by M. avium and persistent IFNγ signaling is pathogen-specific and heterogeneous among primitive HSPCs. Emergence of specific responder cell populations within the HSPC pool may be responsible for enhanced protection against specific infection stimuli, whereas the presence of non-responders may insure long term health of the HSPC pool.