Hematopoietic Stem Cells Undergo Immune Training and Constitute a Long-Term Reservoir for Hyper-Inflammatory Macrophages in a Mouse Model of Chronic Autoimmunity

Auto-immune diseases (AD) are characterized by repeated flares of disease activity separated by periods of remission. Cycles of AD remission and relapse can occur even with therapeutic intervention and contribute to AD morbidity. Paradoxically, during remission myeloid cells retain increased expression of genes related to interferon signaling and antigen presentation. Given the relatively short lifespan of myeloid cells, these observations imply the existence of a clonal reservoir fueling AD relapse. Recent literature describes hematopoietic stem cells (HSC) as a cellular source for trained myeloid cells in response to immune stimuli. Thus, we hypothesize that HSC may also retain a long-term memory of chronic autoimmune inflammation, thereby providing a continuous supply of myeloid cells that promote AD pathogenesis.

To test this hypothesis, we have combined high-throughput molecular and cytokine profiling approaches with functional assays to address heritable changes in immune function using the pristane-induced mouse model of systemic lupus erythematosus (SLE). Eight weeks after pristane injection, we observe significant myeloid lineage expansion in the bone marrow (BM), including mature myeloid cells, granulocyte/monocyte progenitors (GMP) and multipotent progenitor (MPP) populations. To understand the impact of SLE-like disease on the molecular programming of the hematopoietic system, we performed RNA-seq analyses of BM Mon and HSC. As anticipated, BM Mon activated inflammatory programs and antigen presentation genes, which overlapped with gene signatures of human monocytes from SLE patients. HSC also activated innate defense gene programs resembling signatures of trained immunity, thus establishing the potential for autoimmune inflammation to induce immune training in HSC.

To establish whether these molecular programs potentiate myeloid cell function, we generated BM-derived macrophages (BMDM) from control and pristane-induced mice. BMDMs from pristane-induced mice exhibited significantly increased capacity to kill Mycobacterium avium. Further, co-culture of T cells with BMDMs from pristane-induced mice significantly boosted T cell proliferation, indicative of enhanced antigen presentation. To establish whether HSC from pristane-induced mice propagate molecular memory of SLE-like disease to myeloid progeny, we transplanted stringently enriched (LSK/SLAM/CD34 ^-/EPCR ⁺) long-term (LT)-HSC into lethally irradiated recipient mice. 18 weeks post transplant, mice transplanted with LT-HSC from pristane-induced donors had a small but significant reduction in donor BM HSC chimerism but did not exhibit overt changes in lineage output. Strikingly, BMDMs from pristane-induced donors showed increased bacterial killing and inflammatory cytokine generation following M. avium challenge, as well as increased capacity to induce antigen-specific T cell proliferation. Thus, LT-HSC retain and pass on altered functional properties to myeloid cells, even in the absence of AD activity.

To characterize the molecular mechanisms underlying HSC-propagated alterations in myeloid cell function, we ran RNA-seq on donor-derived GMP and BMDMs. Notably, GMP derived from pristane-induced donor LT-HSC had increased expression of Fos and Jun/b/d, key molecular drivers of immune training in stem cells. Likewise, pristane-induced donor BMDMs maintained increased expression of IFN-regulated genes including MHC-I, a gene also overexpressed in PBMCs from human SLE patients. Consistent with these findings, re-stimulation of recipient mice with pristane led to a significant increase in cytokine-producing lymph node T cells versus recipient mice transplanted with control HSC, further supporting a model of immune training.

Overall, these data show that chronic autoimmune inflammation can induce in HSC a heritable trained immunity phenotype that is transmitted to myeloid progeny, enhancing their functional activity. Ongoing studies are establishing the capacity for this phenotype to exacerbate AD. We are also testing the importance of molecular players identified above in establishing AD-related immune training and assessing the potential for therapeutic interventions to disrupt HSC memory in this setting. Our data thus stand to establish a new paradigm for trained immunity in HSC as a key contributor to AD pathology and relapse.