Search
Search articles by title, abstract,author or doi
Advanced search

Central inflammatory memory in human HSCs

Ailin Han , Richard A. Flavell

Vita ››

PDF (2050KB)
Vita > Cutting Edge > DOI: 10.15302/vita.2026.08.0062
Vita Open AccessPublished:

Central inflammatory memory in human HSCs

Author information +
History +
Vita () Cite this article
PDF (2050KB)

Hematopoietic stem cells (HSCs) sustain lifelong blood production while repeated inflammatory insults throughout life can durably reprogram HSCs. In a recent Nature article1, Zeng et al. identified a human HSC subpopulation retaining memory of inflammatory stresses and demonstrated the association of the HSC inflammatory memory program with multiple human diseases including clonal hematopoiesis (CH), sickle cell disease (SCD), and severe coronavirus infection.

登录浏览全文

4963

注册一个新账户 忘记密码

Immunological memory has long been considered a hallmark of adaptive immunity, allowing T and B cells to mount faster and more robust responses upon repeated exposure to the same antigen. The innate immune system, evolutionarily more ancient than adaptive immunity, can also provide enhanced protection against reinfection, leading to the concept of innate immune memory or trained immunity2. HSCs, which give rise to both innate and adaptive immune cells, have been shown to mediate the protective memory response in mice challenged intravenously with Bacillus Calmette–Guérin (BCG) vaccine3, broadening the concept of immunological memory to central trained immunity.
Throughout life, inflammatory stresses caused by infections can durably reprogram HSCs, reshaping hematopoiesis and influencing life-long immune health. Understanding how human HSCs retain long-term inflammatory memory has broad implications for human health and diseases, yet remains a fundamental challenge because mechanistic studies in humans are limited. Epidemiological studies in BCG vaccine and coronavirus infection cases indicate that human HSCs or hematopoietic stem and progenitor cells (HSPCs) can adapt their epigenetic and metabolic programs to establish central trained immunity4,5. However, HSCs and HSPCs are heterogeneous, and it remains largely unknown whether those subpopulations have distinct memory capacities. Moreover, most studies of central trained immunity focused on a single acute inflammatory challenge, primarily revealing a protective outcome. In contrast, humans are repeatedly exposed to inflammatory insults throughout life, which may instead induce maladaptive inflammatory memory. Defining the underlying memory programs is therefore warranted.
Mechanistic and functional studies of human HSCs are limited by their rarity in peripheral blood and poor ex vivo maintenance capacity. To overcome these challenges, the authors used cord blood (CB) samples and humanized mouse models that enable human HSC engraftment and repeated inflammatory challenges. First, heterogeneity of CB HSCs was revealed through single-cell RNA-seq (scRNA-seq) and chromatin accessibility analysis at homeostasis. Two distinct HSC states, characterized by quiescent and inflammatory signatures respectively, were identified, suggesting inflammatory priming in HSCs. To investigate functional changes in HSCs following single or repeated inflammatory stress, the authors engrafted NSG mice with human HSCs and exposed them to one or two rounds of TNF or LPS challenge (Fig. 1). With a single inflammatory challenge, HSC numbers acutely decreased 16 h post-treatment, but fully returned to a baseline if the mice were given 2.5 months of recovery time. In contrast, two rounds of LPS or TNF challenges drove persistent reduction in total human immune cells. Two TNF challenges also decreased HSC numbers even with 2.5-month recovery time. Despite the persistent reduction in humanization level in the primary xenograft mice upon repeated inflammatory stresses, the HSCs obtained in the engrafted mice did not show decreased repopulation capacity, as shown by secondary transplantation. Overall, these humanized mouse studies suggest that HSCs undergo long-lasting changes following repeated inflammatory stresses.
To understand the molecular mechanism of long-lasting reduction in humanization capacity in response to inflammatory stresses, the authors conducted scRNA-seq and scATAC-seq (scMultiome) analysis of the HSPCs isolated from the inflammation-recovery humanized mice. Through scMultiome, they identified two distinct populations of HSCs: HSC-I and HSC-II. Both HSC-I and HSC-II showed a common long-term HSC (LT-HSC) gene set expression, including marker genes like MECOM, AVP, and PREX2. Moreover, HSC-II also specifically exhibited quiescent HSC signatures, while highly expressing a hallmark gene set of TNF-via-NF-κB signaling. Repeated inflammatory stresses increased the differences between HSC-I and HSC-II. Remarkably, HSC-II showed much more significant changes in transcriptome and epigenome than HSC-I responding to repeated inflammatory stresses, suggesting that HSC-II is the subpopulation retaining inflammatory memory. Analysis of TF activity inferred from the scMultiome data showed that, within the HSC-II population, repeated TNF challenges increased the activity of NF-κB, JUNB, REL and FOSL1, whereas repeated LPS challenges increased the activity of MAZ, RXRA, HMGA1 and SPI1. The authors further defined an HSC inflammatory memory (HSC-iM) program using the top 200 marker genes and 3,663 differentially accessible regions of HSC-II, which enabled them to compare their findings with the existing data of various human diseases: the HSC-iM program was enriched in HSCs isolated from patients recovered from severe COVID-19, patients with SCD, and the elderly population. HSC-iM program also predicted increased 5-year all-cause mortality, underscoring its broad clinical relevance across inflammatory and age-associated human diseases.
Repeated inflammatory stresses drive HSC aging. Clonal hematopoiesis (CH), the age-associated expansion of HSC clones carrying somatic mutations6, increases the risk of hematologic malignancy by 0.5–1% annually. Notably, the most common CH mutations, including DNMT3A and TET2, are shared between myelodysplastic syndrome (MDS) and acute myeloid leukemia (AML)7, establishing CH as a bona fide premalignant state. The authors performed TARGET-seq8 of HSPCs isolated from 9 CH cases and 4 age-matched controls. The HSC-iM signature was highly correlated with both aging and CH clone size. Interestingly, in CH cases, both the HSCs bearing the mutations (in TET2 or DNMT3A or both, CH-mut) and those not (CH-WT) showed a higher HSC-iM signature than the HSCs isolated from control donors, suggesting that inflammatory memory is not confined by mutation, but may play an orthogonal role with somatic mutations to promote CH development. Indeed, despite both CH-WT and CH-mut HSCs exhibiting an elevated HSC-iM signature, their downstream functional programs diverged: compared with CH-WT, CH-mut HSCs downregulated the expression of quiescence pathway and increased lineage commitment. Therefore, CH mutations have a huge impact on HSC-iM differentiation, potentially leading to increased myelopoiesis.
The authors identified a novel human HSC-iM subpopulation, and an inflammatory memory program defined by unique transcriptional and epigenetic signatures. This conceptual framework offers a plausible explanation for how lifelong inflammatory exposures are integrated within the human HSCs to influence hematopoietic function, thereby leading to inflammaging and clonal hematopoiesis. In their study, the HSC-iM is identified and defined by transcriptome and epigenome programs from the xenografted mice with a mixture of heterogeneous HSPCs. However, the study of HSCs has been greatly facilitated by the comparison of the engraftment potential of purified different single-cell populations. To further investigate HSC-iM function, it would be interesting if HSC-II cells could be purified, separately engrafted, and compared with their HSC-I counterpart. The existing HSC-iM signature could facilitate the identification of surface markers that distinguish HSC-II from HSC-I, enabling such direct isolation and functional characterization. While NSG mice support human HSC engraftment, myeloid and erythroid lineage development in this strain is deficient. The NSG-SGM3 strain used in this study supports myeloid populations but compromises bone marrow HSC and progenitor maintenance, potentially confounding comparisons among HSC populations. Next-generation humanized mouse models such as MISTRG9 and MISTRGFah−/−10 mice might enable testing the causal role of HSC-iM in CH and functional characterization of HSCs from patients with SCD. Moreover, whether the HSC-iM population is conserved between mice and humans remains an open question, as no equivalent population has yet been identified in mice; a mouse model would be beneficial. Lastly, central inflammatory memory can be either protective or maladaptive depending on the context. Further defining these distinct programs may offer the potential to enable therapies that preserve beneficial immunity and hematopoiesis.

[1]

Zeng, A.G.X. et al. Nature 655, 458–467 (2026).

[2]

Netea, M.G. et al. Science 352, aaf1098 (2016).

[3]

Kaufmann, E. et al. Cell 172, 176–190.e19 (2018).

[4]

Cirovic, B. et al. Cell Host Microbe 28, 322–334.e5 (2020).

[5]

Cheong, J.G. et al. Cell 186, 3882–3902.e24 (2023).

[6]

Jaiswal, S. & Ebert, B.L. Science 366, eaan4673 (2019).

[7]

Papaemmanuil, E. et al. N. Engl. J. Med. 374, 2209–2221 (2016).

[8]

Rodriguez-Meira, A. et al. Mol. Cell 73, 1292–1305.e8 (2019).

[9]

Rongvaux, A. et al. Nat. Biotechnol. 32, 364–372 (2014).

[10]

Song, Y.B. et al. Science 371, 1019–1025 (2021).

RIGHTS & PERMISSIONS

The Author(s) 2026. Published by Higher Education Press. This is an Open Access article distributed under the terms of the CC BY license (https://creativecommons.org/licenses/by/4.0/).

Cite this article

Download citation ▾
Han, A., Flavell, R.  Central inflammatory memory in human HSCs  Vita https://doi.org/10.15302/vita.2026.08.0062 ()
AI Summary AI Mindmap
PDF (2050KB)
Sections
Figures
References

0

Accesses

0

Citation

Detail

Recommended

AI思维导图

/

〈 〉