Stem Cell Aging and Pathways to Pre-Cancer Evolution

“What is it that always is, but never comes to be, and what is it that comes to be but never is?”

Plato, Timaeus

Stem Cells and Pre-Cancer

Like Plato’s description of the enigma of human existence, stem cells may remain dormant for an individual’s lifespan and never fulfill their potential, that is, “never come to be”, or alternatively differentiate into other cell types and thus “come to be” but no longer exist as a stem cell. Perhaps the most unique property of stem cells is that they can divide without differentiating. This property, called self-renewal, allows perpetual generation of all cells in the tissue while maintaining a stem cell pool. However, self-renewal deregulation during aging and in response to microenvironmental and macroenvironmental stressors can lead to cancer.

All cells in the body can acquire mutations, but without self-renewal, they cannot become the roots of cancer. Cumulative data suggest that pre-cancer stem cells (pre-CSCs) arise from clonally mutated tissue stem cells that disrupt normal tissue homeostasis as exemplified by hematopoietic stem cell (HSC) deregulation in pre-leukemic bone marrow disorders.¹ Specifically, in myeloproliferative neoplasms (MPNs) and myelodysplastic syndromes (MDS), pre-leukemia stem cells (pre-LSCs) acquire resistance to apoptosis and programmed cell removal,² have longevity assurance, and evade innate and adaptive immune responses, ultimately leading to the generation of self-renewing leukemia stem cells (LSCs) that fuel therapeutic resistance in secondary acute myeloid leukemia (sAML) by becoming dormant in protective microenvironments.¹ Because CSCs fuel therapeutic resistance,^1,3 interception of CSC generation from pre-CSCs may become a more effective strategy for inducing durable remissions.⁴ However, successful interception strategies will be predicated on determining if tissues with functionally defined stem cells form pre-CSC clones, dissecting the clonal hierarchies that drive pre-CSC evolution in different tissues, and searching for diseases caused by clones that have not fully transformed into an invasive malignancy. Equally importantly, advanced age and systemic inflammation-related mechanisms governing pre-cancer initiation and malignant transformation remain a mystery in many human cancers.⁴

Clonal Hematopoiesis and Pre-Leukemia Stem Cell Generation

The first tissue stem cell isolated prospectively was the HSC.⁵ With the isolation of HSCs in mice⁶ and humans⁷ as well as the identification of progenitor cells that through quantal steps of differentiation make mature blood cells, the potential role of HSCs in disease can be analyzed both clinically and experimentally. Using HSCs and blood formation as the main model, we consider the consequences of somatic tissue stem cell self-renewal in aging,⁸ pre-cancer, and cancer development. In this review, we examine the pathophysiological changes associated with stem cell aging and inflammation and how those processes contribute to the development of cancer.

Only HSCs can regenerate HSCs and blood formation for life in transplanted recipients, and HSCs represent only ~1/100,000 of bone marrow cells. Moreover, HSCs slowly circulate from bone marrow to blood to bone marrow using homing receptors and chemokine receptors and establish long-lived hematopoiesis in other bones.⁹ Recent provocative data suggest that HSC aging may be accelerated by acquisition of somatic DNA mutations earlier in life and that the usually indolent process of age-related mutation acquisition, termed clonal hematopoiesis of indeterminate potential (CHIP),¹⁰ may be superseded later in life by more rapidly dividing, splicing factor gene mutated clones that form the apex of an oligoclonal and ultimately malignant hierarchy (Figure 1).^10,11

Figure 1. Pre-Cancer Stem Cell Progression to Cancer Stem Cell Propagation.

Pre-cancer stem cells in the tissue-specific stem cell compartment give rise to an expanded progenitor population and can undergo malignant regeneration and immune evasion, which promotes propagation of cancer stem cells.³

In mice and humans, at least two sub-populations of mainly quiescent HSCs exist postnatally: 1) balanced and lymphoid-biased HSCs that dominate in early life, and 2) myeloid-biased HSCs that dominate in old age in protective microenvironments or niches.^7,12,13 In the inter-mitotic intervals between cell divisions, HSCs can accumulate mutations that cause strand breaks, as measured by γ-H2AX foci formation.¹⁴ When these HSCs are artificially brought into cell cycle by cytokine cocktails, the cells progress from G0 to a long G1 stage during which most DNA repair systems are turned on, such that entry into S phase at ~30 hours after entering the cell cycle is allowed, and almost all of the HSCs produce clones.¹⁴ The kinds of genes expressed at higher levels in aged mouse HSCs are often the partners in human leukemogenic translocations, implying inappropriate repair of blunt-end double-strand breaks in genes sharing transcription locations. Presumably, the local geography presented by local microbial pathogens, and the resistance to those encountered was carried by memory T and B lymphocytes. In addition, the myeloid-biased HSC both promote and are affected by inflammation.¹⁵

Clonal HSC driver mutations^16–21 that result in myeloid-lineage skewed differentiation, partial loss of dormancy, and a propensity to migrate to extramedullary niches, including the spleen, can lead to the generation of MPNs. An activation mutation of Janus kinase 2 (JAK2), a phosphokinase that normally binds to receptor tyrosine kinases upon cytokine induced dimerization, may give rise to the MPNs. It has been shown that the mutant JAK2 (JAK2V617F) activates the process via STAT3 in the absence of cytokines.¹⁸ It is likely that chronic myelomonocytic leukemia, polycythemia vera, essential thrombocythemia, and myelofibrosis arise from myeloid-biased HSC lineages.¹⁶ These MPNs can progress to sAML. Recent human MPN research shows that APOBEC3C-mediated C-to-T DNA mutations drive further clonal proliferation in MPNs and that an expanded myeloid progenitor population acquires ADAR1-mediated self-renewal and epitranscriptomic A-to-I RNA editing changes following sustained inflammatory cytokine signaling.^22,23

Many HSC-driven disorders undergo progressive genetic and epigenetic changes on the path toward leukemic transformation. Aberrant populations of pre-LSCs can themselves cause adult-onset blood diseases but only a fraction of these clonally expanded cells go on to form the LSCs that drive leukemic propagation and therapeutic resistance. The first discovery of functionally defined LSCs was in AML, which is one of the most aggressive cancers of blood-forming cells. John Dick and colleagues found that CD34⁺CD38^lo cells in AML bone marrow can transfer the disease and contain LSCs,²⁴ while we showed that preleukemic AML1:ETO cells are CD34⁺CD38⁻CD90⁺ HSC stage, and their progeny LSC are CD34⁺CD38⁻CD90⁻ multi-potent progenitor (MPP) stage.²⁵ It has also been reported that CD34− LSCs exist in NPM1mut AMLs.²⁶ Initiating driver mutations occur in HSCs, which, as the only self-renewing cells, can generate clones that give rise to the pre-leukemic clones and MPP LSCs.^25,27 Furthermore, it is possible the initial mutational event occurs in HSCs and ‘hitch-hikes’ on these self-renewing cells until further mutational events give rise to LSCs.^1,7

In chronic myeloid leukemia (CML), it has been shown that the ‘pre-leukemic’ cells arise at the stage of HSCs in the setting of inflammatory cytokine upregulation in the HSC niche.²⁸ The emerging myeloid blast crisis cells were daughter cells of the clone at the stage of the granulocyte-monocyte progenitor (GMP), and usually had translocated the self-renewal agonist, β-catenin, to the nucleus.²⁰ Also, these cells had mis-spliced exon 8 kinase domain out of the enzyme, GSK3β, that normally phosphorylates β-catenin in the cytoplasm thereby leading to ubiquitination and proteasomal degradation. Mis-splicing of GSK3β allowed unphosphorylated β-catenin to enter the nucleus to become a self-renewal inducing transcription factor.²⁸ Subsequent research showed that malignant reprogramming of human pre-leukemic myeloid progenitors into self-renewing LSCs that promote blast crisis transformation was accelerated by pro-survival splicing deregulation²⁹ and inflammatory cytokine-driven activation of the RNA editing enzyme isoform, ADAR1p150.³⁰ Specifically, ADAR1p150 overexpression reduces self-renewal regulatory microRNA biogenesis and tumor suppression³¹ and impairs cell cycle transit.³² In 20 different malignancies, ADAR1 has been linked to therapeutic resistance and immune evasion.^33,34 Receptor tyrosine kinase-like orphan receptor 1 (ROR1) has also been linked to CSC self-renewal and early relapse after therapy, with high expression in chronic lymphocytic leukemia cells conferring a poor prognosis.³⁵

In MDS, most, if not all HSCs belong to a single clone derived from a single cell, some of which have chromosomal anomalies that cause the disease and mark the clones.³⁶ At the stage of GMP or other progenitors, the cells express an ‘eat me’ signal, including cell surface calreticulin, with the phagocytic removal of precursors of red blood cells, or platelets, or neutrophils causing these bone marrow failure disorders.²

Driver and Passenger Mutations

The initial discoveries of oncogenes came from the study of retroviruses and transfer of DNA to tissue culture immortalized fibroblasts.^37–40 One might have suspected that the oncogenes alone could transform normal cells in vivo. But if oncogene action alone is insufficient, and requires other mutational or epigenetic events, the questions become which normal cell initiates the process to develop a frank cancer population in vivo, and is there a particular order and type of event that occurs in series to take a normal cell to a cancer cell?

The original discovery that BCR-ABL and JAK2 V617F transcripts are expressed in HSCs by Jamieson and Weissman and colleagues^18,28 and initiate a chronic phase in CML and myeloproliferative neoplasms, like polycythemia vera,^16,18 respectively, underscores the importance of mutation acquisition by long-lived cell types. Malignant reprogramming of their progeny drives blast crisis transformation thereby suggesting that the cell type and temporal sequence of mutations and epigenetic alterations determine the rate of malignant progression^1,2,41 (Figure 2).

Figure 2. Pre-cancer stem cell clonal architecture is determined by type and temporal sequence of mutations, epigenetic alterations, and epitranscriptomic changes.

^{6,15,20,46, 73} The schematic shows a model for the proposed clonal evolution of secondary acute myeloid leukemia (sAML).⁴⁷ Multipotent progenitors derived from HSCs display decreased self-renewal potential and an increased propensity for differentiation into progenitor cells.⁷⁴ However, as HSCs age, epigenomic deregulation and activation of oncogenes leads to a cascade of events that generates pre-leukemic HSCs, clonal expansion of progenitor populations, and ultimately LSCs with deregulated self-renewal potential. The single-cell analysis identifies an example of sequential acquisition of mutations in pre-leukemic HSCs.

In patients with AML, HSC clones [CD34⁺CD38^lo, CD90⁺Lin⁻] have been directly traced in the marrow, with LSC identified in the CD90⁻ subset of CD34⁺CD38^loLin⁻ cells along with other LSC but not HSC markers. Exome sequencing of the leukemias led to identification of mutations that were absent from long-lived T cells in the same individual; these were used to identify patient-specific somatic LSC mutations. Then DNA primers that encompass these mutations were prepared, and bone marrow HSC from the same patients analyzed, one HSC or in vitro HSC clone at a time to test whether the putative driver or passenger mutations were in the target cell. This first normal single HSC testing for mutations allowed researchers to find the order of mutations that included classical oncogenes such as FLT3-ITD, which activates the receptor tyrosine kinase in the absence of cytokines, KRAS, NRAS, and beta-catenin.⁴² Other recurrent mutations included TET2 (loss of function), DNMT3a, and IDH1 and 2 (altered function), ASXL1, NPM1, and CTCF.^2,43,44

Emerging data indicate that the cell type and temporal sequence of mutations and epigenetic alterations is vital for leukemic transformation.^45,46 Virtually all leukemias studied begin with loss or alteration of function of enzymes that are important either to open chromatin for transcription, such as TET2, IDH1/2, or close it, for example DNMT3A, ASXL1 (Figure 2). These driver mutations increase the frequency of mutated HSC, at the expense of normal HSC, perhaps in part by allowing HSC proliferation but inhibiting differentiation. In fact, Jaiswal and Ebert⁴⁷ found that the expanding CHIP clones were derived from single mutations in TET2, or DNMT3A, or ASXL1, all initiating mutations also seen in AML studies.^2,47

While likely that the classical proliferation inducing oncogenes could be early in the sequence, in fact they were almost always last and associated with the pre-leukemic HSC clone transitioning to a downstream MPP or GMP LSC, neither of which are in nor affected by the candidate HSC niche. Additionally, these clones upregulated anti-phagocytic CD47 and countered the prophagocytic calreticulin signal, which appeared to be a permanent epigenetic change in the clone.¹⁰ CD47 is expressed in many cancers.⁴⁸ The upregulation of CD47 occurs late in the process of pre-cancer development, and ‘saves’ the clones from programmed cell removal.¹⁰ Upregulation of ‘don’t eat me’ signals² allow single tissue stem cells to undergo genetic and epigenetic changes and to expand clonally, thereby contributing to hematologic malignancies. In the context of TNF-NFKB1 signaling pathway activation, CD47 upregulation induced by super-enhancer activation in inflammatory breast cancer thereby underscoring the importance of inflammation in driving cancer immune evasion.⁴⁹

Host-Dependent Inflammaging and Pre-Cancer Stem Cell Generation

In addition to radiation and toxic exposure-induced somatic DNA mutagenesis,⁵⁰ induction of CHIP and generation of pre-CSCs from tissue stem and progenitor cells may also be driven by inflammaging.^3,47 Chronic inflammation has long been linked to accelerated tissue aging, particularly in the hematopoietic system.⁵¹ Inflammaging is a process induced by protracted inflammatory cytokine signaling that promotes accelerated stem cell aging⁵¹ and pre-CSC generation. Recently, both macroenvironmental and microenvironmental drivers of inflammaging in hematopoietic and other tissue-specific stem cells have come to the fore as major arbiters of pre-CSC generation and evolution to self-renewing CSCs, which evade host innate and adaptive immune responses (immune evasion). However, the role of stem cell inflammaging in loss of tissue homeostasis and pre-cancer development has not been clearly elucidated.

While host innate and adaptive immune responses evolved in part to protect stem cells, other cells involved in tissue homeostasis from viral and bacterial pathogens, chronic immune activation is associated with systemic signaling driven by pro-inflammatory cytokines, such as tumor necrosis factor α, interferon (α, β, γ) and interleukins (IL-1, IL-6)^52,53, by activated T cells and tissue resident macrophages, and in some cases by pathogenic fibroblasts.²¹ Both mouse model studies and humanized model systems show that aging is associated with a decrease in neutrophil respiratory burst;⁵¹ a decline in macrophage production of Toll-like receptors as well as chemokines and cytokines thereby resulting in decreased T cell proliferative potential and reduced natural killer cell activity that could lead to diminished immune surveillance against pre-malignant clones.² If evasion of programmed cell removal occurs in HSCs, it may also occur in other tissue stem cells and cause disease and cancer. However, other aspects of immunity increase with aging as evidenced by increased production of pro-inflammatory cytokines by peripheral blood mononuclear cells from elderly compared with young individuals in response to mitogens in vitro.⁵⁵ Moreover, IL-6 levels have been shown to be higher in individuals over 85 years of age.⁵⁴

Inflammaging can induce enzymatic mutagenesis that is governed by activation of primate-specific DNA editing enzymes, such as APOBEC3s, as well as RNA editing induced by ADAR1.²³These inflammation-dependent enzymes may drive the evolution of cancer (Figure 3).²³ In fact, protracted activation of a deaminase such as APOBEC3 in response to chronic pro-inflammatory cytokine signaling has been shown to induce cytidine to thymidine (C-to-T) mutations, thereby promoting clonal somatic mutagenesis in stem cell populations.^23,50,55 Moreover, a groundbreaking study demonstrated that chromosomal instability can be induced by APOBEC3A overexpression in genetic pancreatic ductal adenocarcinoma mouse models and human tumor cells resulting in stimulator of interferon genes (STING)-dependent increased metastatic potential.⁵⁶ Comparative whole genome sequencing of purified hematopoietic stem cells and mature cells in saliva from the same individuals with myeloproliferative neoplasms suggest that pre-existing epigenetic modifier mutations may predispose individuals to chronic inflammatory cytokine signaling that enhances inflammaging, clonal hematopoiesis, and pre-CSC generation.^23,57

Figure 3. DNA and RNA Processing in Normal, Pre-Cancer, and Cancer Stem Cells.

During healthy human aging, normal stem cells undergo differentiation and experience a decrease in pro-survival splicing. As people age, there is an increased risk that epigenetic reprogramming and RNA and DNA editing may induce pre-malignant changes to these stem cells. If a significant proportion of the pre-malignant cancer stem cells undergo further malignant reprogramming and acquire the ability to self-renew and survive, this newly formed microenvironment may drive the evolution of cancer.⁴⁷

Although clonal somatic DNA mutations in epigenetic modifier genes, including TET2 and DNMT3A, in stem cell populations increase the risk of developing AML²⁷ as well as cardiovascular death.⁴⁷ The complexity of clonal stem cell dominance has become apparent as a result of high-resolution single-cell sequencing that demonstrates that some mutations in splicing factor related genes emerge later in life and provide a greater clonal competitive advantage and potential for AML development than classic epigenetic modifier gene mutations.⁴⁷ However, the propensity to develop AML varies substantially between individuals; some individuals rapidly progress to malignancy, yet others do not, thereby suggesting that host environmental exposures, lifestyle factors, and immune dysfunction may also influence the trajectory of these stem cell clone wars.⁴⁷

While RNA splicing deregulation has been shown to drive neo-epitope formation, alternative splicing of self-renewal and survival pathway regulators, like GSK3β,⁵⁸ BCL2 family members,²⁹ ADAR1,^22,59 and STAT3,²³ can enhance pre-CSC evolution to self-renewing CSCs in CML, MPNs, and multiple myeloma.^22,34 BCL-2 inhibition has been shown to reduce oxidative phosphorylation, which typically drives AML LSC populations, suggesting that selective therapeutic targeting of LSCs is possible.^60,61 It was previously shown that inflammatory cytokine-driven activation of the RNA editing enzyme, ADAR1, induced progression of MPN progenitors into self-renewing leukemia stem cells that promote therapeutic resistance in AML.³⁰ This has fueled the development of RNA splicing modulators and RNA editing inhibitors^23,31,32,62 to treat a broad array of hematological malignancies and solid tumors that evolve in the context of inflammatory microenvironments that induce protracted cytokine signaling. As RNA editing and methylation have been shown to alter translational efficiency, transcriptomic and epitranscriptomic processes that cause changes in translation could be used to predict stem cell function and fitness in pre-malignant settings.⁶³

Host-Specific Mechanisms of Aging and Pre-cancer Evolution

The overall physiological decline associated with aging is now starting to be unraveled from a molecular standpoint. Aging mechanisms may be pre-programmed, for example, as described in recent research pointing to de-repression of dormant primate-specific repetitive elements and human endogenous retroviruses.⁶⁴ To better define the mechanisms governing age-related molecular change, twelve interconnected hallmarks of aging have been suggested.⁶⁵ Many of the same molecular, cellular, and systemic changes are also observed in the progression of cancer (Figure 4).¹ For example, age-related changes in the host microbiome⁶⁵ are of interest as microbes may have a role in the maturation of CSCs.⁶⁶ The inflammation-dependent DNA and RNA editing triggered by ADAR1 and APOBEC3C deaminases also warrants further examination as potential drivers of pre-CSCs.

Figure 4. Factors that Contribute to Stem Cell Aging.

Stem cell aging is driven by both microenvironmental and macroenvironmental stressors. While intrinsic exposures have been linked to age-related decline in stem cell function and regulation, extrinsic factors are emerging as major drivers of stem cell aging and pre-cancer development. Inflammaging refers to an increase in pro-inflammatory markers in tissues during aging, is systemic, and is interconnected with other hallmarks of aging.⁶⁵ In stem cells, the genomic instability that results from age-related intrinsic and extrinsic factors can drive pre-cancer and cancer evolution.

Conclusions

Due to significant advances in stem cell whole genome and RNA sequencing combined with single cell genomics, spatial transcriptomics and proteomics and functional analyses that quantify tissue stem cell responses to different environmental exposures, essential insights can be made into intrinsic and extrinsic drivers of stem cell aging and pre-cancer development (Figure 4).^63,67 While current age-related decline in stem cell activity is thought to be largely intrinsic,¹ we propose that extrinsic exposures will become more relevant to accelerated human tissue stem cell aging and pre-CSC generation.^3,68–72 Extrinsic exposure will occur as 1) as human longevity is extended by advances in medicine and early detection of cancer and cardiovascular disease; 2) the global spread of pathogens induces acute and chronic inflammatory responses, likely including myeloid-biased HSCs; and 3) immune dysfunction is elicited by advanced age and stem cell stress-inducing environments, including low-Earth orbit (LEO) as space exploration expands.