Transcription factor networks in cellular quiescence

Abstract

Many of the cells in mammalian tissues are in a reversible quiescent state; they are not dividing but retain the ability to proliferate in response to extracellular signals. Quiescence relies on the activities of transcription factors (TFs) that orchestrate the repression of genes that promote proliferation and establish a quiescence-specific gene expression program. In this review, we discuss how the coordinated activities of TFs in different quiescent stem and differentiated cells maintain reversible cell cycle arrest and establishment of cell protective signaling pathways. We further cover the emerging mechanisms governing the dysregulation of quiescence TF networks with age. We explore how recent developments in single-cell technologies have enhanced our understanding of quiescence heterogeneity and gene regulatory networks. We also discuss how TFs and their activities are themselves regulated at the RNA, protein, and chromatin levels. Finally, we summarize the challenges associated with defining TF networks in quiescent cells.

The eukaryotic cell cycle is a highly controlled and orderly process that ensures proper cell division through mechanisms that are broadly evolutionarily conserved from yeast to mammals¹ (Box 1). When some cycling cells are exposed to triggers to stop proliferating, for instance, in the context of nutrient depletion, organismal development, or tissue homeostasis, they reversibly exit the cell cycle soon after cell division and enter into a quiescent state (also referred to as a reversible G0 state; Box 1). Quiescent cells are defined by their ability to re-enter the cell cycle upon receiving proliferative cues^2–4. Quiescence has been documented in single-celled eukaryotes such as yeast⁵ and a wide range of metazoan stem and differentiated cells^6–9 (Fig. 1a). Quiescent cells are not necessarily “sleepy”, just waiting to be “woken up¹⁰.” Though not dividing, quiescent cells actively re-engineer their signaling, metabolic, and gene expression programs to resist stress^11,12 and perform important functions including ensuring long-term maintenance of adult tissues or participating in immune response^2,4,7 (Fig. 1a). Dysregulation of cellular quiescence is associated with aging, and diseases such as cancer, fibrosis, and autoimmune disorders^13,14.

Box 1. Cell-cycle phases and their relationship to quiescence.

The eukaryotic cell cycle consists of four proliferation-related phases in the following order: Gap 1 or G1, synthesis (S), Gap 2 or G2, and mitosis (M)¹⁴⁹ (see Box Figure). The commitment to enter the cell cycle takes place at G1. This is followed by DNA (chromosome) duplication in S phase that generates cells with 4N DNA content (diploid cells at G1 have the normal 2N DNA content). At the next G2 phase, the cells prepare for the following mitosis (M) phase. M phase involves several steps and lead to the segregation of duplicated chromosomes. After the M phase ends, the cells divide to generate two daughter cells, each with 2N DNA content. The daughter cells produced are in G1 phase and, in proliferating conditions, each daughter cell will transition to the S phase and repeat the above steps.

Intrinsic and extrinsic signals related to DNA integrity, growth factor availability, cell size, and developmental cues control the transition of cells from G1 to S phase¹⁴⁹. The term “restriction point” in mammalian cells or “START” in yeast has been used to describe a point in G1, beyond which the cell fully “commits” to finishing the remainder of the cell cycle, even if proliferation (mitogenic) signals are removed. However, by using single-cell imaging, it has been recently shown that, upon mitogen removal, these committed cells can exit the cell cycle if there is a delay in entering mitosis¹⁵⁰.

Cells can exit at G1 before the restriction point to enter into a non-cycling G0 phase of the cell cycle¹⁴² (see Box Figure). Quiescence is a type of reversible G0 state, that is, the cells can enter the cell cycle when proper signals are present. The “reversibility” of cellular quiescence is in contrast to the irreversible nature of the senescent cell state (an irreversible G0 state) that is encountered during development, aging, and certain diseases, where the cells permanently exit the cell cycle¹⁰⁵ (see Box Figure). The cell-cycle exit of terminally differentiated cells with speciliazed functions, such as neurons, adipocytes, and keratinocytes, is also permanent. However, certain differentiated cells, such as fibroblasts and T cells, remain in a quiescent state and enter into the cell cycle to perform physiological functions related to tissue repair and immune response, respectively.

Fig. 1. Functional roles of quiescent cells in different tissues and significance of quiescence TF networks.

a Quiescence is a resting cell state that is actively and reversibly maintained in different body tissues^2–5,142. (i) Quiescence reversibility is abolished in disease states and aged tissues^13,14, which results in impaired quiescence exit or loss of the pool of quiescent cells due to excessive proliferation. (ii) In response to signals, such as invading pathogens or the presence of cancer cells, resting T cells are activated and participate in the immune response^7,64. (iii) Skeletal muscles contain a pool of quiescent stem cells (MuSCs) that proliferate and differentiate to generate muscle cells that can regenerate muscle after injury^40,41. (iv) In the brains of adult mice, quiescent NSCs proliferate and differentiate in response to injury or stroke to generate new neurons⁶. Whether this process also occurs in humans is not clear. (v) During inflammation or loss of blood, quiescent hematopoietic stem cells (HSCs) in the bone marrow proliferate and differentiate to generate blood cells of all necessary lineages¹³⁴. (vi) In response to a wound, quiescent fibroblasts in the skin proliferate and migrate to the wound site and orchestrate wound repair¹⁴³. b Quiescent and proliferating cells differ in their patterns of gene expression^2,12,15,16. These gene expression changes reflect different activities of transcription factors (TFs) (see Supplementary Table 1 and Fig. 2 and 3). In the schematic shown, gene B is positively regulated by TF2 in proliferating cells, while it is downregulated in quiescent cells by the action of repressive TF3. In contrast, the expression of gene A regulated by TF1 remains unaltered with quiescence (upper panel). These regulatory changes can be computationally described in terms of TF networks¹³⁰, where activating and repressive TFs are connected to their target genes (lower panel). Figure generated with Biorender.

Transitions between proliferating and quiescent states are accompanied by changes in the expression levels of hundreds to thousands of genes^2,12,15,16. Some of these expression changes are essentially universal (e.g., downregulation of cell cycle genes), while others may be specific to the cell type, the quiescence stimulus, or combinations of these factors. Gene expression patterns that establish and maintain a robust quiescent state are largely defined by the combined activities of transcription factor proteins (TFs). Binding of TFs to DNA regulatory regions (promoters, enhancers, or silencers) can lead to activation or repression of target genes (Fig. 1b). Application of next-generation sequencing methods including RNA sequencing (RNA-seq) and Assay for Transposase-Accessible Chromatin using sequencing (ATAC-seq) along with single-cell technologies have greatly facilitated the investigation of TF regulation in bulk samples and individual quiescent cells. This review explores studies in mammalian cells that have addressed the following questions using a variety of experimental and computational tools:

• Which TFs regulate the landscape of quiescence-specific gene expression and what cellular functions do these TFs confer?

• Are similar or different TFs engaged when different types of cells become quiescent?

• What regulatory mechanisms ensure that TFs have peak function in quiescent cells?

• How have emerging technologies enhanced our understanding of quiescence TFs?

• What are the conceptual and technical challenges associated with defining quiescence TF networks (Fig. 1b)?

• Are quiescence TFs dysregulated during aging, and could alterations in quiescence TFs contribute to aging?

Taken together, the studies presented here provide insight into how quiescence, once established, is maintained by TF networks and the gaps in our understanding of these networks. Quiescence in non-mammalian cells and cancer cells is reviewed elsewhere.^5,17

TFs as regulators of cellular quiescence

In this section, we introduce the TFs that coordinate quiescence maintenance by repressing the expression of cell cycle genes, preventing irreversible cell fates, and activating catabolic and protective signaling pathways. Then, we delve into quiescence-related TFs in stem and differentiated cells in different tissues of origin. Finally, we discuss how single-cell sequencing is being used to characterize the quiescence landscape. A complete picture of how quiescence TF networks are wired in a signal- and cell type-specific way is still being compiled. Here we aim to piece together the known components of quiescence TF networks and highlight information that is missing. A more complete list of quiescence TFs is present in Supplementary Table 1.

Regulation of hallmarks of quiescent cells

Below, we highlight cellular processes that are specifically important for viability and homeostasis in quiescent compared with proliferating cells, i.e., the “hallmarks” of quiescence. We explain the functional importance of these hallmarks and the TFs associated with them (Fig. 2 and Supplementary Table 1), recognizing that in most cases these TFs have only been studied in a limited number of cell types. While some cell-fate changes, including proliferation, differentiation, and stemness, are driven by a small number of TF “master regulators,” establishing a quiescent state requires a coordinated activity of multiple TFs.

Fig. 2. TFs and pathways regulating the hallmarks of quiescence.

Quiescence in a variety of cell types is characterized by characteristic features (highlighted in yellow). These hallmarks include repression of cell cycle genes (involved in G1/S and G2/M transitions). Formation of an RB-E2F complex in quiescent cells inhibits the activity of the E2F TF family members E2F1, E2F2 and E2F3 that activate G1/S genes in cycling cells¹⁹. Conversely, the DREAM repressive complex, comprised of multi-vulval class B (MuVB) protein complex, RB, and E2F4 or E2F5¹⁹, inhibits both G1/S and G2/M genes in quiescent cells. The LIN54 protein member of the MuVB complex binds to CHR DNA elements in the promoters of G2/M genes, while E2F4 and E2F5, like E2F1 to −3, binds to E2F DNA motifs in G1/S gene promoters. A second feature is the maintenance of quiescence reversibility by the TF HES1²³. The HES1 gene is a downstream target of the Notch pathway that is active in quiescent cells. Binding of a Notch receptor to its ligand leads to the cleavage and nuclear localization of its intracellular domain (NICD)¹⁴⁴. In the nucleus, the interaction of NICD with the TF RBPJ activates Notch target genes, including HES1. The HES negatively regulates cell cycle genes in quiescent cells; it also regulates its own gene (HES1), establishing a negative feedback loop¹⁴⁵. Upregulation of autophagy genes, a third feature of quiescent cells, is mediated by the TF TFEB³⁰ and FOXO family members³¹. TFEB is localized to the nucleus during quiescence due to low mammalian target of rapamycin (mTOR) activity. The increased expression levels of extracellular matrix (ECM) genes (a fourth hallmark) can also be mediated by the Notch pathway³⁵. Finally, genes that protect cells from reactive oxygen species (ROS) are upregulated in quiescence by FOXO TFs^39,146, which translocate to the nucleus during quiescence due to inactivation of pathways that sequester them into the cytoplasm or induce their degradation. Figure generated with Biorender.

Repression of cell cycle machinery

Repression of cell cycle–related genes is a key common feature of the quiescent gene expression landscape¹ (Fig. 2). In response to signals that include proliferation-inducing mitogens, activator E2Fs, E2F1–3 of E2F TF family, bind chromatin at their conserved E2F recognition motif and induce expression of G1/S genes (e.g. DNA replication genes). In quiescent cells, the members of the retinoblastoma (RB) transcriptional repressor protein family form complexes with these activator E2Fs, thereby sequestering them; these RB:E2F complexes recruit repressive chromatin-modifying factors to E2F binding sites that downregulate G1/S genes (Fig. 2). In proliferating cells, cyclin-dependent kinases (CDKs), phosphorylate RB proteins, causing them to release E2Fs. CDKs are activated by binding to cyclin proteins and their activities are disrupted by CDK inhibitors (CDKI). In quiescent cells, CDK activity is repressed by low expression of cyclins, many of which are E2F targets, and by high expression of CDKIs¹⁸.

G2/M genes that encode mitotic factors are repressed in quiescent cells by the DREAM protein complex composed of the MuvB protein complex along with RB bound to repressive E2Fs, E2F4 or E2F5¹⁹. The DREAM complex represses G1/S genes via E2F4 or E2F5 and G2/M genes through binding of LIN54 TF, a component of MuvB, to G2/M gene promoters containing the cell cycle gene homology region (CHR) motif²⁰. (Fig. 2). Thus, the E2F family of TFs not only activates proliferation-associated genes upon cell cycle entry, but also inhibits expression of cell-cycle genes during quiescence as part of RB-E2F and DREAM repressive complexes.

Maintenance of quiescence reversibility

Reversibility, the ability of quiescent cells to re-enter the cell cycle, distinguishes quiescent from irreversible arrested states (Box 1). The Notch signaling pathway and its downstream target genes, the HES family of repressive TFs, are primary regulators of quiescence reversibility (Fig. 2). In quiescent cells, Notch transmembrane receptors bind to Notch ligands presented by neighboring quiescent²¹ or proliferating²² cells. This interaction leads to cleavage and release of the Notch intracellular domain (NICD) which translocates to the nucleus where it forms a complex with the TF RBPJ leading to activation of HES family genes. Overexpression of HES1 protects quiescent human fibroblasts from irreversible senescence or terminal differentiation, while silencing HES1 makes quiescent fibroblasts more susceptible to these fates²³. In quiescent neural stem cells (NSCs), expression of cell cycle-promoting ASCL1 is suppressed by high levels of HES1²⁴ (Fig. 2). Thus, a role for Notch in repressing irreversible cell fates has been established in multiple types of quiescent cells (Supplementary Table 1).

Metabolism of quiescent cells

In many quiescent cells, the main nutrient-sensing kinase mTORC is off, which results in reduced anabolic metabolism. When quiescent cells are activated, mTORC kinase is activated, which results in activation of the MYC TF that promotes glycolysis and glutaminolysis, and the SREBF TF that promotes lipid synthesis⁷. To reclaim metabolites when nutrients are scarce, many quiescent cells activate the autophagy pathway to selectively sequester old and damaged cytoplasmic macromolecules and transport them to lysosomes which produce metabolites and energy²⁵. Autophagy is important for entry into and maintenance of quiescence in starved cells^26,27. In NSCs and hematopoietic stem cells (HSCs), clearing protein aggregates in lysosomes enhances the ability of quiescent stem cells to re-enter the cell cycle^28,29. TFEB and TF3, members of the microphthalmia/transcription factor E (MiTF/TFE) family of TFs, activate expression of genes involved in autophagy and lysosome biogenesis^28,30. In proliferating cells, TFEB is phosphorylated by mTOR which retains it in the cytoplasm³⁰. Under quiescence-inducing starvation conditions, when mTOR is off, unphosphorylated TFEB translocates to the nucleus, where it binds to a motif in the promoters of target genes³⁰ (Fig. 2). Likewise, FOX TF family member FOXO3 activates a network of target genes that establish autophagy in quiescent cells³¹. These forkhead-box (FOX) TFs are active in quiescent cells, but are negatively regulated by the nutrient-responsive phosphoinositide 3-kinase (PI3K) pathway in proliferating cells^32,33 (Fig. 2). Thus TFs that allow quiescent cells to store and reclaim energy are an important part of the quiescence TF network.

Extracellular matrix protein expression and signaling

An emerging characteristic of quiescent cells in multiple model systems is their expression of higher transcript and protein levels of extracellular matrix (ECM) genes^10,15,34,35. ECM proteins anchor quiescent cells in tissue niches and help maintain quiescence via cell-ECM interactions^35,36. The NICD and its cofactor RBPJ of the Notch pathway can bind regulatory elements and activate genes encoding collagen ECM proteins in quiescent muscle stem cells (MuSCs)³⁵ (Fig. 2). If this pathway is disrupted, the stem cells proliferate inappropriately. As described further below, ECM proteins in stem cell niches, the specialized microenvironments of stem cells, can promote quiescence programs. ECM protein periostin enhances quiescence maintenance in HSCs through integrins on the surface of HSCs by inhibiting the PI3K pathway^36,37.

Protection from external stressors

Quiescent cells exhibit enhanced survival in stressful conditions¹² and improved protection from genotoxic stress, stress due to reactive oxygen species (ROS) that can oxidize and damage macromolecules, and unfolded protein stress. ^10–12 In epithelial cells, quiescent cells are protected against unfolded proteins in part by the integrated stress response TF ATF4 that activates chaperones that improve protein folding¹². Chaperones that protect against proteotoxic stress are also abundant in quiescent muscle stem cells in vivo³⁸. FOXO TFs not only activate autophagy genes, but also induce ROS scavengers, such as catalase, superoxide dismutase, and peroxiredoxin, in quiescent cells³². Disruption of FOXO expression results in elevated ROS levels and quiescence exit³⁹.

Regulation of stem cell quiescence

Most adult stem cells are quiescent and retain the capacity to proliferate to replace lost or damaged cells. We highlight here the quiescence TFs and the associated signaling pathways in muscle, hematopoietic, and neural stem cells, well-studied stem cell types in the context of cellular quiescence. These stem cells reside within distinctive niches and signals from circulating factors, mechanical forces, nearby cells, and the extracellular matrix converge to enforce quiescence in a manner that reflects their physical location. A complete understanding of how each stem cell type establishes and maintains quiescence in response to cell type-specific features requires further investigation.

Muscle stem cells in skeletal muscle

Quiescence in adult muscle stem cells (MuSCs) is important to prevent loss of the stem cell pool and to promote muscle regeneration after injury (Fig. 1a)^40,41. MuSCs reside in a specialized niche between the basal lamina and the muscle myofiber⁴². Interaction between the laminin ECM protein in the basal lamina and an integrin on the MuSC surface preserves their self-renewal capacity^42,43. Muscle fibers, nearby endothelial cells, and mechanical compression present delta ligands that activate the Notch signaling pathway in quiescent MuSCs^39,42,44,45. Quiescent mouse MuSCs deficient in Notch pathway TF RBPJ⁴⁴ or FOXO3³⁹ have impaired self-renewal and tissue regeneration capacities. FOXO3 mediates its effects through the Notch pathway as overexpression of NICD (Fig. 2) rescues the failure to self-renew in FOXO3-deficient MuSCs³⁹. In addition, the PAX7 TF is a specific marker of quiescent MuSCs and is required for the MuSC maintenance⁴⁶. PAX7 can interact with transcriptional corepressor transducin-like enhancer of split (TLE4) to promote stemness by repressing expression of the muscle differentiation-inducing MYF5 TF⁴⁷. Further, heterodimerization between the Wnt pathway TFs β-catenin and LEF1 and TGF-β pathway TF SMAD3 enforces mouse MuSC quiescence⁴⁸, providing an example of how physical interactions between TFs regulate quiescence.

Hematopoietic stem cells in bone marrow

Hematopoietic stem cells (HSCs) reside in the bone marrow and generate cells that populate the blood⁴⁹. The bone marrow stem cell niche maintains quiescence in HSCs and HSCs rapidly lose quiescence when they exit the niche⁵⁰. HSCs in the bone marrow are present in either a niche near bones or a niche near blood vessels⁵¹. HSCs in a bone-adjacent niche interact with bone-forming osteoblasts and bone-resorbing osteoclasts, as well as fibroblasts and adipocytes, while HSCs in vascular regions receive signals from endothelial cells. In addition to surrounding cells, ECM molecules including hyaluronic acid⁵⁰ and circulating factors also regulate HSCs quiescence⁵⁰. Endothelial cells signal to HSCs through JAG1-Notch receptor interactions⁵¹. Osteoblasts in the bone marrow promote HSC quiescence through JAG1-Notch receptor, and secreted cytokines thrombopoietin and osteopontin⁵¹. A large number of TFs regulate HSC quiescence⁵⁰. TP53, a tumor suppressor protein, promotes HSC quiescence and induces expression of TF Gfi1 and TF co-regulator Necdin, both of which restrict HSC proliferation⁵². PU.1, a member of the E26 Transformation Specific (ETS) TF family⁵³, induces quiescence in HSCs by repressing expression of genes regulated by inflammation-associated cytokine IL-1, resulting in reduced protein synthesis and a slowed cell cycle⁵⁴. Depletion of the mouse TF NKX2–3, a member of NKX family of homeodomain TFs, leads to exhaustion of the HSC pool due to impaired HSC quiescence characterized by a defect in autophagy of defective mitochondria⁵⁵. Additional TFs that help to maintain HSC quiescence include HLF, PBX1, NURR1, NRF2, C/EBPa, YY1, GFI-1, and TCF15, while deletion of MEF/ELF4 or ID1 preserves HSC quiescence⁵⁰ (Supplementary Table 1).

Neural stem cells in the brain

Quiescent NSCs can be activated to replace damaged or lost neurons and glial cells during injury^6,56 (Fig. 1a). In rodents, NSCs reside in both the subventricular zone in the lateral ventricle and the subgranular zone in the hippocampus dentate gyrus⁵⁷. NSCs in the subventricular zone receive signals from ependymal cells, a type of ciliated glial cell that lines the brain’s ventricles, neuroblasts, astrocytes, endothelial cells, and cerebrospinal fluid⁵⁷. NSCs interact with adjacent ependymal cells through N-cadherin-mediated cell-cell contacts that maintain NSC quiescence by ensuring the cells are correctly oriented within their niche.⁶ The integrin receptor on the NSC surface that binds to the ECM protein laminin helps the NSCs adhere to blood vessels, and maintains the NSCs in a quiescent state⁵⁷. Endothelial cells also present Notch pathway JAG1 ligands to NSCs to regulate quiescence^6,57,58. Deletion of Rbpj, encoding for Notch TF RBPJ, in mouse NSCs results in increased production of progenitor cells (neuroblasts) in the short term, and a complete loss of the NSC pool and impaired neurogenesis over a longer time frame^6,58–60. Further, Wnt signaling pathway TF LEF1 co-regulates quiescence genes in NSCs with TF NFIX, a member of the nuclear factor I (NFI) family⁶¹. Finally, the FOXO TFs enforce NSC self-renewal and quiescence⁶², as depletion of FOXOs in mice results in NSC pool depletion⁶³. Thus, in NSCs, like HSCs, the coordinated activity of multiple TFs is required for proper quiescence.

Regulation of quiescence in differentiated cells

While quiescence maintenance in stem cells is closely associated with signaling from their niches, quiescence in differentiated cells is less focused on preserving cells within a specific physical location, but still responds to both cell autonomous and paracrine signaling stimuli. Many TFs regulate quiescence in both stem and differentiated cells, including TP53, Notch, FOXO, NF-κB, and Krüppel-like factor (KLF) TFs (Supplementary Table 1). Below, we highlight TFs that regulate quiescence in three types of differentiated cells selected because their quiescent states have received significant attention, they respond to different proliferative stimuli, and they reside in distinct microenvironments: T cells, endothelial cells, and fibroblasts. A full understanding of quiescence TF networks in these different cell types is not yet known and is an area for further study.

T cells of the immune system

Quiescent T cells (also referred to as resting or naïve T cells) activate and differentiate in response to stimuli that include pathogenic infections and cancer cells^7,64 (Fig. 1a). T cells are activated when they interact with an antigen-presenting cell displaying a foreign peptide, and their expansion is supported by cytokines. TFs involved in regulation of quiescence in T cells include FOXO1, FOXO3, FOXP1, KLF2, NF-κB, STAT, RUNX1, BACH2, and TP53 (Supplementary Table 1)^65–67. FOXO3 induces expression of a CDKI, CDKN1B^p27, to arrest cell cycle progression^68,69. Loss of mouse Foxp1 promotes premature activation of naïve T cells, even when their cognate antigen is not present^67,70. In human CD4⁺ T cells, knockdown of FOXP1 induces quiescent CD4⁺ T cells to enter the cell cycle, express activation markers, and upregulate differentiation genes⁷¹. FOXP1 expression is repressed in disorders involving uncontrolled T cell proliferation⁷¹.

Endothelial cells of blood vessels

In adult tissue, quiescent endothelial cells line blood vessels and control the exchange of oxygen and nutrients between the blood and underlying tissues⁸. Endothelial cells receive quiescence signals from other endothelial cells and other cell types such as the mural cells that regulate blood pressure and nearby neurons⁸. In response to mitogens produced when oxygen is limited, angiogenesis results in activation of quiescent endothelial cells and formation of new blood vessels. TFs that control endothelial cell quiescence include ETS family TFs, FOXO1, KLF4, and basic helix-loop-helix (bHLH) TF TAL1, while NF-κB promotes endothelial cell activation (Supplementary Table S1).^72,73 High expression of ETS family TF ERG in resting endothelial cells represses proinflammatory gene expression and endothelial cell proliferation,⁷⁴ while in contrast, ETS1, another ETS family TF, drives a switch from the quiescent to activated state^75–77. Thus, in endothelial cells, signals from surrounding cells and endothelial cells themselves mediate quiescence through multiple pathways and TFs.

Fibroblasts of connective tissues

Fibroblasts, mesenchymal cells that reside within the connective tissue and secrete extracellular matrix, are present in many tissues including skin, lung, and heart⁷⁸. In normal tissue, fibroblasts are largely quiescent^2,79, but when a wound is present, mitogens released in the wound healing environment such as platelet-derived growth factor stimulate fibroblasts to proliferate and migrate toward wound sites where they contribute to the repair process¹⁵. Fibroblasts can be induced into quiescence by removing serum, which contains mitogens and other factors that are also released in wounds. Repressor E2Fs, the DREAM complex, HES1, and TP53 promote quiescence in fibroblasts. In addition, tumor suppressors SALL2, a zinc finger-containing TF, and MXI1, a bHLH TF that inhibits MYC, also positively regulate quiescence in human lung fibroblasts⁷⁹. Knocking down SALL2 or MXI1 increases fibroblast proliferation upon serum deprivation⁷⁹. TWIST1, another bHLH TF, is not expressed in quiescent fibroblasts in healthy tissues⁸⁰, but is induced when fibroblasts are activated in cancerous and fibrotic tissues⁸¹, an example of dysregulation of TFs that establish quiescence, in disease states.

Candidate quiescence TFs from bulk and single-cell sequencing

Recent studies applying bulk and single cell RNA-seq have advanced our understanding of the architecture of the quiescent state. RNA-seq provides information on the global landscape of TF expression levels, TF target genes, and signaling pathways in proliferating versus quiescent cells. For example, bulk RNA-seq of human corneal fibroblasts revealed regulatory roles for E2F, MYC, and NF-κB TFs⁸². ATAC-seq is employed to identify chromatin regions that are more accessible with quiescence genomewide, thus providing clues about putative cis-regulatory DNA elements such as enhancers or silencers in quiescent cells. These DNA elements can be analyzed with TF motif enrichment tools to prioritize candidates for quiescence TFs. Single cell RNA-seq and ATAC-seq, either separately or in combination, can provide insight into subpopulations of quiescent cells that reflect depths or phases of quiescence. Combining single cell RNA-seq with CRISPR-based forward screens has allowed the identification of new TFs that regulate quiescence. Together, these approaches have been used to identify candidate quiescence TFs for further experimental studies.

Single cell RNA-seq to probe quiescence TFs

Single cell RNA-seq has been used to identify quiescence TFs in multiple model system^83–85. Single cell RNA-seq of pancreatic stellate cells revealed that the AP1 family TF c-Jun is an important regulator of the quiescent state⁸³. Using a CRISPR knockout screen combined with barcoding to monitor individual HSCs and their clonal trajectories, the bHLH TF TCF15 was discovered to be required and sufficient for HSC quiescence and renewal⁸⁵. A single cell CRISPR screen in T cells identified IFZF1 as a TF that promotes cell cycle entry by dampening the activity of the TCF1 TF, and ETS1 as a TF that promotes quiescence by inhibiting mTORC1 activity⁸⁶. In addition, a genome-wide CRISPR knockout screen to identify genes important for quiescence induced by high density revealed an important role for the noncanonical NF-κB pathway in control of proliferation arrest induced by high cell density⁸⁷.

Single-cell RNA-seq of multiple types of tissue-isolated or cultured quiescent cells revealed that quiescent cells are not homogeneous and instead can be assigned to one of multiple cell clusters based on their transcriptomic profiles^22,88–91. In mouse MuSCs, two clusters of quiescent cells with different levels of TF Hes1 were present⁸⁹. In mouse NSCs, based only on TF expression levels, a quiescent cell cluster can be identified with high expression of Sox9, Id2, and Id3 TFs, and components of Notch and Bone Morphogenetic Protein (BMP) signaling pathways²². In mouse hepatic stellate cells, a quiescent cluster exists with high expression levels of nuclear receptor and ETS TFs, and NF-κB components⁸⁸. The results reinforce the importance of TFs and their networks in defining the quiescent state and transcriptionally distinct subsets of quiescent cells.

ATAC-seq to identify quiescence TFs

In quiescent mouse MuSCs, ATAC-seq revealed that target motifs for FOX TF family members are enriched in accessible chromatin regions⁹². Promoters of quiescence genes Pax7 and Notch3 are more open in quiescent MuSCs while promoters of differentiation genes such as MyoD and proliferation genes such as Ki-67 were more open in activated MuSCs⁹². Peaks more open in quiescent MuSCs were enriched for genes associated with extracellular matrix; peaks associated with cell-cell communication and stimuli response were enriched in early activation states; and full activation was associated with chromatin opening near genes involved in system development and cell differentiation⁹². Single-nucleus ATAC-seq of mouse MuSCs reveals a quiescent cell cluster with highly accessible chromatin regions that reside near genes encoding for FOX TFs and Notch pathway TFs and receptors⁹³. In quiescent mouse melanocyte stem cells (McSCs), the motifs for NFI, Specificity protein (SP), and KLF TF family are enriched in chromatin regions that are more accessible in quiescent compared to activated McSCs and differentiated cells⁹⁴. Motifs for NFI TF family members are also enriched in accessible chromatin regions of quiescent mouse NSCs⁵⁶. Genes associated with chromatin accessibility changes in quiescent versus activated NSCs are enriched for pathways involving neural identity, differentiation, and proliferation⁵⁶. Finally, ATAC-seq analysis of naïve and activated CD8⁺ T cells revealed increased accessibility for genomic regions with enrichment for bZIP (ATF3), ETS, and Runt (RUNX1, RUNX2) TF families¹⁶. Genomic loci with differential accessibility in naïve versus activated T cells were near genes associated with T cell activation including interleukins and their receptors. Therefore, ATAC-seq of quiescent cells provides further functional validation of the established quiescence TFs by the enrichment of their motifs in accessible genomic regions.

Dysregulation of quiescence-related TFs during aging

Quiescent cells are protected from DNA mutations that result from DNA replication¹³. However, quiescent cells express low levels of DNA repair proteins and rely on error-prone nonhomologous end joining (NHEJ)-mediated DNA repair¹³. These cell-intrinsic factors, in combination with environmental factors (e.g. inflammatory niche signals), can affect the quality of quiescent cells during aging. In this section, we discuss the consequences of aging on quiescent cell maintenance and related TF networks (Fig. 3).

Fig. 3. Dysfunctional TF activities in aged quiescent cells.

Aged quiescent cells show defects in quiescence maintenance that depend on the cell type and the signals from the surrounding niche^{13,40,103,147,148}. Some quiescent cells become more resistant to cell cycle entry with age (first column) as exemplified by hippocampal stem cells⁹⁵ and NSCs.⁹⁶ In NSCs, an inability to exit the quiescent state is due to low levels of the proliferation-inducing ASCL1 TF, which is ubiquitinated and degraded at the proteasome. Inactivation of non-canonical Wnt signaling also contributes to persistence of the quiescent state. In other contexts, aging can make quiescent cells more prone to cell cycle entry (second column). In aged quiescent MuSCs^40,98 and T cells¹⁰⁰, a tendency to activate depends on the FOXO-AKT axis. Signals from the aging niche lead to activation (phosphorylation) of AKT via the PI3K-AKT pathway, and phospho-AKT induces the phosphorylation of quiescence-promoting FOXO TFs in the cytoplasm. Phosphorylated FOXO TFs are either retained in the cytoplasm or degraded by the proteosome, and this prevents them from maintaining quiescence with aging. Aged quiescent cells can also enter a pre-senescent state (third column) due to downregulation of HES1¹⁰⁴, which encodes for the TF HES1 that maintains quiescence reversibility²³. In aged MuSCs, acquisition of a pre-senescent state is due to the activation of p16 (CDKN2A)¹⁰⁶, a cell cycle inhibitor that is expressed in senescent cells. The increased levels of p16 in aged, quiescent cells inactivates the E2F TFs that are required for the activation of cell cycle genes upon reversible quiescence exit. The activities of TFs in aged quiescent cells are also affected by changes in chromatin accessibility (fourth column), which may increase or decrease the binding of TFs to DNA regulatory elements (e.g promoters and enhancers)^56,92. Figure generated with Biorender.

Prolonged quiescence with aging

With age, quiescent mouse hippocampal NSCs enter into a deeper state of quiescence⁹⁵. As mice age, levels of the stem-cell activation–promoting TF ASCL1 are reduced due to its ubiquitination and proteosomal degradation (Fig. 3), which makes aged hippocampal NSCs more likely to re-enter quiescence and less likely to proliferate and differentiate⁹⁵. Inflammatory niche signals and a Wnt antagonist also induce NSC quiescence in the aging mouse brain, and inhibiting these cues leads to more robust activation of old NSCs after injury⁹⁶. Thus, the same TFs that regulate the proliferation-quiescence transition in NSCs also affect the maintenance of healthy NSCs with age. In mouse melanocyte stem cells, genes associated with immune system processes are elevated in quiescent cells, including the gene that encodes the immune checkpoint programmed death-ligand 1 (PD-L1), which promotes immune tolerance, and genes encoding TFs that regulate Pd-l1—Stat3and Irf1⁹⁷. Quiescent melanocyte stem cells expressing PD-L1 persisted with age while others were lost⁹⁷.

Pre-mature exit from the quiescent state

In contrast to the deeper quiescent state upon aging in NSCs, there is a decline in the population of MuSCs with age due to premature activation^40,98. In MuSCs of geriatric mice, quiescent MuSCs become primed for myogenic differentiation⁹⁹ as a result of aging-niche-derived insulin-like growth factor 1 (IGF1), which activates PI3K-AKT signaling, resulting in proteosomal degradation of the quiescence-maintaining FOXO TFs⁹⁹ (Fig. 3). The regulatory FOXO-AKT axis also contributes to a loss of the quiescent naïve T cell pool with age, and the associated weaker response to infections and tumors¹⁰⁰. Higher levels of the AKT suppressor TRIB2 make naïve CD4⁺ T cells more resistant to depletion with age¹⁰¹. Elevating the levels of pro-quiescence factors such as TRIB2 may represent a strategy to counteract the aging-related breakdown of quiescence-maintaining TF networks.

Acquisition of senescence features with age

Proliferating animal cells enter into replicative senescence, a state of irreversible cell cycle arrest (see also Box 1), upon prolonged passaging during cell culture¹⁰².Strikingly, it is shown that human lung fibroblasts that have been quiescent for a long time in culture (≥100 days) also acquire senescent characteristics and DNA damage¹⁰³. Based on a transcriptomic single-cell atlas of aging human skin, aging skin fibroblasts have reduced expression of HES1¹⁰⁴, a Notch target gene that maintains quiescence reversibility²³. Downregulation of HES1 increases senescence in primary skin fibroblasts, demonstrating an age-protective effect of the Notch-HES1 network¹⁰⁴. Geriatric, quiescent mouse MuSCs enter a pre-senescent state with increased levels of senescence factor¹⁰⁵ CDKN2A/p16¹⁰⁶. When stimulated by proliferative signals during muscle injury, these aged MuSCs fail to activate, and instead, enter a fully senescent state¹⁰⁶. High levels of CDKN2A prevent activation of proliferation genes¹⁰⁶, and silencing the gene encoding CDKN2A in aged MuSCs restores them to a quiescent state and renews their ability to respond to injury¹⁰⁶. Thus, as organisms age, cells must actively suppress senescence pathways to stably maintain quiescence.

Chromatin-accessibility changes with aging

The effect of aging on chromatin accessibility depends on the cell type (Fig. 3). Global chromatin accessibility of quiescent NSCs is lower in old than young mice, and the chromatin regions that lose accessibility with age are enriched for binding motifs for NFI⁵⁶. In contrast to NSCs, quiescent MuSCs from old mice have a more accessible chromatin environment compared to young mice with aging-related accessible regions preferentially located near genes involved in cell cycle, senescence, metabolism, and cancer-related pathways⁹². In MuSCs from 74–99 year-old individuals, binding motifs for TFs related to differentiation (NFYA, NFYB, and NFYC) and stress response (ETS2, EGR1) are enriched in accessible compared with inacessible genomic regions, further supporting the links between quiescence TFs and aging-associated changes⁹⁸. Despite these advances, the interplay between chromatin accessibility changes, TF activity, and aging-related defects in quiescence maintenance requires further study.

Mechanisms controlling TF activities in cellular quiescence

During the transition to quiescence, TF activity is regulated at the RNA, protein, and chromatin levels (Fig. 4), and the underlying mechanisms include processing of mRNA transcripts that encode TFs, and TF post-translational modifications, degradation, binding partners, and localization. A full picture of how TFs are regulated to ensure maximal activity in the quiescent state is still emerging.

Fig. 4. Mechanisms controlling the TF activities in quiescent cells at the RNA, protein, and chromatin levels.

a Quiescent cells have higher levels of nuclear-localized transcripts with retained introns compared to proliferating cells^15,107. Intron retention can regulate the activities of TFs in quiescent cells by preventing the cytoplasmic localization and translation of the intron-containing transcripts produced by these TF genes. b TFs can also be regulated in quiescence at the RNA level through alternative polyadenylation (APA), a process that produces transcripts with different 3´ ends^15,108. APA can result in transcript isoforms with different lengths of 3´ untranslated regions (UTRs) (such as the short and long isoforms shown). The different levels and RNA stabilities of these APA isoforms (short versus long) can affect the levels of the translated product, i.e. production of more TF protein in proliferating cells compared to quiescent cells. c At the protein level, the localization (e.g. nuclear versus cytoplasmic) and stability of TF proteins can depend on the type of post-translational modification (e.g. phosphorylation or ubiquitination). In the schematic shown, phosphorylation of a TF (in light blue) in the quiescent state leads to binding of a protein (blue) that retains it in the cytoplasm. A TF (in light red) can also be ubiquitinated and degraded in the cytoplasm to prevent its nuclear localization and activity in quiescence. d The chromatin in quiescent cells is globally more compact and less accessible than in proliferating cells^116–121. Many chromatin regions (mainly enhancers) show differences in chromatin accessibility between quiescent and proliferating cells.^16,56,92 This reduced chromatin accessibility can prevent TFs from binding and activating the genes associated with these enhancers. e Chromatin folds locally into sub-megabase-sized topologically associated domains (TADs)¹²⁵. These TADs exhibit a high degree of DNA-DNA interactions, but only few between neighbouring TADs. Upon proliferation, these TADs are partitioned into smaller TADs that allow for increased enhancer-promoter interactions^16,119, which may be required by proliferation-associated TFs to regulate their target genes. Figure generated with Biorender.

Regulation at the RNA and protein levels

The mRNA post-transcriptional processing landscape of quiescent cells differs from that of proliferating cells. mRNA transcripts in quiescent cells are more likely to retain introns compared to transcripts in proliferating cells^15,107 (Fig. 4a). When quiescent cells are stimulated to divide, these introns can be spliced out to generate translation-ready mRNAs¹⁰⁷. In quiescent fibroblasts, transcripts tend to use more distal polyadenylation sites, which results in increased stability¹⁵ (Fig. 4b). Alternative polyadenylation can regulate quiescence TFs as Pax3 alternative polyadenylation regulates the rate of muscle stem cell activation¹⁰⁸. In T cells, m6A mRNA modifications of STAT TF inhibitors is required for homeostatic expansion¹⁰⁹.

TFs are also dynamically regulated at the protein level in quiescent cells. TF proteins in quiescent cells have a high turnover rate that allows for rapid response to proliferating signals¹¹⁰. In addition, quiescent cells maintain a large pool of idle ribosomes that allow for rapid protein synthesis upon activation¹¹⁰. TF proteins in quiescent cells are also regulated by post-translational modifications^111–113, for instance, PAX7 transcriptional activity and muscle homeostatsis are regulated by PAX7 acetylation¹¹⁴. TFs can also bind to different binding partners in the quiescent state, leading to changes in localization or activity, for example, LEF1, a TF in the Wnt pathway, associates with TGF-β signaling TF Smad3 specifically in quiescent cells⁴⁸ (Fig. 4c).

Chromatin accessibility

ATAC-seq studies show that the chromatin of quiescent cells is less accessible on a genome-wide level, with fewer open chromatin regions compared to proliferating cells^{16,56,64,92,115–121}, and TFs mostly bind to accessible chromatin regions^122–124. Thus, changes in chromatin conformation upon quiescence can impact the time it takes for TFs to find and bind to target DNA regions¹¹⁹. Changes in accessibility with quiescence are mostly localized to regions that are distal (either intergenic or intronic) to a gene transcription start site and these regions are likely sites of binding for activator or repressor TFs^16,56,92 (Fig. 4d). Most promoter regions do not exhibit a change in chromatin accessibility when comparing proliferating and quiescent cells⁹² which may facilitate rapid activation as chromatin remodeling is not required⁵⁶. However, how different TFs bind to these differentially accessible and stable chromatin regions to establish quiescence gene expression patterns is not completely understood.

Three-dimensional genomic architecture

The genome is not linear, rather, it is hierarchically organized in a spatial manner in the nucleus, and TFs must traverse complex genomic conformations to bind their recognition sites¹²⁵. Analysis of 3D genome architecture using high-throughput chromosome conformation capture (Hi-C) techniques has shown that physical contacts between distal enhancers (or silencers) and gene promoters are largely restricted to sub-megabase regions in the genome, called topologically associated domains (TADs)¹²⁵. Activation of resting immune cells is accompanied by remodeling of TAD boundaries, leading to partitioning of TADs into smaller regions that may promote enhancer-promoter contacts^16,119,126 (Fig. 4e). The boundary regions of each TAD are enriched for binding sites for the CCCTC-binding factor (CTCF). CTCF acts as an insulator by favoring the formation of loop structures that block inter-TAD enhancer-promoter contacts¹²⁷. The effect of CTCF activities on quiescence is context-dependent. In one study of HSCs in culture, knockdown of CTCF restricts activation¹²⁸. However, in another study using a CTCF conditional-knockout mouse model, CTCF was found to sustain the adult HSC pool and HSC quiescence¹²⁹. The reduction in HSC quiescence with CTCF inactivation is partially reversed by antioxidants, indicating the importance of CTCF-mediated ROS control in quiescent HSCs¹²⁹.

Mapping TF networks and associated challenges

The TF network of a cell includes all of the direct regulatory links between TFs and their target genes¹³⁰ (Fig. 1b). Understanding quiescence TF networks would reveal relationships among TFs and their targets that maintain quiescence. We present here the progress and obstacles in constructing TF networks for quiescent cells using bulk or single-cell sequencing data (Supplementary Table 2). The considerations presented here would be expected to also affect investigations of other quiescence-related processes.

Preparation of quiescent cells

To reconstruct robust TF networks for quiescence, sequencing data must be obtained from cells that faithfully represent the quiescent state. Quiescence can be established by exposing cultured cells to quiescence signals such as deprivation of the proliferation-promoting mitogens in serum². However, culture conditions may not replicate the signals that cells receive in their in vivo cellular niche. For stem cells especially, a common strategy is to isolate and purify quiescent cells from tissues by taking advantage of stem cell-specific cell-surface markers^38,131,132. Tissue isolation procedures can, however, affect the gene expression profiles of the isolated cells^38,133, which can be partially addressed by fixing the tissue prior to dissociation¹³³.

Annotation of quiescent cells

Across cell types, a lack of robust quiescence markers makes it difficult to positively identify quiescent cells. In practice, annotation of quiescent cells often involves use of cell-state markers, which usually relies on an absence of cell-proliferation markers^9,134,135. However, proliferation markers can also be absent in G1 cycling cells and non-quiescent G0 cells, such as senescent or differentiated cells (Box 1). Catalogs for globally upregulated or downregulated genes with quiescence are available for a variety of cell types and represent potential quiescence markers. Recently, machine-learning models have been used to learn quiescence signatures, which may suggest new candidates for quiescence markers (Supplementary Table 2). Further discoveries of cell-surface protein markers that are quiescence-specific will greatly facilitate the isolation of quiescent cells.

Quiescence state exhibits heterogeneity. Quiescent MuSCs are found in two distinct phases called G0 and G(Alert), the latter representing an “alert” state that rapidly responds to environmental cues¹³⁶. With single-cell gene expression analysis, the quiescent population of NSCs has been divided into “dormant” and “primed-quiescent” states²². Single cell sequencing has also revealed distinct cell clusters in quiescent MuSCs^89,90,93 and resting T cells¹³⁷. Capturing the heterogeneity of quiescent states is, therefore, important for faithful reconstruction of quiescence TF networks.

Another critical issue for assigning cells to a quiescent cell cycle phase in large single-cell RNA-seq datasets is how to clearly define the boundary between the G0 and G1 phases (Box 1). Although many methods for cell cycle scoring exist, most, but not all, have focused predominantly on assigning cells to G1, S, G2, and M phases¹³⁸ and, by default, do not classify cells as G0 (Supplementary Table 2). A continuous phase score may be a more accurate description of the depth of quiescence and the heterogeneity observed in quiescence¹³⁹. Examples of such continuous cell phase scores include mean gene expression of marker genes, cell cycle phase angles, and likelihood of cell phase scores (Supplementary Table 2).

Defining TF-gene regulatory links

Several algorithms have been applied to construct TF-gene networks and to determine TF activity scores in quiescent cells (Supplementary Table 2). Applying the single-cell regulatory network inference and clustering (SCENIC) method¹⁴⁰ to single-cell RNA-seq data of quiescent mouse hepatic stellate cells reveals high activity levels for several nuclear receptor TFs, including NR1H4⁸⁸. Treatment with an NR1H4 agonist results in increased expression of quiescent genes, while loss of NR1H4 reverses these effects⁸⁸. In another study, integrated regulatory network analysis (IReNA) using RNA-seq and ATAC-seq identified a TF network associated with Müller glia quiescence in mouse retina. The TFs in this quiescence network include HES1, KLF15, and NFI members NFIA and NFIX¹⁴¹, supporting the coordinated activity of these TF families in the quiescent state. A systematic comparision of TF networks across cell types using network analysis methods may reveal universal gene-regulatory features of quiescence.

Conclusions and future perspectives

Taken together, the studies discussed here reveal that quiescent cells express an array of TFs that regulate genes that suppress cell-cycle progression, prevent the adoption of irreversible fates, reorganize their metabolism, and activate pathways that maintain their viability during quiescence. Common TFs from a range of TF families are activated as different types of cells enter quiescence. Multiple different TFs have been demonstrated to be required for quiescence of stem and differentiated cells. In some instances, these TFs likely reinforce each other while in other instances the TFs can compete. Emerging technologies suggest that quiescent TFs are less likely to change chromatin accessibility at the promoters of target genes and are more likely to regulate enhancer (or silencer) opening. A full picture of TF-TF crosstalk and the enhancers and silencers to which TFs bind in quiescent cells is not yet available. Further, it is not presently clear whether a universal TF code operates in quiescent cells from different tissues. Deducing the underlying TF networks in quiescent cells will provide a powerful framework for understanding quiescence dysregulation during aging and disease states.

Further application of CRISPR screens as described above will provide additional opportunities to identify new quiescence TFs. Emerging spatial sequencing methods have the potential to characterize quiescent cells and their associated TFs in different locations within intact tissues. This approach may shed light on how tissue niches affect quiescence heterogeneity. Multi-tissue single-cell RNA-seq and ATAC-seq atlases of quiescent cells would provide a map of common and cell-specific TF networks and the signaling pathways that induce them. Finally, proteomic studies will be crucial in gaining insights into TF abundance and post-translational modifications that regulate TF activities in quiescent cells.