The purine-rich box1 (PU.1) gene locus, also known as spleen focus-forming virus (SFFV) proviral integration site 1 (SPI1), was first discovered as the integration target of the SFFV retrovirus that causes Friend erythroblastic tumors in 1988 [1]. Thirteen years later, the SFFV insertional site was found to occur within one of the most critical cis-regulatory elements (CREs), the upstream regulatory element (URE) [2, 3]. The URE exhibits cell context-dependent gene regulatory effects, functioning as either an enhancer or silencer to direct the unique expression pattern critical for PU.1 programming in blood development [4, 5, 6, 7, 8]. Significant progress has been made in understanding the URE-mediated chromatin architecture regulation and the repertoire of CREs involved in normal and malignant hematopoiesis.

In light of recent studies demonstrating the role of noncoding RNAs (ncRNAs) as enhancer regulators of PU.1 and in celebration of the 40th anniversary of the discovery of the first mammalian enhancer in SV40 [9, 10], we dedicate this review to the discussion of CRE-mediated PU.1 expression in normal and malignant hematopoiesis. To that end, we elaborate on the unique expression pattern of PU.1, which dictates its biological roles. The previous literature will then be summarized to demonstrate that the PU.1 promoter is essential but insufficient for its cell-type-specific expression. We further analyze the role of the URE and additional CREs in PU.1 expression in normal and malignant myeloid cell development. The roles of chromatin structure and architecture, as well as transcription factors (TFs) and ncRNAs acting as cell-type–specific regulators, will be discussed. Finally, we will address how genetic abnormalities affect the regulatory functions of the URE, provide perspective on potential approaches for therapeutic targeting of the URE, and address questions that warrant further studies in the future.

A member of the ETS-domain TF family, PU.1 plays a critical role in programming the development of several mammalian lineages. Precise spatiotemporal control of PU.1 expression is crucial for the normal generation of blood cells. PU.1 is detectable in hematopoietic stem and progenitor cells, including hematopoietic stem cells (HSCs), common myeloid progenitors (CMPs), and common lymphoid progenitors (CLPs) [6, 11, 12]. In definitive blood cell lineages, PU.1 is expressed in diverse patterns (Fig. 1).

Fig. 1.

Expression patterns of PU.1 in blood cell lineages. PU.1 levels are represented by gradient bars. Abbreviations: HSCs, hematopoietic stem cells; CMP, common myeloid progenitor; CLP, common lymphoid progenitor; MEP, megakaryocytic-erythroid progenitor; GMP, granulocyte-monocyte progenitor; Pro-E, erythrocyte progenitor; Pro-Mk, megakaryocyte progenitor; Pro-Gran, granulocyte progenitor; Pro-Mono, monocyte progenitor; Pro-DC, dendritic cell progenitor; Pro-T, T cell progenitor; Pro-B, B cell progenitor; Pro-NK, natural killer cell progenitor; Ery, erythrocyte; Mk, megakaryocyte; Gran, granulocyte; Mono, monocyte; Mac, macrophage; DC, dendritic cell; T, T cell; B, B cell; NK, natural killer cell. Figures were created with BioRender.com.

High PU.1 expression is required for myelomonocytic cells, and its expression is the highest in monocytes, macrophages, dendritic cells, and granulocyte-monocyte progenitors (GMPs). It is essential for the development of these cells, and its high amplitude expression plays both an early role in the emergence of GMPs and a later role in the terminal differentiation of monocytes and macrophages [13, 14, 15, 16]. High PU.1 levels play a critical role in the production of macrophages, displaying an early role in the development of a multipotential myeloid precursor and a later role in the development of monocytes/macrophages as disruption of PU.1 in mice caused a complete loss of macrophages [6, 17, 18] (Fig. 1).

Low PU.1 expression is needed for granulocytic and B cells [5, 11, 19]. PU.1 is expressed in precursor and mature B cells [20]. Graded PU.1 levels are needed at different stages of B cell development. As B cells mature, PU.1 expression progressively increases, although not to the same degree as in macrophages [11]. PU.1 activates the proximal promoter of the early B cell factor 1 (EBF1) gene, which regulates the specification of the B cell fate [21]. The balance between PU.1, EBF1, and other TFs, such as E2A and Pax5, orchestrate B cell differentiation [22]. Overexpression of PU.1 in CLPs inhibited B cell development [18]. Nevertheless, deletion of PU.1 in mice resulted in a defect in the generation of progenitors of both lineages [13, 23] (Fig. 1).

PU.1 expression is reduced stepwise during early erythroid and T cell differentiation, an adjustment required for normal development of both lineages. Its expression is relatively high in CLPs and early thymic progenitors (ETPs) but fully extinguished in mature T cells [6, 24]. PU.1 downregulation occurs in the thymus during the transition between the double negative (DN) and double positive (DP) developmental stages. This downregulation is specifically required for the progression of normal T differentiation [25]. For instance, the deletion of PU.1 blocks the differentiation of uncommitted thymocyte progenitors in the fetal thymus [24]. On the other hand, PU.1 expression reduced thymocyte expansion and blocked development at the pro-T cell stage [25]. Notably, PU.1 is expressed and regulates the function of subpopulations of Th2 cells, mature T helper cells involved in humoral immunity. PU.1 interferes with GATA-3 transcriptional activities, establishes a defined cytokine profile, and contributes towards establishing the spectrum of cytokine production in these cells [26]. It is required for IL-9 expression in IL-9-producing Th2 cells (Th9), which are critical in mediating inflammation [25, 27]. Similarly, low levels of PU.1 promote the proliferation of early erythroid progenitors, yet complete suppression is required for their terminal differentiation [28, 29, 30, 31, 32]. PU.1 suppresses the differentiation of erythroid cells by upregulating the key cell cycle regulator Cdk6 expression, thereby promoting cell proliferation [29]. The inhibition of erythroid differentiation is also dictated by the antagonistic effect of PU.1 and GATA-1. PU.1 binds and inhibits GATA-1 from inducing erythroid differentiation. In turn, GATA-1 inhibits the myeloid transcriptional activity of PU.1. This molecular antagonism determines whether the CMP commits toward either a myeloid or erythroid lineage [33, 34]. Knockout of Pu.1 in mice causes a delay in T cell development and impairment in erythroblast maturation, with normal development of megakaryocyte and erythroid progenitors [13, 14].

PU.1 also regulates natural killer (NK) cell differentiation and homeostasis by promoting the production of bone marrow NK cell precursors, splenic mature NK cells, and NK cell proliferation in response to IL-2 and IL-12. Additionally, it induces the expression of the receptors for stem cell factor and interleukin (IL)-7 and the expression of inhibitory and activating members of the Ly49 family [35].

The dynamic expression pattern of PU.1 during hematopoietic development was originally thought to be directed by its promoter. The murine and human PU.1 promoters activate reporter genes in myeloid and B cells but not in T cells [36, 37]. This tissue-specificity is dictated by a core promoter region of ~120 base pairs that span a PU.1 transcription start site populated with TF binding motifs [36, 37] (Fig. 2).

Fig. 2.

A diagram showing the PU.1 core promoter and the effects of transcription factors (TFs) on blood cell lineages. Green and red arrows depict the induction and inhibition of PU.1 expression, respectively. Boxes with colors represent binding sites of TFs involved in regulating PU.1. Distances of TF binding sites from the start site of the human PU.1 transcript NM_001080547 are shown. Binding sites were originally studied in the murine Pu.1 promoter , SP1, Oct1/2, and GATA-1 and human PU.1 promoter for Oct1/2 and C/EBP$\alpha$, $\beta$, $\epsilon$. Figures were created with BioRender.com.

In myeloid cells, a repertoire of TFs contributes to the induction of PU.1 promoter activity. The PU.1 protein is involved in positive feedback regulation of the PU.1 gene through a binding site downstream of the transcription start site [36, 37]. C/EBP family members (C/EBP$\alpha$, C/EBP$\beta$, and C/EBP$\epsilon$), in coordination with AP-1 proteins, activate the PU.1 promoter through an upstream C/EBP-binding site in both murine and human loci [38, 39]. Oct proteins, including Oct-1 and Oct-2, bind an upstream octamer site and provide modest activation because their coactivator BOB1/OBF-1 is not expressed, and Oct-2 is barely expressed in myeloid progenitors [40]. In addition, Sp1 also makes a minor contribution in myeloid cells, as mutation of the Sp1 site results in a slight decrease in PU.1 promoter activity [36] (Fig. 2).

In B cells, the PU.1 protein also engages in positive feedback regulation of the PU.1 gene [37, 40]. In contrast to myeloid cells, B cells depend on Oct proteins for transactivating the PU.1 promoter, corresponding with their robust expression of Oct2 and BOB1/OBF-1 [37, 40]. As in myeloid cells, Sp1 plays only a modest role in transactivating PU.1 promoter activity in B cells [40] (Fig. 2).

In erythroid cells, GATA TFs (GATA-1 and GATA-2) bind a GATA site within the core promoter region to inhibit PU.1 expression, a finding validated in several erythroblastic cell lines [33, 41].

It is worth mentioning that many of the assays for promoter activity employed transfection of episomal reporters and that the observed effects could not be recapitulated in the chromatin context with stable expression and chromatin integration [3, 4, 42]. This indicates the presence of regulatory elements within the genome that dictate PU.1 promoter activities in different blood cell lineages.

As the promoter is crucial but insufficient for driving PU.1 expression, studies over the past twenty years have identified multiple CREs that exert long-range control over the transcription of PU.1 in blood cell lineages (Fig. 3).

Fig. 3.

The role of distant regulatory elements within the PU.1 locus. A diagram showing upstream cis-regulatory elements (CREs) involved in long-range control and cell type-specific expression of PU.1. The solid and dotted arrows indicate studies with and without in vivo experiments in genetic mouse models, respectively. Distances in kilobases (kb) of each CRE from the PU.1 transcription start site (TSS) in human (H) and murine (M) genes are shown. The years and citations when the elements were first described are also shown. Figures were created with BioRender.com.

Multiple lines of evidence from genetic mouse models support the essential contribution of TFs, including lineage-determining TFs, to the opposing functions of the URE in different cell lineages. This CRE comprises two highly conserved ~300 bp homology regions called the distal homology region (DHR or H1) and the proximal homology region (PHR or H2) [2]. The PU.1 protein contributes to a positive autoregulatory loop through its binding site within the PHR region [2, 4] (Fig. 4 and Table 1 (Ref. [2, 4, 6, 39, 43, 47, 48, 49, 50, 51, 52])). Point mutations that abrogate PU.1 binding to its binding site reduce the activity of an integrated reporter gene in myelomonocytic cell lines. In mice, the role of PU.1 binding in this positive feedback loop seems limited to HSCs and progenitor cells, as targeted disruption of the binding site reduced PU.1 expression in HSCs but not in mature macrophages [47]. RUNX1, a known regulator of PU.1, binds three conserved sites within the PHR region of the URE and induces PU.1 promoter activity in the context of chromatin. Mice with mutations disrupting these sites show a strong reduction in PU.1 in HSCs, myeloid, and B cells [4, 47]. Additionally, SATB homeobox 1 (SATB1) induces PU.1 expression in GMP and megakaryocytic-erythroid progenitors (MEPs) in mice and exerts its inducing effect by binding to an AT-rich motif at the DHR region of the URE [48].

Fig. 4.

URE-mediated PU.1 regulation in blood cell lineages. The blue-shaded box represents the URE, containing the PHR and DHR, which are represented by smaller, separated boxes. Their distances from the TSS are 17.2 kb and 16.48 kb, respectively. The green arrows represent induction, and the red represents inhibition. The solid and dotted arrows indicate studies with or without in vivo analyses in genetic mouse models. The colored boxes represent specific TF binding sites that regulate PU.1, and the PrPr box represents the proximal promoter. Figures were created with BioRender.com.

An increasing number of studies also indicate a role for other TFs in the URE activities. For example, the myeloid TF C/EBP$\alpha$ binds as homodimers to two C/EBP-binding sites in the PHR region [39, 52] and transactivates a URE-PrPr reporter gene in transiently transfected 32DPKC myeloid cells [52]. Nuclear factor-kappaB (NF-$\kappa$B) binds to its cognate site in the DHR region in transiently transfected U937 myeloid cells. NF-$\kappa$B also transactivates a URE-driven reporter gene in non-hematopoietic COS-1 cell transfectants [43]. The TCF-family proteins TCF1 and LEF1 bind a TCF motif in the DHR region and repress URE-driven PU.1 promoter reporter activity in the early T cell line EL4. These factors show weaker activity in the more mature T cell Jurkat cell line, indicating a contribution of TCFs to the URE repressing activity restricted to early T lymphocytes [6]. The zinc finger TF Ikaros, which is known to compete with ETS factors such as PU.1 for the same sites [50], exerts distinct influences on URE activity depending on cellular context. It enhances a URE-PrPr reporter in transiently transfected macrophages RAW264.7 cell line and exerts a repressive effect on the same reporter in the pre-B-cell line NFS-25 [51]. It remains to be determined whether C/EBP$\alpha$, NF-$\kappa$B, or Ikaros can control URE-driven PU.1 promoter activity within chromatin and whether abrogation of their respective URE binding sites affects PU.1 mRNA levels in cells or mice (Fig. 4 and Table 1).

URE-mediated fine-tuning of graded levels of PU.1 is crucial in generating blood cell lineages. Disturbances in URE function are associated with blood malignancies. Several genetic mouse models with targeted disruption of the URE have offered insights into the role of this important CRE. Mice with isolated, biallelic loss of the URE exhibit ~80% reduction in PU.1 expression in bone marrow and develop a multilineage differentiation block, which includes accumulations of immature myeloid cells in bone marrow and spleen, T lymphoblasts in the thymus, and stimulated proliferation of B1 cells in peripheral blood. The mice developed B1 cell proliferative syndrome, T cell lymphomas, and AML [5, 6]. Interestingly, mice with uniallelic URE loss and a DNA mismatch repair-deficient background resembling the mutations acquired in aging human individuals exhibit a 35% reduction in PU.1 expression and acquire myeloid-biased preleukemic stem cells. The mice develop myelodysplastic syndromes with subsequent transformation into AML [7]. Thus, differential PU.1 dosage mediated by changes in the URE status and additional genetic changes determine pathways for leukemogenesis and lymphomagenesis.

The URE also undergoes functional disruption by trans-acting factors acting in trans to regulate PU.1 expression in hematologic malignancies. Several leukemia-associated fusion proteins derived from chromosomal translocations alter its activity. In t(8;21) AML cells expressing the oncogenic fusion protein RUNX1-ETO, chromosomal interaction between the URE and the PU.1 promoter is abolished [45]. The RUNX1-ETO protein restricts chromatin accessibility at the URE through local histone deacetylation and inhibits expression of the enhancer lncRNA LOUP, which plays an essential role in the URE–PrPr interaction and PU.1 expression [55]. In addition, RUNX1-ETO shifts the interaction with the URE from the PrPr to the promoter of a PU.1 asRNA (AsPr), resulting in antisense transcription and a myeloid differentiation blockade [45, 46]. In inv(16) AML cells expressing the oncogenic fusion protein CBF$\beta$-MYH11, there is a decreased AsPr and increased PrPr accessibility [45]. Inhibition of CBF$\beta$-MYH11 multimerization by the small molecule inhibitor AI-10-49 [66] reverses the effect, indicating a role by CBF$\beta$-MYH11 in mediating chromatin structure. CBF$\beta$-MYH11 transactivates AsPr promoter activity, resulting in PU.1 asRNA transcription and inhibition of PU.1 expression [46].

In addition to experimentally generated point mutations [4, 47, 52], naturally occurring genetic alterations within the URE can affect TF recruitment. A single nucleotide polymorphism (SNP) at the URE that occurs more frequently in patients with AML with complex karyotype prevents SATB1 binding, decreasing its enhancer activity and reducing PU.1 expression in myeloid progenitors [48]. In mice, the spleen focus forming virus (SFFV), which integrates into the URE in a site flanking the two conserved H1 and H2 regions, is implicated in causing Friend virus-induced erythroleukemia [1, 2, 67]. Thus, disruption of URE function either through the URE sequence or trans-acting factors may represent a driving force in myeloid malignancy.

Approaches to restore normal URE-mediated chromatin architecture have become feasible and could offer therapeutic utility in myeloid malignancies. For example, PU.1 expression can be induced in leukemic cells by endogenous activation of enhancer RNA using a CRISPR activation platform, in which dCAS9 is fused to the VP64 transcriptional activating domain and targeted to the promoter region of the enhancer RNA upstream of the URE [55]. Transcription activator-like effectors (TALEs) fused to VP64 have been used to target the PHR region of the URE that contains the PU.1 binding site [68]. Since TFs, particularly ones with tumor-suppressing functions, harbor inherent challenges for direct therapeutic targeting [69], induction of endogenous PU.1 by CRISPR activation and TALEs holds considerable promise. However, much research and development in this space is needed. For instance, can these platforms be modified for cell type selection, i.e., to be selectively targeted in myeloid and lymphoid cells? Additionally, among PU.1 CREs, which one could be targeted and for which tumor type? Further development of tools to modulate URE-mediated PU.1 expression and renormalize chromatin architecture will likely lead to novel treatments for myeloid malignancies.

Since the discovery of the first and most important CRE of PU.1, the URE, studies over the past two decades have provided a wealth of knowledge regarding the role of CREs in the long-range transcriptional regulation of PU.1 during normal and malignant hematopoiesis. The URE and other CREs act in concert with the PU.1 promoter to mediate their regulatory function via chromatin loops that drive cell-type specific PU.1 expression. Biomolecules such as TFs and ncRNAs cooperate in assembling these complex and dynamic chromatin structures (Table 2). Genomic abnormalities that alter these biomolecules and the CREs themselves can affect functional chromatin architecture and thereby contribute to disease pathogenesis.

A deeper examination of CRE-mediated PU.1 regulation in normal development and disease is merited as many important questions remain. For instance, mechanisms for the dynamic switching by the URE between docking on the PU.1 promoter and the PU.1 asRNA promoter during differentiation represent an important gap in the field. More studies are also needed to understand chromatin signatures underlying the bifunctionality of the URE as an enhancer or a silencer in different blood cell lineages. PU.1 CREs other than the URE must also be further characterized for their contributions to chromatin architecture regulation. As multiple chromosomal translocations are present in hematologic malignancies, it is anticipated that multiple protein fusions could alter the URE functions in leukemia. Rapidly evolving single-cell genomic and epigenomic technologies hold promise in addressing these questions. Accordingly, this will broaden our basic understanding of mammalian gene regulation by cis-acting elements, which comprise up to 10% of the human genome [70]. As strategies to target the URE have shown feasibility [55, 68], further progress could yield clinical tools to modulate CRE-mediated gene expression and normalize chromatin architecture, expanding the therapeutic repertoire for various blood diseases.

Manuscript draft and data accurual: EAK, AKB, and BQT; Figures and Tables: EAK, AKB, EO, and BQT; Idea conception: BQT, ANG, and DGT. Critical review and revision: EAK, AKB, ANG, DGT, and BQT. All authors reviewer and approved the final version of the manuscript. All authors have participated sufficiently in the work to take public responsibility for appropriate portions of the content and agreed to be accountable for all aspects of the work in ensuring that questions related to its accuracy or integrity. All authors contributed to editorial changes in the manuscript.

Not applicable.

We thank Barbara Dziegielewska and members of the Trinh lab for assistance and helpful suggestions. Figures were created with BioRender.com.

This study was supported by the National Heart, Lung, and Blood Institute [grant number: 1P01HL131477-6 A1] to DGT, the National Cancer Institute [grant number: NCI K01 CA222707], and the American Cancer Society [grant number: 134088-IRG-19-143-33-IRG] to BQT.

The authors declare no conflict of interest.