Tumor Cell Dormancy: Threat or Opportunity in the Fight against Cancer

Tumor dormancy, a clinically undetectable state of cancer, makes a major contribution to the development of multidrug resistance (MDR), minimum residual disease (MRD), tumor outgrowth, cancer relapse, and metastasis. Despite its high incidence, the whole picture of dormancy-regulated molecular programs is far from clear. That is, it is unknown when and which dormant cells will resume proliferation causing late relapse, and which will remain asymptomatic and harmless to their hosts. Thus, identification of dormancy-related culprits and understanding their roles can help predict cancer prognosis and may increase the probability of timely therapeutic intervention for the desired outcome. Here, we provide a comprehensive review of the dormancy-dictated molecular mechanisms, including angiogenic switch, immune escape, cancer stem cells, extracellular matrix (ECM) remodeling, metabolic reprogramming, miRNAs, epigenetic modifications, and stress-induced p38 signaling pathways. Further, we analyze the possibility of leveraging these dormancy-related molecular cues to outmaneuver cancer and discuss the implications of such approaches in cancer treatment.


Introduction
Microorganisms adopt different mechanisms to survive under hostile conditions. They undergo drastic changes in cellular physiology to shape the surrounding microenvironment to best fit their requirements [1]. In response to a stressor such as chemotherapy, radiation therapy, or O 2 /nutrient scarcity, stressed tumor cells that survive apoptosis become dormant. After cessation of the therapy, the dormant cells may repopulate, resulting in tumor recurrence and development of chemotherapyresistant cancer cells [2,3].
Dormant cells can be detected as circulating tumor cells (CTCs) in the bloodstream or disseminated cells (DTCs) in secondary sites such as bone marrow (BM). Once metastatic cells find a new home to settle, they may undergo different fates: they either die, remain silent (restrictive soil), or grow with an even more aggressive and lethal behavior than before (permissive soil). The implication of tumor cell dormancy is very well established in the development of multidrug resistance (MDR), minimum residual disease (MRD), tumor outgrowth, and metastatic relapse, leading to cancer treatment failure [4][5][6].
In the context of clinical settings, clinical dormancy is defined as the pre-or post-treatment time point where residual cancer cells are clinically undetectable. Generally, cancer dormancy can be categorized as cellular dormancy or tumor mass dormancy [7]. Cellular dormancy is defined as transient G0-G1 growth arrest in a certain fraction of cancer cells (dormant cells); accordingly, cancer progression occurs when a single dormant malignant cell gains the ability to re-enter the cell cycle. In tumor mass dormancy a stagnation of total tumor growth due to the equilibrium of proliferation and apoptosis is governed by an angiogenic switch or immune escape (two prevailing mechanisms) that eventually shifts the balance in favor of cancer progression [8]. Accordingly, current in vitro, in vivo and ex vivo (organs-on-a-chip) experimental models of cancer dormancy can mimic cellular and tumor mass dormancy (angiogenic and immunologic dormancy), each reflecting distinct growth kinetics [9]. These models were somewhat successful in recapitulating dormancy mechanism of tumors, as they lead to the discovery of cellular dormancy factors/mechanisms including extracellular matrix (ECM) [10], metabolic [11,12], epigenetic [13,14], stemness [15,16], non-coding RNAs [17], and p38 stress-induced signaling pathway [18][19][20], as we intend to discuss in detail in this review (Table 1). However, these models fail to address the full picture of clinically observed dormancy-related cellular processes, such that, the function of angiogenic dormancy in metastatic dormancy is unclear [21]. Also, the paradox role of pro-and anti-proliferative action of the immune system is shown in tumor evolution [22]. Also, inherent difficulties in detecting and characterizing micrometastatic lesions/individual cells in the patient is another hurdle to assess the relative contribution of various mechanisms of dormancy in the clinical setting [23]. However, the new emerging methodologies, namely microfluidics based on analyzing microliters of serum samples may enable detection of single cells, as wells as DTCs and may aid to capture dormancy dynamics at a single-cell resolution (reviewed in [24][25][26][27]). A full discussion of dormancy-related experimental models is recently reviewed [28][29][30]. The focus of the current paper, however, is to employ the data extracted from different dormancy models in the preclinical settings to highlight the stealth mechanisms dictating cancer cell dormancy. Subsequently, we analyze the possibility of leveraging dormancy related molecular cues to outmaneuver cancer. Finally, the implication of each approach in cancer treatment and prognosis is discussed. Upregulation of Angiostatin drive long-term dormancy of primary tumors, inhibit tumor growth, and reduce cancer metastases. [31] Thrombospondin-1 Inducer of angiogenic dormancy Overexpression of Thrombospondin-1 inhibits melanoma angiogenesis, lung colonization, and spontaneous pulmonary metastasis. [32] VEGF/VPF121 Inhibitor of angiogenic dormancy Overexpression of VEGF/VPF121 result in tumor growth and escape from dormancy. [33] (VEGF ( Inducer of angiogenic dormancy Dormant tumors undergo a stable genetic reprogramming during their switch to the fast-growing phenotype by downregulation of angiogenesis inhibitors such as Thrombospondin and decreased the sensitivity of angiogenic tumors to angiostatin along with upregulation of angiogenesis-related genes. [36] EGFR-1, IGF-IR, CD73, PI3K, ESM-1, PIK3CB, TIMP-3 Inhibitor of angiogenic dormancy Mitochondrial dysfunction Inhibitor of metabolic dormancy VLX600 impairs OXPHOS and drives a HIF-1α-dependent switch to glycolysis, which this metabolic pathway can't meet the energy demands of tumor cells, thus induction of autophagy is unavoidable. Yet, due to lack of HIF-1α-stabilization and glucose inaccessibility in metabolically stressed environments, shifting to glycolysis mode will be restricted, consequently, tumor cells undergo apoptosis.   transforming growth factor β2 (TGF-β2); proline hydroxylase I (P4HA1); Eph receptor A5 (EphA5); histone H2BK; insulin-like growth factor binding protein 5 (IGFBP-5); epithelial growth factor receptor (EGFR)-1; insulin-like growth factor type I receptor (IGF-IR); 5′-Ecto-nucleotidase (CD73); endothelial cell-specific molecule 1 (ESM-1); phosphatidylinositol 3-kinase PI3K (PIK3); tissue inhibitor of metalloproteinase-3 (TIMP-3). The RNA polymerase II transcriptional mediator subunit 12 (MED12). ˧ denotes suppression, ↑ denotes upregulation, ↓ denotes downregulation.

Angiogenic Switch
As a foremost contributing factor to tumor dormancy is the incompetence of proliferating tumors to acquire the angiogenic potential to induce the formation of new blood vessels [97]. In the same way, the ability to drive tumor angiogenesis process will promote the escape from latency and initiate tumor outgrowth. This point of cancer progression is recognized as an angiogenic switch, the induction of which is under the control of intricate biological processes comprising the cancer cells, BM-derived endothelial precursors, and tumor associated-stromal microenvironment [98,99].
Angiogenesis switch, or in a better word, angiogenic failure is a key piece of the cancer cell dormancy puzzle. The role of angiogenic dormancy has been highlighted by the experimental as well as clinical observations that anti-angiogenic therapies or angiogenic therapies could induce tumor dormancy or rescue dormant cells, respectively ( Table 1). As such, angiogenic factors display a reduced expression in dormant cells; meanwhile, the high transcription level of angiogenesis inhibitors such as angiostatin, endostatin, and thrombospondin-1 is associated with the tumor dormancy.
Concerning the role of the angiogenic switch in cancer dormancy, two therapeutic approaches follow: in the first approach, long-term dormancy can be induced using anti-angiogenic therapy as validated in different angiogenic models of dormancy in vivo [100]. For example, cancer patients may benefit from bioactive angiostatin gene therapy; however, transgene delivery may necessitate repetitive drug administration to assure persistent latency state of primary tumors and control metastatic expansion, that is not clinically feasible because of off-target effects and also low expression level of transgene as shown in a B16F10 melanoma mouse model [31]. Also, eradication of primary tumors which result in inducing dormant tumor cells is evidenced by the high expression level of angiogenic inhibitor thrombospondin-1 in human melanoma xenografts (D-12, R-18, and U-25). In contrast to wild-type tumors with no thrombospondin-1 expression, a tumor-suppressive index was demonstrated only in mice bearing D-12, U-25, or thrombospondin-1 overexpressing R-18 tumors, as validated by melanoma angiogenesis, lung colonization, and spontaneous pulmonary metastasis [32]. These results suggest a combination of sustained anti-angiogenic therapy besides local treatment for melanoma therapy. Impaired tumor angiogenesis also drives prolonged dormancy of human liposarcoma. This type of tumor with no angiogenic activity produce moderately high amount of the anti-angiogenesis agents TIMP-1 and thrombospondin-1, proposing that the nonangiogenic, microscopic, dormant stage of melanoma might be susceptible to anti-angiogenic treatment years prior to manifestation of malignant disease, or macroscopically detection of anatomical site of a given tumor, by current approaches [101]. Notably, metastatic dormancy can be realized once the proliferation of tumor cell is well-balanced with an equal ratio of tumor cell apoptosis, indicating that anti-angiogenic treatment regulates metastatic outgrowth through increasing of cancer cell death [21].
giogenesis switch, or in a better word, angiogenic failure is a key piece of the cancer cell cy puzzle. The role of angiogenic dormancy has been highlighted by the experimental as well cal observations that anti-angiogenic therapies or angiogenic therapies could induce tumor cy or rescue dormant cells, respectively ( Table 1). As such, angiogenic factors display a expression in dormant cells; meanwhile, the high transcription level of angiogenesis rs such as angiostatin, endostatin, and thrombospondin-1 is associated with the tumor cy.
ncerning the role of the angiogenic switch in cancer dormancy, two therapeutic approaches in the first approach, long-term dormancy can be induced using anti-angiogenic therapy as d in different angiogenic models of dormancy in vivo [100]. For example, cancer patients may from bioactive angiostatin gene therapy; however, transgene delivery may necessitate ve drug administration to assure persistent latency state of primary tumors and control tic expansion, that is not clinically feasible because of off-target effects and also low ion level of transgene as shown in a B16F10 melanoma mouse model [31]. , as validated by melanoma angiogenesis, lung colonization, and spontaneous pulmonary sis [32]. These results suggest a combination of sustained anti-angiogenic therapy besides eatment for melanoma therapy. Impaired tumor angiogenesis also drives prolonged cy of human liposarcoma. This type of tumor with no angiogenic activity produce moderately ount of the anti-angiogenesis agents TIMP-1 and thrombospondin-1, proposing that the iogenic, microscopic, dormant stage of melanoma might be susceptible to anti-angiogenic nt years prior to manifestation of malignant disease, or macroscopically detection of ical site of a given tumor, by current approaches [101]. Notably, metastatic dormancy can be once the proliferation of tumor cell is well-balanced with an equal ratio of tumor cell denotes suppression, ↑ denotes upregulation, ↓ denotes downregulation.

Angiogenic Switch
As a foremost contributing factor to tumor dormancy is the incompetence of proliferating tumors to acquire the angiogenic potential to induce the formation of new blood vessels [97]. In the same way, the ability to drive tumor angiogenesis process will promote the escape from latency and initiate tumor outgrowth. This point of cancer progression is recognized as an angiogenic switch, the induction of which is under the control of intricate biological processes comprising the cancer cells, BM-derived endothelial precursors, and tumor associated-stromal microenvironment [98,99].
Angiogenesis switch, or in a better word, angiogenic failure is a key piece of the cancer cell dormancy puzzle. The role of angiogenic dormancy has been highlighted by the experimental as well as clinical observations that anti-angiogenic therapies or angiogenic therapies could induce tumor dormancy or rescue dormant cells, respectively ( Table 1). As such, angiogenic factors display a reduced expression in dormant cells; meanwhile, the high transcription level of angiogenesis inhibitors such as angiostatin, endostatin, and thrombospondin-1 is associated with the tumor dormancy.
Concerning the role of the angiogenic switch in cancer dormancy, two therapeutic approaches follow: in the first approach, long-term dormancy can be induced using anti-angiogenic therapy as validated in different angiogenic models of dormancy in vivo [100]. For example, cancer patients may benefit from bioactive angiostatin gene therapy; however, transgene delivery may necessitate repetitive drug administration to assure persistent latency state of primary tumors and control metastatic expansion, that is not clinically feasible because of off-target effects and also low expression level of transgene as shown in a B16F10 melanoma mouse model [31]. Also, eradication of primary tumors which result in inducing dormant tumor cells is evidenced by the high expression level of angiogenic inhibitor thrombospondin-1 in human melanoma xenografts (D-12, R-18, and U-25). In contrast to wild-type tumors with no thrombospondin-1 expression, a tumor-suppressive index was demonstrated only in mice bearing D-12, U-25, or thrombospondin-1 overexpressing R-18 tumors, as validated by melanoma angiogenesis, lung colonization, and spontaneous pulmonary metastasis [32]. These results suggest a combination of sustained anti-angiogenic therapy besides local treatment for melanoma therapy. Impaired tumor angiogenesis also drives prolonged dormancy of human liposarcoma. This type of tumor with no angiogenic activity produce moderately high amount of the anti-angiogenesis agents TIMP-1 and thrombospondin-1, proposing that the nonangiogenic, microscopic, dormant stage of melanoma might be susceptible to anti-angiogenic treatment years prior to manifestation of malignant disease, or macroscopically detection of anatomical site of a given tumor, by current approaches [101]. Notably, metastatic dormancy can be realized once the proliferation of tumor cell is well-balanced with an equal ratio of tumor cell apoptosis, indicating that anti-angiogenic treatment regulates metastatic outgrowth through increasing of cancer cell death [21].
Besides, the dormancy process can be reversed when dormant cells are fed with adequate angiogenic factors. The reversibility of dormancy envisages the second approach based on awakening and switching silent cells into proliferative and yet susceptible cancer cells to further face the anti-angiogenic therapy. In this line, studies on engineered organotypic microvascular niches established that while a Thrombospondin-1 secretion from endothelial cells induces maintained BCC dormancy, bioactive Periostin and TGF-β1 function as tumor-promoting factors in sprouting neovasculature which further sparks micrometastatic outgrowth [35]. A detailed role of TGF-β related cellular dormancy is recently reviewed in [102]. Moreover, the shift of resting tumors was shown to be correlated with the downregulation of Thrombospondin and a reduced response to Angiostatin in angiogenic tumors. The transformation of quiescent tumors to highly proliferating tumors was also associated with the activation and regulation of molecular cues connected to tumor dormancy in different in vivo cancer models [36]. Also, engineered WM1341B cells constitutively overexpressing the vascular endothelial growth factor (VEGF)/vascular permeability factor (VPF) isoform is shown to terminate long-term melanoma dormancy and can induce overt and progressively growing tumors. These effects were largely counteracted by neutralizing VEGF activity in mice [33].

Cancer Immune-Mediated Dormancy
The tumor-host interactions are progressively known as crucial constituents of cancer growth or inhibition. Particularly, infiltrating immune cells are critical features of the tumor microenvironment [103]. Cancer cells are encircled by stromal cells, including macrophages, fibroblasts, mast cells, neutrophils, and lymphocytes, which communicate through an intricate system of intercellular signaling pathways, mediated by cytokines, adhesion molecules, and the receptor-ligand network [42]. The immune system can confine the metastatic spread of cancer cells, encouraging their long-lasting latency [104]. Nonetheless, the perturbation of the equilibration in favor of immune escape will result in tumor outgrowth rather than the destruction of cancer cells, otherwise, immune surveillance could maintain angiogenic control and impede cancer cell growth (Figure 1) [105]. In this context, results from a valuable study clearly indicated the immune-mediated dormancy in a nontransgenic mouse model, where injection of MHC1-negative GR9-B11 mouse fibrosarcoma clone results in no apparent spontaneous metastases within 24 months after tumor cell injection receiving no antitumor treatment. Of note, immunodepletion of T lymphocytes or asialo GM1-positive cells results in relieving the restraint dormant cells into disseminated MHC-1 positive metastatic cells in mice lungs [106]. Thus, immune-mediated cancer dormancy at least can be regulated by T lymphocytes, as they can keep spontaneous metastases in permanent dormancy. Recently, CD8+ T cells are shown to play a key role in maintaining tumor dormancy, as tissue-resident memory CD8+ T cells (epidermal CD69+ CD103+ TRM cells) are implicated in the melanoma-immune equilibrium and control the active disease as demonstrated in skin-implanted melanoma cancer cells in mice model [107]. Paradoxically, while expression of B7-H1 on the surface of tumor cells impairs dormant cancer cell-killing potential of T-cells, expression of same surface receptors by natural killer (NK) cells (induced by CXCL10) can activate them to prompt a cytotoxic T lymphocyte (CTL) response resulting in destruction of otherwise CTL-killing resistant dormant tumor cells in vitro and in vivo [108].
Meanwhile, an association of human Tregs, neutrophils and inflammatory mediators such as interferon-gamma (IFNγ) are rather context-dependent [109]. With regards to the role of other interferons, a recent study demonstrated the role of IFN-β-induced immunological dormancy reflected in sustained activation of the IRF7/IFN-β/IFNAR axis in chemo-resistant (high dose doxorubicin and methotrexate) ER-negative 4T1 breast cancer cells in vitro and in vivo [110]. Meanwhile, one study to recapitulate sustained inflammation in lungs as demonstrated in smoker individuals revealed that neutrophil extracellular traps (NETs) proteases (elastase and MMP9) produced during inflammation can degrade laminin which further induces proliferation of dormant cells through activating αvβ3 integrin signaling, resulting in aggressive metastatic tumors in mice [111]. Immunotherapeutics that target disseminated cancer dormancy may aid to control or eliminate cancer state. In this view, treatment modalities that induce or amplify the CTL immune response or abrogate CTL immunosuppression mediated by cancer cells might be beneficial to confine or eradicate metastatic cells [112].  Meanwhile, an association of human Tregs, neutrophils and inflammatory mediators such as interferon-gamma (IFNγ) are rather context-dependent [109]. With regards to the role of other interferons, a recent study demonstrated the role of IFN-β-induced immunological dormancy reflected in sustained activation of the IRF7/IFN-β/IFNAR axis in chemo-resistant (high dose doxorubicin and methotrexate) ER-negative 4T1 breast cancer cells in vitro and in vivo [110]. Meanwhile, one study to recapitulate sustained inflammation in lungs as demonstrated in smoker individuals revealed that neutrophil extracellular traps (NETs) proteases (elastase and MMP9) produced during inflammation can degrade laminin which further induces proliferation of dormant cells through activating αvβ3 integrin signaling, resulting in aggressive metastatic tumors in mice [111]. Immunotherapeutics that target disseminated cancer dormancy may aid to control or eliminate cancer state. In this view, treatment modalities that induce or amplify the CTL immune response or abrogate CTL immunosuppression mediated by cancer cells might be beneficial to confine or eradicate metastatic cells [112].
Dormancy state induced by the elimination of primary tumor can be avoided by a "recombinant T cell receptor ligand therapy", dosing days in advance to tumor resection and lasting throughout the whole treatment course, that is a highly speculative therapy [113]. Alternatively, immunestimulated dormancy pathway can be tuned to reverse dormancy and favor dormant cell eradication. For instance, blocking IDO-kynurenine-AhR metabolic circuitry abolishes immunologic latency mediated by IFN-γ in tumor-repopulating cells (TRCs). Hypothetically, the IFN-γ-STAT1 signaling induces apoptosis in proliferating tumor cells; however, TRCs-induced overexpression of IDO1 and AhR, shifts IFN-γ action and give rise to IDO1/AhR-induced p27 activation and inhibition of STAT1 signaling. Further, these events translate into the inhibition of tumor cell apoptosis and thus switching to dormancy program. Blockade of IDO/AhR metabolic circuitry not only nullifies IFN-γmediated latency but also promotes amplified tumor cell growth arrest by IFN-γ-dependent killing of TRCs in vitro and in vivo [48]. As noted before, dormant tumor cells may hinder CTL-mediated tumor lysis via overexpressing B7-H1 and B7.1 [42]. They also counteract apoptosis by paracrine Dormancy state induced by the elimination of primary tumor can be avoided by a "recombinant T cell receptor ligand therapy", dosing days in advance to tumor resection and lasting throughout the whole treatment course, that is a highly speculative therapy [113]. Alternatively, immune-stimulated dormancy pathway can be tuned to reverse dormancy and favor dormant cell eradication. For instance, blocking IDO-kynurenine-AhR metabolic circuitry abolishes immunologic latency mediated by IFN-γ in tumor-repopulating cells (TRCs). Hypothetically, the IFN-γ-STAT1 signaling induces apoptosis in proliferating tumor cells; however, TRCs-induced overexpression of IDO1 and AhR, shifts IFN-γ action and give rise to IDO1/AhR-induced p27 activation and inhibition of STAT1 signaling. Further, these events translate into the inhibition of tumor cell apoptosis and thus switching to dormancy program. Blockade of IDO/AhR metabolic circuitry not only nullifies IFN-γ-mediated latency but also promotes amplified tumor cell growth arrest by IFN-γ-dependent killing of TRCs in vitro and in vivo [48]. As noted before, dormant tumor cells may hinder CTL-mediated tumor lysis via overexpressing B7-H1 and B7.1 [42]. They also counteract apoptosis by paracrine secretion of cytokines (e.g., IL-3) and simultaneous inactivation of SOCS1, the expression of which negatively controls JAK/STAT signaling [114]. In this context, using BCR/ABL DA1-3b mouse model of acute myelogenous leukemia as a model of sustained tumor immunological dormancy, results indicated that demethylation or gene transfer restored the expression of SOCS1, which further rendered dormant cells sensitive to apoptosis and abrogated resistance to CTL-induced tumor cell destruction. Moreover, the cross-resistance to apoptosis emanating from dormant tumor cell-induced overproduction of Interleukin 3 (IL-3) was also upturned using anti-IL-3 antibody [47].
Alternatively, "recombinant T cell receptor ligand therapy" of overt cancer may retain all undetectable DTCs quiescent providing that the treatment is continued as a plan to control cancer [113]. Thus, preventing metastatic expansion would be more feasible than eliminating an established metastasis [113]. As such, tumor-cell immunization prompts tumor cell latency in mice bearing B-cell aggressive leukemia/lymphoma (BCL1). In this line, while an anti-idiotypic immunity was insufficient to eliminate BCL1 cells, it was capable of suppressing growth activating signals, which further promoted cell cycle arrest, apoptosis and persistent BCL1 dormancy in mice [115]. Also, immunity to BCL1, which was achieved by numerous inoculations of irradiated carcinoma cells, precludes leukemia growth in primary and adoptive transfer recipients even with the lifelong perseverance of residual cancer cells [116].
From a different angle, increasing inflammation instigates tumor recurrence, thus interventions that inhibit inflammatory signaling could inhibit cancer relapse [5]. In this sense, a literature search was reported that inflammation and wound healing process following surgery is enough to induce distant cancer outgrowth in cancerous patients [117], especially when the tumor outgrowth was attenuated by the adaptive immune system [118]. Accordingly, anti-inflammatory medications can benefit cancer patients undergoing surgery by avoiding the awakening of latent micrometastatic cells. Furthermore, it was shown that the lipopolysaccharide treatment-induced localized inflammation in the lungs of mice sends a wake-up signal to metastatic latency in the lung parenchyma through Zeb1 expression, a key regulator of the epithelial-to-mesenchymal transition (EMT). Likewise, Zeb1-orchestrated stimulation of EMT program by itself is sufficient to provoke metastatic spread by triggering the stable entry of tumor cells into a state of metastasis-initiating cells [45]. Accordingly, inhibition of Zeb1, LIFR: STAT3 signaling [52] as well as blocking the actions of inflammatory cytokines such as tumor necrosis factor α (TNFα) and IL-1β cytokines could confer dormancy and thus eliminate cancer outgrowth [46].

Metabolic Reprogramming
Impaired vascular system observed in solid tumors encourages establishing oxygen and nutrients deprived microenvironments that harbor metabolically stressed slow-growing cells. Tumor cells under hypoxia and nutrient deprivation become dormant by reducing or shifting their metabolic needs. A second deprivation assault would shut down the possible energy compensatory options for tumor cells and succumb them to apoptotic death. Such that, small molecule VLX600 displays superior cytotoxic action in nutrient-starved environments in vitro and in vivo and is preferentially active against quiescent cells as validated in HCT116 colon cancer spheroids. The anti-tumor potential is correlated with dampened mitochondrial respiration, culminating in bioenergetic catastrophe and tumor cell apoptosis [11]. Likewise, another study has shown the superior performance of FF-10502, a pyrimidine nucleoside antimetabolite over gemcitabine on pancreatic dormant cells in vitro and in vivo, by inducing dormant cell injury in chemotherapy-resistant cells through blocking DNA polymerase β activity and DNA repair [119]. Further, in vivo and organoid cultures to recapitulate metabolic dormancy models revealed the involvement of metabolic shifts in breast cancer relapse. That, blockade of either transportation of fatty acid into mitochondria or synthesis of cellular fatty acid lessens DNA damages and cellular reactive oxygen species (ROS) levels, connecting these hallmarks to lipid metabolism. Tumor relapse can be prevented by direct disruption of these features, either by mitigating the expansion of the residual breast cell population or scavenging ROS [50]. These metabolic changes were also reflected in transcriptomics and histological signatures of residual cancer cells from neoadjuvant-treated breast cancer patients.
An additional strategy would be reinforcing a quiescence-like metabolic state to block tumor evasion. In this regard, a novel metabolic tumor suppressor, LACTB is shown to reduce mitochondrial phosphatidylserine decarboxylase (PISD) protein abundance and phosphatidylethanolamine (PE)/lysophosphatidylethanolamines (LPE) production, resulting in a mitochondrial state consisted of reduced proliferation, increased epithelial phenotype, and a decrease in mesenchymal and cancer stem cell markers, which associated with tumor regression and inhibition of tumor formation, as demonstrated in in vitro and in vivo breast cancer models in mice and humans. Conversely, LACTB silencing in non-tumorigenic breast cell lines cooperated with oncogenic drivers (HRASG12V and MYCT58A) supporting tumor formation in xenotransplants [49]. Notably, results from a recent study showed the role of autophagy as an important regulator of metabolic tumor cell dormancy. That, the lack of autophagy, which is exploited as an important compensatory mechanism by tumor cells to provide their energy requirements, associates with an early breast cancer recurrence and escape from dormancy as demonstrated in neu-overexpressing mouse mammary carcinoma (MMC) [120].

ECM Remodeling
The likely involvement of communications between ECM and tumor cells in metastatic niches determine tumor latency against the metastatic outbreak. Generally, the inability of cancer cells to appropriately adhere to the ECM may potentiate their entry into a dormancy state [121]. Inhibition of one of the components of fibronectin/uPAR/integrin/ERK/MLCK signaling axis as well as suppressing PI3K/Akt pathway is sufficient to prevent/treat metastatic tumor outgrowth, meanwhile promotes dormancy state through favoring the p38 MAPK signaling [122]. As proof of concept, the switch from dormancy to the proliferating state of D2A1 cells was shown to rely on the production of Fibronectin and β1 Integrin signaling, the formation of filamentous actin (F-actin) stress fiber and cytoskeletal reorganization. In this report, integrin β1-dependent phosphorylation of myosin light chain (MLC) by MLC kinase (MLCK) was a prerequisite for F-actin stress fiber formation and exponentially growing of cancer cells. Further, blockade of β1 integrin or MLCK favored dormancy state and MLCK inhibition significantly reduced metastatic expansion in vivo [56]. Similarly, deposition of type I collagen (Col-I) to induce fibrosis is shown to induce the transition of dormant D2.0R cells to proliferating cells by β1 Integrin-mediated Src and focal adhesion kinase (FAK) activation, causing extracellular signal-regulated kinase (ERK)-dependent phosphorylation of MLC by MLCK and formation of actin stress fiber. Blockade of β1-integrin, MLCK, ERK, Src counteracted Col-I-mediated induction of this signaling cascade, cytoskeletal rearrangement, and proliferating state in the faithfully stimulated metastatic microenvironment in vivo [57]. Recently, targeting the perivascular niche with integrin inhibitors are exploited for chemosensitization of breast cancer cells and the prevention of bone metastasis in mice [123]. Likewise, Src knockdown or pharmacological blocking of SFK signaling was shown to promote p27 localization to the nucleus and thwarts transition of quiescent breast cancer cells (BCCs) into the proliferative and metastatic outbreak; still, SFK inhibition was not enough to eradicate residual cells. ERK1/2 activation was also needed for proliferation and awaking dormant cell. The combined therapy of cells undergoing the transition from dormancy to proliferating state with the MEK1/2 inhibitor (AZD6244) and Src inhibitor (AZD0530) potentiated apoptotic death in a large population of the latent cells and postponed the development of the disseminated disease, none of these was achieved with single-agent therapy. These results were obtained using the fibrosis-induced dormant-to-proliferative switch in vitro, in vivo and also ex vivo lung metastasis assay [59]. Meanwhile, EGFR signaling via activation of PI3K/AKT/mTOR and Ras/Raf/ERK axis is another contribution to tumor proliferation [124]. Conversely, TGFβ signaling represents a therapeutic opportunity to activate dormancy (Table 1, Figure 2) [67].
In the context of ECM modulation, an additional therapeutic approach can be envisioned by preventing cancer cell dormancy. For example, a valuable study reported that proliferating and dormant BCCs home in distinct areas within BM niche, with dormant BCCs, mainly occupy perisinusoidal vascular regions that are rich in stromal cell-derived factor 1 (SDF-1) and E-selectin. SDF-1 and E-selectin coordinate opposing functions in BCC, where the SDF-1/CXCR4 axis facilitates adherence of BCCs to the vascular niche, and E-selectin permits BCC entrance into the BM where they remain dormant. Thus, a combined therapy involving CXCR4/E-selectin inhibition will push cells out of their protective niches and aid trapping of the cells in the vasculature, where they could be destroyed with chemotherapy, thus provide an opportunity to control recurrent disease [68]. These observations were monitored by real-time in vivo microscopy of BM in a breast cancer xenograft model. In the context of ECM modulation, an additional therapeutic approach can be envisioned by preventing cancer cell dormancy. For example, a valuable study reported that proliferating and dormant BCCs home in distinct areas within BM niche, with dormant BCCs, mainly occupy perisinusoidal vascular regions that are rich in stromal cell-derived factor 1 (SDF-1) and E-selectin. SDF-1 and E-selectin coordinate opposing functions in BCC, where the SDF-1/CXCR4 axis facilitates adherence of BCCs to the vascular niche, and E-selectin permits BCC entrance into the BM where they remain dormant. Thus, a combined therapy involving CXCR4/E-selectin inhibition will push cells out of their protective niches and aid trapping of the cells in the vasculature, where they could be destroyed with chemotherapy, thus provide an opportunity to control recurrent disease [68]. These observations were monitored by real-time in vivo microscopy of BM in a breast cancer xenograft model.

Cancer Stem Cells (CSCs)
DTCs or surviving tumor cells during therapy course may or may not be stem cells, in either case, however, stemness is inherent to dormancy phenotype [125]. Cumulative evidence suggests that CSCs are indeed metastasis-initiating cells, or metastatic cells acquire CSC-like phenotype upon infiltration into target organs [16,126]. Tumor cell entrance into and out of dormancy is regulated by contextual cues and intrinsic programs, like those that control the self-renewal ability of mature stem cells [125]. Furthermore, a specialized ECM niche nurse reactivation-undergoing metastatic cells, by

Cancer Stem Cells (CSCs)
DTCs or surviving tumor cells during therapy course may or may not be stem cells, in either case, however, stemness is inherent to dormancy phenotype [125]. Cumulative evidence suggests that CSCs are indeed metastasis-initiating cells, or metastatic cells acquire CSC-like phenotype upon infiltration into target organs [16,126]. Tumor cell entrance into and out of dormancy is regulated by contextual cues and intrinsic programs, like those that control the self-renewal ability of mature stem cells [125]. Furthermore, a specialized ECM niche nurse reactivation-undergoing metastatic cells, by supporting positive cues, such as Notch and Wnt, and attenuating negative cues, such as BMP [127]. Adopting a dormant state, CSCs not only can evade therapeutic killing, but the likely reversibility of this situation also poses the real threat, which can potentiate deadly relapse or recurrence decades later [128]. Notably, CSCs can adjust the expression of different surface markers which help them to colonize their desired target organs, in particular, the bone [129]. For instance, expression of chemokine receptor CXCR4 (SDF-1 receptor) by BCCs facilitates bone metastasis where osteoblasts express high levels of SDF-1 [130]. Moreover, the CXCR4/SDF-1 axis not only derives EMT for bone metastasis, but it also promotes breast cancer cell stemness, plasticity and maintenance of CSC-like properties in vivo [68]. Additional CSC-related mechanisms related to cancer cell dormancy are listed in Table 1.

Epigenetic Modification
As another key regulator of tumor dormancy, epigenetic modifications are involved in the epithelial to mesenchyme transition (EMT) which are associated with CSC phenotype and emergence of drug resistance, tumor dissemination, and a high risk of disease relapse. Identifying genes that encode these reversible alterations is an appealing therapeutic plan to combat metastatic disease by inducing differentiation of the mesenchymal cell into an epithelial phenotype [80].
In this respect, epigenetic upregulation of orphan nuclear receptor NR2F1 is detected in DTCs from the breast [131], and prostate cancer patients harboring lifelong dormant disease and in experimental latency models of head and neck squamous cell carcinoma (HNSCC). The exploitation of these epigenetic dormancy-recapitulating models revealed that NR2F1-induced dormancy depends on RARβ, SOX9 and CDK inhibitors. Also, NR2F1 induces pluripotency gene NANOG, and global chromatin repression, favoring dormancy of DTCs in the BM [82].
Likewise, polycomb-like proteins 1-3 (PCL1-3) are substoichiometric modules of the polycombrepressive complex 2 (PRC2) that are indispensable for complex association with chromatin. Their functional redundancy is due to their opposing roles in the positive and negative regulation of cellular proliferation. Such that, in quiescent cells, expression of PCL1, a p53 target gene is predominant, while proliferating cells express E2F-regulated genes PCL2 and PCL3. Ectopic expression of any PCL protein employs PRC2 to suppress the INK4A gene; nevertheless, only PCL2 and PCL3 render an INK4A-evolved proliferative benefit. Of note, PCL1 confers a PRC2 function which acts independent of chromatin, possessing anti-proliferating effect and induce cellular dormancy through binding to and stabilizing p53 [81]. These results were attained using purified PCL expressing cell populations from the mouse hematopoietic system by flow cytometry.
Also, in vivo genome-wide short hairpin RNA screening has revealed a novel epigenetic-related mechanism involved in bone metastatic latency of estrogen receptor-positive (ER+) breast cancer. Clinical studies were reported that low levels of mitogen-and stress-activated kinase 1 (MSK1) expression links with early relapse in ER+ breast cancer patients. In this study, reduced MSK1 impaired the differentiation of breast cancer cells, and increased their bone homing and growth capacities. From a molecular perspective, MSK1 downregulation induces chromatin remodeling and decreases differentiation traits by modulating promoter chromatin status of genes encoding GATA3 and FOXA1 transcription factors and accelerates bone colonization by cells in distant micrometastatic sites [83]. Results from a recent preclinical study underpinned the role of epigenetic dormancy in patients receiving neoadjuvant letrozole treatment. Acquired resistance to estrogen depletion therapy with aromatase inhibitors was attributed to the dormancy highlighted with a global loss of DNA methylation under extended treatment [132]. Thus, epigenetic alteration and "dormancy phenotype" rather than "acquired resistance" can originate from therapy and that epigenetic markers targeting can be promising for the treatment of resistant hormone-depleted tumors.

Noncoding RNAs (miRNAs)
While the critical role of numerous microRNAs in tumorigenesis is well documented [133], the implication of dormancy miRNAs (Dmirs) is just recently illuminated [134]. For example, mesenchymal stem cell-derived exosomes comprising different miRNA contents, such as miR-222/223, are shown to induce cycling dormancy and early BC quiescent in BM and give rise to drug resistance. Further, administration of MSC-loaded antagomiR-222/223 sensitized BC cells to carboplatin-based therapy and increased survival of the mouse as the model for studying miRNA-regulated breast cancer dormancy [84]. Also, a recent study showed that metastatic outgrowth of the claudin-low mammary tumor cell line, RJ423 can be avoided by re-expression of the miR-200b/200a/429 cluster which promotes epithelial phenotype and induces tumor dormancy [87].
Moreover, a study on experimental models of fast-growing and dormant human osteosarcoma identified three dormancy-associated DmiRs regulating osteosarcoma latency: miR-200c, miR-34a, and miR-93. Accordingly, the expression of these microRNAs is lost upon the shift from avascular dormant into angiogenic fast-growing tumor state. Introduction of these miRNAs by dPG-NH2 polyplexes into MG-63 and Saos-2 cells dampened expression levels of their target genes, including hypoxia-inducible factor 1α (HIF1α), MET proto-oncogene, and moesin, vital to tumor migration and angiogenesis process. Furthermore, therapy with dPG-NH2 containing each of these microRNAs considerably extended the latency state of fast-growing osteosarcomas in mice [85].
Loss of DmiRs expression is also verified in human dormant liposarcoma, breast carcinoma, osteosarcoma, and glioblastoma tumors. In this report, transcriptional reprogramming of tumors via over-expression of DmiR-190, 588, or 580 reduced expressions of pro-angiogenic factors bFGF, TIMP-3, and TGFα while upregulated anti-angiogenic and dormancy stimulating factors angiomotin and EphA5 [135]. In all dormant tumors analyzed, overexpression of miR-190 was predominant where upregulation of this Dmir encouraged long-term dormancy of otherwise proliferative osteosarcomas and glioblastomas [17].

Stress-Induced p38 Signaling
Stress-induced activation of p38 signaling is one of the critical pathways related to tumor dormancy, which regulates a transcriptional network of 46 core genes that includes 16 transcription factors (TFs) [91]. p38 MAPK signaling coordinates the induction of growth arrest and drug-resistance in different models of cancer cell dormancy [136]. Imbalances in the activity ratio of ERK to p38 signaling is fundamental to decide the outcome of dormancy vs. tumorigenicity of different experimentally established cancer models [89].
Stress-dependent p38 activation induces dormancy by adopting a pro-survival mechanism via enhanced activation of the PERK and up-regulating the endoplasmic reticulum (ER) chaperone BiP, conferring dormant cells resistant to drug toxicity. Moreover, up-regulation of BiP suppresses activation of Bax. Thus, p38 signaling via PERK activation and BiP up-regulation secures dormant human squamous carcinoma cells (HEp3) from stress insults, such as chemotherapy in vivo [90]. Besides, the computational analysis showed that p38 activation induces the transcription of the TFs BHLHB3 and p53, while blocks FoxM1 and c-Jun expression. Also, p38-mediated activation of p53 requires down-regulation of c-Jun [91].
ATF6α as another p38-controlled transcription factor is crucial to the survival of the dormancy state. That ATF6α is vital for dormant cells adaptation to nutritional stress, chemotherapy, and, most importantly, the in vivo microenvironment. Accordingly, in dormant cancer cells, MKK6 and p38α/β control nuclear translocation and transcriptional activation of ATF6α. Downstream, ATF6α promotes survival via Akt-independent activation of mTOR signaling and up-regulation of Rheb. Thus, targeting the survival signaling axis ATF6α-Rheb-mTOR in dormant carcinoma cells may help to remove residual disease during dormancy periods [92].
As discussed before, dormant cells and active cells occupy different soils to settle in. As nonproliferative DTCs reside in the BM while metastatic growth occurs in other organs such as the lung. Accordingly, BM niche in patients serves as a metastasis "restrictive soil" by coding dormancy activating clues in DTCs. In this view, in an in vivo HNSCC carcinoma model, specific and strong TGF-β2 signaling in the BM triggered the MAPK p38α/β and resulted in low ERK/p38 signaling ratio. This favored dormancy of malignant DTCs via induction of DEC2/SHARP1 and p27, and CDK4 downregulation. Also, TGF-β2-induced latency nodes to activate SMAD1/5, TGF-β receptor-I (TGF-β-RI) and TGF-β-RIII to induce p27. In lungs, a metastasis "permissive soil" with low TGF-β2 levels, DTC latency state was transitory and continued by metastatic expansion. Importantly, systemic inhibiting of p38α/β or TGF-β-RI activities awaked dormant DTCs, and fueled multi-organ metastasis. This work reveals a "seed and soil" mechanism whereby p38α/β regulates TGF-β2 and TGF-β-RIII signaling to determine the fate of DTC dormancy and delineates permissive (lung) and restrictive (BM) microenvironments for HNSCC metastasis [93].

Conclusions and Clinical Implications
Tumor dormancy is commonly regarded as the most threatening aspect of cancer, leading to the emergence of resistance in a short time and advent of lethal metastatic outbreaks after a long latency period lasting from months to years. Now that we begin to realize dormancy tactics, several therapeutic approaches stick out to outsmart cancer. As illustrated in Figure 3, these schemes include: (i) prolonging dormant state; (ii) eradication of dormant cells; (iii) reawakening/mobilization of dormant cells and (iv) avoiding dormancy state.

Conclusions and Clinical Implications
Tumor dormancy is commonly regarded as the most threatening aspect of cancer, leading to the emergence of resistance in a short time and advent of lethal metastatic outbreaks after a long latency period lasting from months to years. Now that we begin to realize dormancy tactics, several therapeutic approaches stick out to outsmart cancer. As illustrated in Figure 3, these schemes include: (i) prolonging dormant state; (ii) eradication of dormant cells; (iii) reawakening/mobilization of dormant cells and (iv) avoiding dormancy state. Prolonging dormant state is based upon the observation that immunotherapeutic targeting of advanced-stage cancers has extended the survival of cancer patients, yet its curative efficacy is limited due to tumor immunoediting. On the other hand, human vaccines have been able to eradicate and control many infectious diseases. The success has resulted from the administration of vaccines in prophylactic settings or during latency periods to protect an individual during future exposure to the disease rather than curing an established disease. Therefore, administration of immunotherapy at the right time is the key to success. Immunotherapeutic targeting of tumor dormancy could be more promising than targeting of advanced-stage disease to achieve a cure for cancer. Besides immunotherapy, identification of Dmirs advocates novel tools to inverse the malignant aggressive phenotype into a microscopic dormant state and may deliver encouraging targets for timely detection or prevention of cancer. Eradication of dormant cells without their awakening is the second tactic in which tricks of dormancy are leveraged against dormant cells. For example, their restrict dependence on survival pathway or altered energy metabolism could be their Achilles heel to interfere with cancer evolution and relapse. Additionally, dormant and resistant cells could be awakened by external stimuli such as growth factors and angiogenic cues, turning them into proliferative yet therapy-sensitive tumors. Prolonging dormant state is based upon the observation that immunotherapeutic targeting of advanced-stage cancers has extended the survival of cancer patients, yet its curative efficacy is limited due to tumor immunoediting. On the other hand, human vaccines have been able to eradicate and control many infectious diseases. The success has resulted from the administration of vaccines in prophylactic settings or during latency periods to protect an individual during future exposure to the disease rather than curing an established disease. Therefore, administration of immunotherapy at the right time is the key to success. Immunotherapeutic targeting of tumor dormancy could be more promising than targeting of advanced-stage disease to achieve a cure for cancer. Besides immunotherapy, identification of Dmirs advocates novel tools to inverse the malignant aggressive phenotype into a microscopic dormant state and may deliver encouraging targets for timely detection or prevention of cancer. Eradication of dormant cells without their awakening is the second tactic in which tricks of dormancy are leveraged against dormant cells. For example, their restrict dependence on survival pathway or altered energy metabolism could be their Achilles heel to interfere with cancer evolution and relapse. Additionally, dormant and resistant cells could be awakened by external stimuli such as growth factors and angiogenic cues, turning them into proliferative yet therapy-sensitive tumors.
Another alternative yet practical strategy would be pushing DTCs out of their protective niche (e.g., anti-CXCR4) and into a less supportive environment, where they can be targeted and eradicated by anti-cancer drugs. As the final approach is avoiding dormancy in the first place by inhibiting pathways leading to the acquisition of dormancy phenotype. However, due to the nature of some cancer cells and their escape from primary tumors at early stages before detection (they are as little as 5 mm in size), it is nearly impossible to prevent dormancy in all instances. Although most of our understanding arises from the preclinical studies, and implementation of the above-mentioned dormancy-based anti-cancer approaches remain to be weighted in the human cancer biology systems, they are reports for identifying and targeting dormant cancer cells in the clinical settings. To this, using disseminated tumor cells (DTCs) as "keyword" to search in clinicaltrials.gov, they were several clinical trials targeting DTCs in breast and prostate cancer patients. All of these trials were attempted to reduce the reservoir of DTCs in the BM to reduce cancer relapse and increase metastasis-free survival in patients. A detailed description of these trials can be found in Table 2.  Even if we can control cancer dormancy, not all cancer patients will ultimately benefit from this strategy. That is, even the clinically undetectable primary tumor could cause death in a portion of patients along with highly aggressive and metastatic tumors in the rest of patients. Although a considerable portion of disseminated dormant cells success to colonize, outgrow and become resistant; however, there is no strong evidence confirming that every single tumor cell that escape primary tumor and colonize in distant sites become resistant.
Though the discovery of cancer cell-derived by-products such as CTCs, cell-free DNA, exosomes, and apoptotic bodies in the blood circulation and body fluids has opened the door for timely detection and non-invasive monitoring of anti-cancer therapy outcomes, given their rarity, detection of minimal residual disease in the patients is difficult, therefore, tactics to target this population of apparently asleep cells is still in its infancy. Killing may look easier in principal, but not for cancer cells; they pretend to be sleep to pretend innocence but in fact, they are just waiting for the right moment (suitable conditions, e.g., clearance of toxic chemotherapeutics) to strike back. In conclusion, as dormancy is not responsible for all metastatic deaths, targeting dormancy in such cases would be useless. Furthermore, targeting the dormant cells by induction of dormant-to-proliferative switches involves the risk of inducing metastatic progression, while drugs that target DTCs or CSCs might affect normal stem cells. At the end of the day, it is the depth of our knowledge regarding tumor dormancy schemes that will determine the fate of our fight against cancer.