Pattern recognition of tumor dormancy and relapse beyond cell-intrinsic and cell-extrinsic pathways

Abstract

In this thematic issue, several mechanisms of tumor dormancy and relapse are discussed. The reviews suggest mutual interactions and communications between malignant cells and other cells in their niche during tumor dormancy. Nevertheless, a complete understanding of tumor dormancy remains elusive. This is because we are getting lost in details of cell-intrinsic and cell-extrinsic molecular pathways without being able to discover the pattern of tumor dormancy. Here, we discuss some conceptual frameworks and methodological approaches that facilitate pattern recognition of tumor dormancy, and propose that settling on certain biological scale such as mitochondria would be the key to discover the pattern of tumor dormancy and relapse.

1. Tumor dormancy as a network of molecular interactions

In this thematic issue, different environmental factors communicating with malignant cells to support tumor dormancy or trigger tumor relapse have been discussed. For instance, Parker et al. discuss the impact of alternative RNA splicing and noncoding (nc)RNAs in promoting metastatic dormancy and disease recurrence in human cancers focusing on both cell-intrinsic and cell-extrinsic drivers of tumor dormancy [1]. Desai et al. discuss tumor immunoediting mechanism of establishing tumor dormancy (equilibrium) and tumor immunoediting mechanism of tumor escape and relapse [2]. Corthay et al. [3] discuss the SNS, danger and quorum models of immunity for the understanding of immunological tumor dormancy. Ramamoorthi et al. [4] discuss molecular pathways and chemokine-chemokine receptor network in regulating dissemination of dormant cells. Also, phenotypic heterogeneity of disseminated dormant cells, and the role of niches and the immune response in tumor dormancy are explored. They also discuss current therapeutic targeting of tumor dormancy, including anti-HER2 therapy, immunotherapy and endocrine therapy. Werner et al. [5] discuss distinct mutant alleles in the primary tumor cells responsible for tumor dormancy as well as an early or late recurrence, suggesting that dormant tumor cells are originated from the primary tumor sites. Also, the role microenvironment in tumor dormancy and relapse associated with autophagy, aging and obesity is discussed. To evaluate different methods for the detection and isolation of dormant tumor cells, Basu et al. [6] discuss advantages and limitations of experimental approaches used to identify, isolate and study slow cycling dormant tumor cells (SCCs). These include CFSE dye, doxorubicin-induced Histone 2B (H2B)-GFP system for long-term tracking of dormant tumor cells in vivo, the Fluorescent Ubiquitination-based Cell-Cycle Indicator or FUCCI system which can facilitate the tracking and purification of both proliferating (S/G2/M) and nonproliferating (G1) cells. Applying the FUCCI system in melanoma models, Puig et al. [7] showed that the slow cycling dormant cells (Ki67^−/low cells) tend to localize at the center of the melanoma, because of hypoxia, whereas the proliferating tumor cells (Ki67^+/high cells) are distributed more towards the periphery of the tumor mass. These Ki67- dormant cells residing at the site of primary cancer could be the source of disseminated dormant cells. Recurrent disease after prolonged tumor dormancy is a major cause of cancer-associated mortality. Attaran and Bissel [8] discuss critical mediators of tumor progression and their link to cancer dormancy, while also exploring possible therapeutics for targeting relapse. Payne [9] discusses our current understanding of stress responses facilitating malignant cell adaptation and metabolic reprogramming to establish tumor dormancy. Finally, Manjili et al. [10] bring into focus how inflammation caused by surgeries and trauma, or chronic inflammatory diseases contribute to the reawakening of dormant tumor cells and lead to cancer recurrence.

2. Tumor dormancy in the paradox of the Darwinian and Lamarckian selections

Dormancy is a cell survival mechanism ubiquitously found in living organisms, including bacteria, yeast, insects, and mammals. At the cellular level, the dormant state is detected in tissue stem cells [11] and tumor cells [12]. There are two theoretical frameworks that shape our understanding of tumor dormancy and relapse. The Darwinian theory of evolution suggests that tumor dormancy and relapse are determined by cell-intrinsic genetic and epigenetic factors as well as cell extrinsic selection pressure. For instance, low levels of the tumor suppressor p53 facilitates tumor dormancy while high levels induces cell death [13], or suppression of two dormancy genes BHLHE41 and NR2F1 results in the relapse of ER⁺ breast cancer [14]. The Lamarckian theory of evolution emphasizes on the role of microenvironment or niche in the process of tumor dormancy and relapse. For instance, tumor dormancy is perpetuated by hypoxia [1,2]. Another example is the osteoblasts release soluble factors that mediate the dormancy of prostate cancer in the bone environment [15]. Leukemia Inhibitory Factor (LIF), a member of the IL-6 cytokine, is also produced by osteoblasts, and provides a pro-dormancy signal to breast cancer cells [11]. While these theories explain cell-intrinsic and cell-extrinsic aspects of tumor dormancy, respectively, a global understanding of tumor dormancy beyond its individual components/characteristics remains elusive.

The papers published in this thematic issue underscore several mechanisms for tumor dormancy, ruling out a single molecular pathway as the sole driver. In fact, the reviews suggest mutual interactions and communications between malignant cells and other cells in their niche during tumor dormancy. This raises a “chicken-or-egg” question about whether tumor cell dormancy comes before or after receiving signals from its niche. This paradox is due to the cell-intrinsic and cell-extrinsic mechanisms of tumor dormancy, which are not separate entities at the molecular level. In fact, molecules outside the cell, within the cell membrane, and inside the cell, communicate in a multidirectional manner. Such mutual interactions work through the interstitial system which is a contiguous fluid-filled space connecting the cells, tissues and organs [16]. Interpreting the pattern of multidirectional interactions as a one-way cause-effect or a two-way action-reaction can result in the “chicken-or-egg” paradox, and in turn hinder our understanding of tumor dormancy as a whole. In fact, at the sub-molecular level, tumor cells and their niche are rearrangements of quantum particles interacting with one another without borders (Fig. 1). This is our brain that organizes/translates different rearrangement of particles into structures and boundaries similar to neurological organization/translation of the reflection of the light with different wavelengths into different colors whereas no color exists in the nature. The study of illusory colors, i.e., colors that the brain is tricked into seeing, demonstrates that processing of color in the brain is associated with the processing of other properties, such as shape and boundary [17,18]. Structuring the chaotic particles interacting in multidirectional networks into separate entities with boundaries, such as cells, tissues, organs and systems makes learning easier and helps us communicate with our environment. In fact, cell division, cell death and cell cycle arrest or dormancy are different patterns reflecting different rearrangements of cellular particles. Discovering the pattern of multidirectional interactions that shapes tumor dormancy without getting lost in molecular details would enhance our understanding of the tumor dormancy as a whole. A number of immunotherapeutic strategies including adoptive T cell therapy, CAR T cells, engineered T cells, immune checkpoint inhibitors, antibody therapy, and vaccines have been tested, yet tumor relapse has not been overcome [19]. This is because all therapeutics are focused on targeting tumor cells without detecting and modulating the network of molecular interactions that support tumor dormancy. Detection of the molecular network would not be feasible if we continue to take a highly reductionistic approach by focusing on the components rather than looking for patterns. Recent findings show that the brain processes pattern learning, also called probabilistic learning [20]. During pattern recognition, the brain is looking for rules to help make better and faster predictions. To this end, laws of thermodynamics governing molecules may provide a conceptual platform for discovering the pattern of tumor dormancy. Erwin Schrödinger, in his most famous book “What is Life? The Physical Aspect of the Living Cell” [21], described the fundamental law of inheritance as specific arrangements of quantum particles that create a structure carrying information and passing it from generation to generation. Later on, Watson & Crick proposed a pattern for this rule by modeling DNA double helix carrying genetic information [22]. Without such pattern and modeling, it would not have been possible to understand the molecular basis of inheritance. The double helix DNA pattern was discovered by X-ray crystallography and settled at the scale of four nucleotide base-pairs and a triplet genetic code before digging into more molecular details. The challenge we currently are facing is to discover the patterns of tumor dormancy and tumor relapse by settling at a certain biological scale rather than digging into more details.

Fig. 1.

Cellular pattern recognition of organized molecular particles.

3. Thermodynamics of tumor dormancy and relapse

Thermodynamics in biology refers to energy transfers in living organisms. The first law of thermodynamics states that energy cannot be created or destroyed, but it can be transformed. For instance, cellular metabolism and processes are driven by the energy transformed from glucose into high-energy electron carrier NADH and ATP during glycolysis in the cytoplasm, as well as from oxidation of carbohydrates, fat and proteins through electron transfer cascade in the mitochondria. The second law of thermodynamics states that during biological reactions energy is not created or destroyed, but it is transferred or transformed from a more-useful form into a less-useful form as wastes such as heat and toxins in the cell. For instance, only 20 % of the energy produced during glycolysis and ATP synthesis is used and the rest of the energy is converted to heat, reactive oxygen species (ROS) (O₂⁻ and H₂O₂) and other wastes (CO₂ and H₂O), thereby increasing the disorder or entropy of cells [23]. To survive, our body has developed mechanisms for reducing the entropy of cells by releasing wastes and heat into their microenvironment as well as producing anti-oxidants such as superoxide dismutase, catalase, glutathione peroxidase and glutathione reductase. If the heat and wastes are not efficiently neutralized or transferred into the environment, it could induce DNA break or mutation, protein aggregation and mutagenesis which in turn result in apoptosis or cell cycle arrest of nascent transformed cells. The latter establishes cellular hibernation or naturally-occurring tumor dormancy [24] by reducing entropy production through cessation of cell division and reduced metabolism. After a period of natural dormancy, malignant cells become highly efficient in transferring their entropy into their microenvironment and develop cancer. Actively proliferating tumor cells usually have higher temperature and entropy as well as lower pH than their microenvironment, because of high glycolytic activity that produces heat and acids in tumor cells. It has been reported that the excess heat and entropy flows from the tumor to surrounding tissue [25,26], an indication of the ability of tumor cells in lowering their entropy and increasing the entropy of the surrounding tissue as well as changing the state of their entropy through unlimited cell division. This ability allows tumor cells survive indefinitely while altering their microenvironment. Cancer therapies can further increase the entropy of tumor cells such that some entropy overloaded tumor cells undergo apoptosis and some other clones reduce their entropy through cellular hibernation or treatment-induced tumor dormancy [24]. Drug resistance is different from tumor dormancy as dormant cells respond to therapeutics by the cessation of cell division whereas resistant cells continue to grow in the presence of cancer therapies. The latter is due to high ability of tumor cells to transfer their treatment-induced entropy into their microenvironment more efficient than normal cells such that increasing the therapeutic dose would be more toxic for normal cells than malignant cells.

Chemotherapies as well as anti-tumor T cells and macrophages that produce IFN-γ and TNF-α can increase the entropy of tumor cells by elevating ROS including hydrogen peroxide and nitrite oxide levels in tumor cells [27,28]; an excessive ROS production can induce tumor cell death as well as trigger DNA damage response, activation of checkpoints and cell cycle arrest [29]. Anti-inflammatory agents such as aspirin, which inhibit ROS and decrease heat, have been shown to inhibit cancer progression [25]. It was reported that during cancer therapies, some tumor clones can still decrease their entropy by upregulating the anti-oxidant transcription factor NRF2 leading to tumor dormancy [30]. The reduction of energy demands in dormant tumor cells due to cessation of cell division and reduced metabolism can slow down ATP synthesis, thereby reducing entropy production in dormant cells. However, stress hormones or glucocorticoids increase energy demands by affecting mitochondria to enable stress adaptation [31], and also turn anti-tumor immunity into pro-tumor immunity by altering neutrophils and awakening dormant cells [32]. Increased energy demands during stress results in hyperglycemia, glucose oxidation and increased production of energy during glycolysis. All these events are key factors in awakening dormant tumor cells after the cessation of cancer therapeutics. To this end, triple negative breast cancers (TNBC) with highest rates of recurrence express high levels of glucocorticoid receptors [33], rendering them more susceptible to relapse promoting stress hormones. Therefore, entering or coming out of tumor dormancy as well as interactions of cell-intrinsic and cell-extrinsic pathways may be better understood by measuring and modulating the entropy of malignant cells.

4. A future perspective on tumor dormancy: pattern recognition modeling of mitochondria

Molecular interactions and rearrangements can create distinct patterns that cannot be fully understood by taking a reductionistic approach into understanding their components. For instance, drinking water (H₂O) or toxic hydrogen peroxide (H₂O₂) and hydroxyl radical (HO) are made up of different combinations hydrogen and oxygen, producing different molecular pattern/structure with new properties above and beyond the property of their components. Therefore, property of water cannot be understood by breaking it down to hydrogen and oxygen. Nevertheless, current trends in research on tumor dormancy rely on excessive reductionistic approaches trying to pinpoint a single molecular component driving tumor dormancy or relapse as well as focusing on cell-extrinsic mechanisms contributing to tumor dormancy. Accordingly, therapeutic strategies emerging from current trends are focused on molecular targeting of CDK4/6, ERK, Src, Wnt, Scr and MEK1/2, autophagy, MAPK, VEGF, TAMs and more targets to come [34]. Targeting these pathways while showing some effects has failed to control tumor dormancy effectively. To this end, although TAMs can absorb the entropy of tumor cells by the expression of scavenger receptors that take up wastes released by the tumor, depletion of macrophages for the inhibition of tumor resulted in edema [35]. This suggests that a single target strategy such as TAM depletion cannot work effectively because of altering the entropy absorbing function of macrophages that supports other cellular homeostasis. Therefore, tumor dormancy will remain a mystery until we develop a model that detects the pattern of tumor dormancy, i.e., a structure that encompasses key molecular pathways without getting lost in details. To detect the pattern of tumor dormancy, we need to find and settle on certain biological scale rather than keep digging into more redundant molecular details that could distract us from discovering the pattern. For instance, pattern recognition of enzymes helped us understand their functions. The four protein structures that make up all enzymes show how important it is to focus on the right scale to detect the functional pattern. The primary structure includes amino acid chains. The secondary structure includes alpha helices and beta sheets, formed by these amino acid chains. The tertiary structure is the formation of a three-dimensional model due to the interactions between helices and sheets. The quaternary structure is a complete protein made up of multiple amino acid chains. The function of an enzyme depends solely on its interaction with specific substrate due to its three-dimensional pattern. Accordingly, it may be time to step back and focus more on the pattern of tumor dormancy as opposed to digging into more details of signaling molecules, transcription factors, microenvironmental cues etc.

In regards to discovering the pattern of tumor dormancy, the biological scale to settle on seems to be the mitochondria because of its key role in cell death and cell cycle arrest during tumor response to cancer therapies [36]. Interestingly, red blood cells are the only human cells that lack mitochondria, and they do not become malignant. An excessive red blood cell production or polycythemia vera is in fact a disorder of the bone marrow or myeloproliferative neoplasm rather than malignancy of red blood cells [37]. Also, hereditary BRCA1 mutation is not sufficient for cell transformation, and requires alterations in p53 function such that over 90 % of human BRCA1 deficient breast cancers also carry p53 mutations, while p53 alterations are only found in about 40 % sporadic breast cancers [38]. Interestingly, p53 regulates many mitochondrial functions, and reciprocally, p53 activity is regulated through ROS-mediated mechanisms by mitochondria. This suggests the key role for mitochondria in discovering the pattern of tumor cell dormancy. In addition, activation of wild-type BRCA1 is due to its hyperphosphorylation, and hyperphosphorylated form of BRCA1 was detected in the mitochondria for protecting the integrity of mitochondrial (mt)DNA [39]. Therefore, mutant BRCA1 fails to protect the mtDNA integrity and it would take a while for the mtDNA mutations to accumulate and cause cancer because epithelial cells contain numerous mitochondria. These findings indicate multidirectional interactions of mitochondria during tumor cell death and cellular dormancy, and its central place in understanding of the pattern of tumor dormancy and relapse. A broader pattern recognition approach to understand tumor progression or inhibition would be to discover immunological patterns reflecting the complex interactions of the immune cells that generate a new function independent from its constituents.