Extracellular Events Involved in Cancer Cell–Cell Fusion

Fusion among different cell populations represents a rare process that is mediated by both intrinsic and extracellular events. Cellular hybrid formation is relayed by orchestrating tightly regulated signaling pathways that can involve both normal and neoplastic cells. Certain important cell merger processes are often required during distinct organismal and tissue development, including placenta and skeletal muscle. In a neoplastic environment, however, cancer cell fusion can generate new cancer hybrid cells. Following survival during a subsequent post-hybrid selection process (PHSP), the new cancer hybrid cells express different tumorigenic properties. These can include elevated proliferative capacity, increased metastatic potential, resistance to certain therapeutic compounds, and formation of cancer stem-like cells, all of which characterize significantly enhanced tumor plasticity. However, many parts within this multi-step cascade are still poorly understood. Aside from intrinsic factors, cell fusion is particularly affected by extracellular conditions, including an inflammatory microenvironment, viruses, pH and ionic stress, hypoxia, and exosome signaling. Accordingly, the present review article will primarily highlight the influence of extracellular events that contribute to cell fusion in normal and tumorigenic tissues.


Introduction
Fusion processes are understood as a merger of phospholipid bilayer membranes of, e.g., vesicles (e.g., endocytosis and exocytosis), cell organelles (e.g., mitochondria, exosomes), infection of host cells with enveloped viruses, and hybridization of cells. The fusion process itself is a complex, energy-dependent, and tightly and temporally regulated event but only partially understood to date (for review, see: [1][2][3][4][5][6][7][8][9][10][11][12][13][14]). Cell fusion can occur as homofusion (fusion of the same cell type) or heterofusion (fusion between different cell populations) in the course of physiological developments, e.g., during formation of myocytes, osteoclasts, or syncytiotrophoblasts, among others [15]. Pathophysiological heterofusion, such as fusion of breast cancer cells with mesenchymal stroma-/stem-like cells (MSC), has been documented as a rapid event that occurs within a few minutes [16]. The resulting cancer hybrid cells enter a post-hybrid selection process (PHSP). Such PHSP is required to eliminate DNA-unstable karyotypes generated by chromosomal recombination of different nuclei during the previous cell merger [3,17,18]. Cancer hybrid cells have been identified in various tumor types, including breast, ovarian, gastric, and lung tumors following interaction and subsequent heterofusion of corresponding cancer cells with MSC [19][20][21][22]. Thus, cancer cell fusions may elevate the risk to develop new cancer stemlike cells [18,23]. The cancer cell × MSC fusion is predominantly related to an activated presence and the supportive functionality of MSC within the tumor microenvironment (TME). In general, MSC exhibit regenerative and repair activities in damaged tissues and are, therefore, also recruited to tissue lesions of invasively growing tumors [24][25][26][27]. At tumor sites, MSC also display their regenerative capacities, which simultaneously can support tumor growth. Fusion of cancer cells with MSC generates a variety of different hybrid subtypes with altered properties, including enhanced tumor growth and metastatic spreading [17], reduced tumorigenicity [22], or tumor dormancy [28]. Accordingly, cancer cell heterofusion contributes to significantly elevated tumor heterogeneity and plasticity [2,29,30].
In addition to these key elements in cell-cell fusion, other proteins involved in this process have been identified, including cell adhesion molecules, F-actin and F-actin reorganizing proteins, proteases, chemokines, and cytokines [1][2][3][4][5][6][7][8][9][10][11][12][13][14]37]. F-actin and proteins involved in actin cytoskeleton reorganization, such as Arp2/3/WASP, Cdc42, and Rac1, have already been shown to be actively involved in cell-cell fusion of, e.g., myoblasts or breast cancer cells [8, 37,38]. In this process, F-actin-based protrusions penetrate the target cells and cause fusion pores so that the cells fuse with each other. For the other protein classes mentioned, a direct connection with the fusion of the membranes has not yet been demonstrated. Nevertheless, chemokines and cell-cell adhesion molecules, for example, are essential for cell-cell fusion so that two cells can move toward each other and enter into contact [39].
For cell-cell fusion, it is assumed that cells have to acquire pro-fusogenic status in order to fuse [1][2][3][4][5][6][7][8][9][10][11][12][13][14]. In fact, cells are not fusogenic by themselves but must express certain proteins/factors, such as fusogens or PS, the appropriate receptors, to be able to fuse with each other. Similarly, once fusion has occurred, cells must revert to a non-fusogenic state to avoid undesirable subsequent fusion events. Cellular and molecular mechanisms that are important for expression of this hypothetical state remain to be explored. Likewise, clear definitions of pro-fusogenic state need to be articulated.

Physiological and Non-Physiological Cell-Cell Fusion
Placental and muscle maturation represent physiological, genetically determined cellcell fusion processes that are characterized by intrinsic expression of the major components of the cell-cell fusion machinery. These expression levels represent a pro-fusogenic state in a high number of cells, which ensures a high frequency of cell-cell fusion events to occur in parallel [1][2][3][4][5][6][7][8][9][10][11][12][13][14] (Figure 1A). In contrast, bone-marrow-derived cells (BMDCs) represent a population of cells that can regenerate tissue by cell-cell fusion but do not themselves possess intrinsic basal fusogenic properties [40][41][42][43]. While this regenerative capacity underlies regulatory controls, appropriate extracellular processes are also required to shuttle BMDCs into a pro-fusogenic state ( Figure 1A).

Figure 1.
Both intrinsic and extracellular events can convert cells to a pro-fusogenic state, as postulated in this hypothetic model. (A) Intrinsic events likely play a role in physiological cell-cell-fusiondependent processes, such as placentation and myogenesis. Here, intrinsic events lead to time-dependent and developmental-state-dependent expression of the major components of the cell-cell fusion machinery, which ensures a high frequency of cell-cell fusion events. In addition, extracellular events could also convert cells towards a pro-fusogenic state, which would enable BMDCs in tissue regeneration. In a normal state, BMDCs are non-fusogenic but could be converted into a profusogenic state, which is most likely triggered by damaged tissue-derived extracellular signals. (B) Cancer cells could fuse either spontaneously or following induction, suggesting that both intrinsic and/or extracellular events are controlling this process.
In comparison, fusion of cancer cells is different since only a minor proportion of the cells exhibits fusogenic capacities. Given that about 1% of cancer cells can undergo homofusion or fuse with other cells, such as macrophages or MSC, this also means that 99% of cancer cells do not fuse. However, why are these 1% of cancer cells fusogenic, and how are they different as compared to the 99% non-fusogenic cancer cells? Several studies have shown that syncytin-1 is involved in cancer cell fusion [44][45][46][47]. However, the relative fusion rate is usually very low (mostly between 1% and 2%). This is in contrast to the high syncytin-1-mediated fusion frequency of cytotrophoblasts. Immunohistochemical studies of breast cancer, colorectal cancer, lung cancer, and endometrial cancer tissues revealed heterogeneous syncytin-1 expression among the cancer cells [45,46,[48][49][50]. These data indicate that this fusogen is expressed only in a certain compartment of the tumor and/or in a specific subpopulation of cancer cells. In contrast, other studies revealed that syncytin-1 or SyHP were ubiquitously expressed in virtually all cancer cells [44,51]. Interestingly, Figure 1. Both intrinsic and extracellular events can convert cells to a pro-fusogenic state, as postulated in this hypothetic model. (A) Intrinsic events likely play a role in physiological cell-cellfusion-dependent processes, such as placentation and myogenesis. Here, intrinsic events lead to time-dependent and developmental-state-dependent expression of the major components of the cell-cell fusion machinery, which ensures a high frequency of cell-cell fusion events. In addition, extracellular events could also convert cells towards a pro-fusogenic state, which would enable BMDCs in tissue regeneration. In a normal state, BMDCs are non-fusogenic but could be converted into a pro-fusogenic state, which is most likely triggered by damaged tissue-derived extracellular signals. (B) Cancer cells could fuse either spontaneously or following induction, suggesting that both intrinsic and/or extracellular events are controlling this process.
In comparison, fusion of cancer cells is different since only a minor proportion of the cells exhibits fusogenic capacities. Given that about 1% of cancer cells can undergo homofusion or fuse with other cells, such as macrophages or MSC, this also means that 99% of cancer cells do not fuse. However, why are these 1% of cancer cells fusogenic, and how are they different as compared to the 99% non-fusogenic cancer cells? Several studies have shown that syncytin-1 is involved in cancer cell fusion [44][45][46][47]. However, the relative fusion rate is usually very low (mostly between 1% and 2%). This is in contrast to the high syncytin-1-mediated fusion frequency of cytotrophoblasts. Immunohistochemical studies of breast cancer, colorectal cancer, lung cancer, and endometrial cancer tissues revealed heterogeneous syncytin-1 expression among the cancer cells [45,46,[48][49][50]. These data indicate that this fusogen is expressed only in a certain compartment of the tumor and/or in a specific subpopulation of cancer cells. In contrast, other studies revealed that syncytin-1 or SyHP were ubiquitously expressed in virtually all cancer cells [44,51]. Interestingly, all the studies had in common that syncytin-1 was mainly detectable in the cytosolic part and only partially in the cell membrane of the cancer cells [44][45][46][48][49][50][51]. This would be an explanation for the relatively low fusion rate of syncytin-1-expressing cancer cells. Only plasma-membrane-associated syncytin-1 is capable to mediate fusion.
While the process of cell-cell fusion is still not fully understood, some of the involved mechanistic processes have been identified [4,5,10]. However, knowledge regarding temporal orchestration of an appropriate molecular platform to control this process remains scarce. From a physiological point of view, this particularly applies to cells that do not constitutively provide any fusogenic properties but can nevertheless fuse with other cells, including cancer cells. The fact that cancer cells can fuse either spontaneously or following induction suggests the presence of intrinsic and/or extracellular factors controlling this process ( Figure 1B). Intrinsic factors represent, e.g., proteins and phospholipids that are expressed by fusogenic cells themselves. This could be either constitutively or only at specific time points, such as syncytin-1 in cytotrophoblasts [62,63] or myomaker and myomerger in myoblasts [11,64], respectively. Likewise, extracellular factors represent events from the outside of cells to induce/mediate fusion processes. These events include, among others, inflammatory conditions and corresponding cytokines, hypoxia, low pH, viruses, and extracellular vesicles (EVs)/exosomes that will be discussed further in this review article.

Inflammation/Inflammatory Cytokines in Cell Fusion
Inflammation and inflammatory cytokines, such as TNF-α, are well-known triggers/inducers of cell-cell fusion that has been demonstrated in several studies [16,46,[52][53][54][55][56][57][58][59][60]. Yan et al. demonstrated that TNF-α stimulation resulted in increased syncytin-1 expression levels in squamous cell carcinoma cells 9 (SCC-9) and its cognate receptor ASCT2 in human umbilical vein endothelial cells (HUVEC), which was correlated with increased fusion frequency of these cells [46]. These findings substantiate that fusogens and their corresponding receptors are mandatory for cell-cell fusion. Similarly, these data indicate that inflammation/inflammatory cytokines likely prime different cell types for subsequent plasma membrane fusion.

Impact of Proteases and β-Catenin in Cell-Cell Fusion
TNF-α-induced expression of MMP-9 was involved in macrophage fusion and human M13SV1-EGFP-Neo breast epithelial cells × human MDA-MB-435 cancer cell hybridization [56][57][58]60]. Skokos et al. assumed that MMP-9 cleaves substrates, such as E-Cadherin, in which degradation might be necessary for cell-cell fusion [56]. Close cell-cell contact mediated by adhesion molecules is mandatory for cell-cell fusion. Hence, membrane proteins, including adhesion molecules, must be removed/digested to allow the plasma membranes of the two cells to approach each other. The cytoplasmic domain of E-Cadherin binds to β-catenin, which is linked to the cytoskeleton, thereby playing a role in regulation and coordination of cell-cell adhesion [66]. On the other hand, Wnt signaling via transcription co-activator β-catenin controls embryonic development and adult homeostasis [66]. Given that MMP-9 cleaves E-Cadherin in macrophages [56], it can be assumed that release of β-catenin to the cytoplasm could have an impact on cytoskeletal reorganization and an altered gene expression pattern, which might play a role in cell-cell fusion. Takezawa et al. reported that β-catenin might be a molecular switch that regulates transition of cell-cell adhesion to fusion [67]. Thereby, at fertilization, adhesion of sperm to the surface of an oocyte is mediated by the E-Cadherin/β-catenin complex [67]. Moreover, subsequent rapid proteasomal degradation of β-catenin in the oocytes is required for sperm × oocyte fusion [67]. These findings likely indicate that β-catenin-dependent sperm × oocyte fusion is rather attributed to reorganization of the cytoskeleton in the oocyte.
Maretzky and colleagues demonstrated that ADAM10-mediated cleavage of E-Cadherin was associated with β-catenin translocation from the plasma membrane to the cytoplasm and nucleus and expression of β-catenin downstream genes [68]. Similarly, ADAM12directed ectodomain shedding of E-Cadherin potentiated trophoblast fusion, most likely through induction of glial cell missing-1 (GCM1) and syncytin-1 expression [69]. Indeed, studies of Matsuura et al. demonstrated that β-catenin/BCL-9-like (BCL9)/T-cell factor 4 (TCF4) signaling directly targeted the GCM1/syncytin pathway and thereby regulated fusion of human choriocarcinoma BeWo cells [70]. Here, siRNA-mediated knockdown of BCL9L, β-catenin, or TCF4 in forskolin-treated BeWo cells all resulted in significant downregulation of forskolin-induced expression of GCM1 and syncytin-1 and syncytin-2 and fusion [70]. Similarly, TNF-α induced syncytin-1 expression in oral squamous carcinoma cells was dependent on Wnt/β-catenin signaling [46]. These data further substantiate the potential impact of β-catenin signaling in cell-cell fusion due to regulation of syncytins and other fusion-relevant protein expression. With regard to TNF-α-induced and MMP-9-dependent fusion of M13SV1-EGFP-Neo breast epithelial cells and human MDA-MB-435 cancer cells, the potential role of β-catenin signaling is not yet clear. E-Cadherin and MMP-9 are expressed by (TNF-α stimulated) M13SV1-EGFP-Neo breast epithelial cells but not MDA-MB-435 cells [57,58]. Thus, it might be assumed that MMP-9 mediated cleavage of E-Cadherin would result in release of β-catenin to the cytoplasm and nucleus concomitant with expression of fusion-relevant proteins only in M13SV1-EGFP-Neo breast epithelial cells. However, it cannot be ruled out that TNF-α-induced signal transduction pathways directly lead to β-catenin activation. This, for instance, has been demonstrated in glioma cells [71], and, more importantly, also in human breast epithelial cells [72].
Together, inflammation/inflammatory cytokines, such as TNF-α, can trigger cell-cell fusion, and some data point to involvement of β-catenin signaling. This could be either mechanistically (β-catenin degradation leads to actin reorganization) or transcriptionally (β-catenin-dependent gene expression), suggesting that β-catenin signaling might play a role in cell-cell fusion, e.g., by cellular conversion towards a pro-fusogenic state.

Effects of pH and Ions in Cell Fusion
Characteristically, many viral fusogens are initially in a non-fusogenic state. Only transition to an active, pro-fusogenic state after binding to the corresponding receptor by proteolytic cleavage or by pH-dependent irreversible conformational change leads to release in the fusogenic domain [4,5,10,73,74]. Interestingly, only certain viruses with class I, II, and III fusogens, such as influenza virus, HIV, Semliki Forest virus (SFV), dengue virus, and vesicular stomatitis virus (VSV), require low pH to be infectious [4,5,10,73,74]. Strictly speaking, there is no fusion with the plasma membrane of the host cell. Instead, viruses fuse with the membrane of endosomes because low pH values prevail in these vesicles. This rules out a direct role of low pH in cell-cell fusion. Accordingly, no pH dependency has been described for known human fusogens to date. However, a pH dependency was detected for two syncytin homologs in ruminants and lizards [75,76]. Indeed, syncytin-Rum1 and Mab-Env1 are fusogenic only at pH values of 5.0 and 4.0, respectively [75,76].
Data of Liang et al. proposed a pH-dependent mechanism of transmembrane member 16F (TMEM16F) activity [77]. Thereby, a low intracellular pH attenuated and a high intracellular pH potentiated TMEM16F channel and scramblase activity [77]. However, intracellular pH-dependent regulation of TMEM16F activity in normal cells and cancer cells remains elusive. If a change in intracellular pH for stimulation of Ca 2+ -activated phospholipid scramblases (Ca 2+ -PLS) is necessary for cell-cell fusion, this effect should be paralleled by a corresponding mechanism that is still obscure. It is known, however, that dysregulation of intracellular pH is one of the hallmarks of cancer cells [78]. Cellular pH is regulated by a variety of ion transporters and pumps, such as Na + -H + exchangers, Na + -driven bicarbonate transporters, and H + /K + -ATPase proton pump [79,80]. However, it remains to be investigated by which processes/mechanisms these ion transporters and pumps could be specifically engaged to increase intracellular pH and enable Ca 2+ -PLS activation, whose activity is crucial for cell-cell fusion.
Interestingly, fusogenic properties have also been demonstrated for various positively charged ions, such as Ca 2+ and Mg 2+ [81]. For example, Ca 2+ can induce fusion of PS vesicles or PS: phosphatidylcholine (PC) vesicles. This leads to attachment of positively charged Ca 2+ ions to negatively charged PS or PS: PC membrane and destabilization of the structure. Accordingly, the total amount of positively charged Ca 2+ ions as extracellular ionic stress affect phospholipid membranes by induction of aggregation and thereby potentially contributing to fusion [81].
A summary of Sections 2 and 3 is provided in Figure 2.
Viruses can fuse cells by two different mechanisms [88]. In the first mechanism, a host cell is infected by a virus and virus-specific proteins, such as env elements in the cell membrane, are expressed [88]. Since these have fusogenic properties, fusion with other cells and formation of a heterokaryon can subsequently occur. Even sole expression of viral fusogens is sufficient, which is exploited, e.g., in therapy of oncolytic viruses [89][90][91]. In fact, oncolytic-virus-mediated fusion of cancer cells and formation of syncytia increases the effectiveness of therapy and targeted elimination of cancer cells [89][90][91]. To what extent viral membrane-cell membrane fusion can lead to accumulation of viral fusogens in the host cell membrane, including possible cell fusion events, is unknown. It might be assumed that this mechanism is rather unlikely. A certain concentration and possibly also an accumulation of fusogens in the plasma membrane would be necessary for interaction with corresponding receptors and fusion of the neighboring cell to occur. If the number of viral fusogens is too low, a dilution effect will quickly occur as they will spread quickly in the plasma membrane of the host cell.
In the second mechanism, it is believed that viruses do not infect the cells themselves but rather serve as a connective element that links two cells like a bridge, mediating cell fusion [88]. However, it is unclear whether such a mechanism exists and, if so, exactly how it would work. Ultimately, there must be a concerted action between cells and viruses here so that the viruses can fuse in parallel with two cells simultaneously.
With regard to cancer cell × normal cell fusions, however, it is unclear whether these can be mediated by viruses at all since viruses can only infect certain cells due to their tropism. For example, virus "A" can infect (cancer) cell "A" because it has the appropriate receptor "A" to which the virus can dock. (Cancer) cell "A" then expresses virus "A" env proteins so that it can fuse with other (cancer) cells "A". Cell "B", which does not have the receptor "A", cannot be infected by virus "A". Ergo, (cancer) cell "A" that expresses virus "A" env proteins cannot fuse with cell "B".
However, since infection of (cancer) cell "A" with virus "A", as described above, can lead to reactivation of HERV elements [92][93][94][95][96][97], it is conceivable that (cancer) cell "A" can then express HERV env proteins so that fusion with cell "B" can occur. A prerequisite for this is, of course, that cell "B" would have the appropriate receptors for HERV env proteins. Whether such a mechanism can take place at all is completely unclear and also very speculative.
Interestingly, even retroviral transduction of cells, which is routinely used to generate cell lines expressing exogenous non-viral genes, sufficiently induced syncytin-1 expression in transduced PC3 prostate cancer cells [114]. Moreover, exosomes derived from retrovirally transduced cells carried syncytin-1 and were able to transduce target cells via horizontal gene transfer. This supports the assumption that endogenous syncytin-1 mediates delivery of extracellular vesicle cargo into target cells through fusion [114]. Indeed, syncytin-1 was found in the membranes of placenta-derived exosomes, which was important for exosome-mediated intercellular communication [115,116]. However, the data of Uygur and colleagues showed that the sole presence of syncytin-1 in the membrane of exosomes was not sufficient to mediate gene transfer [114]. This was further dependent on MLV-GAG, which was encoded by the retroviral construct used for transduction [114]. The authors hypothesized that MLV-GAG promoted syncytin-1-mediated fusion, possibly by raising local concentration of syncytin-1 [114]. Briefly, these findings indicate that retroviral transduction of cells likely induces HERV element expression in target cells, which might lead to the origin of exosomes sharing some similarities with retroviruses. In this regard, it would be interesting to investigate whether other retroviruses, such as HIV, or reactivation of HERV elements may also lead to generation of syncytin-1 or other env-proteins-positive retroviral-like exosomes.
In summary, an increasing body of evidence reveals a crucial role of reactivated HERV elements, especially syncytin-1 and HERV-K (HML-2), in cancer progression [98][99][100][101][102][103]. However, the mechanism of reactivation of HERV elements is still scarcely understood. Hence, it remains unknown whether aberrant expression of HERV elements might be a direct cause of the disease or solely plays a role in disease progression. A possible link between cell-cell fusion and cancer progression has so far only been demonstrated for HERV env element syncytin-1. In addition to its fusogenic properties, increased syncytin-1 expression also has an impact on cancer cell proliferation [44][45][46]48,50,117,118]. In contrast, the prospective impact of other HERV env elements in cell-cell fusion in cancer remains to be elucidated. Data of Huang et al. demonstrated that the HERV-K (HML-2) env element principally exhibited fusogenic properties [111], but this has only been conducted in melanoma cell lines so far. Thus, additional studies with further cancer cell lines of different tumor entities should be performed first to validate the prospective fusogenic properties of HERV-K (HML-2) env in cancer. Exogenous viruses have been further identified as reactivators of HERV element expression in various cell types [92][93][94][95][96][97]102,112]. In this regard, it would be interesting to investigate whether infection of cancer cells with exogenous viruses would also have an impact on the tumor cells' HERV element expression pattern. Moreover, it would be worthwhile to speculate whether reactivation of a specific HERV subtype, such as HERV-K (HML-2), might also have an impact on the gene expression of other HERV subtypes/elements, such as syncytin-1.

Role of Hypoxia in Cell Fusion
A characteristic feature of advanced cancers is a hypoxic TME [119][120][121][122]. Of particular importance is activation of transcription factor hypoxia-inducible factors-1α (HIF-1α), which leads to expression of numerous hypoxia-associated target genes [119][120][121][122]. The specific role of these genes/proteins in cancer cell-cell fusion remains to be investigated. This further includes the question of which cell types in the hypoxic and inflammatory TME are converted to a pro-fusogenic state by hypoxia. The TME consists of various fusogenic cell populations, such as tumor cells, MSC, macrophages, stem cells, and fibroblasts [119][120][121][122]. All of them respond to hypoxic conditions with altered gene expression profiles and phenotypes [119][120][121][122]. Indeed, several studies suggest an impact of hypoxia on cell-cell fusion of various cell types. These include myoblasts and MSC [123], urine-derived stem cells [124], human hematopoietic progenitors and murine cardiomyocytes [65], ovarian cancer cells [125], breast cancer cells [60], and OSCCs and endothelial cells [126]. In contrast, previous in vitro studies demonstrated that both cell-cell fusion and differentiation of cytotrophoblasts and BeWo cells were impaired by hypoxia due to altered syncytin-1 expression and function [127,128]. This might be an explanation for a reduced syncytia formation in placentas of certain pathological pregnancies [127], indicating a placenta-specific hypoxia-reduced fusion frequency.
Nevertheless, all studies reporting an association between hypoxia and cell-cell fusion provide only very limited explanation for a potential molecular link on how hypoxia might be connected with cell fusion. In this context, fusogens such as syncytin-1 were not investigated [60,65,[124][125][126]. Solely the study by Archaka et al. showed that hypoxia increased expression of the transcription factors myoD and myoG [123], which induced expression of fusogens myomaker and myomerger [64]. In association with cytokines, hypoxia and IL-6 or TNF-α increased the fusion of human peripheral blood CD34 + cells and murine cardiomyocytes through up-regulation of α 4 β 1 -integrin (VLA-4) and VCAM-1 [65]. Inhibition of this interaction using blocking antibodies markedly impaired cell-cell fusion [65]. Hu and colleagues demonstrated that fusion of hypoxia preconditioned human urine-derived stem cells with murine hepatocytes in a chronic injured mouse model was dependent on up-regulation of CXCR4 signaling [124]. Our own data revealed that hypoxia and TNF-α-mediated fusion of human breast cancer cells and human breast epithelial cells were attributed to induction of MMP9 expression in human breast epithelial cells [58,60]. These findings substantiate the importance of chemokine receptor signaling, adhesion molecules, and tight cell-cell contact as prerequisites for cell-cell fusion. Nonetheless, it remains unclear which proteins ultimately facilitated plasma membrane merger in these studies. Huang and colleagues demonstrated that hypoxia promoted spontaneous cell-cell fusion between OSCCs and human immortalized oral epithelial cells partly via epithelialto-mesenchymal transition (EMT) in epithelial cells [126]. Treatment of epithelial cells with the EMT blocker DAPT significantly reduced the fusion rate of the cells [126]. Induction of EMT and DAPT-blocked EMT was validated by Western blot of known EMT markers, including N-Cadherin, E-Cadherin, Vimentin, and Slug [126]. Unfortunately, no further protein expression studies have been performed [126], so the impact of EMT on the fusion of these cells remains ambiguous.
Although an effect of hypoxia on cell-cell fusion is evident, it remains unclear which cell type is primarily converted to a pro-fusogenic state and what exactly happens at the molecular level.

Exosomes/EVs in Cell Fusion
Exosomes represent distinct kinds of EVs contributing to cell fusion. The size, heterogeneity, and complex composition of exosomes/EVs creates challenges for use of conventional analytical tools and strategies. Thus, the approximately 20 to 200 nm vesicles display typical exosome marker proteins, including surface glycoproteins of the tetraspanin transmembrane-4 family, such as CD9, CD63, and CD81 [129]. Exosomes play an important role during intercellular communication, e.g., between cancer cells and MSC and potentially during cell-cell fusion [130]. The presence of PS, particularly in tumor-derived exosomes, represents a prerequisite for fusogenic properties [131]. Exosomes develop from multivesicular bodies of endolysosomal pathway origin followed by budding of late endosomes that subsequently fuse with the plasma membrane to release their content [132,133]. The resulting exosomes contain a double membrane and encapsulate different proteins, DNA fragments, and RNAs. That is, a large variety of RNA components are associated with exosomes, such as mRNAs, circRNAs, tRNAs, long non-coding RNAs, and regulatory microRNAs (miRNAs). The biological content of exosomes differs among the originating cell type and/or the tissue. For example, bone-marrow-derived MSC as compared to adipose tissue-derived MSC contain distinct sets of tRNAs and miRNAs [134].
Released exosomes are shuttled in vivo via biological fluids, such as blood, interstitial fluid, urine, saliva, or cerebrospinal fluid [132,135], to alter the functionality of recipient cells, including cancer cells [136][137][138][139]. Taking clinical advantage of this process, e.g., MSC-derived exosomes can be used to cure diseases [140] or can be loaded with therapeutics to target tumors [141,142]. Uptake of exosomes and EVs by appropriate target cells was suggested by a three-step mechanism: (i) targeting of the acceptor cell by EVs; (ii) entry point and internalization of EVs into the acceptor cells; (iii) delivery of EV content to the acceptor cell [143]. These exosome internalization steps involve clathrin-dependent and -independent pathways, macropinocytosis, phagocytosis, lipidraft-mediated internalization, receptor-mediated endocytosis, and EV fusion with the target cell plasma membrane [136,144].
Although direct proof remains to be elucidated, previous work suggested that exosomes/EVs could directly mediate cell-cell fusion, e.g., by acting as a linker to bridge two individual cells. Previous work has demonstrated that exosomes released by trophoblasts contain fusogens, such as syncytin-1, in their membrane. This seems to be important for exosome × target cell membrane merger and delivery of extracellular vesicle cargo into target cells [145]. However, it remains to be elucidated whether syncytin-1-positive exosomes might be capable to bridge two cells and thus may facilitate cell-cell merger [145,146].
A summary of Sections 4-6 is provided in Figure 3.
contain fusogens, such as syncytin-1, in their membrane. This seems to be important for exosome × target cell membrane merger and delivery of extracellular vesicle cargo into target cells [145]. However, it remains to be elucidated whether syncytin-1-positive exosomes might be capable to bridge two cells and thus may facilitate cell-cell merger [145,146]. A summary of Sections 4-6 is provided in Figure 3. Fusion of cancer cells suggests that this process is controlled by intrinsic and extracellular factors. In the present review, we have focused on those extracellular conditions that act from outside the cells to induce/mediate cell-cell fusion, such as inflammation/inflammatory cytokines, viruses, pH, ion concentration, hypoxia, and EVs/exosomes. Pro- Fusion of cancer cells suggests that this process is controlled by intrinsic and extracellular factors. In the present review, we have focused on those extracellular conditions that act from outside the cells to induce/mediate cell-cell fusion, such as inflammation/inflammatory cytokines, viruses, pH, ion concentration, hypoxia, and EVs/exosomes. Pro-fusogenic effects have been reported predominantly by inflammation/inflammatory cytokines and hypoxia [46,[52][53][54][55]60,65,[123][124][125][126]. However, so far, it is completely unclear what this pro-fusogenic effect looks like at the molecular and cellular level. Do hypoxia and inflammation/inflammatory cytokines induce pro-fusogenic proteins/factors in the target cells or increased translocation of fusogens into the cell membrane, or both? This needs to be clarified in further studies. The relationship between inflammation and hypoxia and cell-cell fusion is also interesting for this reason since the TME is both hypoxic and chronically inflamed [119][120][121][122][147][148][149].
To what extent viruses, a low pH, or exosomes have an influence on the fusogenecity of cells or even contribute to fusion of two (or more) (cancer) cells also remains unclear. It appears conceivable that viruses and exosomes can induce fusion of two or more cells as bridging elements, but it remains to be verified in vivo. This also applies to possible reactivation of HERV env elements with fusogenic properties by extracellular factors and their potential contribution to cancer cell-cell fusion.

Conclusions
The process of cell-cell fusion in the tumor context remains poorly understood. It is now scientific consensus that cell-cell fusion events occur in tumors, even in human cancers [150][151][152][153][154][155][156][157][158][159]. Nonetheless, the questions remain not only by which proteins/factors/conditions this process is mediated but also how it is regulated in general. Likewise, it remains unclear why cancer cells and other cells can fuse with each other and whether this is a random or directed process. Cancer cell fusion increases tumor plasticity via alternative developments, including enhanced metastatic spreading [17], reduced tumorigenicity [22], or tumor dormancy [28]. These diverse outcomes suggest different rather than common pathways or conserved mechanisms that are responsible for both initiation and progression of cancer cell fusion with subsequent PHSP. The previously summarized aneuploidy of individual cell lines from a variety of different tumor entities substantiates potential cancer cell fusion and corresponding tumor heterogeneity [61]. Moreover, underlying inflammatory processes within the TME may similarly differ broadly across different organs and tumor sites.
Although the process of cell-cell fusion has been further elucidated in recent years, much remains unclear. This is primarily true for cancer cell fusion in tumor tissues controlled by intrinsic and extrinsic factors/mechanisms. In this context, extrinsic factors/mechanisms represent potential targets/target structures that could be exploited therapeutically. For example, pro-inflammatory cytokines, such as the presence of TNFalpha, could be modulated to decrease the cell-cell fusion rate in the TME. Similarly, specific protease inhibitors could be used to alter proteolytic degradation of cell-cell contacts and thereby reduce cell-cell fusion. Moreover, exosomes and their potential role in cell-cell fusion also require further investigation. It is known that MSC-derived exosomes have anti-inflammatory properties and confer immunomodulatory effects [140,160]. The TME resembles chronically inflamed tissue and thus represents a potentially profusogenic environment, which could favor fusion of cancer cells with other cells. Due to their anti-inflammatory and immunomodulatory properties, exosomes could interfere with this process.
Together, cell-cell fusion of cancer cells can be promoted by extracellular stimuli, whereas these extracellular factors may not directly mediate the merger of two (or more) cancer cells. Instead, extracellular events support appropriate intrinsic factors that subsequently mediate cell-cell fusion. Orchestrating intrinsic components confer signals for a pro-fusogenic state according to our fusion models (Figures 1-3). The involved factors together with structural and metabolic changes include, among others, G-actin to F-actin cytoskeletal reorganization, phosphatidylserine accumulation in the plasma membrane, and expression of fusogens (e.g., syncytin and corresponding receptors) to promote a profusogenic state. Consequently, extracellular conditions provide an important prerequisite to relay outside-in signaling for further supporting and promoting the complex process of (cancer) cell fusion.