Epithelial cell translocation: New insights into mechanisms of tumor initiation


  • Cheuk T. Leung

    Corresponding author
    1. Department of Cell Biology, Harvard Medical School, Boston, MA, USA
    Current affiliation:
    1. Department of Pharmacology, University of Minnesota Medical School, Minneapolis, MN, USA
    • Department of Cell Biology, Harvard Medical School, Boston, MA, USA
    Search for more papers by this author


original image

A cell translocation mechanism displaces sporadic mutant cells from normal, suppressive epithelial environment during early steps of tumor initiation. This epithelial cell translocation process exerts a selective pressure on early mutant cells to survive and grow in new microenvironment outside of their native niches.


ECM, extracellular matrix; MMP, matrix metalloproteinase.

Biology of tumor initiation

Tumor development is a clonal evolution process originating from sporadic mutant cells that arise within a normal tissue, such as an epithelium 1. Tumor initiation, the processes that contribute to the initial neoplastic outgrowth from histologically normal tissues, represents an early milestone in the cascade of tumor evolution. Survival and expansion of these founding mutant cells facilitates the accumulation of further oncogenic alterations that potentiate tumor development. Despite its importance in the tumor evolution hierarchy, very little is known about the mechanisms that underlie tumor initiation.

Much of our knowledge of tumor initiation is inferred from experimental cell transformation in culture and genetic studies of tumor formation driven by oncogene overexpression or tumor suppressor knockdown in animal models, usually after a long latency of weeks to months. While informative, these models are not amenable to address the cascade of cellular events through which sporadic mutant cells evolve within the native tissue environment. Although genetic alterations are the basic drivers of tumorigenesis, many lines of evidence suggest that the tissue environment can greatly influence cell behavior and, thus, tumor development. Classic studies in chicken embryos and young chickens have demonstrated that normal tissue environment can suppress cell transformation by the Rous sarcoma virus 2, 3. Clinical observations have also indicated that oncogenic alterations known to drive tumor progression are sometimes found in cells within histologically normal epithelial tissues 4, 5. These data suggest that the advancement from dormant mutant cells in intact tissues to neoplastic outgrowth is a critical step of tumor initiation.

The occurrence of sporadic mutant cells within normal epithelia raises new cell biological questions. Previous studies in developing drosophila imaginal discs 6–11 and mammalian epithelial monolayer cultures 12–14 have demonstrated cell-cell interactions between adjacent genetic mosaic cells. A more recent work using three-dimensional (3D) organotypic culture to model the tissue architectural context of tumor initiation has begun to reveal that complex cell-tissue interactions may contribute to the clonal selection of sporadic mutant cells within organized epithelia 15. This study suggested that a cell translocation mechanism that displaces mutant cells from suppressive epithelial environment can promote initial mutant cell outgrowth, but can also serve to eliminate premature mutant cells that fail to survive outside their native niches 15. This emerging research area focusing on the initial stages of human tumorigenesis could provide insights into new strategies for chemoprevention and early cancer detection.

Modeling tumor initiation in three-dimensional organotypic cultures

Sporadic mutant cells can arise within normal epithelial tissues throughout the lifespan of an organism. As epithelial tissues are under tight homeostatic control, the behavior of these sporadic mutant cells is therefore greatly influenced by the epithelial organization. However, the technical challenges of studying single-cell dynamics have precluded detailed mechanistic investigation in native tissues. The use of organotypic cultures to model the genetic and architectural contexts of early stage human tumorigenesis has provided a discovery platform to investigate cellular mechanisms that may govern tumor initiation.

The non-transformed human mammary epithelial cell line, MCF10A, serves as a valuable model to study human glandular epithelial tissues because of its human origin and its ability to establish growth-arrested acinar structures that is not achievable by other epithelial cyst systems, such as that derived from Madin-Darby canine kidney (MDCK) cells, under normal conditions. MCF10A cells form three-dimensional (3D) acinar structures when grown on reconstituted basement membrane (Matrigel) that provides extracellular matrix (ECM) components and physiological tissue stiffness 16, 17. These structures are composed of polarized, mitotically quiescent cells that organize into spherical structures with a hollow lumen, reminiscent of the mammary acinus 18. Moreover, nuclear and chromatin organization of cells in 3D acini are distinct from cells grown in a monolayer, and more closely resemble those found in the epithelial cells of breast tissues 19. Using lentiviral-based inducible oncogene expression vectors to infect 3D mammary acini at limited multiplicity-of-infection can yield sporadic single cells overexpressing the desired oncogenes 15. These models recapitulate the context of early stage tumorigenesis in mammary tissues where mutant cells are adjacent to otherwise normal neighboring cells in an organized epithelial structure.

Cell translocation and clonal expansion from suppressive epithelial environment

Modeling single-cell tumor initiation in organotypic cultures highlights the interplay between oncogenic signals and epithelial architecture in determining cell behavior. Overexpressing oncogenes that perturb the cell cycle machinery (HPV-E7 or CyclinD), activate the c-Myc transcriptional network, or stimulate AKT signaling pathways drives different degrees of hyperproliferation in MCF10A monolayer cultures 17. However, overexpressing the same oncogenes in single cells within organized acinar structures failed to drive cell proliferation, suggesting that cells within the organized acini respond to these oncogenic signals differently compared to monolayer cultures 15.

Interestingly, overexpressing ErbB2 in sporadic single-cells drives cell translocation from the epithelial layer into the lumen and, subsequently, promotes clonal expansion. The ErbB2-overexpressing cells are confined in the lumen by normal acinar cells, resembling the histological features of luminal filling in ductal carcinoma in situ (DCIS) of early-stage breast tumors. ErbB2-mediated cell translocation was also observed in acini derived from primary mouse epithelial cells or MCAS cells, a highly polarized human ovarian epithelial cell line 15. This cascade of translocation and expansion is of particular interest because ErbB2, a EGFR family of receptor tyrosine kinase, is amplified or overexpressed in up to 50–60% of DCIS 20. More striking is the finding that clonal expansion of ErbB2-overexpressing cells is dependent on translocation from the epithelial layer. When translocation is blocked, those mutant cells remain quiescent despite their continued overexpression of ErbB2. These findings suggest that organized epithelia exert a strong suppressive control of cells on broad oncogenic signals, and uncover an epithelial cell translocation mechanism by which mutant cells evade such suppressive control. Importantly, this unexpected link between spatial epithelial cell rearrangement and clonal cell expansion highlights the values of modeling complex cell environments to understand cell behavior in complex biological processes.

Contribution of epithelial cell translocation to clonal selection

Cell displacement has been observed in various epithelial tissues across species and in monolayer epithelial cell culture as a means to remove excessive cells or cell variants, presumably as homeostatic mechanisms 7–11, 14, 21, 22. By following cell fate in the 3D acinar models, cell translocation was found to displace mutant cells from the epithelial layer and expose them to the lumen (Fig. 1). This luminal compartment of the acini is deprived of ECM; attachment to ECM is critical for normal epithelial cell survival. Thus, by displacing cells from their native niches to a different compartment of the acini, cell translocation generates selective pressure on mutant cells to survive and proliferate in this new microenvironment (Fig. 1). While apical cell translocation from glandular epithelial tissue drives cells into the matrix-deprived lumen, translocation in other types of tissues may expose cells to distinct tissue compartments. The outcome of such spatial translocation would depend on whether the new microenvironment is suppressive or permissive for growth and survival.

Figure 1.

A model of early clonal selection driven by cell translocation in acinar structures. A: Sporadic mutant cells with new oncogenic alteration arise and can remain dormant within histological normal epithelial structures. B: Certain oncogenic alterations, such as those activating ERK or MMPs, can induce release of mutant cells from the epithelial structures and drive luminal translocation. C: Cell translocation exposes the mutant cells to the ECM-deprived lumen and drives selection for the ability to survive and grow in this new microenvironment. D: Mutant cells that are able to survive anoikis and grow may lead to clonal expansion (above) while mutant cells that cannot survive detachment from the ECM may die (below, small green apoptotic cell). Illustrations represent cross-sections of acinar structures. Green cells: mutant cells; gray cells: histologically normal cells; white space: acinar lumen; thick black lines: basement membrane; yellow space: stroma.

Mechanisms of cell translocation

The mechanisms that underlie cell translocation and displacement from epithelial layer remain incompletely understood. In the MCF10A acinar system, ErbB2 induces translocation of single cells through ERK- and matrix metalloproteinase (MMP)-mediated mechanisms. This luminal translocation is not dependent on proliferation because blocking proliferation does not block ErbB2-mediated translocation. Weakening of local integrin-matrix attachment by knocking down Talin-1 is also sufficient to induce cell translocation. Furthermore, the integrity of epithelial organization also plays a role in driving translocation, since compromising cell-cell adhesion within the epithelial layer significantly affects translocation 15. These observations are consistent with a model in which detachment of the mutant cells from the basement membrane releases them from the epithelial layer and triggers reorganization of the epithelia maintained by the surrounding normal cells.

Previous studies in MDCK monolayer culture showed that extrusion of Src- and Ras(V12)-expressing cells requires Cdc42 and ROCK activities 14, 23, and extrusion of apoptotic cells involves an actin- and myosin-dependent mechanism 24. Further studies of apoptotic cell extrusion in this monolayer model also showed that the location of the actin-myosin contractile ring can determine the direction of cell extrusion 25, 26. Whether similar force-dependent mechanisms are involved in cell translocation in the mammary acinar model is unclear. Three-dimensional acini are grown on reconstituted basement membrane materials that mimic physiological tissue stiffness while monolayer cultures are grown on rigid support. Epithelia on different substrate stiffness may utilize different mechanisms to execute cell displacement. Preventing the release of transformed cells from suppressive epithelial layers may delay tumor development; alternatively, displacing premature mutant cells could eliminate the lineage from the tissues by anoikis. Therefore, understanding the mechanisms that underlie these dynamic epithelial re-organization processes could provide important insights for early intervention of cancer development.

Cell-cell interaction and competition in tumorigenesis

At the initial stages of tumorigenesis, mutant cells are located adjacent to otherwise normal neighboring cells within organized epithelia. Studies in drosophila imaginal discs have shown that cells of different genotypes within developing epithelial tissues can trigger distinct cellular signaling at their interface, undergo cell competition, and induce apoptosis of adjacent cells (e.g. dMyc-overexpessing cells or Minute mutant cells) 6, 7, 10, 11. These studies demonstrated that survival and growth of individual cells in a tissue could be non-autonomously influenced by the properties of adjacent cells. Later studies showed that this cell-cell interaction and competition observed during metazoan development are also observed in MDCK monolayer culture. Distinct cellular signaling is induced only at the interface between RasV12-expressing cells (or Src-expression cells) and wildtype cells, suggesting that cell variants at the junction can recognize each other 14, 23. Moreover, cell competition was recently reported in MDCK monolayer culture where Lethal giant larvae (Lgl)-, Mahjong-, and Scribble-knockdown cells undergo apoptosis when located adjacent to wildtype cells 12, 13. The tissue-organizational effects of cell-cell interaction and competition between adjacent cells have not been examined in the 3D mammary acinar model. Whether these phenomena also occur in human tissue remains unknown. The contribution of cell-cell interaction and competition in human oncogenesis will be of great interest for future investigations.

Contribution of cell organization to signaling circuitry re-wiring

The majority of cells in mature adult organs is not actively proliferating. Oncogenic signals that induce cell-cycle reentry are key to driving clonal expansion. Epithelial tissues are also under tight homeostatic control to maintain tissue integrity and function. The findings that epithelial cells in organized acinar structures are resistant to various oncogenic stimuli that normally drive hyperproliferation in unorganized cells suggest that cell organization can modulate cellular responses to oncogenic signals. Cells in an organized epithelium are polarized and differentiated to perform specific functions. Therefore, signaling circuitry associated with polarization, differentiation, and homeostasis may interact with oncogenic signals to modulate or diminish the cellular responses. Previous studies have shown that organized mammary acini are resistant to Myc-induced proliferation, although the underlying mechanism remains controversial 27, 28. New findings that cell organization not only suppresses proliferation induced by Myc, but also that induced by HPV-E7, Cyclin-D, or AKT1 indicate that cell organization has a much broader effect in resisting different oncogenic signaling 15. On the other hand, one may speculate that oncogenic alterations in quiescent cells that are suppressed from proliferation may re-wire signaling circuitry in those cells. This ‘rewiring’ may sensitize these mutant, yet quiescent, cells to specific pro-proliferation or pro-survival signals. Exploiting this re-wired signaling circuitry may allow for the development of strategies to target these premalignant mutant cells.

Future prospects

Recent work using elaborate organotypic culture systems has unveiled new insights into the important role of cell-tissue interactions in modulating cell signaling and behavior during early tumorigenesis. Distinct tumor cells-of-origin in different organs are likely subjected to specific architectural context and homeostatic control of that tissue. In addition to genetic alterations, changes in matrix and stromal environment such as those associated with inflammation and aging have also been implicated in tumorigenesis. Studying cell dynamics in the complex cellular context of native tissues remains a major technical challenge. Using organotypic models to reconstitute different aspects of tissue environment should provide valuable platforms to systematically investigate the contribution of matrix components, cell types, and stromal factors to early tumorigenesis. Validating findings from organotypic systems using animal models will be another key step forward in the investigation of early tumor development. Understanding the biology that underlies tumor initiation could foster development of early cancer detection and intervention strategies.


I would like to thank Joan Brugge for her support on the work, and Jonathan Coloff, Lisa Gallegos, and Marcin Iwanicki for their comments and carefully reading of the manuscript. I would also like to thank Dr. Andrew Moore and the anonymous reviewers for their constructive comments. The original work on tumor initiation in organotypic models is supported by a grant from NCI CA080111 (Joan Brugge) and an American Cancer Society postdoctoral fellowship.