The Role of the Microenvironment in Tumor Initiation, Progression, and Metastasis

Half of the secret of the cell is outside the cell: a historical perspective on the role of the microenvironment

In 1889, Stephen Paget published results of an autopsy study he conducted on 735 breast cancer patients. His study revealed that these patients tended to have metastases within four tissues: lung, liver, uterus, and bone. Empowered by these observations, as well as those of peers such as Langenbeck and Fuchs, Paget formulated his now-famous “seed and soil” hypothesis: “Every single cancer cell must be regarded as an organism, alive and capable of development. When a plant goes to seed, its seeds are carried in all directions; but they can only live and grow if they fall on congenial soil.” Perhaps this is where an appreciation for the microenvironment began. Remarkably, the enduring interpretation of Paget’s hypothesis is that certain soils are favorable for tumor growth. This has indeed dominated the landscape of metastasis research and is the subject of the following subsection (Promoting Microenvironments). Inherent in Paget’s message, however, was that certain soils are inhospitable to a tumor’s seed. In the same article, Paget remarks on a colleague’s interpretation that instead of a predisposition to receive a seed, certain organs exhibit “diminished resistance.” Thus the second subsection (Suppressive Microenvironments) deals with this idea.

Promoting Microenvironments

What makes a given microenvironment favorable for the growth of a tumor cell is a topic that is germane to all tumors, disseminated or not. For instance, the probability of harboring an occult (i.e., hidden) neoplasm increases with age to nearly 100% in an organ like the thyroid gland, yet only 0.1% of the population is ever diagnosed with thyroid cancer. What allows the emergence of tumors in some, but not others, is a question that has been pursued since the turn of the 20th century. Peyton Rous and others concluded that a transplanted tumor would not take unless there was a stromal reaction and immediate vascularization of the implant. These properties are common to both inflammation and wounding, and each has long been suspected of creating a tumor-promoting microenvironment. As early as 1863, Virchow noted that chronic irritation and prior injury could precondition tissue for tumor formation. Rous was among the first to show this conclusively by demonstrating that wounding the peritoneal cavities of mice inoculated with tumor cells accelerated the growth of tumors within their visceral organs.

Further evidence for the tumor-promoting power of the wounding microenvironment came from the vast literature on chemical carcinogens. An extensive body of work established that chemicals within coal tar such as benzo[ a ]pyrene derivatives, despite being known mutagens, were not sufficient to cause skin cancer on their own. Despite “initiation” due to chemical exposure, normal skin guards against progression unless the carcinogen dose is so excessive that it damages the tissue in addition to causing mutagenesis. This second step, called promotion, is required and is generally caused by wounding or by other toxic agents, many of which are associated with aberrant tissue repair and fibrosis. The discovery of the first “onco”-virus, referred to as Rous sarcoma virus (RSV), later provided some of the most conclusive evidence that wounding promotes tumor formation. In discovering RSV, Rous took the filtrate of a chicken tumor and noted that this cell-free filtrate would induce sarcomas in recipient chickens, thus proving Koch’s postulate. Decades later, when experimenting with RSV, Bissell and colleagues noted that the injected virus circulated throughout the bird, but tumors tended to arise only at the injection site. Was the wound created by the injection needle the key factor? Nicking the contralateral wing of infected chickens caused tumors to arise also at the sites of abrasions. This phenomenon was mediated by transforming growth factor (TGF)-β1, which was expressed at the site of injection shortly after wounding and shown to induce tumors on its own even in the absence of wounding.

Of course, processes such as wounding and fibrosis are inextricably linked with the formation of new vasculature (e.g., through angiogenesis), but it was not until Judah Folkman’s work in the early 1970s that a causal relationship between tumor growth and angiogenesis was established. Tumor fragments or tumor cells grafted onto the rabbit cornea were observed to induce sprouting from existing vasculature as they grew. Physically preventing microvasculature from reaching the implant resulted in a latent mass where tumor cell proliferation was countered by apoptosis. Folkman’s work demonstrated, for the first time, a nontumor cell—the endothelial cell—that was critical to the growth of a tumor. His work also started a new field focused on “anti-angiogenesis” based on Folkman’s hypothesis: “Solid tumors can grow to visibility only if they can vascularize themselves. Therefore, the mechanism by which tumor implants stimulate neovascularization must be well understood before therapy based upon interference with angiogenesis can be devised.” The angiogenesis inhibitor bevacizumab (Avastin) would become the first therapy explicitly targeting the microenvironment approved by the United States FDA (2004) (see Chapter 17 for more information).

Suppressive Microenvironments

The studies just described established key roles for the microenvironment in promoting tumor growth, which is the primary focus of this chapter. However, it is worth mentioning that much of the milestone research in demonstrating the importance of the microenvironment did so by showing that context could override tumorigenicity—that is, tumor cells could be tricked into thinking they are normal if provided the right cues. The observation that the embryo comprises such a suppressive microenvironment is one that was first made more than 100 years ago, when Askanazy showed that ovarian teratomas could form “normal” tissues composed of the correct embryonic germinal layers when injected into embryos. Decades later, a series of studies from different laboratories provided further evidence that the embryonic microenvironment could induce tumors to function normally in development. Mintz and Illmensee reported to have injected embryonic teratoma core cells from mice with a steel coat genotype into blastocysts from C57-b/b mice (which have black coats). The blastocysts gave rise to functionally “normal” offspring. Their next paper reported that the mice born from the initial experiments produced an offspring that was completely normal and had a mosaic (i.e., striped) coat, implying that the teratocarcinoma could pass through the germline. Although it is true that this work has not been confirmed in other laboratories, there are some dramatic variations on the theme. For instance, Brinster injected two to four teratocarcinoma cells from agouti brown mice into 4-day-old blastocysts of Swiss albino mice. One of the 60 injected mice retained the teratocarcinoma cells (based on the presence of brown hair patches on the white mouse) and proceeded to develop normally as well. Pierce later essentially quantified the balance of power between the embryonic microenvironment and the malignant cell by demonstrating that the embryonic microenvironment could suppress the malignant phenotype of one to a few implanted tumor cells, but that this ability diminished as the number of injected tumor cells increased. Perhaps this offers a hint that our bodies are able to successfully suppress only so many initiated cells, and that this power diminishes with age.

The suppressive effect of the embryonic microenvironment was demonstrated in species other than mice as well. Using RSV, Dolberg and Bissell showed conclusively that cells within injected chick embryos expressed the virus, but early embryos failed to form tumors. Maintaining embryonic architecture was key, however, as dissociating the embryos and placing the PP60 ^src -marked cells (using LacZ ) in culture resulted in rapid transformation of the blue-labeled cells. Hendrix and colleagues more recently reported similar findings for aggressive melanoma cells injected into zebrafish embryos. The lasting impact of these studies is that tissue architecture is dominant even to a powerful oncogene in embryos. These studies also offered the clue that the malignant genotype could be overridden if somehow provided with suppressive cues from the microenvironment.

Taking advantage of this insight required an assay that would allow normal and malignant cells to recapitulate their in vivo phenotypes in culture. This was achieved by Petersen, Bissell, and colleagues culturing cells in a three-dimensional (3D) reconstituted basement membrane (BM) gel. In 3D but not 2D conditions, primary mammary epithelial cells or nonmalignant cell lines formed growth-arrested, polarized acini resembling terminal ductal lobular units of the breast, whereas malignant cells formed disorganized masses that continued to grow. By examining the expression profiles of integrins—heterodimeric receptors on the cellular surface that transduce signals from the extracellular matrix (ECM) through traditional and nontraditional pathways to alter gene expression—Weaver, Bissell, and colleagues noted aberrant overexpression of integrins and a number of other receptors such as EGFR on malignant cells. Suspecting that these receptors were key nodes that integrated signals from the microenvironment to direct cell behavior, the authors began to restore levels of aberrant receptors to normal levels, starting by applying inhibitory antibodies targeting integrin β1 in malignant cells cultured in 3D gels. The result was a dramatic “phenotypic reversion” of malignant breast epithelial cells to structures that looked and behaved like their normal counterparts. To show that this treatment was not somehow selecting for a nonmalignant subpopulation of cells, Weaver and colleagues dissociated tumor cell clusters from 3D gels, replated them onto plastic, and then passaged them back into 3D gels in the absence of blocking antibody. Tumor cells once again formed malignant clusters ( Figure 16-2 ). This strategy led to the discovery of a host of signaling molecules that act in concert to regulate/integrate epithelial phenotype. Many of these molecules effect interactions between cells and their microenvironments, including ECM molecules, growth/ECM receptors, and matrix metalloproteinases (MMPs) that digest ECM components. Remarkably, targeting only one of these aberrantly expressed molecules restores the levels of all the others back to normal, demonstrating the potential of normalizing aberrant microenvironmental signaling to redirect entire signaling webs and impair manifestation of the malignant genotype. It is critical to re-emphasize that these pathways integrate only in 3D; even something as seemingly well-understood as glucose metabolism ties into these non-canonical pathways and regulates malignancy only when interrogated in context.

The Tumor Organ

The historical studies detailed above demonstrate that a tumor is a product of aberrant genomes interacting with an enabling microenvironment. This concept is easier to appreciate if one considers the tumor as a dysfunctional organ. On a basic level, an organ has the following properties:

• 1.

  Organs are multicellular and are composed of multiple tissue layers. Functional tissue layers consist of epithelia, which are tubelike structures that carry fluid, and epithelia are separated from surrounding stroma by a specialized ECM called the basement membrane (BM).

• 2.

  Organs are governed by properties that emerge as a result of the interactions between the cells, ECM molecules, and soluble factors composing the organ, and the sum of these interactions is greater than any one of the individual parts.

• 3.

  Organs have unique signatures of functional differentiation; for instance, the mammary gland produces milk, the pancreas produces insulin, and bone marrow is responsible for maintaining homeostasis of the hematopoietic and lymphatic systems.

Although tumors lack proper function, they do have the first two of these traits in common with organs. The focus of this section is on the second of these traits: the properties that emerge as a result of a tumor’s interactions with its microenvironment. There are two ways to illustrate this concept. The first would be to consider the parallels between a developing organ and a developing tumor. The scope of this discussion would extend well beyond this chapter, however, because just as each organ develops differently, tumors of these organs develop distinctly as well. Instead, a more general way to illustrate the concept of the tumor organ is to consider what happens when an organ is injured—it attempts to heal. The wounding microenvironment shares a great deal in common with the tumor microenvironment (illustrated in Figure 16-3 ), the difference being that wounds eventually stop healing, whereas a tumor’s microenvironment persists. This is why tumors have been called “wounds that do not heal.”