Awakening Immunity

See allHide authors and affiliations

Science  05 Nov 2010:
Vol. 330, Issue 6005, pp. 761-762
DOI: 10.1126/science.1198345

Cancer cells are embedded in stroma, the connective tissue framework of solid tumors. It consists of nonmalignant hematopoietic and mesenchymal cells, as well as extracellular matrix. Whether stromal cells have an essential role in cancer development and growth has been long debated. On page 827 of this issue, Kraman et al. (1) show that deleting a subpopulation of stromal fibroblasts arrests the growth of a solid tumor, an effect that depends on an immune response to the tumor. These results agree with other studies suggesting that immunizing against fibroblasts in tumors can unmask an immune response to cancer (2, 3).

Stromal cells and cancer cells depend on each other for mutual paracrine stimulation. Stromal fibroblasts are probably required for cancer cells to survive and grow (4), but why does their elimination trigger an immune response to the cancer cells? A clue may come from a particular subtype of fibroblast whose removal elicits this response. Fibroblasts from malignant solid tumors show increased expression of genes that are repressed in other tissue fibroblasts (5), particularly the genes encoding the cytoskeletal protein α-smooth muscle actin (α-SMA) and fibroblast activation protein (FAP), a serine protease. Both proteins are also expressed on pericytes (α-SMA+ and FAP+), stromal cells that reside at the interface between tumor endothelium and surrounding tissue. Stromal cells expressing these markers may suppress the immune response to tumors as a consequence of producing massive amounts of stromal cell–derived factor–1 (SDF-1/CXCL12). SDF-1 attracts regulatory T cells (CD4+ subtype) into the tumor (6). It also causes random movement of effector T cells, which interferes with T cell–tumor cell interaction and ultimately hinders tumor destruction (7).

Eliminating neutrophils also produces antitumor immune effects similar to those caused by eliminating FAP+ stromal cells (8, 9). Neutrophils release matrix metalloproteinase 9 and elastase, which enzymatically “free” stromal fibroblast progenitor cells from bone marrow and perivascular reservoirs (10), allowing them to follow the SDF-1 cytokine gradient into the tumor. Metalloprotease released from neutrophils also catalyzes the release and activation of latent transforming growth factor–β (TGF-β) from the extracellular matrix. TGF-β1 activates stromal fibroblasts (α-SMA+ and FAP+), causing them to produce immunosuppressive SDF-1. TGF-β1 also prevents the initiation of effector T cell responses. Furthermore, depending on the stimulus, neutrophils and other leukocytes can themselves produce large amounts of TGF-β1 (11).

Tumor destruction.

Removing FAP+ fibroblasts and pericytes damages the blood supply and causes some cancer cells to die. The resulting damage-associated signals, together with antigens released by dying cancer cells, triggers the production of cytokines (IFN-γ and TNF-α) by cancer antigen-specific T cells in the tumor. This results in the destruction of the remaining cancer and stromal cells by the immune system.


Kraman et al. deleted FAP+ stromal cells from mice bearing solid tumors that arose from injected lung cancer cells. These cancer cells were engineered to express ovalbumin, a xenoantigen capable of eliciting an immune response. The removal of FAP+ fibroblasts did not alter the number or subtypes of tumor-infiltrating T cells, but did result in their activation and secretion of the pro-inflammatory cytokines interferon-γ (IFN-γ) and tumor necrosis factor–α (TNF-α) (see the figure). But in an actual tumor, what type of cancer cell antigens would be suitable for effective tumor destruction by the immune system? Kraman et al. emphasize the importance of unmutated self-antigens on cancer cells as effective elicitors. Although targeting by the immune system of viral antigens such as EBNA3 can eradicate large, bulky masses of lymphomas induced by Epstein Barr virus, it is not known what the critical characteristics of an antigen are that allow effector T cells to destroy large, established solid tumors or aggregates of cancer cells that have dispersed from the primary tumor to the rest of the body. EBNA3 is not expressed on normal cells and is essential for the cancer cells to remain malignant. The closest correlate to antigens on cancers that are not associated with viruses are tumor-specific mutant proteins that are essential for maintaining malignancy.

In this context, FAP as a self-antigen may be problematic because it is expressed on nonmalignant cells and its expression can be lost (12). Nevertheless, FAP is expressed on some fraction of stromal fibroblasts in more than 90% of patients with solid tumors, and patients with higher FAP expression have a worse clinical outcome. Thus, as Kraman et al. suggest, targeting FAP-expressing stroma cells for destruction could unmask the patients' own or adoptively transferred immunity. But in which clinical settings could these results influence future immunotherapy of cancer?

Human cancers when first detected usually have an average diameter of at least 1 cm and contain about 109 cancer cells, including thousands of diverse heritable variants resistant to drugs, radiation, and immunotherapy. Metastatic cells may already be widely dispersed. Kraman et al. treated relatively small tumors in mice soon after they were inoculated with cancer cells. Eliminating FAP+ stromal fibroblasts should inhibit growth of small spontaneous tumors and thus may help eliminate clinically undetectable cancer cells that have already metastasized before excision of the primary tumor, a common cause of relapse. The caveat is that metastatic cancer cells are not necessarily in the milieu of inflammation caused by experimental cancer cell injection. Experimentally, tumors 1 cm in diameter or larger in mice are abolished by cancer-specific T cells that target not only cancer cells but also stromal cells that also present cancer cell antigens (13, 14). The elimination of cancer cell variants by the immune system is presumably due to “bystander killing” that is secondary to elimination of stroma (13, 14). Thus, immunotherapy treatment with both T cells that target cancer cells and an agent that targets FAP-expressing cells for destruction could increase the success of eliminating solid tumors and metastatic cells.


View Abstract

Stay Connected to Science

Navigate This Article