Cancer exome analysis reveals a T-cell-dependent mechanism of cancer immunoediting

Cancer immunoediting, the process by which the immune system controls tumour outgrowth and shapes tumour immunogenicity, is comprised of three phases: elimination, equilibrium and escape. Although many immune components that participate in this process are known, its underlying mechanisms remain poorly defined. A central tenet of cancer immunoediting is that T-cell recognition of tumour antigens drives the immunological destruction or sculpting of a developing cancer. However, our current understanding of tumour antigens comes largely from analyses of cancers that develop in immunocompetent hosts and thus may have already been edited. Little is known about the antigens expressed in nascent tumour cells, whether they are sufficient to induce protective antitumour immune responses or whether their expression is modulated by the immune system. Here, using massively parallel sequencing, we characterize expressed mutations in highly immunogenic methylcholanthrene-induced sarcomas derived from immunodeficient Rag2(-/-) mice that phenotypically resemble nascent primary tumour cells. Using class I prediction algorithms, we identify mutant spectrin-β2 as a potential rejection antigen of the d42m1 sarcoma and validate this prediction by conventional antigen expression cloning and detection. We also demonstrate that cancer immunoediting of d42m1 occurs via a T-cell-dependent immunoselection process that promotes outgrowth of pre-existing tumour cell clones lacking highly antigenic mutant spectrin-β2 and other potential strong antigens. These results demonstrate that the strong immunogenicity of an unedited tumour can be ascribed to expression of highly antigenic mutant proteins and show that outgrowth of tumour cells that lack these strong antigens via a T-cell-dependent immunoselection process represents one mechanism of cancer immunoediting.     

Adaptive immunity maintains occult cancer in an equilibrium state

The capacity of immunity to control and shape cancer, that is, cancer immunoediting, is the result of three processes that function either independently or in sequence: elimination (cancer immunosurveillance, in which immunity functions as an extrinsic tumour suppressor in naive hosts); equilibrium (expansion of transformed cells is held in check by immunity); and escape (tumour cell variants with dampened immunogenicity or the capacity to attenuate immune responses grow into clinically apparent cancers). Extensive experimental support now exists for the elimination and escape processes because immunodeficient mice develop more carcinogen-induced and spontaneous cancers than wild-type mice, and tumour cells from immunodeficient mice are more immunogenic than those from immunocompetent mice. In contrast, the equilibrium process was inferred largely from clinical observations, including reports of transplantation of undetected (occult) cancer from organ donor into immunosuppressed recipients. Herein we use a mouse model of primary chemical carcinogenesis and demonstrate that equilibrium occurs, is mechanistically distinguishable from elimination and escape, and that neoplastic cells in equilibrium are transformed but proliferate poorly in vivo. We also show that tumour cells in equilibrium are unedited but become edited when they spontaneously escape immune control and grow into clinically apparent tumours. These results reveal that, in addition to destroying tumour cells and sculpting tumour immunogenicity, the immune system of a naive mouse can also restrain cancer growth for extended time periods.     

Sporadic immunogenic tumours avoid destruction by inducing T-cell tolerance

The recognition and elimination of tumours by T cells, a process termed cancer immunosurveillance, is effective against certain virus-associated cancers. Spontaneous tumours often induce a specific immune response and are therefore also immunogenic. However, it is not clear whether they can be controlled by T cells. The immunosurveillance hypothesis postulates that tumours, if they eventually grow, escaped T-cell recognition by losing immunogenicity. Here we show, by generating a mouse model of sporadic cancer based on rare spontaneous activation of a dormant oncogene, that immunogenic tumours do not escape their recognition but induce tolerance. In this model, tumours derive from single cells and express a tumour-specific transplantation rejection antigen. Whereas vaccinated mice remain tumour-free throughout their lifetime, naive mice always develop a progressively growing tumour. We also show that despite specific recognition by T cells, the tumours do not lose their intrinsic immunogenicity and are rejected after transplantation in T-cell-competent recipients. Furthermore, in the primary host tumour-induced tolerance is associated with the expansion of non-functional T cells. Together, our data argue against immunosurveillance of spontaneous cancer.     

IFNgamma and lymphocytes prevent primary tumour development and shape tumour immunogenicity

Lymphocytes were originally thought to form the basis of a 'cancer immunosurveillance' process that protects immunocompetent hosts against primary tumour development, but this idea was largely abandoned when no differences in primary tumour development were found between athymic nude mice and syngeneic wild-type mice. However, subsequent observations that nude mice do not completely lack functional T cells and that two components of the immune system-IFNgamma and perforin-help to prevent tumour formation in mice have led to renewed interest in a tumour-suppressor role for the immune response. Here we show that lymphocytes and IFNgamma collaborate to protect against development of carcinogen-induced sarcomas and spontaneous epithelial carcinomas and also to select for tumour cells with reduced immunogenicity. The immune response thus functions as an effective extrinsic tumour-suppressor system. However, this process also leads to the immunoselection of tumour cells that are more capable of surviving in an immunocompetent host, which explains the apparent paradox of tumour formation in immunologically intact individuals.     

IL-23 promotes tumour incidence and growth

Chronic inflammation has long been associated with increased incidence of malignancy and similarities in the regulatory mechanisms have been suggested for more than a century. Infiltration of innate immune cells, elevated activities of matrix metalloproteases and increased angiogenesis and vasculature density are a few examples of the similarities between chronic and tumour-associated inflammation. Conversely, the elimination of early malignant lesions by immune surveillance, which relies on the cytotoxic activity of tumour-infiltrating T cells or intra-epithelial lymphocytes, is thought to be rate-limiting for the risk to develop cancer. Here we show a molecular connection between the rise in tumour-associated inflammation and a lack of tumour immune surveillance. Expression of the heterodimeric cytokine interleukin (IL)-23, but not of its close relative IL-12, is increased in human tumours. Expression of these cytokines antagonistically regulates local inflammatory responses in the tumour microenvironment and infiltration of intra-epithelial lymphocytes. Whereas IL-12 promotes infiltration of cytotoxic T cells, IL-23 promotes inflammatory responses such as upregulation of the matrix metalloprotease MMP9, and increases angiogenesis but reduces CD8 T-cell infiltration. Genetic deletion or antibody-mediated elimination of IL-23 leads to increased infiltration of cytotoxic T cells into the transformed tissue, rendering a protective effect against chemically induced carcinogenesis. Finally, transplanted tumours are growth-restricted in hosts depleted for IL-23 or in IL-23-receptor-deficient mice. Although many strategies for immune therapy of cancer attempt to stimulate an immune response against solid tumours, infiltration of effector cells into the tumour tissue often appears to be a critical hurdle. We show that IL-23 is an important molecular link between tumour-promoting pro-inflammatory processes and the failure of the adaptive immune surveillance to infiltrate tumours.     

Identification and expansion of human colon-cancer-initiating cells

Colon carcinoma is the second most common cause of death from cancer. The isolation and characterization of tumorigenic colon cancer cells may help to devise novel diagnostic and therapeutic procedures. Although there is increasing evidence that a rare population of undifferentiated cells is responsible for tumour formation and maintenance, this has not been explored for colorectal cancer. Here, we show that tumorigenic cells in colon cancer are included in the high-density CD133+ population, which accounts for about 2.5% of the tumour cells. Subcutaneous injection of colon cancer CD133+ cells readily reproduced the original tumour in immunodeficient mice, whereas CD133- cells did not form tumours. Such tumours were serially transplanted for several generations, in each of which we observed progressively faster tumour growth without significant phenotypic alterations. Unlike CD133- cells, CD133+ colon cancer cells grew exponentially for more than one year in vitro as undifferentiated tumour spheres in serum-free medium, maintaining the ability to engraft and reproduce the same morphological and antigenic pattern of the original tumour. We conclude that colorectal cancer is created and propagated by a small number of undifferentiated tumorigenic CD133+ cells, which should therefore be the target of future therapies.     

Senescence and tumour clearance is triggered by p53 restoration in murine liver carcinomas

Although cancer arises from a combination of mutations in oncogenes and tumour suppressor genes, the extent to which tumour suppressor gene loss is required for maintaining established tumours is poorly understood. p53 is an important tumour suppressor that acts to restrict proliferation in response to DNA damage or deregulation of mitogenic oncogenes, by leading to the induction of various cell cycle checkpoints, apoptosis or cellular senescence. Consequently, p53 mutations increase cell proliferation and survival, and in some settings promote genomic instability and resistance to certain chemotherapies. To determine the consequences of reactivating the p53 pathway in tumours, we used RNA interference (RNAi) to conditionally regulate endogenous p53 expression in a mosaic mouse model of liver carcinoma. We show that even brief reactivation of endogenous p53 in p53-deficient tumours can produce complete tumour regressions. The primary response to p53 was not apoptosis, but instead involved the induction of a cellular senescence program that was associated with differentiation and the upregulation of inflammatory cytokines. This program, although producing only cell cycle arrest in vitro, also triggered an innate immune response that targeted the tumour cells in vivo, thereby contributing to tumour clearance. Our study indicates that p53 loss can be required for the maintenance of aggressive carcinomas, and illustrates how the cellular senescence program can act together with the innate immune system to potently limit tumour growth.     

Roles of tumour localization, second signals and cross priming in cytotoxic T-cell induction.

The vertebrate immune system has evolved to protect against infections that threaten survival before reproduction. Clinically manifest tumours mostly arise after the reproductive years and somatic mutations allow even otherwise antigenic tumours to evade the attention of the immune system. Moreover, the lack of immunological co-stimulatory molecules on solid tumours could result in T-cell tolerance; that is, the failure of T cells to respond. However, this may not generally apply. Here we report several important findings regarding the immune response to tumours, on the basis of studies of several tumour types. First, tumour-specific induction of protective cytotoxic T cells (CTLs) depends on sufficient tumour cells reaching secondary lymphatic organs early and for a long enough duration. Second, diffusely invading systemic tumours delete CTLs. Third, tumours that stay strictly outside secondary lymphatic organs, or that are within these organs but separated from T cells by barriers, are ignored by T cells but do not delete them. Fourth, co-stimulatory molecules on tumour cells do not influence CTL priming but enhance primed CTL responses in peripheral solid tumours. Last, cross priming of CTLs by tumour antigens, mediated by major histocompatibility complex (MHC) class I molecules of antigen-presenting host cells, is inefficient and not protective. These rules of T-cell induction and maintenance not only change previous views but also rationales for anti-tumour immunotherapy.     

CCL2 recruits inflammatory monocytes to facilitate breast-tumour metastasis

Macrophages, which are abundant in the tumour microenvironment, enhance malignancy. At metastatic sites, a distinct population of metastasis-associated macrophages promotes the extravasation, seeding and persistent growth of tumour cells. Here we define the origin of these macrophages by showing that Gr1-positive inflammatory monocytes are preferentially recruited to pulmonary metastases but not to primary mammary tumours in mice. This process also occurs for human inflammatory monocytes in pulmonary metastases of human breast cancer cells. The recruitment of these inflammatory monocytes, which express CCR2 (the receptor for chemokine CCL2), as well as the subsequent recruitment of metastasis-associated macrophages and their interaction with metastasizing tumour cells, is dependent on CCL2 synthesized by both the tumour and the stroma. Inhibition of CCL2-CCR2 signalling blocks the recruitment of inflammatory monocytes, inhibits metastasis in vivo and prolongs the survival of tumour-bearing mice. Depletion of tumour-cell-derived CCL2 also inhibits metastatic seeding. Inflammatory monocytes promote the extravasation of tumour cells in a process that requires monocyte-derived vascular endothelial growth factor. CCL2 expression and macrophage infiltration are correlated with poor prognosis and metastatic disease in human breast cancer. Our data provide the mechanistic link between these two clinical associations and indicate new therapeutic targets for treating metastatic breast cancer.     

Identification of human brain tumour initiating cells

The cancer stem cell (CSC) hypothesis suggests that neoplastic clones are maintained exclusively by a rare fraction of cells with stem cell properties. Although the existence of CSCs in human leukaemia is established, little evidence exists for CSCs in solid tumours, except for breast cancer. Recently, we prospectively isolated a CD133+ cell subpopulation from human brain tumours that exhibited stem cell properties in vitro. However, the true measures of CSCs are their capacity for self renewal and exact recapitulation of the original tumour. Here we report the development of a xenograft assay that identified human brain tumour initiating cells that initiate tumours in vivo. Only the CD133+ brain tumour fraction contains cells that are capable of tumour initiation in NOD-SCID (non-obese diabetic, severe combined immunodeficient) mouse brains. Injection of as few as 100 CD133+ cells produced a tumour that could be serially transplanted and was a phenocopy of the patient's original tumour, whereas injection of 10(5) CD133- cells engrafted but did not cause a tumour. Thus, the identification of brain tumour initiating cells provides insights into human brain tumour pathogenesis, giving strong support for the CSC hypothesis as the basis for many solid tumours, and establishes a previously unidentified cellular target for more effective cancer therapies.     

A new immunodeficiency disorder in humans involving NK cells

Immunodeficiency disorders have provided much information on the development and interaction of the various B and T lymphoid components in the immune system of man. As the lymphoid system becomes increasingly divided into functional subsets of cells it will be important to find immunodeficiencies affecting newly discovered cell types. Natural killer (NK) cells are a recently described but ill-defined subpopulation of lymphocytes which is thought to play an important part in surveillance against tumour development. Mice homozygous for the beige gene were found to have a selective deficiency in NK function and were more susceptible to transplantation of syngeneic tumours as predicted. We report here that patients carrying the analogous, autosomal recessive Chediak-Higashi (CH) gene have a profound defect in their ability to spontaneously lyse various tumour cells in vitro by either antibody-dependent or independent mechanisms. Since other cell-mediated cytolytic functions were relatively normal, these results suggest that the beige or Chediak-Higashi gene in both man and mouse controls NK function.     

Selective killing of cancer cells by a small molecule targeting the stress response to ROS

Malignant transformation, driven by gain-of-function mutations in oncogenes and loss-of-function mutations in tumour suppressor genes, results in cell deregulation that is frequently associated with enhanced cellular stress (for example, oxidative, replicative, metabolic and proteotoxic stress, and DNA damage). Adaptation to this stress phenotype is required for cancer cells to survive, and consequently cancer cells may become dependent upon non-oncogenes that do not ordinarily perform such a vital function in normal cells. Thus, targeting these non-oncogene dependencies in the context of a transformed genotype may result in a synthetic lethal interaction and the selective death of cancer cells. Here we used a cell-based small-molecule screening and quantitative proteomics approach that resulted in the unbiased identification of a small molecule that selectively kills cancer cells but not normal cells. Piperlongumine increases the level of reactive oxygen species (ROS) and apoptotic cell death in both cancer cells and normal cells engineered to have a cancer genotype, irrespective of p53 status, but it has little effect on either rapidly or slowly dividing primary normal cells. Significant antitumour effects are observed in piperlongumine-treated mouse xenograft tumour models, with no apparent toxicity in normal mice. Moreover, piperlongumine potently inhibits the growth of spontaneously formed malignant breast tumours and their associated metastases in mice. Our results demonstrate the ability of a small molecule to induce apoptosis selectively in cells that have a cancer genotype, by targeting a non-oncogene co-dependency acquired through the expression of the cancer genotype in response to transformation-induced oxidative stress.     

New signals from the invasive front

Approximately 90% of all cancer deaths arise from the metastatic spread of primary tumours. Of all the processes involved in carcinogenesis, local invasion and the formation of metastases are clinically the most relevant, but they are the least well understood at the molecular level. Revealing their mechanisms is one of the main challenges for exploratory and applied cancer research. Recent experimental progress has identified a number of molecular pathways and cellular mechanisms that underlie the multistage process of metastasis formation: these include tumour invasion, tumour-cell dissemination through the bloodstream or the lymphatic system, colonization of distant organs and, finally, fatal outgrowth of metastases.     

Essential role for oncogenic Ras in tumour maintenance.

Advanced malignancy in tumours represents the phenotypic endpoint of successive genetic lesions that affect the function and regulation of oncogenes and tumour-suppressor genes. The established tumour is maintained through complex and poorly understood host-tumour interactions that guide processes such as angiogenesis and immune sequestration. The many different genetic alterations that accompany tumour genesis raise questions as to whether experimental cancer-promoting mutations remain relevant during tumour maintenance. Here we show that melanoma genesis and maintenance are strictly dependent upon expression of H-RasV12G in a doxycycline-inducible H-Ras12G mouse melanoma model null for the tumour suppressor INK4a. Withdrawal of doxycycline and H-RasV12G down-regulation resulted in clinical and histological regression of primary and explanted tumours. The initial stages of regression involved marked apoptosis in the tumour cells and host-derived endothelial cells. Although the regulation of vascular endothelial growth factor (VEGF) was found to be Ras-dependent in vitro, the failure of persistent endogenous and enforced VEGF expression to sustain tumour viability indicates that the tumour-maintaining actions of activated Ras extend beyond the regulation of VEGF expression in vivo. Our results provide genetic evidence that H-RasV12G is important in both the genesis and maintenance of solid tumours.     

NF-kappaB functions as a tumour promoter in inflammation-associated cancer

The causes of sporadic human cancer are seldom recognized, but it is estimated that carcinogen exposure and chronic inflammation are two important underlying conditions for tumour development, the latter accounting for approximately 20% of human cancer. Whereas the causal relationship between carcinogen exposure and cancer has been intensely investigated, the molecular and cellular mechanisms linking chronic inflammation to tumorigenesis remain largely unresolved. We proposed that activation of the nuclear factor kappaB (NF-kappaB), a hallmark of inflammatory responses that is frequently detected in tumours, may constitute a missing link between inflammation and cancer. To test this hypothesis, we studied the Mdr2-knockout mouse strain, which spontaneously develops cholestatic hepatitis followed by hepatocellular carcinoma, a prototype of inflammation-associated cancer. We monitored hepatitis and cancer progression in Mdr2-knockout mice, and here we show that the inflammatory process triggers hepatocyte NF-kappaB through upregulation of tumour-necrosis factor-alpha (TNFalpha) in adjacent endothelial and inflammatory cells. Switching off NF-kappaB in mice from birth to seven months of age, using a hepatocyte-specific inducible IkappaB-super-repressor transgene, had no effect on the course of hepatitis, nor did it affect early phases of hepatocyte transformation. By contrast, suppressing NF-kappaB inhibition through anti-TNFalpha treatment or induction of IkappaB-super-repressor in later stages of tumour development resulted in apoptosis of transformed hepatocytes and failure to progress to hepatocellular carcinoma. Our studies thus indicate that NF-kappaB is essential for promoting inflammation-associated cancer, and is therefore a potential target for cancer prevention in chronic inflammatory diseases.     

p15Ink4b is a critical tumour suppressor in the absence of p16Ink4a

The CDKN2b-CDKN2a locus on chromosome 9p21 in human (chromosome 4 in mouse) is frequently lost in cancer. The locus encodes three cell cycle inhibitory proteins: p15INK4b encoded by CDKN2b, p16INK4a encoded by CDKN2a and p14ARF (p19Arf in mice) encoded by an alternative reading frame of CDKN2a (ref. 1). Whereas the tumour suppressor functions for p16INK4a and p14ARF have been firmly established, the role of p15INK4b remains ambiguous. However, many 9p21 deletions also remove CDKN2b, so we hypothesized a synergistic effect of the combined deficiency for p15INK4b, p14ARF and p16INK4a. Here we report that mice deficient for all three open reading frames (Cdkn2ab-/-) are more tumour-prone and develop a wider spectrum of tumours than Cdkn2a mutant mice, with a preponderance of skin tumours and soft tissue sarcomas (for example, mesothelioma) frequently composed of mixed cell types and often showing biphasic differentiation. Cdkn2ab-/- mouse embryonic fibroblasts (MEFs) are substantially more sensitive to oncogenic transformation than Cdkn2a mutant MEFs. Under conditions of stress, p15Ink4b protein levels are significantly elevated in MEFs deficient for p16Ink4a. Our data indicate that p15Ink4b can fulfil a critical backup function for p16Ink4a and provide an explanation for the frequent loss of the complete CDKN2b-CDKN2a locus in human tumours.     

Colorectal carcinomas in mice lacking the catalytic subunit of PI(3)Kgamma

Phosphoinositide-3-OH kinases (PI(3)Ks) constitute a family of evolutionarily conserved lipid kinases that regulate a vast array of fundamental cellular responses, including proliferation, transformation, differentiation and protection from apoptosis. PI(3)K-mediated activation of the cell survival kinase PKB/Akt, and negative regulation of PI(3)K signalling by the tumour suppressor PTEN (refs 3, 4) are key regulatory events in tumorigenesis. Thus, a model has arisen that PI(3)Ks promote development of cancers. Here we report that genetic inactivation of the p110gamma catalytic subunit of PI(3)Kgamma (ref. 8) leads to development of invasive colorectal adenocarcinomas in mice. In humans, p110gamma protein expression is lost in primary colorectal adenocarcinomas from patients and in colon cancer cell lines. Overexpression of wild-type or kinase-dead p110gamma in human colon cancer cells with mutations of the tumour suppressors APC and p53, or the oncogenes beta-catenin and Ki-ras, suppressed tumorigenesis. Thus, loss of p110gamma in mice leads to spontaneous, malignant epithelial tumours in the colorectum and p110gamma can block the growth of human colon cancer cells.     

Lysyl oxidase is essential for hypoxia-induced metastasis

Metastasis is a multistep process responsible for most cancer deaths, and it can be influenced by both the immediate microenvironment (cell-cell or cell-matrix interactions) and the extended tumour microenvironment (for example vascularization). Hypoxia (low oxygen) is clinically associated with metastasis and poor patient outcome, although the underlying processes remain unclear. Microarray studies have shown the expression of lysyl oxidase (LOX) to be elevated in hypoxic human tumour cells. Paradoxically, LOX expression is associated with both tumour suppression and tumour progression, and its role in tumorigenesis seems dependent on cellular location, cell type and transformation status. Here we show that LOX expression is regulated by hypoxia-inducible factor (HIF) and is associated with hypoxia in human breast and head and neck tumours. Patients with high LOX-expressing tumours have poor distant metastasis-free and overall survivals. Inhibition of LOX eliminates metastasis in mice with orthotopically grown breast cancer tumours. Mechanistically, secreted LOX is responsible for the invasive properties of hypoxic human cancer cells through focal adhesion kinase activity and cell to matrix adhesion. Furthermore, LOX may be required to create a niche permissive for metastatic growth. Our findings indicate that LOX is essential for hypoxia-induced metastasis and is a good therapeutic target for preventing and treating metastases.