An Essential Role for the Immune System in the Mechanism of Tumor Regression Following Targeted Oncogene Inactivation

Abstract

Tumors are genetically complex and can have a multitude of mutations. Consequently, it is surprising that the suppression of a single oncogene can result in rapid and sustained tumor regression, illustrating the concept that cancers are often “oncogene addicted.” The mechanism of oncogene addiction has been presumed to be largely cell autonomous as a consequence of the restoration of normal physiological programs that induce proliferative arrest, apoptosis, differentiation, and/or cellular senescence. Interestingly, it has recently become apparent that upon oncogene inactivation, the immune response is critical in mediating the phenotypic consequences of oncogene addiction. In particular, CD4⁺ T cells have been suggested to be essential to the remodeling of the tumor microenvironment, including the shut down of host angiogenesis and the induction of cellular senescence in the tumor. However, adaptive and innate immune cells are likely involved. Thus, the effectors of the immune system are not only involved in tumor initiation, tumor progression, and immunosurveillance, but are also involved in the mechanism of tumor regression upon targeted oncogene inactivation. Hence, oncogene inactivation may be an effective therapeutic approach because it both reverses the neoplastic state within a cancer cell and reactivates the host immune response that remodels the tumor microenvironment.

Introduction: oncogene addiction

Oncogene addiction describes the dramatic and sustained regression of some cancers following the targeted inactivation of a single oncogene [1–12]. Oncogene addiction has been described as a cell autonomous phenomenon that is associated with proliferative arrest, apoptosis, senescence, and/or differentiation [13]. The mechanism of oncogene addiction is not known but has been suggested to be a consequence of the restoration of physiological programs [11,14], synthetic lethality [15], and/or differential decay of survival and apoptosis programs [16]. The mechanisms of oncogene addiction appear to be generally similar for different oncogenes and different tumor types. The most robust oncogene targets may be the essential driver mutations that are critical for the molecular scaffolding that maintains the cancer phenotype [17,18]. However, it is not clear if an oncogene must be mutated in order to establish addiction [12]. Regardless, until recently, it has been largely presumed that the mechanisms of oncogene addiction are cell autonomous. Current literature illustrates that this is unlikely to be the case [19–26].

In the clinic, oncogene addiction has been used as a therapeutic approach. Thus, the inactivation of an oncogene with a targeted small molecule or antibody therapy can be associated with significant tumor regression, although tumors often recur [27–29]. Drugs that target oncogenes such as BCR-ABL, c-Kit, HER2/Neu, EGFR, and PML-RARα have significant clinical activity against leukemia, breast cancer, and lung cancer [30–32]. More recently, vemurafenib, which targets the BRAF V600E mutation, has been shown to have activity against melanoma [33]. Drugs that target oncoproteins such as ALK, JAK2, MDM2, and PI3 kinase are currently in clinical trials [34–43]. However, the most important driver oncoproteins, such as K-RAS and MYC, have remained undruggable, although strategies to target RAS [44] and MYC [45,46] have been recently described.

Experimental mouse models have been a particularly tractable approach to interrogating the mechanism of oncogene addiction. Transgenic mouse models employing strategies that enable the conditional expression of oncogenes have been used to illustrate that cancers initiated by an oncogene, such as MYC, RAS, BCR-ABL, MET, and BRAF, are generally reversible upon oncogene suppression [1,2,4,5,8,47,48]. From these mouse models, insight has been gleaned into the mechanisms of oncogene addiction [49]. Oncogene inactivation has been shown to induce rapid proliferative arrest, apoptosis, differentiation, and/or senescence of tumors, as previously described [50]. The specific consequences depend upon the particular tumor type. Thus, oncogene inactivation in hematopoietic tumors induces proliferative arrest and apoptosis [4], whereas in sarcoma, it induces sustained differentiation [51], and in epithelial tumors it induces apoptosis in most but not all tumor cells (a small population can remain dormant) [52]. Genetic and cellular contexts influence the consequences of tumor regression [53].

However, it also has been appreciated that oncogene inactivation in a tumor elicits a profound influence not only on tumor cells but also on the tumor microenvironment. Oncogenes can regulate angiogenesis; MYC inactivation can result in the shut down of angiogenesis, contributing to tumor regression [54–57]. Inactivation of an oncogene recruits a marked change in the tumor microenvironment associated with the recruitment of immune effector cells, and this plays a direct role in the mechanism of oncogene addiction [12,58–61]. The innate and the adaptive immune systems are known to play a critical role in tumor initiation and progression [62–65] as well as immunosurveillance [66–69]. Hence, it appears that the immune system is integral to both the construction as well as the destruction of tumors [59]. Oncogene inactivation appears to be an unanticipated mechanism by which these immune-based mechanisms against cancer can be activated.

Immune system and the anti-tumor response

The immune system is a critical barrier to tumorigenesis [70]. Hosts with suppressed immune systems have a greatly increased incidence of cancer [71,72]. In particular, drugs that suppress host adaptive immunity are associated with multiple tumor types, including lymphoma and skin cancer [73–75]. Patients with a compromised immune system seem to have decreased overall and progression-free survival in several different solid and hematologic malignancies [76,77]. During tumorigenesis, cancers appear to be immune edited. For a tumor to arise, they must evolve to bypass immune-mediated mechanisms [78–80]. Hence, immune surveillance can prevent cancer formation and tumors that arise have learned to coexist or bypass these effector mechanisms.

Similarly, the immune system appears to play a role in the mechanism by which therapeutics prompt tumor regression [75,81]. Evidence that conventional therapeutics, such as certain chemotherapies, mediate their efficacy at least in part through immune activation has been defined [82,83]. Many immune-based therapies are emerging as efficacious treatments for cancer, including engineered antibodies that target tumor cells, such as rituximab [84] and trastuzumab [85], as well antibodies or drugs that manipulate immunostiumulatory or immunoinhibitory signals that can be used to induce immune effectors to recognize cancer cells [86–88], such as anti-CTLA-4, which can enable T cell activation via stimulation of CD80 and CD86 [89], anti-PD-L1, which can reverse immunosuppression [90] or anti-CD47, which enhances phagocytosis of cancer cells [91]. Thus, a robust host immune response may be useful or necessary for the complete clearance of tumor cells [86,87]. By specifically activating effectors of the immune response, one can elicit a robust therapeutic response.

Oncogene addiction and the immune response

Multiple studies provocatively suggest that targeted oncogene inactivation and oncogene addiction are mediated in concert with immune activation against tumor cells (summarized in Table 1, Fig. 1). Hence, oncogene inactivation appears to elicit tumor regression by reversing a cancer phenotype both from within tumor cells through cell autonomous mechanisms as well as outside the tumor cells through the activation of a host immune response. Together, this results in the remodeling the tumor microenvironment (Fig. 1) [92]. Thus, following oncogene inactivation, there appear to be discrete, temporally associated changes that occur in a tumor. Initially, within the first 2 days, tumor cells undergo proliferative arrest and apoptosis. Then, CD4⁺ T cells, as well as other immune effectors, are recruited to the tumor site and this response appears to be causally required to elicit cellular senescence of tumor cells and the shut down of angiogenesis in the host. Finally, additional immune effector cells may contribute to the clearance of remaining tumor cells (Fig. 1).

Table 1.

Select Oncogenes that, when targeted, recruit immune effectors.

Target Protein(s)	Tumor Type	Mechanism of Inactivation	Immune Effectors Recruited	Refs
MYC	T cell acute lymphoblastic lymphoma	Conditional Animal Models	CD4+ T cells, cytokines	[58]
BCR-ABL	Pro-B cell acute lymphocytic leukemia, CML, GIST	Conditional Animal Models, Imatinib	CD4+ T cells, CD8+ T cells, NK cells, cytokines,	[58,119]
BRAF	Melanoma	Vemurafenib	CD+ T cells, CD8+ T cells, reduction in immunosuppressive cytokines	[120,123]
PDGFR, RET, or KIT	Kidney Cancer	Sunitinib	T cell activation, reduction of regulatory T cells, Influencing MHC class I antigen presentation	[126–128]
PML	Acute promyelocytic leukemia, prostate cancer	Arsenic+All-Trans-Retinoic Acid (ATRA)	CD8+ T cells	[129]
HER2/Neu	Breast Cancer	Trastuzumab	NK Cells	[132,133]
Proteosome, NF-kB	Multiple Myeloma	Bortezemib	CD8+ T cells, Dendritic cells	[130]
EGFR	Non-small cell lung cancer	Erlotinib	TRAIL+ Dendritic cells	[131]

Fig. 1. The immune system is essential to the mechanism of oncogene addiction.

The interaction between tumor cell-intrinsic and immune-mediated mechanisms is crucial for sustained tumor regression upon oncogene inactivation. Following oncogene inactivation in a mouse model or targeted therapy in the clinic, the tumor (first panel) will initially regress via proliferative arrest, differentiation, and/or apoptosis (second panel). If the host immune system is activated, immune effectors are recruited to the tumor site, leading to senescence and the shutdown of angiogenesis (third panel). As time progresses, both an innate and an adaptive immune response result in the clearance of residual tumor cells and a potent anti-tumor memory (fourth panel).

Prior studies suggested that oncogenes might regulate the ability of a tumor to recruit immune cells. Thus, oncogene activation can modulate the expression of some critical immune regulatory receptors [92], perturb normal immune cell development [93], and modulate the expression of critical immune regulatory cytokines [94]. Thus, it seemed possible that oncogene inactivation could restore an immune response. Indeed, oncogene inactivation depends upon an immune response to induce tumor regression (Table 1) through many mechanisms (Fig. 1). In conditional transgenic mouse models of MYC or BCR-ABL tumorigenesis [95], oncogene inactivation only results in sustained tumor regression in immune intact hosts [58]. Notably, the kinetics of tumor regression, the extent of tumor regression, and the ability to maintain sustained tumor regression were all perturbed by up to 1000-fold in an immune compromised host [58] (Fig. 2).

Fig. 2. Oncogene inactivation combined with immune therapy may be more efficacious as a treatment for cancer.

In an immune-competent host, oncogene inactivation induces an immune response that leads to more rapid, complete, and sustained tumor regression (left). Conversely, oncogene inactivation in the absence of an immune system impedes the kinetics and extent of tumor elimination and is associated with tumor recurrence (center). The combination of oncogene inactivation together with immunostimulatory therapy may cooperate to induce more sustained tumor regression, in particular in circumstances where hosts are immune compromised (right).

Surprisingly, in CD4^−/− but not CD8^−/− hosts, oncogene inactivation failed to elicit sustained tumor regression. In CD4^−/− hosts, oncogene inactivation continued to result in proliferative arrest and apoptosis, but cellular senescence and shut down of angiogenesis was impeded [58]. Reconstitution of RAG1^−/− hosts with CD4⁺ T cells alone was sufficient to enable oncogene inactivation to induce sustained tumor regression. The absence of immune cells, generally, and CD4⁺ T cells, specifically, was likely to suppress tumor regression through an influence on other immune effectors and the expression of cytokines. Upon oncogene inactivation, there is a dramatic recruitment of immune cells to the tumor as measured by bioluminescence imaging of luciferase-labeled effector cells [58]. Thus, CD4⁺ T cells are causally involved in the mechanism of tumor regression upon oncogene inactivation.

Changes in cytokines are also likely important in the mechanism of tumor regression. Notably, there is an induction of anti-tumor and a suppression of pro-tumor cytokines after oncogene inactivation but only in immunocompetent, not immunocompromised, hosts [58]. Although many cytokines are likely to be involved, thrombospondin-1 (TSP-1) was found to be one critical cytokine [58]. Notably, immune cells that were TSP^−/− could not reconstitute the ability of RAG^−/− mice to elicit tumor regression upon oncogene inactivation. TSP-1 has been suggested to be a key regulator of not only angiogenesis but also senescence [96]. Moreover, CD47, the receptor of TSP, is a key regulator of the immune response [97]. Finally, TSP-1 and CD47 have been suggested to regulate cellular senescence [96,98,99]. Thus, there are likely multiple mechanisms by which TSP-1 could contribute to tumor regression.

CD4⁺ T cells are likely to contribute to tumor regression through many mechanisms. Of note, CD4⁺ T cells can express a variety of cytokines that have been implicated in the regulation of cellular senescence and/or angiogenesis [100–103]. But, CD4⁺ T cells may also be working via direct cellular interactions with the tumor cells or host stromal cells resident in the tumor microenvironment. Finally, CD4⁺ T cells may recruit other immune or host cells that contribute to the mechanism of tumor regression.

How MYC inactivation results in the activation of immune effectors, and specifically CD4⁺ T cells, is not clear. There are several non-mutually exclusive possibilities. First, MYC has been suggested to regulate the expression of molecules that may be immunosuppressive and/or regulate angiogenesis [104]. Hence, MYC inactivation could lead to the direct change in expression of cytokines by tumor cells, thereby recruiting immune cells [104]. Second, MYC inactivation could activate an immune response through immunogenic cell death [105]. Third, MYC inactivation could result in such a dramatic reduction in tumor growth that immune effectors are able to effectively compete to further suppress tumor growth. Identifying the specific mechanism of the immune activation and response could suggest important strategies for monitoring and implementing a therapeutic response [19].

Importantly, many other immune effectors are likely to contribute to the response of targeted therapies. This is potentially governed by the unique genetic and cellular context of each tumor [106,107]. In other mouse models of oncogene-induced tumors, investigators have noted that innate immune cells such as mast cells [108], macrophages [109], and other antigen-presenting cells (APCs) may function as barriers to tumor growth and facilitators of tumor regression. Thus, it would seem likely that these other innate and adaptive immune cells as well as other host cells may contribute to the mechanism of tumor regression.

Combining immune therapy with targeted therapy could cooperate to induce efficacious treatment for human cancers (Fig. 2). In an immune-competent host, oncogene-targeted therapy may result in a robust immune response that leads to a swift and thorough tumor regression (left panel). However, the lack of an immune system dampens both the kinetics and the magnitude of tumor clearance, and tumor recurrence is almost guaranteed (center panel). To prevent disease relapse, immune-targeted therapy may cooperate with oncogene-targeted therapy (right panel). There are many types of immune therapy that could be useful, including boosting NK activity [110], modulating cytokine production by T cells and cytokine signaling [111], and transferring memory T cells [112,113]. Engineering of other specific cytotoxic and T helper cells [114], NK cells [138–142], gamma delta T cells [143], and APCs such as dendritic cells [144] could also be useful. Finally, modulating the humoral immune response could be effective [145,146]. Defining the precise contributions of different effectors could be integral to designing the most efficacious strategies for utilizing the immune system to treat cancer.

Targeted oncogene inactivation and immune response in human patients

The host immune response has been shown to be essential for the optimal response to cancer therapy in human patients [115]. Select chemotherapeutics and radiotherapy mediate their anti-tumor effects in part by inducing an immune response [116,117]. Because conventional therapeutics often also impede the immune system, they may be limited in their efficacy [118]. In contrast, targeted oncogene inactivation provides a strategy to selectively target tumor cells that may not inhibit immune activation and, ideally, may cooperate with immune therapies.

Significant evidence suggests that targeted oncogene inactivation recruits an immune response (Table 1). In patients with BCR-ABL⁺ gastrointestinal stromal tumors treated with imatinib, a better prognosis has been associated with IFN-γ secretion by NK cells in the peripheral blood [119]. Similarly, CD8⁺ T cells were essential for optimal responses to imatinib, and this may be due to a reduction in the immunosuppressive enzyme indoleamine 2,3-dioxygenase (Ido) within the tumor cells. Thus, the best clinical outcomes may be obtained through combination of oncogene-targeted therapy together with immune-modulating therapy.

Oncogene inhibition appears to induce tumor regression both through direct effects on tumor cells as well as recruitment of an immune response; this is true for many oncogenes, as summarized (Table 1). The inhibition of BRAF has a direct influence on tumor growth but also appears to activate the immune system [120]. The combination of immune therapy and vemurafenib may be synergistic [121]. Vemurafenib increased intratumoral accumulation of adoptively transferred T cells [122]. Similarly, vemurafenib increased the number of CD4⁺ or CD8⁺ T cells in tumors in Stage III and Stage IV melanoma [120]. The presence of tumor-infiltrating lymphocytes after BRAF inhibition is associated with a better clinical prognosis [120]. Lastly, targeting BRAF V600E appears to diminish the expression of immunosuppressive cytokines and chemokines [123]. Ongoing investigations will determine whether optimal combinations and doses of vemurafenib in combination with immunotherapy are more clinically efficacious [124,125]. Sunitinib, which can target PDGFR, RET, and KIT, may also recruit and/or depend on the immune response for its optimal function as a tyrosine kinase inhibitor [126]. Sunitinib treatment increases IFN-γ-producing T cells [127] and decreases regulatory T cells [127,128].

Other targeted therapies may induce an immune response in addition to their tumor-specific effects (Table 1). The combination of arsenic and all-trans-retinoic acid is used against PML-RARα, and can modulate antigen presentation [129]. Bortezomib, an agent used for multiple myeloma that targets the proteasome and NF-κB, can affect numbers of CD8⁺ T cells and dendritic cells [130]. Erlotinib, a successful non-small cell lung cancer therapeutic that targets EGFR, can also induce an influx of dendritic cells [131]. Lastly, the efficacy of trastuzumab, a breast and ovarian cancer therapeutic that targets HER2/neu, may depend on a robust NK cell response [132,133]. Thus, targeted inhibition of oncogenes may be efficacious via activation of an immune response and may synergize with immunotherapy (Fig. 2).

However, the targeted inactivation of oncogenes may have undesired effects on the immune response. The MAPK/extracellular signal-regulated kinase kinase (MEK) inhibitors can induce potent T cell inhibition [134]. Imatinib has been shown experimentally to influence the immune response in a multitude of ways [34,58,135–139]. Similarly, targeted inactivation of oncogenes such as MYC and RAS could have profound effects on immune activation [34]. Consequently, it will be pivotal to consider that targeted oncogene inactivation can have positive and negative influences on the contribution of the immune system to the therapeutic response.

Implications: targeted oncogene inactivation and the immune response

Generally, targeted oncogene inactivation can elicit an immune response that may be critical to tumor regression (Table 1). In preclinical identification of therapeutic agents that target oncogenes, it is important to utilize model systems that have an intact host immune system. Thus, xenograft models may underestimate the efficacy of therapeutics. Monitoring the immune response may be useful to predict the efficacy of targeted therapeutics both during preclinical and clinical studies. This could include the evaluation of immune cellular effectors as well as specific cytokines.

The immune response dynamically evolves because of interactions between the tumor and the host [140]. The immune state of a host cannot just be measured prior to therapy, but also will have to be examined after a therapeutic response. It is likely that this response may predict the outcome. A deliberate understanding of how targeted oncogene therapies influence immune activation should enable a directed approach to combining immune and targeted therapies.

Oncogene inactivation alone appears to be able to induce tumor regression both through a direct effect on tumor cells as well as by recruiting immune effectors that can remodel the tumor microenvironment. The judicious combination of oncogene-targeted therapy with specific immunomodulatory therapy may further increase the clinical response and long-term survival of patients [124,141,142]. Pointedly, immune activation may be essential to prevent the emergence of therapy-resistant tumor cells, which can lead to tumor recurrence [143,144] (Fig. 2). Hence, the identification of the best agents to elicit oncogene addiction will require examination of the efficacy of these therapies and consideration of their ability to induce both cell autonomous and host-dependent mechanisms of tumor regression.