A time and place for inhibiting autophagy

Abstract

Autophagy is an attractive therapeutic target in cancer. Successful autophagy-focused clinical intervention will require a detailed understanding of when and where autophagy is important during tumorigenesis. In this issue of Cancer Research, Khayati and colleagues use state-of-the-art genetically engineered mouse models to demonstrate that transient systemic inhibition of autophagy can irreversibly impair the growth of established lung tumors with a good tolerability in normal tissues, suggesting a therapeutic strategy for cancer treatment.

Autophagy refers to a cellular process that degrades cellular contents (such as organelles and proteins) and recycles their building blocks for metabolic adaptation and homeostasis maintenance (1). Genetic ablation of essential autophagy genes (ATGs) has been shown to impair tumor growth in diverse tumor models, suggesting that inhibiting autophagy can be explored as a therapeutic strategy in cancer treatment (2,3). Many of these preclinical studies involve genetically engineered mouse models (GEMMs) that use the Cre-loxP conditional knockout system to delete an ATG with concurrent activation of an oncogene and/or deletion of a tumor suppressor to induce tumor development in a given mouse tissue (4-6). For example, in a previous study, Guo, White, and colleagues developed Kras^LSL-G12D/+;Trp53^L/L (KP);Atg7^L/L GEMMs and showed that conditional ablation of the autophagy essential gene Atg7 suppressed oncogenic Kras-induced lung tumor development (4). However, from a clinical translation perspective, such experimental designs carry several inherent limitations. First, these preclinical models are designed to study the role of autophagy in tumor initiation (in which a target gene is deleted before tumor formation), not tumor maintenance (in which a target gene is deleted in established tumors). However, studying tumor maintenance would be more relevant to clinical treatment that occurs after cancer has already developed. Second, a presumed autophagy inhibitor given to a patient would lead to systemic autophagy inhibition (i.e., inhibit autophagy in both normal tissues and tumors). However, in the conditional knockout system, autophagy is only impaired in one organ/tissue. Since autophagy is also important for normal tissue development (1), it remains a concern that systemic autophagy inhibition could cause intolerable toxicities in patients. Finally, to minimize potential side effects, drugs are often given to patients at certain periods of time. In contrast, in typical GEMM studies, a target gene is deleted permanently. To begin addressing these issues, White and colleagues subsequently generated a model in which Atg7 was deleted in both established tumors and normal tissues, and demonstrated that acute and systemic Atg7 deletion compromised the growth of established KP lung tumors (7). However, since Atg7 deletion is irreversible in this model, systemic autophagy inhibition also resulted in various defects in normal tissues, and the mice died 2-3 months after Atg7 deletion (7).

In this issue of Cancer Research, Khayati et al. further refined the model, which now allows for reversible and systemic deletion of Atg5 (another essential autophagy gene) in established tumors (8). This elegant study involves complex mouse model engineering and crosses. Specifically, a recent study generated mouse models that harbor an allele expressing doxycycline (Dox)-controlled Atg5 shRNA crossed with another allele expressing a reverse tetracycline-controlled transactivator (rtTA) (9). In this system, Atg5 shRNA expression is stimulated by rtTA only in the presence of Dox treatment, therefore affording temporal control of Atg5 expression. In addition, rtTA expression was regulated by a stopper element flanked by two LoxP sites (Loxp-Stopper-Loxp or LSL) or a ubiquitous promoter without the LSL. This design adds another layer of spatial control of Atg5 expression, such that rtTA-mediated Atg5 shRNA expression can occur either in the whole body (in rtTA;Atg5 shRNA mice) or in specific tissues (in LSL-rtTA;Atg5 shRNA mice, in which Cre can remove LSL and induce rtTA expression in a tissue-specific manner ). By crossing these mice with the KP model, the authors subsequently generated LSL-rtTA;Atg5 shRNA;KP and rtTA;Atg5 shRNA;KP GEMMs for tissue specific or systemic reversible inhibition of autophagy (Fig. 1).

Figure 1. Transient systemic, not tumor-intrinsic, Atg5 knockdown impairs the growth of established Kras//Trp53-mutant (KP) lung tumors without inducing significant toxicities in normal tissues.

Both LSL-rtTA;Atg5 shRNA;KP and rtTA;Atg5 shRNA;KP mouse models were intranasally infected with adenovirus-Cre to delete p53 and activate Kras to induce lung tumors. At this stage, Atg5 expression was not affected. Four weeks post lung tumor induction, mice were then fed with a doxycycline (Dox)-supplemented diet to induce rtTA-dependent expression of Atg5 shRNA, resulting in tumor-intrinsic Atg5 knockdown in LSL-rtTA;Atg5 shRNA;KP mice or systemic Atg5 knockdown in rtTA;Atg5 shRNA;KP mice. Notably, systemic, but not tumor-intrinsic, Atg5 knockdown impaired established KP lung tumor growth by altering tumor metabolic programs and promoting immune evasion. Long-term systemic Atg5 knockdown also resulted in toxicities in several normal tissues. In contrast, short-term systemic Atg5 knockdown (by discontinuing Dox diet administration) irreversibly inhibited KP lung tumor growth but rescued the toxicities induced in normal tissues. Figure adapted from images created in Biorender.

In both models, mice were intranasally infected with adenovirus-Cre to activate Kras and delete p53 and simultaneously induce rtTA expression in the lung; however, without Dox treatment, rtTA is incapable of stimulating Atg5 shRNA expression. Four weeks post lung tumor induction, mice were then fed a Dox-supplemented diet to induce Atg5 shRNA and suppress Atg5 expression in the whole body (in rtTA;Atg5 shRNA;KP mice) or only in lung tumors (in LSL-rtTA;Atg5 shRNA;KP mice). Surprisingly, despite strong evidence demonstrating autophagy inhibition in lung tumors, KP tumor growth was not affected in Dox-treated LSL-rtTA;Atg5 shRNA;KP mice (Fig. 1). In stark contrast, Dox treatment in rtTA;Atg5 shRNA;KP mice markedly suppressed the growth of established KP lung tumors (Fig. 1). These data indicated that systemic autophagy, but not tumor-intrinsic autophagy, is important for maintaining established KP lung tumors (8).

The authors then conducted a series of functional analyses, including immune profiling, in vivo isotope tracing and metabolic flux analyses, in these two models to pinpoint the underlying mechanisms. These experiments revealed that systemic, but not tumor-intrinsic, autophagy inhibition resulted in several metabolic defects, including impaired glucose carbon flux to major metabolic pathways and inhibition of gluconeogenesis for serine biosynthesis, as well as increased lymphocyte infiltration in established KP lung tumors (Fig. 1); importantly, co-depletion of CD4⁺ and CD8⁺ T cells at least partly abolished the effect of systemic autophagy inhibition on suppressing established KP tumors. These observations suggested that systemic autophagy supports the growth of established KP lung tumors by promoting both cancer cell metabolism and immune evasion (8).

As noted above, previous studies also demonstrated that systemic autophagy inhibition suppressed the growth of established KP tumors (7); however, long-term systemic autophagy inhibition resulted in toxicities in normal tissues and premature animal death (7). Khayati and colleagues confirmed toxicities in normal tissues in the current model with systemic autophagy inhibition. To study the effect of restoring autophagy on both tumor growth and normal tissue toxicities, at 15 weeks post-Dox treatment, they ceased Dox treatment for three weeks and confirmed the restoration of Atg5 expression and autophagy markers. Remarkably, they found that short-term systemic autophagy restoration reversed defects in normal tissues but the suppression of tumor growth was maintained. Finally, the authors demonstrated that intermittent Dox diet administration (with on-off-on cycles) limited systemic autophagy inhibition–induced toxicities in normal tissues and significantly extended animal survival (8).

Together, the findings by Khayati et al. reveal that transient systemic autophagy inhibition selectively impairs the growth of established lung tumors with minimal toxicities in normal tissues (8), perhaps because these tumors are more sensitive to transient autophagy inhibition than normal tissues. This suggests that intermittent administration of autophagy inhibitors to patients could be an effective strategy for treating cancer and minimizing on-target off-tumor toxicity. This study also raises several outstanding questions for future investigations. Although intermittent Dox administration significantly prolonged the survival of rtTA;Atg5 shRNA;KP mice, these animals still eventually succumbed to lung tumors, suggesting that autophagy inhibition is insufficient to completely abolish established KP tumors. The GEMMs established in this study will provide important preclinical models to study combination therapies of autophagy inhibition and other therapeutic strategies. Given the impressive immune phenotypes revealed in this study, it will be particularly interesting to explore how autophagy inhibition induces lymphocyte infiltration in tumors and whether combining transient systemic autophagy inhibition with immunotherapy achieves a better therapeutic effect. Finally, additional preclinical studies are required to address whether these findings can be extended to lung cancers with additional mutation profiles (such as KRAS/LKB1-mutant lung cancer) or other cancer types (such as pancreatic cancer).