Regulation of PTEN translation by PI3K signaling maintains pathway homeostasis

The PI3K pathway is mutated in a substantial set of tumors which includes alterations in PIK3CA and PTEN genes. However PI3K mutations often co-occur with other primary drivers like RAS, RAF, HER2. Also PI3K pathway inhibitors have not been successful in PI3K mutant tumors, attributed primarily to toxicity related issues and adaptive resistance contributed by relief of negative feedback mechanisms (induction of RTKs). Now we show that PTEN, the major negative regulator of PI3K signaling is regulated by the pathway by mTOR-4EBP1 dependent translation thereby constituting a feedback loop. We show that inhibition of PI3K signaling pharmacologically (eg PI3K inhibitors) or physiologically (amino acid or serum starvation) reduces PTEN and contributes to the resuscitation of the pathway and reduces the efficacy of the PI3K inhibitors. Physiological (ligands) or oncogenic stimulation of the pathway induces PTEN levels thereby regulating the duration and amplitude of signaling. Oncogenic induction of the pathway by PIK3CA activating mutations or HER2 overexpression induces the pathway to different degrees and mirrored by the levels in PTEN protein. Therefore PTEN levels maybe an indicator of PI3K pathway output and also serve to limit the output of PIK3CA mutants. We find that co-alterations in PIK3CA and PTEN increases PI3K and mTORC1 signaling greatly as compared to PIK3CA mutation alone, thereby creating a rationale for selection for co-mutations. Consistent with this hypothesis we find that a subset of tumors co-select for these mutations, especially endometrial cancers in which they co-occur in 70% of tumors. The mechanism of this phenomenon was found to be PI3K control of mTOR and its regulation of the cap-dependent translation inhibitor 4E-BP1. This mechanism was further confirmed by removing the 5'UTR of PTEN, conditions in which, PTEN protein was insensitive to PI3K dependent regulation and resulted in increased duration of pathway inhibition and increased efficacy of PI3K inhibitors. We have also built a computational model of the PI3K network consisting of the FOXO-RTK and 4E-BP1-PTEN feedback loops to quantitatively confirm the role of the 4E-BP1-PTEN feedback in determining the steady state and kinetic responses of PTEN and pAKT to PI3K inhibition in multiple cell lines. Overall, our investigation has identified a major homeostatic regulation of the PI3K network that involves regulation of the second most mutated tumor suppressor genes, PTEN, by one of the most mutated oncogenes PI3K. This has major implications for cancer therapy, understanding the biology of PI3K pathway driven tumors and mechanisms that drive homeostasis in the growth factor network.

PI3K信号通路在众多肿瘤中发生突变，其中包括PIK3CA和PTEN基因的变异。然而，PI3K突变常与RAS、RAF、HER2等主要驱动因素共现。此外，PI3K通路抑制剂在PI3K突变型肿瘤中未能取得成功，这主要归因于与毒性相关的问题以及由负反馈机制（RTKs的诱导）缓解所导致的适应性耐药。现在我们展示，PTEN，作为PI3K信号的主要负调节因子，通过mTOR-4EBP1依赖的翻译受到该通路的调控，从而构成一个反馈环路。我们发现，通过药物（例如PI3K抑制剂）或生理学方式（氨基酸或血清饥饿）抑制PI3K信号可降低PTEN水平，并有助于该通路的复苏，同时降低PI3K抑制剂的功效。该通路的生理（配体）或癌基因刺激会诱导PTEN水平，从而调节信号传导的持续时间和幅度。通过PIK3CA激活突变或HER2过表达诱导的癌基因通路，其诱导程度与PTEN蛋白水平相呼应。因此，PTEN水平可能成为PI3K通路输出的指标，并有助于限制PI3CA突变体的输出。我们发现，与PIK3CA单独突变相比，PIK3CA和PTEN的共变异显著增加了PI3K和mTORC1信号，从而为共突变的选择提供了理论依据。与此假设一致，我们发现部分肿瘤共选择这些突变，尤其是在子宫内膜癌中，它们在70%的肿瘤中同时出现。这种现象的机制被发现是PI3K对mTOR的控制及其对帽子依赖性翻译抑制剂4E-BP1的调节。通过去除PTEN的5'UTR，进一步证实了这一机制，在这些条件下，PTEN蛋白对PI3K依赖性调节不敏感，导致通路抑制持续时间增加和PI3K抑制剂的功效提高。我们还构建了一个包含FOXO-RTK和4E-BP1-PTEN反馈环路的PI3K网络计算模型，以定量证实4E-BP1-PTEN反馈在确定PTEN和pAKT对PI3K抑制的稳态和动力学响应中的角色。总的来说，我们的研究确定了PI3K网络的一种主要稳态调节机制，该机制涉及第二常见突变肿瘤抑制基因PTEN的调节，这一调节由最常见突变的癌基因PI3K之一所执行。这对于癌症治疗、理解PI3K通路驱动的肿瘤的生物学以及驱动生长因子网络稳态的机制具有重大意义。