PI3King the Environment for Growth: PI3K Activation Drives Transcriptome Changes That Support Oncogenic Growth

The PI3K pathway evolved to be one of the central components of signaling for cellular proliferation and survival. Activated downstream of a myriad of receptor tyrosine kinases (RTK), PI3K plays a critical role in modifying cellular activities and coordinating cellular metabolism to meet the demands of these processes (1, 2). While PI3K-mediated functions (e.g., the coordination of glucose uptake, storage, release, and utilization) are critical for mammalian survival and the ability to respond to changing environments, aberrant activation of this pathway can also support hyperproliferative growth. Evolutionarily, there appears to have been considerable pressure to balance activation and inhibition of PI3K signaling so that the pathway could be used to support proliferation without allowing for overgrowth, particularly as influenced by available nutrients (3, 4). Numerous mechanisms to control PI3K signal transduction have evolved to insulate cells from the potential catastrophic effects of allowing for over- or underactivation of the pathway, and, as a result, several feedback systems actively modulate PI3K signaling in order to maintain cellular homeostasis (5, 6). For example, insulin release from the pancreas is regulated by glucose availability, thus tying PI3K activation by insulin receptor to systemic glucose availability. Balancing this systemic regulation, PI3K signaling also regulates the phosphorylation and inhibition of glycogen synthase kinase 3-β in the liver (7). In this manner, PI3K activity controls the balance of storage and release of glycogen and therefore glucose availability, thus regulating the stability of its own signaling in the absence of significant environmental change (e.g., depletion of carbohydrate stores or overwhelming intake of sugar, such as consumption of a soft drink). At the cellular level, activation of PI3K leads to internalization of activated RTKs, limiting the receptor availability after acute stimulation, which again functions to maintain signaling homeostasis (8).

Components of the PI3K signaling cascade are amongst the most frequently mutated genes in cancer, with the most common alterations being activating hotspot mutations in PIK3CA (the gene encoding p110α, the catalytic subunit of PI3K) and loss-of-function mutations in PTEN (the phosphatase that catalyzes the inverse reaction of PI3K), both of which lead to increased PI3K signaling (9). The findings reported by Ladewig and colleagues in this issue of Cancer Research suggest that in cells with chronic PI3K activation, PI3K signaling also acts to modify normal RNA splicing programs (10). Integrating data from patient- and cell line–derived breast cancer models, including isogenic cell lines, the authors used transcriptomic analysis to (i) assess the impact of activating mutations in PIK3CA upon the transcriptome, (ii) gauge the impact of PI3K activation on transcription factor activity, and (iii) identify shifts in pre-mRNA processing, transcription, and use dynamics. Using pre- and post-PI3K inhibitor–treated patient samples, they confirmed that many of the PI3K-driven expression changes in transcripts related to proliferation, metabolism, and splicing were reversed in the context of PI3K inhibition. Their results demonstrate the profound impact of PIK3CA mutations upon the transcriptome and splicing in cancer (Fig. 1). These findings have important implications for the potential to develop therapeutic regimens that target both PI3K signaling and components of the splicing machinery for the treatment of PIK3CA-mutant cancer because these regimens may undercut the feedforward effects of alternative splicing observed in these tumor cells.

While the impact of PI3K activation on proliferation and metabolism are well appreciated, the impact of PI3K signaling upon splicing has garnered less attention. Based on the findings from Ladewig and colleagues, much like other downstream effects of PI3K signal transduction, the changes in splicing and the resulting shift in the spectrum of isoforms of various target proteins seem likely to be evolutionarily selected for, as the distribution of isoforms has the potential to act as a rheostat for the metabolic and proliferative effects of prolonged signal activation. In this view, cells that activate PI3K signaling transiently in response to a single activating event, e.g., in response to an acute spike in insulin, would have different target isoforms present as compared with cells that experience chronic PI3K activation and thus already have had time to process and translate alternatively spliced transcripts. Such a bifurcation in cell states would allow cells and organisms to adapt to time of plenty, preparing them to respond differently to a single activating event as compared with constant activation. This adaptive mechanism could be beneficial cyclically during the seasons, as nutrient sources may wax and wane. The differential response driven by alternative splicing may enable cells coming from a nutrient-restricted (low-signaling state) to a nutrient-rich (high-signaling state) environment to replenish their energy stores rather than proliferating, while those cells that had already been experiencing high/recurrent signaling activation may grow or replicate as the continued activation of the pathway suggests sufficient resources to support growth. Similar effects could have evolved for pathologic responses, such as in the context of wound healing where prolonged PI3K signaling in response to VEGF (and other wound healing-associated growth factors) could induce alternative splicing to help prime cells for the rapid proliferation required for wound repair, while transient signals would not prime recipient cells in such a way. Thus, by altering transcripts and preferentially selecting for specific isoforms of various proteins, e.g., PTK2/FRNK or AFMID as highlighted by Ladewig and colleagues, cells with chronic or repeated PI3K activation would demonstrate differential responses as compared with cells that had only experienced single or transient activation. As follows, prolonged PI3K signaling activation would have the potential to elicit a different and more profound response in cells that would be beneficial to tumor progression.

In evaluating the transcriptomic impact of PIK3CA mutations and the reciprocal effects of PI3K inhibition in breast cancer, the authors have presented insight into yet another mechanism by which PI3K signaling has the capacity to govern cellular responses to external cues. Their studies reveal a new layer of signal modification through which PI3K activation feeds forward to maintain cell survival and growth. The work provides novel reasoning for why PIK3CA mutations observed in cancer have such profound effects on tumor cell growth and proliferation, as the use of alternately spliced mRNAs driven by PI3K activation may prime the cells for more robust responses to continued signaling. Their studies support a model in which modulation of splicing downstream of PI3K activation creates a window poststimulation where, due to the presence of proteins translated from alternatively spliced transcripts, cells would exhibit enhanced sensitivity to further simulation, rendering these cells better situated to respond when the environment allows for rapid growth. In addition to demonstrating a new mechanism through which hotspot mutations in PIK3CA may contribute to tumor progression, their findings also suggest the possibility that cells evolved to show differential sensitivity to RTK activation under conditions of acute versus chronic PI3K signal transduction.

B.D. Hopkins reports nonfinancial support and other support from Faeth Therapeutics outside the submitted work; in addition, B.D. Hopkins has a patent for combining PI3K inhibitors with ketogenic diet licensed and with royalties paid from Faeth Therapeutics. No disclosures were reported by the other author.

The Hopkins Laboratory is supported by an R00 from the NCI CA230384 and an award from the Gray Foundation. Figure 1 was created with BioRender.com.