Ready or Not, Here We Go

Abstract

their expected metastatic organs. This suggested an association between exosomal distribution and organ specificity in which exosomal signatures could be used to predict organotropism. Interestingly , exosomes were also found to have the capacity to "educate" cancer cells to a new destination. For example, injecting lung-metastatic exosomes could redirect bone metastatic cells to spread to the lung. Further investigation into exosomal cargo as a potential source of organotropism identified certain integrins (ITGs) that could steer the metastasis of tumor cells. Mass spectrometry analysis showed an association between integrins ITGa 6 b 4 and ITGa 6 b 1 and lung-tropic exo-somes. ITGb 5 and its associated integrin ITGa v were highly expressed in liver-tropic exosomes. Importantly, brain-tropic exosomes contained abundant ITGb 3 integrins. Also worth noting was the identification of ITGa 2 b 1 , an integrin found in all analyzed metastatic cell lines but absent in nonmetastatic cell lines. Not only were exosomes specific for a metastatic organ, but the cell type at the site responsible for exosome uptake was also organ specific. Returning to their labeled isolated exosomes, the investigators found that CD31 1 brain endothelial cells may be responsible for the initial uptake of exosomes into the metastatic organ, potentially as a result of the extracellular matrix-rich associations (Figure). Next, they showed that specific integrins not only were expressed but also were required for organ-specific metastasis. Using shRNA to observe ITGb 4 knockdown in lung-metastasis cells, they found a decrease in exosomes localized to the lung relative to the control exosomes; conversely, when ITGb 4 was overexpressed, it increased exosome uptake in the lung. Exosomes isolated from knockdown cell lines also reduced the metastatic potential of the tumor cells. Integrin-activated downstream analysis revealed that genes associated with cell migration and inflammation, namely genes belonging to the S100 family, were highly unregulated, adding to previous reports that exosomes could increase proliferation, met-astatic potential, and migration. 3-5 Proof of concept for clinical translatability was suggested in the investigation of exosomal integrins in plasma from lung-or liver-metastasis patients by enzyme-linked immunosorbent assay. Expectedly, exosomes in patients who developed lung or liver metastasis express high ITGb 4 or ITGa v , respectively. Dr Paget's "seed and soil" hypothesis argues that a supportive microenvironment and a well-adapted tumor cell will create successful metastasis; therefore, the interactions between the resident cells specific to exosome fusion and the integrin-mediated exosomes could help cells establish successful metastasis. Hoshino et al investigated exosomal organotropism from 2 fronts: the tumor cell releasing the exosomes and the microenviron-ment of the metastatic site receiving the exosomes. The identification of integrins that may be responsible for guiding cells to a metastatic site, ITGb 3 in the case of brain metastasis, provides a tumor cell-based target. The identification of the cell type responsible for the uptake of exosomes such as CD31 1 brain endothelial cells indicates which cells to target in the microenvironment. The authors not only present a compelling biological narrative for how metastasis is driven and directed but also provide molecular targets in the neoplastic cell and metastatic sites. A strategy to limit the initial breach of circulating tumor cells for cancers with clinical propensity for brain metastasis could offer an upfront combinatorial approach that addresses the fundamental step in the formation of brain metastases: vascular leakiness in the intact BBB. REFERENCES 1. Fong MY, Zhou W, Liu L, et al. Breast-cancer-secreted miR-122 reprograms glucose metabolism in premetastatic niche to promote metastasis. Nat Cell Biol. 2015;17(2):183-194. 2. Hoshino A, Costa-Silva B, Shen TL, et al. Tumour exosome integrins determine organotropic metastasis. Nature. 2015;527(7578):329-335. 3. Zaharie F, Muresan MS, Petrushev B, et al. Exoso-me-carried microRNA-375 inhibits cell progression and dissemination via Bcl-2 blocking in colon cancer. J Gastrointestinal Liver Dis. 2015;24(4):435-443. 4. Gorczynski RM, Erin N, Zhu F. Serum-derived exosomes from mice with highly metastatic breast cancer transfer increased metastatic capacity to a poorly metastatic tumor. The regulation of cancer cell migration by lung cancer cell-derived exosomes through TGF-beta and IL-10. F or the last few decades, computational neuroscientists have devoted substantial resources to improving the performance of artificial intelligence (AI) on classic games, including chess, checkers, backgammon, and Scrabble. Expert-level play by AI has been achieved largely via algorithms that test all possible combinations of moves and outcomes and choose the move combination that optimizes the score. This brute force strategy is computationally expensive , however, and whereas it may be feasible for games with relatively limited numbers of moves, it is difficult to extend to increasingly complex games. In the ancient Chinese game Go, the number of possible board configurations increases rapidly as the game progresses, making it exceedingly more complicated than chess, in which the number of possible board configurations decreases with time. Thus, although the IBM supercomputer Deep Blue defeated reigning chess world champion Garry Kasparov 10 years ago, human Go experts have consistently outperformed AI-until now. In a recently published article in Nature, Silver et al 1 described a novel decision-making algorithm that allowed AI to learn the game of Go and defeat the human European Go champion. This computing achievement, one of the "grand challenges" of AI, 2-4 had not been projected to occur for another several years. Tackling this challenge was a fitting endeavor for the team from Google DeepMind, given the vastly larger parameter space of Go compared with chess, a difference measurable in factors of a googol (10 100). The authors' approach uses deep neural networks (DNNs), computational techniques originally developed for processing complex visual information such as image classification or facial recognition. DNNs consist of layers of "neurons" (computational units) tiled on top of each other. Increasingly deep layers contain increasingly abstract representations of the actual image. In this rendition, the researchers passed the Go board configuration (ie, the pattern of tokens on the board) into the DNN as a 19 · 19 image with several layers. Each point in the 19 · 19 image represented a location on the game board, and each of the 48 layers of the image contained a different sphere of information relevant to game play (eg, which player occupied a location, how many turns had passed since a location had been occupied, and whether placing a piece at a location was legal by game rules). The DNN was then trained by a combination of machine learning strategies (Figure, A). These strategies allow an AI to "learn" a task by making future predictions based on previously observed patterns without being explicitly instructed how to respond. This step is crucial to imbue the AI with flexibility and efficiency. The first training step consisted of supervised learning, in which the DNN was instructed by moves chosen by an expert. The authors exposed the algorithm to 30 million positions, from which it could start building a decision policy. This allowed the AI to create a policy network that could somewhat SCIENCE TIMES NEUROSURGERY