Document resource
Background The tumor microenvironment (TME) of GBM consists of not only tumor cells, but stromal cells, oligodendrocytes, and myeloid cells. Due to its highly invasiveness and frequent mutations, treatment options are limited. GBM tumors often exhibit dysregulation of tyrosine kinases, such as EGFR, VEGFR, and PDGFRa, making tyrosine kinase inhibitors (TKIs) a potential treatment option. Although preclinical studies have yielded positive outcomes in slowing tumor growth, the overall impact on patient survival is limited. Hereafter, this study uses Bayesian inference to understand the aberrant kinase activities in tumor tissues.Methods We established a co-culture of U87 cells and THP1-induced macrophages. Co-cultures were then digested and phosphopeptides were then enriched for pTyr. We used KSTAR 1 to compute the kinase activities of both co-culture and GBM tissues from CPTAC-3 project.2 We first established thresholds for individual batches and identify the phosphosites to be included in the calculation. We then visualized the activities using ‘DotPlot’ function. We used Cibersort3 to deconvolve GBM tissues. We borrowed a previously published signature matrix.4 We used S-model for batch correction, disabled quantile normalization, and used ran 100 permutations. In the Bayesian model, the prior represents the distributions of the parameters: EGFR amplification (EGFRamp) status, the proportions of tumor and myeloid cells, and a global shrinkage prior. The model was built using pymc package.Results In-vitro cultures of macrophages and U87 show heterogeneous kinase profiles. EGFR, ERBB2, and PDGFRB increased with the proportions of U87, while BTK, CSF1R, and LCK increased with the proportion of macrophages ( figure 1). In most tumor tissues, there were 10-30% of myeloid cells and 60-80% tumor cells (including stem cells). Bayesian results (figure 2) showed in tumor tissues, there were a lot more nuances compared to co-cultures. While tumor cells and macrophages in tissues can exhibit various phenotypes, some of the observations from co-cultures still held true. The infiltration of myeloid cells increased CSF1R and did not change EGFR activities, suggesting EGFR mostly relied on the presence of tumor cells. This also suggested myeloid cells were donating CSF1R-relevant substrates and promoting CSF1R activity. EGFR, ERBB2/3/4, and TYK2 were higher in EGFRamp tissues, suggesting EGFRamp may depend on these kinases for signaling.Conclusions Here, we presented a Bayesian framework to identify the influences of EGFRamp status and cell type proportions in kinase activity prediction While results merited further experimental validation, this facilitated our understanding on tumor heterogeneity in terms of kinase signaling.Acknowledgements Research reported in this publication was supported by the National Cancer Institute of the National Institutes of Health under Award Number U01CA284193. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. Interdisciplinary Training in Systems & Biomolecular Data Science Statement: ‘Research reported in this publication was supported by the National Institute of General Medical Sciences of the National Institutes of Health under Award Number T32GM145443. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. Croucher Foundation.References Crowl S, Jordan BT, Ahmed H, et al. KSTAR: an algorithm to predict patient-specific kinase activities from phosphoproteomic data. Nat Commun. 2022;13:4283. https://doi.org/10.1038/s41467-022-32017-5Multi-scale signaling and tumor evolution in high-grade gliomas. Liu, Jingxian, Agarwal, Anupriya, et al. Cancer Cell. 42:7:1217–1238.e19Newman AM, Steen CB, Liu CL, et al. Determining cell type abundance and expression from bulk tissues with digital cytometry. Nat Biotechnol. 2019;37:773–782. https://doi.org/10.1038/s41587-019-0114-2Varn FS, Johnson KC, Martinek J, Huse JT, Nasrallah MP, Wesseling P, Cooper LAD, Malta TM, Wade TE, Sabedot TS, Brat D, Gould PV, Wöehrer A, Aldape K, Ismail A, Sivajothi SK, Barthel FP, Kim H, Kocakavuk E, Ahmed N, White K, Datta I, Moon HE, Pollock S, Goldfarb C, Lee GH, Garofano L, Anderson KJ, Nehar-Belaid D, Barnholtz-Sloan JS, Bakas S, Byrne AT, D’Angelo F, Gan HK, Khasraw M, Migliozzi S, Ormond DR, Paek SH, Van Meir EG, Walenkamp AME, Watts C, Weiss T, Weller M, Palucka K, Stead LF, Poisson LM, Noushmehr H, Iavarone A, Verhaak RGW; GLASS Consortium. Glioma progression is shaped by genetic evolution and microenvironment interactions. Cell. 2022 Jun 9;185(12):2184-2199.e16. doi: 10.1016/j.cell.2022.04.038. Epub 2022 May 31. PMID: 35649412; PMCID: PMC9189056.Abstract 1104 Figure 1Kinase activity prediction in U87 and macrophage co-culturesAbstract 1104 Figure 2Bayesian model shows the effects of EGFR amplification status, myeloid proportion, and tumor cell proportions