DPP inhibition alters the CXCR3 axis and enhances NK and CD8+ T cell infiltration to improve anti-PD1 efficacy in murine models of pancreatic ductal adenocarcinoma.

BACKGROUND: Pancreatic ductal adenocarcinoma (PDAC) is projected to be the second leading cause of cancer death in the USA by 2030. Immune checkpoint inhibitors fail to control most PDAC tumors because of PDAC's extensive immunosuppressive microenvironment and poor immune infiltration, a phenotype also seen in other non-inflamed (ie, 'cold') tumors. Identifying novel ways to enhance immunotherapy efficacy in PDAC is critical. Dipeptidyl peptidase (DPP) inhibition can enhance immunotherapy efficacy in other cancer types; however, the impact of DPP inhibition on PDAC tumors remains unexplored. METHODS: We examined the effects of an oral small molecule DPP inhibitor (BXCL701) on PDAC tumor growth using mT3-2D and Pan02 subcutaneous syngeneic murine models in C57BL/6 mice. We explored the effects of DPP inhibition on the tumor immune landscape using RNAseq, immunohistochemistry, cytokine evaluation and flow cytometry. We then tested if BXCL701 enhanced anti-programmed cell death protein 1 (anti-PD1) efficacy and performed immune cell depletion and rechallenged studies to explore the relevance of cytotoxic immune cells to combination treatment efficacy. RESULTS: In both murine models of PDAC, DPP inhibition enhanced NK and T cell immune infiltration and reduced tumor growth. DPP inhibition also enhanced the efficacy of anti-PD1. The efficacy of dual anti-PD1 and BXCL701 therapy was dependent on both CD8+ T cells and NK cells. Mice treated with this combination therapy developed antitumor immune memory that cleared some tumors after re-exposure. Lastly, we used The Cancer Genome Atlas (TCGA) to demonstrate that increased NK cell content, but not T cell content, in human PDAC tumors is correlated with longer overall survival. We propose that broad DPP inhibition enhances antitumor immune response via two mechanisms: (1) DPP4 inhibition increases tumor content of CXCL9/10, which recruits CXCR3+ NK and T cells, and (2) DPP8/9 inhibition activates the inflammasome, resulting in proinflammatory cytokine release and Th1 response, further enhancing the CXCL9/10-CXCR3 axis. CONCLUSIONS: These findings show that DPP inhibition with BXCL701 represents a pharmacologic strategy to increase the tumor microenvironment immune cell content to improve anti-PD1 efficacy in PDAC, suggesting BXCL701 can enhance immunotherapy efficacy in 'cold' tumor types. These findings also highlight the potential importance of NK cells along with T cells in regulating PDAC tumor growth.

Deep learning for automated classification and characterization of amorphous materials.

It is difficult to quantify structure-property relationships and to identify structural features of complex materials. The characterization of amorphous materials is especially challenging because their lack of long-range order makes it difficult to define structural metrics. In this work, we apply deep learning algorithms to accurately classify amorphous materials and characterize their structural features. Specifically, we show that convolutional neural networks and message passing neural networks can classify two-dimensional liquids and liquid-cooled glasses from molecular dynamics simulations with greater than 0.98 AUC, with no a priori assumptions about local particle relationships, even when the liquids and glasses are prepared at the same inherent structure energy. Furthermore, we demonstrate that message passing neural networks surpass convolutional neural networks in this context in both accuracy and interpretability. We extract a clear interpretation of how message passing neural networks evaluate liquid and glass structures by using a self-attention mechanism. Using this interpretation, we derive three novel structural metrics that accurately characterize glass formation. The methods presented here provide a procedure to identify important structural features in materials that could be missed by standard techniques and give unique insight into how these neural networks process data.

Serum deprivation/starvation leads to reactivation of HIV-1 in latently infected monocytes via activating ERK/JNK pathway.

Despite the high success rate, antiretroviral therapy does not cure the disease completely due to presence of latent viral reservoirs. Although several studies have addressed this issue earlier, the role of serum starvation/deprivation in HIV-1 latency has not been studied. So, we investigated the role of serum starvation in regulating HIV-1 latency. The impact of serum starvation on HIV-1 latency was assessed in latently infected monocytes U1 and T-cells J1.1. Serum starvation breaks HIV-1 latency in U1 cells. Under similar conditions, J1.1 cells failed to show reactivation of virus. We investigated the involvement of cell death pathway and autophagy during the serum starvation in viral reactivation. Inhibition of these pathways did not affect viral reactivation. Furthermore, other crucial factors like NF-κB, SP1 and AKT did not play any role in regulating viral latency. Here, we report that serum deprivation up-regulates ERK/JNK pathway. This leads to phosphorylation of c-Jun which plays an important role in viral reactivation. Treatment of cells with U0126, an ERK kinase inhibitor, potently inhibited viral replication. In summary, we show that serum starvation leads to reactivation of HIV-1 in latently infected monocytes through the ERK/JNK pathway.

Predicting molecular properties with covariant compositional networks.

Density functional theory (DFT) is the most successful and widely used approach for computing the electronic structure of matter. However, for tasks involving large sets of candidate molecules, running DFT separately for every possible compound of interest is forbiddingly expensive. In this paper, we propose a neural network based machine learning algorithm which, assuming a sufficiently large training sample of actual DFT results, can instead learn to predict certain properties of molecules purely from their molecular graphs. Our algorithm is based on the recently proposed covariant compositional networks framework and involves tensor reduction operations that are covariant with respect to permutations of the atoms. This new approach avoids some of the representational limitations of other neural networks that are popular in learning from molecular graphs and yields promising results in numerical experiments on the Harvard Clean Energy Project and QM9 molecular datasets.

HIV-1 Tat potently stabilises Mdm2 and enhances viral replication.

Murine double minute 2 (Mdm2) is known to enhance the transactivation potential of human immunodeficiency virus (HIV-1) Tat protein by causing its ubiquitination. However, the regulation of Mdm2 during HIV-1 infection and its implications for viral replication have not been well studied. Here, we show that the Mdm2 protein level increases during HIV-1 infection and this effect is mediated by HIV-1 Tat protein. Tat appears to stabilise Mdm2 at the post-translational level by inducing its phosphorylation at serine-166 position through AKT. Although p53 is one of the key players for Mdm2 induction, Tat-mediated stabilisation of Mdm2 appears to be independent of p53. Moreover, the non-phosphorylatable mutant of Mdm2 (S166A) fails to interact with Tat and shows decreased half-life in the presence of Tat compared with wild-type Mdm2. Furthermore, the non-phosphorylatable mutant of Mdm2 (S166A) is unable to support HIV-1 replication. Thus, HIV-1 Tat appears to stabilise Mdm2, which in turn enhances Tat-mediated viral replication. This study highlights the importance of post-translational modifications of host cellular factors in HIV-1 replication and pathogenesis.