Exotic invasive species can influence the behavior and ecology of native and resident species, but these changes are often overlooked. Here we hypothesize that the ghost ant, Tapinoma melanocephalum , living in areas that have been invaded by the red imported fire ant, Solenopsis invicta , displays behavioral differences to interspecific competition that are reflected in both its trophic position and symbiotic microbiota. We demonstrate that T . melanocephalum workers from S . invicta invaded areas are less aggressive towards workers of S . invicta than those inhabiting non-invaded areas. Nitrogen isotope analyses reveal that colonies of T . melanocephalum have protein-rich diets in S . invicta invaded areas compared with the carbohydrate-rich diets of colonies living in non-invaded areas. Analysis of microbiota isolated from gut tissue shows that T . melanocephalum workers from S . invicta invaded areas also have different bacterial communities, including a higher abundance of Wolbachia that may play a role in vitamin B provisioning. In contrast, the microbiota of workers of T . melanocephalum from S . invicta -free areas are dominated by bacteria from the orders Bacillales, Lactobacillales and Enterobacteriales that may be involved in sugar metabolism. We further demonstrate experimentally that the composition and structure of the bacterial symbiont communities as well as the prevalence of vitamin B in T . melanocephalum workers from S . invicta invaded and non-invaded areas can be altered if T . melanocephalum workers are supplied with either protein-rich or carbohydrate-rich food. Our results support the hypothesis that bacterial symbiont communities can help hosts by buffering behavioral changes caused by interspecies competition as a consequence of biological invasions.

Methylation of RNA plays an important role in cancer. Classical forms of such modifications include N6-methyladenine (m6A), 5-methylcytosine (m5C), and N1-methyladenine (m1A). Methylation-regulated long non-coding (lnc) RNAs are involved in various biological processes, such as tumor proliferation, apoptosis, immune escape, invasion, and metastasis. Therefore, we performed an analysis of transcriptomic and clinical data of pancreatic cancer samples in The Cancer Genome Atlas (TCGA). Using the co-expression method, we summarized 44 m6A/m5C/m1A-related genes and obtained 218 methylation-associated lncRNAs. Next, with COX regression, we screened 39 lncRNAs that are strongly associated with prognosis and found that their expression differed significantly between normal tissues and pancreatic cancer samples (P < 0.001). We then used the least absolute shrinkage and selection operator (LASSO) to construct a risk model comprising seven lncRNAs. In validation set, the nomogram generated by combining clinical characteristics accurately predicted the survival probability of pancreatic cancer patients at 1, 2, and 3 years after diagnosis (AUC = 0.652, 0.686, and 0.740, respectively). Tumor microenvironment analysis showed that the high-risk group had significantly more resting memory CD4 T cells, M0 macrophages, and activated dendritic cells and fewer naïve B cells, plasma cells, and CD8 T cells than the low-risk group (both P < 0.05). Most immune-checkpoint genes were significantly different between the high- and low-risk groups (P < 0.05). The Tumor Immune Dysfunction and Exclusion score showed that high-risk patients benefited more from treatment with immune checkpoint inhibitors (P < 0.001). Overall survival was also lower in high-risk patients with more tumor mutations than in low-risk patients with fewer mutations (P < 0.001). Finally, we explored the sensitivity of the high- and low-risk groups to seven candidate drugs. Our findings indicated that m6A/m5C/m1A-associated lncRNAs are potentially useful biomarkers for the early diagnosis and estimating the prognosis of, and ascertaining the responses to immunotherapy in, patients with pancreatic cancer.

IntroductionTo identify proteins and corresponding genes that share sequential and structural similarity with programmed cell death protein-1 (PD-1) in patients with type 1 diabetes mellitus (T1DM) via bioinformatics analysis.Research design and methodsAll proteins with immunoglobulin V-set domain were screened in the human protein sequence database, and the corresponding genes were obtained in the gene sequence database. GSE154609 was downloaded from the GEO database, which contained peripheral blood CD14+ monocyte samples from patients with T1DM and healthy controls. The difference result and the similar genes were intersected. Analysis of gene ontology and Kyoto encyclopedia of genes and genomes pathways was used to predict potential functions using the R package ‘cluster profiler’. The expression differences of intersected genes were analyzed in The Cancer Genome Atlas pancreatic cancer dataset and GTEx database using t-test. The correlation between the overall survival and disease-free progression of patients with pancreatic cancer was analyzed using Kaplan-Meier survival analysis.Results2068 proteins with immunoglobulin V-set domain similar to PD-1 and 307 corresponding genes were found. 1705 upregulated differentially expressed genes (DEGs) and 1335 downregulated DEGs in patients with T1DM compared with healthy controls were identified. A total of 21 genes were overlapped with the 307 PD-1 similarity genes, including 7 upregulated and 14 downregulated. Of these, mRNA levels of 13 genes were significantly increased in patients with pancreatic cancer. High expression ofMYOM3andHHLA2was significantly correlated with shorter overall survival of patients with pancreatic cancer, while high expression ofFGFRL1,CD274, andSPEGwas significantly correlated with shorter disease-free survival of patients with pancreatic cancer.ConclusionsGenes encoding immunoglobulin V-set domain similar to PD-1 may contribute to the occurrence of T1DM. Of these genes,MYOM3andSPEGmay serve as potential biomarkers for the prognosis of pancreatic cancer.

Abstract. Word representation is the basic research content of natural language processing .At present, distributed representation of monolingual words has shown satisfactory application effect in some Neural Probabilistic Language (NPL) research, while as for distributed representation of cross-lingual words, there is little research both at home and abroad. Aiming at this problem given distribution similarity of nouns and verbs in these two languages, we embed mutual translated words, synonyms, super-ordinates into Chinese corpus by the weakly supervised learning extension approach and other methods, thus Laos word distribution in cross-lingual environment of Chinese and Laos is learned. We applied the distributed representation of the cross-lingual words learned before to compute similarities of bilingual texts and classify the mixed text corpus of Chinese and Laos, Experimental results show that the proposal has a satisfactory effect on the two tasks.