The biomedical literature contains a wealth of chemical-protein interactions (CPIs). Automatically extracting CPIs described in biomedical literature is essential for drug discovery, precision medicine, as well as basic biomedical research. However, the existing methods do not consider the impact of overlapping relations on CPI extraction. This leads to the extraction of sentences with overlapping relations becoming the bottleneck of CPI extraction. In this paper, we propose a novel neural network-based approach to improve the CPI extraction performance of sentences with overlapping relations. Specifically, the approach first employs BERT to generate high-quality contextual representations of the title sequence, instance sequence, and knowledge sequence. Then, the Gaussian probability distribution is introduced to capture the local structure of the instance. Meanwhile, the attention mechanism is applied to fuse the title information and biomedical knowledge, respectively. Finally, the related representations are concatenated and fed into the softmax function to extract CPIs. We evaluate our proposed model on the CHEMPROT corpus. Our proposed model is superior in performance as compared with other state-of-the-art models. The experimental results show that the Gaussian probability distribution and external knowledge are complementary to each other. Integrating them can effectively improve the CPI extraction performance. Furthermore, the Gaussian probability distribution can significantly improve the extraction performance of sentences with overlapping relations in biomedical relation extraction tasks. Data and code are available at https://github.com/CongSun-dlut/CPI_extraction.