Existing low-light image enhancement networks have made remarkable progress in either the spatial or frequency domain. However, spatial-domain methods often overlook the spectral differences between clear and degraded images, which may lead to the loss of image details or the introduction of noise. In contrast, frequency-domain methods lack the flexibility to effectively restore image features. To address these issues, we propose a novel dual-domain low-light image enhancement network (DDLIE-Net), which consists of a Frequency Interaction Enhancement (FIE) module, a Spatial-Domain Structure-Guided Attention (SDSGA) module, and a Dual-Domain Illumination Restorer (DDIR). The FIE module adopts a multi-branch, content-aware framework that dynamically decomposes features into different frequency sub-bands and adaptively enhances them for restoration. The SDSGA module leverages structural features to guide non-local interaction modeling, enhancing the frequency structures processed by the FIE module in the spatial domain. The integration of the FIE and SDSGA modules within the DDIR facilitates the effective restoration of frequency information, thereby improving the overall performance of DDLIE-Net. Experimental results demonstrate that the proposed model achieves outstanding performance across twelve benchmark tests.