Enhancing low-light images is a key area of research in the domain of image restoration. While existing enhancement methods achieve impressive results, they often involve large parameter scales and high computational complexity. Lightweight approaches can improve computational efficiency but frequently suffer from subpar enhancement performance. To overcome these difficulties, we introduce a low-light image enhancement framework that combines both spatial and frequency domains, ensuring low computational cost and superior enhancement quality. Specifically, our method introduces a Retinex-based luminance enhancement component in the spatial domain to optimize the illumination information of images. At the same time, multi-scale wavelet decomposition is used to extract information from the frequency domain, enabling the independent optimization of low-frequency and high-frequency components. For the low-frequency component, Mamba is incorporated to enhance illumination and structural restoration by leveraging global information modeling. For the high-frequency component, we exploit the correlation between low- and high-frequency information to assist image recovery and improve detail expression. By avoiding frequent alternations between spatial and frequency domains, which can lead to instability, our method fully capitalizes on the advantages of both domains. Experimental results show that the proposed framework outperforms existing methods on various low-light enhancement datasets, while also ensuring low computational complexity and parameter efficiency, highlighting its potential for real-world applications.