In this paper, single snapshot direction-of-arrival (DOA) estimation under mutual coupling (MC) is considered for arbitrary array structures. A compressed sensing approach is utilized and a joint-sparse recovery algorithm is proposed for DOA and MC coefficient estimation. In this respect, both spatial source directions and MC coefficients are embedded into a joint-sparse vector. A new dictionary matrix is defined using the symmetricity of the MC matrix. The proposed approach does not depend on the structure of MC matrix and it is suitable for any type of array geometry.