Moving mass flight vehicles (MMFVs) with high-mass-ratio suffer from a more severe nonlinear attitude-servo coupling than the traditional configurations that are under particle moving mass hypothesis. Therefore, the coupled attitude-servo control problem is investigated in this paper, and an adaptive backstepping sliding mode controller with extended state observers and neural-networks (ABSMC-ENN) is proposed. Based on the formulated control model with strict-feedback form, a sliding mode control law (SMC) with exponential reaching rate and a nonsingular terminal sliding mode control law (NTSMC) are combined to perform the fast and robust tracking of the command angle-of-attack. Considering the disturbances and uncertainties existing in the coupled dynamics, two nonlinear extended state observers (ESOs) are utilized to estimate the total disturbances online for compensation. To further enhance the adaptiveness and dynamic response performances of the controller in various flight conditions, two radial basis function neural networks (RBFNNs) are exploited to optimize the switching gains of the controller in real-time with performance indexes being the functions of tracking errors. Closed-loop stability of the whole system is proved via Lyapunov methodology. Comparison simulation studies considering multiple disturbances, various flight conditions, and different controllers successfully demonstrate the superiority of the ABSMC-ENN in robustness and adaptiveness.