This paper proposes, and evaluates the benefit of, one particular hybrid satellite-HAPS-ground network, where one high-altitude-platform-station (HAPS) connected to one geo-satellite assists the ground base-stations (BSs) at serving ground-level users. The paper assumes that the geo-satellite is connected to the HAPS using free-space-optical backhaul link. The HAPS, equipped with multiple antennas, aims at transmitting the geo-satellite data to the users via radio-frequency (RF) links using spatial-multiplexing. Each ground BS, on the other hand, is equipped with multiple antennas and serves users through RF links. The paper then focuses on maximizing the network-wide throughput, subject to user-connectivity constraints, HAPS and BSs power constraints, and backhaul constraints, so as to jointly determine the user-association strategy of each user (i.e., user to geo-satellite via HAPS, or user to BS), and their associated beamforming vectors. We tackle such a mixed discrete-continuous optimization problem using an iterative approach, where the user-association is determined using a combination of linear integer programming and generalized assignment problems, and where the beamforming strategy is found using a weighted-minimum-mean-squared-error approach. The simulations illustrate the appreciable gain of our proposed algorithm, and highlight the prospects of augmenting the ground networks with beamforming-empowered HAPS for connecting the unconnected, and super-connecting the connected.