<p>Pinching-Antenna Systems (PASS) have emerged as a promising low-power alternative to traditional high-power 6G Multiple-Input Multiple-Output (MIMO) architectures by facilitating beamforming through selective antenna activation along dielectric waveguides instead of power-hungry phase shifters. Although PASS can potentially reduce the hardware power consumption of the antenna by over 95%, existing models mostly neglect the practical influence of mutual coupling between the active and pinched elements. In this study, we introduce a joint optimization approach based on an advanced radiation model that explicitly considers interelement coupling and waveguide spacing. The antenna selection and beamforming problems are then formulated as Mixed-Integer Nonlinear Programming (MINLP). We further develop efficient solution strategies to overcome this challenge: (i) a globally optimal Adaptive Tree Search (ATS) method using branch-and-prune principles; (ii) a low-complexity heuristic based on zero forcing; and (iii) a matching theoretic Cooperative Element Selection (CES) algorithm tailored for millisecond-level execution. Finite-time convergence guarantees and approximation bounds are provided with a thorough performance evaluation under realistic 6G channel conditions. The results indicate that the proposed coupling-aware ATS framework exhibits a 21.3% improvement in power efficiency and a 38.7% gain in spectral efficiency compared with conventional PASS designs. Moreover, the CES algorithm operates within 3% of the global optimum while reducing the computational complexity by 80%, highlighting PASS as a practical and energy-efficient solution for future green wireless networks.</p>