In this paper, a novel scheme of command to line-of-sight guidance law regarding the outstanding practical limitations is proposed. With the aid of formulating problem in pursuer's line of sight frame, the exact feed-forward acceleration commands are derived for a general target motion in three-dimensional engagement geometry. Then, a nonlinear control scheme constructed from feed-forward terms along with a nested saturated feedback strategy is suggested. Two momentous practical limitations are considered. The first is related to existence of pursuer's acceleration saturation which is ignored in all existing solutions. The required conditions to insure the global stability are derived regarding this inevitable constraint. The second is related to existence of guidance filter which is assumed to be ideal in existing literature. The importance of guidance filter is highlighted in the present work employing a cubature Kalman filter. The effectiveness of proposed scheme in comparison with the well-known augmented proportional navigation law, with the absence and presence of guidance filter, autopilot dynamics and target maneuvers is illustrated through benchmark scenarios.