We consider two synchronization problems, namely, state synchronization for networks of identical linear agents, and regulated output synchronization for networks of non-identical linear agents, under a time-varying topology. Agents are assumed to be multiple-input multiple-output (MIMO), minimum-phase, right invertible and non-introspective (i.e., they only receive relative output information from their neighboring agents, but lack independent information about their own states or outputs). The time-varying topology is assumed to switch among a finite number of directed graphs containing a directed spanning tree with an arbitrarily a priori given non-vanishing dwell time. Moreover agents are unable to exchange of controller states with their neighboring agents. For solving each problem, we propose a distribute dynamic protocol based on a combination of low-gain and high-gain design techniques.