A system-of-systems (SoS) is inherently open in configuration and evolutionary in lifecycle. For the next generation of cooperative cyber-physical system-of-systems, safety and security constitute two key issues of public concern that affect the deployment and acceptance. In engineering, the openness and evolutionary nature also entail radical paradigm shifts. This paper presents one novel approach to the development of qualified cyber-physical system-of-systems, with Cooperative Intelligent Transport Systems (C-ITS) as one target. The approach, referred to as knowledge-in-the-loop, aims to allow a synergy of well-managed lifecycles, formal quality assurance, and smart system features. One research goal is to enable an evolutionary development with continuous and traceable flows of system rationale from design-time to post-deployment time and back, supporting automated knowledge inference and enrichment. Another research goal is to develop a formal approach to risk-aware dynamic treatment of safety and security as a whole in the context of system-of-systems. Key base technologies include: (1) EAST-ADL for the consolidation of system-wide concerns and for the creation of an ontology for advanced run-time decisions, (2) Learning Based-Testing for run-time and post-deployment model inference, safety monitoring and testing, (3) Provable Isolation for run-time attack detection and enforcement of security in real-time operating systems.