The "digital transformation" is perceived as the key enabler for increasing wealth and well-being by politics, media and the citizens alike. In the same vein, digital government steadily receives more and more attention. Digital government gives rise to complex, large-scale state-level system landscapes consisting of many players and technological systems -and we call such system landscapes digital government ecosystems. In this paper, we systematically approach the state-level architecture of digital government ecosystems. We will discover the primacy of the state's institutional design in the architecture of digital government ecosystems, where Williamson's institutional analysis framework supports our considerations as theoretical background. Based on that insight, we will establish the notion of data governance architecture, which links data assets with accountable organizations. Our investigation results into a digital government architecture framework that can help in large-scale digital government design efforts through (i) separation of concerns in terms of appropriate categories, and (ii) a better assessment of the feasibility of envisioned digital transformations. With its focus on data, the proposed framework perfectly fits the current discussion on moving from ICT-driven to data-centric digital government.