I've read somewhere (but cannot locate the source) that the following classification holds: simply connected homogeneous 3-manifolds are either isometric to $S^2 \times \mathbb{R}$ or to a metric Lie group.
I'm looking for a reference with a proof of the above statement. In particular, I'm curious if there is a proof of the compact case that does not rely on the Poincaré conjecture (since, if I'm not mistaken, $S^3$ is the only simply connected homogeneous closed 3-manifold, and is a Lie group).