In a wider sense: The sphere is the smallest surface that encloses a given volume.
The sphere is the solution to one of the most general optimization problems in three-dimensional geometry.
This makes it, in isotropic nature, the solution to many physical optimization problems, for example finding the minimal potential energy configuration of a blob of matter, the surface of least potential energy of a soap bubble enclosing an amount of gas, the simplest solution for a single electron swinging around a proton etc. If you want, that is a consequence of an isomorphism (or, perhaps, homomorphism) of math and nature. That structural similarity is not surprising; mathematics, practically as well as conceptually — because it is conceived by humans living in the world as it is —, is rooted in nature.
Natural systems, absent external influences, can spontaneously only move in the direction of least potential energy because the path there from any other state releases energy (photons are emitted, heat is generated). Leaving the state of lowest potential energy requires an influx of energy (heating, absorption of photons) or, more generally, outside influence. The result in the latter case is, of course, the lowest potential energy for the overall system.
The sphere is the ideal solution for such problems in the absence of any disturbances, which is why none of the actual existing shapes are spheres — for example, none of the celestial bodies is exactly spherical ;-).