a conditional gaussian p(x1|x2) is said to be proportional to the joint density where it comes from, p(x1, x2)
In a sense, every conditional density is proportional to the corresponding joint density, since p(x1 | x2) = p(x1, x2) / p(x2), so maybe that's what they mean; it's not a property which is special to the Gaussian distribution. However, that's not "proportional" in the usual sense, because the factor 1/p(x2) is not a constant.