I was reading the slides from an interesting talk by Raphael Bousso. On slide 12 he wrote:
Virtual particles contribute different fractions of the mass of different materials (e.g., to the nuclear electrostatic energy of aluminum and platinum) I If they did not gravitate, we would have detected this difference in tests of the equivalence principle (in this example, to precision $10^6$)
I was wondering if there is a good reference which discusses this in more detail?