Maybe I'm missing something, but I think physics prevents this. An excellent lens has a numerical aperture of maybe 0.95 at best, so with Abbe's formula you get a resolution limit of maybe 200 nm at best. With an oil immersion microscope (still a light microscope, albeit not an ordinary one) you might get as low as 100 nm, but that doesn't mean you could see that fact that the tubes are square. It's a long time since I've heard physics and that I've been in an laboratory, so maybe I'm wrong.
No you are right - I kind of buried the lead there. The cubes themselves have a side-length of 100-120nm, but it means the longest diagonal length is actually more like ~211nm.
But there's also the 2D diagonal which is 172nm, so what seems to happen is you end up seeing two slightly super-imposed and different shaped blurs, whereas normally you'd see just the 1 if they were perfect spheres.
