Part 1 must be printed with support structure, part 2 must be rotated 180° before printing. Printing must be very slow, increase print bed adhesion with glue stick if necessary. I’m still testing the best slicer settings myself, I’m using Voronoi printing as a guide.
Addition: I have revised part 1 again, on the one hand to make the individual spike proteins a little more stable and on the other hand to improve the detachment of the hemisphere after printing. For this purpose, I created a small recess for the attachment of a lever.

A small note on post-processing: it happened to me that a few spikes broke off when removing the support structure. With a soldering iron they can be “glued” back on without any problems and then hold as freshly printed