You do not 3D-print protein structures (unless you mean 3D models of proteins in which case ignore the rest of this post). Not in the sense of "here's an arbitrary position of atoms, give me a molecule". There's a reason why Folding@Home takes a lot of processing power. It's very hard to figure out how chains of amino acids will interact with themselves.

The best you can do is do gene synthesis using a method called Oligonucleotide Synthesis

http://en.wikipedia.org/wiki/Oligonucleotide_synthesis

Using this technique, you can take an arbitrary gene sequence (ATGC sequence) and get a plasmid out of it. This will specify a sequence of amino acids that will fold into a protein. Then you just have to get a bacterial host to take it up into it's DNA and then make sure the chemical conditions of the host are right to generate your protein (you can't synthesize a protein if there aren't the right types of Amino Acids floating around, for example).

Companies like Genscript (which I found by randomly googling) will manufacture a specified gene, and send you back frozen plasmids, which you can then inject into your favorite bacterial host.

http://www.genscript.com/gene_synthesis.html

> Does anyone know how to 3D-print protein structures?

Use pdb2stl. It's part of mayavi.

Most people use ribosomes to do this.

I can connect you to http://www.deskgen.com/ who might be able to help?