We went over the AND gate paper in class, and generated a lot of constructive criticism that I hadn't thought of by just reading it on my own. I guess this is what class is for, huh.
In particular, we addressed the question of whether the AND gate is truly modular. As I discussed in the previous post, in principle this AND gate is modular in the sense that it can be plugged into different inputs and outputs. However, this isn't the whole story. "Plugging it in" to a different input is not so simple as just putting a new promoter in front of Input Gene 1. The whole promoter/RBS/coding-region assembly has to be tuned to have an appropriate strength. If you swap in a stronger promoter, you might have to weaken the RBS (ribosome binding site) in order to get just the right amount of protein expression to feed in to the rest of the AND gate.
In fact, that's exactly what the authors had to do when they first put their AND gate together, which I kind of glossed over. Recall that the system works by expressing an mRNA with amber stop codons in, and the amber-suppressor tRNA that can read those stop codons; the mRNA codes for a protein that transcribes the output promoter. Now, in principle this should Just Work. But in practice, if there's enough mRNA around, you can get spontaneous read-through even if the amber-suppressor tRNA is supposedly turned "off", for two reasons. One, even if the tRNA is "off", it might still be produced at a low basal level. Two, the amber stop codon is relatively "weak", and sometimes just gets read through anyway. (That is, it's not very good at recruiting the translation-stopping machinery, which is supposed to disassemble the ribosome and cut loose the newly translated protein.) So the authors had to adjust the RBS that governed translation of the mRNA, so there wouldn't be too much of it floating around and you wouldn't get this spurious effect.
Apparently this sort of adjustment is called "impedance matching", for those of you who are more familiar with electronics than cells. I don't know enough about electronics to explain exactly what impedance matching is or why it's a suitable analogy here, but it boils down to "make sure all the wires are carrying appropriate amounts of current, and if you connect something new to a wire you might have to add a resistor or something to fix the current back to how it used to be."
So, no, it's not plug-and-play quite yet. More like plug, mutagenize the RBS, and play... but maybe we'll get there eventually.
The other main criticism of this AND gate, from a modularity point of view, is that you can't have two copies of the gate in the same cell and expect them to operate independently. mRNAs and tRNAs float around, and if one AND gate is expressing mRNA and the other is expressing tRNA, then both of them will output ON, even though both of them ought to be OFF. This is a more serious problem, because you can't just tune an RBS and expect it to go away -- this design for an AND gate is in principle not modular with respect to other AND gates placed in the same cell. (Possible workarounds include hiding different AND gates in different cells and mixing several populations together, but then you have to work with cell-to-cell cooperation, which is a whole different ballgame.)