How many AU pairs are bad?

So I have been trying to focus on using as few GC pairs as possible and can design some pattern with almost no GC pairs. When I try that on the lab the dot chart looks like poopy. How few GC’s is bad and what is really a good balance?

Although GC pairs can be troublesome as they can more easily sway a sequence into folding in alternative, stable structures (and sequences with GC excess can also be a pain to synthesize because of their interference with PCR [the method we use to create the DNA that will be turned into RNA]), they are definitely useful to stabilize the structure at key points. Lack of GC base pairs makes the RNA fold into several alternative and unstable structures, because of the lack of rigidity in AU and GU pairs (GC pairs have three hydrogen bonds, AU only two).
When designing sequences, I usually start using AU pairs and then use GC base pairs to stabilize the structure at key points (normally these somewhere in the middle of long helices and at closing loops or for stacking effects).
Finally, if you look at natural RNA sequences that use their structure to perform their function, their average GC content is around 50%, indicating that GC pairs are indeed as important as the others.

It is not bad to have too many AU pairs, but on the short stacks, GC pairs are needed because they are stronger.

the natural ratio in the aptamer database is 11%GU, 31%AU, 58%GC