If you're sure they're 3-betting light, then 4-betting light can be profitable.

Play around with pokerstove. Click on Player 1 and assign a light 3-betting hand range and see what percentage of all hands that is. Take away the hands you think will fold to a 4-bet and see how much of his range you're folding. This is based on some back-of-the-envelope EV calculations, but if it isn't well more than half then I don't think he's 3-betting light enough to make 4-betting light profitable.

Also play around with pokerstove and see how various hands stack up against a typical 4-bet calling range. Most villains won't call (or shove over) a four-bet without QQ+/AK. Against this range...

TT = 36.4%
22 = 34%
65s = 31.1%
T9s = 29.9%
97s = 29.5%
ATs = 29.3%
A2s = 28.9%
A2o = 24.8%
AQo = 24.4%

Some villains won't call a 4-bet with AK, and if you take it away AK and add JJ, it looks a lot different...

AQo = 30%
ATs = 29.6%
A2s = 29.4%
A2o = 25.5%
65s = 22.4%
97s = 20.5%
T9s = 20.3%
TT = 18.6%
22 = 18%

What looks good on both? Suited aces.

Something counterintuitive? If you're going to 4-bet light with suited connectors, better to choose 65s or 87s b/c they have more equity when called by overpairs or overcards than JTs or T9s.

Cliffnotes: Be as sure as you can that villain is 3-betting light before you 4-bet light. Most of your +EV from 4-betting light comes from fold equity. But, if you're worried about equity if called, suited aces and low-to-medium suited connectors are good choices for some typical 4-bet calling ranges.