That’s kind of the Holy Grail there; down pitching without artifacts. I don’t think there is any approach that works great in every situation. It takes manual sound design to do it well.
One approach is to just resample (no “real” processing), layering the sound with versions of itself at different pitches, EQd, saturated, compressed etc to blend nicely. That way, you completely avoid the pitch shift artifacts, but still have proper material to cover the desired frequency range.
Oh, and speaking of blending; it can be really useful to have sounds intermodulate in various ways, to make them sound like one, rather than separate layers. Apply distortion or compression to the mixed layers, use a bit of AM, FM or ring modulation; that sort of stuff. Make the layers interact in an organic fashion, like they would if they were actual sounds emitted from the same physical object.