From your tests it sounds as if the AC/DC coupling in the scope is actually working - thanks for checking this.
The problem here really comes down to feeding a 400V signal into a scope with a 20V range. When the scope is set to DC coupling, both the AC and DC parts of your signal will be attenuated, using a x10 probe you will still see 40V which is an overload for the scope.
When the scope is set to AC, then the AC part of the signal will be attenuated by a factor of 10, but your DC signal will not be attenuated by your probe, the scope input will see a 400V signal which is an overload.
You have three options here - one is to use an external capacitor as you have done. This has obviously to be rated at > 400V and be of a low leakage type.
The second option is to attenuate your signal before the scope - for automotive use we sell a 20:1 external attenuator. http://www.picotech.com/auto/attenuator.html
This has a max input of 300V, but using this at 400V is going to be much less of an overload that putting 400V into a 20V scope input.
The third option is to use an active probe such as the x10/x100 probe from http://accessories.picotech.com/oscillo ... robes.html
I hope this helps.