You don't specify if the control voltage is with respect to ground, or if it can float.
Circuit 3 is the most practical N-channel scheme. The source is at a fixed voltage with respect to ground, which means you can provide a fixed gate-source voltage to control it. The MOSFET will be 'on' anywhere from +2.5 to +12V above ground, depending on the device.
Circuit 1 is tricky. When the MOSFET is off, the source is somewhat of a floating node (imagine a resistor divider with the top resistor enormous) sitting somewhere close to zero. When the MOSFET is on, the source will be very close to 400V assuming saturation. A moving source means that the gate-to-ground control voltage would have to move as well to keep the MOSFET on.
Circuit 1 is better if you reference the control voltage to the source of the MOSFET and not to ground. This is trivial if you're intending to drive the MOSFET with a PWM signal with sufficiently small on-time to allow use of a pulse transformer or charge-pump driver. Fixing the control voltage to the source of the MOSFET means the MOSFET can float up and down as it wants to, without impacting the drive.
Circuit 2 is straightforward like circuit 3. If the control voltage is referenced to ground, proving 397.5V to 388V from gate to ground (-2.5 to -12V from gate to source) will turn the MOSFET on. The source is fixed (always at +400V) so controlling the gate means a fixed voltage is all you need. (Unless your 400V bus collapses, but that's another issue).
Circuit 4, like circuit 2, is tricky. When the MOSFET is off, the source sits near 400V. When it is on, it will fall to near zero. A variable source means a variable gate supply with respect to ground, which is again a messy proposition.
In general, keep your sources fixed where possible, or if they have to float, use a floating supply to control them.