I really "love" those absurd Proteus schematics with unspecified signals coming from nowhere.
The gate voltage controlling Q1 must not only swing above the input supply voltage but also swing below cicruit ground. I can only guess, that's it's an unipolar square wave referred to ground now, which explains the observed results. It would be easy to provide a floating gate voltage referred to Q1 source in a simulation, but you'll have difficulties to implement it in a real circuit.
A NMOSFET is unsuitable for an inverting buck-boost converter, that's the simple truth. Use PMOS, as suggested.