Have you tried swapping the connections to A1 and A2 over?
That can only change the direction of steps. Swapping A1<->A2 or B1<->B2 or swapping both As for both Bs will each
change the direction of rotation. Miswired states have a winding connected to one A terminal and one B terminal,
which is wrong. A's drive one winding, B's drive another. Direction can be sorted out in software or by swapping
one winding's wires.
If winding A is trying to turn in one direction and winding B in the other, then the motor won't turn. It may 'chatter', depending on the frequency of the PWM, but it won't turn. Reversing the connections to BOTH windings will reverse the direction the motor turns, only if both windings are trying to turn the motor in the SAME direction.
Bipolar stepper motors are driven in quadrature, I don't think you understand this. Each winding by itself has
no say in the direction, its the relative phase of the A and B windings that determines the direction. Reverse
the sense of the current in A (leaving B unchanged) and the direction will reverse, ditto for B, ditto for swapping the roles
of A and B. Each acts just like a reflection of the motor geometry in space, and a reflection in space reverses
the direction of rotation.
I've a stepper motor right here in front of me, trust me this is how they work.