Mutual event observations between the two components of 90 Antiope were carried out in 2007-2008. The pole position was refined to lambda0 = 199.5+/-0.5 eg and beta0 = 39.8+/-5 deg in J2000 ecliptic coordinates, leaving intact the physical solution for the components, assimilated to two perfect Roche ellipsoids, and derived after the 2005 mutual event season (Descamps et al., 2007). Furthermore, a large-scale geological depression, located on one of the components, was introduced to better match the observed lightcurves. This vast geological feature of about 68 km in diameter, which could be postulated as a bowl-shaped impact crater, is indeed responsible of the photometric asymmetries seen on the "shoulders" of the lightcurves. The bulk density was then recomputed to 1.28+/-0.04 gcm-3 to take into account this large-scale non-convexity. This giant crater could be the aftermath of a tremendous collision of a 100-km sized proto-Antiope with another Themis family member. This statement is supported by the fact that Antiope is sufficiently porous (~50%) to survive such an impact without being wholly destroyed. This violent shock would have then imparted enough angular momentum for fissioning of proto-Antiope into two equisized bodies. We calculated that the impactor must have a diameter greater than ~17 km, for an impact velocity ranging between 1 and 4 km/s. With such a projectile, this event has a substantial 50% probability to have occurred over the age of the Themis family.