On the critical end point in a two-flavor linear sigma model coupled to quarks