The aim of this paper is to establish formulas for the subdifferentials of the sum and the composition of convex functions in terms of the subdifferentials of the data functions at nearby points. Applications to general optimization problems lead to a new notion of sequential Lagrange multipliers.