Detection, immobilization and purification of carbohydrates can be done using molecular probes that specifically bind to targeted carbohydrate epitopes. Carbohydrate-binding modules (CBMs) are discrete parts of carbohydrate-hydrolyzing enzymes that can be engineered to bind and detect specifically a number of carbohydrates. Design and engineering of CBMs have benefited greatly from structural studies that have helped us to decipher the basis for specificity in carbohydrate-protein interactions. However, more studies are needed to predict which modifications in a CBM would generate probes with predetermined binding properties. In this report, we present the crystal structures of two highly related engineered CBMs with different binding specificity profiles: X-2, which is specific for xylans and the L110F mutant of X-2, which binds xyloglucans and β-glucans in addition to xylans. The structures of the modules were solved both in the apo form and complexed with oligomers of xylose, as well as with an oligomer of glucose in the case of X-2 L110F. The mutation, leucine to phenylalanine, converting the specific module into a cross-reactive one, introduces a crucial hydrogen-π interaction that allows the mutant to retain glucan-based ligands. The cross-reactivity of X-2 L110F is furthermore made possible by the plasticity of the protein, in particular, of residue R142, which permits accommodation of an extra hydroxymethyl group present in cellopentaose and not xylopentaose. Altogether, this study shows, in structural detail, altered protein-carbohydrate interactions that have high impact on the binding properties of a carbohydrate probe but are introduced through simple mutagenesis.