The influence of linker design on fusion protein production and performance was evaluated when a family 9 carbohydrate-binding module (CBM9) serves as the affinity tag for recombinant proteins expressed in Escherichia coli. Two bioinformatic strategies for linker design were applied: the first identifies naturally occurring linkers within the proteome of the host organism, the second involves screening peptidases and their known specificities using the bioinformatics software MEROPS to design an artificial linker resistant to proteolysis within the host. Linkers designed using these strategies were compared against traditional poly-glycine linkers. Although widely used, glycine-rich linkers were found by tandem MS data to be susceptible to hydrolysis by E. coli peptidases. The natural (PT)(x)P and MEROPS-designed S(3)N(10) linkers were significantly more stable, indicating both strategies provide a useful approach to linker design. Factor X(a) processing of the fusion proteins depended strongly on linker chemistry, with poly(G) and S(3)N(10) linkers showing the fastest cleavage rates. Luminescence resonance energy transfer studies, used to measure average distance of separation between GFP and Tb(III) bound to a strong calcium-binding site of CBM9, revealed that, for a given linker chemistry, the separation distance increases with increasing linker length. This increase was particularly large for poly(G) linkers, suggesting that this linker chemistry adopts a hydrated, extended configuration that makes it particularly susceptible to proteolysis. Differential scanning calorimetry studies on the PT linker series showed that fusion of CBM9 to GFP did not alter the T(m) of GFP but did result in a destabilization, as seen by both a decrease in T(m) and DeltaH(cal), of CBM9. The degree of destabilization increased with decreasing length of the (PT)(x)P linker such that DeltaT(m) = -8.4 degrees C for the single P linker.