We have studied the significance of the N-terminal presequence of watermelon (Citrullus vulgaris) glyoxysomal malate dehydrogenase [gMDH; (S)-malate:NAD+ oxidoreductase; EC 1.1.1.37] in microbody targeting. The yeast Hansenula polymorpha was used as heterologous host for the in vivo expression of various genetically altered watermelon MDH genes, whose protein products were localized by immunocytochemical techniques. It is shown that the presequence of gMDH is essential and sufficient for peroxisomal targeting; it can target the mature part of the mitochondrial MDH to microbodies, whereas deletion of the presequence results in accumulation of the mature form of gMDH in the cytosol. Alignment of the N termini of several peroxisomal proteins that are assumed to contain a peroxisomal targeting signal at the N terminus (PTS2) suggested the consensus seqence RL-X5-HL. A similar motif is present in the presequence of watermelon gMDH--namely, 10RI-X5-17HL. Mutational analysis revealed that substitutions of 10RI into DD or 17HL into DE destroyed the topogenic information, whereas substitutions of 25M into I and 26EE into LV did not. By combining our data with recent analyses of others on the presequences of mammalian thiolases, it is concluded that the peroxisomal targeting information of PTS2 is contained in the consensus sequence RL/I-X5-HL. In contrast to the higher plant and mammals, the Hansenula yeast peroxisomes seem to lack an enzyme capable of removing microbody presequences of higher eukaryotes.