Not all members of the order Carnivora are carnivorous. Some are omnivorous, and a few, such as the giant panda, Ailuropoda melanoleuca, are almost exclusively herbivorous. Although a number of adaptations to increased plant-eating are recognized within Carnivora, few have been studied at the molecular level. One molecular adaptation to diet that is spread widely across Mammalia is the differential intracellular targeting of the intermediary metabolic enzyme alanine:glyoxylate aminotransferase (AGT), which tends to be mitochondrial in carnivores, peroxisomal in herbivores, and both mitochondrial and peroxisomal in omnivores. In the present study, we have analyzed the targeting of AGT in Carnivora in relation to species' natural diets. We show not only that there has been an adaptive shift in AGT targeting from the mitochondrion toward the peroxisome as diets have shifted from being mainly carnivorous to ones that are more omnivorous and herbivorous but also that in one lineage, namely that of the giant panda, there is evidence for positive selection pressure at the molecular level on the AGT mitochondrial targeting sequence to decrease its efficiency, thereby allowing more AGT to be targeted to the peroxisomes.

Unlike most organellar proteins, some peroxisomal proteins are often found in significant amounts in the cytosol. Such apparent import inefficiency is very marked in guinea pig (Caviaporcellus) hepatocytes in which the cytosolic levels of two peroxisomal proteins, catalase and alanine:glyoxylate aminotransferase (AGT), are much higher than those found in human (Homosapiens) hepatocytes, for example. In an attempt to provide an explanation for this phenomenon, we have cloned the guinea pig CpPEX5 gene, which encodes the peroxisomal targeting sequence type 1 (PTS1) import receptor Pex5p, and functionally compared it with its human homologue, HsPex5p. Our results showed the following: (1) CpPEX5, like its human homologue, encodes two splice variants differing by the presence or absence of an internal region of 37 amino acids; (2) both variants were expressed in all guinea pig tissues studied; (3) both variants were equally able to complement peroxisomal import of PTS1 proteins in microinjected ∆pex5 human fibroblasts; (4) CpPex5p was as efficient as HsPex5p in mediating the peroxisomal import of proteins possessing the consensus PTS1, Ser-Lys-Leu, but much less efficient in mediating the import of proteins possessing non-consensus PTS1s (i.e. Lys-Lys-Leu of human AGT and Ala-Asn-Leu of human catalase); (5) reporter proteins with the consensus PTS1, Ser-Lys-Leu, inhibited the peroxisomal import of endogenous catalase, whereas AGT with the non-consensus Lys-Lys-Leu did not; (6) high concentrations of HsPex5p, but not CpPex5p, markedly inhibited the import of AGT, but not catalase or proteins ending in Ser-Lys-Leu; and (7) in the yeast two-hybrid system, AGT-Ser-Lys-Leu interacted with the tetratricopeptide repeat domain of HsPex5p, but AGT-Lys-Lys-Leu did not. In addition, AGT-Ser-Lys-Leu was targeted to peroxisomes in Saccharomycescerevisiae, whereas AGT-Lys-Lys-Leu was not. These data suggest that the inefficient peroxisomal import of AGT and catalase in guinea pig cells is due to the inefficiency with which CpPex5p mediates the peroxisomal import of proteins containing non-consensus PTS1s. They also suggest that the non-consensus PTS1 of human AGT might interact with HsPex5p very differently compared with the consensus PTS1, Ser-Lys-Leu.

Liver microsomes contain two carnitine acyltransferase activities. One of these has properties closely corresponding to those of 88 kDa mitochondrial carnitine palmitoyltransferase-1 (CPT-1). Antisera against CPT-1 cross-react with an 88 kDa microsomal protein, suggesting that CPT-1 may be targeted to both microsomal and mitochondrial membranes. However, no experiments using cDNAs corresponding to CPT-1 involving in vitro translation with microsomes or involving in vivo COS-1 cell transfection provided any evidence to support this hypothesis.