Broad distribution of TPI-GAPDH fusion proteins among eukaryotes: evidence for glycolytic reactions in the mitochondrion?

Abstract

Glycolysis is a central metabolic pathway in eukaryotic and prokaryotic cells. In eukaryotes, the textbook view is that glycolysis occurs in the cytosol. However, fusion proteins comprised of two glycolytic enzymes, triosephosphate isomerase (TPI) and glyceraldehyde-3-phosphate dehydrogenase (GAPDH), were found in members of the stramenopiles (diatoms and oomycetes) and shown to possess amino-terminal mitochondrial targeting signals. Here we show that mitochondrial TPI-GAPDH fusion protein genes are widely spread across the known diversity of stramenopiles, including non-photosynthetic species (Bicosoeca sp. and Blastocystis hominis). We also show that TPI-GAPDH fusion genes exist in three cercozoan taxa (Paulinella chromatophora, Thaumatomastix sp. and Mataza hastifera) and an apusozoan protist, Thecamonas trahens. Interestingly, subcellular localization predictions for other glycolytic enzymes in stramenopiles and a cercozoan show that a significant fraction of the glycolytic enzymes in these species have mitochondrial-targeted isoforms. These results suggest that part of the glycolytic pathway occurs inside mitochondria in these organisms, broadening our knowledge of the diversity of mitochondrial metabolism of protists.

Figure 1. N-terminal leader sequences of TPI-GAPDH fusion proteins.

Alignment of the N-terminal region of TPI-GAPDH fusion proteins with four stand-alone TPI proteins. Amino acid residues on the leader sequences are colored as follows. Yellow: hydrophobic residues, blue: hydroxylated residues, orange: positively charged residues, purple: negatively charged residues and white: other polar uncharged residues. Results of prediction analysis are summarized on the right side of sequences. “Mitochondrial” suggests a mitochondrial localization based on the results of four different prediction programs (see main text).

Figure 2. Maximum likelihood tree of GAPDH protein sequences.

GAPDH ML tree constructed using RAxML with the LG+I+G+F model. GAPDH sequences fused with TPI are highlighted by colored boxes. The numbers at nodes are ML bootstrap values (left: RAxML, right: PhyML). Bootstrap values below 50% are not shown except values for nodes of particular interest. Nodes received bootstrap values ≥90% in both RAxML and PhyML analyses are indicated by thick line. The root was arbitrarily chosen for display purposes. Scale bar shows the number of inferred amino acid substitutions per site.

Figure 3. Maximum likelihood tree of PK protein sequences.

PK tree constructed using RAxML with the LG+G model. Putative mitochondrial-targeted PK sequences are highlighted by colored boxes. All other details are as in Figure 2.

Figure 4. Maximum likelihood tree of PGK protein sequences.

PGK tree constructed using RAxML and the LG+I+G+F model. Putative mitochondrial-targeted PGK sequences are highlighted by colored boxes. See Figure 2 for other presentation details.