PML and PML-like exonucleases restrict retrotransposons in jawed vertebrates.

We have uncovered a role for the promyelocytic leukemia (PML) gene and novel PML-like DEDDh exonucleases in the maintenance of genome stability through the restriction of LINE-1 (L1) retrotransposition in jawed vertebrates. Although the mammalian PML protein forms nuclear bodies, we found that the spotted gar PML ortholog and related proteins in fish function as cytoplasmic DEDDh exonucleases. In contrast, PML proteins from amniote species localized both to the cytoplasm and formed nuclear bodies. We also identified the PML-like exon 9 (Plex9) genes in teleost fishes that encode exonucleases. Plex9 proteins resemble TREX1 but are unique from the TREX family and share homology to gar PML. We also characterized the molecular evolution of TREX1 and the first non-mammalian TREX1 homologs in axolotl. In an example of convergent evolution and akin to TREX1, gar PML and zebrafish Plex9 proteins suppressed L1 retrotransposition and could complement TREX1 knockout in mammalian cells. Following export to the cytoplasm, the human PML-I isoform also restricted L1 through its conserved C-terminus by enhancing ORF1p degradation through the ubiquitin-proteasome system. Thus, PML first emerged as a cytoplasmic suppressor of retroelements, and this function is retained in amniotes despite its new role in the assembly of nuclear bodies.

Altering the binding determinant on the interdigitating loop of mandelate racemase shifts specificity towards that of d-tartrate dehydratase.

The enolase superfamily (ENS) has served as a paradigm for understanding how enzymes that share a conserved structure, as well as a common partial reaction (i.e., metal-assisted, Brønsted base-catalyzed enol(ate) formation), evolved from a common progenitor to catalyze mechanistically diverse reactions. Enzymes of the mandelate racemase (MR)-subgroup of the ENS share interdigitating loops between adjacent, 2-fold symmetry-related protomers of the tightly associated homodimers that comprise their quaternary structures. For the MR-subgroup members MR and d-tartrate dehydratase (TarD), the tip of the loop contributes a binding determinant to the adjacent active site (i.e., Leu 93 and Lys 102, respectively). To assess the role of Leu 93 of MR in substrate specificity and catalysis, we constructed L93 variants bearing hydrophobic (L93A, L93F, and L93W), polar neutral (L93N), acidic (L93D), or basic (L93K and L93R) residues at position 93. Gel filtration-HPLC revealed that wild-type MR and all L93 MR variants, apart from L93R MR (dimeric), were tetrameric in solution. The catalytic efficiency (kcat/Km) was reduced in the R→S and S→R reaction directions for all variants, primarily due to reduced turnover (kcat). Substitution of Leu 93 by Lys or Arg to mimic Lys 102 of TarD enhanced the binding of malate and tartrate, with meso- and d-tartrate exhibiting linear mixed-type inhibition of L93K MR. Despite the striking 500-fold increase in the binding affinity of d-tartrate, relative to (S)-mandelate, L93K MR exhibited no TarD activity. MD simulations suggested that the failure of L93K MR to catalyze α-deprotonation (i.e., H-D exchange) arises from inappropriate positioning of the Brønsted base (Lys 166). Thus, a change in binding determinant on the interdigitating loop can play a significant role in governing substrate specificity within the ENS, but does not necessarily confer 'new' catalytic activity despite similarities in catalytic machinery.