Discussion
Structures of DPAGT1 allowed us to explain the mechanism of this key enzyme in protein N-glycosylation. We showed that missense variants in DPAGT1 associated with CMS and CDG-Ij alter DPAGT1 function via diverse mechanisms. For many cases of milder CMS disease, severely reduced activity from one allele is combined with an allele with partial activity. In two cases, Val264Gly and Gly160Ser, it appeared that errors in splicing reduced the levels of correct mRNA were partially compensated by 2-fold increases in enzymatic activity. In CDG-Ij, either one allele produces protein with 20% activity or, alternatively, two alleles producing 5%–10% activity, leading to much greater disease severity. In all cases some active protein is present, with a threshold of symptoms and increasing disease severity between no disease at 50% activity and severe disease with 5%–10% of activity. It is significant that DPAGT1 activity can be increased by point mutations at single sites, suggesting it may be possible to increase enzymatic activity and/or modulate stability with small molecules, e.g., pharmacological chaperones (Convertino et al., 2016, Sánchez-Fernández et al., 2016).
DPAGT1 is an “off-target” for the natural bactericidal tunicamycin. Comparison of the human PNPT DPAGT1 and bacterial PNPT MraY structures revealed a gating loop (residues Cys299-Arg303) in DPAGT1 next to where the N-2′′ atom of tunicamycin binds, that is absent in the more open structure of MraY. This difference allowed design of analogues TUN-X,X with two lipid chains targeted to bind MraY, but not DPAGT1. This circumvented the toxicity problem normally observed with tunicamycin. Additive modes of action against other carbohydrate-processing enzymes, such as Mtb WecA or TagO/TarO (Ishizaki et al., 2013); (Santa Maria et al., 2014), may also be important for the effects of the analogues.
Mtb is responsible for ∼1.3 million deaths per annum with increasing spread of drug resistant strains requiring new strategies (Young et al., 2008, Zumla et al., 2013). We have shown that the TUN-X,X lipid analogues have much lower toxicity than tunicamycin itself, are effective in killing Mtb in vitro, in cellulo, and in vivo. The analogues did show some toxicity in mice after more than 2 weeks in diseased (but not healthy) animals, yet they are still much less toxic than tunicamycin. While these lead versions of the TUN-X,X lipid analogues are not as effective as the frontline drugs rifampicin and isoniazid in macrophages and in mice in vivo, details of the effects in mice are often not recapitulated in humans. In addition, we do not yet have data on intracellular uptake in animals, which may be affecting the outcome. MICs show these compounds are excellent leads for the design of novel antibiotics with a new mechanism of action. TUN-X,X lipid analogues are effective antibacterials, with limited toxicity in human cells and in mice (at least with short term dosing), and these and other analogues (Price et al., 2017b) (Price et al., 2017a) suggest a novel approach to development of antibiotics against Gram-positive bacteria.