Un-phased: Lipid Droplets Modulate the Bioavailability of Antibiotics

Abstract

In a paper recently published in Science (Greenwood et al., 2019), Greenwood and colleagues now describe a fascinating example of how partitioning of a small lipophilic molecule into a phase-separated cellular constituent, the lipid droplet (LD), contributes to its antibacterial action against Mycobacterium tuberculosis.

One of the first lessons an experimental biochemist learns is “if it does not mix, it will not react,” indicating that, for example, reaction rates are predictable only for homogeneous solutions. But of course, life is not based on homogeneous mixtures, but rather compartmentalization of biochemical reactants. This allows particular reactants to reach high local concentrations and separates distinct reactions competing for the same substrates. Much of this segregation is achieved through the evolution of membrane-bound organelles, such as mitochondria or the endo-plasmatic reticulum (ER). However, we are increasingly realizing that compartments based on phase separation of biomolecules can provide an additional way of segregating molecules. For example, the cell nucleolus consists of phase-separated proteins and RNAs dedicated to forming new ribosomes.

Lipid droplets (LDs) are highly unusual cellular organelles (Olzmann and Carvalho, 2019; Walther and Farese, 2012). Their hydrophobic, neutral-lipid oil-phase core is bounded by a monolayer of phospholipids that prevents LD coalescence. Specific proteins embedded into this monolayer mediate many of the biochemical reactions involving LDs. With LDs forming a dispersed oil phase, cells are effectively converted to an emulsion, with the cytosol forming the aqueous phase. Due to their composition, LDs diffract light, and as a consequence, LDs were among the first organelles discovered by light microscopy in the late 1800s (Altmann, 1890).

The LD oil phase typically contains mostly the neutral lipids triacylglycerols and sterol esters as storage pools for metabolic energy or membrane precursors. However, depending on the cell type, the neutral lipid core can also contain other hydrophobic molecules, such as retinol esters or lipophilic vitamins that partition into the oil phase. In addition, the surface of LDs appears to have evolved a number of unanticipated functions, such as storing otherwise toxic proteins (e.g., histones) and participating in metabolic pathways (e.g., phosphatidylcholine biosynthesis).

Since the availability and need for lipophilic molecules stored in LDs fluctuates, the LD population of cells is highly dynamic. LDs form and grow in conditions where excess lipids are available for storage, and LDs shrink when the lipids are catabolized during times of need. For these dynamic reactions, LDs may interact with other organelles, such as mitochondria, peroxisomes, or the ER (Schuldiner and Bohnert, 2017).

As another fascinating function, LDs participate in the replication of a number of intracellular pathogens. These include viruses, such as hepatitis C virus (HCV), which requires targeting of HCV core and other proteins to the LD surface for viral assembly (Meyers et al., 2016), and Toxoplasma gondii, which may utilize the lipids in LDs for replicative growth (Nolan et al., 2017). Moreover, some bacterial pathogens interact with LDs, presumably to tap their hydrocarbon cores as an energy source for replication. One such pathogen is Mycobacterium tuberculosis, the cause of one of the world’s deadliest infectious diseases, tuberculosis. M. tuberculosis replicates in tissue macrophages and can remain dormant in these cells for years. Intracellular M. tuberculosis not only interacts with LDs but can also be found in a number of other niches within host macrophage cells, including compartments of the endosomal system or “membranous vacuoles” (Barisch and Soldati, 2017).

In the current study, Greenwood and colleagues asked where in macrophages antibiotics targeting M. tuberculosis localize and whether uneven distribution of antibiotics affects their efficacy in targeting bacteria. Many antibiotics, including bedaquiline, an inhibitor of the bacteria’s ATP-synthase that is used to combat M. tuberculosis infection, are hydrophobic molecules, which allows them to cross membranes. As such, one might expect that they partition predominantly into the LD oil phase, effectively removing them from the pool of bioactive molecules.

Surprisingly, Greenwood et al. (2019) report a different outcome. Using sophisticated ion imaging, they found that bedaquiline indeed accumulated in macrophage LDs but that this localization unexpectedly enhanced its antibiotic effect. The authors attribute this surprising result to the effects of the remodeled cellular metabolism induced by M. tuberculosis. Intracellular bacteria first induce LD formation in macrophages and then rapidly utilize the nutrients within them, leading to LD shrinkage and disappearance. When macrophages are treated with bedaquiline, the antibiotic accumulates in the abundant LDs, and as the organelles are utilized rapidly, high amounts of the antibiotic are liberated and transferred to the pathogen, thus more potently poisoning it.

Although unexpected in the context of the pharmacology of lipophilic molecules, these findings are consistent with earlier reports on the interaction of lipophilic toxins with LDs. For instance, it has been known for years that yeast strains engineered to lack LDs are more sensitive to the anti-fungal compound terbinafine (Sorger et al., 2004), possibly because the drug is not sequestered in LDs.

The new findings thus unexpectedly suggest that lipid partitioning of antibiotics may, at least in some cases, be a desired characteristic of antimicrobial compounds. In addition to M. tuberculosis, a number of pathogens may interact with LDs, including Plasmodium, Trypanosoma, and Toxoplasma species (Vallochi et al., 2018), and this insight into lipid partitioning might be useful to combat these other pathogens. Moreover, one could imagine that partitioning into other phase-separated cellular compartments could be exploited in the development of new therapeutic approaches.

However, many questions remain. For example, how is the benefit of partitioning antibiotics into macrophage LDs counteracted by sequestering the drug in other cell types that don’t directly participate in the infection, but contain large amounts of the oil phase, such as adipose tissue? Adipocytes contain by far the largest reservoir of triacylglcyerol in the human body and thus might trap the largest amount of the antibiotic. Whether adipose tissue droplets provide a dead-end pool or act as a dynamic buffer prolonging the effect of bedaquiline and other molecules is unclear. For macrophages themselves, does the composition of LDs (i.e., triacylglycerols versus sterol esters) have consequences for their ability to store and mobilize antibiotic compounds? Would the use of acyl CoA:cholesterol acyltransferase (ACAT) and acyl CoA:diacylglycerol acyltransferase (DGAT) inhibitors, which block the formation of LDs, acutely increase the concentrations and killing power of bedaquiline?

Also, are the effects of LD partitioning significant in systemically combating the infection in humans? Interestingly, Greenwood et al. (2019) found that bedaquiline mostly removed cells with a large number of LDs and M. tuberculosis from the population. It will be interesting to investigate how this may impact the evolution of the infection or the emergence of resistant strains.

In any case, the current study introduces a fascinating new concept to pharmacodynamics—the partitioning of hydrophobic drugs into intracellular LDs—that may be exploited to combat specific intracellular pathogens.