Understanding the Lipid Droplet Proteome and Protein Targeting

Abstract

Reporting in this issue of Developmental Cell, Bersuker et al. (2018) adapt APEX technology to lipid droplets for a more accurate view of the droplet proteome and Prévost et al. (2018) provide important insights into the basis of droplet protein targeting that altogether extend the understanding of this organelle.

The notion of a protein component to cytoplasmic lipid droplets would have been absurd before 1991. Droplets were simply thought to be neutral lipid inclusions in the cytoplasm, containing mainly triacylglycerols (TGs) and steryl esters (SEs). With the shocking discoveries of a hormone-responsive protein, termed perilipin (now Perilipin 1) in the “fat cake” (the inconvenient and generally discarded top layer after centrifuging a cell extract to purify organelles) of adipocytes (Greenberg et al., 1991) and the existence of a phospholipid monolayer surrounding the droplet core (Tauchi-Sato et al., 2002), the notion of a protein component to droplets began to seem not so farfetched. Meanwhile, several proteins involved in neutral lipid synthesis or mobilization were found to associate with droplets from various cell types.

The utility of identifying the protein components of a cellular compartment in an unbiased way is hard to overstate, because it provides essential information on function. The rub with most organelles used for protein analysis has been challenges in their purification. Fortunately, lipid droplets have proven to be more facile than other organelles in that they float. There is no need for complex protocols of differential centrifugation, taking into account mass, shape, and density, to tease organelles apart from each other and minimize contamination. For droplets, one can simply subject a crude lysate to centrifugal force; the droplet will float to the top (hence a fat cake), often through a cushion of buffer above the lysate. A few washes with buffer following this to liberate contaminants that adventitiously adhere, and the purified prep is ready for direct interrogation by mass spectroscopy.

Through these simple steps, the droplet proteomes have been determined from cells representing most branches of life: eubacteria, plants, fungi, and animals. Efforts are now underway to determine differences in droplet protein composition in different tissues with a species or in cells subjected to different nutritional states. Besides the expected characters—lipases, acyl-CoA synthases, and acyltransferases—droplets also apparently house many unexpected proteins, such as redox proteins with unknown substrates, and a multiplicity of proteins with trafficking functions, such as Rabs. Although secondary purification steps can minimize artifacts of contamination from other organelles, the frequency and strength of real inter-organellar contact sites on droplets (Gao and Goodman, 2015) makes elimination of all “contaminants” very difficult, if not impossible. Indeed, virtually all lipid droplet proteomics studies have found proteins that are markers of other organelles, especially ER, mitochondria, and peroxisomes. Certainly proteins that reside in the ER lumen, such as BiP, or in the inner mitochondrial membrane, such as components of oxidative phosphorylation, are in droplet proteomes as impurities. But then, how to rule out other proteins whose localization is more ambiguous?

In this issue of Development Cell, a Resource article from Bersuker et al. (2018) adapted APEX technology to reexamine the lipid droplet proteome. In this method, sequences encoding an organellar marker are fused to those encoding a plant ascorbate peroxidase (APEX). In the presence of biotin-phenol and a pulse with H₂O₂, the peroxidase generates a short-lived biotinphenoxyl radical that biotinylates nearby proteins. The bio tinylated proteins can then be isolated on streptavidin beads and subjected to proteomics analysis. The group used a point mutant of APEX, termed APEX2, that is more stable and active within cells (Lam et al., 2015). To increase the confidence and robustness of the procedure, the group fused APEX2 to two established droplet surface proteins, the perilipin PLIN2 and the lipase ATGL (with an inactivating mutation to prevent lipolysis). Furthermore, two human cell lines were used: U2OS, an osteosarcoma line, and Huh7, a liver-derived line. Because the biotinphenoxyl radical may travel 5–10 nm from its origin on the enzyme into the surrounding cytoplasm, the group purified droplets after subjecting a crude lysate to the peroxidase reaction, to avoid cytosolic contamination, before the addition of streptavidin.

As the authors hoped, the number of proteins from other organelles was greatly reduced, and the identified proteins provide a compelling dataset for exploring their targeting and functional roles. There were no ER luminal proteins identified, in stark contrast to previous analyses, and only one mitochondrial protein, the ADP/ATP carrier, was detected. It is curious that this protein is a known interloper of droplet proteomes. Could it actually have a role closer to droplets? Interestingly, several proteins were identified in this analysis that were missed by earlier approaches of simply subjecting droplets directly to mass spectrometry, indicating that the streptavidin affinity purification step allowed proteins of low concentration to be detected. Several enzymes of TG and SE synthesis and lipolysis were observed on the droplet proteome for the first time from both cell lines, and four novel redox enzymes were also detected. In addition to these proteins, Bersuker et al. (2018) add to the observation of the association of multiple Rab proteins with the droplet proteome, first observed by two groups in 2004 (Brasaemle et al., 2004; Liu et al., 2004) and a long-standing mystery. The analysis by APEX technology found 26 Rabs in the droplet proteomes in both cell lines (and an additional 10 found exclusively in UY2OS or Huh7 cells). Certainly this is highly significant. The simplest interpretation is that droplets are the depot for this class of small GTPases, which are then recruited to other membranes as required. However, in light of all of the droplet inter-organellar contacts that have been uncovered, is it possible that droplets play a more intimate role in vesicular trafficking by directly donating Rabs to the target membranes where they are needed?

A controversy in the field is the role of lipid droplets in the unfolded protein response. Past proteomics studies have shown that proteins in the endoplasmic-reticulum-associated degradation (ERAD) pathway colocalize with droplets, and several players in protein degradation are similarly found in the new study. From their proteomic data, Bersuker et al. (2018) identified a droplet protein with a short half-life, c18orf32, and elegantly showed that its degradation actually occurred in the ER upon exiting the droplet. With these findings, the role of droplet-localized proteins like VCP/p97 (an ATPase that extracts proteins in ERAD) and several members of the ubiquitin protein-degrading system await future resolution.

Finally, we must ask how all of these newly discovered and verified proteins target to the droplet. In this same issue of Developmental Cell, work from Prévost et al. (2018) examining the mechanism of protein targeting to lipid droplets suggests that the droplet surface is quite different from that of a normal bilayer in that the underlying neutral lipid might be transiently exposed to the surface, increasing the surface tension. They find that hydrophobic regions of proteins may have increased affinity to this surface, thereby lowering the surface tension. They further show that bulky hydrophobic residues are critical to the targeting of amphipathic helices. Hydrophobic hairpins, reminiscent of plant oleosins, established shortly after the discovery of perilipin (Tzen et al., 1992) are also important targeting elements to droplets. Given the prevalence of amphipathic helices with bulky hydrophobic residues to proteins that target to other organellar membranes, the authors also make the conclusion that rather than there being specific mechanisms to target proteins to lipid droplets, it is likely that mechanisms of specificity are in place to prevent promiscuous protein association with lipid droplets. In light of the significantly refined proteome now established by Bersuker et al. (2018), it will be interesting to test this hypothesis, as well as explore how many of the droplet proteins contain these established targeting domains, determine how many are bound to the droplet indirectly through protein-protein interactions, and further build our understanding of the assembly and function of this fascinating and essential organelle.