Background

Lipid droplets (LDs) are organelles composed by a neutral lipid core surrounded by a phospholipid monolayer where LD proteins are inserted (Tzen et al., 1993). LDs were first described as stable compartments for storage of high-energy lipids in specialized tissues. However, recent reviews describe LDs as highly dynamic organelles present in most cell types, with functions that transcend mere energy storage (Shimada et al., 2018; Huang, 2018; Shao et al., 2019).

In plants, early studies on LDs were conducted mostly in oilseeds, where they are very abundant and act as energy source in the post-germination stage, before photosynthetic capacity is acquired (Yatsu and Jacks, 1972; Huang, 1996; Tzen et al., 1997). Subsequent works revealed that these organelles are also formed in pollen grains (Ischebeck, 2016), senescent leaves (Brocard et al., 2017) and pathogen-infected tissues (Shimada et al., 2014; Fernández-Santos et al., 2020), although their functions are poorly understood. It is therefore imperative to have efficient protocols for LD isolation from vegetative tissues, where their abundance is comparatively low.

LDs are the least dense cellular organelles, hence, most isolation protocols are based on their capacity to float after centrifugation. Tissue is homogenized in a suitable buffer, and then the extract is clarified and centrifuged with an overlay of a less dense buffer (floating buffer). Upon centrifugation, LDs float from the extraction buffer and through the floating buffer, to accumulate in a fat pad visible on top of the tube.

The choice of floating and extraction buffers poses a dilemma to the experimenter. Their stringency and the number of centrifugations will establish the balance between obtaining cleaner LDs, but devoid of some loosely bound, bona fide LD proteins, or more “complete” LDs with carryover contaminants attached during preparation (Huang, 2018). Early protocols from oilseeds often included harsh buffers, containing detergents, organic solvents, chaotropic agents or high salt concentrations (Tzen et al., 1997; Jolivet et al., 2004; Buchanan-Wollaston et al., 2005; Katavic et al., 2006). As LDs are very abundant in this tissue, such conditions guaranteed high yields of pure LDs where core proteins were identified (Tzen et al., 1990; Chen et al., 1999; Lin et al., 2002).

In order to identify less tightly bound LD proteins, later protocols designed for proteomics applications have tended to less stringent conditions, as in Horn et al. (2013) and Brocard et al. (2017). In these cases, an enrichment factor was calculated for each protein between LDs and other fractions to rule out contaminants. “True” LD proteins were thus defined as those above an arbitrary threshold. However, proteins with dual localization can be missed with this strategy, especially if the studied condition is dynamic and LD formation is transient. Unfortunately, there is no perfect negative control, and “true” LD components will always need to be confirmed by complementary methods such as microscopy. Another problem in previous protocols from leaves has been the large amount of sample needed for extraction, due to lower LD abundance in this tissue compared with seeds or fruit mesocarp (Brocard et al., 2017).

Based on this background, we developed an LD isolation protocol from Arabidopsis leaves, with the following aims: i) simplicity ii) maximization of LD yield and iii) low to medium stringency in order to identify loosely bound LD proteins while minimizing contaminants. Briefly, we extract LDs in a 0.6M sucrose buffer with 0.1% Tween-20 followed by two successive flotations in the same buffer with less sucrose (0.4 M and 0.2 M). The amount of Tween-20 can be variable and is critical in two aspects: increasing stringency and LD yield. The other critical point in our hands was to use fresh tissue, as yield and integrity of LDs is affected when extracted from frozen samples. We obtained approx. 1,700 LDs/mg fresh tissue using 6-week-old wild type plants and were able to identify and confirm known LD proteins such as CLO3 and α-DOX1, as well as proteins newly associated to LDs in plants such as GPAT4, GPAT8 and PAD3 (Fernández-Santos et al., 2020). LD quality can be checked by microscopy or flow cytometry. By using transgenic CLO3:GFP Arabidopsis plants, we were also able to follow the isolation process by western blot using an anti-GFP antibody.