The Biology of Lipid Droplet-bound Mitochondria

Abstract

Proper regulation of cellular lipid storage and oxidation is indispensable for the maintenance of cellular energy homeostasis and health. Mitochondrial function has been shown to be a main determinant of functional lipid storage and oxidation, which is of particular interest for the adipose tissue as it is the main site of triacylglyceride storage in lipid droplets (LDs). Recent studies have identified a subpopulation of mitochondria attached to LDs, peridroplet mitochondria (PDM) that can be separated from cytoplasmic mitochondria (CM) by centrifugation. PDM have distinct bioenergetics, proteome, cristae organization and dynamics that support LD build-up, however their role in adipose tissue biology remains largely unexplored. Therefore, understanding the molecular basis of LD homeostasis and their relationship to mitochondrial function and attachment in adipocytes is of major importance.

Introduction

Proper regulation of lipid storage and utilization is critical to maintain cellular energy homeostasis and health. In this context, the adipose tissue is the main energy storing tissue in mammals, where energy derived from nutrients is stored in form of triacylglycerides (TAGs) in lipid droplets (LDs). In addition to its function in lipid storage, distribution and oxidation, the adipose tissue has an important role as an endocrine organ [1]. Disorders of the adipose tissue can lead to numerous diseases, from obesity, type 2 diabetes to lipodystrophy and cachexia. Importantly, excessive lipid storage and lack of lipid storage both can lead to lipotoxicity [2–4]. Although generally higher body mass index (BMI) and adiposity is associated with reduced insulin sensitivity, being obese is not sufficient to induce insulin resistance. Studies have shown a significant fraction of obese individuals do not develop hyperglycemia or hyperlipidemia [5]. Moreover, there are lean individuals with lipodystrophy due to dysfunctional adipocytes, who develop severe insulin resistance [6]. Consequently, white adipose tissue dysfunction rather than just expansion, might be the major pathogenic event leading to insulin resistance. In this regard, multiple studies have found that mitochondrial function is a main determinant of healthy adipocyte function both in lipid storing and as an endocrine organ [7–9]. Therefore, understanding the molecular basis of LD homeostasis and their relationship to mitochondrial function in adipocytes is of major importance. This review aims to discuss our current understanding of the role of LDs and mitochondria in adipose tissue and whole body metabolism.

Adipose Tissue Lipid Storage and Endocrine Function

There are two major types of adipose tissue in mammals, the white adipose tissue (WAT) and the brown adipose tissue (BAT). The majority of adipose tissue in adult humans is WAT, however under certain conditions there is an increase in BAT depots and the biogenesis of other thermogenic adipocytes called beige or brite adipocytes. Detailed discussions of the different types of adipocytes, their cell lineage, and differentiation were published in recent reviews [10,11].

Apart from its function as lipid storing tissue, WAT has an important role as an endocrine organ, secreting adipokines such as leptin and adiponectin, involved in whole body metabolic regulation [1]. Similar to WAT, BAT also serves as lipid storing and endocrine tissue, however in addition to that, BAT plays a crucial role in adaptive thermogenesis. In this sense, stimuli such as cold exposure, stress or certain diets will increase circulating norepinephrine levels, which then activate the β3-adrenergic receptors on BAT, promoting an increase in nutrient oxidation and the generation of heat [12].

Being the main tissue for lipid storage, the adipose tissue is specialized in storing and releasing fatty acids upon stimulation. Thus, the adipose tissue plays a central role in controlling whole body lipid levels, and protecting from toxicity induced by high levels of non-esterified fatty acids (NEFA), namely lipotoxicity [2]. Under conditions such as starvation, adipose tissue lipolysis is activated and fatty acids are released to other tissues for the generation of ATP by fatty acid oxidation (FAO) [13,14]. This process needs to be highly regulated, as uncontrolled release of NEFA can, not only be detrimental to surrounding tissues, but can also cause damage to the adipocyte itself. Therefore, in order to reduce circulating fatty acids and prevent damage such as ER stress, part of these liberated fatty acids are re-esterified, forming a futile energy consuming cycle. This process was shown to be dependent on DGAT1, as adipocyte specific knock-out of DGAT1 resulted in increased levels of circulating fatty acids, and ER stress in adipocytes [15]. White adipocyte mitochondria provide the ATP required for fatty acid esterification and release, as shown by decreased fatty acid release induced by oligomycin and the absence of adipose tissue expansion in mice with deficiencies in mitochondrial oxidative function [9,19–21]. Therefore, in white adipocytes, mitochondria are expected to play an integral role in the futile cycle of lipolysis and fatty acid re-esterification [16], as the ability of mitochondria to oxidize fatty acids is blocked in mature white adipocytes [17,18]. In contrast, brown and beige adipocyte mitochondria are specialized in oxidizing fatty acids to generate heat via UCP1-mediated and other mechanisms [22–24], thereby reducing susceptibility to lipotoxicity Here, NEFA play a role activating uncoupled respiration and fatty acid oxidation [25].

Mitochondria in Adipose Tissue

Mature white adipocytes typically consist of one large LD, with a diameter in the 100 μm range (unilocular adipocytes), which covers most of the cytoplasm (Figure 1D). The characteristic unilocular phenotype is mainly observed in in-vivo differentiated adipocytes, and difficult to reproduce in commonly used adipocyte cell lines such as 3T3-L1 cells (Figure 1B). Numerous studies have observed various levels of heterogeneity among white adipocytes, both in terms of their LD size and composition and function. Detailed reviews on this topic were published recently [26,27]. Compared to brown adipocytes, white adipose tissue and adipocyte cell lines have significantly fewer mitochondria (Figure 1). However, during WAT differentiation mitochondrial biogenesis plays an important role, as mitochondria are critical to meet the ATP demand of the differentiation process and provide substrates for lipogenesis. Interestingly, terminal differentiation to mature white adipocytes is associated with removal of mitochondria by autophagy [28– 30]. Accordingly, adipose-specific knock-out of autophagy gene Atg7 leads to a concurrent impairment of the adipocyte differentiation program and accumulation of mitochondria, as well as showing small LDs similar to immature white adipocytes [30]. The fact that mature WAT removes mitochondria generated during differentiation seems to suggest that mitochondrial function needs to be limited in mature WAT. However, several studies have demonstrated the importance of mitochondrial function in WAT physiology [7,9,17,19,29,31,32]. Indeed mitochondrial function was found to be crucial for adiponectin production [33]. Furthermore, adipocyte specific knock-out of TFAM, a key regulator of mitochondrial transcription, resulted in severe lipodystrophy, insulin resistance and hepatosteatosis [9]. Supporting the contribution of decreased mitochondrial function in WAT to metabolic diseases, there seems to be a negative correlation between mitochondrial mass in the WAT and obesity. In this context, white adipocytes from ob/ob mice had significantly lower mitochondrial mass and oxygen consumption rates compared to lean control mice, which could partially be improved by PPARgamma agonist rosiglitazone [31].

Figure 1. Mitochondria and Lipid Droplets in white and brown adipocytes:

(A) Cultured primary mouse brown adipocyte differentiated for 7 days. Mitochondria were stained with MitoTracker Deep Red FM (red) and lipid droplets were stained with BODIPY 493/503 (green). Zoom-in box illustrates peridroplet mitochondria. N denotes nucleus. (B) Cultured 3T3-L1 cells differentiated for 15 days. Cells were fixed in 4% PFA and mitochondria were stained with a primary antibody rabbit-anti-grp75 (green) and lipid droplets with goat-anti-Plin1 (red). Secondary antibodies used were donkey-a-rabbit-Alexa488 and donkey-a-goat-Alexa568. Zoom-in box illustrates peridroplet mitochondria. N denotes nucleus. (C) Mouse brown adipose tissue fixed and imaged as whole mount. Tissues were fixed in 4% PFA, mitochondria were stained with a primary antibody rabbit-anti-grp75 (red), and lipid droplets with goat-anti-Plin1 (green). Secondary antibodies used were donkey-anti-rabbit-Alexa568 and donkey-anti-goat-Alexa488, nuclei were visualized with DAPI. Zoom-in box illustrates peridroplet mitochondria. (D) Mouse gonadal adipose tissue fixed and imaged as whole mount. Tissues were fixed in 4% PFA, mitochondria were stained with a primary antibody rabbit-anti-grp75 (red), and lipid droplets with goat-anti-Plin1 (green). Secondary antibodies used were donkey-anti-rabbit-Alexa568 and donkey-anti-goat-Alexa488, nuclei were visualized with DAPI. Zoom-in box illustrates peridroplet mitochondria. All scale bars = 20 microns; Zoom-in box scale bars = 1 micron. All images are unpublished data (Anton Petcherski and Michaela Veliova).

With a remarkable resemblance to immature white adipocytes [14], mature brown adipocytes are characterized by numerous small and medium sized LDs, with diameters ranging from nm range to around 15 μm (multilocular adipocytes), which are surrounded by numerous mitochondria (Figure 1 A and C) [34,35]. Studies observed that BAT undergoes “whitening”, a process that is accompanied by reduced mitochondrial mass and a decrease in the number of lipid droplet, becoming unilocular [36]. Conversely, the process of “browning” is typically associated with an increase in mitochondrial mass and activity, the expression of thermogenic markers such as UCP1 and Prdm16, and the number of lipid droplets per adipocyte in white adipose tissue, which leads to improvements in whole body metabolism [37]. Remarkably, rosiglitazone was shown to increase both fatty acid oxidation and lipolysis/re-esterification in white adipose tissue, before browning is found [31]. Therefore, one could propose that, in white adipose tissue, the fatty acid oxidation stimulation induced by rosiglitazone is confined to browning of white adipocytes, while mature white adipocytes only showed enhanced re-esterification (not browned).

Both WAT and BAT are innervated by the sympathetic nervous system (SNS), and their activation is dependent on hormonal stimulation by the SNS [38]. In this context, mitochondria play an indispensable role in adipose tissue thermogenesis, a process enabled by the uncoupling protein 1 (UCP1), a mitochondrial protein expressed exclusively in thermogenic adipocytes. UCP1 uncouples mitochondrial nutrient oxidation from ATP synthesis to generate heat [22,23]. UCP1-mediated thermogenesis is activated after stimulation of β3-adrenergic receptors by Norepinephrine released from sympathetic neurons, which induce fatty acid release from LDs and mitochondrial fragmentation in a PKA-dependent manner [12,39]. Mitochondrial fragmentation and the fatty acids released activate UCP1 to mediate uncoupling. Activated BAT elevates its metabolic activity, causing an increase in glucose uptake as well. Thus, active BAT is detectable in mice and humans using positron emission tomography–computed tomography (PET-CT) to monitor 18flurodeoxyglucose (18FDG) uptake, this glucose uptake being reduced with aging or obesity [40–44]. In this context, a recent study showed that mitochondrial lipoylation is reduced in aged BAT and serves likely as a key driver in age-related decline in BAT thermogenic activity [45]. In addition to an increase in mitochondrial oxygen consumption upon BAT stimulation, mitochondrial dynamics play a central role in BAT thermogenic capacity. Wikstrom et al. showed that upon adrenergic stimulation BAT mitochondria undergo PKA mediated fragmentation by activating the mitochondrial fission protein Drp1 [46]. Moreover, the study showed that Drp1 activation and mitochondrial fragmentation are necessary for BAT thermogenesis using a dominant-negative form of Drp1 [46]. Furthermore, our recent study has shown a tight interaction between mitochondria and LDs, and that BAT mitochondria detach from LDs upon cold exposure [35].

What are peridroplet mitochondria and why do we need them?

Various studies have observed a tight or transient (also known as “kiss-and-run”) interaction between LDs and mitochondria [35,47–53]. Interestingly multiple groups observed an increase in mitochondria-LD interaction upon conditions of starvation and were thus hypothesized to be a mechanism for increasing mitochondrial fatty acid oxidation [48,50,51] (Figure 2B). However, only recently it was shown that mitochondria attached to LD have distinct bioenergetics, proteome, cristae organization and dynamics compared to cytoplasmic mitochondria (CM) (Figure 3). These mitochondria, termed peridroplet mitochondria (PDM) were successfully isolated from mature BAT by utilizing high-speed centrifugation to separate mitochondria from their associated lipid droplets [35]. Benador et al. show that that PDM have higher capacity to oxidize pyruvate and malate, but lower capacity for fat oxidation, compared to CM. Interestingly, PDM were shown to have a higher ATP synthesis capacity and higher expression levels of ATP synthase [35]. ATP synthesis in PDM enabled esterification of lipids and expansion of LDs (Figure 2A). Furthermore, it was found that PDM differ from CM in their dynamics, as they do not fuse and thus do not share content with their cytoplasmic neighbors, despite higher expression of mitochondrial fusion protein Mfn2 [35]. However, due to their attachment to LDs, their motility is significantly lower, which is likely the reason for their reduced fusion activity [35]. Data in primary brown adipocytes using overexpression models of a PDM tethering protein suggest that forcing mitochondria to attach to the LD may be sufficient to give them the unique bioenergetics characteristics associated with PDM [35]. However, it remains to the determined how LD attachment “transforms” mitochondria into PDM.

Figure 2: Two proposed roles for mitochondria association with lipid droplets.

(A) Model 1: Peri-Droplet Mitochondria (PDM) support LD expansion. PDM have a high capacity to synthesize ATP which fuels TAG synthesis for LD expansion. In Model 1 PDM feed the LD. (B) Model 2: Association of PDM with the lipid droplet facilitate trafficking of fatty acids from LDs to mitochondria. In Model 2 LD feed the PDM.

Figure 3: Peridroplet Mitochondria (PDM) and Cytoplasmic Mitochondria (CM):

PDM have a higher capacity to oxidize pyruvate and synthesize ATP. They provide ATP for TAG synthesis and thereby promote lipid droplet buildup. CM have a high capacity to oxidize fatty acids. The two mitochondrial populations do not fuse with each other.

Despite the role of PDM in providing ATP for lipid synthesis in brown adipocytes, little is known about their function in other tissues, including white adipocytes. However, the finding that at least two types of mitochondria co-exist within a cell provides a mechanism of how seemingly antagonistic metabolic programs, such as lipogenesis and fatty acid oxidation can occur at the same. Thus, PDM may be particularly important for BAT and beige adipocytes, as this tissue needs to manage storage and oxidation of fatty acids, often at the same time.

Other metabolically highly active tissues could also benefit from having a population of mitochondria capable of assisting FFAs esterification. In cardiac tissue, conditions that induce a rapid increase in lipid droplet mass such as fasting or β-agonist treatment invariably led to induction of Plin5 expression and increase in mitochondria-lipid droplet contact sites [54]. PDM function may also protect cells from oxidative stress, through a yet unknown mechanism. In the HepG2 liver cell line, Tan et al., demonstrated that PLIN5 protein expression and the incidence of PDM was increased upon the induction of oxidative stress. Moreover, PLIN5 overexpression conferred protection from H₂O₂ induced apoptosis leading to the speculation that PDM can confer protection from oxidative stress [55]. Interestingly, the transfer of anti- and pro-apoptotic factors from the outer mitochondrial membrane to the lipid droplet membrane has been previously shown to be part of the stress response in yeast and mammalian cells [56].

The possible role of PDM in other tissues such as muscle and liver and cancer were recently reviewed [57].

PDM visualization and isolation methodologies

While previous studies have observed mitochondria associated with LD, the type of visualization and quantitation varied greatly between these studies. Therefore, this section of the review focuses on the various methodologies to visualize and quantify PDM association, and furthermore discuss published isolation protocols and their limitations. Noteworthy is that the different analyses and isolation methods may detect different types of interactions of PDM to LDs. Different levels of strength of association (permanent vs transient) are detected with varying isolation and detection methods, which are outlined below, and might explain why different groups propose opposing roles for PDM (Figure 2). However, it remains to be determined whether the type of strength of PDM interaction with LD corresponds to a different biological function.

Electron Microscopy

The first observations of mitochondria-LD interactions were made using electron microscopy (EM) by Palade in 1995 [58]. Indeed EM images provide the highest resolution, and multiple groups have observed seemingly tight interaction of mitochondria and LD using this method [35,47–49,59,60]. This tight interaction can be seen as an electron dense region that connects the PDM and LD [35,48,49,59]. A clear advantage of this method is the high resolution of EMs, which allows the detection of PDM even in cells that do not have many or big LDs [49,61]. Furthermore, EM images can give interesting information on cristae arrangement in PDM. In that regard, several groups observed a perpendicular arrangement of mitochondrial cristae towards the LD [35,48]. However, the biological significance of this observation remains to be determined.

Live-cell confocal imaging

A variety of studies have observed mitochondria LD interaction using live-cell fluorescence imaging [35,47,51,62,63]. A clear advantage of live-cell imaging to monitor mitochondria-LD interactions is the real-time component. Along with other organelles in the cells, both mitochondria and LDs are in constant movement and so too might be their interactions. A life-cell imaging approach thus can provide important information on the nature of PDM interaction; whether these are permanent or transient interactions. Furthermore, responses to pharmacological interventions can be studied more easily using a life-cell imaging approach. Recent studies have described approaches for spatial life cells imaging with novel analysis approaches that provide high-resolution information and a more accurate assessment of organelle interactions. [64,65]. Improved imaging technology paired with accurate analysis platforms will likely improve our understating of the nature of PDM and their interactions with LDs.

Analyses of cell-free LD fraction

Multiple studies have isolated LD fractions by centrifugation and measured mitochondrial content in this lipid fraction by florescence microscopy [35,52], western blot analyses [47,52,53], quantitative proteomics [53,66] or with fluorescent plate reader assay (unpublished data). Using these approaches only strong interactions between mitochondrial and LD will be detected, but relatively weakly attached PDM might be missed. Furthermore, depending on the speed of centrifugation different groups may have assessed a different pool of PDM, depending on the strength of the interaction.

PDM isolation heterogeneity

Several studies have attempted to isolate PDM and separate them from the LDs using centrifugation with or without salts or detergents [35,52,53]. While Benador et al. showed successful isolation of PDM using a centrifugation of 10.000xg [35], other studies showed that centrifugation is not sufficient to strip all mitochondria from the lipid droplet, requiring proteinase treatments to isolate them [52,53]. These discrepancies could attributed to; 1. differences in the ionic strength of the buffers used for tissue homogenization 2. the type of homogenization technique (e.g. dounce homogenizer, filtration through mesh) 3. the strength of homogenization (amount of strokes used). Further work needs to be done on optimizing isolation protocols, tissue differences and properly defining the nature of the interaction between mitochondria and LDs. In addition, these differences might be suggesting an exciting aspect of peridroplet mitochondria, which is that peridroplet mitochondria are not homogeneous. It is conceivable that depending on the strength of interaction there is a subgroup of PDM specialized to provide ATP, while another subgroup could potentially be involved in lipogenesis or other processes regulating lipid metabolism.

Mechanisms for PDM interaction with Lipid Droplets

Although mitochondria and LD contact has been described in multiple tissues and conditions, the exact mechanism by which this interaction is mediated remains inconclusive. In fact, several mechanisms for mitochondria-LD tethering were proposed in the literature, opening the possibility that PDM interaction with LDs is regulated on multiple levels, or in a tissue specific manner (Table 1).

Table 1:

Mechanisms of Mitochondria-Lipid Droplet Interaction

Proposed Tether	Mechanism of Tethering	Cell Type	Technique	Interaction with ER	Effects on de novo lipogenesis and fatty acid esterification	Reference
Perilipin5 (Plin5)	C-terminus of Plin5 regulates interaction with mitochondria	CHO, AML12, HL-1, primary brown adipocytes, INS1	site-directed mutagenesis, confocal fluorescence microscopy	N/A	de novo: N/A FA esterification: increased	Wang et al., 2011; Benador et al., 2018
DGAT2	N-terminus of DGAT2	COS-7	site-directed mutagenesis, confocal fluorescence microscopy, cell fractionation	Yes	de novo: increased FA esterification: N/A	Stone et al., 2009
Mfn2	Mfn2 interaction with Plin1	primary brown adipocytes	co-immunoprecipitation	Yes	de novo: N/A FA esterification: N/A	Boutant et al., 2017
MIGA2	C-terminus of MIGA2 targets mitochondria and N-terminus targets LDs	COS-7, 3T3-L1	site-directed mutagenesis, confocal fluorescence microscopy and electron microscopy	Yes	de novo: increased FA esterification: N/A	Freyre et al., 2019

One of the proposed PDM tethering proteins is Perilipin 5 (Plin5), a member of the Perilipin family of LD-coating proteins. Plin5 was shown to co-localize with mitochondria and LDs. Overexpression of Plin5 is sufficient to promote mitochondria-LD association in multiple cell types [63,67,35]. Importantly Plin5 is also a negative regulator of lipolysis via its interaction with ATGL, which makes studies analyzing the involvement of PDM recruitment to LD expansion challenging [67]. Interestingly, Wang et al. found that a conserved sequence at the C-terminus of the protein is required for Plin5 to tether mitochondria [63]. However, expression of Plin5-mitochondria tethering sequence alone did not co-localize with mitochondria, indicating that Plin5 might form a complex with another protein that directly binds to mitochondria, or may indicate that specific protein folding is required for the tethering effect (unpublished data). A recent study showed that Plin5 phosphorylation at S155, a PKA target, induces lipolysis without detaching mitochondria from the LD in cardiomyocytes [68]. Adrenergic stimulation in brown adipocytes leads to detachment of PDM [35], however it remains to be determined if it is through PKA-mediated signaling on Plin5, or an alternative mechanism. Interestingly, Gallardo-Montejano et al. published that Plin5 translocates to the nucleus in a PKA dependent manner [69], suggesting that Plin5 translocation to the nucleus could induce transcriptional changes to maintain PDM detached from the LD.

Another proposed mitochondria-LD tether is Diglyceride-Acyltransferase 2 (DGAT2). DGAT2, together with DGAT1, is responsible for the last step in TAG synthesis. DGAT2 was shown to co-localize with LDs, mitochondria and the ER, thereby bringing these three organelles in direct contact [62]. Stone et al. showed that the N-terminus of DGAT2 contains a mitochondrial targeting sequence, which is necessary and sufficient for its mitochondrial localization. Accordingly, introducing the targeting sequence alone into Cos7 cells was sufficient to target a fluorescent probe to mitochondria. Interestingly Irshad et al. showed that, in contrast to DGAT1, which is required for TAG synthesis from circulating free fatty acids, DGAT2 was necessary for esterification of TAGs derived from de novo lipogenesis [70]. It remains to be determined, if loss of DGAT2 leads to a decrease in PDM, and if that further affects TAG synthesis from de novo lipogenesis.

Furthermore, the outer mitochondrial membrane protein Mitofusin 2 (Mfn2) has been suggested as possible tethering protein between LD and mitochondria [71], and between mitochondria and the ER [72]. Knock-out of Mfn2 in BAT leads to a 50% reduction in mitochondria-LD interaction, and Mfn2 pull-down experiments showed co-immunoprecipitation of Mfn2 with LD- coating protein Plin1 in MEFs [71]. Moreover, the outer membrane mitochondrial protein Mfn2 was recently demonstrated to participate in the transfer phospholipid precursors from the ER to mitochondria to enable phospholipid synthesis (PE) [73]. LDs have a monolayer of phospholipids to encapsulate the neutral lipids stored. Consequently, one could hypothesize that the interaction between mitochondria and LDs might allow an efficient transfer of de novo synthesized phospholipids to facilitate LD expansion. Various reviews have discussed the function of Plin5 [67,74], DGAT2 [75,76] and Mfn2 [77,78] aside from their proposed role as mitochondria-LD tethering proteins.

A recent study identified Mitoguardin 2 (MIGA2) as a tethering protein between mitochondria and LDs in white adipose tissue [47]. Detailed analysis of various truncations of MIGA2 revealed that an amphipatic sequence at the C-terminus of the protein was necessary and sufficient to mediate mitochondria-LD interaction[47]. Indeed, expression of this mitochondria-recruiting sequence was able to target a fluorescent probe to mitochondria [47]. In addition, fusion of this sequence to Calnexin, an ER protein that does not co-localize with mitochondria, was sufficient to mediate Calnexin-mitochondria interaction [47]. Interestingly, the study supports that MIGA2 acts as a connection between mitochondria, the ER and LDs, similarly to DGAT2 [47,62]. MIGA2 interaction with the ER was mediated by a FFAT-motif binding to VAP-A/B in the ER. While MIGA2-mediated tethering promotes de novo lipogenesis, the role of MIGA2-mediated PDM recruitment in the esterification of pre-existing fatty acids into triglycerides remains to be determined [47]. On the other hand, Plin5-mediated PDM recruitment increased TAG esterification from pre-existing fatty acids, but whether Plin5 increases de novo lipogenesis remains to be determined [35]. The observation that DGAT2, a PDM tethering protein, was shown to be required for TAG synthesis coming from de novo lipogenesis [62,70], similar to MIGA2 [47], as opposed to PDM recruitment by Plin5, which promoted esterification of external fatty acids [35], suggests that different tethers might assign different functions to PDM. Since two of the proposed tethers for PDM are also tethering the ER to the LD, and the ER is known to be the site of LD-budding [79,80], it is interesting to speculate that PDM are involved in the generation of new LDs, or the expansion of existing LDs, or both.

The observation that MIGA2 acts as a tethering protein for PDM in WAT, a tissue that expresses little to no Plin5 [47,81], supports that PDM attachment and function might be regulated in a tissue-specific manner. Furthermore, it is conceivable that depending on the tether used, PDM could have beneficial vs. detrimental effects on systemic lipid metabolism. Additionally, it is possible that certain tethers like MIGA2 are expressed during cell differentiation to aid in this process, whereas others, like Plin5 or DGAT2 are needed for metabolic regulations in the fully differentiated state.

Potential for PDM as therapeutic target

Evidence that PDM can be beneficial (Table 2)

Table 2:

Benefits and Pathologies related to PDM

Tissue/Cell type	Model	PDM increase or decrease	Observation	Reference
Whole body	DGAT2 KO	N/A	Lethal due to Lipopenia	Stone et al., 2004
Heart	Plin5 KO	Decreased (little to no lipid droplets)	Reduced lipid storage and increase in fatty acid oxidation; Increased sensitivity to ischemic injury by lipid induced oxidative stress; Aggravated myocardial hypertrophy in the model of heart failure	Kuramoto et al., 2012 Drevinge et. al., 2016; Zheng et al., 2017; Wang et al., 2019
Heart	Plin5 OE	increased	Increase in TAG content and reduced expression of FA oxidizing genes; cardiac steatosis with only mild impairment to cardiac function	Pollak et al., 2013; Wang et al., 2013
Hepatic stellate cells (HSC)	Plin5 OE	N/A	HSC lower Plin5 expression upon activation; exogenous expression of Plin5 lowers oxidative stress and inhibits HSC activation	Lin et al., 2016;
Hepatocytes	Plin5 KO	decreased	Reduced insulin sensitivity, reduced fatty acid uptake and storage	Keenan et al., 2019
Hepatocyte	Plin5 OE	increased	Resistance against H2O2 induced oxidative stress	Tan et al., 2019
Pancreatic β-cells	Plin5 OE	N/A	Improved glucose tolerance; Protection from lipotoxicity and ER stress	Trevino et al., 2015; Zhu et al., 2019
Brown adipose tissue	Mfn2 KO	decreased	Impaired cold tolerance, but increased resistance to HFD induced obesity in female mice; increased mitochondrial capacity to oxidize fatty acids	Boutant et al., 2017; Mahdaviani et al., 2017
Skeletal muscle	Plin5 OE	N/A	Increased LD size and improved mitochondrial function while maintaining insulin-sensitivity; Protection against lipotoxicity	Bosma et al., 2013; Laurens et al., 2016

Previous publications suggest that one important role of PDM is to provide ATP for TAG synthesis and LD expansion [35]. Although increasing PDM content may not be a good approach to increase fat consumption, it may represent an approach to prevent lipodystrophy and/or lipotoxicity by securing free fatty acids into TAG (Figure 4). Indeed, TAG accumulation was shown to protect from FFA-induced lipotoxicity by channeling palmitate into neutral lipid pools [3]. Furthermore, ectopic accumulation of LDs in skeletal muscle is not only associated with increased body weight and type 2 diabetes, but, paradoxically, also observed in endurance athletes, further suggesting that TAG accumulation can have positive effects [82,83].

Figure 4: Possible role of PDM and CM in protection from lipotoxicity.

Both PDM and CM contribute to the removal of FFA and the prevention of lipotoxicity. While PDM support removal of FFA into storage, CM consume FFA as a fuel source. When lipids are not consumed or stored, lipotoxicity can impair mitochondrial function.

While there are only few relevant studies assessing the role of PDM in adipose tissue to date, we can gain some ideas on the role of PDM in fat storing cells from other cell model and over-expression models. Most of our understanding of the possible role of PDM in protecting from lipotoxicity comes from studies using Plin5 overexpression (OE) models, or Plin5 knock-out (KO) models. Interpreting these studies in the context of PDM comes with various caveats, as Plin5 OE will not only induce PDM attachment, but also inhibit lipolysis through its function inhibiting ATGL [67]. Results from studies using Plin5 KO models might also be difficult to interpret, as these cells or mice usually have uncontrolled lipolysis, which could come from lack of Plin5 regulating ATGL, or a lack of PDM, or a combination thereof. Keeping these caveats in mind there are still some interesting things to learn about PDM from Plin5 KO/gain-of-function models.

In agreement with the hypothesis that PDM are essential to protect from lipid induced damage, Plin5 KO in cardiac muscle caused increased sensitivity to ischemic injury [84–86], while Plin5 gain-of-function did not cause major cardiac damage, and was even cardio-protective despite LD accumulation [87,88]. In a recent study, Du and colleagues found endogenous Plin5 expression to be increased in cardiac epithelial cells in response to gluco-lipotoxicity, conferring a protective effect against FFA-ROS mediated microvascular injury. On the other hand, this type of injury was exacerbated by Plin5 KO in a model of type-2-diabetes with high circulating FFAs and in models of heart failure [89,90]. Furthermore, Plin5 deletion in hepatocytes led to reduced insulin sensitivity, which was associated with decreased fatty acid uptake and storage [91]. Remarkably, PDM may be linked to inflammatory processes in the liver, as Plin5 OE was shown to restore quiescence in hepatic stellate cells in a liver fatty acid-binding protein dependent manner [92–94]. Endogenous Plin5 expression in pancreatic islets was stimulated upon fasting induced increases in serum FFAs, supported post-prandial insulin secretion, and Plin5-OE improved glucose tolerance [95]. Moreover, overexpression of Plin5 in INS1 pancreatic β-cells protected from palmitate induced lipotoxicity and ER stress [96]. Plin5 expression was found to be elevated in endurance conditioned human muscle, increasing metabolic flexibility, insulin sensitivity, and protection against lipotoxicity, which can also be recapitulated by Plin5 OE further implicating PDM involvement in the metabolic health of the whole body and skeletal muscle [97,98]. Furthermore, Gemmink et al. found that fasting induced endogenous Plin5 expression correlated with increased insulin sensitivity, and LDs generated de novo in the fasting phase were exclusively associated with Plin5, suggesting PDM as the essential protective mechanism against lipotoxicity in skeletal muscle [99].

Although it may seem counterintuitive to increase LD content as a therapy for diabetes, studies using the insulin sensitizers Thiazolidinediones (TZDs) suggest that increased adiposity is not in conflict with improved insulin sensitivity [100]. TZDs are PPARgamma agonist that show very promising results in improving diabetic insulin sensitivity, despite their stimulatory effect on adipocyte differentiation and expansion. Indeed, Rosiglitazone has been shown to briten human WAT ex vivo in part through stimulation of PLIN5 expression, and the formation of small, satellite, PDM coated LDs [101]. These studies are in favor of increasing PDM content to promote LD synthesis, as they suggest that an increase in overall fat mass may not be the problem, if the fat is partitioned to LD, and thereby stored in a safe place.

Evidence that detaching PDM from the LD can be beneficial (Table 2)

Previous studies show that upon adrenergic stimulation, BAT PDM detach from the LD, and that mitochondria undergo fragmentation [35,46]. While mitochondrial fragmentation is necessary for Norepinephrine-stimulated energy expenditure, it remains to be determined, whether PDM detachment is necessary as well, in order to reach maximal thermogenic capacity. In addition, the observation that OCR upon adrenergic stimulation are not completely insensitive to the mitochondrial ATP synthase inhibitor oligomycin, may suggest that even under adrenergic stimulation there are PDM that remain attached, and possibly play a role in adrenergically stimulated lipid cycling. Detaching PDM from LDs may represent a promising approach to increase fat oxidation and energy expenditure, particularly in WAT or BAT. However, it is important to note that this strategy might result in accumulation of toxic free fatty acids, if it is not concurrent to UCP1 activation or increased mitochondrial fat oxidation. Although a direct assessment of the toxic effect of lack of PDM remains to be established, we can infer interesting information from studies using PLIN5 or DGAT2 KO models. In this context, DGAT2 KO mice die shortly after birth, due to severe lipopenia, which could not be rescued by OE of DGAT1 [102]. Even though the exact mechanism remains unclear, it is conceivable that part or the mechanism, by which DGAT2 KO is lethal, is through a lack of PDM. Possibly DGAT2 KO are void of PDM and are unable to esterify fatty acids, leading to exaggerated lipotoxicity.

Nonetheless, acute detachment in brown adipocytes, such as under Norepinephrine stimulation, could represent a valuable tool to promote fat oxidation. In this context, BAT-specific knock-out of Mfn2 resulted in resistance to diet induced obesity, and increased capacity of mitochondria to oxidize fatty acids in female mice, despite impaired cold tolerance [103]. Given that Mfn2 KO in these mice also showed reduced interaction of mitochondria and LD, this may suggest that PDM detachment contributes to resistance to HFD and further suggests that PDM are necessary for regulation of inducible thermogenesis [71,103].

Concluding Remarks

We propose that understanding the regulation of the interaction between mitochondria and LDs in adipose tissue, as well as its role in lipid metabolism and homeostasis might provide novel opportunities to prevent lipotoxicity and the ensuing metabolic complications and present an avenue to fine-tune energy expenditure.