The HepG2 cell line was obtained from the American Type Culture Collection (ATCC HB-8065). Cells were cultured in Dulbecco's modified Eagle's medium (DMEM 11965-09; Thermo Fisher Scientific), supplemented with 10% fetal bovine serum (FBS), 1% penicillin-streptomycin antibiotics (Biological Industries) and maintained at 37 °C with a 5% CO₂ atmosphere. Hepatocytes were exposed to 200 μM OA (P1383, Sigma-Aldrich) or 200 μM PA (P0500; Sigma-Aldrich) conjugated to fatty acid-free bovine serum albumin (BSA) (A8806; Sigma-Aldrich) for 2, 6, 18 and 24 h, as previously described (18). Cells treated with BSA were used as controls.

Cell protein extracts were obtained using T-PER lysis buffer (78510, Thermo Fisher Scientific), supplemented with protease (#04693159001, Roche) and phosphatase (#04906845001, Roche) inhibitors. Homogenates were centrifuged at 14,500 G for 10 min, and the supernatants were collected for measuring protein concentrations with the BCA protein assay (23227, Thermo Fisher Scientific). Protein extracts of 50 μg were resolved in 10% SDS-polyacrylamide gels, transferred onto PVDF membranes previously activated with methanol, and blotted overnight with primary antibodies at 4 °C. The antibodies and dilutions were: anti-PLIN2 (1:1000; #MA5-24797, Thermo Fisher Scientific), anti-PLIN5 (1:1000; #GP31, Progen), and anti-MFN2 (1:1000; #ab50838, Abcam). The membranes were then incubated for 1 h with the following secondary HRP-coupled antibodies: anti-mouse (1:5000; #402335, Calbiochem), anti-guinea pig (1:5000; #ab6908, Abcam) or anti-rabbit (1:5000; #401315, Calbiochem) and revealed with the EZ-ECL detection kit (DW1029, Biological Industries). A C-DiGit® Blot Scanner and the Image Studio software version 3.1 (LI-COR) were used for image acquisition and densitometric analysis. GAPDH (1:5000; #MAB374, Sigma-Aldrich) was used as a loading control.

Intracellular ATP content was determined using the luciferin/luciferase-based ATP detection kit CellTiter-Glo Luminescent Cell Viability Assay (Promega) following the manufacturer's instructions as described in (19). Briefly, HepG2 cells were cultured in 96-well Petri dishes and washed 3 times with PBS before incubation with the reagent. Sample luminescence was quantified in a Synergy 2 microplate reader (BioTek Instruments). Data were normalized as fold of changes over control. Treatment with oligomycin (5 μg/mL) for 3 h was used as a negative control.

Cells were seeded in 60 mm Petri dishes at 80% confluence and treated according to the experiment. After the different treatments, measurements were performed as previously described (20–22). In brief, cells were washed with PBS, trypsinized for 3 min, and centrifuged at 200 G for 5 min. Then, the cells were resuspended in PBS and placed in the chamber of a Clark electrode (Oxygraph Plus, Hansatech). Basal respiration, proton leak, or ATP-unlinked respiration (Oligomycin, 400 μM) and uncoupled respiration (FCCP 20 μM) were measured sequentially for 3 min. Data obtained were standardized to basal control respiration.

Cells were plated on 12 mm Petri dishes and treated according to the experiment. To label LDs and mitochondria, cells were stained with BODIPY 493/503 (2 μM; D3922, Invitrogen) and MitoTracker Orange (400 nM; M7510, Invitrogen) for 25 min at 37 °C. Then, the cells were fixed with 4% paraformaldehyde and 0.01% Hoechst in PBS, and mounted in Dako Fluorescence Mounting Medium (S3023, Dako-Agilent). Fixed cells were imaged using a Nikon C2 Plus-SiR confocal microscope. 6–20 cells were registered for 3 independent experiments.

Images acquired were deconvoluted, background-subtracted, thresholded, and analyzed with ImageJ software (NIH). LDs and mitochondria number and individual volume were quantified using the 3D Object Counter plugin, as previously described (20–23). LD-mitochondria colocalization was determined within one focal plane using the JACoP plugin (20, 22, 23). The mitochondrial potential was defined in relation to MitoTracker Orange fluorescence intensity, and was analyzed within a single plane at the cell equator with the Analyze Particles function (20, 22). The mitochondrial potential of bound LD and non-bound-LD was performed by constructing a compartment consisting of 10 z-planes of the sites where mitochondria colocalize with LDs. Then, the intersected or excluded mitochondrial fluorescence of bound LD and non-bound-LD was quantified, respectively. Image intersections were obtained using the Image Calculator command of ImageJ (“AND” operator) (22).

All statistical analyses were performed using GraphPad Prism software, version 6 (San Diego, CA, USA). Data are expressed as mean ± SEM of at least three independent experiments. Data were analyzed by one-way ANOVA, and, when appropriate, comparisons between groups were performed using Tukey's or Dunnett's post-hoc tests. A two-tailed Pearson's coefficient was used for correlation analysis. Differences were considered significant at P < 0.05 (that is, a confidence level of 95%, α = 0.05).

To evaluate whether OA or PA affects the morphology of LDs, HepG2 hepatocytes were treated with OA or PA between 0-24 h. Whereas OA (Figure 1B) stimulated the appearance of big and bright droplets in hepatocytes, mainly in the central region of the HepG2 cells, PA triggered the emergence of evenly distributed smaller droplets. To quantify this effect, LD morphology was also assessed according to the number and volume of individual isolated elements through 3D reconstitution of confocal stacks (20–23) and the total fluorescence of BODIPY. Compared to BSA, OA significantly increased the number of LDs per cell at all the times analyzed, reaching a ~30-fold increase as early as 2 h after treatment (Figure 1C). Subsequently, OA also led to a gradual increase in the volume of the LDs, reaching ~4 and ~6-fold, compared to controls after 18 and 24 h, respectively (Figure 1D). In terms of total BODIPY fluorescence, only OA triggered a significant increase in the fluorescence after 6 h of treatment; moreover, this increase was maintained at later times of 18 and 24 h (Figure 1E). These results agree with the rapid nucleation of new LDs, which later grow over time. On the other hand, treatment with PA led to a slower increase in the number of LD per cell, which was significant (~20-fold) at 6 and 18 h of treatment (Figure 1C). In contrast to OA, PA did not change the volume of the LDs, suggesting slow nucleation of new LD without a growing phase (Figure 1D).

We next evaluated mitochondrial morphology in HepG2 hepatocytes using the mitochondrial-specific MitoTracker Orange probe (400 nM, 25 min) and 3D-reconstruction imaging (Figures 2A–C), as we have previously reported (21, 23). In contrast to the steep changes in LD morphology, OA only caused a slight but significant increase in the number of mitochondria per cell, concomitant with a decrease in their volume after 24 h, indicative of the induction of mitochondrial fission (Figures 2B,C). Meanwhile, PA treatment triggered a faster and more intense process of mitochondrial fragmentation, noticeable after 2 h, reaching a ~2-fold increase in the number of mitochondria per cell and a significant ~2-fold reduction in mitochondrial volume (Figures 2B,C). The magnitude of these differences reveals maintenance in the total size of the mitochondrial network, mainly suggesting changes in the mitochondrial fusion/fission equilibrium. Our results with PA are consistent with their reported effects on mitochondrial morphology in other cell types (24–26).

To evaluate whether the changes in organelle dynamics affect their physical coupling, we then assessed LD-mitochondria proximity, evaluated as Mander's colocalization coefficients between LD and mitochondria (20, 22, 27) (Figures 2D,E). Both fatty acids increased LD-mitochondria proximity starting at 2 h. However, OA treatment induced a significantly higher increase at 2 h (>2-fold), which later decreased to values similar to PA (~1.5-fold). Taken together, these results suggest that OA treatment triggers the early appearance of more and bigger LDs than PA, which are in closer contact with mitochondria in HepG2 hepatocytes.

PLINs are the main structural proteins of LDs, and determine their structure, metabolism, and interaction with other organelles (28, 29). While PLIN2 is ubiquitously expressed, PLIN5 is especially enriched in tissues with high levels of mitochondrial oxidation, such as the liver (29). PLIN5 reportedly regulates mitochondrial recruitment to LD, thereby regulating mitochondrial metabolism (8, 29, 30). On the other hand, MFN2 not only promotes mitochondrial fusion and boosts mitochondrial metabolism (31, 32), but also binds to PLIN1 to facilitate the LD-mitochondria interaction in adipose tissue (17, 33). Due to their importance, we measured their relative abundance in our experimental model through Western blot analysis (Figure 3A). As shown in Figures 3A,B, PA triggered a significant acute increase in PLIN2 at 6 h of treatment, without altering MFN2 or PLIN5, compared to control. On the other hand, OA increased both PLIN5 and MFN2 (Figures 3A,C,D) at later times (18–24 h). These results suggest that OA promotes PLIN5-MFN2-mediated LD-mitochondria interaction, while PA favors the presence of PLIN2 at the surface of LDs.

Mitochondria act as oxidative centers for different metabolites, such as fatty acids, which ultimately fuel a chain of redox reactions driven by oxygen-mediated oxidation. The protein complexes that catalyze this redox process (termed the electron transport chain, ETC) pump protons across the inner mitochondrial membrane. This creates an electrochemical gradient, which entails the generation of a mitochondrial transmembrane potential (Δψm). The resulting proton-motive force drives ATP production by transporting the protons back across the mitochondrial inner membrane through the ATP synthase enzyme (23, 34). Of note, this mechanism implies that ATP production dissipates the Δψm (Figure 4A). Thus, to characterize the differential effects of OA and PA treatments on mitochondrial metabolism, we analyzed three defining parameters of mitochondrial bioenergetics (34): Δψm, O₂ consumption, and ATP levels.

We first evaluated the Δψm by using the mitochondria-specific MitoTracker Orange potentiometric probe and confocal microscopy. Because PA treatment triggers mitochondrial fragmentation and a concomitant decrease in mitochondrial volume, to quantify the Δψm, we evaluated both the mean mitochondrial fluorescence and the total fluorescence per cell of MitoTracker Orange. The first fluorescence parameter addresses the bioenergetic state of individual mitochondria, and the second parameter evaluates the metabolic state of the cell as a whole. Figures 4B,C shows that both OA and PA treatments increased the mean and total fluorescence levels of MitoTracker Orange, starting at 2 h, indicating a boost in the Δψm. In the case of OA, the Δψm peaked at 6 h and remained high for all the analyzed times. On the other hand, PA treatment led to a subsequent decrease in the Δψm to baseline levels after 6 and 18 h, followed by a second rise in the Δψm, noticeable at 24 h, thus suggesting a two-phase response for this fatty acid.

Next, we assessed the O₂ consumption rate (OCR) as a measure of mitochondrial oxidative activity. Treatments with OA or PA increased basal OCR but with different increment profiles (Figure 4D). In the case of OA, the increment was slower, becoming significant at 6 h. PA triggered a faster increase in the OCR, as early as 2 h, which continued at 6 h (Figure 4D). We also measured OCR in the presence of oligomycin, an inhibitor of ATP synthesis, which reveals the amount of mitochondrial respiration not associated with ATP production (Figure 4D). Neither OA nor PA changed the OCR in the presence of oligomycin, thereby suggesting that neither treatment affects the mitochondrial coupling between respiration and ATP production. In other words, it appears that mitochondria remain similarly “efficient”. Additionally, we also measured OCR in the presence of FCCP, which dissipates the proton gradient at the inner mitochondrial membrane (i.e., Δψm). Given that mitochondrial respiration is a process that works against the proton gradient, the elimination of the Δψm allows respiration to reach its maximum levels. As with oligomycin, neither OA nor PA changed the magnitude of FCCP-induced OCR (Figure 4D), implying that mitochondria maintain a similar “maximal performance” across treatments. Taken together, these results suggest that the changes in basal OCR would not be due to differences in mitochondrial functional capacity (for example, increases in mitochondrial functional units or damage of existing ones), but rather to regulatory changes, such as differential substrate availability or upstream signaling cascades.

Similar to OCR, PA treatment triggered a significant increase in ATP production as soon as 2 h, which remained high at least until 6 h. However, OA treatment led to a slower ATP increase, which reached significance at 6 h (Figure 4E). As an operational indicator of the mitochondrial functional profiles, we used the relationship between ATP production and the Δψm (Figure 4F). At 2 h, the ATP/Δψm ratio remained relatively constant for both conditions. Nonetheless, at 6 h there was a divergence: the mitochondrial network from PA-treated cells appeared more ATP production-oriented, while the network from OA-treated cells seemed to prefer Δψm build-up. Altogether, these results suggest that both treatments stimulate mitochondrial bioenergetics, but with different functional profiles.

In sum, both PA and OA fatty acids induce distinct changes in mitochondrial morphology, LD-mitochondria interaction, and their bridging proteins and at the level of mitochondrial functionality. However, our previous studies have shown that the mitochondrial network and its interactions are highly heterogeneous (22, 23). Thus, the observed changes probably involve a subpopulation of mitochondria as a sort of “specialization mechanism”. Moreover, mounting evidence underscores an association between mitochondrial size and function, although in a cell type-dependent fashion (35). In brown adipocytes, which avidly oxidize FA, Benador et al. showed that LD-associated mitochondria are larger in size and mainly dedicated to ATP production and LD expansion, while smaller mitochondria associate mainly with FA oxidation (8). To address whether OA or PA replicate this behavior in our hepatocyte-derived cell line, we first compared the Δψm level of the LD-bound mitochondrial network with the rest of the mitochondrial network (hereafter termed “bulk mitochondria”). In Figures 5A,B, we show that although both OA and PA increase the Δψm of bulk mitochondria starting at 2 h, only OA has a steady effect. In contrast, PA again displays a biphasic response, which faded at 6 h, reappeared at 18 h, and finally disappeared after 24 h. In terms of LD-bound mitochondria, OA caused a minor accretion in their Δψm, which became significant after 24 h of treatment (Figure 5C). On the other hand, PA augmented the Δψm after 6 h, which faded at 18 h, but reappeared at 24 h. Curiously, the Δψm increments in bulk mitochondria interleaved with those of the LD-bound mitochondria. Taken together, these results suggest that OA induces steep nucleation and gradual expansion of LDs, which strongly interact with mitochondria exhibiting a relatively “weak” Δψm. Contrarily, PA triggers the biogenesis of multiple but small LDs, which show a moderately increased interaction with mitochondria harboring a relatively “strong” Δψm, associated with higher OCR and ATP production.

Finally, to assess the association between mitochondrial bioenergetics and LDs dynamics, we correlated the size of each individual LD with the Δψm of the associated mitochondria. We chose the temporality of 2 h because this is our earliest determination of LD nucleation, where the differences between OA and PA are most marked. In agreement with Figure 1, the histogram of Figure 5D shows that OA favors the emergence of larger LDs (cross-sectional area > 0.2 μm²), while PA leads to the accumulation of smaller ones with smaller areas (< 0.2 μm²). Then, we analyzed each LD from 18-20 cells in both conditions and correlated the size of each LD with the associated total MitoTracker Orange fluorescence (Figure 5E). That is, we evaluated how active were the mitochondria associated with each LD (in terms of the Δψm). We found that the different fatty acids promoted the appearance of distinct populations of LDs, respective to their size and the activity of the attached mitochondria. In either case, the “bulk LDs” consisted of units with smaller size (< 1 μm²), associated with mitochondria with lower Δψm. As initially noted, OA led to the emergence of larger LDs, characterized by their colocalization with mitochondria with lower Δψm. On the contrary, PA treatment did not promote the enlargement of LDs but triggered the appearance of a sub-population of small-sized LDs associated with mitochondria with higher Δψm. Altogether, these observations indicate that the unsaturated fatty acid OA promotes lipid accumulation in larger LDs proximal to mitochondria that build-up lower Δψm and that the saturated fatty acid PA favors smaller LDs in close apposition to mitochondria with higher Δψm.