LDs (lipid droplets) are cellular organelles which can be found in nearly all eukaryotic cells. Despite their importance in cell biology, the mechanism underlying LD biogenesis remains largely unknown. In the present study we report that conditions of ER (endoplasmic reticulum) stress stimulate LD formation in Saccharomyces cerevisiae. We found that LDs accumulated in yeast mutants with compromised protein glycosylation or ER-associated protein degradation. Moreover, tunicamycin and Brefeldin A, agents which induce ER stress, were found to stimulate LD formation. In contrast, the restoration of protein glycosylation reduced LD accumulation. Interestingly, enhanced neutral lipids synthesis and LD formation under conditions of ER stress was not dependent on Ire1p. Lastly, we demonstrated that the absence of LDs did not compromise cell viability under ER stress. Our results suggest that although more LDs are produced, LDs are not essential to cell survival under ER stress.
- endoplasmic reticulum stress
- lipid droplet
- neutral lipids storage
- Saccharomyces cerevisiae
- unfolded protein response
LDs (lipid droplets) contain a highly hydrophobic core of neutral lipids, mainly TAGs (triacylglycerols) and SEs (sterol esters), and are surrounded by a monolayer of phospholipids with proteins embedded [1,2]. Varying in size and composition, LDs can be found in nearly all eukaryotic cells including yeast, plants and mammals [3,4]. LDs have long been regarded as an inert storage depot for energy and more bioactive lipids, such as fatty acids and cholesterol. Recent studies, however, suggest that LDs may play important roles in various cellular processes, since biosynthetic enzymes and proteins involved in signalling and membrane trafficking are found to localize exclusively to or associate with this organelle [5–7]. It has also been shown that reduced synthesis of LDs by genetic approaches causes programmed cell death [8,9]. Rather than simply serving as an inert lipid storage depot, LDs may be a ubiquitous cellular organelle that participates in various cellular activities, such as protein and lipid trafficking, protein degradation and cell signalling . More importantly, changes in the cellular dynamics of LDs are associated with a number of devastating human diseases, such as obesity, Type 2 diabetes, fatty liver and atherosclerosis.
Despite the obvious physiological and pathological importance of LDs, very little is known about their biogenesis, maturation, degradation (lipolysis), and especially the regulation of these processes. Budding of nascent LDs from the ER (endoplasmic reticulum) is a prevailing model of LD formation . In this model, droplets of neutral lipids originate between the two leaflets of the ER bilayer and bud into the cytosol. Consistent with this model, freeze-fracture electron microscopy revealed sites of continuity between the membrane surface of LDs and the outer membrane leaflet of the ER . It is noteworthy, however, that alternative models of LD biogenesis exist [12,13]. Although these models conflict with each other, they all agree that the ER is the site of LD biogenesis. However, no ER-related signalling pathways have been identified to regulate or associate with the biogenesis of LDs.
The budding yeast Saccharomyces cerevisiae is a simple but powerful model genetic system and has proven to be invaluable to the understanding of cellular lipid metabolism, membrane trafficking and cell signalling. The biochemical pathways leading to the synthesis of TAGs and SEs are defined at the molecular level in yeast . In addition, a number of resident LD proteins have been identified . As in animal cells, LDs in yeast are also believed to originate from microdomains of the ER, where most of the enzymes for lipid synthesis reside. In an effort to identify gene products that may affect the dynamics of LDs, we took a ‘reverse genetic’ approach to screen the entire collection of yeast deletion mutants for changes in the quantity and morphology of LDs . Among the mutants that caused a significant increase in the size and/or number of LDs, many are involved in ER stress. Here we further investigate the role of ER stress in the production of LDs and the role of LDs under ER stress.
Yeast strains, medium, and agents
Wild-type (BY4741) and haploid deletion mutants were obtained from EUROSCARF (EUROpean Saccharomyces Cerevisiae ARchives for Functional analysis) collection centre developed by the Saccharomyces Genome Deletion Project . W303 (MATa leu2-3,112 trp1-1 can1-100 ura3-1 ade2-1 his3-11,15) and its derivative are1Δare2Δdga1Δlro1Δ (are1Δ::HIS3 are2Δ::LEU2 dga1Δ::URA3 lro1Δ::URA3), which is deficient in neutral lipids synthesis and hence devoid of LDs, were also used. Cells were grown with rotary shaking at 30°C in liquid YPD medium [1% yeast extract, 2% bacto peptone, 2% dextrose (Difco)]. Tm (tunicamycin; Sigma) was dissolved in DMSO and added to a final concentration of 10 μg/ml. BFA (Brefeldin A; Sigma) was prepared in ethanol and added to a final concentration of 75 μg/ml. Manganese chloride was purchased from Sigma. Oleic acid (Sigma) was solubilized with sodium hydroxide and bound to BSA at a molar ratio of 5.5:1 . When used, the final concentration of oleate was 0.5 mM.
Nile Red staining and fluorescence microscopy
Cells were stained with 20 μg/ml of Nile Red (Sigma). Fluorescence imaging was performed on a Leica DMLB microscope (Wetzlar) with a Curtis ebq 100 fluorescent lamp. A UV-filter set (436/7 nm bandpass excitation filter, 455 nm dichromatic mirror, 470 nm long-pass emission filter) was used to observe Nile Red fluorescence. The number of intracellular LDs was directly counted under the microscope; 100 cells were counted to calculate the average number of intracellular LDs, which was presented as mean±S.D.
Quantification of cellular neutral lipids
Lipid extraction was performed as described by Zhang et al.  and the quantification of neutral lipids was performed as described by Zweytick et al.  with modifications. Briefly, cells were grown in YPD until they reached the required growth phase as determined by D600, harvested, washed twice with 0.5% Nonidet P-40, once with distilled H2O, and freeze-dried. The dried cell pellets were resuspended in lyticase (Sigma) solution (1700 units/ml in 10% glycerol) and incubated at 37°C for 15 min, at −70°C for 1 h and at 37°C for 15 min. Lipids were extracted with hexane, blown dry and dissolved in chloroform/methanol (2:1, v/v). Samples were applied to Silica gel 60 F254 plates (Merck) and chromatograms were developed in hexane/diethyl ether/acetic acid (85:15:1, by vol.) with triolein and cholesteryl oleate (Matreya) as the standards. For quantification of SEs and TAGs, plates were dipped into methanolic MnCl2 solution (0.63 g MnCl2·4H2O, 60 ml water, 60 ml methanol and 4 ml concentrated sulfuric acid), dried and heated at 120°C for 15 min. Densitometric scanning was performed at 500 nm with a CAMAG TLC scanner.
Preparations of antibodies and protein immunoblotting
Antibodies against Are1p and Lro1p were raised by immunizing rabbits with GST-fused Are1p-12–190 and Lro1p-440–661 corresponding to the 179 near-N-terminal amino acids of Are1p and the 222 C-terminal amino acids of Lro1p respectively. Anti-Vti1p, used as a loading control, was a gift from Dr Wanjin Hong (Institute of Molecular and Cell Biology, Singapore).
For protein immunoblotting, the Bradford protein assay was used to determine the protein concentration of the samples . BSA was used as the protein standard. SDS/PAGE was performed according to the method of Laemmli  using 8% or 10% acrylamide gel. The proteins were visualized after electrophoresis by staining with Coomassie Blue . For immunoblotting, the proteins were transferred from the acrylamide gel to a nitrocellulose membrane as previously described . The membranes were blocked with 5% non-fat milk before probing with the primary and secondary antibodies. Detection of membrane-bound antibodies was performed using the enhanced chemiluminescence detection system (Pierce).
RNA isolation and RT–PCR (reverse transcription–PCR)
Total RNA was extracted using the RNeazy kit (QIAGEN). On-column DNase digestion (QIAGEN) was performed on all RNA samples. RNA concentrations were determined by measuring the absorbance of the samples at 260 nm and 280 nm. RNA purity and integrity was assessed by gel electrophoresis. RT–PCR was performed using an Access RT–PCR kit (Promega). Total RNA (1 μg for each sample) was used in a 50 μl system. Both ribosomal RNA and ACT1 which encodes yeast actin were used as loading controls. Primers are shown in Supplementary Table S1 (at http://www.BiochemJ.org/bj/424/bj4240061add.htm). PCRs were performed for 25 cycles. The final products (10 μl) were loaded on a 1% agarose gel and separated via electrophoresis. The intensity of the bands under UV light was compared. The 25-cycle PCR was still in the exponential phase, as the intensity of the PCR products was significantly stronger after a 30-cycle amplification compared with a 25-cycle amplification in all the samples.
Construction of deletion mutants
IRE1 knockout mutants were generated by replacing the entire IRE1 open reading frame with the HIS3 gene amplified from pFA6-HIS3MX6 via PCR-mediated homologous recombination in single deletion mutants . Proper integration of the HIS3 fragment was determined by colony PCR. The primers used are shown in Supplementary Table S1.
To identify gene products that may affect the dynamics of LDs, we took a ‘reverse genetic’ approach to screen the ∼4800 viable yeast deletion mutants (from EUROSCARF) with the vital dye Nile Red, which is specific for LDs. We identified two major classes of mutants: the mld (many lipid droplets) mutants that showed increased number of LDs, and the fld (few lipid droplets) mutants, which had significantly less LDs. The screening process and the characterization of one of the fld mutants has been recently reported . Here, we focus on the mld mutants and identify a role for ER stress in the regulation of LD biogenesis.
Mutants defective in N-linked glycosylation accumulated more LDs
Physiologically, ER stress can result from mutations of genes encoding glycosylation enzymes . Not surprisingly, we found that mutants defective in protein glycosylation, including anp1, erd1, mnn10, mnn11, och1, ost4 and pmr1, produced significantly more LDs when the cells entered stationary phase (Table 1 and Supplementary Figure S1 at http://www.BiochemJ.org/bj/424/bj4240061add.htm). In addition, lipid analysis showed that the mutants had elevated TAGs and/or SEs when compared with the wild-type. As shown in Figure 1(A), anp1 and pmr1 displayed increased synthesis of both TAGs (by 30% and 70% respectively) and SEs (by 40% and 110% respectively). In contrast, erd1, och1 and ost4 had a 70–100% increase in TAG synthesis only, whereas mnn10 and mnn11 strains mainly displayed increases in cellular SE level (by 11% and 16% respectively). The LDs of mnn10 and mnn11 cells appeared smaller under the microscope than those of the wild-type cells, which explained why neutral lipids synthesis increased only modestly in mnn10 and mnn11 strains, though the number of LDs increased by approx. 50% as compared with the wild-type (Table 1).
Mutations in ERAD components resulted in more LD accumulation
ERAD (ER-associated protein degradation) is responsible for the degradation of misfolded or unassembled proteins from the ER. The ERAD pathway translocates misfolded proteins back into the cytosol, where they are eliminated by the proteasome . Walter and co-workers found that impaired ERAD leads to an accumulation of unfolded proteins in the ER, thus constitutively causing ER stress . In the present study, we also found that LD biosynthesis was increased when DOA10 or HRD1, nonessential genes required for ERAD were mutated (Table 1 and Supplementary Figure S1). Meanwhile, doa10 displayed a 35% increase in TAGs and 60% increase in SEs and hrd1 displayed a 30% increase in TAGs and 40% increase in SEs, all compared with the wild-type (Figure 1B).
Tm or BFA treatment induced LD synthesis
To directly test ER stress and the production of LDs, we experimentally induced ER stress by treating cells with Tm, which inhibits N-linked glycosylation, and found that Tm treatment led to increased intracellular LD production. Wild-type cells, grown to early exponential phase (D600∼0.5), were either harvested immediately or treated with Tm or an equivalent volume of DMSO (final 0.2% v/v). After 1 h of incubation in the presence of Tm, cells accumulated more LDs (3.1±1.3) compared with the group receiving no additional treatment (1.5±0.8, P<0.01). DMSO alone had no obvious effect on LD formation (Figure 2A). Lipid quantification analysis showed that Tm treatment increased TAG and SE synthesis by 30% and 130%, respectively (Figure 2B). Furthermore, Tm treatment enhanced oleate-induced LD synthesis. As shown in Figures 2(A) and 2(B), compared with the control group (YPD+DMSO), cells pretreated with Tm displayed more LDs and also a higher level of TAGs and SEs after the addition of oleate. When the control group was incubated in the presence of 0.5 mM oleate for 1 h, the number of LDs per cell on average was 5.2±2.1, cellular TAG synthesis increased 6.5-fold compared with the YPD group, and SE synthesis increased 1.5-fold. In comparison, when cells were pretreated with Tm for 1 h, the average number of intracellular LDs after oleate supplementation was 9.1±2.9 (P<0.001, compared with 5.2±2.1), cellular TAGs increased 8-fold and SEs increased 5-fold (both compared with the YPD group).
Similarly, BFA induced neutral lipids synthesis and LD formation in erg6 cells (Figure 2). Since BFA cannot easily penetrate the cell wall of wild-type cells, we selected the erg6 mutant. When the erg6 cells were incubated in the presence of 75 μg/ml BFA, the number of intracellular LDs per cell on average increased from 2.6±1.1 to 3.6±1.3 (P<0.01), cellular TAG synthesis was elevated by 8-fold and SE synthesis by 80%. In addition, when oleate was added to the medium, cells pretreated with BFA also synthesized more LDs compared with the control group (11.3±3.5 against 6.3±2.8, P<0.001). Likewise, BFA treatment also accelerated oleate-induced cellular TAG and SE synthesis.
Alleviation of ER stress conditions by restoration of protein glycosylation reduced the ‘mld’ phenotype
To further verify our hypothesis that conditions of ER stress induce neutral lipids and LD synthesis, we studied whether the removal of ER stress could reduce LD accumulation. Pmr1p is a Ca2+ ATPase implicated in protein glycosylation; pmr1 cells have defects in outer chain glycosylation . It is suggested that the defect is primarily due to a failure to transport Mn2+ into the secretory pathway, and hence the addition of Mn2+ greatly alleviates the glycosylation defect in pmr1 cells . As expected, supplementation of Mn2+ effectively reduced the production of LDs in pmr1 cells. With increasing concentrations of Mn2+, pmr1 cells gradually decreased LD formation and synthesis of neutral lipids (Supplementary Figures S2A and S2B at http://www.BiochemJ.org/bj/424/bj4240061add.htm). When the concentration of Mn2+ was increased to 4.5 μM, its effect reached a plateau. Higher levels of Mn2+ inhibited the growth of pmr1 cells, with 450 μM causing complete growth arrest (results not shown). In contrast, supplementation of Mn2+ of equivalent concentrations to the wild-type cells did not result in any significant change in either the number of LDs (Supplementary Figure S2C) or in the amount of cellular neutral lipids (Supplementary Figure S2D).
Expression levels of enzymes catalysing the synthesis of neutral lipids were not increased when LD formation was stimulated under conditions of ER stress
The increased synthesis of neutral lipids and LDs in conditions of ER stress could have many possible causes. One possibility could involve the increased expression of the enzymes catalysing the synthesis of TAGs and SEs. Are1p, Are2p, Dga1p and Lro1p are the four enzymes catalysing the final step of the formation of SEs and TAGs. To be precise, Are1p and Are2p are responsible for the formation of SEs and a minor portion (approx. 5%) of TAGs, whereas Dga1p and Lro1p are responsible for the majority of TAG synthesis . In the present study, we successfully raised antibodies against Are1p and Lro1p and used them to observe the cellular levels of Are1p and Lro1p via immunoblotting when ER stress was initiated. As indicated in Figure 3(A), Tm treatment did not increase the protein levels of Are1p and Lro1p. In addition, both ERAD mutants and four selected protein glycosylation mutants that were found containing significantly more LDs in this study had similar (or even lower) levels of Are1p and Lro1p to those of the wild-type cells. Furthermore, supplementation of Mn2+ reduced LD accumulation of pmr1 cells, but failed to change the cellular levels of Are1p and Lro1p (Figure 3B).
Are2p and Dga1p are the other two enzymes involved in neutral lipids synthesis. We did not have antibodies against Are2p and Dga1p, and therefore chose RT–PCR to compare the mRNA levels of ARE2 and DGA1. First, we compared the mRNA levels of the wild-type cells before and after Tm treatment. As seen in Figure 3(C), after Tm treatment, the mRNA levels of ARE1 and LRO1 remained unchanged, which was consistent with our results that Tm treatment did not result in a significant change in the expression levels of Are1p and Lro1p, as shown in Figure 3(A). Importantly, the mRNA levels of ARE2 and DGA1 were not elevated after Tm treatment either. Furthermore, the mutants defective in protein glycosylation or ERAD did not display higher mRNA levels of ARE1, ARE2, DGA1 or LRO1 than the wild-type cells (Figure 3D). On the contrary, the mRNA level of ARE2 was even lower in anp1 and mnn11. Taken together, our results demonstrate that the expression levels of enzymes involved in neutral lipids synthesis are not upregulated under conditions of ER stress, despite the increased synthesis of LDs. These results suggest that more lipid substrates are diverted to the ER for the synthesis of TAGs and SEs under ER stress.
Stimulated LD production in conditions of ER stress was not Ire1p-dependent
It has been demostrated that Ire1p senses the accumulation of unfolded proteins in the ER lumen and signals the information across the ER membrane, whereas Hac1p executes the activation of transcription of UPR (unfolded protein response) genes . We therefore addressed whether Ire1p was essential for LD formation in conditions of ER stress. We knocked out the IRE1 gene from anp1, mnn10, mnn11, pmr1 and doa10 strains. All the double deletion mutants demonstrated hypersensitivity to 0.5 μg/ml Tm (results not shown), confirming the deletion of the IRE1 gene. However, both the level of neutral lipids and the quantity of intracellular LDs in the double deletion mutants showed no significant difference from those of the corresponding single deletion strains (Figure 4 and Supplementary Figure S3 at http://www.BiochemJ.org/bj/424/bj4240061add.htm). Moreover, Tm treatment also led to elevated synthesis of neutral lipids and accelerated formation of LDs in ire1 cells. As seen in Figure 5, Tm treatment dramatically increased LD formation in exponentially growing ire1 cells, and the number of LDs increased from 1.6±0.5 to 3.1±1.2 (P<0.01). In addition, TAG synthesis was elevated by 10% and SEs by 90%. On the basis of these two experiments, we conclude that increased LD formation under conditions of ER stress is not dependent on Ire1p.
LD formation was not essential to cell survival in conditions of ER stress
Since LD synthesis was stimulated in conditions of ER stress, we tested whether this process was essential for the viability of cells under ER stress. The mutant QKO (are1Δare2Δdga1Δlro1Δ) is a strain in which all the four enzymes involved in neutral lipids synthesis have been knocked out. Consequently, this mutant is completely devoid of intracellular LDs (Figure 6A). We spotted both the wild-type w303 and the mutant QKO cells onto a YPD plate containing 0.5 μg/ml Tm and incubated the plate for 48 h at 30°C. The viability of QKO was not compromised in the presence of 0.5 μg/ml Tm (Figure 6B). In addition, we did not observe any difference in growth between the wild-type and the QKO cells with increasing concentrations of Tm (results not shown). These results suggest that LD formation is not essential to cell survival in conditions of ER stress.
LDs have increasingly been recognized as an important cellular component. The synthesis of the core lipids TAG and SE is known to be the major contributor in LD formation. However, little is known about how other factors/proteins are involved in LD biogenesis. Nor is it clear how the synthesis of TAGs and SEs is regulated during the biogenesis and expansion of the LDs. In the present study, we established a strong link between conditions of ER stress, synthesis of neutral lipids and LD formation. We further showed that the Ire1p-mediated transcriptional programme was not involved in enhanced lipid synthesis and LD formation under ER stress. Most significantly, we demonstrated that LDs were dispensable for cell viability under ER stress.
It is well established that the accumulation of unfolded and/or misfolded proteins in the ER induces ER stress . These proteins need to be stored and degraded properly to avoid cytotoxicity. For instance, partially unfolded proteins expose hydrophobic residues and tend to aggregate, thereby resulting in cytotoxicity. The LDs can serve as a temporary safe depot for proteins with exposed hydrophobic patches . Therefore, having more LDs under ER stress may be a self-protective mechanism employed by most eukaryotic cells. Supporting this hypothesis, we provided convincing evidence that conditions of ER stress stimulate LD formation: (1) mutants defective in protein glycosylation or ERAD displayed more LDs than the wild-type cells; (2) agents that induce ER stress triggered LD synthesis; (3) alleviation of ER stress via restoration of glycosylation decreased LD production. Interestingly, the number of LDs increases in many disease states, including osteoarthritis and liver degeneration . Such an increase in droplet number might be a protective response that generates more hydrophobic surface to sequester misfolded proteins, most likely in response to ER stress.
Since the UPR is initiated whenever protein folding in the ER is compromised, we investigated whether LD formation in conditions of ER stress was mediated by UPR. Although the scope of UPR is very broad and many aspects of secretory function are under the regulation of UPR , it is evident that stimulated LD biosynthesis in conditions of ER stress does not belong to traditional UPR because this process is not Ire1p-dependent. In agreement with this, we did not observe any significant increase in the protein levels of key enzymes involved in neutral lipids synthesis or mRNAs encoding those enzymes under ER stress, although more lipids were synthesized. This is not surprising, as most of these enzymes are regulated allosterically by the substrates . One possibility is that lipid trafficking from other cellular compartments to the ER is enhanced when the cells are under stress, thereby providing more substrates for the synthesis of SEs and TAGs. It should also be noted that, although the mutants identified in this study all accumulated more LDs than the wild-type, their impact on neutral lipids synthesis varied between strains. erd1, och1 and ost4 mainly upregulated TAG synthesis, mnn10 and mnn11 mainly upregulated SE synthesis, whereas anp1, pmr1, hrd1 and doa10 upregulated both TAG and SE synthesis (Figure 1). In addition, after Tm treatment, the increase of SE synthesis was more prominent than that of TAG synthesis, whereas BFA treatment appeared to have a greater impact on TAG synthesis (Figure 2). These results suggest that different types of ER stress may differ in their impact on neutral lipids synthesis.
A surprising yet important discovery is that the yeast strain completely devoid of LDs (the QKO mutant) was still viable in the presence of 0.5 μg/ml of Tm, indicating that accelerated neutral lipids synthesis and LD formation is not essential in conditions of ER stress. This result does not negate a protective role of LDs under physiological conditions. However, it does suggest that yeast cells are equipped with multiple means to cope with unfolded/misfolded proteins, especially under extreme conditions.
In summary, the results described here offer new insights into the regulation of the synthesis of storage neutral lipids and the biogenesis of LDs. ER stress may not only control lipid synthesis, but also regulate the budding, fusion or degradation of LDs, and other aspects of lipid metabolism. For instance, it has been reported that ER stress links obesity, insulin action and Type 2 diabetes in mammals . Further studies in this area will undoubtedly bring beneficial outcomes to LD biology and human health.
Hongyuan Yang and Weihua Fei planned the experiments and wrote the manuscript. Weihua Fei performed most of the experiments. Han Wang and Xin Fu raised anti-sera and undertook immunoblotting. Christopher Bielby helped with part of the fluorescence microscopic work.
This research was supported by a grant from the Biomedical Research Council, Singapore [BMRC grant number 04/1/21/19/322] and a grant from the Academic Research Council [grant number DP0984902], Australia.
We thank Dr Davis Ng (Temasek Life Sciences Laboratory, Singapore) for his advice on this study. We also thank Dr Wanjin Hong for the Vti1p antibody.
Abbreviations: BFA, Brefeldin A; ER, endoplasmic reticulum; ERAD, ER-associated protein degradation; EUROSCARF, EUROpean Saccharomyces Cerevisiae ARchives for Functional analysis; fld, few lipid droplets; LD, lipid droplet; mld, many lipid droplets; QKO mutant, are1Δare2Δdga1Δlro1Δ; RT–PCR, reverse transcription–PCR; SE, sterol ester; TAG, triacylglycerol; Tm, tunicamycin; UPR, unfolded protein response
- © The Authors Journal compilation © 2009 Biochemical Society