The evolutionarily conserved protein CG9186 is associated with lipid droplets, required for their positioning and for fat storage

Abstract

Lipid droplets (LDs) are specialized cell organelles for the storage of energy-rich lipids. Although lipid storage is a conserved feature of all cells and organisms, little is known about fundamental aspects of the cell biology of LDs, including their biogenesis, structural assembly and subcellular positioning, and the regulation of organismic energy homeostasis. We identified a novel LD-associated protein family, represented by the Drosophila protein CG9186 and its murine homolog MGI:1916082. In the absence of LDs, both proteins localize at the endoplasmic reticulum (ER). Upon lipid storage induction, they translocate to LDs using an evolutionarily conserved targeting mechanism that acts through a 60-amino-acid targeting motif in the center of the CG9186 protein. Overexpression of CG9186, and MGI:1916082, causes clustering of LDs in both tissue culture and salivary gland cells, whereas RNAi knockdown of CG9186 results in a reduction of LDs. Organismal RNAi knockdown of CG9186 results in a reduction in lipid storage levels of the fly. The results indicate that we identified the first members of a novel and evolutionarily conserved family of lipid storage regulators, which are also required to properly position LDs within cells.

Keywords: Drosophila melanogaster; Lipid droplets; Lipid metabolism; Organelle clustering; Organelle positioning.

Fig. 1. CG9186 is an evolutionarily conserved LD-associated protein

(A) Phylogenetic tree of CG9186 and related sequences from different species. (B) Sequence alignment of CG9186 and homologous sequences of insects and mammals. Sequence identity in five or six species is shown in dark blue, identity in three or four species is shown in lighter shades of blue. The orange shaded boxes indicate the predicted catalytic center with the catalytic triad consisting of serine 119, aspartate 254 and histidine 283. (C) Schematic view of the CG9186 genomic locus, associated transcripts (turquoise) and resulting protein (blue), generated protein variants (yellow) and RNAi targeting sequences (black). (D) Homology-based model of CG9186. The model demonstrates accessibility of the protein targeting region (turquoise; amino acids 141–200) and the hypothetical arrangement of the annotated catalytic triad (black rectangle; the inset on the top left depicts an enlarged view). (E) Whole-mount in situ hybridization of Drosophila embryos using a CG9186 RNA antisense probe. CG9186 is maternally provided (embryonic stage 4) and later on (embryonic stages 12–13) is restricted to the salivary glands (asterisk). (F) Detection of CG9186 transcripts by in situ hybridization in combination with a fluorescent antibody staining for crebA, a marker for embryonic salivary glands (Fox et al., 2010). Embryos are oriented anterior to the left and dorsal side up.

Fig. 2. Subcellular localization of CG9186:eGFP

(A) CG9186:eGFP localizes to LDs in Drosophila Kc167 cells and induces LD clusters (asterisk) whereas LDs of untransfected cells are dispersed (arrowhead). (B) Co-localization of CG9186:eGFP and PLIN2:TdT in Kc167 cells resulted in fused, irregular LDs, which are also seen following overexpression of CG9186:eGFP in cells treated with brummer RNAi (C) or 800 μM OA (D). (E-G) Quantification of the triacylglycerol increase induced by PLIN2:TdT overexpression, brummer RNAi or oleic acid loading. Acylglycerol levels in E were measured with the zenbio triglyceride measurement kit and in F and G with a colorimetric assay (Hildebrandt et al., 2011). Acylglycerol levels are normalized to total protein. Scale bars: 5 μm.

Fig. 3. Localization of endogenous CG9186 and specificity of the CG9186 antiserum

(A-D) Immunofluorescent stainings using the CG9186 antiserum, of (A) Kc167 tissue culture cells, (B) third instar larval salivary glands, (C) and third instar larval fat bodies of control animals (Fb-Gal4 driver with no activated transgene and (D) fat-body-enriched CG9186 RNAi transgene expression. (E) Fat body of w[−] control animals stained with the CG9186 pre-immune serum. LDs were counterstained with HCS LipidTOX. (F) Western blot of third instar larval fat body and tissue culture cell extracts. The fat body extracts were obtained from animals carrying a Gal4 activator driving fat-body-enriched expression (Fb-Gal4) of a CG9186:eGFP transgene, a CG9186 knockdown transgene (VDRC105945) or no transgene as a control (VDRC w[−]). The Kc167 cell extracts came from wild-type cells (mock treated), or cells targeted by two different dsRNAs reducing CG9186 transcript levels (RNAi treatment lasted six days). β-Tubulin served as a loading control. Scale bars: 5 μm.

Fig. 4. CG9186 targets LDs via the ER

(A) In the absence of OA Drosophila ML-DmBG3-c2 cells show almost no LDs. Under these conditions CG9186:eGFP localizes in a reticular pattern reminiscent of the ER. (B) Upon addition of 400 μM OA, the ML-DmBG3-c2 cells deposit copious amounts of LDs decorated by CG9186:eGFP. (C) ML-DmBG3-c2 cells co-transfected with CG9186:eGFP and KdelR:TdT demonstrate ER localization of CG9186:eGFP. (D) Biochemical analysis of CG9186 subcellular localization. Drosophila Kc167 cells were grown in the absence or presence of OA. Cell extracts were prepared and the subcellular compartments were purified by subcellular fractionation. Equal volumes (20 μl of the LD fraction, ‘LDs’) or equal amounts of protein (10 μg of cytosolic or membrane proteins) of the respective fractions were subsequently probed in western blots with antisera detecting CG9186, PLIN2 or β-Tubulin. Please note that the cytosolic fractions contained much more total protein than the microsomal pellet. (E) Enlarged image frames from supplementary material Movie 1. The movie depicts the translocation of CG9186:eGFP from the ER to newly formed LDs. Kc167 cells stably transfected with CG9186:GFP were kept in low serum medium to avoid the deposition of LDs. At time point 0, OA-containing complete medium was added and every 10 seconds a z-stack of images were recorded with a spinning disc confocal system. Cells at 1 minute 30 seconds, 4 minutes, 10 minutes and 20 minutes post OA addition are shown. The cells of the polyclonal cell line differ in CG9186:eGFP expression level (arrow, arrowhead, asterisk). Cells that start with prominent CG9186:eGFP show LD-associated CG9186:eGFP by 10 minutes post OA addition (cells marked with asterisk or arrowhead). Cells lacking CG9186:eGFP expression need about 10 minutes longer until clear-cut localization can be seen (cells marked by arrow). Scale bars: 5 μm.

Fig. 5. CG9186 regulates LD positioning and is not involved in Lipid remobilization

(A,B) Maximum intensity projections of z-stacks of Drosophila Kc167 cells expressing CG9186:eGFP (A) and CG9186S119A:eGFP (B) 40 hours post transfection. LDs were induced with 400 μM OA. (C) Cells as shown in A and B, as well as neighboring untransfected cells, were analyzed for their respective LD phenotype. Bars represent the averaged results from three biologically independent experiments. In total, the LD patterns of 638 untransfected cells, 699 CG9186:eGFP-transfected and 650 CG9186S119A:eGFP-transfected cells were visually analyzed. Error bars indicate s.d. (D) The lipid content of cells stably expressing eGFP, CG9186:eGFP or CG9186S119A:eGFP was analyzed by thin layer chromatography. (E) Acylglycerol content of wild-type Kc167 cells and cells stably expressing CG9186 was analyzed with a colorimetric assay (Hildebrandt et al., 2011). LDs were induced overnight with 400 μM OA. (F) eYFP-tagged CG9186 protein targets LDs in COS-7 cells and induces LD clustering. (G) Lipolytic activity towards trioleoylglycerol (TAG), dioleoylglycerol (DAG) or monooleoylglycerol (MAG) of COS-7 cells transiently transfected with constructs expressing β-Gal (LZ), hormone sensitive lipase (HSL) or CG9186. Scale bars: 5 μm (A,B) and 10 μm (F).

Fig. 6. Structure–function analysis of CG9186

(A-E) eGFP-tagged CG9186 deletion constructs expressed in Drosophila Kc167 cells in the presence of 400 μM OA. (F) Cells expressing CG9186:eGFP (174 cells), CG9186(Δ1–140):eGFP (121 cells) or CG9186(Δ201–307):eGFP (151 cells) were analyzed visually for the LD-clustering phenotype. Bars represent the mean of three separate experiments. Error bars indicate s.d. (G) Helical wheel plot of amino acids 179–196 (FNFLPVWLRLMLIQIYFL′) of CG9186. Hydrophilic residues are shown as circles, hydrophobic residues as diamonds and potentially positively charged residues as pentagons. Hydrophobicity is represented by green shading (darker green is more hydrophobic) and hydrophilicity by red colouring. Potentially charged residues are shown in blue. The line indicates the border between hydrophobic and hydrophilic residues. Scale bars: 5 μm.

Fig. 7. CG9186 function is evolutionarily conserved

(A) Expression of an eGFP-tagged fusion protein of the murine CG9186 ortholog MGI:1916082 in COS-7 cells. In the absence of lipid loading the protein localized at the ER. Upon incubation with OA MGI:1916082 translocated to LDs. The images in the third row show a wider field of view of cells expressing MGI:1916082:eGFP (arrowhead) next to untransfected cells (asterisk) reveal that LDs are dispersed in non-expressing cells in contrast to the clustered LDs in the transfected cells. (B) Expression of GFP-tagged MGI:1916082 in Drosophila Kc167 cells results in LD targeting and clustering. (C) Northern blot analysis of the expression of MGI:1916082 encoding transcripts in various tissues. (D) qRT-PCR of the different tissue samples. Relative expression levels normalized to MGI:1916082 expression in brown adipose tissue (BAT) and β-Actin as a reference gene. WAT, white adipose tissue. Scale bars: 10 μm.

Fig. 8. CG9186 is involved in organismic lipid storage regulation

(A-D) Salivary glands of Drosophila third instar larvae store dispersed LDs as shown by: (A) localization of the bona fide LD-associated protein PLIN1:eGFP (Beller et al., 2010) to LipidTOX-positive spherical particles; (B-D) different neutral lipid stains. (E) Transgene-derived CG9186:eGFP overexpression results in a clustering of LDs in third instar larval salivary glands. (F) This clustering phenotype is enhanced when the mutated CG9186S119A:eGFP protein is overexpressed. (G) Ubiquitous or salivary gland enriched RNAi-mediated targeting of CG9186-encoding transcripts results in a reduction of salivary gland LDs. A–G are maximum intensity projections of z-stacks. (H) Ubiquitous RNAi-mediated knockdown of CG9186 results in decreased overall TAG storage levels in 6-day-old flies, fed ad libitum, compared with control flies lacking the transgene (VDRC w[−]). (I) Functional domains of CG9186 identified in the present study. Schematic drawing of the CG9186 sequence depicting important protein regions with the respective functional information uncovered by our studies. Scale bars: 20 μm.