Grasp55−/− mice display impaired fat absorption and resistance to high-fat diet-induced obesity

The Golgi apparatus plays a central role in the intracellular transport of macromolecules. However, molecular mechanisms of Golgi-mediated lipid transport remain poorly understood. Here, we show that genetic inactivation of the Golgi-resident protein GRASP55 in mice reduces whole-body fat mass via impaired intestinal fat absorption and evokes resistance to high-fat diet induced body weight gain. Mechanistic analyses reveal that GRASP55 participates in the Golgi-mediated lipid droplet (LD) targeting of some LD-associated lipases, such as ATGL and MGL, which is required for sustained lipid supply for chylomicron assembly and secretion. Consequently, GRASP55 deficiency leads to reduced chylomicron secretion and abnormally large LD formation in intestinal epithelial cells upon exogenous lipid challenge. Notably, deletion of dGrasp in Drosophila causes similar defects of lipid accumulation in the midgut. These results highlight the importance of the Golgi complex in cellular lipid regulation, which is evolutionary conserved, and uncover potential therapeutic targets for obesity-associated diseases.

| Grasp55 -/mice have growth retardation. a, Schematic diagram of the strategy used to generate Grasp55 -/mice (Gorasp2 tm1(KOMP)Vlcg , strain; KOMP Repository, Oakland, CA, USA). The genotyping primer regions are marked with black and yellow arrows. b, PCR genotyping of genomic DNA samples from mice generated by heterozygote (+/-) crosses. c, d, Photographs of wild-type (Grasp55 +/+ ) and Grasp55 -/mice at embryonic day 18.5 (c), postnatal day 7 (d). Source data are provided as a Source Data file.

Supplementary Fig. 2 | Brown adipose tissues are reduced in Grasp55 -/mice. a, b,
Photograph (a) and weight (b) of brown adipose tissue (BAT) of Grasp55 +/+ and Grasp55 -/mice at 4 weeks of age (n = 6). c, d, Histologic examination of BAT (H&E staining). Representative light microscopic tissue images are presented in (c). Quantitative analyses of the H&E images, that inversely correlate with fat contents of BAT by counting the number of adipocyte nuclei in a light microscopic field (450  340 μm), are depicted in (d, n = 10). Data are shown as mean ± SEM. Scale bars: 50 μm. *p < 0.05, **p < 0.01. All p values were calculated by unpaired two-tailed Student's t tests. Source data are provided as a Source Data file.

Supplementary Fig. 4 | GRASP55 deficiency reduces dietary lipid absorption but does
not affect amino acid absorption in mice. a, An oral fat tolerance test (OFTT) was performed in Grasp55 +/+ and Grasp55 -/mice without tyloxapol treatment. Mice received oral gavage with olive oil (10 μl/g of body weight) after 16 h fasting. Triglyceride (TG) concentrations were determined in blood plasma (n = 6). b-d, Intestinal absorption of amino acids was measured in Grasp55 +/+ and Grasp55 -/mice at the age of 12 weeks using 14 Cradiolabeled glycine (b), L-proline (c), and L-tryptophan (d) as described in Methods (each n = 4). Data are shown as mean ± SEM. n.s.: not significant, *p < 0.05, **p < 0.01. P values were calculated by unpaired (a) or paired (b, c and d) two-tailed Student's t tests. Source data are provided as a Source Data file.

Supplementary Fig. 5 | Measurement of cholesterol in plasma fractions and intestines. a, b,
Cholesterol content was determined in each blood plasma fraction after fast performance liquid chromatography of plasma lipids from Grasp55 +/+ and Grasp55 -/mice. Blood samples were taken 2 h after oral gavage with olive oil. The cholesterol content in each fraction is presented in (a). The area under curve (AUC) of plasma lipoprotein fractions was measured using ImageJ software and is summarized in (b). The data represent three independent experiments using pooled blood plasma from three to four mice. GRASP55 deletion reduces cholesterol contents in chylomicron (CM)/VLDL fractions. c, d, Intestinal cholesterol (c) and cholesteryl ester (d) levels were measured in the jejunum in 12-week-old Grasp55 +/+ and Grasp55 -/mice fasted for 16 h (HFD -), followed by high-fat diet feeding for 4 h (HFD +) (n = 6). Data are shown as mean ± SEM. n.s.: not significant, *p < 0.05. P values were calculated by paired (b) or unpaired (c and d) two-tailed Student's t tests. Source data are provided as a Source Data file. Supplementary Fig. 6 | The maximum luminal width of the Golgi cisternae is increased by GRASP55 deficiency. a, The maximum lumen widths of the Golgi cisternae were measured using EM images of mouse intestinal cells as detailed in Methods. Jejunum tissues of 12-week-old male mice were prepared 16 h after fasting. Bidirectional arrows indicate the maximum luminal width. b, Quantitative analyses of the maximum luminal width of the Golgi cisternae are summarized (Grasp55 +/+ , n = 138 from five mice; Grasp55 -/-, n = 146 from five mice). GRASP55 deficiency did not cause Golgi unstacking. However, it appeared to increase the luminal width of some Golgi cisternae (>0.1 m, dashed circle in b). On average, GRASP55 deficiency increased the maximum luminal width from 49.9 ± 2.2 nm to 58.1 ± 3.8 nm with a marginal statistical significance (p = 0.065, Student's t test). Scale bars: 200 nm. Source data are provided as a Source Data file. Fig. 7 | Golgi cisternal adhesive proteins are not altered by GRASP55 deficiency in the mouse jejunum. a, b, Immunoblot analysis of GM130, Golgin97, GRASP65, and Golgin45. Representative immunoblots are shown in (a). Densitometric analyses are shown in (b) (n = 4 for each). β-actin was monitored as a cytosolic protein loading control. c, Immunohistochemistry was performed using jejunums of Grasp55 +/+ and Grasp55 -/mice. GRASP55 deletion did not alter the Golgi localization of GM130, Golgin97, GRASP65, and Golgin45 (green, apical side of perinuclear regions in the jejunal epithelia). Nuclei were counterstained with 4′,6-diamidino-2-phenylindole (blue). Three independent experiments showed similar results. Unprocessed blots are presented in Supplementary

Supplementary Fig. 9 | GRASP55 deficiency reduces the Golgi localization of ATGL.
Immunohistological images of ATGL and the Golgi marker protein GM130 in intestinal epithelia. Jejunum tissues were prepared from Grasp55 +/+ and Grasp55 -/mice with free access to food (a) or 4 h after olive oil bolus (b, olive oil, 10 μl/g of body weight). The results of multiple experiments (n = 5) are summarized in (c). MCC, Manders' colocalization coefficient. Scale bars: 20 μm. Data are shown as mean ± SEM. **p < 0.01. All p values were calculated by unpaired two-tailed Student's t tests. Source data are provided as a Source Data file.

Supplementary Fig. 10 | GRASP55 deficiency reduces the Golgi localization of MGL.
Immunohistological images of MGL and the Golgi marker protein GM130 in intestinal epithelia. Jejunum tissues were prepared from Grasp55 +/+ and Grasp55 -/mice with free access to food (a) or 4 h after olive oil bolus (b, olive oil, 10 μl/g of body weight). The results of multiple experiments (n = 5) are summarized (c). MCC, Manders' colocalization coefficient. Scale bars: 20 μm. Data are shown as mean ± SEM. *p < 0.05. All p values were calculated by unpaired two-tailed Student's t tests. Source data are provided as a Source Data file.  Fig. 5b. The levels of pHSL, HSL, and MTP were not altered by GRASP55 depletion. e, f, The protein expression of MGL was analyzed by immunoblotting. Representative immunoblots are shown in (e) and densitometric analysis of multiple experiments is shown in (f, n = 4). The level of βactin was monitored as a cytosolic protein loading control. g, h, Quantitative PCR analysis of ATGL (g, n = 5) and MGL (h, n = 5) mRNAs. GRASP55 depletion reduces cytosolic MGL levels without affecting its mRNA levels. Unprocessed blots can be found in Supplementary Fig. 22 For the induction of ATGL protein expression, cells were treated with 400 μM oleic acid for 16 h. Representative immunoblots of LDs and cytosols are shown in (a). In panels (b-d), the results of multiple experiments of RAB18, ADRP, and ATGL, respectively, are summarized (n = 3 each). Aldolase A was used as a cytosolic protein loading control. MG132 does not rescue the decreased RAB18, ADRP, or ATGL levels induced by SAR1-T39N and ARF1-T31N. Unprocessed blots can be found in Supplementary Fig. 22. Data are shown as mean ± SEM. *p < 0.05. **p < 0.01. All p values were calculated by ANOVA followed by Tukey's multiple comparison tests. Source data are provided as a Source Data file. Supplementary Fig. 15 | GRASP55 does not alter the protein stability of GBF1 and ARF1 in Caco-2 cells. Immunoblot analysis of GBF1 and ARF1 levels were performed in control and GRASP55-depleted Caco-2 cells, stably transfected with shRNA against GRASP55 (shGRASP55). Some cells were incubated with the proteasomal inhibitor MG132 (0.5 μM, 16 h). Representative immunoblots are shown in (a). Densitometric analyses of GBF1 and ARF1 are shown in (b) and (c), respectively (n = 3 for each). GBF1 and ARF1 levels were not altered by GRASP55 depletion or by treatment with MG132 (compare with the results obtained for ATGL, shown in Supplementary Fig. 13). Unprocessed blots are presented in Supplementary Fig. 22. Data are shown as mean ± SEM. n.s.: not significant. All p values were calculated by unpaired two-tailed Student's t tests. Source data are provided as a Source Data file.

Supplementary Fig. 16 | Autophagic flux is not altered by Grasp55 deletion in the mouse jejunum.
To analyze the effects of GRASP55 deficiency on autophagy, conversion of microtubule-associated protein light chain 3 (LC3) into its phosphatidylethanolamineconjugated form (LC3-II) as a parameter of autophagic flux was examined. The jejunum samples were taken from ad libitum-fed mice. Some samples were taken 2 h after olive oil feeding (10 μl/g body weight). Under both conditions, deletion of Grasp55 did not affect LC3-II conversion. Representative immunoblots (a) and the results of densitometric analyses (b) (n = 4) are shown. Unprocessed blots are presented in Supplementary Fig. 22. Data are shown as mean ± SEM. n.s.: not significant. All p values were calculated by unpaired two-tailed Student's t tests. Source data are provided as a Source Data file. Supplementary Fig. 17 | GRASP55 deficiency decreases the cellular levels of ATGL and triglyceride contents in mouse hepatocytes. a-d, Immunoblot analysis of ATGL, MTP, and HSL in the mouse liver. Liver tissues were prepared from 12-week-old Grasp55 +/+ and Grasp55 -/mice fed on the normal diet. A representative immunoblot is shown in (a), and the results of multiple experiments (n = 4) are summarized in (b-d). The amount of β-actin was monitored as a cytosolic protein loading control. e, f, Morphological analysis of mouse livers fasted for 16 h or after olive oil bolus (olive oil, 10 μl/g of body weight) by H&E staining and electro-microscopy (EM). White arrows indicate the cytosolic LDs. Note that LD size is reduced in Grasp55 -/mice. Quantitative analyses of LD diameters are summarized in (f, Grasp55 +/+ , n = 296 from five mice; Grasp55 -/-, n = 150 from five mice). GRASP55 deficiency reduced the LD size from 1.23 ± 0.04 m to 0.73 ± 0.03 m (p < 0.01). g, The amount of triglyceride (TG) per g liver was measured (n = 6). Unprocessed blots can be found in Supplementary Fig. 22. Data are shown as mean ± SEM. Scale bars: 50 μm (H&E) and 5 μm (EM). n.s.: not significant, **p < 0.01. P values were calculated by paired (b, c and d) or unpaired (g) two-tailed Student's t tests. Source data are provided as a Source Data file.  Organ weight was measured at age of 4 weeks (n = 6). *p < 0.05, **p < 0.01, relative to Grasp55 +/+ (Student's t test). Source data are provided as a Source Data file.