Intestinal Failure and Aberrant Lipid Metabolism in Patients With DGAT1 Deficiency.

BACKGROUND & AIMS: Congenital diarrheal disorders are rare inherited intestinal disorders characterized by intractable, sometimes life-threatening, diarrhea and nutrient malabsorption; some have been associated with mutations in diacylglycerol-acyltransferase 1 (DGAT1), which catalyzes formation of triacylglycerol from diacylglycerol and acyl-CoA. We investigated the mechanisms by which DGAT1 deficiency contributes to intestinal failure using patient-derived organoids.
METHODS: We collected blood samples from 10 patients, from 6 unrelated pedigrees, who presented with early-onset severe diarrhea and/or vomiting, hypoalbuminemia, and/or (fatal) protein-losing enteropathy with intestinal failure; we performed next-generation sequencing analysis of DNA from 8 patients. Organoids were generated from duodenal biopsies from 3 patients and 3 healthy individuals (controls). Caco-2 cells and patient-derived dermal fibroblasts were transfected or transduced with vectors that express full-length or mutant forms of DGAT1 or full-length DGAT2. We performed CRISPR/Cas9-guided disruption of DGAT1 in control intestinal organoids. Cells and organoids were analyzed by immunoblot, immunofluorescence, flow cytometry, chromatography, quantitative real-time polymerase chain reaction, and for the activity of caspases 3 and 7.
RESULTS: In the 10 patients, we identified 5 bi-allelic loss-of-function mutations in DGAT1. In patient-derived fibroblasts and organoids, the mutations reduced expression of DGAT1 protein and altered triacylglycerol metabolism, resulting in decreased lipid droplet formation after oleic acid addition. Expression of full-length DGAT2 in patient-derived fibroblasts restored formation of lipid droplets. Organoids derived from patients with DGAT1 mutations were more susceptible to lipid-induced cell death than control organoids.
CONCLUSIONS: We identified a large cohort of patients with congenital diarrheal disorders with mutations in DGAT1 that reduced expression of its product; dermal fibroblasts and intestinal organoids derived from these patients had altered lipid metabolism and were susceptible to lipid-induced cell death. Expression of full-length wildtype DGAT1 or DGAT2 restored normal lipid metabolism in these cells. These findings indicate the importance of DGAT1 in fat metabolism and lipotoxicity in the intestinal epithelium. A fat-free diet might serve as the first line of therapy for patients with reduced DGAT1 expression. It is important to identify genetic variants associated with congenital diarrheal disorders for proper diagnosis and selection of treatment strategies.

Background and Context

Mutations in DGAT1 have recently been identified in patients with congenital diarrheal disorders (CDDs), but the underlying molecular pathomechanisms have remained largely elusive.

New Findings

The authors identified 10 patients with DGAT1 deficiency representing the largest cohort to date, linking gut epithelial lipid metabolism and lipotoxicity to CDD; and rescued aberrant lipid metabolism with isoenzyme DGAT2.

Limitations

Although the authors show exogenous DGAT1 or DGAT2 expression or proteasome inhibitors may overcome defects, future studies may need to address how that knowledge can be translated to targeted therapies.

Impact

The authors highlight the importance of identifying the genetic defect in patients with CDD, and showcase further use of gut organoid technology to study rare diseases of the gastrointestinal tract.

Congenital diarrheal disorders (CDDs) are a group of rare inherited intestinal disorders that are characterized by intractable, sometimes life-threatening, diarrhea and nutrient malabsorption. CDDs can be classified based on their aberrations in absorption and transport of nutrients and electrolytes, enterocyte differentiation and polarization, enteroendocrine cell differentiation, or dysregulation of the intestinal immune response. Congenital protein-losing enteropathy (PLE) is a type of CDD that is characterized by increased protein loss from the gastrointestinal (GI) system. Patients with PLE often suffer from hypoproteinemia, fat malabsorption, fat-soluble vitamin deficiencies, and malnutrition. Recently, we have identified germline loss-of-function mutations in CD55 as a major monogenic etiology for congenital PLE.

Previously, mutations in the gene encoding diacylglycerol-acyltransferase 1 (DGAT1) were found to underlie a syndrome of diarrhea and congenital PLE.3, 4, 5, 6 DGAT1 and its isozyme DGAT2 (encoding for diacylglycerol-acyltransferase 2) are responsible for the conversion of diacylglycerol (DG) and fatty acyl-CoA to triacylglycerol (TG) in humans.7, 8 TG is the main energy substrate stored in human adipose tissue, is essential for milk production in the mammary gland, and is part of the very low-density lipoprotein–mediated transport of lipids to peripheral tissue.9, 10 In the human small intestine, DGAT1 is the only highly expressed enzyme, whereas DGAT2 is mainly expressed in the liver.3, 11 In enterocytes, TG is stored in lipid droplets or packaged into chylomicrons before transport into the lymphatic system.8, 12, 13

The pathomechanism responsible for intestinal failure and PLE in DGAT1 deficiency has remained unclear. Through next-generation sequencing, we identified 10 additional patients from 6 unrelated pedigrees with 5 different, novel bi-allelic mutations in DGAT1 leading to severe, sometimes fatal course of PLE and fat intolerance. We took this unique opportunity to further shed light on the fundamental pathomechanisms of human DGAT1 deficiency.

Materials and Methods

Study Approval

The study was approved by the responsible local ethics committees (Ethics Commission of the Medical University of Vienna and Institutional Review Board of the University Medical Center Utrecht). All participants provided written informed consent for the collection of samples and subsequent analysis.

DNA Sequencing

Whole-exome sequencing was performed on patients as previously described.14, 15 Targeted panel sequencing was performed as previously described. Conventional Sanger sequencing was performed for validation and segregation analysis of variants.

Cell Culture

Organoids were generated from duodenal biopsies that were obtained from 3 healthy controls and 3 patients during duodenoscopy for diagnostic purposes, as described in detail in the Supplementary Materials and Methods. The healthy controls were patients suspected of celiac disease or inflammatory bowel disease, who did not show abnormalities on endoscopic and histological examinations.

Caco-2 cells and patient-derived fibroblasts were cultured in Dulbecco’s modified Eagle’s medium with/without GlutaMax and high glucose (Life Technologies, Carlsbad, CA) supplemented with 10% heat-inactivated fetal bovine serum (GE Health Care, Little Chalfont, UK), 100 U/mL penicillin (Gibco, Waltham, MA), 100 μg/mL streptomycin (Gibco), and 1 mM sodiumpyruvate (Gibco) at 37°C and 5% CO2. Patient-derived Epstein-Barr virus B lymphoblastoid cell line (B-LCL) was maintained in RPMI 1640 with 10% heat-inactivated fetal bovine serum, 100 U/mL penicillin (Gibco), and 100 μg/mL streptomycin (Gibco).

CRISPR/Cas9 Knockout of DGAT1

Plasmid constructs for the expression of DGAT1 single-guide RNAs (sgRNA) and Cas9 nuclease were generated as previously described and outlined in Supplementary Material and Methods. Three DGAT1 sgRNAs targeting exon 7 and 1 sgRNA targeting intron 6-7 were designed. Two different sgRNA (mix)-plasmids were used for transfection: sgRNA#2 and a mix of sgRNA#6, sgRNA#7, and sgRNA#8 (sgRNA#678). Hygromycin-resistance was achieved by co-transfection with the PiggyBac (PB) Transposon System (plasmids PB-Hygromycin and PB-Transposase were kindly provided by Bon-Kyoung Koo).

Transfection of healthy intestinal organoids was performed by electroporation, as described previously and extensively in Supplementary Materials and Methods.

Lipid Droplet Assays

Oleic acid (OA) was conjugated to bovine serum albumin (BSA) as described in Supplementary Materials and Methods. Organoids were grown in expansion medium (EM) on black clear-bottom 96-well imaging plates (Corning Life Sciences, Corning, NY). On day 6, the organoids were incubated with 1 mM OA/BSA for 17 hours in presence or absence of 0.1 μM DGAT1 inhibitor (AZD 3988; Tocris, Bristol, UK). Organoids were then fixed in 4% formaldehyde for 30 minutes at room temperature. Cells were washed in phosphate-buffered saline (PBS) and stained with 0.025 mg/mL LD540 and 4′,6-diamidino-2-phenylindole (DAPI) (Sigma-Aldrich, St Louis, MO) in PBS for 15 minutes at room temperature in the dark. Cells were washed and stored in PBS. Imaging of the organoids was performed using a Leica (Wetzlar, Germany) SP8X laser-scanning confocal microscope outfitted with a white light laser. The acquired stacks were processed and analyzed with Fiji/ImageJ (National Institutes of Health, Bethesda, MD)20, 21; shown are maximum projections of approximately 15-μm stacks.

For flow cytometry analysis, organoids were grown and treated with OA/BSA in the same manner as for the confocal analysis. After overnight incubation with 1 mM OA/BSA, the cells were harvested by pipetting and dissociated using TrypLE Express (ThermoFisher, Waltham, MA) until single cells were acquired. The cells were then fixed and stained in the same manner as for the confocal analysis and assayed using a BD FACS Canto II (BD Biosciences, San Jose, CA).

Fibroblasts were seeded at 6 × 104 cells in a 6-well plate and allowed to adhere overnight. Cells were then treated with OA positive control from the Lipid Droplet Fluorescence Assay (#500001; Cayman Chemical, Ann Arbor, MI) at 1:4000 dilution for 24 hours. Cells were fixed and stained according to the manufacturer’s protocol and assayed using BD FACS Fortessa (BD Biosciences, Franklin Lakes, NJ).

Cloning and Retrovirus Production

DGAT1 (ENST00000528718.5) and DGAT2 (ENST00000228027.11) complementary DNAs (cDNAs) were amplified by polymerase chain reaction from HEK293 cDNA library and cloned into pDONR221 using BP reaction according to the manufacturer’s protocol (Thermo Fisher). LR reaction was performed into pFMIG for wild-type (WT) DGAT1 and DGAT2 with N-terminal Streptavidin-HA tag.

Quantitative Real-Time Polymerase Chain Reaction

RNA was isolated from Caco-2 cells or organoids grown in either EM or differentiation medium (DM) for 5 days, used to synthesize cDNA by using the iScript cDNA synthesis kit (Bio-Rad, Hercules, CA) and amplified with SYBR green supermix (BIO-Rad) in a Light Cycler96 (BIO-Rad) according to the manufacturer’s protocol. Details on analysis and primers are given in Supplementary Materials and Methods.

Western Blotting

Cell lysates were generated and Western blotting was performed as described in the Supplementary Materials and Methods.

Thin Layer Chromatography

Organoids were grown in EM for 7 days and then prepared for Folch extraction as described previously and in the Supplementary Materials and Methods. Thin layer chromatography was performed by spotting the isolated lipid phase on aluminium-backed silica plates (Merck Millipore, Burlington, MA). As reference samples 1,3-dipentadecanoin (DG, C15:0/-/C15:0), tripentadecanoin (TG, C15:0/C15:0/C15:0) and triheptadecanoin (TG, C17:0/C17:0/C17:0) were included. The plates were then developed in a mobile phase of hexane:diethylether:acetic acid (60:15:2). The lipid bands were visualized by spraying the plates with a solution of 10% CuSO4 (wt/vol) in 10% H2SO4 (vol/vol) and subsequently heating the plates to 120°C for 30 minutes, as described previously to calculate DG/TG ratios, band intensities were quantified by Fiji/ImageJ.20, 21

Lipotoxicity Assays

For the propidium iodide staining, organoids were grown and incubated with varying concentrations of OA as described for the lipid droplet confocal assay. The organoids were then washed with Hank’s balanced salt solution (HBSS) (Gibco), and stained with Hoechst 1 μg/mL (Sigma-Aldrich) and propidium iodide 0.1 mg/mL (ThermoFisher) in HBSS at room temperature for 15 minutes. Organoids were imaged by an inverted Olympus IX53 epifluorescence microscope (Tokyo, Japan).

For the Caspase-Glo 3/7 assay (Promega, Madison, WI), organoids were grown on TC-treated 96-well plates (Greiner, Kremsmunster, Austria) and incubated with OA as was done for the propidium iodide staining. Organoids were washed and resuspended in HBSS and transferred to a white-walled 96-well plate (Greiner). The assay was performed according to the manufacturer’s protocol and luminescence was measured on a Tristar 2 luminometer (Berthold Technologies, Oak Ridge, TN).

Statistical Analysis

Data are presented as mean ± SD. Experiments were performed with a minimum of 3 replications. Statistical significance was determined at P ≤ .05 using 2-way analysis of variance with Tukey’s multiple comparison test, a Mann-Whitney U test, or a Student t test where appropriate. Significance is indicated as P ≤ .05 (*), P ≤ .01 (**), or P ≤ .001 (***)or P < .0001 (∗∗∗∗).

Results

Clinical Phenotype

We investigated 10 patients from 5 consanguineous families and 1 family of unknown consaguinity with unaffected parents. All of the patients had a history of intestinal failure due to congenital diarrhea and/or vomiting, resulting in failure to thrive. (Figure 1A, Table 1, Supplementary Table 1, see Supplementary Materials and Methods for clinical details). These 10 patients come from 4 Turkish families originating from Turkey and 2 Caucasian families from The Netherlands. In summary, 8 of 10 cases showed early-onset PLE characterized by hypoalbuminemia, hypogammaglobulinemia, and intractable diarrhea, with 1 patient developing marked steatorrhea. Patients 7 and 8 presented with severe vomiting only, which resulted in failure to thrive in the older sibling. Food containing fat induced abdominal pain and vomiting soon after ingestion and serum lipid profiles of the 10 patients were variable (Supplementary Table 2). Some patients had normal levels of serum TG, whereas 1 patient exhibited hypertriglyceridemia. Most showed a reduced level of high-density lipoprotein, with normal levels of low-density lipoprotein, very low-density lipoprotein, and cholesterol.

Figure 1

Pedigrees, mutations, and genetic location of 6 families with DGAT1 deficiency. (A) Pedigrees of families with DGAT1 deficiency and chromatograms showing mutation in affected patients. Filled shapes indicate affected individuals, half-filled are heterozygous for mutation indicated, and empty shapes indicate WT. (B) Exonic scheme of DGAT1 showing mutations identified in this study in black and previously identified mutations in red.3, 4, 5, 6

Table 1

Patient Characteristics

Patient ID	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10
Demographics
Current age and gender	4.5 y, M	Deceased, F	2 y, F	8 y, M	2 y, M	6 y, M	14 y, M	17 y, M	10 y, F	10 y, F
Country of origin and ethnicity	Turkey, Turkish	Turkey, Turkish	Turkey, Turkish	Turkey, Turkish	Turkey, Turkish	Turkey, Turkish	The Netherlands, Caucasian	The Netherlands, Caucasian	The Netherlands, Caucasian	The Netherlands, Caucasian
Age of clinical onset	Birth	Birth	3 wk	2 mo	40 d	2.5 mo	First month	First month	Birth	Birth
Disease manifestations
Failure to thrive	Yes	Yes	Yes	Yes	Yes	Yes	No	Yes	Yes	Yes
Vomiting	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Diarrhea	Yes, fatty	Yes	Yes	Yes	Yes, bloody and watery	Yes, watery	No	No	Yes	Yes
Hypoalbuminemia	Yes	Yes	Yes	Yes	Yes	Yes	No	No	Yes	Yes
Hypogammaglobulinemia	Yes	Yes	Yes	Yes	Yes	Yes	ND	ND	Yes	Yes
Edema	Yes	Yes	Yes	No	No	No	No	No	No	No
GI examinations (endoscopy and imaging)	Normal	Normal	Normal	Normal	ND	Normal	Normal	Normal	Normal	Normal
Hematoxylin-eosin and electron microscopy	Normal	Normal	Duodenal enterocytic lipid accumulation, microvilli are shortened and rarefied	Normal	ND	Focal vacuolization at one area and partially blunted villi	Normal histology (on fat-free diet)	Normal histology (on fat-free diet)	Misdiagnosis of atypical MVID based on CD10 positive globules on LM and laterally located microvilli on EM	Misdiagnosis of atypical MVID based on CD10 positive globules on LM and laterally located microvilli on EM
Extra-GI manifestations	Recurrent infections, otitis media	Recurrent infections	Corneal cystine crystal accumulation and intermittent metabolic acidosis	ND	Hepatomegaly and jaundice	Recurrent infections	Gilles de la Tourette syndrome	Gilles de la Tourette syndrome	Recurrent infections	Recurrent infections
Clinical course
Treatments	Fat-free formula and MCT, albumin infusions	Albumin infusions	Albumin, TPN	Cholestyramine	Creon pancreatic lipase, hydrolyzed formula	Creon pancreatic lipase	Monthly infusion of Intralipid and Omegaven suppletion of lipid-soluble vitamins	Monthly infusion of Intralipid and Omegaven suppletion of lipid-soluble vitamins	TPN and small bowel transplantation	TPN, recently started with fat-free formula
Outcome	Asymptomatic and normal growth with low-fat diet + MCT oil + fat-free formula	Patient passed away at 6 mo due to sepsis	Stool frequency once a day with Basic F formula feeding	Reduced stool volume and frequency on cholestyramine treatment	Stool frequency reduced on Creon treatment; weight and height are still below 3rd percentile	Stool frequency reduced on Creon treatment; after 2 y, symptoms resolved spontaneously	Enteral feeding without fat	Enteral feeding without fat	Tolerates enteral feeding, but still stunted	Tolerates enteral feeding, but still stunted

EM, electron microscopy; F, female; GI, gastrointestinal; LM, light microscopy; M, male; MCT, medium-chain triglyceride; MVID, microvillus inclusion disease; ND, not determined; TPN, total parenteral nutrition.

Supplementary Table 1

Genetic Description of DGAT1-deficient Patients in This Study and in Previous Reports

Patient	Position Chr. 8	Reference	Alternate	DGAT1 mutation	DGAT1 mutated protein	CADD score
1	145540731	C	T	c.1202G>A	p.W401X	38
2	145540731	C	T	c.1202G>A	p.W401X	38
3	145542123	CCT	CAGATGGGCAGGGTGGGATGGG	c.573_574delAGinsCCCATCCCACCCTGCCCATCT	-	25.7
4	145541252	C	T	c. 937-1G>A	-	24.9
5	145541233	T	TG	c.953insC	p. I319Hfs *31	35
6	145541233	T	TG	c.953insC	p. I319Hfs *31	35
7	145541968	TAGG	T	c.629_631delCCT	p.S210_Y211delinsY	18.9
8	145541968	TAGG	T	c.629_631delCCT	p.S210_Y211delinsY	18.9
9	145541968	TAGG	T	c.629_631delCCT	p.S210_Y211delinsY	18.9
10	145541968	TAGG	T	c.629_631delCCT	p.S210_Y211delinsY	18.9
Haas et al.	145541756	A	G	c.751+2T>C	p.A226_R250del	22.9
Stephen et al. (Fam 1)	145541756	A	G	c.751+2T>C	p.A226_R250del	22.9
Stephen et al. (Fam 2)	145541457	A	G	c.884T>C	p.L295P	31
Gluchowski et al.	145542706	A	G	c.314T>C	p.L105P	29.3
Ratchford et al.	145541074	TAGA	T	c.1013_1015delTCT	p.F338del	22.3
	145540567	C	G	c.1260C>G	p.S420R	31

CADD, combined annotation dependent depletion; Chr, chromosome.

Supplementary Table 2

Serum Lipid Values of DGAT1-deficient Patients in This Study

Patient	Age on test date	Triglyceride (mg/dL)	HDL (mg/dL)	LDL (mg/dL)	VLDL (mg/dL)	Cholesterol (mg/dL)
1	5 mo	135 (<150)	25a (40–60)	12 (<110)	27 (<30)	63 (<170)
2	5 mo	84 (<150)	16a (40–60)	28 (<110)	16 (<30)	62 (<170)
3	11 mo	370a (<200)	9.5a (40–60)	23 (<110)	ND	96 (<170)
4	8 y	123 (<180)	Normal	110a (<110)	ND	181a (<170)
5	ND	ND	ND	ND	ND	ND
6	2 y	184 (<200)	21a (40–60)	13 (<110)	36a (<30)	59 (<170)
7	9 y	142 (<180)	32a (40–60)	ND	ND	139 (<170)
8	12 y	88 (<180)	30a (40–60)	ND	ND	112 (<170)
9	22 mo	124 (<180)	24a (40–60)	ND	Normal	139 (<170)
10	20 mo	165 (<180)	ND	ND	ND	128 (<170)

NOTE. Parentheses indicate normal values.

HDL, high-density lipoprotein; LDL, low-density lipoprotein; ND, not determined; VLDL, very low-density lipoprotein.

Indicates abnormal value

Endoscopy was performed on patients 1 to 3 and 7 to 10, which showed no macroscopic abnormalities of the duodenum and colon (Supplementary Figures 1 and 2). Histopathology of a duodenal biopsy from patient 1 showed marked flattening of the villi (Figure 2 and Supplementary Figure 2) and patient 3 showed marked shortening of the villi (Supplementary Figure 1A), Electron microscopy of a duodenal biopsy of patient 3 showed lack of microvilli (Supplementary Figure 1A). Patients 7 and 8 showed normal pathology, although the biopsies were taken under fat-restricted diet (Supplementary Figure 1B). Duodenal biopsies from patient 10 showed lateral microvilli and cytoplasmic CD10 staining (Supplementary Figure 1C), which led to the misdiagnosis of atypical microvillus inclusion disease.

Supplementary Figure 1

Histologic and endoscopic characteristics of intestines of DGAT1-deficient patients. (A) Colonoscopy (left) and electron microscopy (EM, right) revealed normal intestinal structures in patient 1. (B) Periodic acid-Schiff and EM of patient 7 were obtained during fat-free diet and do not show abnormalities. (C) Patient 10 shows cytosolic CD10 staining and lateral microvilli on EM.

Supplementary Figure 2

Hematoxylin-eosin (H&E) and immunohistochemical staining for DGAT1 in (A) healthy controls, (B) patient 2, and (C) patient 1.

Figure 2

DGAT1 protein expression in patient-derived material. (A) Immunohistochemistry of DGAT1 in control and patient 1 and patient 2 ileal and duodenal biopsy, respectively. (B) Western blot showing lack of DGAT1 and DGAT2 in patient 3 fibroblast lysate, but normal expression of DGAT1 in healthy control fibroblasts. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as a loading control. (C) Western blot showing lower expression of potentially nonfunctional DGAT1 protein and normal DGAT2 protein level from Epstein-Barr virus–derived B lymphoblastoid cell line of patient 4, a parent, and a healthy control. GAPDH was used as a loading control. (D) Western blot for DGAT1 protein expression in undifferentiated (EM) and differentiated (DM) organoids from 2 healthy controls and patients 7 and 8. HSP90 was used as loading control. Results are representative of 3 independent experiments.

Eight patients received various treatments that led to resolution or significant improvement of the GI symptoms. Most patients were placed on a fat-restricted diet, which alleviated their GI symptoms. Addition of medium-chain TG was tolerated, and some patients were infused with intravenous essential fatty acid supplements Intralipid and Omegaven and fat-soluble vitamins. Patient 4 received cholestyramine and patients 5 and 6 received pancreatic lipase due to low fecal elastase level, both of which reduced daily stool frequency. Patients 9 and 10 are twins and patient 9 received an intestinal transplant. The various treatments and outcomes are summarized in Supplementary Table 3.

Supplementary Table 3

Compilation of Clinical Characteristics and Specific Treatment of DGAT1-deficient Patients in This Study

Patient ID	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10
Recurrent infections	Yes	Yes	No	No	No	Yes	No	No	Yes	Yes
Type of infections	Upper respiratory tract infections	NA	Central line–related fungal infection	NA	NA	Upper respiratory infections, recurrent otitis media	NA	NA	(1) Sepsis: Klebsiella and Streptococcus (2) Sepsis: Escherichia coli	(1) NA (2) Sepsis: Klebsiella (3) Sepsis: Bacillus cereus (4) Sepsis: Staphylococcus aureus
Treatment for infectious episodes	Antibiotics IV or PO	NA	Amphotericin B IV	NA	Prophylactic immunoglobulins IV	Antibiotics PO	NA	NA	Antibiotics IV	Antibiotics IV
Response to treatment for recurrent infection	Well responded	NA	NA	NA	Well responded	Well responded	NA	NA	Well responded	Well responded
Signs of liver disease	NA	NA	NA	Elevated liver enzymes	Jaundice and hepatomegaly.Elevated liver enzymes.Liver biopsy: hepatocanalicular and ductular cholestasis, paucity of bile ducts, mixed type; hepatosteatosis (10%), porto-portal fibrosis and fibrotic activity of 3/6.	Elevated liver enzymes	NA	NA	Elevated liver enzymes and jaundice.Toxic liver injury due to effects of long-term treatment with antifungal drugs (confirmed by liver biopsy).	Elevated liver enzymes and jaundice. Liver biopsy normal.
Fecal elastase 1 levels(normal range: > 500 μg/g)	ND	NA	ND	NA	130 μg/g	ND	NA	NA	>500 μg/g	>500 μg/g
Serum albumin level before treatment (normal range: >3.2 g/dL)	1.8 g/dL	1.8 g/dL	2.2 g/dL	2.8 g/dL	2.2 g/dL	2.59 g/dL	NA	4.35 g/dL	2.38 g/dL	3.2 g/dL
Fecal-alpha-1-antitrypsin level(normal range: <5 mg/g)	NA	ND	Low	ND	NA	NA	NA	NA	9.1 mg/g	7.6 mg/g
Specific treatment	Fat-free formula and medium chain triglycerideAlbumin IV	Albumin IV	Albumin IV, TPNFat-free formula and supplementation of medium chain triglyceride	Cholestyramine	Creon hydrolyzed formula	Creon	Monthly infusion of Intralipid and Omegaven supplementation of lipid-soluble vitamins	Monthly infusion of Intralipid and Omegaven supplementation of lipid-soluble vitamins	TPN and small bowel transplantation	TPN and fat-free formula
Clinical course after treatment	Significant improvement	Deceased	Significant improvement of diarrhea	Significant improvement;diarrhea stopped after treatment	Mild improvement. His daily stool frequency had reduced with the administrations Creon, but his weight and height are below 3rd percentile.	No improvement of failure to thrive, but significant improvement of diarrhea.After 2 years of age, his symptoms improved spontaneously. His weight is now 16 kg (3rd–10th percentile), height 104 cm (3rd percentile)	Significant improvement	Significant improvement	Completely resolved PLE. Tolerates polymeric enteral feeding.Still stunted and underweight by causes secondary to intestinal transplantation and several autoimmune problems.	NA

IV, intravenous; NA, not available; ND, not done; PLE, protein-losing enteropathy PO, per os; TPN, total parenteral nutrition.

Some patients developed extra-GI manifestations. In patients 1, 2, and 6, recurrent episodes of unspecific infections were recorded. Patients 9 and 10 had recurrent episodes of septicemia due to catheter-related blood stream infections. None of our patients showed any pattern of unusual or opportunistic infections that might be overtly associated with fecal loss of immunoglobulins or lymphopenia. Treatment of infections was successful with antibiotics, antifungal drugs, or prophylactic intravenous immunoglobulin, as outlined in Supplementary Table 3. Patient 5 had hepatomegaly and jaundice, and liver biopsy revealed fibrosis and cholestasis. Patients 7 and 8 were diagnosed with Gilles de la Tourette syndrome and were treated with dexamphetamine.

Identification of Novel DGAT1 Mutations

Whole-exome sequence analysis was performed in patients 1, 3, 7, 8, 9, and 10, while targeted panel sequencing was performed for patients 4 and 5. Variants in affected siblings (patients 2 and 6) were identified using conventional Sanger sequencing. Collectively, we identified 5 novel homozygous variants in the gene DGAT1 (Mendelian Inheritance in Man: 604900, GenBank: NM_012079.5). These variants segregated with the disease under the assumption of autosomal recessive inheritance, with heterozygous carriers being unaffected (Figure 1A). The variants are neither present nor reported as rare heterozygous variants in the gnomAD database, and were predicted to be deleterious using Combined Annotation Dependent Depletion prediction tool (Supplementary Table 1).

In family 1, we identified a homozygous nonsense mutation at amino acid 401 leading to an early stop codon (c.1202G>A, p.W401X). In family 2, we identified a homozygous insertion deletion (c.573_574delAGinsCCCATCCCACCCTGCCCATCT) in exon 6 of DGAT1. In family 3, we identified a homozygous splice site acceptor mutation (c. 937-1G>A) preceding exon 12. In family 4, we identified a homozygous single base-pair insertion, leading to a frameshift and early stop codon (c.953insG, p.I319Hfs*31) in exon 12. In families 5 and 6, we identified a homozygous 3 base-pair deletion (c.629_631delCCT, p.S210_Y211delinsY) in exon 7 (Figure 1, Supplementary Figure 3). Altogether, we identified 5 novel disease-causing homozygous mutations in DGAT1 in 6 patients of Turkish origin and 4 of Dutch origin.

Supplementary Figure 3

Family pedigrees and Sanger sequencing histograms of DGAT1 deficiency. Filled shapes indicate affected individuals, half-filled are heterozygous for mutation indicated, and empty shapes indicate wild type.

Consequences of DGAT1 Mutations

We proceeded to study the consequences of DGAT1 mutation on available material. Immunohistochemistry on GI biopsies from patient 1 and patient 2 showed a lack of DGAT1 protein expression specifically in the epithelium of the duodenum, ileum, and colon (Figure 2A and Supplementary Figure 2), whereas DGAT1 protein was not detected in the gastric mucosa of control or patient material (Supplementary Figure 2). In patient 3, reverse-transcription polymerase chain reaction of cDNA from patient-derived fibroblast showed aberrant splicing (Supplementary Figure 4), which ultimately led to undetectable protein expression on Western blot (Figure 2B). In patient 4, Epstein-Barr virus–derived B-lymphoblastic cell line showed a highly reduced expression of DGAT1 (Figure 2C). Intestinal organoids derived from patients 7 and 8 showed normal mRNA levels on differentiation (data not shown), but protein was absent (Figure 2D). Similar data were obtained from patient 9 (data not shown). Together, we show that all the novel DGAT1 mutations identified led to aberrant protein expression.

Supplementary Figure 4

Defective mRNA splicing in patient 3. (A) Gel showing aberrant DGAT1 transcripts in mRNA isolated from patient 3 fibroblasts with and without a 4-hour puromycin treatment. * denotes aberrant transcript. ACTB was used as control.

To confirm that the DGAT1 c.629_631delCCT mutation specifically leads to reduced DGAT1 protein levels without affecting mRNA levels, Caco-2 cells were stably transfected with Flag-DGAT1 WT or Flag-DGAT1 c.629_631delCCT. Indeed, DGAT1 mRNA expression was similar in both cell lines (Supplementary Figure 5A), but DGAT1 protein was not detectable when DGAT1 was mutated (Supplementary Figure 5B). Incubation with the proteasome inhibitor MG132 shows that the loss of protein is at least partially due to proteasomal degradation of the mutant DGAT1 (Supplementary Figure 5B). To further investigate the increased proteasomal degradation of DGAT1 c.629_631delCCT, we determined the level of ubiquitination of the mutant protein. Therefore, Caco-2 cells were co-transfected with His-ubiquitin and Flag-DGAT1 WT or Flag-DGAT1 c.629_631delCCT and treated with MG132. A ubiquitin pulldown assay showed that ubiquitination of Flag-DGAT1 c.629_631delCCT was increased compared with ubiquitination of Flag-DGAT1 WT (Supplementary Figure 5C).

Supplementary Figure 5

DGAT1 c.629_631delCCT results in increased proteasomal degradation of DGAT1. Caco-2 cells were stably transfected with indicated constructs. (A) qRT-PCR analysis of DGAT1 mRNA expression, relative to HPRT1, in Caco-2 cells stably transfected with indicated constructs. Average and standard deviation of 3 independent experiments are shown. (B) Transfected Caco-2 cells were untreated or treated with 2 μM MG132 for 16 hours. Protein expression was determined by Western blot analysis using anti-Flag and anti-HSP90 antibodies. Results are representative of 3 independent experiments. (C) Caco-2 cells were transiently transfected with His-ubiquitin and indicated constructs and treated with 2 μM MG132 for 16 hours. Ubiquitination of DGAT1 was determined by a His pulldown and subsequent Western blot analysis using anti-Flag antibodies. Results are representative of 3 independent experiments.

Loss of DGAT1 Leads to Aberrant Lipid Metabolism

Free fatty acids (FFAs) can be processed for energy production through beta oxidation, or stored in the form of lipid droplets on its incorporation into TG. In enterocytes, this lipid droplet formation is required for the transport of long-chain FFAs into chylomicrons, before being excreted across the basolateral membrane of enterocytes. Recently, it was shown that DGAT1 mutant fibroblasts accumulated less lipid droplets when incubated with OA, an 18-carbon FFA. We hypothesized that DGAT1 deficiency will lead to aberrant lipid droplet formation in patient-derived cells.

We performed a staining for LD540, which binds neutral lipids,19, 26 on intestinal organoids from patients 7, 8, and 9 after 16 hours incubation with BSA-coupled OA. Using fluorescence imaging, we observed an increase in lipid droplet formation in healthy control organoids, which was significantly reduced in DGAT1 mutant patient-derived organoids from patients 7, 8, and 9 and DGAT1 knockout (DGAT1KO) organoids that were generated using CRISPR/Cas9 genome editing for DGAT1 in healthy control organoids (Figure 3A and Supplementary Figure 6 and Supplementary Table 4). Formation of lipid droplets was at least partially dependent on DGAT1, as the use of a selective DGAT1 inhibitor (DGAT1i) AZD 3988 on healthy control organoids resulted in a reduction of lipid droplet accumulation as shown by decreased LD540 staining (Figure 3A). This was further confirmed and quantified using a flow cytometry–based assay, in which healthy control organoids showed an increased granularity due to the accumulation of lipid droplets (side scatter area, SSC-A) and increased LD540 staining on incubation with OA. Absence of DGAT1 (patients or KO) or DGAT1 inhibition caused a significant reduction in granularity and LD540 staining compared with healthy controls (Figure 3B).

Figure 3

Loss of DGAT1 results in decreased lipid droplet formation in organoids. (A) Immunofluorescent images of 4′,6-diamidino-2-phenylindole (DAPI) (blue) and LD540 (yellow) staining of organoids from healthy control, DGAT1 mutant patient 8 (P8) and DGAT1KO organoids after 17-hour incubation with vehicle control (BSA), 1 μM OA, or 1 μM OA + 0.1 μM DGAT1 inhibitor (OA+DGAT1i). Representative images of 3 healthy controls, 3 patients (patients 7–9), and 3 CRISPR/Cas9 genome-edited DGAT1-knock-out (DGAT1KO) organoids. (B) Representative histograms of SSC and LD540 staining in organoids from controls, patients, and DGAT1KO organoids as described in (A). Upon OA stimulation, control organoids accumulate lipid droplets and show increased SSC and LD540, which was severely reduced in patient-derived and DGAT1KO cells. Mean fluorescence intensity (MFI) of SSC and LD540 was plotted for n = 3 per group. Statistical analysis was done using a 2-way analysis of variance with Tukey’s multiple comparison test. Mean ± SD is indicated; * P ≤.05, *** P ≤.001.

Supplementary Figure 6

Generation of DGAT1 knockouts in healthy human intestinal organoids. (A) Targeting strategy for the generation of DGAT1 knockout organoids using CRISPR/Cas9 genome editing. (B) Schematic view of insertion-deletion mutations of 3 DGAT1KO clones. Gray areas represent deleted sequences from the WT DGAT1 gene. (C) Western blot analysis of DGAT1 expression in DGAT1KO organoids compared with healthy control 1 (HC1) and patient 8 (P8). Actin, loading control.

Supplementary Table 4

sgRNAs Used to Target DGAT1 for CRISPR/Cas9 Gene Editing

sgRNA ID	Core sequence DNA	Position DGAT1 gene	Predicted cleavagea
sgRNA#2	GGCACCATGAGTTGACGTCG	Exon 7	13621-13622 (AC)
sgRNA#6	TGTGCGCCATCAGCGCCAGC	Exon 7	13570-13571 (TG)
sgRNA#7	GTGCGCCATCAGCGCCAGCA	Exon 7	13569-13570 (CT)
sgRNA#8	GTGGCTGGCCCGAGACAGAT	Intron 6–7	13486-13487 (CT)

sgRNA, single-guide RNA.

According NCBI Reference Sequence: NG_034192.1; Homo sapiens diacylglycerol O-acyltransferase 1 (DGAT1).

In addition, we incubated normal donor and patient 3–derived fibroblasts with OA and quantified lipid droplet formation by flow cytometry. Normal donor fibroblasts accumulated lipid droplets on incubation with OA, which was shown by increased granularity (SSC-A) and Nile Red staining, a dye that binds to neutral lipids such as TG. In contrast, patient-derived dermal fibroblast from patient 3 failed to accumulate lipid droplets, as they showed a lack of granularity (Figure 4A) and Nile Red staining (Figure 4B). In addition, we show that this lack of lipid droplet formation was not detected in B-LCL derived from patient 4 (Supplementary Figure 7), presumably due to the expression of DGAT2 in this cell type (Figure 2).

Figure 4

Loss of DGAT1 results in decreased lipid droplet formation in fibroblasts. (A) Left: Representative contour plots for forward (FSC-A) and SSC-A of patient 3 and 2 control fibroblasts with and without OA addition. Right: Mean SSC-A of 5 technical replicates. (B) Left: Representative histogram of Nile Red mean fluorescence intensity (MFI) of patients 3 and 2 control fibroblasts with and without OA addition. Right: MFI of 5 technical replicates. (C) Western blot showing retroviral-mediated delivery of exogenous DGAT1 and DGAT2 on patient 3 fibroblasts. (D) Mean SSC-A and (E) MFI of Nile Red staining on patient 3 fibroblasts reconstituted with empty vector (EV), WT DGAT1, or WT DGAT2. Statistical analysis was done using 2-way analysis of variance with Tukey’s multiple comparison test. Mean ± SD is indicated; **P< .01 or ****P < .0001.

Supplementary Figure 7

Lipid droplet staining in Epstein-Barr virus–derived B-lymphoblastic cell line (EBV B-LCL) of patient 4 with and without OA. (A) Mean SSC-A and (B) Nile Red mean fluorescence intensity (MFI) staining of lipid droplet stained EBV B-LCLs.

WT DGAT1 and DGAT2 Rescues Lipid Droplet Formation in DGAT1 Deficiency

We reconstituted fibroblasts from patient 3 with WT DGAT1 and DGAT2 protein using a retroviral delivery system. We successfully reconstituted DGAT1 protein expression, and overexpressed DGAT2 in these cells (Figure 4C). The DGAT1-reconstituted fibroblasts were able to incorporate OA into lipid droplets as seen by the increased granularity and increased Nile Red fluorescence (Figure 4D and E). The phenotype was also rescued by overexpression of DGAT2, resulting in restored granularity and Nile Red staining on addition of OA (Figure 4D and E). We concluded that reconstitution of DGAT1 rescues the altered lipid metabolism phenotype, and that DGAT2 can partially rescue altered lipid droplet formation in DGAT1-deficient fibroblasts.

Loss of DGAT1 Specifically Inhibits TG Formation

To substantiate the specificity of DGAT1 deficiency for causing aberrant TG formation, we performed thin layer chromatography to measure the levels of TG and DG in organoids derived from 3 healthy controls and patients 7, 8, and 9 (Figure 5A). By quantification of respective DG and TG band intensities, we determined that the TG/DG ratio in healthy control organoids was significantly higher compared with patient-derived organoids (Figure 5B). These results indicate a DGAT1-dependent loss of TG synthesis in patient-derived intestinal organoids, whereas levels of DG were comparable.

Figure 5

Loss of DGAT1 results in decreased triglyceride (TG) formation. (A) Organoids from 3 healthy controls and patients 7, 8, and 9 were grown in EM for 7 days. Lipid extracts were isolated and run by thin layer chromatography to determine levels of diacylglycerol (DG) and triglyceride (TG). Relative positions were indicated by reference samples for DAG and TG. (B) Ratio of intensity of TG over DAG for each sample. Statistical analysis was done using a Mann-Whitney U test. Mean ± SD is indicated, ∗P < .05.

Loss of DGAT1 Results in Increased Sensitivity to Lipid-induced Toxicity

Most DGAT1-deficient patients reported in this study suffered from PLE after ingestion of dietary lipids, which can be the result of either intestinal mucosal injury or lymphatic abnormalities. To assess whether exposure to lipids leads to mucosal injury in DGAT1-deficient patients, we performed lipotoxicity assays in healthy control and DGAT1-deficient organoids. We incubated the organoids with varying concentrations of BSA-coupled OA in EM and assessed cell death by brightfield microscopy and propidium iodide staining (Figure 6A). Remarkably, we observed 100% cell death at 4 mM OA in patient-derived organoids, whereas cell death was still almost absent at 6 mM OA in control organoids. To further quantify these findings, we determined lipid-induced caspase-mediated cell death using a Caspase-Glo 3/7 assay. We show that DGAT1-deficient cells are more sensitive to lipotoxic stress compared with healthy control organoids and undergo programmed cell death on treatment with OA (Figure 6B). By nonlinear regression analysis, the median lethal dose was determined to be approximately 7 mM OA for healthy control organoids and 4 mM OA for patient-derived organoids (P < 0.001). Overall, these results indicate that DGAT1-deficient organoids are more susceptible to lipid-induced toxicity, which may reflect the clinical feature of PLE in DGAT1-deficient patients that occurs on ingestion of fat.

Figure 6

Loss of DGAT1 results in increased sensitivity to lipotoxic stress. Organoids of 3 healthy controls and 3 patients (7–9) were grown in EM and incubated overnight with a range of oleic acid (OA) concentrations. (A) Representative images showing brightfield and propidium iodide staining for cell death of the organoids after incubation with OA. (B) Caspase-Glo 3/7 assay for apoptotic cells after incubating organoids with OA. Samples were normalized for vehicle control values and the maximum value for each sample was set to 100% assay response. Median lethal dose (LD50) was calculated by regression analysis. Statistical analysis on LD50 values was done using a Student’s t test, ***P ≤ .001.

Discussion

Lipid metabolism is an important physiological function within the human body that includes the digestion and absorption of lipid products from food. Inborn errors of lipid metabolism can result in a wide range of symptoms, from neurological impairment to hypertriglyceridemia. Recently, lipid metabolism disorders have been linked to CDD, such as in the case of Niemann-Pick disease type C with inflammatory bowel disease due to impaired autophagy. In the case of cytoplasmic TG metabolism disorders, none of the other deficiencies have been reported to develop any GI phenotype.

Previous studies on DGAT1 deficiency have identified a total of 3 distinct homozygous and 2 compound heterozygous mutations in DGAT1.3, 4, 5, 6 These patients suffered from severe congenital diarrhea and PLE, clinical features that are shared with most of our patient cohort. Although this shared phenotype further confirms the involvement of DGAT1 in intestinal failure, limited functional data or potential therapeutic options have been reported thus far.

Currently, there is no genotype-phenotype correlation in DGAT1 deficiency, as patients develop varied clinical history ranging from complete resolution of GI symptoms to a lethal course of disease. As previously described and also observed in our study, discordant phenotypes associated with identical mutations in siblings and in different, unrelated families are of interest. Whether this phenomenon, as well as the novel clinical features, such as normal serum TG and hepatic involvement, observed in our patients indeed extend the spectrum of the clinical phenotype of DGAT1 deficiency, or are unspecific secondary effects of treatment, cannot be inferred from our study. By analogy, the unsatisfactory clinical outcome described for patient 9 after the small bowel transplantation may be attributed to known posttransplant complications rather than unsuccessful correction of the DGAT1 deficiency in the intestine.

To provide further insight into the pathomechanism of DGAT1 deficiency, we have produced cell models that recapitulate an altered lipid metabolism in vitro. Here, we show that patient-derived organoids can recapitulate the molecular pathomechanism of DGAT1 deficiency. We have demonstrated aberrant lipid metabolism as evidenced by reduced lipid droplet and TG formation on incubation with OA in both patient-derived organoids and fibroblasts. In addition, we show for the first time that DGAT1-deficient cells are more susceptible to lipid-induced toxicity, which provides a plausible explanation for the clinical PLE symptoms in DGAT1 deficiency and possibly other forms of PLE, such as primary intestinal lymphangiectasia. The fact that dietary fat restriction can even restore normal fecal protein clearance in PLE further supports the concept that cellular lipotoxicity may be one of the driving forces for ongoing fecal protein loss in untreated PLE. Whether this lipotoxicity-induced enterocyte dysfunction is caused by endoplasmic reticulum stress or induced autophagy, which has been implicated in lipid metabolism disorders such as Niemann-Pick disease type C with inflammatory bowel disease, remains to be determined. Moreover, elucidation of the precise role of DGAT1 in common lipotoxicity-related diseases, such as type 2 diabetes, nonalcoholic fatty liver disease, and metabolic cardiomyopathy, may be the first step toward new therapy strategies for obesity-related disorders.

Of note, the novel technology of patient-derived intestinal organoids provided an unprecedented look into pathobiology of the cells in the GI tract. In the future, a more systematic use and collection of organoids from potential DGAT1 deficiency and other patients with intestinal failure will allow for better dissection of the disease mechanism involving the gut epithelium.

As described in this study and in previous literature, DGAT1-deficient patients usually do well on a fat-restricted diet. An early introduction of such a diet might even have prevented the development of a full-blown PLE in patients 7 and 8. In addition, the administration of short chain fatty acid proves to be a good supplement to the diet for some patients. Intravenous administration of essential fatty acid was also well-tolerated, presumably as it bypasses absorption through the gut epithelium as well. Whether patients benefit from novel ways of treatment with available drugs, such as cholestyramine and pancreas enzymes, as empirically applied for patient 4 and patients 5 and 6, respectively, would have to be attempted and evaluated in additional patients and in carefully designed trials. Before treatment, neither diagnosis of bile acid diarrhea nor exocrine pancreas insufficiency was formally tested or firmly established, because the observation of slightly decreased fecal elastase levels may be unspecific due to fecal dilution in severe chronic diarrhea. Although the course of disease evidently can be mild, early diagnosis of DGAT1 remains vital, as all patients were severely ill during the first few months of life and thus far one patient suffered from a lethal course of disease. Additionally, the impact on the quality of life during chronic treatment (monthly infusions or through a central line that includes risks of infections) and the variable genotype-phenotype relationships of DGAT1 patients should not deter the search for new therapeutic options for these patients.

In pursuit of possible therapeutic strategies, we determined that proteasome inhibitors could provide a potential therapeutic option in case of proteasome-mediated protein degradation, as shown for the mutation in families 5 and 6 (Supplementary Figure 4). In addition, we determined that DGAT2 might be able to compensate for the lack of DGAT1 function. DGAT2 is an isozyme of DGAT1 that does not share any homology in protein sequence or domains. However, DGAT2 shares functional characteristics with DGAT1, catalyzing the formation of TG from DG and fatty acyl-CoA in the liver. In this study, we show that DGAT2 might be able to compensate for this phenotype, as shown by OA addition on DGAT2-expressing B-LCL and exogenous expression of DGAT2 in DGAT1-deficient fibroblasts. A single heterozygous, autosomal dominant DGAT2 mutation has been described in a family with Charcot-Marie-Tooth syndrome, but no GI involvement was reported, possibly because DGAT2 is not highly expressed in the human intestine. Therapeutic strategies to induce DGAT2 expression might potentially provide an additional, viable treatment strategy in DGAT1 deficiency.

Previous studies performed with Dgat1 mice implicated a beneficial role of DGAT1 inhibition in obesity through changing metabolic landscapes.33, 34, 35 These studies resulted in the development of DGAT1 inhibitors for use in human obesity. However, participants in a clinical trial of DGAT1 inhibitor Pradigastat developed side effects of diarrhea and nausea,37, 38 similar to the phenotype of DGAT1 deficiency. Dgat1 mice did not develop a GI phenotype, presumably due to the intestinal expression of DGAT2 and diacylglycerol transacylase, which might compensate for the lack of DGAT1. Our data suggest that in humans, lack of TG formation and packaging is detrimental in the context of GI epithelium, and that a serious note of caution on the clinical use of DGAT1 inhibitors should be provided. This finding accentuates the lack of available knowledge of intestinal lipid uptake and metabolism, and emphasized how the use of clinically relevant in vitro systems can help to fill this gap.

In summary, we here described a large cohort of DGAT1-deficient patients, and extended our knowledge of the pathomechanism of DGAT1 deficiency. Our findings expand the differential diagnosis of vomiting and congenital diarrhea in neonates. Clinicians should maintain a high suspicion for DGAT1 deficiency in cases of unexplained vomiting, especially when associated with failure to thrive and PLE, and could consider a fat-free diet with supplementation of essential fatty acids and fat-soluble vitamins as a first line of therapy. In conclusion, the findings described in this article show that DGAT1 mutations not only cause congenital diarrhea and PLE, but are also linked to fat intolerance.