TM6SF2 Rs58542926 Impacts Lipid Processing in Liver and Small Intestine

TM6SF2 rs58542926 impacts lipid processing in liver and small intestine

Elizabeth A. O’Harea*, Rongze Yanga*, Laura YergesArmstronga, Urmilla Sreenivasana Rebecca McFarlanda, Carmen C. Leitcha Meredith H. Wilsonc, Shilpa Narina a. Alexis Gorden a,b, Kathy Ryana, Alan R. Shuldinera, Steve A. Farberc, G. Craig Wood d, Christopher D. Still e, Glenn S. Gerhard d,e, Janet D. Robishaw d, Carole Sztalryda,f, **, and Norann A. Zaghloula, **

aDivision of Endocrinology, Diabetes and Nutrition, Department of Medicine, University of Maryland School of Medicine, Baltimore, MD 21201, USA

b Current affiliation: Mid Atlantic Permanente group, Largo, MD 20774, USA cCarnegie Institution for Science, Department of Embryology, Baltimore, MD 21218, USA.

dGeisinger Clinic, Geisinger Obesity Research Institute, Danville PA 17822, USA eCurrent affiliation: Department of Medical Genetics and Biochemistry, Temple University School of Medicine, Philadelphia, PA 19140, USA fBaltimore VA Medical Center, VA Research Service, Geriatric Research, Education and Clinical Center (GRECC) and VA Maryland Health Care System, 10N Green Street Baltimore 21201, USA

Keywords: triglyceriderich lipoproteins, missense mutation, endoplasmic reticulum stress, cytosolic lipid droplets, nonalcoholic fatty liver disease

This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process which may lead to differences between this version and the Version of Record. Please cite this article as doi: 10.1002/hep.29021

Footnotes

*These Authors Contributed Equally

**Authors for Correspondence:

Norann A. Zaghloul Carole Sztalryd

660 W. Redwood Street 660 W. Redwood Street

Howard Hall 487 Howard Hall 445A

Baltimore, MD 21201 Baltimore, MD 21201

Phone: 4107061646 Phone: 4107064047

Fax: 4107061622 Fax: 4107061622

Email: [email protected] Email: [email protected]

***List of Abbreviations:

TM6SF2

transmembrane 6 superfamily member 2

TRLs

triglyceriderich lipoproteins

ACDRP

Amish Complex Disease Research Program

NAFLD

Nonalcoholic fatty liver disease

2 Hepatology

CVD

Cardiovascular disease

E167K

Glutamate with lysine at residue 167

ApoB

apolipoprotein B

triglyceride

LDL-C

lowdensity lipoprotein cholesterol

NASH

nonalcoholic steatohepatitis

endoplasmic reticulum

shRNA

short hairpin RNA

VLDL

very low density lipoprotein

VLDL-C

very low density lipoprotein cholesterol

EBCT

electronbean computerized tomography

RNA

ribonucleotide acid

CLD

cytosolic lipid droplet

3 Hepatology

TEM

transmission elctron microscopy

BMI

Body mass index

ALT

Alanine transaminase

ALKP

Alkaline phohatase

HDL-C

high density lipoproteincholesterol

IDL

Intermediate density lipoprotein

Morpholino antisense

dpf

day postfertilization

mRNA

messenger ribonucleic acid

gRNA

guide RNA

ApoB-TRL

apoBtriglyceride rich lipoprotein

LDL

Low density lipoprotein

SEM

Standard error of mean

4 Hepatology

non significant

foigr

foie gras

DGAT2

AcylCoA:diacylglycerol acyltransferase 2

HFD

High fat meal

LDs

Lipid droplets

OOA

Older Order Amish

****Financial Support:

This research was supported by R01DK102001 (NAZ), T32AG000219 (EAO),

RO1HL121007 (EAO), F32DK19592 (WIL), RODK1093399 (FAB), the MidAtlantic

Nutrition Obesity Research Center (P30DK072488) and the Geriatric Research,

Education and Clinical Center, Baltimore Veterans Affairs Health Care Center.

5 Hepatology

Abstract:

The transmembrane 6 superfamily member 2 (TM6SF2) lossoffunction variant,

rs58542926, is a genetic risk factor for nonalcoholic fatty liver disease and progression to

fibrosis, but is paradoxically associated with lower levels of hepaticderived triglyceriderich

lipoproteins (TRLs). TM6SF2 is expressed predominately in liver and small intestine, sites for

triglyceride rich lipoprotein biogenesis and export. In light of this, we hypothesized that TM6SF2

may exhibit analogous effects on both liver and intestine lipid homeostasis. To test this, we

genotyped rs58542926 in 983 bariatric surgery patients from the Geisinger Medical Center for

Nutrition and Weight Management, Geisinger Health System (GHS) in PA and from 3,556 study

participants enrolled in the Amish Complex Disease Research Program (ACDRP). Although

these two cohorts have different metabolic profiles, carriers in both cohorts had improved fasting

lipid profiles. Importantly, following a high fat challenge, carriers in the ACDRP cohort exhibited

significantly lower postprandial serum triglycerides suggestive of a role for TM6SF2 in the small

intestine. To gain further insight into this putative role, effects of TM6SF2 deficiency were

studied in a zebrafish model and in cultured human Caco2 enterocytes. In both systems

TM6SF2deficiency resulted in defects in small intestine metabolism in response to dietary lipids

including significantly increased lipid accumulation, decreased lipid clearance and increased

endoplasmic reticulum stress. Conclusions: These data strongly support a role of TM6SF2 in

regulation of postprandial lipemia potentially through a similar function for TM6SF2 in the

lipidation and/or export of both hepatically and intestinallyderived TRLs.

6 Hepatology

Nonalcoholic fatty liver disease (NAFLD) is highly prevalent among obese and insulin resistant individuals who are also at risk of developing elevated circulating lipids and cardiovascular disease (CVD) (1, 2). Recently, however, a paradoxical increased susceptibility to NAFLD with low circulating lipids was found in carriers of a loss of function variant, rs58542926, in the transmembrane 6 super family member 2 gene (TM6SF2) (35). Given that dyslipidemia is a major contributor to atherosclerosis and CVD (68), this paradox offers the potential to define mechanisms by which hepatic steatosis can be dissociated from increased

CVD risk factors. Understanding the biological function of TM6SF2 may offer critical insight into the etiology of both pathologies.

The TM6SF2 rs58542926 variant is a cytosine to thymine substitution at coding nucleotide 499, resulting in a glutamate to lysine substitution at residue 167 (E167K). While overproduction of apolipoprotein B (apoB)containing lipoproteins by the liver and intestine is one hallmark of obesityassociated NAFLD and an independent risk factor of CVD, carriers of

TM6SF2 E167K have increased liver lipid stores and unexpectedly lower serum triglyceride

(TG), lower lowdensity lipoproteincholesterol (LDLC), and total cholesterol. They also appear protected from CVD despite a higher risk of nonalcoholic steatohepatitis (NASH) (35, 913).

TM6SF2 is localized mainly in the endoplasmic reticulum (ER) membrane and is predominantly expressed in liver and small intestine, two organs involved in packaging and secretion of dietary lipids in TGrich lipoproteins (TRL) and regulation of whole body lipid homeostasis (3). Previous mechanistic studies focused primarily on the impact of TM6SF2 lossoffunction in liver. These reports confirmed the lossoffunction nature of the E167K variant, demonstrating that it results in reduced levels of the protein compared to wild type TM6SF2 in cultured hepatocytes (3).

Moreover, knockdown of Tm6sf2 in mice with short hairpin RNA (shRNA) increased liver TG content and decreased very low density lipoprotein (VLDL) secretion, offering a hepatic

7 Hepatology

mechanism underlying cooccurrence of NAFLD and reduced circulating lipids (3). The role of

intestinal TM6SF2, however, remains unexplored.

The major objective of this study was to investigate whether TM6SF2 may impact

intestinal lipid homeostasis similar to its role in liver. Specifically, we tested the hypothesis that

TM6SF2 lossoffunction results in increased lipid accumulation and decreased lipid secretion in

the small intestine. Given the association of E167K in humans with NAFLD and progression to

NASH (3, 11, 1315), we also characterized potential subcellular mechanisms by which

TM6SF2 lossoffunction contributes to both. We evaluated TM6SF2 lossoffunction on ER

stress due to ER localization of the endogenous protein (5) and the crucial role of ER stress in

development of steatosis and progression to NASH (16). We performed association studies for

lipid traits in the Geisinger Health system (GHS) bariatric surgery and in the Amish Complex

Disease Research Program (ACDRP) cohorts followed by in vivo and in vitro mechanistic

studies in zebrafish and cultured human enterocytes, respectively. Our findings confirmed the

association of E167K with NAFLD and NASH in the GHS cohort and with improved lipid profiles

in the GHS and ACDRP cohorts despite wide differences in energy homeostasis status between

these two cohorts. Upon further examination of the association of circulating lipids with

TM6SF2 lossoffunction in the ACDRP cohort we found evidence for reduced fasting and

postprandial lipemia after a high fat challenge, suggesting a role for TM6SF2 in regulating

intestinal postprandial lipid homeostasis. We then modeled TM6SF2 lossoffunction in vivo

using larval zebrafish, an optically transparent model useful for whole animal studies of lipid

metabolism (17, 18) and cultured human Caco2 enterocytes (19). Our results indicated that

loss of TM6SF2 function leads to lipid accumulation in enterocytes, similar to hepatocytes, as

well as a similar increase in ER stress in both tissues. Together, these findings suggest that

TM6SF2 regulates both hepatic and intestinal lipid and ER homeostasis,.

8 Hepatology

Experimental Procedures

Study Participants:

The TM6SF2 rs58542926 variant was genotyped in 983 bariatric surgery patients in the GHS

cohort and 3556 participants in the ACDRP cohort. Tests of association were carried out for

effect of E167K on serum lipids, liver histology (GHS cohort only) and hepatic fat content by

electronbeam computerized tomography (EBCT; ACDRP cohort only). A high fat feeding

intervention was performed in the ACDRP cohort to evaluate effect of E167K on fasting and

postchallenge lipid traits (20). Institutional review board approval was obtained from the

Geisinger Clinic and University of Maryland and all subjects provided written informed consent.

Zebrafish Experiments

Zebrafish embryos were microinjected with morpholino antisense (control, tm6sf2, or dgat2)

alone or with human TM6SF2 RNA. Larvae were assessed for morpholino efficacy, gene expression, LDLC levels, hepatic steatosis via oil red O staining, and ER dilation and cytosolic lipid droplet (CLD) number and size via transmission electron microscopy (TEM) in liver and intestine. Zebrafish care and experimental procedures were carried out in accordance with the

Animal Care and Use Committees of the University of Maryland.

Cell culture experiments

Caco2 cells from ATCC (Manassas, VA) were treated with lentiviral short hairpin RNA (shRNA) targeting TM6SF2 and assessed for CLD accumulation by BODIPY staining and confocal imaging analysis (21). TG synthesis and secretion were determined by [3H]oleic acid pulse chase experiments (22) and [3H]oleic acid incorporated into TG or secreted into culture media as previously described (21).

All experimental details and statistical analyses are provided in the online supplemental material

9 Hepatology

Results:

Association of TM6SF2 rs58542926 with improved fasting and postprandial lipid profiles in

humans:

We first examined the relevance of TM6SF2 function to NAFLD in a large bariatric

surgery cohort at the Geisinger Medical Center (Danville, PA) (total N=983) (Supplemental

Table 1). Individuals with at least one copy of the rs58542926 variant (TM6SF2; c. C>T; p.

Glu>Lys; N=130) exhibited a higher average steatosis grade of 1.40±0.08 versus 1.14±0.03 in

noncarriers (N=853; p=0.04; Supplemental Table 2). In addition, 43.8% of T allele carriers

exhibited lobular inflammation versus 32.4% carrying the major allele (p=0.02; Supplemental

Table 2). Consistent with previous reports (12, 13), rs58542926 was associated with perivenular

fibrosis in 26.6% of carriers versus 17.5% of noncarriers (p=0.008; Supplemental Table 2).

These data suggest that rs58542926 is associated with increased liver steatosis and increased

risk for progression to NASH in extreme obesity. Despite the dyslipidemia typically associated

with obesity, fasting lipid levels, total cholesterol, TG and LDLC were also lower in carriers of

the T allele and statistically significant only in the subgroup also exhibiting NAFLD (p=0.02 for

total cholesterol and p=0.004 for TG; Supplemental Table 1 and Table 3).

The rs58542926 minor allele frequency (MAF=6.7%) in the GHS bariatric surgery cohort

translated to a small number of homozygous participants (n=3), hindering our ability to assess

the impact of carrying two copies of the T allele for this variant. To address this, we analyzed a

large founder population cohort from the ACDRP (N=3556), inclusive of nonobese subjects, in

which the rs58542926 T allele is significantly enriched (MAF=12%) compared to outbred

European populations and thus resulting in a larger number of homozygotes (n=51; 1.4%).

Association analysis of carriers in the ACDRP cohort revealed: i) no association to BMI, fasting

glucose or insulin, steatosis or alanine transaminase (ALT) (steatosis: β=0.03±0.09, p=0.68;

ALT: β=0.005±0.016, p=0.75), ii) lower alkaline phosphatase (ALKP) (β= 0.03±0.01, p=0.003),

10 Hepatology

iii) lower TG, total cholesterol and LDLC (p<7.1x106, p<3.6x106; and p<2.1x108, respectively), and iv) higher high density lipoprotein cholesterol (HDLC), (p<5.0x105) (Supplemental Tables 4

and 5).. Lipid subfraction data from a subset of participants (n=833) indicated that rs58542926

carriers had lower total nonHDLC (p=2.7x10−9), very low density lipoprotein (VLDLC;

−4 −4 −6 p=2.4x10 ), VLDL3C (p =2.2x10 ), intermediate density lipoprotein (IDLC; p=1.9x10 ), real

LDLC (p=3.5x10−8) and remnant lipoproteinC (p=9.8x10−6), but not HDL or HDLrelated sub fractions (p>0.09; Supplemental Table 6). Lipid profiles of 807 genotyped individuals from the

ACDRP cohort who underwent highfat challenge were available for analysis (20) (Figure 1). T allele carriers of rs58542926 exhibited lower postprandial triglyceridaemia (p<9.9x103) with lower total area under the curve (ß=9.5, p<0.005) and incremental area under the curve (ß=

4.25, p<0.02; Figure 1). Moreover, apoB48 levels measured at 2 hours postmeal were significantly lower in the rs58542926 T allele homozygotes (19.1±7.5 vs 13.1±6.0, values are ±

SEM, n=12, p<0.02 adjusted for age and sex) This suggests a substantial delay in appearance of ingested lipid into circulating TG and impaired lipid processing by the small intestine. These data therefore prompted us to pursue mechanistic studies in a zebrafish model of TM6SF2 loss offunction.

Modeling TM6SF2 loss of function in zebrafish

The rs58542926 variant is a lossoffunction allele (35), To model its function, we

carried out genetic knockdown in a model organism, the zebrafish, in which

hepatic and intestinal lipid homeostasis could be easily observed. We first

verified that tm6sf2 was expressed in larval zebrafish liver and gut by wholemount in situ

hybridization and qRTPCR (Supplemental Fig 1AB). Expression continued to be abundant in

liver and intestine at adult stages (Supplemental Fig 1C). We then targeted endogenous

expression of the zebrafish ortholog of TM6SF2, tm6sf2, by antisense morpholino

oligonucleotide (MO) injected into onecell embryos. Both spliceblocking MOs, targeting exon

11 Hepatology

3 (tm6sf2_e3 MO) or exon 4 (tm6sf2_e4 MO), suppressed tm6sf2 expression, with tm6sf2_e4

MO exhibiting more potency at earlier stages and lack of MOrelated offtarget effects as

detected by no change in p53∆113 gene expression, a proapoptotic gene isoform associated

with MO offtarget effects (Supplemental Figure 2AC; (23)). We therefore used the tm6sf2_e4

MO to model tm6sf2 loss of function in all subsequent experiments. Suppression of tm6sf2

through this approach increased the proportion of larvae exhibiting visible hepatic lipid (p<1x10

10; Figure 2B) versus control MOinjected larvae (Figure 2A,B), correlating with an increased

number of CLDs quantified within a 49 nm2 area of the liver. The average CLD number was

29.9±2.3 and 8.9±3.6 for exon 4 and control, respectively (values are ± SEM; n≥75; p<1x1020;

Figure 2B and Supplemental Figure 3). To further validate our model of hepatic steatosis

induced by tm6sf2 loss of function in the zebrafish model, we performed CRISPR/Cas9

mediated disruption of the zebrafish tm6sf2 locus by coinjection of larvae with Cas9 mRNA and

gRNA targeting either exon 3 or exon 4. An increase of hepatic steatosis was observed in both

CRISPR mutant lines (Supplemental Figure 4AC). The mismatch at the genomic locus was

confirmed by T7 endonuclease assays and sequencing (24) (Supplemental Figure 4).

We quantified LDLC in tm6sf2-depleted and control MO larvae after dissecting livers

and measuring LDLC in the remaining carcass (25). With a control diet, carcass LDLC

decreased by tm6sf2 disruption: 0.17±0.09 versus 1.90±0.19 for tm6sf2_e4 MO and control MO,

(p<0.003; Figure 2C). Taken together, these data validate the zebrafish model for study of

TM6SF2 function and suggest a tm6sf2 role in ApoBTRL secretion since ApoBTRL and LDL

are directly interrelated.

We next assessed the relevance of rs58542926 to systemic lipid homeostasis in

zebrafish by coinjection of tm6sf2_e4 MO with either wild type human TM6SF2 RNA or E167K

RNA. Addition of wild type TM6SF2 RNA expression but not E167K RNA completely rescued

steatosis (Figure 2A and B), supporting the lossoffunction nature of E167K (3, 5, 13). The

average number of liver CLDs was 31.3±3.6 and 32.9±3.2 with MO alone or E167K RNA co

12 Hepatology

injection, respectively, and 9.2±1.5 with coinjection of MO and wild type RNA (p<1x1024) which was not significantly different from control MO (values are ± SEM, n≥75, ns).

tm6sf2 disruption is associated with both ER stress and steatosis

ER stress contributes to hepatic steatosis and progression to NASH (16). We therefore performed zebrafish studies to evaluate whether loss of tm6sf2 also results in ER stress. TEM examination of liver ultrastructure in unfed 5 dpf larvae (Figure 3A) revealed an increase in

CLDs in hepatocytes with tm6sf2 knockdown versus control MO, consistent with hepatic CLD accumulation observed by whole mount oil red O staining (Figures 2A and B, 3A and B).

Average CLD area/cell was 1825±238 versus 2878±325 with tm6sf2_e4 MO versus control MO, respectively (p<0.005; values are ± SEM, n=12 larvae and 31 cells) (Figure 3B). Knockdown of

tm6sf2 but not control MO resulted in ER cisternae dilation (supplemental Figure 6A and B):

cisternae width was 63.7±7.3 nm versus 30.0±4.4 nm with tm6sf2_e4 MO versus control MO,

respectively (p≤3x1064; values are ± SEM, n>100 ER cisternae). Markers of ER stress, including chop, bip, edem1 and spliced xbp1, examined by qRTPCR at 5 dpf, were also significantly increased in tm6sf2 morphants versus controls (Figure 3C).

Similar results were obtained in human liver biopsies from rs58542926 carriers from the

bariatric surgery cohort from the Geisinger Health System (Danville, PA). We examined the

expression of ER stress markers including CHOP, HSPA5, ATF6, ATF4 and spliced XBP1 by

qRTPCR and found a significant increase in carriers of the variant compared to noncarriers

with comparable degree of steatosis (p<0.05, p<0.01, p<0.001, and p<0.05, respectively).

Increased ER was present irrespective of fibrosis (Figure 3D).

Together, these observations indicate that steatosis and ER stress are concurrently

associated with TM6SF2 loss of function in both zebrafish and human livers (Figure 3AD and

supplemental Figure 6) warranting further investigation,

13 Hepatology

tm6sf2 is sufficient to rescue foiegras mutant steatosis and ER stress

Given that over expression of TM6SF2 mRNA reduced the proportion of wild type larvae

exhibiting accumulation of hepatic lipid (Figure 2A and B), its over expression should also

ameliorate lipid accumulation in a zebrafish model of NAFLD caused by ER stress. To test this

possibility, we examined the foie gras (foigr) zebrafish mutant, which exhibits hepatic steatosis

as a result of ER stress (26, 27). We injected progeny of heterozygous foigr carriers with RNA

encoding wild type TM6SF2. Relative to untreated mutant larvae at 5 dpf, over expression of

TM6SF2 resulted in a significant decrease in the proportion of animals exhibiting oil red O

staining in the liver (16% in tm6sf2_e4 MO larvae vs 23% in control MO larvae; Figure 4A and

Supplemental Figure 4). Expression of the ER stress marker genes, chop and edem1, was also

significantly reduced in livers of wild type TM6SF2 RNAinjected larvae, relative to untreated

control mutants (Figure 4B). In contrast, over expression of E167K into foigr larvae did not

rescue hepatic steatosis as 85% of larvae displayed steatosis and produced an upregulation of

all assessed genes associated with induction of ER stress (Figures 4A and 4B, Supplemental

Figure 4).

tm6sf2 knockdown enhances tunicamycin-induced ER stress gene expression

To further explore the role of tm6sf2 knockdown in ER stress, we examined whether

pharmacological induction of ER stress phenocopies the tm6sf2 knockdown ER stress

phenotype in zebrafish liver. We used tunicamycin (Tm), an antibiotic that blocks protein N

linked glycosylation in ER by inhibiting the transfer of GlcNAc to dolichol phosphate and typically

causes full activation of most UPR target genes (26, 28, 29). To examine the possibility of an

epistatic relationship between tm6sf2 and ER stress, we used a combination of tm6sf2 MO and

tunicamycin treatment (30). Combining MO mediated knockdown of tm6sf2 and Tm treatment

enhanced ER stress gene expression (Supplemental Figure 7). However, TM6SF2 over

expression did not rescue Tm induced ER stress (Supplemental Figure 7), suggesting that the

14 Hepatology

mechanism by which tm6sf2 deficiency disrupts ER homeostasis likely differs from the one

underlying Tm induced ER stress.

Reduced TG synthesis ameliorates tm6sf2 depletion-induced phenotypes

To determine if tm6sf2induced ER stress could be alleviated by reduced production of

TG, the major component of hepatic CLD, we examined the consequences of reducing TG

production on the steatosis and ER stress phenotypes in zebrafish larvae lacking tm6sf2. We

reduced TG formation by inhibiting AcylCoA: diacylglycerol acyltransferase 2 (dgat2), a key

enzyme for the final step of TG biosynthesis that is highly expressed in liver (31). dgat-2 MO

treatment resulted in effective dgat-2 gene knockdown at day 5.(Supplemental Figure 2).

Knockdown of dgat2 reduced tm6sf2 knockdowninduced hepatic steatosis, as indicated by oil red O staining, to a level comparable to MO control treated larvae (Figure 5A). 89% of larvae injected with tm6sf2_e4 MO exhibited liver oil red O staining, a significant increase compared to control MOtreated larvae (25%) or larvae injected with both tm6sf2 and dgat2 (21%; Figure

5B). In addition, dgat2 inhibition ameliorated tm6sf2 depletioninduced ER stress. Elevated levels of expression for ER stress markers chop, bip and edem1 by tm6sf2 deletion were ameliorated by loss of dgat2 (Figure 5C). These findings suggest that inhibition of TG synthesis reduced both hepatic steatosis and ER stress resulting from loss of tm6sf2.

Disruption of tm6sf2 perturbs intestinal clearance of dietary lipid

TM6SF2 expression is highest in small intestine, in both humans and mice

(Supplemental Figure 5AB) and delayed postprandial TG excursion was observed in human

E167K carriers, suggesting TM6SF2 has a role in intestinal lipid homeostasis similar to that in

liver (Figure 1). To test this possibility, we assessed CLD content of enterocytes in zebrafish at

5 dpf, when larvae begin to rely on dietary lipid, by comparing histology and intestinal CLD in

ultrathin section TEM of larval anterior intestine in three fed states: 1) unfed, 2) immediately

15 Hepatology

after a high fat meal (3 hour HFD), 3) 18 hour after the 3 hour HFD, and 4) 24 hour after the 3

hour HFD. Unfed control larvae did not exhibit CLD accumulation in enterocytes (Figure 6A), but

the average CLD area per cell surface area was about 30 fold greater in unfed tm6sf2_e4 MO

larvae. ER cisternae dilation was also increased (Figure 6AD), consistent with a tm6sf2 role in

enterocyte ER and lipid homeostasis in the absence of dietary lipid similar to its role in liver

(Figure 3AC and Supplemental Figure 6A and B). After a 3 hour HFD, CLDs increased in

control larvae and then reduced following 18 hour clearance. The area of CLD relative to cell

surface area was 1059±261.21 nm2 after the 3 hour HFD and then reduced to 96.53±52.94 nm2

after 18 hour clearance (values are ± SEM, n=412 larvae, p<0.01; Figure 7C). With tm6sf2

depletion, an increase in CLD was also observed after 3 hour HFD (unfed 781.78±388.4 nm2 vs

3 hour HFD 3221.4±681.62 nm2, values are ± SEM, n=12 larvae, p<0.001 (Figures 6A and 7A

C). However, the CLD accumulation observed in tm6sf2 deficiency enterocytes after 3hr HFD

was significantly greater than the CLD accumulation in control MO enterocytes (Figure 7AD).

Importantly, by 18hr after a 3hr HFD, while 90% of the accumulated fat was cleared in

enterocytes from zebrafish treated with control MO, only 40% of accumulated fat was cleared in

enterocytes from zebrafish treated with tm6sf2 MO.. By 24hr, there was no longer a statistical

difference observed among the groups for total CLD area although tm6sf2-deficient enterocytes

still contained a greater number of CLDs compare to control enterocytes (control 4.2±0.95 vs

tm6sf2 MO 12.9±2.6 average CLD/cell; values are ± SEM, n=12 larvae, p<0.001) (Figure 7CD).

Together with our human postprandial lipid measurement, these data support a likely role of

TM6SF2 in small intestine lipid clearance.

TM6SF2 modulates lipid content in human Caco-2 enterocytes and is associated with ER stress

To assess whether our observations in zebrafish could be extrapolated to human cells,

we transduced human Caco2 enterocytes with lentiviral shRNA targeting TM6SF2

(shTM6SF2), resulting in 98% reduction of mRNA versus a scrambled shRNA control treatment

16 Hepatology

(Figure 8A). Visualization of intracellular lipid by BODIPY 493/503 staining (21) indicated that

TM6SF2 depletion increased intracellular lipid content by 3.7 fold relative to control shRNA treatment (p<0.01; 79 independent experiments; Figure 8AB) in Caco2 cells incubated with oleic acid (600M for 16 hours) but not without exogenous oleic acid (data not shown), suggesting that TM6SF2 depletion increases CLD in enterocytes in response to exogenous lipids. Further, examination of TG synthesis and secretion in enterocytes by pulsechase studies with [(3)H]oleate indicated that TM6SF2 depletion did not alter incorporation of exogenous oleate into intracellular TG, measured after 16 hours, but reduced incorporation into TG secreted into the media after a 1 hour “chase” (p<0.01; 34 independent experiments; Figure 8

CD). Finally, we confirmed the ER localization of TM6SF2 in Caco2 cells and upon treatment of TM6SF2 shRNAtreated cells with oleic acid (600M for 16 hours) we observed significantly increased expression of ER stress markers, including HSPA5, ATF6, ATF4, and CHOP compared to control (p<0.001, p<0.05, p<0.05, and p<0.01, respectively; Figure 8E and

Supplemental Figure 4D). These results corroborate our observations in zebrafish and strongly support a TM6SF2 role in processing of dietary lipids in human enterocytes.

Discussion:

Our findings indicate that TM6SF2 plays a role in both liver and enterocyte lipid homeostasis. Upon depletion of TM6SF2 gene expression in human Caco2 enterocytes, or its homolog in zebrafish, we observed significant lipid accumulation in response to exogenous lipid in either tissue. These observations are consistent with observations in human carriers of a

TM6SF2 lossoffunction variant, which exhibit hepatic steatosis and ER stress progressing to fibrosis and inflammation along with significant decreases in fasting circulating lipids but also reduced postprandial triglyceride excursion. Our findings also reveal that increased ER stress as a result of loss of TM6SF2 expression is concurrent with the observed increase accumulation

17 Hepatology

of CLDs in both human and zebrafish tissues. The rescue of both phenotypes from depletion of

TM6SF2 gene expression by simply blocking production of TG, and the observation of epistatic

interaction with Tm. provide strong evidence that the TM6SF2 gene function involves lipid

homeostasis, which has a downstream impact on ER stress. Taken together, these findings

suggest an important role for TM6SF2 in mediation of circulating TG through a previously

unidentified intestinal function that likely parallels its functions in liver. There is convincing

evidence that hypertriglyceridemic states are associated with accelerated atherosclerosis and

CVD (8, 32). Most TG is transported in the plasma by TRL that are either hepaticallyderived

VLDL or intestinallyderived chylomicrons, with considerable overlap of molecular players and

pathways involved in processing across the repertoire of TRLs. Our data provide strong support

for a role of TM6SF2 in TRL processing, in both liver and intestine tissues.

In fasting conditions, TG is predominately carried in VLDL particles, whereas in the

postprandial state TG is additionally carried in chylomicrons and their remnants (8). The role of

TM6SF2 in modulating VLDL secretion from liver is well documented and has been proposed as

the primary mechanism by which fasting lipoprotein profiles are improved in carriers of the loss

offunction variant (3, 5). Importantly relevant to its function, ApoBassociated lipoproteins are

primarily affected in carriers of the lossoffunction variant, as evidenced by the highly significant

effects on LDLC, VLDLC, IDLC, and TG. Our findings offer an additional mechanism by

which TM6SF2 depletion may contribute to improved CVD profile through its role in regulating

postprandial intestinederived TRLs. Our observation that TM6SF2 carriers of the Tallele have

lower postprandial triglyceridaemia in response to a high fat challenge is in agreement with a

recent published report (33). Although about 80% of the increase in circulating TG after a fat

load meal comes from apoB48containing lipoproteins (34), approximately 80% of the increase

in particle number is accounted for by VLDL particles (35, 36). ApoB48 and apoB100

containing particles are cleared from the circulation by common pathways and therefore

compete for clearance (37). Our data support the need to further detailed human kinetic studies

18 Hepatology

to understand the full impact of TM6SF2 in processing of apoBlipoproteins, and relative

contribution of liver and small intestine to the improved lipid profile.

The improved lipid profile in carriers of the E167K variant is in apparent conflict with the variant’s adverse effects on NAFLD and progression to NASH. This suggests that impaired TRL secretion may be directly relevant to retention of intracellular lipid in hepatocytes. Intriguingly, we observed improved lipid profiles in carriers of the variant from two cohorts with vastly different metabolic profiles. The Old Order Amish (OOA) is a relatively healthy population with less diabetes, more physical activity and less alcohol consumption than the general population.

The variant in TM6SF2 was strongly correlated with reduced circulating lipids in the OOA, even in the absence of NAFLD or liver dysfunction (42, 43). In contrast, the variant was associated with increased steatosis, fibrosis and inflammation in the metabolically stressed conditions found in extremely obese bariatric surgery patients. Similarly, increased association with NASH was reported in obese children carrying the rs58542926 variant (10). These observations suggest a role for TM6SF2 in both liver and intestinal response, but perhaps underscore the role of metabolic stress in unmasking its deleterious effects in liver. Importantly, these findings support the possibility of a common intracellular mechanism by which TM6SF2 functions across tissues.

This suggests that the strength of the association of TM6SF2 T allele with NAFLD and

NAFLD progression may depend on additional factors that compromise hepatic lipid homeostasis. Indeed, our observations of increased association with steatosis, fibrosis, and inflammation in bariatric surgery patients might indicate that extreme obesity is a metabolic stress that impacts the presence of NASH associated with rs58542926. Similarly increased association with NASH was reported in obese children (10). This may be clinically relevant as it may indicate that interventions aiming at either reducing hepatic lipid flux or increasing lipid utilization may help circumvent the deleterious health consequences of the TM6SF2 variant on liver lipid homeostasis. Overall, our observations in two cohorts with vastly different metabolic

19 Hepatology

profiles support a role for TM6SF2 in both liver and intestinal response to dietary lipid with

improved systemic lipid homeostasis. These data potentially support the possibility of a unifying

mechanism at the level of TM6SF2 intracellular function.

Dietary fat absorption is a multistep process that involves temporary storage of dietary

fat in CLDs of enterocytes and hepatocytes, and repackaging in lipoprotein particles for

transport (38). It is accepted that both CLDs and apoBrich lipoproteins both originate within the

ER bilayer from newly synthesized esterified lipids, but they mature on opposite sides of the

bilayer (38). The current model of chylomicrons and VLDL assembly proposes a twostep

process. In the first step newly synthesized apoB is lipidated during translocation across the ER

into the lumen, yielding a primordial particle. In the second step, bulk transfer of core lipids is

believed to take place postranslationally, involving lumenal LDs (39). Several lines of evidence

support a role of TM6SF2 in the packaging and secretion of esterified lipids. Previous reports

have indicated that TM6SF2 localizes to the ER in hepatocytes, consistent with our

observations in enterocytes (5). In addition, our zebrafish and cell culture studies indicate that

TM6SF2 deficiency reduced the flux of TG toward export in both tissues, and consequently CLD

number is concurrently increased.

Our data also suggest that it is unlikely that processes upstream of the ER are impaired.

It is not likely that fatty acid uptake is impaired given that the introduction of dietary lipid resulted

in a significant increase in enterocyte lipid content. The absence of an uptake defect suggests

that the localization of a defect is limited to an ERlocalized defect in lipid packaging and

synthesis or to postER trafficking and secretion.

The role of perturbations in the ER in NAFLD has become a subject of considerable

interest in recent years (16). Notably, depletion of tm6sf2 in zebrafish hepatocytes and

enterocytes led to both increased CLD accumulation and ER stress in our zebrafish and human

studies. Both of these phenotypes can be successfully rescued simply by dgat2 knockdown, an

indication that TG biogenesis and the tm6sf2 loss associated lipid homeostasis phenotype might

20 Hepatology

be causative of ER stress. Nonetheless, TM6SF2 rescued both steatosis and ER stress in a

zebrafish mutant, foigr, where steatosis is thought to be secondary to ER stress (26). foigr is

highly conserved; TRAPPC11 is the human ortholog, but its function remains elusive. Published

data suggest that TRAPPC11 functions in the anterograde secretory pathway at an early stage

in ERto Golgi transport (40). In addition to its trafficking function, recent published data

suggests that TRAPPC11 may also act upstream of the dolichylphosphate N

acetylglucosaminephosphotransferase (DPAGT1CDC), the target enzyme responsible for Tm

induced ER stress (41). However, results from our epistasis experiments suggest that TM6SF2

over expression does not rescue Tm induced ER protein homeostasis, an indication that Tm

and tm6sf2 depletion activate ER stress by different mechanisms. If TRAPPC11 deletion

causes a backup of VLDL in the smooth ER (26), our data suggest that TM6SF2 overexpression

can compensate for the interrupted trafficking. VLDL and chylomicron precursors are

assembled in the ER and their maturation occurs in the postER compartments (42). However,

how this maturation is achieved is only partially understood. Our data support the novel

hypothesis that TM6SF2 may be key to the process involved in ApoB lipoprotein maturation.

Further examination of the intracellular mechanisms underlying lipid accumulation will be necessary to fully clarify the function of TM6SF2 in hepatic and intestinal lipid accumulation. The identification of an important role of a single gene in both tissues which are central to regulation of systemic lipemia offers the exciting possibility of a single gene target to address elevated circulating TRLs from both sources. Further study will also potentially shed light onto other interactors that may collaborate with TM6SF2, revealing important insight into the regulation of

TRL regulation and secretion. The elucidation of an intestinal role for TM6SF2 offers a first step towards such understanding.

References

21 Hepatology

1. Choi SH, Ginsberg HN. Increased very low density lipoprotein (VLDL) secretion, hepatic steatosis, and insulin resistance. Trends Endocrinol Metab 2011;22:353363. 2. Tolman KG, Fonseca V, Dalpiaz A, Tan MH. Spectrum of liver disease in type 2 diabetes and management of patients with diabetes and liver disease. Diabetes Care 2007;30:734743. 3. Kozlitina J, Smagris E, Stender S, Nordestgaard BG, Zhou HH, TybjaergHansen A, Vogt TF, et al. Exomewide association study identifies a TM6SF2 variant that confers susceptibility to nonalcoholic fatty liver disease. Nat Genet 2014;46:352356. 4. Holmen OL, Zhang H, Fan Y, Hovelson DH, Schmidt EM, Zhou W, Guo Y, et al. Systematic evaluation of coding variation identifies a candidate causal variant in TM6SF2 influencing total cholesterol and myocardial infarction risk. Nat Genet 2014;46:345351. 5. Mahdessian H, Taxiarchis A, Popov S, Silveira A, FrancoCereceda A, Hamsten A, Eriksson P, et al. TM6SF2 is a regulator of liver fat metabolism influencing triglyceride secretion and hepatic lipid droplet content. Proc Natl Acad Sci U S A 2014;111:89138918. 6. Taskinen MR, Boren J. New insights into the pathophysiology of dyslipidemia in type 2 diabetes. Atherosclerosis 2015;239:483495. 7. Nordestgaard BG, Varbo A. Triglycerides and cardiovascular disease. Lancet 2014;384:626635. 8. Chapman MJ, Ginsberg HN, Amarenco P, Andreotti F, Boren J, Catapano AL, Descamps OS, et al. Triglyceriderich lipoproteins and highdensity lipoprotein cholesterol in patients at high risk of cardiovascular disease: evidence and guidance for management. Eur Heart J 2011;32:13451361. 9. Roh YS, Loomba R, Seki E. The TM6SF2 variants, novel genetic predictors for nonalcoholic steatohepatitis. Gastroenterology 2015;148:252254. 10. Grandone A, Cozzolino D, Marzuillo P, Cirillo G, Di Sessa A, Ruggiero L, Di Palma MR, et al. TM6SF2 Glu167Lys polymorphism is associated with low levels of LDLcholesterol and increased liver injury in obese children. Pediatr Obes 2015. 11. Dongiovanni P, Petta S, Maglio C, Fracanzani AL, Pipitone R, Mozzi E, Motta BM, et al. Transmembrane 6 superfamily member 2 gene variant disentangles nonalcoholic steatohepatitis from cardiovascular disease. Hepatology 2015;61:506514. 12. Daly AK, Day CP, Liu YL, Anstee QM. TM6SF2 as a genetic risk factor for fibrosis. Hepatology 2015;62:1321. 13. Liu YL, Reeves HL, Burt AD, Tiniakos D, McPherson S, Leathart JB, Allison ME, et al. TM6SF2 rs58542926 influences hepatic fibrosis progression in patients with non alcoholic fatty liver disease. Nat Commun 2014;5:4309. 14. Dongiovanni P, Romeo S, Valenti L. Genetic Factors in the Pathogenesis of Nonalcoholic Fatty Liver and Steatohepatitis. Biomed Res Int 2015;2015:460190. 15. Sookoian S, Castano GO, Scian R, Mallardi P, Fernandez Gianotti T, Burgueno AL, San Martino J, et al. Genetic variation in transmembrane 6 superfamily member 2 and the risk of nonalcoholic fatty liver disease and histological disease severity. Hepatology 2015;61:515525. 16. Zhang XQ, Xu CF, Yu CH, Chen WX, Li YM. Role of endoplasmic reticulum stress in the pathogenesis of nonalcoholic fatty liver disease. World J Gastroenterol 2014;20:17681776.

22 Hepatology

17. Schlegel A. Studying lipoprotein trafficking in zebrafish, the case of chylomicron retention disease. J Mol Med (Berl) 2015;93:115118. 18. Schlegel A, Gut P. Metabolic insights from zebrafish genetics, physiology, and chemical biology. Cell Mol Life Sci 2015;72:22492260. 19. Schlegel A. Studying nonalcoholic fatty liver disease with zebrafish: a confluence of optics, genetics, and physiology. Cell Mol Life Sci 2012;69:39533961. 20. Pollin TI, Damcott CM, Shen H, Ott SH, Shelton J, Horenstein RB, Post W, et al. A null mutation in human APOC3 confers a favorable plasma lipid profile and apparent cardioprotection. Science 2008;322:17021705. 21. Bell M, Wang H, Chen H, McLenithan JC, Gong DW, Yang RZ, Yu D, et al. Consequences of lipid droplet coat protein downregulation in liver cells: abnormal lipid droplet metabolism and induction of insulin resistance. Diabetes 2008;57:20372045. 22. Caviglia JM, Sparks JD, Toraskar N, Brinker AM, Yin TC, Dixon JL, Brasaemle DL. ABHD5/CGI58 facilitates the assembly and secretion of apolipoprotein B lipoproteins by McA RH7777 rat hepatoma cells. Biochim Biophys Acta 2009;1791:198 205. 23. Robu ME, Larson JD, Nasevicius A, Beiraghi S, Brenner C, Farber SA, Ekker SC. p53 activation by knockdown technologies. PLoS Genet 2007;3:e78. 24. Hwang WY, Fu Y, Reyon D, Maeder ML, Kaini P, Sander JD, Joung JK, et al. Heritable and precise zebrafish genome editing using a CRISPRCas system. PLoS One 2013;8:e68708. 25. O'Hare EA, Wang X, Montasser ME, Chang YP, Mitchell BD, Zaghloul NA. Disruption of ldlr causes increased LDLc and vascular lipid accumulation in a zebrafish model of hypercholesterolemia. J Lipid Res 2014;55:22422253. 26. Cinaroglu A, Gao C, Imrie D, Sadler KC. Activating transcription factor 6 plays protective and pathological roles in steatosis due to endoplasmic reticulum stress in zebrafish. Hepatology 2011;54:495508. 27. Sadler KC, Amsterdam A, Soroka C, Boyer J, Hopkins N. A genetic screen in zebrafish identifies the mutants vps18, nf2 and foie gras as models of liver disease. Development 2005;132:35613572. 28. Chen Z, Ballar P, Fu Y, Luo J, Du S, Fang S. The E3 ubiquitin ligase gp78 protects against ER stress in zebrafish liver. J Genet Genomics 2014;41:357368. 29. Lehle L, Tanner W. The specific site of tunicamycin inhibition in the formation of dolicholbound Nacetylglucosamine derivatives. FEBS Lett 1976;72:167170. 30. Levine AJ, Hu W, Feng Z, Gil G. Reconstructing signal transduction pathways: challenges and opportunities. Ann N Y Acad Sci 2007;1115:3250. 31. Yen CL, Stone SJ, Koliwad S, Harris C, Farese RV, Jr. Thematic review series: glycerolipids. DGAT enzymes and triacylglycerol biosynthesis. J Lipid Res 2008;49:22832301. 32. Miller M, Stone NJ, Ballantyne C, Bittner V, Criqui MH, Ginsberg HN, Goldberg AC, et al. Triglycerides and cardiovascular disease: a scientific statement from the American Heart Association. Circulation 2011;123:22922333. 33. Musso G, Cassader M, Paschetta E, Gambino R. TM6SF2 may divert postprandial cholesterol toxicity away from the vessel walls to the liver in NAFLD. J Hepatol 2015.

23 Hepatology

34. Cohn JS, Johnson EJ, Millar JS, Cohn SD, Milne RW, Marcel YL, Russell RM, et al. Contribution of apoB48 and apoB100 triglyceriderich lipoproteins (TRL) to postprandial increases in the plasma concentration of TRL triglycerides and retinyl esters. J Lipid Res 1993;34:20332040. 35. Karpe F, Bell M, Bjorkegren J, Hamsten A. Quantification of postprandial triglyceriderich lipoproteins in healthy men by retinyl ester labeling and simultaneous measurement of apolipoproteins B48 and B100. Arterioscler Thromb Vasc Biol 1995;15:199207. 36. Schneeman BO, Kotite L, Todd KM, Havel RJ. Relationships between the responses of triglyceriderich lipoproteins in blood plasma containing apolipoproteins B 48 and B100 to a fatcontaining meal in normolipidemic humans. Proc Natl Acad Sci U S A 1993;90:20692073. 37. Adiels M, Matikainen N, Westerbacka J, Soderlund S, Larsson T, Olofsson SO, Boren J, et al. Postprandial accumulation of chylomicrons and chylomicron remnants is determined by the clearance capacity. Atherosclerosis 2012;222:222228. 38. Sturley SL, Hussain MM. Lipid droplet formation on opposing sides of the endoplasmic reticulum. J Lipid Res 2012;53:18001810. 39. Lehner R, Lian J, Quiroga AD. Lumenal lipid metabolism: implications for lipoprotein assembly. Arterioscler Thromb Vasc Biol 2012;32:10871093. 40. Scrivens PJ, Noueihed B, Shahrzad N, Hul S, Brunet S, Sacher M. C4orf41 and TTC15 are mammalian TRAPP components with a role at an early stage in ERtoGolgi trafficking. Mol Biol Cell 2011;22:20832093. 41. DeRossi C, Vacaru A, Rafiq R, Cinaroglu A, Imrie D, Nayar S, Baryshnikova A, et al. trappc11 is required for protein glycosylation in zebrafish and humans. Mol Biol Cell 2016;27:12201234. 42. Sundaram M, Yao Z. Recent progress in understanding protein and lipid factors affecting hepatic VLDL assembly and secretion. Nutr Metab (Lond) 2010;7:35.

Figure legends

Figure 1. Tallele carriers (EK and KK) have significantly reduced TG levels compared to their EE counterparts in response to a liquid meal: Triglycerides levels before and during the high fat challenge by E167K genotype. Shown are covariateadjusted geometric means with 95% confidence intervals. Filled grey squares and hatched lines indicate individuals carrying the T allele and filled diamonds indicate noncarriers. Postprandial lipid responses were calculated as the AUC of the TG measurements at fasting, 1, 2, 3, 4 and 6 hours (p=0.05, p=0.03, p=0.008, p=0.04, p=0.05, p=0.05, respectively). The response relative to the baseline level was determined by calculating the incremental area under the curve (iAUC) of TG. The total area under the curve (AUC) of TG and postprandial TG levels and the incremental AUC (IAUC) were calculated and are shown in the insert with their respective p values.

Figure 2. Loss of tm6sf2 function in zebrafish results in hepatic lipid accumulation and reduced LDL-C. (A) Representative lateral view images of livers of Oil red O stained zebrafish larvae at 5 dpf. Shown from top left to bottom right: control MO, tm6sf2_e4 MO, coinjection of tm6sf2_e4 MO and wild type TM6SF2 RNA, coinjection of tm6sf2_e4 MO with TM6SF2 E167K mutant RNA, and injection of wild type TM6SF2 RNA only. SB, swim bladder; L, liver. (B) Quantification of lipid accumulation shown as average number of lipid droplets per 49 nm2

24 Hepatology

square area in livers of larvae treated as indicated (n=5). (C) LDLC levels were quantified after removal of livers from control MO or tm6sf2_e4 MO larvae given a control diet for 48 hr starting at 5 dpf. Data represent average of triplicated values from 2 pooled groups of 100 larvae each per experiment for two separate experiments. *** p≤1x1010.

Figure 3. Increased hepatic lipid accumulation in tm6sf2 morphants and ER stress. Similar increased expression of ER stress markers is observed in liver biopsies of TM6SF2 Glu167Lys (EK) carriers compared to TM6SF2 Glu167Glu (EE) counterparts. (A) Representative transmission electron microscopy (TEM) images of ultrathin 70 nm sections from livers of unfed control MO or tm6sf2_e4 MO larvae at 5 dpf assessed for hepatic lipid accumulation. CLD, cytosolic lipid droplet. (B) Quantification of hepatic lipid droplet accumulation shown as the average number of lipid droplets per nucleus. (C) Representative TEM images of ER cisternae (arrows) in livers of unfed control MO or tm6sf2_e4 MO treated 5 dpf larvae. (D). qRTPCR quantification of ER stress markers from 24 liver biopsies of TM6SF2 Glu167Glu (EE) and TM6SF2 Glu167Lys (EK) carriers. Six EE and EK carriers had been diagnosed with grade 0 or 1 steatosis but no fibrosis, 6 EE and EK with both steatosis (grade 1 and 2) and fibrosis. Patients were all female and morbidly obese. Results of triplicated assays, n = 6 each group, mean + SEM; *p<0.05, **p<0.01 for EE vs EE with fibrosis or EE vs EK or EE vs EK with fibrosis).

Figure 4. ER stress is ameliorated byTM6SF2 wild-type overexpression in foie gras mutants. (A) Representative lateral view images of livers of oil red O stained foigr mutants injected with control MO, TM6SF2 WT RNA, or TM6SF2 E167K mutant RNA. SB, swim bladder; L, liver. (B) Quantification of gene expression by qRTPCR for markers of ER stress in 5 dpf foigr larvae injected with either control MO, TM6SF2 WT RNA, TM6SF2 E167K mutant RNA or tm6sf2_e4 MO. Data represent average of triplicated values from 2 pooled groups of 40 larval livers each per experiment for two separate experiments. * p<0.05; **p<0.005

Figure 5. Inhibition of TG synthesis reduces hepatic steatosis and ER stress in tm6sf2 morphants. (A) Representative lateral view images of livers from oil red O stained control larvae, larvae injected tm6sf2_e4 MO alone, or coinjected with dgat2 MO. SB, swim bladder; L, liver. (B) Quantification of the proportion of larvae exhibiting steatotic (dark grey) or normal livers (light grey). Numbers of total larvae used across two experiments shown on bars. (C) Quantification of ER stress marker gene expression by qRTPCR from homogenized whole 5 dpf larvae injected control MO, tm6sf2_e4 MO alone, or tm6sf2_e4 MO and dgat2 MO together. Data represent average of triplicated values from 2 pooled groups of 20 larvae each per experiment for two separate experiments. ns: not significant; * p<0.01;**p<0.0005

25 Hepatology

droplet:cell size area as a proportion relative to control. n=12 larvae. (C) Representative TEM images of ER cisternae (arrows) from ultrathin 70 nm sections of enterocytes in anterior intestine. CLD, cytosolic lipid droplet. (D) Quantification of average ER lumen width measured from TEM images of larval intestine from unfed control MO or tm6sf2_e4 MO larvae. Data represent average width of 10 randomly selected smooth ER lumens per cell, three measurements per lumen. n=15 cells per treatment. ** p<1x1010; ***p<1x1010

Figure 7. Loss of tm6sf2 perturbs intestinal clearance of dietary lipid. Representative TEM images of ultrathin 70 nm sections from enterocytes in anterior intestine of larvae injected with control or tm6sf2_e4 MO treated with (A) high fat meal (HFD) for 3 hours or (B) a 3 hour HFD followed by 18 hour clearance. CLD, cytosolic lipid droplet. BB, brush border (C) Quantification of enterocyte CLD accumulation in control MO or tm6sf2_e4 MO larvae treated with a 3 hour HFD followed by 18 or 24hour clearance. Data calculated average CLD area relative to total cell area measured in individual cells on images in which at least three contiguous cells with visible nuclei could be visualized, n=412 larvae. ** p<0.01 relative to control MO at same condition. # p<0.01 relative to control MO 3hr HFD; & p<0.05 relative to tm6sf2 MO 3hr HFD.. (D) Quantification of the number of enterocyte cytosolic lipid droplet per cell in control MO or tm6sf2_e4 MO larvae treated a 3 hour HFD followed by 18 or 24hour clearance. n=412 larvae. *** p<0.001 relative to control MO at same condition. & p<0.05 relative to tm6sf2 MO 3hr HFD.

Figure 8. Effect of silencing TM6SF2 on cytosolic lipid droplet content of human Caco-2 cells, on TG synthesis and secretion and on ER stress gene expression. (A, B)Human Caco2 cells were analyzed 48 hours after shRNA inhibition with TM6SF2specific or control shRNA lentiviral vectors. Cells were incubated overnight with 400 m oleic acid. The following day, cells were stained with DAPI and BODIPY493/503, fixed and analyzed by confocal microscopy. (A) Representative image (Scale bar, 10 m). (B) Imaging analysis. Ratio of Bodipy fluorescence intensity to nuclei number normalized per area. Data are means±SE. from 7 9 independent experiments. **p<0.001. (C, D) During the pulse period, cells were incubated for 16 hours in culture medium containing 0.6 mM [13H]oleic acid complexed to fatty acidfree bovine serum albumin (BSA). Subsequently, during the washout period, the cells were incubated in culture medium containing 5 g fatty acidfree BSA per liter of medium for 1 hour. (C) Incorporated radioactivity in TG fraction of cells at the end of the pulse period. Data are mean±SE (n=3 independant experiments) (D) Incorporated radioactivity in TG fraction of media at the end of the 1hr washout, *p<0.05. Data are mean±SE(n=3 independent experiments) (E) Incubation of Caco2 cells with OA (0.65 mM) for 16 h caused increased ER stress as evaluated by GRP78, ATF6, ATF4, CHOP and spliced XBP1 (SXBP1). *p<0.05, **p<0.01, ***p<0.01. Data are mean±SE (n=3 independent experiments).

Acknowledgment We thank Dr. Simeon Taylor for helpful discussion and guidance. We also thank our Amish liaisons and Amish Research Clinic staff, as well as the Amish community for cooperation and support and the patients and staff of the Geisinger Obesity Institute

26 Hepatology

Figure 1. T-allele carriers (EK and KK) have significantly reduced TG levels compared to their EE counterparts in response to a liquid meal: Triglycerides levels before and during the high fat challenge by E167K genotype. Shown are covariate-adjusted geometric means with 95% confidence intervals. Filled grey squares and hatched lines indicate individuals carrying the T allele and filled diamonds indicate noncarriers. Postprandial lipid responses were calculated as the AUC of the TG measurements at fasting, 1, 2, 3, 4 and 6 hours (p=0.05, p=0.03, p=0.008, p=0.04, p=0.05, p=0.05, respectively). The response relative to the baseline level was determined by calculating the incremental area under the curve (iAUC) of TG. The total area under the curve (AUC) of TG and postprandial TG levels and the incremental AUC (IAUC) were calculated and are shown in the insert with their respective p values.

Figure 1 254x338mm (300 x 300 DPI)

Hepatology

Figure 2. Loss of tm6sf2 function in zebrafish results in hepatic lipid accumulation and reduced LDLC. (A) Representative lateral view images of livers of Oil red O stained zebrafish larvae at 5 dpf. Shown from top left to bottom right: control MO, tm6sf2_e4 MO, coinjection of tm6sf2_e4 MO and wild type TM6SF2 RNA, coinjection of tm6sf2_e4 MO with TM6SF2 E167K mutant RNA, and injection of wild type TM6SF2 RNA only. SB, swim bladder; L, liver. (B) Quantification of lipid accumulation shown as average number of lipid droplets per 49 nm2 square area in livers of larvae treated as indicated (n=5). (C) LDLC levels were quantified after removal of livers from control MO or tm6sf2_e4 MO larvae given a control diet for 48 hr starting at 5 dpf. Data represent average of triplicated values from 2 pooled groups of 100 larvae each per experiment for two separate experiments. *** p≤1x1010. Figure 2 254x338mm (300 x 300 DPI)

Hepatology

Figure 3. Increased hepatic lipid accumulation in tm6sf2 morphants and ER stress. Similar increased expression of ER stress markers is observed in liver biopsies of TM6SF2 Glu167Lys (EK) carriers compared to TM6SF2 Glu167Glu (EE) counterparts. (A) Representative transmission electron microscopy (TEM) images of ultrathin 70 nm sections from livers of unfed control MO or tm6sf2_e4 MO larvae at 5 dpf assessed for hepatic lipid accumulation. CLD, cytosolic lipid droplet. (B) Quantification of hepatic lipid droplet accumulation shown as the average number of lipid droplets per nucleus. (C) Representative TEM images of ER cisternae (arrows) in livers of unfed control MO or tm6sf2_e4 MO treated 5 dpf larvae. (D). qRT-PCR quantification of ER stress markers from 24 liver biopsies of TM6SF2 Glu167Glu (EE) and TM6SF2 Glu167Lys (EK) carriers. Six EE and EK carriers had been diagnosed with grade 0 or 1 steatosis but no fibrosis, 6 EE and EK with both steatosis (grade 1 and 2) and fibrosis. Patients were all female and morbidly obese. Results of triplicated assays, n = 6 each group, mean + SEM; *p<0.05, **p<0.01 for EE vs EE with fibrosis or EE vs EK or EE vs EK with fibrosis).

Hepatology

Figure 3 254x338mm (300 x 300 DPI)

Hepatology

Figure 4. ER stress is ameliorated byTM6SF2 wild-type overexpression in foie gras mutants. (A) Representative lateral view images of livers of oil red O stained foigr mutants injected with control MO, TM6SF2 WT RNA, or TM6SF2 E167K mutant RNA. SB, swim bladder; L, liver. (B) Quantification of gene expression by qRT-PCR for markers of ER stress in 5 dpf foigr larvae injected with either control MO, TM6SF2 WT RNA, TM6SF2 E167K mutant RNA or tm6sf2_e4 MO. Data represent average of triplicated values from 2 pooled groups of 40 larval livers each per experiment for two separate experiments. * p<0.05; **p<0.005

Figure 4 254x338mm (300 x 300 DPI)

Hepatology

Figure 5. Inhibition of TG synthesis reduces hepatic steatosis and ER stress in tm6sf2 morphants. (A) Representative lateral view images of livers from oil red O stained control larvae, larvae injected tm6sf2_e4 MO alone, or co-injected with dgat2 MO. SB, swim bladder; L, liver. (B) Quantification of the proportion of larvae exhibiting steatotic (dark grey) or normal livers (light grey). Numbers of total larvae used across two experiments shown on bars. (C) Quantification of ER stress marker gene expression by qRT-PCR from homogenized whole 5 dpf larvae injected control MO, tm6sf2_e4 MO alone, or tm6sf2_e4 MO and dgat2 MO together. Data represent average of triplicated values from 2 pooled groups of 20 larvae each per experiment for two separate experiments. ns: not significant; * p<0.01;**p<0.0005

Figure 5 254x338mm (300 x 300 DPI)

Hepatology

Figure 6. Depletion of tm6sf2 results in lipid accumulation and ER stress in small intestine. (A) Representative TEM images of ultrathin 70 nm sections of enterocytes in anterior intestine from 5 dpf unfed control MO and tm6sf2_e4 MO larvae assessed for lipid accumulation. N, nucleus; BB, brush border; LD, lipid droplet. (B) Quantification of enterocyte lipid droplet accumulation in unfed control MO or tm6sf2_e4 MO larvae. Data calculated average lipid droplet area relative to total cell area measured in individual cells on images in which at least three contiguous cells with visible nuclei could be visualized. Data shown represent lipid droplet:cell size area as a proportion relative to control. n=12 larvae. (C) Representative TEM images of ER cisternae (arrows) from ultrathin 70 nm sections of enterocytes in anterior intestine. CLD, cytosolic lipid droplet. (D) Quantification of average ER lumen width measured from TEM images of larval intestine from unfed control MO or tm6sf2_e4 MO larvae. Data represent average width of 10 randomly selected smooth ER lumens per cell, three measurements per lumen. n=15 cells per treatment. ** p<1x10-10; ***p<1x10-10

Hepatology

Figure 6 254x338mm (300 x 300 DPI)

Hepatology

Figure 7. Loss of tm6sf2 perturbs intestinal clearance of dietary lipid. Representative TEM images of ultrathin 70 nm sections from enterocytes in anterior intestine of larvae injected with control or tm6sf2_e4 MO treated with (A) high fat meal (HFD) for 3 hours or (B) a 3 hour HFD followed by 18 hour clearance. CLD, cytosolic lipid droplet. BB, brush border (C) Quantification of enterocyte CLD accumulation in control MO or tm6sf2_e4 MO larvae treated with a 3 hour HFD followed by 18- or 24-hour clearance. Data calculated average CLD area relative to total cell area measured in individual cells on images in which at least three contiguous cells with visible nuclei could be visualized, n=4-12 larvae. ** p<0.01 relative to control MO at same condition. # p<0.01 relative to control MO 3hr HFD; & p<0.05 relative to tm6sf2 MO 3-hr HFD.. (D) Quantification of the number of enterocyte cytosolic lipid droplet per cell in control MO or tm6sf2_e4 MO larvae treated a 3 hour HFD followed by 18- or 24-hour clearance. n=4-12 larvae. *** p<0.001 relative to control MO at same condition. & p<0.05 relative to tm6sf2 MO 3hr HFD. Figure 7 254x338mm (300 x 300 DPI)

Hepatology

Figure 8. Effect of silencing TM6SF2 on cytosolic lipid droplet content of human Caco-2 cells, on TG synthesis and secretion and on ER stress gene expression. (A, B)Human Caco-2 cells were analyzed 48 hours after shRNA inhibition with TM6SF2-specific or control shRNA lentiviral vectors. Cells were incubated overnight with 400 µm oleic acid. The following day, cells were stained with DAPI and BODIPY493/503, fixed and analyzed by confocal microscopy. (A) Representative image (Scale bar, 10 µm). (B) Imaging analysis. Ratio of Bodipy fluorescence intensity to nuclei number normalized per area. Data are means±SE. from 7 -9 independent experiments. **p<0.001. (C, D) During the pulse period, cells were incubated for 16 hours in culture medium containing 0.6 mM [13H]oleic acid complexed to fatty acid-free bovine serum albumin (BSA). Subsequently, during the washout period, the cells were incubated in culture medium containing 5 g fatty acid-free BSA per liter of medium for 1 hour. (C) Incorporated radioactivity in TG fraction of cells at the end of the pulse period. Data are mean±SE (n=3 independant experiments) (D) Incorporated radioactivity in TG fraction of media at the end of the 1hr washout, *p<0.05. Data are mean±SE(n=3 independent experiments) (E) Incubation of Caco-2 cells with OA (0.65 mM) for 16 h caused increased ER stress as

Hepatology

evaluated by GRP78, ATF6, ATF4, CHOP and spliced XBP1 (SXBP1). *p<0.05, **p<0.01, ***p<0.01. Data are mean±SE (n=3 independent experiments).

254x338mm (300 x 300 DPI)

Hepatology

Supplementary Appendix

Table of Contents

Detailed Material and Methods 2-10 1. Human Studies 2 2. Zebrafish Studies 4 3. Cell culture Studies 8 4. Material and Methods References 11 Tables 13-19 Table 1 13 Table 2 14 Table 3 15 Table 4 16 Table 5 17 Table 6 18 Table 7 19 Figures legend 20-21 Figures Figure 1 22 Figure 2 23 Figure 3 24 Figure 4 25 Figure 5 26 Figure 6 27 Figure 7 28 Figure 8 29 Figure 9 30

Hepatology

Detailed Material and Methods

1. Human studies

i) Study populations and phenotypic measurements:

Bariatric Cohort:

The TM6SF2 rs58542926 variant was genotyped in 983 bariatric surgery patients at the Geisinger Health System in Danville, PA. The protocol was approved by the Geisinger Clinical Institutional Review Board, and all subjects provided written informed consent. Prior to surgery, patients were extensively phenotyped, which included a comprehensive medical history and physical examination providing anthropometry, fasting serum glucose and lipids, liver and kidney function and medication usage. Intra-operative liver biopsy specimens were processed for histocytochemistry and scored for steatosis grade (0=<5%, 1=5%-33%, 2=33%-66%, 3=>66%), lobular inflammation (0=no foci, 1=<2 foci per 200X field, 2=2-4 foci per 200X field, 3=>4 foci per 200X field) hepatocyte ballooning and perivenular fibrosis, as previously described (1, 2).

Old Order Amish of Lancaster County, PA cohort:

The Old Order Amish (OOA) of Lancaster County, PA is a founder population that our group at the University of Maryland School of Medicine has been studying since 1993. The current Lancaster OOA population numbers ~34,000 individuals, including ~12,000 adults. We estimate that these individuals are descendants of 554 founders, with 128 founders contributing to 95% of the present day gene pool (3). Through leadership of coauthor Dr. Alan Shuldiner, as of 1/20/2016 we had 7,236 Amish adults with DNA samples enrolled in one or more of our studies, where ~6,000 have extensive phenotype information in one or more of the following areas: glucose homeostasis/diabetes, cardiovascular health, bone health, longevity, and platelet function and response to anti-platelet agents. This rich collection of well-phenotyped subjects makes up the Amish Complex Disease Research Program (ACDRP). For our study, we selected 3,556 of these subjects for whom we had both TM6SF2 genotypic and relevant phenotypic information. Specifically, the main data set included only subjects with non-missing values for body mass index (BMI), and serum LDL-cholesterol, HDL-cholesterol and triglycerides (TG) (3-5).

Standardized protocols and procedures were used for phenotyping across all studies. Height and weight were measured using a stadiometer and calibrated scale, and BMI (kg/m2) calculated. Fasting serum TG, total cholesterol and HDL-cholesterol (HDL-C) were measured by Quest Diagnostics (Horsham, PA). LDL-cholesterol was calculated by the Friedewald equation where TG<400 mg/dl. Glucose concentrations were assayed with a glucose analyzer (Beckman Coulter, Brea, CA or YSI, Yellow Springs, OH). Insulin levels were determined by radioimmunoassay (Linco Research Inc., St. Charles, MO). HOMA-IR was calculated using the following formula: HOMA-IR = [Fasting glucose (in mg/dl)*Fasting insulin (in μU/ml)]/405.

Lipid sub-fraction data and triglyceride intervention data were measured in the Heredity and Phenotype Intervention (HAPI) study (5). HAPI is an ACDRP intervention study that evaluated short-term cardiovascular interventions. The current analysis includes baseline data and data after a high-fat challenge. The high-fat challenge consisted of a whipping cream milk shake

Hepatology

standardized to be 782 calories/m2 body surface area and with the following composition: 77.6% fat, 19.2% carbohydrate, 3.1% protein. Blood was drawn before intervention (fasting) and at 1, 2, 3, 4 and 6 hours after intervention. Lipoprotein sub-fractions were measured by Vertical Auto Profile (VAP) technology (Athrotech, Birmingham, AL) at baseline (6). Plasma apoB-48 was assessed with an enzyme-linked immunosorbent assay (ELISA) (Biovendor R&D, Ashville NC). Hepatic fat content (determined via spleen:liver density ratio, where a higher ratio indicates more fat) was measured in a subset of subjects by thoracic electron-bean computerized tomography (EBCT) scans on an Imatron C-150 EBCT scanner. Two 1.0 cm2 regions of interest (ROI) were measured for the liver and one was measured from the spleen in a manner that minimized measurements of vessels, focal lesions, artifacts or the edge of the organ. AccuView (Accuimage Diagnostics Corp., San Francisco, CA) software was used to calculate the attenuation coefficient in Hounsfield Units for each ROI (7). The average liver attenuation ratio was divided by the spleen measure for standardization (liver:spleen ratio). The liver to spleen ratio was inverse normally transformed for analysis. All scans were performed and analyzed blinded to genotype. Due to strict cultural and religious principles, alcohol consumption in the Amish population is frowned upon and very rare and thus we believe unlikely to confound our measurements of liver fat. In order to maintain cultural appropriateness, we do not ask participants about alcohol consumption.

ii) rs58542926 genotyping:

The bariatric surgery patients were recruited through the Geisinger Medical Center for Nutrition and Weight Management and were prospectively enrolled into a clinical research program in obesity. The research was approved by the Institutional Review Board of the Geisinger Clinic and all patients provided informed consent. Genomic DNA was isolated from patient whole- blood samples (8) and was then genotyped on the Illumina HumanOmniExpress-12 v1.0 DNA analysis bead chip. Only SNPs passing quality control were used for haplotype phasing and imputation, as detailed in the eMERGE platform manuscript (9). Accuracy of imputed results including rs58542926 was determined based on high concordance rates (>99.8%) in the masked analysis (9).

In the Amish cohort, Taqman (Applied Biosystems, Foster City, CA) genotyping was conducted using manufacturer recommended procedures. The variant conformed to the expectations of Hardy-Weinberg equilibrium (p=0.76), had a high call rate (99%) and 99.3% concordance rate was observed among duplicate samples. iii) Analyses of quantitative traits:

In the Geisinger sample continuous variables (e.g., LDL-C/HDL-C) were modeled using linear regression and comparing carriers of the T allele to the CC genotype while simultaneously adjusting for age, sex, and current use of lipid medication. Liver histology measures were dichotomized and analyzed using logistic regression adjusting for age, sex and current use of lipid medication. Analyses were conducted using SAS (v9.3, Carey, NC).

Association analyses of quantitative traits in the OOA was performed using the measured genotype approach that models variation in the trait of interest as a function of measured environmental covariates, measured genotype and a polygenic component to account for phenotypic correlation due to relatedness. Sex and age were included as covariates and SNPs were coded using an additive model. The polygenic component was modeled using the relationship matrix derived from the complete 14-generation pedigree structure to properly 3

Hepatology

control for the relatedness of all subjects in the study. These analyses were carried out using the mixed model software MMAP developed in our group at the University of Maryland. The 25 traits examined include serum lipids (i.e. total cholesterol, HDL-cholesterol, LDL-cholesterol, TG and subfraction data), insulin and glucose after fasting, triglyceride excursion after high-fat feeding and hepatic fat content. Fasting insulin, glucose measures and HOMA-IR (skewed traits) were log transformed before analysis and untransformed means were reported for ease of interpretation. These association studies were not impacted by exclusion of carriers of ApoB, ApoC3, PNPLA3 and HSL variants (data not shown).

2. Zebrafish studies:

i) Zebrafish husbandry, embryo culture, and adult gene expression analysis:

Adult wild-type Tubingen or foigr (foie gras) (10) zebrafish embryos were collected from natural matings and maintained at 28.5°C. Embryos were cultured in embryo medium (11) at 28.5°C until harvesting at time points between 1 and 7 days post-fertilization (dpf). Larval stages were verified by measurement of larval length (Supplementary Figure 8). Adult animals were isolated at 6 months of age and euthanized prior to dissection. Liver and intestine was separated from whole gastrointestinal tracts dissected out from adults (n=5). Liver tissue was verified by qRT- PCR using fabp10a as a marker, which was expressed at very high levels in liver samples and very low levels in intestine (data not shown).

ii) Quantitative RT-PCR:

cDNA was generated by isolation of whole embryo or extracted liver RNA and subsequent reverse transcription (RevertAid First Strand cDNA Synthesis Kit, Thermo Scientific, Waltham, MA). Samples and controls/standards were run in triplicates on a Roche LightCycler 480 instrument (Roche, Basel, Switzerland) and mRNA expression levels were quantified relative to β-actin. Experiments were repeated at least three times. Primer sequences available upon request.

iii) Morpholino injection and validation:

Antisense oligonucleotide morpholinos (Gene Tools, LLC, Philomath, OR), designed to individually target exon 3 or 4 of tm6sf2, were injected into wild-type embryos between the 1 and 2-cell stage (n>200). Each morpholino (MO) was injected at three different concentrations (n>200 per concentration) to determine dose-response as well as the optimal morpholino concentration for future studies. A non-specific control MO was utilized in all experiments. Morpholino sequences are as follows: tm6sf2_e3: 5'-ACCACTGGCCTGAAATACAAAACA-3'; tm6sf2_e4: 5'-CCTCATGTCATTACGCTAACCGTTT-3'; control MO: 5'- CCTCTTACCTCAGTTACAATTTATA-3'; dgat2_e2: 5'-CCGAGCCTACACAGGATTTAAAGGA- 3'. Embryos injected with each MO were cultured at 28.5°C. To validate disruption of splicing and removal of a targeted exon, cDNA was generated from control, tm6sf2_e3, tm6sf2_e4, and dgat2_e2 morphants, as outlined above, at each day of development starting at 24 hours post fertilization (hpf) through 5 dpf. A total of 20 embryos were pooled for RNA extraction at each

Hepatology

time point. Experiments were repeated three times. Primer sequences are available upon request.

iv) CRISPR design and validation:

To confirm the results obtained by MO-injection, CRISPR-Cas9 mutants were generated for both tm6sf2 exons (3 and 4). Briefly, guide RNAs composed of a 22 bp target sequence flanked by a 5' T7 promoter sequence (5'-TAATACGACTCACTATA-3') and a 3' overlap sequence (5'- GTTTTAGAGCTAGAAATAG-3') were annealed to a generic oligo (5'- AAAAGCACCGACTCGGTGCCACTTTTTCAAGTTGATAACGGACTAGCCTTATTTTAACTTGC TATTTCTAGCTCTAAAAC-3'). The assembled oligos were transcribed in vitro using the MaxiScript Kit (Ambion, AM1314M). Cas9 transcript was generated using the mMESSAGE mMACHINE T3 kit (Life Technologies, AM1348). Approximately 300 single-celled wild-type (Tubingen) embryos were co-injected with 25 pg guide RNA and 300 pg Cas9 RNA. For each tm6sf2 gRNA construct (exon 3 and exon 4), 50 embryos were fixed at 5 dpf, stained with oil Red O (see methods below), and imaged for lipid accumulation. Genomic DNA was extracted for T7 endonuclease-based confirmation of mismatch (12) following PCR amplification of the target region and sequencing. The remaining embryos were used to propagate tm6sf2 exon3 and exon 4 mutant genetic lines.

v) Molecular cloning and transcript expression:

The coding sequence of human TM6FS2 cDNA was amplified by PCR using Phusion DNA polymerase using forward primer 5’-AGATCTCATatggacatcccgccgctgg and reverse primer 5’- GTCGACtcaatgctgcttcttgtggag-3’ from human liver cDNA. The PCR products were purified and subcloned into pCR2.1-TOPO vector (ThermoFisher, Waltham, MA). To generate the human TM6SF2 E167K mutant construct, a DNA fragment was amplified from the wild-type human TM6SF2 plasmid using a mutant forward primer containing a XmcI restriction enzyme sequence (5’-tcgccatgagcatcctggtgttccttacaggaaacattcttggcaaatacagctccaa-3’) and reverse primer (5’- GTCGACtcaatgctgcttcttgtggag-3)’. The 3’ end of the wild-type cDNA fragment from the XmcI cut site to the stop codon of the wild-type vector was replaced by mutant PCR fragment by T4 DNA ligase after digestion with XmcI and SalI. Each ORF was then subcloned into the pCS2+ vector (Clontech, MountainView, CA). Capped RNA was generated from each TM6SF2-pCS2+ construct using the mMESSAGE mMACHINE T7 Transcription Kit (Ambion, ThermoFisher Scientific, Waltham, MA) after linearization with NotI or SacII. TM6SF2 RNA or mutated (E167K) TM6SF2 was injected into embryos either alone or along with 6ng tm6sf2_e4 MO or 2ng tm6sf2_e3 MO.

All constructs were verified by restriction enzyme digestion and by DNA sequencing analysis.

vi) Oil red O (ORO) Staining:

5 dpf control and morphant larvae were fixed in 4% paraformaldehyde in 1x phosphate buffered saline and incubated overnight at 4°C. The following day, embryos were subjected to oil red O staining and imaged as previously described (14).

Hepatology

vii) Lipid Droplet Assessment and Quantification:

ORO stained embryos (wild-type, foigr, and morphant) were visualized under Brightfield and imaged on a glass slide within 3% methyl cellulose. Embryos were measured for length and size of swim bladder in order to adjust and identify any aging discrepancies; outliers were removed. Those embryos with appropriate 5 dpf sizing were then assessed for number of cytosolic lipid droplets in the liver (49 nm2 square region). Specifically, cytosolic lipid droplets were quantified via manual counting within 5 separate 49 nm2 liver regions while lipid droplet size (in the same five 49 nm2 regions) was quantified using ImageJ software. Moreover, the number of morphant larvae affected versus not affected was also recorded to determine morpholino efficacy. A minimum of 50 embryos were subjected to quantification per injection group.

viii) Diets and feeding:

Starting at 5 dpf, larvae were fed a control diet (CD) [Zeigler AP100 (Aquatic Habitats, Inc., Cary, NC) which was replaced twice daily until 7 dpf (48 hours of feeding) (15). At 7 dpf, larvae were fixed in 4% paraformaldehyde in 1x phosphate buffered saline and incubated overnight at 4°C in preparation for ORO staining. For transmission electron microscopy (TEM), embryos were fed a high-fat diet (HFD) for 3 hours starting at 5 dpf and then processed for TEM (for TEM fixation and sectioning methods see “Transmission electron microscopy” below). Preparation of the 10% HFD is as follows: fresh organic egg yolk isolated, forced through a 28G1/2 needle, and then added to embryo medium to obtain a 10% egg yolk-embryo medium solution. The solution was then vortexed and forced pipetted for ~10 minutes until lipid micelles formed (sized at ~1 mm diameter via stage micrometer).

ix) LDL-cholesterol quantification:

Control diet-fed control MO and tm6sf2_e4 MO embryo homogenates, comprised of 100 embryos per group, were utilized in the LDL/VLDL cholesterol assay kit (Cell Biolabs, Inc., San Diego, CA) following the protocol previously outlined (17). Exceptions to the protocol include utilization of de-livered embryos rather than whole embryos. Experiments were repeated four times. Values obtained by fluorimetric analysis were calculated relative to total protein concentration.

x) Transmission electron microscopy:

Three separate experiments were performed. TEM processing and imaging was performed at two locations, the University of Maryland and the Carnegie Institution for Sciences.

At 5 dpf, unfed, 3 hour HFD-fed, and 3 hour HFD/18- or 24-hour clearance control MO and tm6sf2_e4 MO embryos were immediately placed in PIPES and incubated at 4°C overnight. The following morning, the head (anterior of the heart) and tail (posterior of the urogenital pore) portions were removed and the midsection was mounted and transversely sectioned to obtain both liver and anterior intestine. Silver colored ultrathin sections of 70 nm were mounted onto 6

Hepatology

grids and imaged using a FEI Tecnai T12 high-resolution transmission electron microscope (UMB Core Facility, Baltimore MD). Images were acquired with AMT-XR611 digital camera using AMTV600 software (Advanced Microscopy Techniques). Lipid (size, number, and morphology), ER dilation, and location (liver, intestine) were measured in at least 2 larvae per condition.

Addtionally, At 5 dpf, unfed, 3 hour HFD-fed, and 3 hour HFD/18- or 24-hour clearance control MO and tm6sf2_e4 MO embryos were screened for “full” intestines (except for unfed embryos) and immediately fixed in a 3% glutaraldyhyde, 1% formaldehyde, 0.1 M cacadylate solution (pH 7.4) and incubated at 4°C overnight. Postfixation was done in 1% osmium tetroxide and En Bloc-stained with uranyl acetate in maleate (a detailed protocol is available upon request). Zebrafish were orientated in resin (Epon+Quetol (2:1), Spurr (3:1), 2%BDMA) blocks. Transverse sections were made to obtain liver and anterior intestinal sections using an ultramicrotome (Porter-Blum MT-2; Sorvall Instruments, Newton, CT), mounted on Formvar- coated grids, and stained with lead citrate. Images were captured with a Phillips Tecnai 12 microscope (Carnegie Institution for Science, Baltimore, MD) and recorded with a Gatan multiscan CCD camera using Digital Micrograph software. Lipid (size, number, and morphology), ER dilation, and location (liver, intestine) were measured in 4-8 larvae per condition.

Data from both instruments (UMB Core Facility and Carnegie Institution for Science) were combined for at least n=12 larvae per condition except for the 18 hr time point (n=4).

xi) Cytosolic lipid droplet and ER lumen quantification:

High-quality, high-magnification images were obtained using a FEI Tecnai T12 high-resolution transmission electron microscope or the Phillips Tecnai 12 microscope. Lipid droplet counts (size and number) were analyzed using ImageJ software (image magnification: 2-10 μm). Eight embryos were observed under TEM per condition (control MO vs. tm6sf2_e4 MO; unfed, 3 hour HFD, 3 hour HFD/24 hour clearance). For ER lumen quantification, a total of 10 random smooth ER lumen widths were quantified per cell with three measurements taken per ER lumen. At least 20 cells were assessed per condition. ImageJ software was utilized (image magnification: 500 nm).

xii) Whole-mount in situ hybridization:

In situ hybridization was carried out as per standard protocols (Thisse & Thisse 2008 PMID:18193022). Riboprobes for tm6sf2 were synthesized by in vitro transcription from vectors containing the open reading frame cloned from 5 dpf larval cDNA (primers available upon request). Transcription was carried out using the DIG RNA Labeling Kit (T7) (Roche, Indianapolis, IN) and precipitated with ammonium-acetate.

xiii) Drug treatments:

Injected larvae (control MO, tm6sf2_e4 MO, tm6sf2_e4 + TM6SF2 WT RNA, tm6sf2_e4 + TM6SF2 E167K RNA, TM6SF2 WT RNA, or TM6SF2 E167K RNA) were cultured until 96 hpf at which point embryos were incubated in one of three culture mediums for 48 hours: i) 1% DMSO 7

Hepatology

in embryo medium; ii) 1 μg/mL Tunicamycin (T7765-1MG; Sigma-Aldrich, St. Louis, MO) in embryo medium. At 5 dpf, livers from individual larvae were harvest and processed for RNA (TRIzol® reagent (Invitrogen, Grand Island, NY)) and cDNA synthesis (RevertAid First Strand cDNA Synthesis Kit, Thermo Scientific, Waltham, MA).

xiv) Statistical analyses:

A Student’s t-test was used to determine significance. Bonferroni correction was used to adjust for multiple comparisons (at least four independent experiments in all cases).

3. Cell culture studies

i) RNA extraction and qRT-PCR:

Total RNA of pooled human multiple tissue samples (control) were purchased (Clontech laboratory Inc., Mountain View, CA). Total RNA of from 24 liver biopsies of EE and EK carriers were obtained from bariatric surgery patients at the Geisinger Health System in Danville, PA. Six EE and EK carriers had been diagnosed with grade 0 or 1 steatosis but no fibrosis, 6 EE and EK with both steatosis (grade 1 and 2) and fibrosis. Patients were all female and morbidly obese. Multiple mouse tissues from adult C56/BL6J mice were homogenized in TRIzol® reagent (Invitrogen, Grand Island, NY). Total RNA was extracted according to manufacturer’s instruction, and cDNA synthesized using the Transcriptor First Strand cDNA Synthesis Kit (Roche, Basel, Switzerland). Quantitative real time PCR was performed in triplicate for each sample on a LightCycler 480 (Roche, Basel, Switzerland) using SYBR Green I PCR mixture (Roche Diagnostics Corporation, Indianapolis, IN). Relative fold changes of TM6SF2 to β- ACTIN or 18S mRNA expression levels were calculated using the 2-∆CT method. Primer sequences for SYBR Green assays are shown in Supplemental Table 7.

For Caco-2 cultured cells, total RNA was extracted using Qiagen RNease Kit (Qiagen, Valencia, CA), and cDNA synthesized using Transcriptor First Strand cDNA Synthesis Kit (Roche, Basel, Switzerland). Steady-state mRNA levels were determined by two-step quantitative real time PCR (qRT-PCR) using the LightCycler 480 (Roche, Basel Switzerland) and Taqman probe/primer sets (Applied Biosystems, Carlsbad, CA), with 18S as an internal control for normalization. Primer sequences for Taqman assays are shown in Supplemental Table 7.

ii) Molecular cloning:To generate the YFP fusion vector, cDNAs of TM6SF2 were amplified from human cDNAs by PCR to include BgIII/SalI restriction sites at the 5’ and 3’ ends; product was subsequently cloned in frame with monomeric pEYFP-C1 (13). See 2. Zebrafish studies: v) Molecular cloning and transcript expression for more information as to initial human TM6SF2 vector construction, which was subsequently utilized to generate the above construct.

Two sets of DNA oligonucleotides encoding shRNA targeting TM6SF2 (shRNA1: GCACTGTTCACTACCTCCTCT; shRNA2: GTGCTTGATTGCCCCACAGAT) and one set of scrambled shRNA (control shRNA GCACCAACATCGAAAGTGACT), which does not match any known human genes, were first cloned into the pEntra shuttle vector (Clontech, Mountain View, CA). The TM6SF2 and control shRNA inserts were then subcloned into the lentiviral destination 8

Hepatology

vector pSMPUW-CMV-DEST (Cell Biolabs, San Diego, CA) under a CMV promoter using standard Gateway LR cloning protocol according to the manufacturer’s instructions (Invitrogen, Carlsbad, CA). All constructs were verified by restriction enzyme digestion and by DNA sequencing analysis.

iii) Caco-2 cells cultures and gene knockdown by shRNA:

Caco-2 cell cultures were obtained from American Type Culture Collection (Mannassas, VA) and grown in DMEM with 4.5 g/L glucose, 4 mM glutamine, 100 U/mL penicillin, 100 mg/mL streptomycin and 20% FBS in a 95% air-5% CO2 atmosphere at 37°C, as previously described (18). The medium was changed every other day. Cells were plated at a density 1-2 104 cells/cm2 in 25-cm2 flasks and split with 0.25% trypsin-1 mM EDTA when they reached 70–90% confluence. For experiments, cells were plated at 107 cells/cm2 in 12-well plates for gene expression studies or in confocal dishes for imaging. Lentiviral plasmid vectors containing shRNA inserts targeting human TM6SF2 (GenBank accession NM_001001524) were obtained (see below). Caco-2 cells were transduced with an equal mix of lentiviral plasmid vector containing shRNA targeting human TM6SF2 (GeneBank accession NM_001001524) or scrambled shRNA (see ii) Molecular cloning section above) that does not match any known human genes, with identical transduction procedures. After 48 hours, cells were analyzed for gene expression or by confocal image analysis for CLD content. Confocal imaging of live cells was performed at 37°C and 5% CO2 using a Zeiss LSM510 microscope equipped with an S-M incubator (Carl Zeiss MicroImaging, Inc. Thornwood, NY), and controlled by a CTI temperature regulator along with humidification and an objective heater as previously described (19). For pulse/chase radiolabeling experiments, cells were differentiated as previously described (20) and lentiviral transduction performed 48 hours before the start of the pulse experiments (see below).

iv) Lentiviral vector production:

HEK293T cells were cultured in Dulbecco's modified Eagle's medium (DMEM) containing 10% FBS, 100 μg/mL streptomycin and 100 U/mL penicillin. Cells were allowed to reach 90% confluence at which time they were transfected in the presence of DMEM with 1.2 μg of pSMPUW lentiviral expressing vector containing shRNA targeting TM6SF2 or scrambled shRNA in (Cell Biolabs, San Diego, CA) (see ii) Molecular cloning section above) 1.2 μg of pCD/NL- BH*DDD (Addgene plasmid 17531; plasmid was generously made available to Addgene biobank by Dr. Jacob Reiser) and 0.2 μg of pVSVG (Cell Biolabs, San Diego, CA), for each well of a 6-well plate using LipoD293 (SignaGen Laboratories, Rockville, MD). Liposome/DNA complex-containing medium was removed and replaced with fresh medium 16 hours post transfection. The supernatant containing the viral vector particles was collected at 48 and 72 hours following initial transfection, combined and aliquoted for storage at −80°C. Lentiviral vectors used for all experiments had a minimum titer of 5×105 IFU/mL.

v) Detection lipid content by Bodipy 493/503 staining:

Hepatology

Lipid droplet visualization and cell counting (nuclei) via image analysis was performed as previously described (21).

vi) Cell incubation:

Radiolabeled [3H]oleic acid was used to track synthesis and secretion of lipids; these experiments had two incubation periods designated pulse and washout (22). The experimental design was similar to a protocol previously described (22). During the pulse period, cells were incubated for 16 hours in culture medium containing 0.6 mM [3H]oleic acid complexed to fatty acid-free bovine serum albumin (BSA) (Lot number 99, EDM Millipore, MA) at a 4:1 molar ratio. For the washout period, cells were incubated in culture medium containing 5 g fatty acid-free BSA per liter of medium for 1 hour, providing sufficient time for the secretion of lipoproteins assembled during the pulse period. Separate samples of cells and media were collected after the pulse and pulse/washout. Lipids were extracted by the Dole extraction method (23), and 12.5% of the total lipid from the cell samples and 25% from the media samples were analyzed by thin layer chromatography (24). [3H]oleic acid incorporation into the TG fraction in the media and in the cells was calculated as picomoles and nanomoles, respectively.

vii) Statistical analysis: Statistical significance was tested using either one-way analysis of variance or a two-tailed Student's t-test (GraphPad Software Inc., La Jolla, CA).

Hepatology

Supplemental Material and Methods references:

1. Kleiner DE, Brunt EM, Van Natta M, Behling C, Contos MJ, Cummings OW, Ferrell LD, et al. Design and validation of a histological scoring system for nonalcoholic fatty liver disease. Hepatology 2005;41:1313-1321. 2. Gerhard GS, Chokshi R, Still CD, Benotti P, Wood GC, Freedman-Weiss M, Rider C, et al. The influence of iron status and genetic polymorphisms in the HFE gene on the risk for postoperative complications after bariatric surgery: a prospective cohort study in 1,064 patients. Patient Saf Surg 2011;5:1. 3. Lee WJ, Pollin TI, O'Connell JR, Agarwala R, Schaffer AA. PedHunter 2.0 and its usage to characterize the founder structure of the Old Order Amish of Lancaster County. BMC Med Genet 2010;11:68. 4. Hsueh WC, Mitchell BD, Aburomia R, Pollin T, Sakul H, Gelder Ehm M, Michelsen BK, et al. Diabetes in the Old Order Amish: characterization and heritability analysis of the Amish Family Diabetes Study. Diabetes Care 2000;23:595-601. 5. Mitchell BD, McArdle PF, Shen H, Rampersaud E, Pollin TI, Bielak LF, Jaquish C, et al. The genetic response to short-term interventions affecting cardiovascular function: rationale and design of the Heredity and Phenotype Intervention (HAPI) Heart Study. Am Heart J 2008;155:823-828. 6. Pollin TI, Damcott CM, Shen H, Ott SH, Shelton J, Horenstein RB, Post W, et al. A null mutation in human APOC3 confers a favorable plasma lipid profile and apparent cardioprotection. Science 2008;322:1702-1705. 7. Albert JS, Yerges-Armstrong LM, Horenstein RB, Pollin TI, Sreenivasan UT, Chai S, Blaner WS, et al. Null mutation in hormone-sensitive lipase gene and risk of type 2 diabetes. N Engl J Med 2014;370:2307-2315. 8. Chu X, Erdman R, Susek M, Gerst H, Derr K, Al-Agha M, Wood GC, et al. Association of morbid obesity with FTO and INSIG2 allelic variants. Arch Surg 2008;143:235-240; discussion 241. 9. Verma SS, de Andrade M, Tromp G, Kuivaniemi H, Pugh E, Namjou-Khales B, Mukherjee S, et al. Imputation and quality control steps for combining multiple genome-wide datasets. Front Genet 2014;5:370. 10. Cinaroglu A, Gao C, Imrie D, Sadler KC. Activating transcription factor 6 plays protective and pathological roles in steatosis due to endoplasmic reticulum stress in zebrafish. Hepatology 2011;54:495-508. 11. Westerfield M. The Zebrafish Book. A Guide for the Laboratory Use of Zebrafish (Danio rerio). 4 ed. Eugene, OR: University of Oregon Press, 2000. 12. Hwang WY, Fu Y, Reyon D, Maeder ML, Tsai SQ, Sander JD, Peterson RT, et al. Efficient genome editing in zebrafish using a CRISPR-Cas system. Nat Biotech 2013;31:227- 229. 13. Wang H, Bell M, Sreenivasan U, Hu H, Liu J, Dalen K, Londos C, et al. Unique regulation of adipose triglyceride lipase (ATGL) by perilipin 5, a lipid droplet-associated protein. J Biol Chem 2011;286:15707-15715. 14. O'Hare EA, Wang X, Montasser ME, Chang YP, Mitchell BD, Zaghloul NA. Disruption of ldlr causes increased LDL-c and vascular lipid accumulation in a zebrafish model of hypercholesterolemia. J Lipid Res 2014;55:2242-2253. 15. O'Hare EA, Wang X, Montasser ME, Chang Y-PC, Mitchell BD, Zaghloul NA. Disruption of ldlr causes increased LDL-cholesterol and vascular lipid accumulation in a zebrafish model of hypercholesterolemia. Journal of Lipid Research 2014. 11

Hepatology

16. Walters James W, Anderson Jennifer L, Bittman R, Pack M, Farber Steven A. Visualization of Lipid Metabolism in the Zebrafish Intestine Reveals a Relationship between NPC1L1- Mediated Cholesterol Uptake and Dietary Fatty Acid. Chemistry & Biology 2012;19:913-925. 17. O’Hare EA, Wang X, Montasser ME, Chang Y-PC, Mitchell BD, Zaghloul NA. Disruption of ldlr causes increased LDL-c and vascular lipid accumulation in a zebrafish model of hypercholesterolemia. Journal of Lipid Research 2014;55:2242-2253. 18. Levy E, Mehran M, Seidman E. Caco-2 cells as a model for intestinal lipoprotein synthesis and secretion. FASEB J 1995;9:626-635. 19. Bell M, Wang H, Chen H, McLenithan JC, Gong DW, Yang RZ, Yu D, et al. Consequences of lipid droplet coat protein downregulation in liver cells: abnormal lipid droplet metabolism and induction of insulin resistance. Diabetes 2008;57:2037-2045. 20. Luchoomun J, Hussain MM. Assembly and secretion of chylomicrons by differentiated Caco-2 cells. Nascent triglycerides and preformed phospholipids are preferentially used for lipoprotein assembly. J Biol Chem 1999;274:19565-19572. 21. Beller M, Sztalryd C, Southall N, Bell M, Jackle H, Auld DS, Oliver B. COPI complex is a regulator of lipid homeostasis. PLoS Biol 2008;6:e292. 22. Caviglia JM, Sparks JD, Toraskar N, Brinker AM, Yin TC, Dixon JL, Brasaemle DL. ABHD5/CGI-58 facilitates the assembly and secretion of apolipoprotein B lipoproteins by McA RH7777 rat hepatoma cells. Biochim Biophys Acta 2009;1791:198-205. 23. Dole VP. Fractionation of plasma nonesterified fatty acids. Proc Soc Exp Biol Med 1956;93:532-533. 24. Sztalryd C, Bell M, Lu X, Mertz P, Hickenbottom S, Chang BH, Chan L, et al. Functional compensation for adipose differentiation-related protein (ADFP) by Tip47 in an ADFP null embryonic cell line. J Biol Chem 2006;281:34341-34348.

Hepatology

Supplemental Table 1. Demographic, Anthropometric and Lipid characteristics in the study participants enrolled in the Bariatric Surgery Program at the Geisinger Clinic Center for Nutrition and Weight Management According to TM6SF2 T allele Genotype

Mean (SE) by Genotype P CC - / T value N=853 N=130 Sex M/F M/F 169/684 24/101 Ethnic White/Caucasian White/Caucasian Age 45.9 (0.4) 46.5 (1.0) 0.61 Body Mass Index* (kg/m2) 49.3 (0.3) 49.8 (0.8) 0.55 Total Cholesterol* (mg/dL) 189.2 (1.4) 183.4 (3.6) 0.10 HDL-C* (mg/dL) 46.8 (0.4) 47.3 (0.9) 0.52 LDL-C* (mg/dL) 107.3 (1.2) 104.4 (3.1) 0.14 Triglycerides*,† (mg/dL) 182.6 (4.6) 160.9 (6.7) 0.20

*Age-, sex- and lipids medication adjusted mean and standard error (SE) are presented for each genotype group

† trait was log-transformed for analysis (p-value) and untransformed values are presented for genotype mean and SE values

& CT n=127 and TT n=3

Hepatology

Supplemental Table 2. Liver histology features in the study participants enrolled in the Bariatric Surgery Program at the Geisinger Clinic Center for Nutrition and Weight Management According to TM6SF2 T allele Genotype

Mean (SE) by Genotype P CC - / T value N=853 N=130 Steatosis *,§ (mean (SE)) 1.14 (0.03) 1.40 (0.08) 0.04

¥ Lobular inflammation (% > grade 0) 32.4 43.8 0.02 ¥ Hepatocyte Ballooning (% > grade 0) 28 33.3 0.26 ¥ Perivenular fibrosis (% > grade 0) 17.5 26.6 0.008 Cirrhosis¥ (% > grade 0) 2.1 1.6 0.9

*Age-, sex- and lipids medication adjusted mean and standard error (SE) are presented for each genotype group

§ Steatosis was examined histologically and graded in severity from 0 (no steatosis) to 3 (severe steatosis)

¥Dichotomized grade 0 vs. >0 for inflammation, ballooning, fibrosis and cirrhosis for statistical analysis

Hepatology

Supplemental Table 3. Lipid traits in the study participants with a steatosis grade superior of 0 enrolled in the Bariatric Surgery Program at the Geisinger Clinic Center for Nutrition and Weight Management According to TM6SF2 T allele Genotype

Steatosis grade > 0 Mean (SE) by Genotype Beta (SE) p-value CC -/T

N=584 N=103 Total Cholesterol (mg/dl)* 191.2 (1.7) 181.5 (4.1) -11.24 (5.34) 0.04 HDL (mg/dl)* 45.5 (0.5) 46.5 (1.0) 1.41 (1.27) 0.27 LDL (mg/dl)* 107.7 (1.4) 103.0 (3.5) -7.06 (4.57) 0.12 Triglycerides (mg/dl)*† 200.5 (6.2) 162.5 (7.2) -0.15 (0.05) 0.02

SE=Standard Error

*adjusted for age, sex, lipid medication

† trait was log transformed for analysis (p value) and untransformed values are presented for genotype mean, beta and SE values

Hepatology

Supplemental Table 4. Demographic, Antropometric and Lipid Characteristics of Old Order Mean (SE) by Genotype Clinical Characteristics CC CT TT Beta (SE) p-value N=2771 N=734 N=51 Sex M/F M/F M/F 0.003 1260/1511 341/393 16/35 (0.020) 0.90 Ethnic White/Caucasian White/Caucasian White/Caucasian Age 46.2 (0.3) 47.2 (0.6) 48.3 (2.5) 0.95 (0.58) 0.10 Body Mass Index (kg/m2)* 27 (0.1) 26.9 (0.2) 27.4 (0.6) -0.11 (0.20) 0.59 Total Cholesterol (mg/dL)* 210.3.0 (0.9) 207.5 (1.7) 178.1 (7.8) -8.54 (1.84) 3.6x10-6 HDL-C (mg/dL)* 57.4 (0.3) 58.8 (0.6) 64.8 (2.6) 2.48 (0.61) 5.0x10-5 LDL-C (mg/dL)* 136.8 (0.8) 133.9 (1.6) 101.0 (6.9) -9.45 (1.68) 2.1x10-8 Triglycerides (mg/dL)*,† 80.3 (1.0) 73.9 (1.7) 61.7 (6.6) -0.10 (0.02) 7.1x10-7

Amish Study Participants (N=3556), According to TM6SF2 T allele Genotype

SE=Standard Error

*adjusted for age, sex, study

† trait was log transformed for analysis (p value) and untransformed values are presented for genotype mean, beta and SE values

Hepatology

Supplemental Table 5. Metabolic Characteristics of Old Older Amish Study Participants (N=2059), According to TM6SF2 T allele Genotype

Mean (SE) by Genotype CC CT TT Beta (SE) p-value N=1597 N=439 N=23 101.6 Glucose (mg/dl)* 90.5 (0.5) 88.9 (0.8) (10.4) 0.88 (0.8) 0.29 Insulin (mu/ml)* 10.9 (0.2) 10.6 (0.3) 10.8 (0.8) 0.01 (0.02) 0.57 HOMA-IR* 2.5 (0.1) 2.4 (0.1) 2.8 (0.4) 0.02 (0.03) 0.49

SE=Standard Error

*adjusted for age, sex, study and diabetes status

Hepatology

Supplemental Table 6. Effect of TM6SF2 T allele Genotype on the Lipoprotein Subclass Profile in Old Order Amish Study Participants (N=833)

Mean (SE) by Genotype CC CT TT Beta (SE) p-value N=631 N=191 N=11 210.7 205.1 157.6 Total Cholesterol* (1.8) (3.6) (12.8) -17.7 (3.4) 2.5x10-7 56.1 HDL-C* (0.5) 57.9 (1.0) 52.6 (3.7) 1.7 (1) 0.09 14.7 HDL2-C*,† (0.2) 15.5 (0.5) 13.2 (1.9) 0.88 (0.48) 0.10 41.4 HDL3-C* (0.3) 42.3 (0.6) 39.5 (1.9) 0.84 (0.58) 0.15 160.6 153.9 Total Non-HDL-C* (1.7) (3.3) 112.0 (9.7) -19.2 (3.2) 2.7x10-9 10.2 IDL-C*,† (0.4) 8.5 (0.6) 5.5 (1.4) -2.93 (0.66) 1.9x10-6 126.2 121.7 LDL-C* (1.4) (2.7) 82.7 (8.3) -14.8 (2.7) 3.5x10-8 17.4 Total VLDL-C*,† (0.2) 16.7 (0.3) 16.3 (1.3) -1.4 (0.39) 2.4x10-4 VLDL3-C*,† 9.5 (0.1) 9.1 (0.2) 8.7 (0.6) -0.83 (0.23) 2.2x10-4 Remnant Lipoprotein- 19.4 C*,† (0.5) 17.3 (0.8) 14.0 (2.0) -3.9 (0.87) 9.8x10-6 69.5 Triglycerides*,† (1.7) 64.1 (2.7) 64.8 (13.7) -9.2 (3.0) 8.7x10-4

SE=Standard Error

*adjusted for age, sex, study

† trait was log transformed for analysis (p value) and untransformed values are presented for genotype mean, beta and SE values

Hepatology

Supplemental Table 7. Sequences of human and mouse primers used for qRT-PCR measurements

Gene Forward primer Reverse primer Taqman Human 18s CTCAACACGGGAAACCTCAC CGCTCCACCAACTAAGAACG Human GRP78 CAGATGAAGCTGTAGCGTATGG ACATACATGAAGCAGTACCAGGTC Human CHOP CAGAGCTGGAACCTGAGGAG TGGATCAGTCTGGAAAAGCA Human ATF4 GGTCAGTCCCTCCAACAACA CTATACCCAACAGGGCATCC Human ATF6 CTTTTAGCCCGGGACTCTTT TCAGCAAAGAGAGCAGAATCC Human TM6SF2 TTCCTTACAGGAAACATTCTTGG GCAGGTAGGGGATGGTGAG SYBR Green Mouse Β-actin CAGCTTCTTTGCAGCTCCTT CACGATGGAGGGAATACAG Mouse tm6sf2 GGTATTTGCTGGAGCCATTG AGGTAGCCCAGGTGTCCTCT Human Β-ACTIN AGAAAATCTGGCACCACACC GGGTGTTGAAGGTCTCAAA Human TM6SF2 GCTGCCTATGCTCTCACCTT ACACGGTAGGTGAAGGGTGT Human Spliced CTGAGTCCGAATCAGGTGCAG ATCCATGGGGAGATGTTCTGG XBP1 Human Unspliced CAGCACTCAGACTACGTGCA ATCCATGGGGAGATGTTCTGG XBP1

Hepatology

Supplemental Figure Legends

Supplemental Figure 1 Expression of tm6sf2 in larval zebrafish. (A) Wholemount in situ hybridization in 4 dpf larvae using antisense riboprobe for tm6sf2. Robust expression in posterior intestine (dashed box, magnified below). (B) qRT-PCR expression of either beta-actin or tm6sf2, relative to beta-actin, in cDNA generated from mRNA extracted from 5 dpf larval liver. (C) qRT-PCR expression of either beta-actin or tm6sf2, relative to beta-actin, in cDNA generated from mRNA extracted from adult zebrafish liver or intestine.

Supplemental Figure 2. Validation of efficacy and specificity of tm6sf2 MOs in zebrafish. (A) qRT-PCR quantification of tm6sf2 transcript levels in embryos injected with control MO, tm6sf2_e3 MO or tm6sf2_e4 MO at 1, 3, and 5 dpf. Data represent average expression from homogenates of 20 pooled embryos from each of two experiments per treatment. (B) Gel electrophoresis results of RT-PCR from embryo homogenates at 1, 3 or 5 dpf treated with tm6sf2_e3 MO, tm6sf4_e4 MO or control MO. A 261 bp from the tm6sf2_e3 MO and a 271 bp from the tm6sf2_e4 MO was detected through 5 dpf, in addition to the 365 bp wild-type transcript. (C) qRT-PCR quantification of ∆113 p53 expression, diagnostic marker for off-target MO toxicity. (D) Gel electrophoresis results of RT-PCR from embryo homogenates at 1, 3 or 5 dpf injected with dgat2 MO and control MO. A spliced 373 bp from the dgat2 MO was detected through 5 dpf, in addition to the 502 bp wild-type transcript. ** p≤0.005. (E) mRNA expression levels for dgat2, qPCR analysis (n = 3 for each group, data are means + SEM

Supplemental Figure 3. Representative 5 dpf larvae used to assess lipid accumulation. Representative image of a control MO (A-A”) and tm6sf2_e4 MO (B-B’) zebrafish embryos at 5 dpf stained for neutral lipids using oil red O staining. Features used for assessment of phenotypes: yellow line, liver; red box, area imaged under higher magnification in lateral view liver images; green box, 49 nm2 area used for quantification of total lipid droplet number and size.

Supplemental Figure 4. Targeting tm6sf2 by CRISPR/Cas9 recapitulates hepatic steatosis phenotype. (A-C) Representative lateral images of liver in oil red O stained 5 dpf zebrafish F0 larvae targeted by CRISPR/Cas9 against tm6sf2 exons 3 or 4. (D) The severity of hepatic lipid accumulation by targeted exon was similar to that elicited by MO against each exon. (E-F) Gel electrophoresis images of PCR products amplified from the targeted region of genomic DNA extracted from individual embryos and digested by T7 endonuclease to detect mismatch. (G-H) Sequencing results of individual mutations imparted in each exon targeted by CRISPR/Cas9.

Hepatology

Supplemental Figure 5. TM6SF2 is predominately expressed in liver and enterocytes, is regulated by dietary feeding and is mainly located in ER. (A) Expression of TM6SF2 mRNA in human tissues. Expression of TM6SF2 was normalized to that of ACTB. (B) Expression of Tm6sf2 mRNA in mouse tissues (C57/BL strain). (C) Caco-2 cells were co-transfected with KDEL-DsRed (ER retention peptide) and TM6SF2-GFP and imaged by confocal microscopy after 24 hours. Micrographs depict one representative cell of ten observed in two experiments. Bar: 10 μm.

Supplemental Figure 6. Morphological characterization of hepatic liver stress in tm6sf2 morphants. (A) Representative TEM images of larval livers from unfed control MO or tm6sf2_e4 MO larvae. (B) Quantification of average ER lumen width measured from. Data represent average width of 10 randomly selected smooth ER lumens per cell, three measurements per lumen, n=15 cells per treatment. (E) qRT-PCR quantification of gene expression for markers of ER stress from livers dissected from 5 dpf control MO or tm6sf2_e4 MO larvae. Data represent average of triplicated values from 2 pooled groups of 40 larval livers each per experiment for two separate experiments. *p<0.05; **p<0.005; *** p≤3x10-64

Supplemental Figure 7. ER stress markers measured in zebrafish liver after MO-mediated gene knockdown or human TM6SF2 overexpression and Drug Treatment. (A) Expression of ER stress markers in larvae treated 1mg/ml Tunicamycin for 48 hours, (B) Expression of ER stress markers in larvae treated with DMSO for 48 hours, Significant expression differences were observed in tm6sf2_e4 MO, overexpression for chop, bip, and xbp1. Values relative to control MO; gene expression normalized to β-actin. N=40 larval livers * p≤0.01; ** p≤0.001

Supplemental Figure 8. Validation of human TM6SF2 over expression in livers extracted from tunicamycin treated zebrafish larvae. Expression of human TM6SF2 mRNA in larvae with or without co-injection tm6sf2_e4 MO and treated with 1mg/ml Tunicamycin for 48 hours. Significantly increased expression was observed in zebrafish injected with WT mRNA compared to control MO (control), gene expression normalized to β-actin. N=20 pooled larvae, triplicated., ** p≤0.001

Supplementary Figure 9. Larval length measurement. Average length of 5 dpf larvae injected with control MO, tm6sf2_e4 MO, Cas9 mRNA only, or Cas9 mRNA plus gRNA targeted against tm6sf2 exon 4 (tm6sf2 exon4 CRISPR).

Hepatology

Supplemental Figure 1

B 3 Relative expression in zebrafish liver actin

- 2.5

1.5

0.5

Expression relativebeta to 0 beta-actin tm6sf2 C 5 Gene expression in adult zebrafish 4.5 4 3.5 3 2.5 beta-actin 2 tm6sf2 1.5 1 0.5 0 Expression relative to beta actin LIVER INTESTINE

Hepatology

Supplemental Figure 3

250 μm control MO A’ A’’

50 μm 50 μm control MO B B’

tm6sf2_e4 MO

Hepatology

Supplemental Figure 4

Oil Red O Staining A D _e3 _e4 only - only - SB Cas9 tm6sf2 CRISPR tm6sf2 CRISPR Cas9 L

50 μm no staining severe B T7 Endonuclease I Assay

_e3 CRISPR E F only only - - tm6sf2 tm6sf2_e3 CRISPR tm6sf2_e4 CRISPR Cas9 Cas9

C 438 bp 510 bp 400 bp * 450 bp * * 400 bp _e4 CRISPR tm6sf2

intron 3 G tm6sf2 wild-type >aaatttgctctaatatttactattgattattgttttgtattttcagGCCAGTGGTTATTCTTGAA> tm6sf2_e3 CRISPR >aaatttgctctaatattt------>

tm6sf2 wild-type >ATTGGCGTGGCTGTTCTTGTGGCAGTTTTCCTCCTGGTCTACCTCGCTACGCGCTTCAATCCTCC> tm6sf2_e3 CRISPR >------> intron 4 tm6sf2 wild-type >TAAGGATCCTTTGTTCTATGgtatgatttgattggttgattcttaaaaacactgaatttatccac> tm6sf2_e3 CRISPR >------gttgattcttaaaaacactgaatttatccac>

intron 4 tm6sf2 wild-type >taaacaagtattactgtattgactttgtttccagTTTTTGCAGAGTTTTCCTTCACGTGTGTCAT> H tm6sf2_e4 CRISPR >taaacaagtattactgtattgactttgtttccagTTTTTGCAGAGTTTTCCTTCACGTGTGTCAT>

tm6sf2 wild-type >TGACCTCACGAGTGCCCTAGAATATGATGGCTTCGCCTCTGGGTTCATGGAGTTCTACCAGAAAA> tm6sf2-e4 CRISPR >TGACCTCACGAGTGCCCTAGAATATGATGGCTTCGCCTCTGGGCTCATGGAGTTCTACCAGAAAA> intron 5 tm6sf2 wild-type >CGgttagcgtaatgacatgaggagaaagacaagttgatcttatcgattgcagtataaatcttccc> tm6sf2_e4 CRISPR >CGgttagcgtaatgacatgaggagaaagacaagttgatcttatcgattgcagtataaatcttccc> 25

Hepatology

Supplemental Figure 5

A human

B mouse

TM6SF2-GFP RFP-KDEL Merge D

Hepatology

Supplemental Figure 6

A A control MO, unfed tm6sf2MO, unfed B

***

0.1 ? m 0.1 ? m

Hepatology

Supplemental Figure 7

A ER Stress Gene Expression - Tunicamycin-Treated Embryos 2.5 ** control MO tm6sf2_e4 MO 2 Ratio ** tm6sf2_e4 MO + WT RNA tm6sf2 WT RNA actin actin - β 1.5 * *

(% Relative to control MO) 0.5 Gene Expression:

0 chop bip edem1 xbp1_splice

B ER Stress Gene Expression - DMSO-Treated Embryos 6

5 ** Ratio 4 actin actin - β 3 **

2 ** * (% Relative to control MO)

Gene Expression: 1

0 chop bip edem1 xbp1_splice

Hepatology

Supplemental Figure 8

5.00 Tunicamycin Control Tunicamycin tm6sf2_e4 aO + WT RNA

actin 4.00 Tunicamycin + WT RNA β C ** 3.00

2.00 ** mRNA levels

Relative TM6SF2 Relative 1.00

0.00

Hepatology

Supplemental Figure 9

Larvae Length 4.5 4 3.5 3 2.5 2 1.5 1 0.5 Larvae Length (mm) 5 dpf at 0 control MO tm6sf2_e4 MO Cas9-only injected tm6sf2 exon 4 control CRISPR

Hepatology