Introduction
Hyperhomocysteinemia is a common, independent
risk factor for cardiovascular disease (1–9). Up to 40%
of patients diagnosed with premature coronary artery,
cerebrovascular, or peripheral vascular disease are
reported to have hyperhomocysteinemia. Patients
with severe hyperhomocysteinemia due to cystathio-
nine β-synthase (CBS) deficiency also exhibit a wide
range of clinical manifestations, including athero-
sclerosis, thrombosis, mental retardation, ectopia
lentis, osteoporosis, and skeletal abnormalities. These
patients also develop hepatic steatosis or “fatty liver,”
which is characterized by enlarged, multinucleated
hepatocytes containing microvesicular lipid droplets
(10, 11). Consistent with these findings, homozygous
CBS-deficient mice having severe hyperhomocys-
teinemia also develop hepatic steatosis (12).
Several mechanisms have been proposed to explain
the pathological changes associated with hyperhomo-
cysteinemia. Homocysteine has been shown to cause
cell injury when administered to animals (13, 14) or
when added directly to cultured mammalian cells
(15–17). Furthermore, cultured vascular endothelial
cells from patients with heterozygous CBS deficiency
are more sensitive to homocysteine-induced damage
than are wild-type cells (17). Because homocysteine is
generated intracellularly and can accumulate in cells
(18, 19), it has been suggested that homocysteine
could act intracellularly to directly modulate the activ-
ity of both large (enzymes, receptors) and small mole-
cules (nitric oxide, glutathione) (20).
We have reported that homocysteine causes protein
misfolding in the endoplasmic reticulum (ER) and acti-
vates the unfolded protein response (UPR), leading to
The Journal of Clinical Investigation | May 2001 | Volume 107 | Number 10 1263
Homocysteine-induced endoplasmic
reticulum stress causes dysregulation
of the cholesterol and triglyceride biosynthetic pathways
Geoff H. Werstuck,1 Steven R. Lentz,2 Sanjana Dayal,2 Gazi S. Hossain,1
Sudesh K. Sood,1 Yuan Y. Shi,1 Ji Zhou,1 Nobuyo Maeda,3 Skaidrite K. Krisans,4
M. Rene Malinow,5 and Richard C. Austin1
1Department of Pathology and Molecular Medicine, McMaster University and the Hamilton Civic Hospitals Research Centre,
Hamilton, Ontario, Canada
2Department of Internal Medicine, University of Iowa College of Medicine and Veterans Affairs Medical Center, Iowa City,
Iowa, USA
3Department of Pathology and Program for Molecular Biology and Genetics, University of North Carolina, Chapel Hill,
North Carolina, USA
4Department of Biology, San Diego State University, San Diego, California, USA
5Oregon Regional Primate Research Center, Beaverton, Oregon, USA
Address correspondence to: Richard C. Austin, Hamilton Civic Hospitals Research Centre, 711 Concession Street, Hamilton,
Ontario, L8V 1C3, Canada. Phone: (905) 527-2299 ext. 42628; Fax: (905) 575-2646; E-mail: raustin@thrombosis.hhscr.org.
Received for publication October 17, 2000, and accepted in revised form April 9, 2001.
Hepatic steatosis is common in patients having severe hyperhomocysteinemia due to deficiency for
cystathionine β-synthase. However, the mechanism by which homocysteine promotes the develop-
ment and progression of hepatic steatosis is unknown. We report here that homocysteine-induced
endoplasmic reticulum (ER) stress activates both the unfolded protein response and the sterol regu-
latory element–binding proteins (SREBPs) in cultured human hepatocytes as well as vascular
endothelial and aortic smooth muscle cells. Activation of the SREBPs is associated with increased
expression of genes responsible for cholesterol/triglyceride biosynthesis and uptake and with intra-
cellular accumulation of cholesterol. Homocysteine-induced gene expression was inhibited by over-
expression of the ER chaperone, GRP78/BiP, thus demonstrating a direct role of ER stress in the acti-
vation of cholesterol/triglyceride biosynthesis. Consistent with these in vitro findings, cholesterol
and triglycerides were significantly elevated in the livers, but not plasmas, of mice having diet-induced
hyperhomocysteinemia. This effect was not due to impaired hepatic export of lipids because secre-
tion of VLDL-triglyceride was increased in hyperhomocysteinemic mice. These findings suggest a
mechanism by which homocysteine-induced ER stress causes dysregulation of the endogenous sterol
response pathway, leading to increased hepatic biosynthesis and uptake of cholesterol and triglyc-
erides. Furthermore, this mechanism likely explains the development and progression of hepatic
steatosis and possibly atherosclerotic lesions observed in hyperhomocysteinemia.
J. Clin. Invest. 107:1263–1273 (2001).
See related Commentary on pages 1221–1222.
increased expression of the ER stress-response genes,
GRP78/BiP and GADD153 (21, 22). These observa-
tions suggest that homocysteine acts intracellularly to
cause cell dysfunction by perturbing the ER. Consistent
with our findings, another group has also demonstrat-
ed that homocysteine activates the UPR in cultured
vascular endothelial cells, leading to increased expres-
sion of GRP78/BiP as well as two novel ER stress-
response genes, RTP and Herp (23, 24). However, the
mechanism by which the ER stress-inducing effects of
homocysteine contribute to the pathophysiology of
hyperhomocysteinemia has not been established.
An association between UPR activation and lipid
biosynthesis has been demonstrated in yeast (25, 26)
and human fibroblasts (27). In concordance with these
studies, we have demonstrated in cultured human vas-
cular endothelial cells that homocysteine increases
expression of the sterol regulatory element–binding
protein-1 (SREBP-1) (21), an ER membrane-bound
transcription factor that functions to activate genes
encoding enzymes in the cholesterol/triglyceride
biosynthesis and uptake pathways (28, 29). However, it
is unknown whether homocysteine alters the expression
of SREBP-dependent genes that are involved in choles-
terol/triglyceride biosynthesis or whether this effect of
homocysteine influences lipid metabolism in vivo.
In this study, we used cultured human cells and
murine models of hyperhomocysteinemia to test the
hypothesis that homocysteine-induced ER stress alters
cholesterol/triglyceride metabolism. Our results show
that activation of the UPR by homocysteine induces
SREBP-dependent genes that are essential for biosyn-
thesis and uptake of cholesterol and triglycerides and
leads to the accumulation of cholesterol in cultured
cells. Furthermore, a significant increase in the levels of
cholesterol and triglycerides were observed in the livers,
but not plasmas, of hyperhomocysteinemic mice.
These results indicate that homocysteine-induced ER
stress leads to the transcriptional activation of genes
responsible for lipogenesis that likely contribute to
hepatic steatosis in hyperhomocysteinemia.
Methods
Cell culture and treatment conditions. Primary human
umbilical vein endothelial cells (HUVECs) were isolat-
ed by collagenase treatment of human umbilical veins
(30) and cultured in M199 medium (Life Technologies
Inc., Burlington, Ontario, Canada) containing 20
µg/ml endothelial cell growth factor, 90 µg/ml porcine
intestinal heparin, 100 µg/ml penicillin, 100 µg/ml
streptomycin, and 20% FBS (HyClone Laboratories,
Logan, Utah, USA). Cells from passages two to four
were used in these studies. Human aortic smooth mus-
cle cells (HASMCs) were purchased from Cascade Bio-
logics (Portland, Oregon, USA) and cultured in M231
media (Cascade Biologics) containing smooth muscle
cell growth supplement (Cascade Biologics). The
human hepatocarcinoma cell line, HepG2, was
obtained from the American Type Culture Collection
(ATCC; Rockville, Maryland, USA) and cultured in α-
DMEM (Life Technologies Inc.) containing 10% FBS.
The human transitional bladder carcinoma cell line,
T24/83, was obtained from ATCC and cultured in
M199 medium containing 10% FBS. All cells were
maintained in a humidified incubator at 37°C with 5%
CO2. DL-Homocysteine, L-methionine, DL-cysteine,
glycine, DTT, tunicamycin, A23187, and β-mercap-
toethanol were purchased from Sigma Chemical Co.
(St. Louis, Missouri, USA). These compounds were pre-
pared fresh in culture medium, sterilized by filtration,
and added to the cell cultures.
Determination of intracellular levels of homocysteine.
HepG2 cells exposed to 1 or 5 mM homocysteine for 0
to 24 hours were washed three times in α-DMEM and
three times in PBS. Cells were lysed in water by three
freeze/thaw cycles and cellular debris removed by cen-
trifugation. Total homocysteine, defined as the total
concentration of homocysteine after quantitative
reductive cleavage of all disulfide bonds (31), was deter-
mined in cellular lysates using the IMx System (Abbott
Laboratories, Mississauga, Ontario, Canada) and nor-
malized to total protein concentration.
Animals and diets. Wild-type C57BL/6 mice (CBS+/+)
were obtained from Charles River (Montreal, Québec,
Canada) or The Jackson Laboratory (Bar Harbor,
Maine, USA). Heterozygous CBS-deficient mice (CBS+/–)
(12) were crossbred to CBS+/+ mice. Genotyping for the
targeted allele was performed by PCR (12). At the time
of weaning, offspring were fed one of four diets: (a) a
control diet that contained 4.1 g L-methionine/kg and
7.5 mg folate/kg (LM-485; Harlan Teklad Laboratory,
Madison, Wisconsin, USA); (b) a high-methionine diet
that was identical to the control diet except that the
drinking water was supplemented with 0.5% L-methio-
nine; (c) a high-methionine/low-folate diet that con-
tained 1.5 mg folate/kg and drinking water that was
supplemented with 0.5% L-methionine; or (d) a very
high-methionine (24.6 g/kg), low-folate (1.5 g/kg) diet
(TD98272; Harlan Teklad Laboratory) (32). Succinyl-
sulfathiazole (1.0 mg/kg) was added to the low-folate
diets. After 2 to 20 weeks on an experimental diet, mice
were sacrificed with an intraperitoneal injection of 75
mg sodium pentobarbital, plasma was collected in
EDTA (final concentration 5–10 mM) for measure-
ment of total homocysteine, and their tissues removed
and snap-frozen in liquid N2 before storage at –70°C.
Plasma total homocysteine was measured by HPLC and
electrochemical detection as described previously (33).
The experimental protocols were approved by the Uni-
versity of Iowa and Veterans Affairs Animal Care and
Use Committees or the McMaster University Animal
Research Ethics Board.
Hepatic VLDL-triglyceride secretion rate. CBS+/+ mice fed
control or hyperhomocysteinemic diets for 2 weeks
were fasted overnight and hepatic VLDL-triglyceride
secretion rates quantified by the intravenous adminis-
tration of Triton WR1339 (500 mg/kg body weight;
Sigma Chemical Co.) as described previously (34).
1264 The Journal of Clinical Investigation | May 2001 | Volume 107 | Number 10
Blood samples (50 µl) were taken at 0, 2, and 4 hours
after Triton WR1339 injection, and plasma cholesterol
and triglycerides were measured enzymatically as
described below. Hepatic cholesterol and triglyceride
secretion rates were calculated from the slope of the
curve and expressed as micromole per hour per kilo-
gram of body weight.
Histological analysis. Liver tissue from mice fed control
or hyperhomocysteinemic diets for 16 to 20 weeks was
fixed in 4% formalin, embedded in paraffin, and tissue
sections were stained with hematoxylin and eosin, as
described previously (35).
Preparation of total RNA. Total RNA was isolated from
cultured cells or tissues using the RNeasy total RNA kit
(QIAGEN Inc., Mississauga, Ontario, Canada) and
resuspended in diethyl pyrocarbonate–treated water.
Quantification and purity of the RNA was assessed by
A260/A280 absorption, and RNA samples with ratios
above 1.6 were stored at –70°C for further analysis.
Northern blot analysis. Total RNA (10 µg/lane) was size
fractionated on 2.2 M formaldehyde/1.2% agarose gels,
transferred to Zeta-Probe GT nylon membranes (Bio-
Rad Laboratories Inc., Mississauga, Ontario, Canada),
and hybridized using radiolabeled cDNA probes as
described previously (21, 22). Signal intensities were
quantified by densitometric scanning of the autoradi-
ograms using the ImageMaster VDS and Analysis Soft-
ware (Amersham Pharmacia Biotech, Baie d’Urfé,
Québec, Canada). To correct for differences in gel load-
ing, integrated optical densities were normalized to
human GAPDH. The human isopentenyl diphos-
phate:dimethylallyl diphosphate (IPP) isomerase cDNA
encodes an 837-bp DNA fragment from the 3′-untrans-
lated region of the IPP isomerase gene. The cDNA
probes encoding human 3-hydroxy-3-methylglutaryl
coenzyme A (HMG-CoA) reductase and farnesyl
diphosphate (FPP) synthase were kindly provided by
Skaidrite Krisans (San Diego State University, San
Diego, California, USA). Human SREBP-1 cDNA
(number AA568572) was purchased from Genome Sys-
tems (St. Louis, Missouri, USA), and LDL receptor
cDNA was purchased from ATCC. The cDNA probes
encoding GRP78/BiP or GADD153 have been
described previously (21, 22).
Construction of mammalian expression plasmid encoding
human GRP78/BiP. The cDNA encoding the open read-
ing frame of human GRP78/BiP (approximately 1.95
kb) was amplified by RT-PCR using total RNA from
primary HUVECs. Primers used for RT-PCR were syn-
thesized at the Institute for Molecular Biology,
McMaster University. GRP78/BiP cDNA was generat-
ed using SuperScript RNase H– reverse transcriptase
(Life Technologies Inc.) and a primer complimentary
to a sequence in the 3′-untranslated region of the
human GRP78/BiP mRNA transcript (AB10230: 5′-
TAT TAC AGC ACT AGC AGA TCA GTG-3′). For PCR
amplification, the forward primer AB10231 (5′-CTT
AAG CTT GCC ACC ATG AAG CTC TCC CTG GTG
GCC GCG-3′) contained a Kozak consensus sequence
(boldface) prior to the initiating ATG and a terminal
HindIII restriction site (underlined). The reverse
primer AB10232 (5′-AGG CCT CGAG CT ACA ACT
CAT CTT TTT CTG CTG T-3′) contained a terminal
XhoI restriction site (underlined) adjacent to the
authentic termination codon of the GRP78/BiP
cDNA. After PCR was performed, the amplified
GRP78/BiP cDNA was gel-purified using the QIAEX
gel extraction kit (QIAGEN Inc.) and ligated into T-
ended pBluescript KS (Stratagene, La Jolla, California,
USA). The ligation mixture was then used to trans-
form competent DH5α cells (Life Technologies Inc.).
Plasmid DNA was isolated from transformed cells
using the QIAEX miniprep kit (QIAGEN Inc.), digest-
ed with HindIII and XhoI, and the GRP78/BiP cDNA
purified from agarose, as described above. The
GRP78/BiP cDNA was ligated into the HindIII/XhoI
site of the mammalian expression vector pcDNA3.1(+)
(Invitrogen Corp., Carlsbad, California, USA) to pro-
duce the recombinant plasmid, pcDNA3.1(+)-
GRP78/BiP. Authenticity of the human GRP78/BiP
cDNA sequence was confirmed by fluorescence-based
double-stranded DNA sequencing.
Establishment of stable T24/83 cell lines overexpressing
GRP78/BiP. T24/83 cells grown to 30% confluence
were transfected with 5 µg of the pcDNA3.1(+)-
GRP78/BiP expression plasmid using 30 µl of Super-
Fect transfection reagent (QIAGEN Inc.), as described
by the manufacturer. As a vector control, T24/83 cells
were transfected with pcDNA3.1(+) under the same
conditions. Stable transfectants were selected in com-
plete medium containing 1.2 mg/ml G418 (Life Tech-
nologies Inc.) for 2 weeks. G418-resistant clones were
subsequently isolated and cultured in complete medi-
um containing 1.0 mg/ml G418. Overexpression of
GRP78/BiP was assessed using immunoblotting and
indirect immunofluorescence, as described below.
The Journal of Clinical Investigation | May 2001 | Volume 107 | Number 10 1265
Figure 1
Intracellular homocysteine levels in HepG2 cells. HepG2 cells were
cultured in the absence or presence of 1 mM (circles) or 5 mM (tri-
angles) homocysteine for the indicated time periods, washed, and
lysed by three freeze/thaw cycles. Total intracellular homocysteine
was determined and normalized to total protein. Data are present-
ed as the mean ± SD of three separate experiments.
Immunoblot analysis. The anti-KDEL mAb (SPA-
827), which recognizes both GRP78/BiP and
GRP94, was purchased from StressGen Biotech-
nologies Corp. (Victoria, British Columbia, Cana-
da). The anti-GADD153 polyclonal Ab (sc-575) was
purchased from Santa Cruz Biotechnology Inc.
(Santa Cruz, California, USA). The mAb’s reactive
against human SREBP-1 or -2 (clones IgG-2A4 and
IgG-1C6, respectively) were purchased from
PharMingen (Mississauga, Ontario, Canada). Total
protein lysates from mouse tissues or cultured cells
were solubilized in SDS-PAGE sample buffer and
separated on SDS-polyacrylamide gels under reduc-
ing conditions, as described previously (21, 22).
After incubation with the appropriate primary and
horseradish peroxidase-conjugated (HRP-conjugat-
ed) secondary Ab’s (Life Technologies Inc.), the
membranes were developed using the SuperSignal
chemiluminescent substrate (Pierce Chemical Co.,
Rockford, Illinois, USA).
Immunohistochemistry and image analysis. Immunohis-
tochemistry using polyclonal Ab’s against GRP78/BiP
(sc-1050; Santa Cruz Biotechnology Inc.) was per-
formed as described previously (21, 22). Images were
captured with Northern Eclipse image
analysis/archival software (Empix Imaging Inc., Mis-
sissauga, Ontario, Canada).
Uptake of BODIPY FL LDL. Cells treated in the
absence or presence of homocysteine were washed
with PBS and incubated in media containing 10
µg/ml BODIPY FL LDL (Molecular Probes, Eugene,
Oregon, USA). After incubation at 37°C for 2 hours,
cells were washed with PBS, fixed in 3% formaldehyde
in PBS, and the uptake of LDL was detected by fluo-
rescence microscopy.
Total cholesterol and triglyceride levels. Cultured cells or
tissues were homogenized in lysis buffer containing
0.1% Triton X-100. Lysates were saponified, and lipids
were extracted with hexane/isopropanol (3:2) (36). Col-
orimetric cholesterol and triglyceride assays were car-
ried out using the Sigma Diagnostics Cholesterol and
1266 The Journal of Clinical Investigation | May 2001 | Volume 107 | Number 10
Figure 2
Homocysteine induces the expression of GRP78/BiP and GADD153 in HepG2
cells. (a) Northern blot analysis of the steady-state mRNA levels of GRP78/BiP
and GADD153 in HepG2 cells cultured for 4 hours in the absence (control) or
presence of either 5 mM homocysteine, 5 mM cysteine, 5 mM methionine, 10
µg/ml tunicamycin, 2.5 mM DTT, 5 mM homoserine, or 5 mM glycine. Total
RNA (10 µg/lane) was size fractionated by agarose-gel electrophoresis, trans-
ferred to nylon membranes, and subjected to blot hybridization using radiola-
beled cDNA probes encoding human GRP78/BiP or GADD153. Control for
equivalent RNA loading was assessed using a radiolabeled GAPDH cDNA
probe. (b) Immunoblot analysis of GRP78/BiP and GADD153 protein in
HepG2 cells cultured in the absence or presence of 5 mM homocysteine for the
indicated time periods. HepG2 cells were also treated with 2.5 mM DTT or 10
µg/ml tunicamycin for 8 hours. Total protein lysates (40 µg/lane) were sepa-
rated on 12% SDS-polyacrylamide gels under reducing conditions, transferred
to nitrocellulose membranes and immunostained with Ab’s against either
GRP78/BiP (anti-KDEL) or GADD153.
Figure 3
Homocysteine induces the expression of SREBP-1 in HepG2 cells. (a) Immunoblot
analysis of SREBP-1 protein in HepG2 cells cultured in the absence or presence of 5
mM homocysteine for the indicated time periods. Total protein lysates (40 µg/lane)
were separated on 10% SDS-polyacrylamide gels under reducing conditions, trans-
ferred to nitrocellulose membranes, and immunostained with an mAb that recog-
nizes both the precursor (P) and mature (M) forms of SREBP-1. (b) Northern blot
analysis of the steady-state mRNA levels of SREBP-1 in HepG2 cells cultured in the
presence of 5 mM homocysteine for the indicated time periods. Total RNA (10
µg/lane) was size fractionated by agarose-gel electropho