Antioxidants reduce endoplasmic reticulum stress
and improve protein secretion
Jyoti D. Malhotraa, Hongzhi Miaob, Kezhong Zhanga, Anna Wolfsona, Subramaniam Pennathurc, Steven W. Pipeb,
and Randal J. Kaufmana,b,d,1
Departments ofaBiological Chemistry,cInternal Medicine, andbPediatrics, anddHoward Hughes Medical Institute, University of Michigan Medical Center,
Ann Arbor, MI 48109
Communicated by David Ginsburg, University of Michigan Medical School, Ann Arbor, MI, September 26, 2008 (received for review July 22, 2008)
Protein misfolding in the endoplasmic reticulum (ER) contributes to the
pathogenesis of many diseases. Although oxidative stress can disrupt
other has not been explored. We have analyzed expression of coagu-
response (UPR), causes oxidative stress, and induces apoptosis in vitro
and in vivo in mice. Strikingly, antioxidant treatment reduces UPR
activation, oxidative stress, and apoptosis, and increases FVIII secre-
tion in vitro and in vivo. The findings indicate that reactive oxygen
species are a signal generated by misfolded protein in the ER that
cause UPR activation and cell death. Genetic or chemical intervention
to reduce reactive oxygen species improves protein folding and cell
survival and may provide an avenue to treat and/or prevent diseases
of protein misfolding.
factor VIII ? oxidative stress ? unfolded protein response
processes are poorly understood (1). Reactive oxygen species (ROS)
originate during oxygen-using cellular metabolic processes, such as
oxidative phosphorylation within mitochondria. The ER provides a
unique oxidizing environment for protein folding and disulfide bond
formation before transit to the Golgi compartment. During disulfide
bond formation ROS are formed as a product of electron transport
cysteine residues during disulfide bond formation in the ER may
significantly contribute to oxidative stress (2, 3). The unfolded protein
response (UPR) is an adaptive signaling pathway designed to prevent
the accumulation of misfolded proteins in the ER lumen. Studies also
folding (2). The UPR is signaled through the protein kinases inositol-
requiring protein 1? and PKR-related ER kinase and the activating
transcription factor 6? (4, 5). Chronic unresolved accumulation of
unfolded proteins in the ER leads to apoptosis. To elucidate the
oxidative stress, and apoptosis, we have analyzed the secretion of
the X chromosome-linked bleeding disorder hemophilia A. As FVIII
is prone to misfolding in the ER lumen, FVIII expression provides a
unique approach to manipulate the ER stress response that does not
rely on pharmacological intervention, where any connection between
ER stress and ROS would be difficult to dissect.
FVIII is comprised of three domains in the order A1-A2-B-A3-
levels in vivo, the requirements for FVIII secretion have been charac-
terized in cultured cells that express heterologous FVIII genes. These
studies demonstrated that FVIII forms non-disulfide-bonded high
molecular weight aggregates that are retained within the ER through
lthough endoplasmic reticulum (ER) stress and oxidative stress
interaction with the protein chaperones Ig-binding protein (BiP/
GRP78), calnexin, and calreticulin (8–11). In addition, FVIII traffick-
the lectin LMAN1/MCFD2 complex (12, 13). As FVIII is susceptible
to misfolding in the ER, its expression induces transcription of ER
stress-response genes through the UPR (14). Here we show that
unfolded FVIII accumulation in the ER lumen activates the UPR,
causes oxidative stress, and induces apoptosis. Furthermore, antioxi-
dants prevent ER stress-induced oxidative damage, activation of the
UPR, and apoptosis, and improve FVIII secretion. The findings dem-
onstrate an unprecedented link by which protein misfolding in the ER
and ROS act in concert to activate the UPR and cause cell death. In
addition, ROS can cause protein misfolding in the ER and prevent
analyze the relationship between protein misfolding in the ER and
oxidative stress, we analyzed Chinese hamster ovary (CHO)-H9 cells
that were engineered for transcriptional induction of wild-type human
of wtFVIII mRNA and protein, although the majority of the newly
synthesized wtFVIII protein is not secreted from the cell, but rather
of wtFVIII synthesis in these cells causes distention of the ER lumen,
genes (14). TUNEL staining indicated that NaB treatment led to
not induce apoptosis in CHO cells that did not express wtFVIII (data
in the ER lumen can generate ROS, cells were stained with dichlo-
superoxide ion (16). Although DCF fluorescence did increase in
control CHO cells upon treatment with NaB, DCF fluorescence
dramatically increased more than 100 fold upon induction of wtFVIII
expression in CHO-FVIII cells (Fig. 1C).
Antioxidants Prevent Oxidative Stress and Improve wtFVIII Secretion
In Vitro. To begin to address whether oxidative stress induced by
wtFVIII expression interferes with wtFVIII secretion, we asked
whether antioxidants can influence wtFVIII secretion. Butylated hy-
medium of CHO-H9 cells at the time of NaB treatment significantly
reduced apoptosis (Fig. 1B) and DCF fluorescence observed upon
Author contributions: J.D.M. designed research; J.D.M., H.M., K.Z., A.W., and S.P. per-
formed research; J.D.M., S.P., S.W.P., and R.J.K. analyzed data; and J.D.M. and R.J.K. wrote
This article contains supporting information online at www.pnas.org/cgi/content/full/
November 25, 2008 ?
vol. 105 ?
no. 47 ?
did not reduce the apoptosis observed at 24 h (data not shown). BHA
with the amount of activity (data not shown), suggesting that the
secreted FVIII was properly folded. Treatment with weaker antioxi-
dants—N-acetylcysteine or ascorbic acid—had a lesser, more incon-
sistent effect on reducing DCF fluorescence (data not shown) and
treatment reduces oxidative stress and apoptosis in response to in-
of functional wtFVIII. In addition, as the BHA-mediated increase in
wtFVIII secretion into the medium corresponded with a decrease in
intracellular antigen, the increase in wtFVIII production was likely a
consequence of increased secretion and not increased cell survival.
wtFVIII expression, we analyzed CHO cells that were engineered for
tetracycline-induced expression of the vitamin K-dependent clotting
lated within the ER, activated the UPR, and caused apoptosis [sup-
porting information (SI) Fig. S1 A–C]. Under these conditions, there
by addition of BHA to medium (Fig. S1D). BHA treatment increased
wtFVIII or FVII, secreted proteins that are prone to misfolding in the
ER lumen, generates ROS and induces the UPR and apoptosis.
Furthermore, BHA treatment reduces ROS accumulation and apo-
ptosis, and increases secretion of FVII and wtFVIII in cultured mam-
Compared with wtFVIII and B Domain Deletion, 226/N6 Is More
Efficiently Secreted In Vivo. We then asked whether FVIII accumu-
lation in the ER induces ROS in vivo. For these studies, we compared
the effects of expression of FVIII variants that exhibit different secre-
tion efficiencies. Previous studies demonstrated that deletion of the
heavily glycosylated B domain (residues 741-1647; B domain deletion
(226/N6, which retains an additional 226 aa with 6 N-linked glycosyl-
ation sites from the B domain) is secreted ?10-fold more efficiently
ment with the lectin chaperone machinery of the cell (Fig. 1A) (20).
wtFVIII, BDD, and 226/N6 each contain eight disulfide bonds so that
to form hydrogen peroxide. Therefore, comparison of the ROS pro-
duction upon expression of these different forms of FVIII should
hydrodynamic delivery of FVIII DNA expression vectors into the tail
of hepatocytes (Fig. 2A). Injection of these vectors did not cause
significant liver damage or necrosis. However, the levels of serum
alanine aminotransferase and aspartate aminotransferase were slightly
elevated in wtFVIII- and BDD- injected mice after 24 h (Fig. S2C).
Injection of either wtFVIII or BDD vector DNA produced detectable
In contrast, injection of vector encoding 226/N6 produced approxi-
mately seven- to eightfold greater levels of both FVIII antigen and
For all constructs, the amount of functional activity in the plasma
correlated with levels of FVIII antigen, suggesting that the majority of
the secreted antigen was functional. In addition, analysis of FVIII
antigen in liver extracts demonstrated that 226/N6 accumulated to a
improved secretion efficiency (20). These findings show that hydrody-
the majority of the hepatocytes and that 226/N6 is more efficiently
secreted from the liver into the plasma than either wtFVIII or BDD.
wtFVIII and BDD Cause Oxidative Stress, Activate the UPR, and Induce
Apoptosis In Vivo. We then characterized whether accumulation of
FVIII in the ER causes UPR activation. Western blot analysis dem-
onstrated increased expression of BiP, a sentinel marker of UPR
extracts from mice injected with vector encoding wtFVIII or BDD,
2C). In addition, phosphorylation of the ? subunit of eukaryotic
translation initiation factor 2 (eIF2?), a marker for PKR-related ER
kinase activation, and splicing of Xbp1 mRNA, a marker for inositol-
of either wtFVIII or BDD compared with 226/N6 or empty vector.
UPR, expression of 226/N6 did not significantly activate the UPR.
We next explored whether the accumulation of FVIII in the ER
matic shows the domain structure of FVIII and
deletion constructs used in these experiments.
Positions of disulfide bonds and N-linked glyco-
CHO cells (CHO-FVIII) were treated with NaB (5
TUNEL-positive cells were quantified from three
separate experiments. (C) Control CHO cells and
CHO-FVIII cells were treated with NaB for 24 h
before staining with DCF for analysis by flow
cytometry. Where indicated, cells were treated
with NaB in the presence of BHA (10 ?M). CHO
cells were treated with H2O2for 30 min before
DCF staining as a positive control. (D) CHO-FVIII
cells were treated with NaB in the presence or
absence of BHA, ascorbic acid (500 ?M; AA), or
N-acetylcysteine (500 ??; NAC). Anti-oxidants
conditioned medium and cells were harvested
for analysis FVIII antigen. Data represent the
Induction of FVIII expression causes ox-
www.pnas.org?cgi?doi?10.1073?pnas.0809677105Malhotra et al.
direct analysis of lipid peroxidation (malondialdehyde) in liver extracts
(Fig. 2 D and E). Mice that received either wtFVIII or BDD vectors
ROS was also associated with depletion of intracellular glutathione
that received 226/N6 vector or empty vector.
studied whether expression of poorly secreted FVIII causes apoptosis
cells in mice injected with wtFVIII or BDD vectors, but not in mice
and eosin-stained liver sections (Fig. S2 A and B).
Chop Deletion Protects from Oxidative Stress and Apoptosis upon
wtFVIII and BDD Expression. As the ER stress-induced apoptotic cell
CHOP has been implicated in ROS production (23–25), we asked
whether CHOP is required for apoptosis and ROS production in
with wtFVIII or BDD vectors, apoptosis was significantly reduced in
livers from Chop?/? mice that were injected with these vectors (Fig.
3A and Fig. S2D). The plasma levels of wtFVIII and BDD expressed
in Chop?/? mice were modestly but significantly increased compared
226/N6 were not significantly affected by Chop deletion. Whereas
expression of wtFVIII or BDD significantly increased lipid peroxida-
tion (hydroxyoctadecadienoic acid [HODE]) and protein oxidation
(i.e., carbonyls)—sensitive and quantitative markers of ROS produc-
increased in Chop?/? mice (Fig. 3 C and D). In addition, in contrast
conclude that both oxidative stress and the apoptotic response to
wtFVIII or BDD expression requires CHOP.
To provide insight into the mechanism by which Chop deletion
expression of wtFVIII or BDD, but not 226/N6, induced expression of
UPR adaptive functions including BiP, p58IPK, the ER degradation-
enhancing mannosidase-like protein EDEM1 (Edem1), spliced Xbp1
downstream of CHOP, Gadd34, and Trb3, were also induced upon
wtFVIII or BDD expression. In contrast, the induction of these genes
was significantly attenuated in Chop-null mice that were injected with
the wtFVIII or BDD vectors (Fig. 3E). In addition, whereas the
expression of ER oxidases 1? and 1? (Ero1?/? promote oxidation in
vectors, the expression of these genes was not induced in Chop?/?
mice. Western blot analysis of liver extracts from Chop?/? mice
protein level (Fig. S4). The gene expression analysis also indicated the
expression levels of some genes encoding an antioxidant response
vectors (Fig. 3E). These results demonstrate that expression of either
wtFVIII or BDD, but not 226/N6, induce the UPR, apoptosis, and
oxidative stress in a manner that requires CHOP.
Antioxidant Treatment Attenuates UPR Activation and Apoptosis
stress for the UPR and apoptotic response, we analyzed the effect of
antioxidant treatment. WT mice were fed chow supplemented with
BHA for 4 days before DNA delivery. BHA feeding dramatically
reduced apoptosis (Fig. 4A), reduced glutathione depletion (Fig. 4B),
upon expression of wtFVIII or BDD. BHA feeding had no effect on
these parameters in mice that received the 226/N6 vector or empty
vector. mRNA expression analysis demonstrated that expression of
wtFVIII or BDD induced the UPR, as previously observed in
Chop?/? mice (Fig. 3E vs. Fig. 4D). In contrast, BHA feeding
attenuated UPR activation of genes encoding adaptive, as well as
apoptotic, functions, upon delivery of wtFVIII or BDD vectors (Fig.
UPR activation at the protein level (Fig. S4). Interestingly, BHA
feeding also increased expression of the antioxidative stress response
suggesting that antioxidant treatment and Chop deletion may act
through a common mechanism to improve hepatocyte function.
Antioxidant Treatment Improves wtFVIII and BDD Secretion In Vivo.
reduced intracellular accumulation of wtFVIII and BDD in the liver,
and this correlated with increased secretion into the plasma, by eight
in vivo expression in liver. DNA expression vectors were delivered by tail-vein
injection into WT C57BL/6 mice. After 24 h, blood and liver tissues were
isolated for analysis. (A) Liver tissue sections were analyzed for immunolocal-
ization of FVIII antigen. (B) FVIII antigen in plasma samples and liver extracts
was measured by ELISA. FVIII activity in plasma samples was measured using
the COAMATIC assay kit. For activity measurements, the background of
murine FVIII activity was subtracted (0.35 U/ml). (C) Western blot analysis of
liver tissue for detection of BiP, phospho-eIF2?, and CHOP. Densitometry
in mice injected with wtFVIII and BDD compared with 226/N6. Spliced Xbp1
mRNA in liver tissue was measured by real-time RT-PCR. (D) Fresh frozen liver
for 30 min at 37 °C. Sections were analyzed by fluorescence microscopy. (E)
Malondialdehyde was measured in liver homogenates. (F) Liver lysates were
analyzed for GSH and GSH and oxidized glutathione content. Data represent
the mean and SD from three different animals in B, E, and F.
Malhotra et al.
November 25, 2008 ?
vol. 105 ?
no. 47 ?
Fviii?/? mice (Fig. 5A). The increase in FVIII antigen was propor-
the secreted FVIII was folded properly. In contrast, BHA feeding did
not have a significant effect on secretion of the well secreted 226/N6
molecule. In addition, injection of mice with the superoxide dismutase
mimetic Mn (iii)tetrakis(4-benzoic acid) porphyrin (MnTBAP) also
improved BDD secretion and attenuated both UPR induction and
apoptosis (Fig. S5 A-C). We also extended these observations to
delivery (26). Upon adenovirus delivery, BHA feeding improved the
secretion of BDD (Fig. S6).
molecule having a missense mutation Arg593Cys (R593C) that causes
protein misfolding and retention within the ER (27). This common
mutation has been identified in patients with mild hemophilia A
characterized by reductions in both FVIII antigen and activity. BHA
feeding attenuated apoptosis and UPR gene induction observed upon
expression of R593C-BDD (Fig. 5 B and D). BHA feeding preferen-
tially increased the secretion of the folding-defective R593C-BDD
mutant compared with BDD (Fig. 5C). The findings show that reduc-
tion in ROS can increase secretion of proteins prone to misfolding in
an animal model in vivo.
Our findings show that accumulation of unfolded protein in the ER
lumen is sufficient to produce ROS and that both ROS and unfolded
findings suggest that unfolded protein in the ER lumen signals ROS
production as a second messenger to activate the UPR and induce
apoptosis (Fig. 5E). Although it is not presently known how protein
exist. First, misfolded proteins bind protein chaperones, such as BiP,
that consume ATP that may stimulate mitochondrial oxidative phos-
phorylation to produce ROS as a byproduct. Second, ROS may be
produced as a consequence of disulfide bond formation in the ER
during the transfer of electrons from thiol groups in folding substrates
with normal chow or chow supplemented with BHA for 4 days and then DNA expression vectors were injected into the tail vein. After 24 h, plasma and liver samples were
www.pnas.org?cgi?doi?10.1073?pnas.0809677105Malhotra et al.
through PDI and ERO1 to molecular oxygen to produce hydrogen
number of disulfide bonds, the increased ROS production in response
likely a direct consequence of de novo disulfide bond formation.
Alternatively, GSH may be consumed during reduction of unstable
protein misfolding in the ER lumen can cause Ca2?leak from the ER
chain. Although further studies are required to elucidate how protein
misfolding in the ER lumen produces ROS, our studies demonstrate
that, when unresolved, leads to cell death.
protein have suggested there is an association between abnormal
protein degradation, are susceptible to ROS production (2, 33). Our
findings extend these observations by showing that protein misfolding
essential component in the events leading to protein misfolding in the
ER and ER stress-induced apoptosis. ROS could exacerbate protein
or modifying chaperone and/or ERAD functions, thereby amplifying
for function and survival of cells that have a high protein-folding load
and/or are susceptible to oxidative stress, such as B lymphocytes or
pancreatic ?-cells, or in cells that are exposed to a variety of environ-
mental insults (34, 35) (Fig. 5E).
The ER stress-induced apoptotic cell death pathway is, at least in
protected from ER stress-induced apoptosis (21, 22, 25, 36). Although
the precise mechanism by which CHOP mediates apoptosis is un-
including Gadd34, Ero1, Bim, and Trb3. Of note, Gadd34 encodes a
subunit of protein phosphatase 1 that directs dephosphorylation of
eIF2? to promote protein synthesis and oxidation of the ER under
conditions of ER stress (37, 38). In addition, increased expression of
ERO1 causes hyperoxidation of the ER in yeast (28). We have shown
and prevents apoptosis in hepatocytes during conditions of unfolded
protein accumulation in the ER lumen. Intriguingly, antioxidant treat-
ment also reduced UPR induction (including CHOP), reduced oxida-
tive stress, and reduced apoptosis. Moreover, the ER stress-induced
alterations in gene expression observed upon Chop deletion mirrored
the changes in gene expression caused by antioxidant treatment. The
findings suggest that antioxidants and Chop deletion may improve ER
function through a common mechanism.
disease, Alzheimer disease, and Parkinson disease, as well as diabetes,
gene delivery. (A) FVIII antigen in plasma and liver samples was measured at 24 h after DNA injection of the indicated vectors. (B–D) DNA vectors encoding FVIII-BDD or
Malhotra et al.
November 25, 2008 ?
vol. 105 ?
no. 47 ?
evidencethatsupportstheideathatantioxidantsanddietmodification Download full-text
can alleviate oxidative stress and prove beneficial in these diverse
of this idea, chemical chaperones 4-phenyl butyric acid and taurour-
sodeoxycholic acid, which are thought to improve protein folding,
Traditional therapy for hemophilia A involves protein replacement
with plasma-derived and, more recently, recombinant-derived FVIII.
However, this costly approach is hampered by development of anti-
FVIII inhibitory antibodies, limited supply, potential for pathogen
transmission, and poor access to the venous circulation. FVIII gene
transfer offers one potential solution to these problems. As the B
strategies use B domain-deleted FVIII, similar to the BDD we have
described here. Unfortunately, to date, clinical studies using retroviral-
mediated and adenoviral-mediated delivery of FVIII have not pro-
with adenovirus suggests hemophilia gene therapy may be limited by
adenovirus (50). The ROS produced as a consequence of FVIII
FVIII mutant R593C that is known to cause hemophilia A (27).
Therefore, antioxidants may provide a useful adjuvant to improve
FVIII production in patients who receive gene therapy or who have
mutations that disrupt FVIII folding. These findings should also en-
courage the evaluation of antioxidant treatment to improve folding of
different substrates and in different diseases associated with protein
Mice. Male C57BL/6 mice were purchased from Jackson Laboratory. Control and
Fviii?/? (exon 16 deletion) mice in a C57BL/6 background at 6 to 8 weeks were
housed under pathogen-free conditions at the University of Michigan Laboratory
David Ron (New York, NY) and bred into a C57BL/6 background. The University
Hydrodynamic Tail Vein Injections. The expression vectors for wtFVIII, BDD, and
226/N6 were previously described (20). Vectors for FVIII-BDD and R593C-BDD were
kindly provided by J. Voorberg (Amsterdam, The Netherlands) (27). Plasmid DNA
ACKNOWLEDGMENTS. We thank Dr. D.T. Rutkowski and D. Ginsburg for critical
reading of this paper. This work was supported by National Institutes of Health
Grants DK042394, HL052173, and HL057346 and Human Frontier Science Program
1. Malhotra JD, Kaufman RJ (2007) Endoplasmic reticulum stress and oxidative stress: a vicious
3. Tu BP, Weissman JS (2004) Oxidative protein folding in eukaryotes: mechanisms and conse-
4. Ron D, Walter P (2007) Signal integration in the endoplasmic reticulum unfolded protein
5. Schroder M, Kaufman RJ (2005) The mammalian unfolded protein response. Annu Rev
6. Do H, Healey JF, Waller EK, Lollar P (1999) Expression of factor VIII by murine liver sinusoidal
7. Hollestelle MJ, et al. (2001) Tissue distribution of factor VIII gene expression in vivo-a closer
chain-binding protein association with the secretion of glycoproteins. J Cell Biol 105:2665–
9. Swaroop M, Moussalli M, Pipe SW, Kaufman RJ (1997) Mutagenesis of a potential immuno-
globulin-binding protein-binding site enhances secretion of coagulation factor VIII. J Biol
10. Tagliavacca L, Wang Q, Kaufman RJ (2000) ATP-dependent dissociation of non-disulfide-
12. Zhang B, et al. (2003) Bleeding due to disruption of a cargo-specific ER-to-Golgi transport
and interact with coagulation factor VIII in the early secretory pathway. J Biol Chem
14. Dorner AJ, Wasley LC, Kaufman RJ (1989) Increased synthesis of secreted proteins induces
15. Rutkowski DT, et al. (2006) Adaptation to ER stress is mediated by differential stabilities of
16. Ryan TC, et al. (1990) Measurement of superoxide release in the phagovacuoles of immune
17. Vercammen D, et al. (1998) Inhibition of caspases increases the sensitivity of L929 cells to
18. Kamata H, et al. (2005) Reactive oxygen species promote TNFalpha-induced death and sus-
23. McCullough KD, et al. (2001) Gadd153 sensitizes cells to endoplasmic reticulum stress by
25. Song B, et al. (2008) Genetic deletion of C/EBP homologous protein CHOP reduces oxidative
28. Sevier CS, et al. (2007) Modulation of cellular disulfide-bond formation and the ER redox
29. Cuozzo JW, Kaiser CA (1999) Competition between glutathione and protein thiols for disul-
transition, mitochondrial outer membrane permeabilization and apoptosis. Oncogene
33. Haynes CM, Titus EA, Cooper AA (2004) Degradation of misfolded proteins prevents ER-
35. Masciarelli S, Sitia R (2008) Building and operating an antibody factory: redox control
during B to plasma cell terminal differentiation. Biochim Biophys Acta 1783:578–588.
36. Silva RM, et al. (2005) CHOP/GADD153 is a mediator of apoptotic death in substantia nigra
37. Novoa I, Zeng H, Harding HP, Ron D (2001) Feedback inhibition of the unfolded protein
GADD34 targets protein phosphatase 1 alpha to the endoplasmic reticulum and promotes
39. Merad-Boudia M, et al. (1998) Mitochondrial impairment as an early event in the process of
40. Nakamura T, Lipton SA (2007) Molecular mechanisms of nitrosative stress-mediated protein
42. NathanC(2008)Epidemicinflammation:ponderingobesity. MolMed14:485–492.
43. WilliamsKJ,TabasI(2002)Atherosclerosisandinflammation. Science297:521–522.
46. Robertson RP, Harmon JS (2006) Diabetes, glucose toxicity, and oxidative stress: a case of
47. Houstis N, Rosen ED, Lander ES (2006) Reactive oxygen species have a causal role in multiple
50. High KA (2007) Update on progress and hurdles in novel genetic therapies for hemophilia.
www.pnas.org?cgi?doi?10.1073?pnas.0809677105Malhotra et al.