Double-Stranded RNA-Dependent Protein
Kinase Links Pathogen Sensing
with Stress and Metabolic Homeostasis
Takahisa Nakamura,1MasatoFuruhashi,1Ping Li,1Haiming Cao,1Gurol Tuncman,1Nahum Sonenberg,2CemZ. Gorgun,1
and Go ¨khan S. Hotamisligil1,3,*
1Department of Genetics & Complex Diseases, Harvard School of Public Health, Boston, MA 02115, USA
2Department of Biochemistry, McGill University, Montreal, Quebec H3G1Y6, Canada
3Harvard-MIT Broad Institute, Cambridge, MA 02142, USA
As chronic inflammation is a hallmark of obesity,
pathways that integrate nutrient- and pathogen
sensing pathways are of great interest in under-
standing the mechanisms of insulin resistance, type
2 diabetes, and other chronic metabolic pathologies.
Here, we provide evidence that double-stranded
RNA-dependent protein kinase (PKR) can respond
to nutrient signals as well as endoplasmic reticulum
(ER) stress and coordinate the activity of other
critical inflammatory kinases such as the c-Jun
N-terminal kinase (JNK) to regulate insulin action
and metabolism. PKR also directly targets and
modifies insulin receptor substrate and hence inte-
grates nutrients and insulin action with a defined
pathogen response system. Dietary and genetic
obesity features markedactivation ofPKR in adipose
and liver tissues and absence of PKR alleviates
metabolic deterioration due to nutrient or energy
excess in mice. These findings demonstrate PKR
as a critical component of an inflammatory complex
that respondstonutrients andorganelle dysfunction.
Metabolic diseases appear as clusters including obesity, insulin
resistance, type 2 diabetes, and cardiovascular disease and
constitute a major global health problem with limited treatment
options. In the past decade, it has been realized that the emer-
gence of this cluster has strong inflammatory underpinnings
(Hotamisligil, 2006). During the course of obesity, a broad array
of inflammatory and stress responses are evoked in metabolic
tissues, leading to chronic, low-grade local inflammation that
plays a central role in the inhibition of insulin receptor signaling
and disruption of systemic metabolic homeostasis. This atypical
state, which we refer to as metaflammation (Hotamisligil, 2006),
involves immune and nonimmune cells and the engagement of
immune response pathways with nutrients and metabolites.
However, the mechanistic basis of these extensive functional
links and molecules that coordinate this network of responses
remains to be understood.
If the nutrient and pathogen response systems were truly inte-
grated, the involvement of pathogen sensors in metabolic regu-
lation, especially during exposure to excess nutrients, would be
anticipated. Such anticipation has stimulated the pursuit of
pattern recognition receptors (PRRs) such as the Toll-like recep-
tors (TLRs) for a role in metabolism (Shi et al., 2006). Other than
PRRs, there are only a few molecules that can assume such
a role to carry the potential ability for direct recognition of path-
ogens and possess catalytic activity to directly couple to meta-
bolic pathways. One such molecule is the double-stranded
RNA-dependent protein kinase (PKR), which has been originally
identified as a pathogen sensor and a proposed regulator of the
otes (Samuel, 1993). Virus-derived double-stranded RNA mole-
cules are recognized and bound by PKR through the N-terminal
double-stranded RNA-binding motifs, resulting in autophos-
phorylation through the activation of the intramolecular kinase
domain (Garcia et al., 2006; Williams, 2001). Interestingly, in
the context of infections, PKR can regulate or act in conjunction
with major inflammatory signaling pathways that are implicated
in metabolic homeostasis, including the c-Jun N-teminal kinase
(JNK) and IkB kinase (IKK) (Bonnet et al., 2000; Goh et al., 2000;
Takada etal.,2007). Inmetabolic disease, it isunclear how these
and other inflammatory signaling molecules are coordinately
regulated to disrupt metabolism. However, it is feasible to
consider a model where these multiple signaling pathways act
in concert by forming a response node or acting in complexes
that yield to metabolic surplus.
Interestingly, among the very few substrates identified for the
kinase activity of PKR, the major one is the eukaryotic initiation
factor 2a (eIF2a), which regulates general protein synthesis (Hol-
cik and Sonenberg, 2005; Ron and Walter, 2007). It has been
postulated that PKR-mediated phosphorylation of eIF2a is
a strategy to inhibit viral protein synthesis in host cells (Holcik
and Sonenberg, 2005). However, in other contexts, inhibition of
general translation through the same eIF2a phosphorylation
338 Cell 140, 338–348, February 5, 2010 ª2010 Elsevier Inc.
represents a major arm controlling endoplasmic reticulum (ER)
homeostasis (Ron and Walter, 2007). ER stress also plays an
important role in the development of insulin resistance and
diabetes, at least in part, by triggering JNK activity (Gregor
and Hotamisligil, 2007; Ozcan et al., 2004). In both mouse
and human, there is a consistent and marked increase in
phosphorylation of eIF2a and JNK activity in obese metabolic
tissues (Gregor et al., 2009). Hence, coordination of protein
synthesis through eIF2a and stress signaling through JNK likely
represent a relevant mechanism in the integration of inflamma-
tory signals with metabolic outcomes, especially in the context
of nutrient and energy surplus. PKR features properties to coor-
dinate pathogen responses, endoplasmic reticulum function,
inflammatory signaling, and translational regulation. Therefore,
we postulate that PKR may represent a core component of
a putative ‘‘metabolic inflammasome,’’ in other words, a mecha-
nism integrating pathogen response and metabolic pathways
that plays a critical role in metabolic homeostasis by controlling
the action of major players such as JNK, IKK, and/or other
PKR Activity in Obesity and Metabolic Stress
If PKRwere involved
responses, it would be anticipated that its activity is altered in
is altered in obesity and metabolic stress, we first examined
from leptin deficiency (ob/ob, also known as Lepob/ob). There
was a striking increase in PKR activity, which was assessed by
autophosphorylation level of PKR using32P-gATP, in both white
adipose tissue (WAT) and liver of ob/ob mice compared to lean
PKR activity (p-PKR)
IP: PKR, IB: PKR
IP: PKR, IB: PKR
Figure 1. Regulation of PKR Activity in Obesity and
(A) A genetic mouse model of obesity (ob/ob) was used to
fied PKR and32P-gATP in white adipose tissue (WAT) and
liver compared with age- and sex- matched lean controls.
Fatty acid synthase (FAS) and b-tubulin proteins are
shown as controls.
(B) PKR activity was examined in white adipose tissue
(WAT) and liver of the male WT mice kept either on regular
diet (RD) or high-fat diet (HFD) for 20 weeks.
(C) PKR activity in liver of 10-week-old male WT mice,
which were infused with lipid or saline for 5 hr. Total S6
protein is shown as control.
(D) PKR activity in primary MEFs. Cells were cultured inthe
absence or presence of 0.5 mM palmitic acid for 2 hr.
See also Figure S1.
icant upregulation of PKR activity in both WAT
and liver tissues of mice fed with a high-fat diet
(Figure 1B). In both models, we also observed
an increase in PKR protein and mRNA levels,
consistent with the reported autoregulation of
PKR expression upon stimulation of its activity
(Figures 1A and 1B and Figures S1A and S1B available online)
(Gusella et al., 1995; Samuel, 1993). However, there was only
a minor regulation of skeletal muscle tissue PKR activity in the
ob/ob model and no regulation was evident in dietary obesity
(Figures S1C and S1D). In adipose tissue, the principle source
of increased PKR expression in dietary obesity was found in
mature adipocyte fraction instead of stromal-vascular (SV) frac-
tion (Figure S1A). Interestingly, in addition to PKR itself, mRNA
levels of some other interferon target genes were also increased
in adipocytes of obese WAT and liver (Figures S1A and S1B). As
these mice were maintained in a specific pathogen-free environ-
ment, activation of PKR likely involves metabolic signals associ-
ated with nutrient and energy excess.
To determine whether PKR activity is related to metabolic
stress and triggered by a nutrient, we next used an in vivo lipid
infusion system, which acutely increases circulating free fatty
acids (FFA) and causes insulin resistance (Kim et al., 2004;
Shi et al., 2006). In this setting, there was also a significant
increase in PKR activity in lean mice following lipid exposure
(Figure 1C). These results demonstrate that PKR is activated
by nutrient excess and metabolic stress in vivo. Given that
PKR activity is induced by lipid infusion, we next explored
whether FFA itself can also activate PKR by treating primary-
isolated mouse embryonic fibroblast cells (MEFs) with palmitic
acid. As shown in Figure 1D, palmitic acid exposure resulted in
PKR activation. Since TLR4 has been implicated in fatty acid
response (Shi et al., 2006), we asked whether PKR activation
by lipids involved this PRR. However, in TLR4-deficient primary
MEFs, palmitic acid-induced PKR activation was also detected
at levels similar to those in wild-type (WT) controls (Figure S1E).
These experiments demonstrated that PKR is activated by
lipids in vitro and in vivo, and this activation is not dependent
Cell 140, 338–348, February 5, 2010 ª2010 Elsevier Inc. 339
PKR Controls JNK Activity in Response to Metabolic
JNK activity is modulated by PKR under several stress condi-
tions (Goh et al., 2000). As JNK activation is also related to fatty
acid exposureandERstress(Gaoetal.,2004;Urano etal.,2000)
and critical for insulin sensitivity (Hirosumi et al., 2002), we next
examined the impact of PKR on fatty acid- and ER stress-medi-
ated JNKactivation. To assess effects of PKR deficiency in cells,
ing the PKR kinase domain (Abraham et al., 1999). Treatment of
lated MEFs with palmitic acid demonstrated that this fatty acid
failed to activate JNK in the absence of PKR (Figure 2A). In
a similar fashion, treatment of Pkr+/+and Pkr?/?MEFs with thap-
sigargin, an agent that induced ER stress, resulted in prominent
induction of JNKphosphorylation in Pkr+/+relative to Pkr?/?cells
(Figure 2B). ER stress-induced PKR activation was decreased in
the presence of translation inhibitor cycloheximide. Therefore,
new protein synthesis, at least in part, appears to be required
these results demonstrate that PKR is a required component for
JNK activation in response to lipid exposure and ER stress.
One mechanism by which JNK negatively regulates insulin
signaling is through induction of IRS1 serine phophorylation in
response to fatty acid exposure and ER stress (Gao et al.,
2004; Ozcan et al., 2004). If PKR controls JNK activation, it is
conceivable that PKR may, indirectly or directly, target IRS1
function, which is critical in insulin action. To address whether
fatty acid- or ER stress-induced IRS1 phosphorylation is depen-
dent on PKR, we examined the effect of palmitic acid or thapsi-
gargin treatment on IRS1 serine phosphorylation in Pkr+/+and
Pkr?/?MEFs. Both of these treatments resulted in induction of
IRS1 serine 307 phosphorylation in Pkr+/+MEFs but not in
Pkr?/?cells (Figures 2C and 2D). As these data suggest that
activated PKR modulates IRS1 serine phosphorylation, we
next assessed the effect of a direct PKR activator, a virus-
derived double-stranded RNA mimetic, polyinosinic-polycyti-
dylic acid (polyIdC) (Garcia et al., 2006; Williams, 2001), on
IP: IRS1, IB: p85
IP: IRS1, IB: IRS1
IP: IRS1, IB: pTyr
Relative pTyr/IRS1 levels
IP: Flag (PKR)
IB: Flag (PKR)
IP: Flag (PKR)
IB: Flag (PKR)
Figure 2. PKR Regulates JNK Activity Resulting in Inhibition of Insulin Signaling
(A) Primary Pkr+/+and Pkr?/?MEFs weretreatedwith 0.5 mM palmitic acid for 2hr.JNK activity was assessed by a kinase assayusing recombinantc-Jun protein
ined with anti-phospho-JNK (Thr183/Tyr185) antibody.
(C and D) Induction of IRS1 phosphorylation after 0.5 mM palmitic acid (C) or 300 nM thapsigargin (D) treatment for 2 hr in Pkr+/+and Pkr?/?MEFs. Phosphor-
ylation level of IRS1 on Ser307 was examined with anti-phospho-IRS1 (serine 307) antibody.
(E)InductionofIRS1 phosphorylation inretrovirally PKR-reconstituted Pkr?/?MEFs. Thecellswere serum-starved for14hrfollowed by western blotanalysiswith
anti-phospho-IRS1 (serine 307) antibody.
(F) The PKR-reconstituted Pkr?/?MEFs were stimulated with 10 nM insulin for 3 min. The cell lysates were immunoprecipitated with anti-IRS1 antibody followed
shown as the mean ± standard error of the mean (SEM). *p < 0.05.
(G and H) Induction of palmitic acid- and thapsigargin-induced PKR activity requires intact RNA-binding domain of PKR. Pkr?/?MEFs were reconstituted with
vector,flag-tagged wild-type(WT),RNA-binding domain mutant (K64E),orkinase-dead mutant (K296R)of PKRbyretrovirus-mediated gene transfer. Thesecells
(H). The cell lysates were immunoprecipitated with anti-Flag antibody followed by PKR kinase assay and western blot analysis with anti-Flag antibody.
See also Figure S2.
340 Cell 140, 338–348, February 5, 2010 ª2010 Elsevier Inc.
IRS1 phosphorylation in Pkr+/+and Pkr?/?MEFs. This treatment
also induced IRS1 serine phosphorylation with a marked
increase in PKR expression in Pkr+/+MEFs. However, polyId
C-induced IRS1 phosphorylation was not observed in Pkr?/?
MEFs (Figure S2B). We also reconstituted PKR expression in
the Pkr?/?MEFs as increased PKR expression itself induces
PKR activity thus allowing comparisons in the same cellular
background without stimulation of other signaling pathways
(Figure S2C). As shown in Figure 2E, serine phosphorylation of
IRS1 was induced by reconstitution of PKR in Pkr-deficient cells.
In these cells, the level of insulin-induced tyrosine phosphoryla-
tion of IRS1 and insulin-induced interaction between IRS1 and
p85, a regulatory subunit of phosphoinositide 3-kinase (PI3K),
were both significantly diminished in PKR-reconstituted cells
compared to the Pkr?/?controls expressing the empty vector
(Figure 2F). Hence, PKR can induce IRS serine phosphorylation
and block insulin action.
We next asked whether RNA-binding ability of PKR is required
for palmitic acid- or ER stress-induced PKR activation. RNA-
binding capacity of PKR is abolished by introducing a point
mutation to its lysine 64 residue in the RNA-binding motif
(McCormack et al., 1994; Wu and Kaufman, 1996). We reconsti-
tuted Pkr?/?MEFs with WT or RNA-binding defective (K64E)
PKR by retrovirus-mediated gene transfer and examined PKR
kinase activity after treatment with palmitic acid or thapsigargin.
Wild-type PKR is activated by both palmitic acid and thapsigar-
gin (Figures 2G and 2H). However, PKR with defective RNA
binding (K64E) was not activated in palmitic acid- or thapsigar-
gin-treated cells (Figures 2G and 2H). A mutation in the kinase
domain of PKR abolished all activity and was used as a control
(Figures 2G and 2H). These data demonstrate that PKR is
a required component for JNK activation and IRS1 inhibition
when induced by a nutrient, a pathogen component, or ER
stress, and the RNA-binding domain of PKR is indispensable in
IRS1, a Direct Substrate of PKR
We next treated Pkr+/+and Pkr?/?cells with TNF-a, which is well
known to both induce PKR activity and block insulin action
through IRS1 phosphorylation in WT cells (Hotamisligil et al.,
1996). In these experiments, we noticed that PKR was detect-
able by western blot analysis following immunoprecipitation
with an anti-IRS1 antibody, demonstrating a potential interaction
between these proteins (Figure 3A). To verify the interaction
iment, where we immunoprecipitated PKR from protein extracts
of TNF-a-treated Pkr+/+MEFs and attempted to detect IRS1 by
western blot analysis. PKR-deficient cells were used in the
same setting as controls. These experiments also demonstrated
the interaction between these two proteins (Figure 3B). To
assess the specificity of PKR-IRS1 interaction, we also exam-
ined the relation between IRS1 and the PKR-like endoplasmic
reticulum kinase (PERK), which is the closest homolog of PKR
in mammals, and did not detect any interaction between these
proteins with or without induction of ER stress (Figure S3).
We next performed an in vitro pull-down assay using recombi-
nant IRS1 and PKR proteins. These experiments demonstrated
that PKR directly interacts with IRS1 protein (Figure 3C). The
robust interaction between PKR and IRS1 raised the possibility
that PKR phosphorylates IRS1 directly. To address this possi-
bility, we performed kinase assays using recombinant IRS1
and active PKR in vitro. As shown in Figure 3D, PKR directly
phosphorylates IRS1 including the serine 307 residue. We next
performed in vitro kinase assays after immunopurification of
PKR from Pkr+/+MEFs and used Pkr?/?cell extracts as a nega-
tive control. We found that IRS1 was phosphorylated by immu-
nopurified PKR from TNF-a- or thapsigargin-treated Pkr+/+
MEFs (Figures 3E and 3F). More importantly, the increased
serine phosphorylation of IRS1, which is detected by a phos-
pho-specific antibody, was induced only by activated PKR
gin-treated Pkr?/?cells in control experiments (Figures 3E and
3F). We have previously shown that JNK1 plays a critical role
in stress-induced IRS1 phosphorylation and insulin resistance
(Hirosumi et al., 2002). To address whether JNK1 is required
for PKR-mediated IRS1 phosphorylation, we next assessed
effects of exogenous expression of PKR on IRS1 phosphoryla-
tion in primary Jnk1+/+and Jnk1?/?MEFs. By introducing PKR
through adenovirus-mediated gene transfer, serine phosphory-
lation of IRS1 is dramatically increased in Jnk1+/+MEFs
(Figure 3G). In Jnk1?/?MEFs, PKR still retained the ability to
induce serine phosphorylation of IRS1 (Figure 3G), although
this was significantly reduced in magnitude. These data suggest
that PKR assumes a role in regulation of IRS1 phosphorylation in
both a JNK1-dependent and -independent manner in cultured
cells and illustrate the importance of functional interaction
between these kinases to interfere with insulin action.
Metabolic Regulation and Insulin Action in Pkr?/?Mice
Given thatPKRactivityisstronglyregulated in obesityandlinked
to signaling pathways interfering with metabolic homeostasis,
we then used several different in vivo models to test the func-
tional significance of PKR in the pathogenesis of obesity, insulin
resistance, and type 2 diabetes. First, mice lacking PKR (Pkr?/?)
and WT control (Pkr+/+) were placed on a high-fat diet (HFD)
along with a control group of each genotype on regular diet
(RD). The PKR-deficient model used in our study was generated
by removing the kinase domain and hence retains no kinase
activity (Abraham et al., 1999). On HFD, Pkr+/+mice developed
obesity compared to mice kept on RD, while weight gain in
Pkr?/?mice was significantly lower, starting to become evident
after 10 weeks of HFD (Figure 4A). Dual energy X-ray absorption
(DEXA) analysis demonstrated reduced total body adipose mass
in Pkr?/?mice (Figure 4B). In addition, consistent with reduced
adiposity, serum leptin level in Pkr?/?mice was lower than that
in Pkr+/+mice (Figure 4C). Total serum adiponectin levels were
not significantly different between genotypes (Figure 4D). In
Pkr?/?mice, blood glucose levels were significantly lower than
those in Pkr+/+controls (Figure 4E). Examination of serum insulin
levels revealed thatthehyperinsulinemia observedin Pkr+/+mice
was not evident in the Pkr?/?animals, indicating that these
animals might exhibit enhanced insulin sensitivity (Figure 4F).
To investigate systemicinsulin sensitivity, weperformed glucose
after the start of HFD. Since body weights of Pkr+/+and Pkr?/?
Cell 140, 338–348, February 5, 2010 ª2010 Elsevier Inc. 341
mice did not exhibit significant differences until week 10–12,
experiments performed at an early stage allowed evaluation of
insulin sensitivity without changes in body weight. After 6 weeks
of HFD, where the body weights were similar between geno-
types, Pkr?/?mice already showed significantly lower glucose
levels after a glucose challenge compared to Pkr+/+mice (Fig-
ure 4G). This increased glucose tolerance continued to be
evident in Pkr?/?mice after 14 weeks of HFD (Figure S4A). Simi-
larly, during insulin tolerance tests (ITT), the hypoglycemic
response to insulin was also significantly enhanced in Pkr?/?
mice compared to Pkr+/+controls after 16 weeks of HFD (Fig-
ure S4B). Glucose disposal curves and the hypoglycemic
response to insulin in mice on RD were similar between geno-
types (Figures S4C and S4D). Although no significant changes
were observed in heat production, food intake, or total physical
activity between genotypes (Figures S4E–S4H), rates of oxygen
consumption and carbon dioxide production of Pkr?/?mice
were modestly but significantly higher than those of Pkr+/+
mice (Figures S4I and S4J), indicating that energy expenditure
may be a potential mechanism for body weight reduction.
Biochemical and Molecular Alterations in Pkr?/?Mice
insulin action by examining in vivo insulin receptor-signaling
capacity in WAT and liver tissues of mice on HFD. In intact
animals, insulin-stimulated AKT phosphorylation on serine 473
was significantly increased in WAT extracts of Pkr?/?mice
compared with those of Pkr+/+controls (Figure 5A). There was
also significantly increased insulin-stimulated AKT phosphoryla-
tion in liver tissue of Pkr?/?mice (Figure 5B). On the other hand,
++++ ++ +++
Relative pIRS1/IRS1 levels
PKR activity (p-PKR)
Figure 3. PKR Directly Regulates IRS1 Phosphorylation
(A) Induction of interaction between IRS1 and PKR after TNF-a treatment in Pkr+/+and Pkr?/?MEFs. IRS1 and PKR protein levels were examined either with
immunoprecipitation (IP) followed by immunoblotting (IB) or by direct immunobloting in cells treated with 5 ng/ml TNF-a treatment for 3 hr.
(B) Physical interaction between IRS1 and PKR in TNF-a-treated MEF cells. Cell lysates were prepared from Pkr+/+or Pkr?/?MEFs treated with 5 ng/ml TNF-a for
3 hr followed by immunoprecipitation with anti-PKR antibody and western blot analysis with anti-IRS1 antibody.
(C) Physical interaction between IRS1 and PKR in a pull-down assay in vitro using recombinant IRS1 and PKR proteins.
(D) Direct phosphorylation of IRS1 by PKR in kinase assay in vitro using recombinant IRS1 and PKR proteins. Phosphorylation level of IRS1 was assessed by
autoradiography or western blot analysis with anti-phospho-IRS1 (serine 307) antibody.
(E and F) In vitro PKR kinase assay. IRS1 phosphorylation by immunopurified PKR prepared from 5 ng/ml TNF-a (E) or 300 nM thapsigargin (F) treated MEFs and
analyzed by autoradiography or western blot analysis with anti-phospho-IRS1 (serine 307) antibody.
(G) Effects of exogenous expression of PKR on IRS1 serine phosphorylation in primary Jnk1+/+and Jnk1?/?MEFs. Flag-tagged human PKR was introduced to
primary Jnk1+/+and Jnk1?/?MEFs by adenovirus-mediated gene transfer. Phosphorylation level of IRS1 on serine 307 was examined with anti-phospho-IRS1
(serine 307) antibody. Both exogenous and endogenous PKR expression was detected by anti-PKR antibody. The graph on the right shows the quantification of
See also Figure S3.
342 Cell 140, 338–348, February 5, 2010 ª2010 Elsevier Inc.
we did not observe significant alterations in insulin signaling in
the skeletal muscle of Pkr?/?mice (Figure S5A). These results
demonstrate significant contribution of PKR to high-fat diet-
induced insulin resistance. As eIF2a is a substrate of PKR, we
determined the phosphorylation level of eIF2a on Ser52 using
a specific antibody. These experiments demonstrate markedly
decreased eIF2a phosphorylation in both adipose tissue and
liver extracts of Pkr?/?mice compared with those of Pkr+/+
controls (Figures 5C and 5D). Dietary obesity is also character-
ized by increased JNK activity (Hirosumi et al., 2002), which is
correlated with PKR activity (Figure 1). In liver and adipose
tissues, JNK1 activity was significantly reduced in Pkr?/?mice
compared with Pkr+/+controls on HFD (Figures 5C and 5D and
Figure S5B). These biochemical alterations were also reflected
in the expression of inflammatory mediators induced by obesity
in adipose tissue. In Pkr?/?mice, expression levels of several
inflammatory cytokines such as Tnfa, Il6, and Il1b, and an anti-
et al., 2008), were significantly reduced in comparison to levels
seen in Pkr+/+controls on HFD (Figure 5E). In addition, expres-
sion of a macrophage marker F4/80 and an ER stress marker
Grp78 was also significantly reduced in adipose tissue of Pkr?/?
mice (Figure 5E). These changes in gene expression between
genotypes were not observed in mice fed RD (Figure S5C).
Consistent with the HFD-induced changes in inflammatory
gene expression, examination of the WAT sections revealed
reduced inflammatory cells in Pkr?/?mice compared to Pkr+/+
controls (Figure 5F). There was also a reduction in hepatic fatty
infiltration (Figure 5G). Biochemical measurement of liver triglyc-
erides demonstrated the reduction in liver lipid content in Pkr?/?
mice (Figure 5H). Serum alanine aminotransferase (ALT) level
was not significantly different between genotypes (Figure 5I).
A Role of PKR in Lipid Infusion-Induced Acute Insulin
As shown in Figure 1, lipid exposure leads to PKR activation.
To further explore the impact of PKR on insulin sensitivity in
another setting and without the potential confounding effects
of body weight or adiposity, we studied the effects of PKR
activation on acute insulin resistance induced by lipids
Figure 4. Glucose Metabolism and Insulin Sensitivity in Pkr?/?Mice
mean ± SEM. *p < 0.05.
(B) Analysis of body fat by dual energy X-ray absorptiometry (DEXA). Data are shown as the mean ± SEM. *p < 0.05.
(C and D) Serum leptin (C) and adiponectin (D) levels after 6 hr daytime food withdrawal in Pkr+/+(n = 5) and Pkr?/?(n = 6) mice on HFD for 15 weeks. Data are
shown as the mean ± SEM.
(E and F) Blood glucose (E) and serum insulin (F) levels after 6 hr daytime food withdrawal in Pkr+/+(n = 6) and Pkr?/?(n = 6) mice on HFD for 8 weeks. Data are
shown as the mean ± SEM.
(G) Glucose tolerance tests were performed on Pkr+/+(n = 6) and Pkr?/?mice (n = 6) on RD and HFD for 6 weeks. Data are shown as the mean ± SEM. *p < 0.05.
See also Figure S4.
Cell 140, 338–348, February 5, 2010 ª2010 Elsevier Inc. 343
(Shi et al., 2006). In this system, we administered lipid intrave-
nously into the Pkr+/+and Pkr?/?mice on RD and performed
hyperinsulinemic-euglycemic clamp studies to examine the
whole-body insulin sensitivity and glucose metabolism. Glucose
mice required significantly higher levels of glucose infusion to
maintain blood glucose consistent with increased insulin sensi-
tivity (Figure 6A). In Pkr+/+mice, GIR required to maintain eugly-
cemia was reduced by 50.2% in comparison with saline-infused
controls (Figure 6B). In contrast, lipid infusion exerted a signifi-
cantly smaller effect on the GIR in Pkr?/?mice (Figure 6B).
Consistent with this result, the whole-body glucose disposal
rates (Rd) observed in Pkr?/?mice during the clamp were signif-
icantly higher than those in Pkr+/+controls (Figure 6C). A similar
trend was also observed in hepatic glucose production (HGP)
levels, although this did not reach statistical significance (Fig-
ure 6D). Examination of insulin-stimulated tissue glucose uptake
revealed significant increase in muscle tissue of Pkr?/?mice
compared to Pkr+/+controls (Figure 6E). Although WAT glucose
uptake was not significantly different between genotypes,
a similar trend was also observed (Figure 6F). Under these con-
ditions, lipid infusion did not result in differential activation of
proapoptotic pathways between genotypes, indicating that the
utor to the phenotype seen in the Pkr?/?mice (Figure S6). Taken
together these data clearly show that lipid-induced PKR
Figure 5. Biochemical and Molecular Alterations in Pkr?/?Tissues
(A and B) Phosphorylation level of Akt on serine 473 in WAT (A) and liver (B) of Pkr+/+and Pkr?/?mice on HFD for 20 weeks. The graphs on the right of each blot
show the quantification of the results. Data are shown as the mean ± SEM. *p < 0.05, **p < 0.01. AU: arbitrary unit.
(CandD)Phosphorylation levelofeIF2aonserine52andJNK1kinase activity,whichwasdetectedbyakinaseassayusingimmunopurified JNK1,32P-gATP,and
recombinant c-Jun protein as substrate in WAT (C) and liver (D) of Pkr+/+and Pkr?/?mice on HFD for 20 weeks. b-tubulin is shown as a control.
(E) Gene expression in WAT including proinflammatory cytokine levels in Pkr+/+and Pkr?/?mice on HFD for 20 weeks. Data are shown as the mean ± SEM.
*p < 0.05, **p < 0.01.
(F and G) Haematoxylin and eosin staining of WAT (F) and liver (G) sections of Pkr+/+and Pkr?/?mice, respectively. Scale bar, 200 mm.
(H) Triglyceride contents in liver of Pkr+/+(n = 6) and Pkr?/?(n = 6) mice on HFD for 20 weeks. Data are shown as the mean ± SEM.
(I) Serum alanine aminotransferase level after 6 hr daytime food withdrawal in Pkr+/+(n = 6) and Pkr?/?(n = 6) mice on HFD for 15 weeks. Data are shown as the
mean ± SEM.
See also Figure S5.
344 Cell 140, 338–348, February 5, 2010 ª2010 Elsevier Inc.
activation is an important negative regulator of insulin action and
that PKR directly regulates insulin sensitivity in multiple settings.
Although the integration of nutrient and pathogen response
systems has been put forward as an attractive model to explain
the inflammatory origin of metabolic disease, identification of
specific molecules and mechanisms directly coordinating these
extensive links has been a major challenge (Hotamisligil and
Erbay, 2008). In this study, we uncovered a link between an
established pathogen-sensing pathway mediated by PKR and
metabolic homeostasis through regulation of JNK activity and
insulin action (Figure 7). This raises an intriguing possibility that
PKR may act as a central integrator in the inflammatory compo-
nent of metabolic control by linking nutrient- and pathogen-
One of the remaining important questions regarding the rela-
tionships between inflammatory pathways and insulin action is
how the action of many molecules with similar biological func-
tions are coordinated. Previous studies have shown direct inter-
actions of PKR with the IKKb-NF-kB pathway (Bonnet et al.,
2000). In this study, we have demonstrated PKR’s direct inter-
action and modulation of IRS, a critical molecule in insulin
action, and the major regulatory role over JNK activation. Taken
together, these findings make PKR a very attractive molecule at
the core of a complex that features major inflammatory and
stress roads that intersect with metabolism. As activation of
several stress signaling pathways produces similar outcomes
in metabolic regulation, the possibility that these molecules
work as part of a complex or a signaling node assembled and
coordinated bya molecule ofinnate immunitywith direct nutrient
recognition potential is of critical importance. Such a role is high-
lighted for PKR in the integration of nutrient and pathogen
sensing in relation to metabolic homeostasis (Figure 7). In this
scenario, it is possible that the activity of one component is
highly regulated by the other, and thus amplified feedback
mechanisms may be in place. For example, the interaction and
function of JNK or IKK may also influence PKR and PKR activa-
tion, in turn contributing to organelle stress. Hence, we suggest
tive metabolic inflammasome (or metaflammasome). These
interesting possibilities merit further investigation.
It is noteworthy that PKR responds to pathogens, nutrients,
and organelle stress. Then, PKR plays a role in the mounting of
adaptive and survival responses including suppression of
general protein translation through phosphorylation of eIF2a
and inhibition of anabolic effects of insulin action. The marked
increase in PKR activity observed in multiple models of obesity
featuring energy and nutrient excess may represent an adaptive
attempt to interfere with synthetic pathways that would further
accumulate energy. This is consistent with the late-onset effects
Figure 6. PKR Mediates Lipid-Induced Insulin Resistance
(A–F) Hyperinsulinemic–euglycemic clamp studies performed in Pkr+/+(n = 5) and Pkr?/?mice (n = 5) infused with lipid for 5 hr. (A) Glucose infusion rates (GIR)
throughout the clamp procedure. (B) Average GIR. (C) Whole-body glucose disposal rates (Rd). (D) Hepatic glucose production (HGP) during the clamp. Tissue
glucose uptake in gastrocnemius muscle (E) and epididymal fat (F) tissues of Pkr+/+and Pkr?/?mice. Data are shown as the mean ± SEM. *p < 0.05.
See also Figure S6.
Cell 140, 338–348, February 5, 2010 ª2010 Elsevier Inc. 345
of PKR activity on adiposity. The function of PKR activation may
be limited to coordinated launching of an inflammatory response
together with kinases such as JNK and IKK, as supported by the
evidence provided in this study, which results in alterations in
insulin action and possibly other metabolic outcomes. The
most attractive implication of our observation is that PKR can
directly sense nutrients or other metabolic products including
endogenous RNA species produced during metabolic or ER
stress. This possibility is strongly supported by the observation
that PKR’s defined RNA-binding domain is required for its acti-
vation by lipids or ER stress (Figures 2G and 2H).
ogens directly and to control the inflammatory output and meta-
bolic consequences illustrates the diversity of PKR-mediated
responses to both extrinsic and intrinsic stress signals. This
functional diversity also raises the possibility that the role of
PKR in metabolic regulation not only may be critical in dietary
exposures and insulin resistance but also may include other
metabolic abnormalitiesthatemerge duringthecourseof certain
infections. For example, both type 1 and type 2 diabetes have
been associated with viral infections, but mechanisms linking
viral infections to diabetes are not well understood (Antonelli
et al., 2009; von Herrath, 2009). It is known that one-third of
patients with chronic hepatitis C, which can trigger PKR activity,
develop type 2 diabetes (Bahtiyar et al., 2004; Delhem et al.,
2001; Tokumoto et al., 2007). Molecular mechanisms by which
b cells are destroyed during the development of type 1 diabetes
have also remained elusive, although recent studies suggest
a potential involvement of metabolic status, ER stress, and
inflammatory responses (Eizirik et al., 2008; Ron and Walter,
2007). As PKR modifies insulin signaling and contributes to
apoptosis induced by double-stranded RNA (Scarim et al.,
2001), it may offer critical insights into the pathogenesis of
both forms of diabetes and metabolic deregulation triggered
by viral infections.
Taken together with the previous reports, our results showing
that PKR can act in conjunction with major inflammatory kinases
and directly interact with a critical insulin signaling component
lead us to suggest PKR as a core component of a putative
‘‘metabolic inflammasome’’ that consists of major elements in
inflammatory signaling and insulin action (Figure 7). This PKR-
coordinated sensing and signaling complex may represent
a central mechanism for the integration of pathogen response
and innate immunity with insulin action and metabolic pathways
that are critical in chronic metabolic diseases. If small molecules
can modulate in vivoPKR activity, therapeutic opportunities may
arise from such efforts.
Animal care and experimental procedures were performed with approval from
animal care committees of Harvard University. Two different types of targeted
mutations of PKR have been established and reported in mice, RNA-binding
domain-defective and kinase-domain-defective models (Abraham et al.,
1999; Baltzis et al., 2002; Yang et al., 1995). In this study, the kinase-
domain-defective PKR-deficient mice have been used. Male Pkr+/+and
Pkr?/?mice in the 129Sv 3 BALB/C mixed background (Abraham et al.,
1999) were kept on a 12 hr light/12 hr dark cycle and were placed on a HFD
(D12492: 60% kcal% fat; Research Diets), beginning at 3 weeks of age ad
libitum. After 6 and 14 weeks on HFD, GTTs were performed by intraperitoneal
glucose injection (1.5 g/kg) following an overnight food withdrawal. After
16 weeks on HFD, ITTs were performed by intraperitoneal insulin injection
(1 IU/kg) following 6 hr daytime food withdrawal. After 20 weeks, these mice
were sacrificed and tissues were collected for further analysis. Total body
fat mass was assessed by dual energy X-ray absorptiometry (DEXA; PIXImus).
For metabolic measurements, mice were placed in an indirect open circuit
calorimeter (Columbus Instruments). Serum insulin, leptin, and adiponectin
levels were measured with ELISA (Alpco). Liver triglycerides were determined
with a colorimetric system (Sigma-Aldrich) adapted for microtiter plate assays
(Furuhashi et al., 2007). Serum alanine aminotransferase level was measured
with Piccolo-lipid panel plus (Abaxis).
Hyperinsulinemic-Euglycemic Clamp Studies
(Furuhashi et al., 2007). For the lipid-induced acute insulin resistance mouse
model, 6-month-old male Pkr+/+and Pkr?/?mice were anesthetized and the
right jugular vein was catheterized with a PE-10 polyethylene tube filled with
heparin solution (100 U/ml) (United States Pharmacopeia). After a 3 day
recovery, overnight-fasted mice were preinfused with lipid (5 ml/kg/hr; Intrali-
The Intralipid we used contains 20% Soybean oil. The Soybean oil is a refined
naturalproductconsisting of amixtureofneutraltriglyceridesof predominantly
linoleic, oleic, palmitic, linolenic, and stearic acids. After the basal period,
a 120 min hyperinsulinemic-euglycemic clamp was conducted with a primed-
continuous infusion of human insulin (Novolin; Novo Nordisk) at a rate of
diate measurement of plasma glucose concentration, and 25% glucose was
infused at variable rates to maintain plasma glucose at basal concentrations.
Insulin-stimulated whole-body glucose disposal was estimated with a contin-
uous infusion of3H-glucose throughout the clamps (0.1 mCi/min). To estimate
insulin-stimulated glucose uptake in individual tissues, 2-14C-deoxyglucose
Figure 7. Regulation of Systemic Metabolic Responses by PKR
PKR senses and responds to obesity, ER stress, and pathogen-related stress
in concert with JNK, leading to metabolic disease under diverse physiological
and pathological conditions. In this capacity, PKR not only integrates immune
stasis and unfolded protein response (UPR) to translational regulation through
eIF2a and insulin signaling through IRS1. Finally, the kinases IRS1 and eIF2a
may represent a ‘‘metabolic inflammasome’’ complex assembled and regu-
lated by PKR.
346 Cell 140, 338–348, February 5, 2010 ª2010 Elsevier Inc.
(2-14C-DG; Perkin Elmer) was administered as a bolus (10 mCi) 75 min after the
start of clamps. Blood samples were collected for the determination of plasma
muscles from both hindlimbs and epididymal adipose tissue were harvested
and immediately frozen in liquid N2and stored at ?80?C until further analysis.
Calculations for the determination of metabolic parameters are described in
the Determination of Metabolic Parameters in Hyperinsulinemic-Euglycemic
Clamp Study section of the Extended Experimental Procedures.
3H2O, and 2-14C-DG concentrations. At the end of clamps,
Portal Vein Insulin Infusion and Protein Extraction from Tissues
Following 6 hr food withdrawal, mice were anesthetized and insulin (2 IU/kg)
or PBS was injected into mice through the portal vein. Three minutes
after injection, tissues were removed, frozen in liquid nitrogen, and kept at
?80?C until processing. For protein extraction, tissues were placed in a cold
lysis buffer (25 mM Tris-HCl [pH 7.4], 1 mM EGTA, 1 mM EDTA, 10 mM
Na4P2O7, 10 mM NaF, 2 mM Na3VO4, 1% NP-40, 1 mM PMSF, 1% protease
inhibitor cocktail [Sigma-Aldrich]). After homogenization on ice, the tissue
lysates were centrifuged, and the supernatants were used for western blot
MEF Culture and Analysis
Primary Pkr+/+and Pkr?/?MEFs were used to assess phosphorylated JNK
level or JNK activity under palmitic acid, TNF-a, thapsigargin, or polyino-
sinic-polycytidylic acid stimulation. Both Pkr+/+and Pkr?/?cell lines were
established using the standard 3T3 immortalization protocol and used to
assess phosphorylation level of IRS1 and effects of PKR reconstitution on
insulin action. At 70%–80% confluency, cells were serum depleted for 3 hr
or overnight prior to the stimuli. Sodium palmitate was dissolved in water at
65?C and prepared as 20 mM solution. In cell culture experiments, the
20 mM palmitic acid preparation was diluted with 0.5% BSA containing
DMEM to obtain the 0.5 mM palmitic acid concentration. Reagents and
recombinant cytokines were gently added to the culture dishes in the incu-
bator to prevent any environmental stress. For retrovirus production,
BOSC23 packaging cells were transfected at 70% confluence with Lipofect-
amine 2000 (Invitrogen) and a retroviral vector containing flag-tagged human
PKR coding region. After 48 hr, the viral supernatant was harvested and
filtered. Cells were incubated overnight with the viral supernatant, supple-
mented with 8 mg/ml polybrene. Expression vectors of adenovirus were con-
structed by cloning flag-tagged human PKR in adenovirus vector pAD/CMV/
V5-DEST, and the viruses were produced as described in Virapower adeno-
virus system (Invitrogen). Amplified-virus was used to infect primary Jnk1+/+
Tissue or cell lysates containing 100–300 mg of protein were mixed with
agarose-conjugated PKR antibody (Santa Cruz). The mixture was agitated at
4?C, pelleted by centrifugation, and washed with lysis buffer followed by addi-
tional washes with PKR kinase buffer (15 mM HEPES [pH7.4], 10 mM MgCl2,
40 mM KCl, 2 mM DTT) for equilibration. The beads were incubated in kinase
SDS-PAGE. To assess the phosphorylation of IRS1 by PKR, in vitro kinase
assays were performed with anti-PKR immunoprecipitates from lysates of
Pkr+/+and Pkr?/?MEFs, which were treated by TNF-a or thapsigargin. The
anti-PKR immunoprecipitates were mixed with IRS1, which was immunopuri-
fied with anti-IRS1 antibody (Upstate Biotechnology) from serum-starved WT
MEFs. Further procedures are provided in the PKR Kinase Assay with
Recombinant Protein section of the Extended Experimental Procedures.
Quantitative Real-Time PCR Analysis
Total RNA was isolated using Trizol reagent (Invitrogen). For reverse transcrip-
tion, total RNA was converted to first strand cDNA using a high capacity cDNA
reverse transcription system (Applied Biosystems). Quantitative real-time PCR
analysis was performed using SYBR Green in a real-time PCR machine (7300
Real Time PCR system; Applied Biosystems). Primers are listed in Table S1.
To normalize expression data of WAT and liver, 36B4 and GAPDH mRNAs
were used as an internal control gene, respectively.
The mean values for biochemical data from each group were compared by
Student’s t test. Comparisons between multiple time points were analyzed
using repeated-measures analysis of variance, ANOVA. In all tests, p < 0.05
was considered significant.
Supplemental Information includes Extended Experimental Procedures, six
figures, and one table and can be found with this article online at doi:10.1016/
butions, especially Brenna Baccaro, Sara Vallerie, Margaret F. Gregor, and
Ling Yang for support and discussions and Kristen Gilbert, Rebecca Foote,
and Julie Gound for administrative assistance. We are grateful to Jun Eguchi
and Atsuo Sasaki for helpful discussions. This study is supported by a grant
from the National Institutes of Health (to G.S.H.). T.N. is supported by fellow-
ships from the International Human Frontier Science Program and the Uehara
for the Promotion of Science and the American Diabetes Association. H.C. is
betes Association. G.S.H. is a share owner and on the Scientific Advisory
Board of Syndexa Pharmaceuticals.
Received: June 26, 2009
Revised: October 21, 2009
Accepted: December 31, 2009
Published: February 4, 2010
Abraham, N., Stojdl, D.F., Duncan, P.I., Methot, N., Ishii, T., Dube, M., Vander-
hyden, B.C., Atkins, H.L., Gray, D.A., McBurney, M.W., et al. (1999). Charac-
terization of transgenic mice with targeted disruption of the catalytic domain
of the double-stranded RNA-dependent protein kinase, PKR. J. Biol. Chem.
Antonelli, A., Ferri, C., Ferrari, S.M., Colaci, M., Sansonno, D., and Fallahi, P.
(2009). Endocrine manifestations of hepatitis C virus infection. Nat. Clin. Pract.
Endocrinol. Metab. 5, 26–34.
Bahtiyar, G., Shin, J.J., Aytaman, A., Sowers, J.R., and McFarlane, S.I. (2004).
Association of diabetes and hepatitis C infection: epidemiologic evidence and
pathophysiologic insights. Curr. Diab. Rep. 4, 194–198.
Baltzis,D.,Li,S.,and Koromilas, A.E.(2002).Functional characterizationofpkr
gene products expressed in cells from mice with a targeted deletion of the
N terminus or C terminus domain of PKR. J. Biol. Chem. 277, 38364–38372.
Bonnet, M.C., Weil, R., Dam, E., Hovanessian, A.G., and Meurs, E.F. (2000).
PKR stimulates NF-kappaB irrespective of its kinase function by interacting
with the IkappaB kinase complex. Mol. Cell. Biol. 20, 4532–4542.
Chakrabarti, A., Sadler, A.J., Kar, N., Young, H.A., Silverman, R.H., and Wil-
liams, B.R. (2008). Protein kinase R-dependent regulation of interleukin-10 in
response to double-stranded RNA. J. Biol. Chem. 283, 25132–25139.
Delhem, N., Sabile, A., Gajardo, R., Podevin, P., Abadie, A., Blaton, M.A.,
Kremsdorf, D., Beretta, L., and Brechot, C. (2001). Activation of the inter-
feron-inducible protein kinase PKR by hepatocellular carcinoma derived-
hepatitis C virus core protein. Oncogene 20, 5836–5845.
Eizirik, D.L., Cardozo, A.K., and Cnop, M. (2008). The role for endoplasmic
reticulum stress in diabetes mellitus. Endocr. Rev. 29, 42–61.
Furuhashi, M., Tuncman, G., Gorgun, C.Z., Makowski, L., Atsumi, G., Vaillan-
court, E., Kono, K., Babaev, V.R., Fazio, S., Linton, M.F., et al. (2007). Treat-
ment of diabetes and atherosclerosis by inhibiting fatty-acid-binding protein
aP2. Nature 447, 959–965.
Cell 140, 338–348, February 5, 2010 ª2010 Elsevier Inc. 347
Gao, Z., Zhang, X., Zuberi, A., Hwang, D., Quon, M.J., Lefevre, M., and Ye, J.
(2004). Inhibition of insulin sensitivity by free fatty acids requires activation of
multiple serine kinases in 3T3-L1 adipocytes. Mol. Endocrinol. 18, 2024–2034.
Garcia, M.A., Gil, J., Ventoso, I., Guerra, S., Domingo, E., Rivas, C., and Este-
ban, M. (2006). Impact of protein kinase PKR in cell biology: from antiviral to
antiproliferative action. Microbiol. Mol. Biol. Rev. 70, 1032–1060.
Goh, K.C., deVeer, M.J., and Williams, B.R. (2000). The protein kinase PKR is
required for p38 MAPK activation and the innate immune response to bacterial
endotoxin. EMBO J. 19, 4292–4297.
Gregor, M.F., Yang, L., Fabbrini, E., Mohammed, B.S., Eagon, J.C., Hotamisli-
gil, G.S., and Klein, S. (2009). Endoplasmic reticulum stress is reduced in
tissues of obese subjects after weight loss. Diabetes 58, 693–700.
Gregor, M.G., and Hotamisligil, G.S. (2007). Adipocyte stress: The endo-
plasmic reticulum and metabolic disease. J. Lipid Res. 48, 1905–1914.
Gusella, G.L., Musso, T., Rottschafer, S.E., Pulkki, K., and Varesio, L. (1995).
Potential requirementof afunctional double-stranded RNA-dependent protein
kinase(PKR)for the tumoricidal activation of macrophagesby lipopolysaccha-
ride or IFN-alpha beta, but not IFN-gamma. J. Immunol. 154, 345–354.
Hirosumi, J., Tuncman, G., Chang, L., Gorgun, C.Z., Uysal, K.T., Maeda, K.,
Karin, M., and Hotamisligil, G.S. (2002). A central role for JNK in obesity and
insulin resistance. Nature 420, 333–336.
Holcik, M., and Sonenberg, N. (2005). Translational control in stress and
apoptosis. Nat. Rev. Mol. Cell Biol. 6, 318–327.
Hotamisligil, G.S. (2006). Inflammation and metabolic disorders. Nature 444,
Hotamisligil, G.S., and Erbay, E. (2008). Nutrient sensing and inflammation in
metabolic diseases. Nat. Rev. Immunol. 8, 923–934.
Hotamisligil, G.S., Peraldi, P., Budavari, A., Ellis, R., White, M.F., and Spiegel-
man, B.M. (1996). IRS-1-mediated inhibition of insulin receptor tyrosine kinase
activity in TNF-alpha- and obesity-induced insulin resistance. Science 271,
Kim, J.K., Fillmore, J.J., Sunshine, M.J., Albrecht, B., Higashimori, T., Kim,
D.W., Liu, Z.X., Soos, T.J., Cline, G.W., O’Brien, W.R., et al. (2004). PKC-theta
knockout mice are protected from fat-induced insulin resistance. J. Clin.
Invest. 114, 823–827.
McCormack, S.J., Ortega,L.G., Doohan, J.P.,and Samuel, C.E. (1994).Mech-
anism of interferon action motif I of the interferon-induced, RNA-dependent
protein kinase (PKR) is sufficient to mediate RNA-binding activity. Virology
Ozcan, U., Cao, Q., Yilmaz, E., Lee, A.H., Iwakoshi, N.N., Ozdelen, E., Tunc-
man, G., Gorgun, C., Glimcher, L.H., and Hotamisligil, G.S. (2004). Endo-
plasmic reticulum stress links obesity, insulin action, and type 2 diabetes.
Science 306, 457–461.
Ron, D., and Walter, P. (2007). Signal integration in the endoplasmic reticulum
unfolded protein response. Nat. Rev. Mol. Cell Biol. 8, 519–529.
Samuel, C.E. (1993). The eIF-2 alpha protein kinases, regulators of translation
in eukaryotes from yeasts to humans. J. Biol. Chem. 268, 7603–7606.
Scarim, A.L., Arnush, M., Blair, L.A., Concepcion, J., Heitmeier, M.R., Scheu-
anisms of beta-cell death in response to double-stranded (ds) RNA and
interferon-gamma: dsRNA-dependent protein kinase apoptosis and nitric
oxide-dependent necrosis. Am. J. Pathol. 159, 273–283.
Shi, H., Kokoeva, M.V., Inouye, K., Tzameli, I., Yin, H., and Flier, J.S. (2006).
TLR4 links innate immunity and fatty acid-induced insulin resistance. J. Clin.
Invest. 116, 3015–3025.
Takada, Y., Ichikawa, H., Pataer, A., Swisher, S., and Aggarwal, B.B. (2007).
Genetic deletion of PKR abrogates TNF-induced activation of IkappaBalpha
kinase, JNK, Akt and cell proliferation but potentiates p44/p42 MAPK and
p38 MAPK activation. Oncogene 26, 1201–1212.
Tokumoto, Y., Hiasa, Y., Horiike, N., Michitaka, K., Matsuura, B., Chung, R.T.,
and Onji, M. (2007). Hepatitis C virus expression and interferon antiviral action
is dependent on PKR expression. J. Med. Virol. 79, 1120–1127.
Urano, F., Wang, X., Bertolotti, A., Zhang, Y., Chung, P., Harding, H.P., and
by transmembrane protein kinase IRE1. Science 287, 664–666.
von Herrath, M. (2009). Can we learn from viruses how to prevent type 1 dia-
betes?: the role of viral infections in the pathogenesis of type 1 diabetes and
the development of novel combination therapies. Diabetes 58, 2–11.
Williams, B.R. (2001). Signal integration via PKR. Sci. STKE 2001, RE2.
Wu, S., and Kaufman, R.J. (1996). Double-stranded (ds) RNA binding and not
dimerization correlates with the activation of the dsRNA-dependent protein
kinase (PKR). J. Biol. Chem. 271, 1756–1763.
Yang, Y.L., Reis, L.F., Pavlovic, J., Aguzzi, A., Schafer, R., Kumar, A., Williams,
B.R., Aguet, M., and Weissmann, C. (1995). Deficient signaling in mice devoid
of double-stranded RNA-dependent protein kinase. EMBO J. 14, 6095–6106.
348 Cell 140, 338–348, February 5, 2010 ª2010 Elsevier Inc.
EXTENDED EXPERIMENTAL PROCEDURES
Anti-IRS1 and anti-phospho-IRS1 (Ser307) were from Upstate Biotechnology (Lake Placid, NY). Antibody against PKR, JNK1, Akt,
phospho-Akt, insulin receptor b subunit, b-tubulin, Fatty acid synthase (FAS), and Actin were from Santa Cruz Biotechnology (Santa
Cruz, CA). Anti-phospho-eIF2a (Ser52) antibody was purchased from Invitrogen (Carlsbad, CA). Anti-phospho-insulin receptor
(Tyr1162/1163) was purchased from Calbiochem (Gibbstown, NJ). Anti-phospho-JNK (Thr183/Tyr185) antibody was purchased
from Cell Signaling Technology (Danvers, MA). Recombinant IRS1, agarose-conjugated PKR, and anti-phopho PKR antibody
were purchased from Millipore (Billerica, MA). Active PKR recombinant protein was purchased from SignalChem (Richmond, BC,
Adipocyte and Stromal-Vascular (SV) Fractionation
Epididymal fat pads were isolated from 11-week-old mice fed either RD or HFD for 8 weeks. Following mincing in albumin containing
Krebs-Ringer phosphate (KRP) buffer, 1 mg/ml Collagenase II (Sigma) and 0.2 mg/ml DNase I (Sigma) were added and incubated at
37?C for 20 min on a shaking platform. The cell suspensions were filtrated through 250 mm nylon mesh to remove undigested tissues.
Adipose tissues were then separated by their ability to float after low-speed centrifugation. To obtain total SV fractions, the cells
below the adipocyte layer were collected and centrifuged. The pellets were washed in albumin containing KRP buffer.
Determination of Metabolic Parameters in Hyperinsulinemic-Euglycemic Clamp Study
was determined by the difference between3H counts with and without drying. Tissue 2-14C-DG-6-phosphate content was deter-
mined in homogenized samples that were subjected to an ion-exchange column to separate 2-14C-DG-6-phosphate from 2-14C-
DG. Rates of basal hepatic glucose production and insulin-stimulated whole-body glucose uptake were determined as the ratio
of the3H-glucose infusion rate to the specific activity of plasma glucose at the end of the basal period and during the final 30 min
of clamps, respectively. Hepatic glucose production during the hyperinsulinemic-euglycemic clamps was determined bysubtracting
the glucose infusion rate from the whole-body glucose uptake. Glucose uptake in individual tissues was calculated from the plasma
2-14C-DG profile, which was fitted with an exponential curve, and tissue 2-14C-DG-6-phosphate content.
PKR Pull-Down Assay
An agarose-conjugated PKR (Millipore) was mixed with recombinant full-length IRS1 protein (Millipore) in interaction buffer (5 mM
Tris-HCl [pH 7.4], 25 mM KCl, 1 mM MgCl2, 0.25% Triton X) and then agitated at 4?C. After the agitation, the agarose-conjugated
PKR was pelleted by centrifugation and washed with the interaction buffer followed by SDS-PAGE. One part of supernatant was
kept to detect unbound IRS1.
PKR Kinase Assay with Recombinant Protein
For kinase assay with recombinant protein, full-length IRS1 (Millipore) was dephosphorylated by l protein phosphatase (New
England BioLabs) and used as a substrate. The kinase assay was performed with an active PKR (SignalChem) and the dephosphory-
lated IRS1 in kinase buffer (5 mM Tris-HCl [pH 7.4], 25 mM KCl, 1 mM MgCl2, 0.25% Triton X, 10 mM ATP, 10 mCi32P-gATP) followed
JNK Kinase Assay
For JNK kinase assay, tissue lysates containing 500 mg of protein were mixed with JNK1 antibody (Santa Cruz) and protein G sephar-
ose beads. The mixture was agitated at 4?C, pelleted by centrifugation, and washed with the lysis buffer followed by additional
washes with JNK kinase buffer (25 mM HEPES [pH 7.4], 20 mM MgCl2, 20 mM b-glycerophosphate, 0.5 mM EGTA, 0.5 mM NaF,
0.5 mM Na3VO4, 1 mM PMSF) for equilibration. The beads were incubated in kinase buffer containing 10 mCi32P-gATP and c-Jun
protein at 30?C for 20 min followed by SDS-PAGE.
Cell 140, 338–348, February 5, 2010 ª2010 Elsevier Inc. S1
PkrIrf7 Isg15 2 5 Oas1a
Relative mRNA levels
Pkr Irf7Isg15 2 5 Oas1a
Relative mRNA levels
Relative PKR/ -Tubulin levels
Relative PKR/ -Tubulin levels
IB: p-PKR (Thr 451)
IP: PKR -> PKR kinase assay
Figure S1. Related to Figure 1
(A and B) Expression of PKR and other interferon a/b target genes in adipocytes, stromal vascular (SV) fraction (A), and liver (B) of WT mice fed high-fat diet for 8
weeks. Adipocytes and SV fraction were isolated from white adipose tissues. Data are shown as the mean ± SEM. *p < 0.05.
(C and D) Regulation of PKR activity and expression in skeletal muscle. Genetic (ob/ob) (C) and dietary (D) mouse models of obesity were used to examine PKR
activity by a kinase assay using immunopurified PKR and32P-gATP in skeletal muscle with age- and sex-matched lean controls. b-tubulin protein is shown as
controls. The graphs show the quantification of the data. *p < 0.05.
90 min. Phosphorylation level of PKR was examined with anti-phopho-PKR (Thr451) antibody. PKR activity was assessed by autophosphorylation level of PKR
S2 Cell 140, 338–348, February 5, 2010 ª2010 Elsevier Inc.
IB: p-PKR (Thr 451)
Figure S2. Related to Figure 2
(A) Thapsigargin-induced PKR expression and activation in the presence of cycloheximide (CHX). Wild-type MFEs were pretreated with 20 mg/ml CHX for 30 min
before addition of 100 nM thapsigargin for 1 hr. Cell lysates were analyzed by western blot analyses with antibodies as indicated. Phosphorylation level of PKR
was examined with anti-phospho-PKR (Thr451) antibody.
(B) Polyinosinic-polycytidylic acid (polyIdC)-induced IRS1 phosphorylation in PKR-dependent manner. Induction of IRS1 phosphorylation after 100 mg/ml poly-
inosinic-polycytidylic acid (polyIdC) treatment for 2 hr in Pkr+/+and Pkr?/?MEFs. Phosphorylation level of IRS1 on serine 307 was examined with anti-phospho-
(C) Expression level of retrovirus-mediated Flag-tagged PKR in Pkr?/?MEFs. Flag-tagged PKR was introduced by retrovirus-mediated gene transfer in Pkr?/?
MEFs. After puromycin selection, the cell lysates were analyzed by western blot analyses with antibodies as indicated.
Cell 140, 338–348, February 5, 2010 ª2010 Elsevier Inc. S3
Cell lysate IP: IRS1
Figure S3. Related to Figure 3
Association of IRS1 with PKR but not with PERK.
(A) Cell lysates were prepared from 300 nM thapsigargin-treated or nontreated MEFs for 3 hr followed by immunoprecipitation with anti-IRS1 antibody and
western blot analyses with anti-PKR or anti-PERK antibodies.
(B and C) Cell lysates were prepared from 100 nM thapsigargin-treated or nontreated MEFs for 1 hr followed by immunoprecipitation with anti-IRS1 antibody and
western blot analyses with anti-PKR or anti-PERK antibodies.
S4 Cell 140, 338–348, February 5, 2010 ª2010 Elsevier Inc.
Time after insulin injection (min)
Blood glucose (mg/dl)
015 30 6090 120
Time after glucose injection (min)
Blood glucose (mg/dl)
0 1530 60 90 120
Time after glucose injection (min)
015 30 6090 120
Blood glucose (mg/dl)
Time after insulin injection (min)
Blood glucose (mg/dl)
p = 0.273
Amount of food (g)
p = 0.195
p = 0.153
p = 0.560
*p = 0.001
*p = 0.001
Activity (Xamb) VO2
Figure S4. Related to Figure 4
(A–D)Glucose/insulintolerancetests inPkr+/+and Pkr?/?miceonRDand HFD.GlucosetolerancetestswereperformedinPkr+/+(n=5)andPkr?/?mice(n=6)on
the mean ± SEM. *p < 0.05.
(E–J)Metabolic studiesinPkr+/+and Pkr?/?miceon HFD. MetabolicstudieswereperformedinPkr+/+(n=6) and Pkr?/?(n= 7)mice onHFD for10 weeks.(E)Heat
production, (F) Dailyfoodintake, (Gand H) Physical activity determined by automatic sensors on x(G) and zaxes (H), (Iand J)rates of oxygen consumption(I) and
carbon dioxide production (J). Data are shown as the mean ± SEM.
Cell 140, 338–348, February 5, 2010 ª2010 Elsevier Inc. S5
++ + ++ +
Relative mRNA levels
Figure S5. Related to Figure 5
(A and B) PKR expression and activation in skeletal muscle in obese mouse models. (A) Insulin-stimulated IRb tyrosine 1162/1163 and Akt serine 473 phosphor-
ylation in skeletal muscle of Pkr+/+and Pkr?/?mice on HFD. Tissue lysates were analyzed by western blot with antibodies as indicated. (B) JNK1 activity was
detected by a kinase assay using immunopurified JNK1,32P-gATP, and recombinant c-Jun protein as substrate. b-tubulin is shown as a control.
(C) Gene expression in WAT including proinflammatory cytokine levels in Pkr+/+and Pkr?/?mice on RD. Data are shown as the mean ± SEM.
S6 Cell 140, 338–348, February 5, 2010 ª2010 Elsevier Inc.
IB: Cleaved PARP
IB: Caspase 12
Figure S6. Related to Figure 6
Apoptotic gene expression after lipid infusion in liver of Pkr+/+ and Pkr?/?mice. Tissue lysates were analyzed by western blot analyses with antibodies as indi-
cated. Tunicamycin (TM) treated liver sample was used as a control for induction of liver injury.
Cell 140, 338–348, February 5, 2010 ª2010 Elsevier Inc. S7
Table S1. Primers for Real-Time PCR Download full-text
GeneSpeciesAccession #Forward Reverse
CCC TCA CAC TCA GAT CAT CTT CT
ACA ACC ACG GCC TTC CCT ACT T
GCA ACT GTT CCT GAA CTC AAC T
GCT CTT ACT GAC TGG CAT
CCA CTC ACC TGC TGC TAC TCA
CCC CAG TGT CCT TAC AGA GTG
TCA TCG GAC GCA CTT GGA A
AAA ACA AGG TGG ATT GTC ACA CG
CTG GAG CCA TGG GTA TGC A
TGG GAC CTA AAG GTG AAG ATG CTG -3' 5'- TCA GGC GCA AAT GCT TGA TCA C
GAG GTT CAG CAT GAG AGA CGT T -3' 5'- ACA CAG TTG GTA CCA GTG CTT G
CAC TGG TCT AGG ACC CGA GAA -3' 5'- AGG GGG AGA TGT TCA GCA TGT
ATC ACC ATC TTC CAG GAG CG -3' 5'- CCA TGC CAT CAC TGC CAC CC
-3' 5'- GCT ACG ACG TGG GCT ACA G
-3' 5'- CAC GAT TTC CCA GAG AAC ATG TG
-3' 5'- ATC TTT TGG GGT CCG TCA ACT
-3' 5'- CGC AGC TCT AGG AGC ATG TG
-3' 5'- TGG TGA TCC TCT TGT AGC TCT CC
-3' 5'- GTG CCC AGA GTG GAT GTC T
-3' 5'- CAA CCA CCT TGA ATG GCA AGA
-3' 5'- GTT GGG CTC ACA CTG TTC ATA AT
-3' 5'- AAG CAC AAG CCG AGA CTG CT