MOLECULAR AND CELLULAR BIOLOGY, Nov. 2011, p. 4417–4429
Copyright © 2011, American Society for Microbiology. All Rights Reserved.
Vol. 31, No. 22
A Novel Role for hSMG-1 in Stress Granule Formation?
James A. L. Brown,1,2† Tara L. Roberts,1,3† Renee Richards,1,3Rick Woods,1Geoff Birrell,1
Y. C. Lim,1,3Shigeo Ohno,4Akio Yamashita,4Robert T. Abraham,5
Nuri Gueven,1and Martin F. Lavin1,3*
Radiation Biology and Oncology Laboratory, Queensland Institute of Medical Research, Brisbane, Queensland 4029, Australia1;
Chromosome Biology Centre, National University of Ireland, Galway, Ireland2; The University of Queensland Centre for
Clinical Research, Brisbane, Queensland 4029, Australia3; Department of Molecular Biology,
Yokohama City University School of Medicine, Yokohama 236-0004, Japan4; and
Wyeth Oncology Research, 401 N. Middletown Road, Pearl River, New York 109655
Received 27 July 2011/Returned for modification 25 August 2011/Accepted 29 August 2011
hSMG-1 is a member of the phosphoinositide 3 kinase-like kinase (PIKK) family with established roles in
nonsense-mediated decay (NMD) of mRNA containing premature termination codons and in genotoxic stress
responses to DNA damage. We report here a novel role for hSMG-1 in cytoplasmic stress granule (SG)
formation. Exposure of cells to stress causing agents led to the localization of hSMG-1 to SG, identified by
colocalization with TIA-1, G3BP1, and eIF4G. hSMG-1 small interfering RNA and the PIKK inhibitor
wortmannin prevented formation of a subset of SG, while specific inhibitors of ATM, DNA-PKcs, or mTOR had
no effect. Exposure of cells to H2O2and sodium arsenite induced (S/T)Q phosphorylation of proteins. While
Upf2 and Upf1, an essential substrate for hSMG-1 in NMD, were present in SG, NMD-specific Upf1 phos-
phorylation was not detected in SG, indicating hSMG-1’s role in SG is separate from classical NMD. Thus, SG
formation appears more complex than originally envisaged and hSMG-1 plays a central role in this process.
Cells are exposed to a variety of genotoxic stresses that
impact on DNA integrity, gene regulation, subcellular organ-
elles, and metabolic events. The phosphoinositide 3 kinase-like
kinase (PIKK), hSMG-1, is an ?400-kDa protein that plays an
important role in cellular viability which is demonstrated by the
embryonic lethality observed in hSMG-1-deficient mice (39).
In addition, hSMG-1 plays a central role in maintaining
mRNA quality through the process of nonsense-mediated
mRNA decay (NMD), where it has been demonstrated to be
crucial for initiating the signaling cascade through phosphor-
ylation of Upf1 at S1078 and S1096, resulting in degradation of
mRNA containing premature termination codons (PTC) (4, 7,
24, 42, 59). PTC-containing mRNAs can be produced through
genomic mutations, alternative splicing, or RNA damage, and
NMD is responsible for the elimination of aberrant PTC-con-
taining mRNAs which could encode nonfunctional truncated
proteins that could interfere with their endogenous counter-
parts (39, 59). NMD is elicited by recognition of the SURF
complex (hSMG-1, Upf1, eRF1, and eRF3) when the termi-
nation codon is situated within ?50bp of the last exon junction
complex (EJC) (7, 23, 24). This results in SMG-1 phosphory-
lating Upf1, leading to NMD-mediated mRNA degradation
(24, 42, 58). Recent work has also implicated two cofactors in
hSMG-1 regulation: SMG-8 and -9 (58). These proteins form
a trimeric complex with hSMG-1 and are required for NMD to
occur. SMG-8 acts to inhibit hSMG-1 kinase activity prior to
interaction with the EJC. In addition to NMD, mRNAs can be
regulated through storage in cytoplasmic stress granules (SG)
or by degradation in the related and often associated struc-
tures, processing bodies (P bodies) (3, 12, 13, 31). SG are
formed in response to cellular stress such as heat shock and
oxidative stress that results in the phosphorylation of eukary-
otic translation initiation factor 2? (eIF2?) (32). SG are com-
posed of accumulated mRNA and their associated proteins,
such as TIA-1, eIF4G, and G3BP1 (32, 33). That SG are only
transiently formed suggests that they are active sites where
individual mRNAs are processed for storage, translation dur-
ing stress and recovery, or shuttled to the associated structures,
PB, for degradation (3, 28, 49).
Brumbaugh et al. (7) and Gewandter et al. (19) demon-
strated that hSMG-1 is a genotoxic stress-activated protein
kinase that displays some functional overlap with the related
kinase, ATM. Expression of hSMG-1 was required for optimal
activation of p53 in response to ionizing radiation (IR) and
small interfering RNA (siRNA) depletion of hSMG-1 caused
constitutive activation of p53 and Chk2, leading to an in-
creased sensitivity to IR (7). As in the case of NMD, Upf1 was
shown to be a substrate for hSMG-1 in response to radiation
damage. hSMG-1 has also been shown to regulate the G1/S
checkpoint in response to prolonged oxidative stress by p53
activation and p53-independent proteolysis of p21 (18).
hSMG-1 also plays a role in telomere stability. Telomeric re-
peats are transcribed into noncoding RNA known as TERRA.
hSMG-1 negatively regulates TERRA association with telo-
meres, and hSMG-1 depletion increased the number of
TERRA-positive chromosomes and resulted in telomere desta-
bilization (6, 9). In addition, depletion of hSMG-1 in tumor
cells markedly increased the extent and accelerated the rate of
apoptosis induced by tumor necrosis factor alpha (TNF-?)
(46). Furthermore, hSMG-1 was demonstrated to be required
for granzyme B-mediated apoptosis in a primary tumor cell
* Corresponding author. Mailing address: Queensland Institute of
Medical Research, 300 Herston Road, Herston, Queensland 4006,
Australia. Phone: 617 3362 0341. Fax: 617 3362 0106. E-mail: martin
† J.A.L.B. and T.L.R. contributed equally to this study.
?Published ahead of print on 12 September 2011.
line (41). Inactivation of smg-1 has also been shown to increase
the life span of Caenorhabditis elegans, which appears to be
related to resistance to oxidative stress (38). hSMG-1 has also
been shown to negatively regulate hypoxia-inducible factor 1?
(HIF-1?) in part by blocking mitogen-activated protein (MAP)
kinase activation (10). Thus, it is evident that hSMG-1 is a key
player not only in NMD but also functions in the DNA damage
response, oxidative stress response, hypoxia, and apoptosis.
We describe here a new role for hSMG-1 in the formation of
SG. This role appears to be separate from its role in active
NMD, since although we demonstrated that Upf1 localized to
SG, Upf1 was not phosphorylated on residues known to play a
key role in NMD. hSMG-1 colocalized with a number of SG-
specific markers, and knockdown by siRNA prevented SG for-
mation. Inhibition of PIKK activity by wortmannin treatment
reduced SG formation similar to hSMG-1 knockdown, but
overexpression of kinase dead hSMG-1 did not prevent SG
formation. Our data point to a novel and complex role for
hSMG-1 in SG formation as part of the stress response.
MATERIALS AND METHODS
Cell culture conditions. Cells were cultured in RPMI 1640 (lymphoblastoid
cell lines [LCLs]) or Dulbecco modified Eagle medium supplemented with 10%
fetal calf serum (Invitrogen), 100 U of penicillin (Invitrogen)/ml, and 100 U of
streptomycin (Invitrogen)/ml. Primary normal foreskin fibroblasts (NFF) were
established from primary tissue (Glen Boyle, Queensland Institute of Medical
Research). Approximately 11 different NFF strains, obtained from separate
donors, were used during this study. Primary keratinocytes were supplied by N.
Saunders (University of Queensland), and human kidney proximal tubular cells
were supplied by C. Percy (University of Queensland). Primary melanocytes were
also supplied by G. Boyle. The cell lines used were HeLa, U2OS, and A549 and
LCLs. All cells were maintained at 37°C in a humidified atmosphere of 5% CO2
and 95% air.
IR treatment of cells. Irradiation of cells was performed at room temperature
using a137Cs source delivering gamma rays at a dose rate of 1.096 Gy/min (MDS
Gammacell irradiator; Nordion, Ottawa, Ontario, Canada).
Hydrogen peroxide treatment of cells. Adherent cell lines were washed twice
in serum-free medium. The cells were resuspended in 1? original culture volume
of serum-free medium containing H2O2(at the indicated concentrations) and
returned to the incubator.
Heat shock treatment of cells. Adherent cells were placed in a 45°C water bath
and allowed to equilibrate for 5 min. Cells were held at this temperature for 1 h
prior to fixation.
Kinase inhibitor treatment of cells. Wortmannin, AMA-37, and Ku55933
(Calbiochem, La Jolla, CA) were dissolved in dimethyl sulfoxide (DMSO). Ad-
herent cells were grown to the desired density (?70% confluent), and inhibitors
were added directly into the cell culture medium at final concentrations of 10 ?M
for wortmannin and Ku55933 and 30 ?M for AMA-37. Rapamycin (a gift from
Dianne Watters, Griffith University) was used at 20 ?M. Cells were returned to
the incubator for 2 h. Inhibitor treatment was continued in conjunction with any
NaAs treatment of cells. Sodium arsenite (NaAs) was dissolved in water to
create a 1 M stock solution. NaAs was added directly into the cell culture
medium at a final concentration of 1 mM for up to 1 h.
Antibodies. Constructs encoding fragments of hSMG-1 fused to glutathione
S-transferase (GST; antibody 1 [Ab1] against amino acids [aa] 24 to 101, Ab2
against aa 155 to 222, and Ab3 against aa 2458 to 2851) were generated in
pGEX5. Recombinant protein production and immunization of sheep was per-
formed as described previously (15). Antibodies were purified from sheep sera by
affinity chromatography. Sera were diluted 1/10 in phosphate-buffered saline
(PBS) and passed over a GST column to deplete the anti-GST antibodies. Eluted
sera were then passed over GST–hSMG-1 columns to isolate anti-hSMG-1 an-
tibodies (the columns were produced using the same protein fragments used for
immunization or a fragment including the N-terminal region detected by Ab1
and Ab2, aa 24 to 222). hSMG-1-specific antibodies were eluted, buffer ex-
changed, and concentrated into a final buffer of PBS, 50% glycerol, and 0.1%
sodium azide. hSMG-1 antibodies were used for immunofluorescence at a 1/100
to 1/600 dilution. hSMG-1 Ab3 was used for Western blotting at 1/1,000. Upf1
was detected using anti-Upf1 antibody (rabbit serum) and Upf2 detected with
anti-Upf2 (affinity-purified rabbit antibody [a gift from L. E. Maquat, University
of Rochester, Rochester, NY]). The phospho-UPF1 antibody that recognizes
pS1078 and pS1096 was supplied by S. Ohno and has been described previously
(59). The phosphorylated (S/T)Q motif antibody (rabbit serum) at a dilution of
1/100 to 1/400 was generated in Robert. T. Abraham’s Laboratory at the Burn-
ham Institute for Medical Research (La Jolla, CA). Two batches of this antibody
were used; although the staining intensity differed between the batches of anti-
body, the overall pattern of staining was consistent. The commercial antibodies
and the respective dilutions used for immunofluorescence were as follows: goat
polyclonal TIA-1 antiserum (C20, Sc-1751; Santa Cruz Biotechnology), 1/100;
G3BP1 antibody clone 23 (catalog no. 611126; BD Biosciences), 1/100; rabbit
anti-eIF4G antibody (Santa Cruz catalog no. SC11373, H300), 1/500; and mouse
anti-?H2AX (Upstate Biotechnology). HA was detected by using mouse anti-
hemagglutinin (anti-HA; Covance) at 1/2,000. The antibodies used for Western
blotting were mouse anti-ATM (Genetex) at 1/2,000, rabbit anti-phospho-1981
ATM (R&D Systems) at 1/2,000, mouse anti-DNA-PKcs(Oncogene/Merck) at
1/1,000, and rabbit anti-phospho-2056 DNA-PKcs(Abcam) at 1/2,000.
Conditions for immunofluorescent detection of hSMG-1. Cells were fixed in
1% paraformaldehyde–PBS, followed by permeabilization in 0.1% Triton X-100
and PBS. Nonspecific sites were blocked using 10% fetal calf serum in PBS. All
primary antibodies diluted in 1% newborn calf serum in 0.1% Triton X-100 and
PBS. Alexa Fluor 488- or Alexa Fluor 594-conjugated (Molecular Probes) sec-
ondary antibodies were used at 1/200 to 1/1,000. DNA was visualized using
Hoechst 3339 or DAPI (4?,6?-diamidino-2-phenylindole). Images were captured
using a digital camera (Zeiss AxioCam MRm) attached to a fluorescence micro-
scope (Zeiss Axiovert 2 Mot Plus) with a ?63/1.4 Zeiss Plan-Apochromat oil lens
or using a DeltaVision fluorescence microscopy system (Applied Precision) with
?60/1.4 Olympus Plan-Apochromat oil lens, and images were analyzed with
Softworx imaging software. Deconvolution was performed for five cycles, using a
conservative setting with high noise filtering (Applied Precision). All images
were acquired at room temperature. Colocalization analysis was performed using
Softworx imaging software. Line profiles were determined by drawing a line
through the cytoplasm of SG-positive fibroblasts. Pearson correlation coefficients
(PCCs) were determined for a defined region of the cytoplasm of SG-positive
cells. Between five and ten measurements were performed for each of these
analyses, and representative data are shown.
Mammalian expression vectors. Portions (5 ?g) of plasmid DNA (pSR-HA-
hSMG-1, pSR-HA-hSMG-1-DA, or pSR-HA-Upf1-4SA) (24, 59) were trans-
fected into the indicated cells in a six-well plate using Lipofectamine 2000
(Invitrogen) according to the manufacturer’s instructions. The following day,
transfected cells were replated onto coverslips in a 24-well plate and allowed to
adhere for 6 to 8 h prior to treatment with 6 mM sodium butyrate for ca. 16 h
prior to stress treatment. At approximately 48 h posttransfection, the cells were
either left untreated or subjected to various treatments and then rinsed once with
PBS before being fixed in 4% (wt/vol) paraformaldehyde for 10 min. The cells
were then permeabilized in 0.5% (vol/vol) Triton X-100 in PHEM buffer (60 mM
PIPES, 25 mM HEPES, 10 mM EGTA, 2 mM MgCl2[pH 6.9]) for 10 min. After
a brief rinse in PBS, the cells were blocked in 5% (vol/vol) normal horse serum
plus PBS or fetal calf serum plus PBS for 1 h. The remainder of the staining
protocol was performed as described above.
siRNA knockdown. hSMG-1 and Upf1 stealth siRNA duplex oligoribonucle-
otides were obtained from Invitrogen (hSMG-1-HSS118096-8; Upf1 HSS109171-
3). The duplex oligoribonucleotides were resuspended to a concentration of 20
?M. NFF were grown to ca. 70% confluence prior to use. The volumes of
reagents listed are per well for a six-well plate. siRNA (120 pmol) was added to
final volume of 400 ?l of Opti-MEM (Invitrogen) at 37°C, followed by mixing.
Concurrently, 20 ?l of Lipofectamine 2000 (Invitrogen) was added to 400 ?l of
Opti-MEM, followed by mixing. Each mixture was incubated at room tempera-
ture for 5 min, after which the two mixtures were combined and incubated at
room temperature for a further 20 min. While the mixture was incubating, each
well was washed twice with serum-free medium. After the final wash, 400 ?l of
Opti-MEM and 400 ?l of transfection mix were added to the well. The plate was
returned to the incubator for 6 h, followed by the addition of 2 ml of medium. At
approximately 48 h (hSMG-1) or 64 h (Upf1) after the treatment had com-
menced, the cells were used for assays.
Statistical analysis. Statistical analysis was performed using GraphPad Prism
5 software. The samples were analyzed using a paired two-tailed t test. Statisti-
cally significant differences are marked with asterisks in the figures (*, P ? 0.05;
**, P ? 0.01).
Sample preparation for and analysis by MS. Samples were processed as
described previously (35). Briefly, samples were separated by SDS-PAGE, incu-
bated in fixing solution (40% ethanol, 10% acetic acid, 50% H2O), and re-
4418BROWN ET AL.MOL. CELL. BIOL.
buffered in sensitizing solution (30% ethanol, 6.8% [wt/vol] sodium acetate, 0.5%
[wt/vol] sodium thiosulfate), followed by washing. The gel was soaked in silver
solution (0.25% [wt/vol] silver nitrate, 0.015% formaldehyde) and briefly washed
with developing solution (2.5% [wt/vol] sodium carbonate, 0.0074% formalde-
hyde). The reaction was terminated by the addition of stop solution (1.46%
[wt/vol] EDTA). The bands were excised, and the silver stain was removed prior
to tryptic digestion using an adapted form of the method of Gharahdaghi et al.
(20); the method for tryptic digestion was modified from that of Shevchenko et
al. (50). The pellet returned from tryptic digest was resuspended in 5 ?l of 50%
acetonitrile and 0.1% trifluoroacetic acid. This was then used for mass spectro-
photometric (MS) analysis.
For protein identification, peptides were analyzed using a Microflex MALDI-
TOF-PSD (Bruker Daltonics, Bremen, Germany) operated in positive-ion re-
flectron mode. MS data were acquired using 350 shots of a nitrogen laser at 355
nm with a 20-Hz repetition rate and various intensities. The MS data were
calibrated via close external calibration using peptide standards (New England
Biolabs) containing angiotensin I (MH?1,296.69), neurotensin (MH?1,672.92),
ACTH (1 to 17 clip, MH?2,093.09), ACTH (18 to 39 clip, MH?2,465.20), and
ACTH (7 to 38 clip, MH?3,657.93). The Mascot search engine and the Homo
sapiens taxonomic subset of the NCBI nonredundant database were used to
identify the MS data. Mass tolerance was set at 150 ppm. Searches took into
account carbamidomethylated cysteine and oxidized methionine. For the pur-
poses of protein identification, no other posttranslational modifications were
Cell cycle analysis. After indicated treatments, the cells were incubated with
propidium iodide (50 ?g/ml; Sigma) and RNase A (100 ?g/ml; Roche) for 30 min
before cell cycle profiles were quantified using a FACScan (BD Biosciences).
Analysis was performed using ModFit LT (BectonDickinson).
Western blotting. Protein samples analyzed on either hand-poured 4.2% poly-
acrylamide gels or on 4 to 12% Bis/Tris gradient gels (Invitrogen). The proteins
were then transferred onto nitrocellulose membrane, and Western blotting was
performed as described previously (47).
hSMG-1 characterization. We developed three different an-
tibodies that recognize hSMG-1 with a view to investigating
hSMG-1 function in the stress response. The regions of
hSMG-1 recognized by the three antibodies are shown in Fig.
1A. Immunoblot analysis revealed that the antibodies recog-
nized a protein of the expected size for hSMG-1 in extracts
from NFF and LCLs. For all three antibodies, a single high-
molecular-weight band was observed (Fig. 1B), though follow-
ing further separation multiple isoforms of hSMG-1 could be
detected (see Fig. 5A; also data not shown). The bands were
confirmed to be hSMG-1 using immunoprecipitation, followed
by MS; of the 35 peptides identified from tryptic digests, 25
matched hSMG-1, resulting in 7% protein coverage (data not
shown). The specificity of the antibodies was further confirmed
by using siRNA knockdown of hSMG-1 that resulted in a loss
of signal for both immunoblotting and immunostaining (see
Fig. 5A and B). Cellular fractionation followed by immuno-
blotting confirmed that hSMG-1 is present in both the cyto-
plasm and the nucleus (Fig. 1C), as demonstrated previously
(7, 24). Detection of ATM only in the nuclear fraction dem-
onstrated that nuclei were intact. Immunostaining of primary
cells also showed the presence of hSMG-1 in both subcellular
compartments (Fig. 1D, untreated). In response to genotoxic
stress, other members of the PIKK family, including ATM and
ATR (ATM and Rad3 related), localize to nuclear foci at sites
of DNA damage (1, 17). Exposure of primary fibroblasts to IR
gave rise to ?H2AX foci at sites of DNA damage, but there was
no evidence that hSMG-1 localized to these sites under these
conditions (Fig. 1D). However, in a small percentage of cells
(up to 10%) hSMG-1 localized to cytoplasmic granules in re-
sponse to treatment with the DNA-damaging agent hydrogen
peroxide (H2O2) (Fig. 1E). These granules were detected by all
three anti-hSMG-1 antibodies after H2O2treatment (Fig. 1F).
hSMG-1 localization to SG. In fibroblasts, three major types
of cytoplasmic granules have been described: vaults, processing
bodies (P bodies) and SG. These structures are involved in
regulation of translation, mRNA degradation and in stabiliza-
tion and intracellular transport of mRNA (31–33, 49). Of these
structures, only SG are significantly induced by stress-inducing
agents (31). SG form in response to protein translation inhi-
bition and contain the 48S complex and mRNA bound to
T-cell internal antigen 1 (TIA-1) and many other proteins (19,
32). These granules contain proteins that either promote
mRNA stability or lead to destabilization and degradation of
specific mRNA (2). hSMG-1-positive granules, induced by
treatment with H2O2, showed colocalization with TIA-1, indi-
cating that these were indeed SG (Fig. 2A). SG were not
detected with preimmune serum, copurified anti-GST antibody
or in the absence of primary antibody (data not shown). A
commonly used agent to induce SG is sodium arsenite (NaAs)
(30). NaAs causes oxidative stress in cells, increases eIF2?
phosphorylation, and causes translational arrest (2, 37). Expo-
sure of cells to NaAs gave rise to SG, as determined by eIF4G
and hSMG-1 staining (Fig. 2B). Heat was also investigated as
an SG-inducing agent since it induces SG containing unique
SG components (14, 29). Heat treatment induced hSMG-1-
positive SG (Fig. 2B). SG induction in response to NaAs and
heat occurred in at least 70% of cells (see Fig. 5C). Further
confirmation of granule induction was carried out with over-
expressed HA-tagged hSMG-1. Due to the difficulty in trans-
fecting primary cell lines, transient transfection of HeLa cells
with HA-hSMG-1 was performed. SG detected by anti-eIF4G
antibody revealed colocalization with HA-hSMG-1 in SG after
heat shock and H2O2treatment (Fig. 2C). To confirm that the
hSMG-1-positive structures were SG, we used another SG
marker, G3BP1 (52). G3BP1 and hSMG-1 colocalized after
treatment with heat (Fig. 2D). The number of SG per cell and
the number of SG-positive cells varied (particularly with
H2O2), and these granules showed a great deal of heterogene-
ity in size (data not shown), findings consistent with previous
observations (27, 31).
The universality of SG induction in other primary cells was
also investigated. Figure 3A shows merged images of hSMG-1
(red) and TIA-1 (green) staining for SG induced in kidney
proximal tubular cells, melanocytes, and human undifferenti-
ated keratinocytes. With both H2O2and heat treatment,
hSMG-1 and TIA-1 colocalized for all of the primary cells. In
contrast, the carcinoma cell lines U2OS and A549 were unable
to form hSMG-1 staining granules but formed TIA-1-positive
granules after treatment with NaAs (Fig. 3B).
Since hSMG-1 has an established role in NMD, we investi-
gated the possibility that its localization to SG was connected
with NMD and/or mRNA decay. Recruitment of hSMG-1 in
this process is generally mediated by Upf2 and Upf3b stably
associated with the EJC (24). Therefore, we determined
whether Upf2 was localized to SG. Upf2 localized with TIA-1
to SG in response to heat (Fig. 4A). Another key member of
this mRNA surveillance complex, Upf1, was also present in SG
induced by H2O2, heat, and NaAs, as shown by colocalization
with eIF4G or hSMG-1 (Fig. 4B and C). This colocalization
was quantified, yielding a PCC of 0.7832, where a PCC value
VOL. 31, 2011 ROLE FOR hSMG-1 IN STRESS GRANULE FORMATION4419
between 0.5 and 1 is considered a positive association. A line
profile of fluorescence in the red and green channels also
confirmed the overlapping signals (Fig. 4D). A key step in the
targeting of NMD is phosphorylation of Upf1 at four specific
sites by hSMG-1 (S1073, S1078, S1096, and S1116), which
appears to be critical for recognition of the PTC (24). A Upf1
phospho-specific antibody recognizing pS1078 and p1096
failed to detect phosphorylation in SG, suggesting that Upf1
phosphorylation is not required for SG formation or that it
occurs at different sites (Fig. 4E). The localization of a form of
Upf1 (HA4SAUpf1), nonphosphorylatable at all four sites,
confirms the findings obtained with the P-Upf1 antibody (Fig.
4F). These observations are in keeping with a report by Gard-
ner (16), which showed that, in response to hypoxia and arse-
nic, Upf1 localized to SG and NMD was inhibited (16). How-
ever, hSMG-1 mediated NMD-dependent phosphorylation at
the remaining 23 predicted PIKK phosphorylation sites
not be ruled out. To determine whether hSMG-1 could be
regulating another form of mRNA decay, we determined
FIG. 1. hSMG-1 is cytoplasmic and nuclear and localizes to cytoplasmic granules in response to stress. (A) Schematic of the regions recognized
by of hSMG-1 antibodies. GST fusion proteins were prepared using the regions indicated (region 1, Ab1; region 2, Ab2; region 3, Ab3) and used
to generate polyclonal antibodies in sheep. (B) Specificity of detection of hSMG-1 by immunoblotting. Extracts were prepared from LCLs and
NFF, and proteins were separated by SDS-PAGE prior to immunoblotting with the three different hSMG-1 antibodies. (C) hSMG-1 is present both
in the cytoplasm and in the nucleus. Immunoblot analysis was performed for hSMG-1 in nuclear and cytoplasmic extracts from primary normal
foreskin fibroblasts (NFF). ATM was used as a control to demonstrate fractionation. (D) hSMG-1 does not localize to sites of DNA damage. DNA
damage induced nuclear foci in response to 10 Gy of IR. NFF were fixed and stained with anti-?H2AX (red) and anti-hSMG-1 (Ab2, green)
antibodies. Nuclei were detected with DAPI. (E) Detection of cytoplasmic granules in NFF in response to H2O2. The cells were incubated for 1 h
after exposure, fixed, and stained with hSMG-1 (Ab2) antibody. (F) Immunofluorescent detection of hSMG-1 with three different antibodies in
NFF. hSMG-1 antibodies (red) directed to different regions of hSMG-1 show cytoplasmic granule formation in response to heat.
4420 BROWN ET AL.MOL. CELL. BIOL.
whether hSMG-1 localized to P bodies. We used a construct
encoding the P-body-specific marker decapping protein 1
(DCP1), which is required for mRNA decapping and degra-
dation (49). No significant colocalization between hSMG-1-
and DCP-1-positive granules was observed after treatment of
fibroblasts with H2O2, providing further confirmatory evidence
that hSMG-1 was localized specifically to SG under these con-
ditions (Fig. 4G).
hSMG-1 is required for NaAs-induced SG formation or
maintenance. To further establish the importance of hSMG-1
in SG formation, three different siRNA duplexes were used to
reduce cellular levels of hSMG-1. In these experiments we
focused on heat and NaAs as inducing agents because they
reproducibly induced SG in a high percentage of cells. Figure
5A shows that all siRNA sequences significantly reduced but
did not ablate hSMG-1 protein expression. After siRNA treat-
ment of fibroblasts, SG formation was decreased after NaAs
treatment but not after exposure to heat (Fig. 5B). The per-
centage of cells containing eIF4G-positive SG was the same
after heat treatment but was significantly reduced in NaAs-
treated cells (Fig. 5C). siRNA knockdown of hSMG-1 also
appeared to decrease SG formation in response to H2O2(data
not shown), although the very low and variable percentage of
cells forming SG in response to H2O2meant these data were
not statistically significant. These data point to an important
role for hSMG-1 in the formation or stability of a subgroup of
SG, although more complete inhibition of SG formation may
occur if total hSMG-1 knockdown was achieved. This experi-
ment also demonstrates that heat-induced SG formation is
hSMG-1 independent. To further examine the role NMD
might play in SG regulation, we knocked down Upf1 expres-
sion with siRNA (Fig. 5D). As determined by Western blot-
ting, the protein level of Upf1 was decreased by two indepen-
dent siRNA sequences. The Upf1 siRNA treatment did not
affect the protein levels of either hSMG-1 or TIA-1 (Fig. 5D).
siRNA treatment in this experiment did not completely knock
down Upf1 expression; however, if the amount of siRNA used
or the length of incubation was increased, fibroblast cell via-
bility was severely compromised (data not shown). Therefore,
we used these conditions to examine the role of Upf1 in SG
formation. After siRNA knockdown of Upf1, the fibroblasts
were treated with either NaAs or heat for 1 h to induce SG
formation. The cells were then immunostained for Upf1 and
TIA-1, and SG formation was quantified. Knockdown of Upf1
showed a small decrease in SG formation in response to both
NaAs and heat, although this was not statistically significant
(data not shown). However, in many cells treated with anti-
Upf1 siRNA, sufficient Upf1 protein was present to see it
recruited to SG (Fig. 5E). Decreased SG formation after Upf1
knockdown was not due to inhibition of hSMG-1 recruitment
to SG since hSMG-1 clearly localized in SG even after Upf1
depletion (Fig. 5F). Therefore, whether Upf1 plays a direct
role in SG formation or whether depletion of Upf1 generally
suppresses cellular responses is currently unclear.
FIG. 2. hSMG-1 localizes to stress granules. (A) hSMG-1 (detected by Ab1, red) colocalizes with the SG marker TIA-1 (green) in response to
H2O2. NFF were stained after H2O2treatment (1.5 mM) for 1 h. (B) Colocalization of hSMG-1 (Ab2, red) with a second SG marker eIF4G (green)
after treatment of NFF with heat (45°C) or 1 mM NaAs for 1 h. DAPI (blue) was used to stain nuclei. Colocalization (yellow) is indicated in a
merged panel on the right. (C) An HA-tagged form of hSMG-1 localizes to SG. HeLa cells were transiently transfected with HA-hSMG-1 and
incubated for 48 h in fresh medium prior to exposure to heat or H2O2, followed by staining with anti-HA or anti-eIF4G antibodies. (D) hSMG-1
(Ab1) (red) colocalizes with G3BP1 (green) in SG formed in NFF in response to heat. Hoechst was used to stain the nuclei.
VOL. 31, 2011 ROLE FOR hSMG-1 IN STRESS GRANULE FORMATION4421
Role of hSMG-1 kinase activity in SG formation. hSMG-1
may be required for SG formation for either of two reasons: (i)
hSMG-1 kinase activity may be required for formation or sta-
bility of SG or, (ii) alternatively, the hSMG-1 protein may
physically be required. In common with other members of the
PIKK family, hSMG-1 phosphorylates substrates at (S/T)Q
motifs in response to DNA damage (7, 59). Since there is no
specific hSMG-1 inhibitor, we investigated the role of hSMG-1
kinase activity in SG formation using the PIKK inhibitor wort-
mannin (48). Wortmannin exposure for 2 h prior to stress
treatment blocked SG formation after treatment with NaAs
but did not inhibit SG formation induced by heat treatment
(Fig. 6A). This further supported the hSMG-1-independent
formation of heat-induced SG previously observed (Fig. 5B
and C). Wortmannin appeared to also inhibit SG formation in
the small number of cells that responded to H2O2(data not
shown). To investigate whether inhibition of other PIKK fam-
ily members may have resulted in the observed decrease in SG
formation, specific inhibitors of the related PIKK family mem-
bers were used. ATM (Ku55933) and DNA-PK (AMA-37)
FIG. 3. hSMG-1 plays a central role in SG formation in primary cells but does not localize to SG in tumor cells. (A) Appearance of SG in
primary cells in response to H2O2or heat treatment. The images represent merged images between hSMG-1 (Ab1) (red) and TIA-1 (green). Nuclei
were visualized with Hoechst stain. Abbreviations: PTC, human proximal tubular cells; MEL, primary melanocytes; HEK, undifferentiated
keratinocytes. (B) A549 and U2OS cells were treated with NaAs for 1 h and stained for hSMG-1 (Ab1) (red) and TIA-1 (green). TIA-1-positive
but hSMG-1-negative SG were formed in response to NaAs.
4422 BROWN ET AL.MOL. CELL. BIOL.
FIG. 4. Localization of NMD proteins to SG but lack of evidence for a role for hSMG-1 in mRNA decay. (A) Upf2 colocalizes with TIA-1 in
SG. NFF were fixed with or without heat treatment and stained for TIA-1 (green) and Upf2 (red). Colocalization (yellow) following heat treatment
is shown in the merged panel on the right. (B) Upf1 also localizes to SG. HeLa cells were transiently transfected with HA-Upf1. After 48 h, the
cells were treated with H2O2and NaAs and then stained with anti-HA (green) and anti-eIF4G (red) antibodies. (C) Endogenous Upf1 localizes
to hSMG-1. NFF were treated with NaAs or heat and after 1 h stained for Upf1 (green) and hSMG-1 (Ab2, red). (D) Upf1 colocalizes with
hSMG-1. The line profile shows overlapping fluorescence signals for Upf1 (green) and hSMG-1 (red). Image analysis was performed using the
Softworx computer program. Line profiles were determined by drawing a line through the cytoplasm of SG positive fibroblasts. (E) Upf1 is not
phosphorylated in heat- and NaAs-induced SG. NFF were treated with NaAs or heat and, after 1 h, stained with phospho-specific antibodies
recognizing sites phosphorylated on Upf1 during NMD. P-Upf1 (green) was not detected in hSMG-1 (Ab2, red) positive SG but could be clearly
seen in the nucleus. (F) Localization of a nonphosphorylatable form of Upf1, HA4SAUpf1, to SG. NFF were transiently transfected with
HA4SAUpf1 and incubated for 48 h in fresh medium prior to exposure to H2O2and heat, followed by staining with anti-HA and anti-Upf1
antibodies. (G) hSMG-1 (Ab1) does not localize to P bodies. NFF were transiently transfected with the P-body-specific marker DCP1-YFP and
incubated for 48 h prior to treatment with H2O2for 1 h and followed by the detection of YFP and staining with antibodies against hSMG-1.
VOL. 31, 2011 ROLE FOR hSMG-1 IN STRESS GRANULE FORMATION4423
inhibitors did not significantly inhibit SG formation in response
to either NaAs or heat treatment (Fig. 6A). We also checked
for the involvement of another member of the PIKK family
mammalian target of rapamycin (mTOR) that controls cell
growth and survival (51). Inhibition of mTOR by rapamycin
also failed to interfere with heat- or NaAs-induced SG (Fig.
6A). All inhibitors were shown to be active under these con-
ditions, as evidenced by the inhibition of radiation-induced
FIG. 5. Effect of hSMG-1 or Upf1 disruption on SG formation. (A) Knockdown of hSMG-1 in NFF by three different siRNA oligonucleotides.
hSMG-1 protein was detected by immunoblotting (Ab3), and the loading control was DNA-PKcs. (B) At 48 h after siRNA treatment, the NFF were
treated with NaAs or heat for 1 h and then stained for hSMG-1 (Ab2, red) and eIF4G (green). Colocalization appears as yellow. Nuclei were
detected with DAPI (blue). After siRNA treatment, the numbers of eIF4G SG-positive cells were reduced in NFF that had been treated with
NaAs but not in heat-treated NFF. Control siRNA had no affect on SG formation in response to either agent. (C) Quantification of SG
formation following treatment with two different anti-hSMG-1 or control siRNA. The percentage of NFF with eIF4G-positive granules was
scored. The data represent the averages of three independent experiments, and error bars show the standard errors of the mean. Asterisks
denote statistically significant differences in SG formation (*, P ? 0.05;**, P ? 0.01). (D) Knockdown of Upf1 expression in NFF using two
different siRNA. Western blotting confirmed knockdown of Upf1 protein level but no affect on hSMG-1 or TIA-1 expression levels. GAPDH
was probed for as a loading control. An irrelevant lane has been removed from the image, but all samples were run on the same gel. (E) Upf1
knockdown does not block stress granule formation. Approximately 64 h after anti-Upf1 siRNA treatment SG were induced with either NaAs
or heat for 1 h. Fibroblasts were stained for TIA-1 (green) and Upf1 (red). SG formation could be seen in all treated samples. The data are
representative of three independent experiments. (F) Cells were treated as for panel E but stained for hSMG-1 (green) and Upf1 (red).
Nuclei were visualized with DAPI (blue). hSMG-1 was still recruited to SG following Upf1 knockdown. The data are representative of three
4424 BROWN ET AL.MOL. CELL. BIOL.
ATM autophosphorylation measured by S1981 phosphoryla-
tion, inhibition of DNA-PKcsautophosphorylation on S2056,
and rapamycin inhibition of mTOR, which blocked the move-
ment of cells from G1into S phase (Fig. 6B to D). Quantifi-
cation of the immunofluorescent images showed far less colo-
calization after wortmannin treatment than after treatment
with any other inhibitor (a PCC of 0.8082 for DMSO versus a
PCC of 0.4798 for wortmannin) (Fig. 6E). These results sug-
gest that PIKK activity is essential for SG formation in re-
sponse to NaAs but not to heat.
We looked for the presence of potential PIKK substrates in
SG by staining with a phospho-specific antibody against p(S/
T)Q sites. In response to NaAs and heat treatment, speckles of
p(S/T)Q staining could be observed in the cytoplasm of cells
(Fig. 7A). These sites were not completely colocalized with
hSMG-1 but were often overlapping or associated with hSMG-
1-positive granules (PCC of 0.4497; see also the left panel of
Fig. 7B). In contrast, SG formed in response to heat were
strongly hSMG-1 positive, but no phosphorylated (S/T)Q sites
were detected (Fig. 7A, lower panel). In response to H2O2, a
much stronger phosphorylated (S/T)Q signal, more completely
colocalizing with hSMG-1, was detected (PCC of 0.6988) (Fig.
7A, right panel of Fig. 7B). These data imply that PIKK activity
may be important in the regulation or stability of NaAs- and
H2O2-induced SG but not heat-induced SG.
To further investigate the specific requirement of hSMG-1
FIG. 6. Role for PIKK kinase activity in SG formation. (A) Wortmannin inhibits SG formation in response to NaAs treatment but not heat
treatment. Cells were NaAs or heat treated after 2 h of incubation with the indicated inhibitor. NFF were then stained with anti-hSMG-1 (Ab2,
red) and anti-eIF4G (green) antibodies, with the colocalization appearing as yellow. Nuclei were detected with DAPI (blue). Inhibition of ATM
(Ku55933), DNA-PKcs(AMA37), or mTOR (rapamycin) did not interfere with formation of SG in response to either treatment. The data are
representative of two to four independent experiments. (B) Effect of ATM-specific inhibitor on ATM autophosphorylation at S1981. NFF were
incubated with Ku55933 for 2 h prior to 10 Gy of IR and then incubated for 30 min or 1 h prior to the preparation of cell extracts. Immunoblotting
was carried out with a phospho-specific antibody for ATM S1981. Nbs1 protein was used as a loading control. (C) Effect of different inhibitors on
DNA-PK autophosphorylation at S2056. NFF were incubated with either Ku55933 (ATM inhibitor), wortmannin (general PIKK inhibitor), or
AMA37 (DNA-PKcsinhibitor) for 2 h prior to exposure to 10 Gy of IR. Extracts were prepared and immunoblotted either with the phospho-
specific antibody (P2056-DNA-PK) or an anti-DNA-PKcsantibody. Total DNA-PKcsdetection was used as a loading control. (D) Rapamycin
prevents passage of NFF from G1to S phase. Cells were exposed to concentrations of rapamycin from 20 to 100 ?M prior to analysis by flow
cytometry using propidium iodide (PI) staining. (E) PCCs for cells treated with PIKK inhibitors prior to SG formation. These coefficients were
determined using Softworx software for a defined region of the cytoplasm of fibroblasts. The data show the averages of at least five measurements
from different cells from at least two independent experiments and the standard deviations of the measurements.
VOL. 31, 2011 ROLE FOR hSMG-1 IN STRESS GRANULE FORMATION4425
kinase activity in SG formation in response to NaAs and H2O2,
a kinase-deficient mutant of hSMG-1 (HA-hSMG-1-DA) was
transiently transfected into HeLa cells. Kinase-dead hSMG-1
localized as efficiently as wild-type hSMG-1 to SG induced by
H2O2or NaAs (Fig. 7C). In SG containing wild-type HA-
SMG-1, phosphorylated (S/T)Q signal was observed in re-
sponse to both H2O2and NaAs (Fig. 7C). In SG containing the
kinase-deficient HA-hSMG-1-DA, phosphorylated (S/T)Q
sites were undetectable in response to H2O2treatment (Fig.
7C), suggesting that hSMG-1 is responsible for the observed
phosphorylation. However, phosphorylated (S/T)Q sites were
still present in NaAs-induced SG containing the kinase-dead
hSMG-1, indicating that another member of the PIKK family
is capable of phosphorylation of these sites. Collectively, the
experiments described here show that hSMG-1 is recruited to
SG and its presence, but not kinase activity, is essential for
NaAs- and H2O2-induced, but not heat-induced, SG forma-
hSMG-1 has been shown to play several distinct roles in the
cellular response to stress, including its involvement in NMD
(59), in the protection of cells against TNF-?- or granzyme
B-induced apoptosis (41, 46), and in activating the G1/S check-
point (18). More recently Chen et al. (10) found that hSMG-1
inhibited HIF-1? transactivation activity in part by suppressing
MAP kinase ERK1 in hypoxia. These results suggested that
hypoxic conditions were efficient in activating hSMG-1 to re-
strain hypoxia-induced malignancy. Furthermore, mutations of
the hSMG-1 kinase domain have been observed in breast can-
cer, lung adenocarcinoma, and kidney and stomach cancer (11,
short, SMG-1 has known roles in NMD and genome mainte-
nance and is implicated in regulation of oxidative stress, apop-
tosis and hypoxia responses. We have identified here a novel
role for hSMG-1 in the formation of SG.
FIG. 7. Detection of phosphorylated proteins in SG and importance of protein kinase activity of hSMG-1. (A) NaAs and H2O2treatment of
NFF induced the appearance of proteins phosphorylated at (S/T)Q sites, and these sites partially colocalize in SG with hSMG-1. Heat induced SG
formation (hSMG-1 positive) without detectable phosphorylation at (S/T)Q sites. Staining in the upper panel used an older batch of rabbit
anti-P(S/T)Q antibody: P(S/T)Q (green), hSMG-1 (Ab2, red), and DAPI (blue). In the lower panel, a newer batch of rabbit anti-P(S/T)Q antibody
was used: P(S/T)Q (red), hSMG-1 (Ab2, green), and DAPI (blue). (B) P(S/T)Q signals associate with hSMG-1 after SG induction. The left panel
shows a line profile indicating regions of overlapping fluorescence signals for P(S/T)Q (green) and hSMG-1 (red) following NaAs treatment. The
right-hand panel shows greater overlap of the P(S/T)Q (green) signal with hSMG-1(red) after H2O2treatment. Image analysis was performed using
the Softworx computer program. Line profiles were determined by drawing a line through the cytoplasm of SG-positive fibroblasts. (C) Overex-
pression of kinase-deficient hSMG-1 does not inhibit SG formation. HeLa cells were transfected with either HA-hSMG-1 or HA-hSMG-1-DA
(kinase deficient). The cells were treated with 1 mM H2O2or NaAs for 1 h and then stained with anti-HA (green) and anti-P(S/T)Q (red)
antibodies. In cells transfected with HA-hSMG-1-DA, no P(S/T)Q sites were observed in response to H2O2treatment.
4426 BROWN ET AL.MOL. CELL. BIOL.
SG are related to other mRNA-protein complexes, including
P bodies, germ cell granules, and neuronal RNA granules (8),
all of which contain a variety of RNA-binding proteins, trans-
lation factors, and RNA decay machinery that can change
depending on exposure to cellular stress (31). However, P
bodies and SG are the only types present in fibroblasts. Typi-
cally, SG contain the 48S preinitiation complex composed of
the small ribosomal subunit and mRNA, together with a num-
ber of translation initiation factors, poly(A)-binding protein,
proteins that regulate mRNA translation, and proteins in-
volved in cell signaling pathways (2, 44). SG are commonly
described as structures that sequester RNA during times of
cellular stress either to promote degradation, to stabilize
mRNA for rapid translation once the stress has abated, or to
promote translation of specific mRNA during stress (2). Nas-
cent mRNA transcripts can also be exported from the nucleus
and targeted directly to SG (8). This finding shows that while
SG form in response to translational arrest, not all transcripts
in SG are associated with stalled ribosomal complexes. More
recently, alternative functions for SG have been described. SG
have been shown to harbor proteins involved in the regulation
of apoptosis such as TRAF2, RACK, and FAST (5, 34, 36).
These data suggest that SG act to limit apoptosis, while cells
adapt to stress, since inhibition of SG formation during stress
can lead to decreased cell survival (8). Furthermore, SG have
been implicated in cellular responses to viral infection (8),
although the role SG play here is unclear since many viruses
disrupt or induce SG formation to benefit their own life cycle.
Previous work has shown that in mammalian cells NMD fac-
tors, including Upf1 and NMD mRNA targets, can traffic
through P bodies (12) and localize to SG (16) when NMD is
inhibited by specific inhibitors or in response to hypoxia.
We showed here that hSMG-1 is recruited to SG in response
to a range of cellular stresses and that knockdown of hSMG-1
strongly reduced SG formation in response to NaAs but not to
heat. Although use of the PIKK inhibitor wortmannin pre-
vented SG formation in response to NaAs, overexpression of a
kinase-dead version of hSMG-1 did not. These data suggest
that the role of hSMG-1 in stress responses is likely to be
dependent on the type of cellular stress encountered. There
are potentially three facets to the involvement of hSMG-1 in
SG: (i) the mechanism of hSMG-1 recruitment to SG, (ii) the
requirement for hSMG-1 as a protein facilitating SG formation
following certain stresses, and (iii) the role of PIKK, including
hSMG-1, kinase activity in either SG formation, function, or
Mechanism of hSMG-1 recruitment to SG. Stress-induced
signaling leading to phosphorylation of eIF2? and SG forma-
tion can be initiated by at least four different kinases (GCN2,
PERK, HRI, and PKR) (22, 54). These kinases function in
conjunction with other signaling pathways to coordinate the
cellular response to a specific stress. Some of these pathways,
converging on eIF2?, may have additional parallel effects fa-
cilitating recruitment of hSMG-1 to SG. The recruitment of
hSMG-1 to SG in response to all stresses tested may be linked
to inhibition of NMD. During the response to hypoxia Upf1
localized to SG and under the same conditions NMD was
inhibited, although a causative link between these phenomena
was not established (16). Furthermore, a very recent study
showed that a variety of cellular stresses (including the SG-
inducing agent NaAs) resulted in the phosphorylation of eIF2?
and inhibition of NMD (55). Although in neither study was a
causative link between SG formation and NMD inhibition
established. In addition, a hyperphosphorylated form of Upf1
accumulated in P bodies in response to chemical inhibition of
NMD, which blocks NMD at a step following Upf1 phosphor-
ylation (12). In the present study, we observed the recruitment
of hSMG-1, Upf1, and Upf2 to SG, but Upf1 was not phos-
phorylated at known NMD sites under these conditions, indi-
cating that active NMD was not occurring within SG. This
result is supported by the recent finding (55) that NMD was
inactivated after treatment with the SG-inducing agent NaAs
used here. Since Upf1 detected in P bodies after NMD inhi-
bition was hyperphosphorylated, the unphosphorylated form
we detected in SG may have an NMD independent role, or it
may represent a different form of NMD inhibition more similar
to that observed during hypoxia (12, 16). It is likely that under
all of the stresses examined here, NMD is inhibited (55).
Therefore, the recruitment of hSMG-1 and Upf1 to SG, with-
out being essential for SG formation, may provide the basis of
a mechanism for NMD inhibition in response to the phosphor-
ylation of eIF2? (55). Alternatively, hSMG-1 may be required
to traffic or process specific transcripts in SG under stress
conditions such as heat shock, where it is not essential for SG
Requirement of hSMG-1 as a protein facilitating SG forma-
tion following certain stresses. hSMG-1 clearly plays a role in
the formation or stability of SG after treatment with NaAs or
H2O2but not heat treatment. Knockdown of hSMG-1 with
siRNA reduced the formation of SG in response to these
agents, but a kinase-dead version of hSMG-1 also strongly
localized to SG. This construct has previously been shown to
act in a dominant-negative manner (24). Therefore, the data
suggest that the presence of the hSMG-1 protein, but not its
kinase activity, is required for SG formation in response to
some stresses. This role for hSMG-1 has not been described
before. Using an RNA-mediated interference-based screen,
Ohn et al. (45) identified 100 human genes required for SG
assembly. hSMG-1 was not identified in that screen, which may
be due to differences in the cell lines used or that knockdown
of hSMG-1 induces apoptosis (7), removing these cells from
further analysis. To determine whether the requirement for
hSMG-1 in SG formation was related to NMD, we attempted
to define the role of the hSMG-1 substrate Upf1 in response to
NaAs using siRNA knockdown. However, substantial Upf1
depletion decreased cell viability. This is not surprising since
Upf1 has previously been shown to be required for embryonic
development (40) and another NMD component Upf2 was
shown to be essential for the viability of hematopoietic stem
cells (56). Consequently, the decrease in SG formation ob-
served with Upf1 knockdown can be interpreted in two ways:
(i) Upf1 is required for SG formation in response to both
NaAs and heat, but sufficient Upf1 depletion could not be
achieved in order to see a dramatic effect, or (ii) Upf1 is not
required for SG formation, and the small decrease observed
here may be due to decreased viability or functionality of the
cells. At this stage, it is still unclear what role Upf1 plays in SG
formation. Combined, these data suggest that hSMG-1 is re-
quired for SG formation in response to NaAs in a manner that
may be related to its role in NMD, but its retention in SG is
VOL. 31, 2011ROLE FOR hSMG-1 IN STRESS GRANULE FORMATION 4427
independent of its kinase activity. Known roles for hSMG-1 are
associated with its kinase activity. An understanding of this
novel role for hSMG-1 will require further biochemical anal-
ysis of the protein and its stress induced interaction partners
and/or isolation of different classes of SG to see what is unique
about SG where hSMG-1 is required for their formation. Cur-
rently, isolation of SG in this manner is not possible. The
differential requirement of hSMG-1 for SG formation may
reflect the ability of cells to survive after insult by different
stresses or reflect different mechanisms of translation or NMD
inhibition induced (8).
Role of PIKK activity in SG formation, function, or disas-
sembly. The final aspect of hSMG-1 function in SG is the
potential role of PIKK activity in the cellular response to
stress. Phosphorylation of PIKK target (S/T)Q sites was de-
tected in or associated with SG in response to NaAs and H2O2
treatment. Wortmannin inhibition blocked formation of SG in
response to these stresses, indicating that PIKK activity is
essential for SG formation under these conditions. Interest-
ingly, overexpression of a kinase-dead version of hSMG-1 in
HeLa cells did not prevent SG formation and, in fact, the
kinase-dead form localized to SG with an efficiency similar to
that of wild-type hSMG-1 in response to both stimuli, although
this may be due to the ability of hSMG-1 to dimerize (42). To
further complicate the picture, H2O2-induced phosphorylation
of (S/T)Q sites was inhibited by kinase-dead hSMG-1, but
NaAs-induced phosphorylation was not. These data suggest
either that hSMG-1 kinase activity is not involved in SG for-
mation or that hSMG-1 kinase activity can be compensated for
by another PIKK family member in this situation. However,
hSMG-1 kinase activity may be involved in SG regulation in
response to H2O2since hSMG-1-dependent phosphorylation
sites were observed in these SG. Other PIKK family members
have been implicated in SG formation. In the RNAi screen
discussed above, the PIKK family member, DNA-PKcswas
detected, indicating that its kinase activity may also be impor-
tant (45). However, in the presence of a specific DNA-PKcs
inhibitor we did not observe interference with SG formation in
response to either NaAs or heat treatment. We also failed to
prevent SG formation using the ATM-specific inhibitor
KU55933 (21) or rapamycin, an inhibitor of mTOR (51), de-
spite wortmannin strongly inhibiting SG formation. At the
concentration used wortmannin inhibition of activity of pro-
teins other that members of the PIKK family should be mini-
mal. Potentially, multiple PIKK family members may be in-
volved in an overlapping or redundant manner in SG
regulation, as they are in the nuclear DNA damage response.
If so, inhibition of more than one would be required to see a
reduction in SG formation in response to NaAs or H2O2.
Interestingly, PIKK family members have also been impli-
cated in NMD-independent RNA degradation involving Upf1.
Histone mRNA stability may also be controlled by DNA-PK-
mediated phosphorylation of Upf1 (43), and ATR may also
phosphorylate Upf1 during histone mRNA degradation (25,
26). hSMG-1 involvement in this process has not been inves-
tigated. How this process may relate to a role for PIKK in SG
regulation will require further investigation.
Overall, we show that hSMG-1 is recruited to SG in re-
sponse to heat, NaAs, and H2O2treatment. Our data suggest
that the physical presence of the hSMG-1 protein is required
for formation of a subset of SG independently of its protein
kinase activity and that protein phosphorylation by PIKKs,
including hSMG-1, may be involved in the regulation and/or
turnover of SG in response to specific stresses.
We thank Paul Anderson and Nancy Kedersha (Harvard Medical
School) for the DCP1-eYFP construct, L. E. Maquat (University of
Rochester Medical Centre) for the Upf1 and Upf2 antibodies and
constructs, and Dianne Watters (Griffith University) for the rapamy-
cin. We thank Glen Boyle (Queensland Institute of Medical Research)
for the primary fibroblasts, Nicholas Saunders (University of Queens-
land) for the primary keratinocytes, C. Percy (University of Queens-
land) for the human kidney proximal tubular cells, Aine Farrell for her
tissue culture expertise, and the Lavin laboratory for stimulating dis-
cussions. We also thank Tracey Laing for typing the manuscript.
We thank the Australian Research Council for funding and the
National Health and Medical Research Council of Australia for a
Peter Doherty Fellowship to T.L.R.
1. Andegeko, Y., et al. 2001. Nuclear retention of ATM at sites of DNA double-
strand breaks. J. Biol. Chem. 276:38224–38230.
2. Anderson, P., and N. Kedersha. 2006. RNA granules. J. Cell Biol. 172:803–
3. Anderson, P., and N. Kedersha. 2009. Stress granules. Curr. Biol. 19:R397–
4. Arias-Palomo, E., et al. 2011. The nonsense-mediated mRNA decay SMG-1
kinase is regulated by large-scale conformational changes controlled by
SMG-8. Genes Dev. 25:153–164.
5. Arimoto, K., H. Fukuda, S. Imajoh-Ohmi, H. Saito, and M. Takekawa. 2008.
Formation of stress granules inhibits apoptosis by suppressing stress-respon-
sive MAPK pathways. Nat. Cell Biol. 10:1324–1332.
6. Azzalin, C. M., P. Reichenbach, L. Khoriauli, E. Giulotto, and J. Lingner.
2007. Telomeric repeat containing RNA and RNA surveillance factors at
mammalian chromosome ends. Science 318:798–801.
7. Brumbaugh, K. M., et al. 2004. The mRNA surveillance protein hSMG-1
functions in genotoxic stress response pathways in mammalian cells. Mol.
8. Buchan, J. R., and R. Parker. 2009. Eukaryotic stress granules: the ins and
outs of translation. Mol. Cell 36:932–941.
9. Chawla, R., and C. M. Azzalin. 2008. The telomeric transcriptome and SMG
proteins at the crossroads. Cytogenet. Genome Res. 122:194–201.
10. Chen, R. Q., et al. 2009. Kinome siRNA screen identifies SMG-1 as a
negative regulator of hypoxia-inducible factor-1? in hypoxia. J. Biol. Chem.
11. Ding, L., et al. 2008. Somatic mutations affect key pathways in lung adeno-
carcinoma. Nature 455:1069–1075.
12. Durand, S., et al. 2007. Inhibition of nonsense-mediated mRNA decay
(NMD) by a new chemical molecule reveals the dynamic of NMD factors in
P-bodies. J. Cell Biol. 178:1145–1160.
13. Eulalio, A., I. Behm-Ansmant, and E. Izaurralde. 2007. P bodies: at the
crossroads of posttranscriptional pathways. Nat. Rev. Mol. Cell. Biol. 8:9–22.
14. Farny, N. G., N. L. Kedersha, and P. A. Silver. 2009. Metazoan stress granule
assembly is mediated by P-eIF2?-dependent and -independent mechanisms.
15. Frangioni, J. V., and B. G. Neel. 1993. Solubilization and purification of
enzymatically active glutathione S-transferase (pGEX) fusion proteins. Anal.
16. Gardner, L. B. 2008. Hypoxic inhibition of nonsense-mediated RNA decay
regulates gene expression and the integrated stress response. Mol. Cell. Biol.
17. Gatei, M., et al. 2001. Ataxia telangiectasia mutated (ATM) kinase and ATM
and Rad3-related kinase mediate phosphorylation of Brca1 at distinct and
overlapping sites: in vivo assessment using phospho-specific antibodies.
J. Biol. Chem. 276:17276–17280.
18. Gehen, S. C., R. J. Staversky, R. A. Bambara, P. C. Keng, and M. A. O’Reilly.
2008. hSMG-1 and ATM sequentially and independently regulate the G1
checkpoint during oxidative stress. Oncogene 27:4065–4074.
19. Gewandter, J. S., R. A. Bambara, and M. A. O’Reilly. 2011. The RNA
surveillance protein SMG1 activates p53 in response to DNA double-strand
breaks but not exogenously oxidized mRNA. Cell Cycle 10:15.
20. Gharahdaghi, F., C. R. Weinberg, D. A. Meagher, B. S. Imai, and S. M.
Mische. 1999. Mass spectrometric identification of proteins from silver-
stained polyacrylamide gel: a method for the removal of silver ions to en-
hance sensitivity. Electrophoresis 20:601–605.
21. Hickson, I., et al. 2004. Identification and characterization of a novel and
4428 BROWN ET AL.MOL. CELL. BIOL.
specific inhibitor of the ataxia-telangiectasia mutated kinase ATM. Cancer
22. Holcik, M., and N. Sonenberg. 2005. Translational control in stress and
apoptosis. Nat. Rev. Mol. Cell. Biol. 6:318–327.
23. Isken, O., and L. E. Maquat. 2008. The multiple lives of NMD factors:
balancing roles in gene and genome regulation. Nat. Rev. Genet. 9:699–712.
24. Kashima, I., et al. 2006. Binding of a novel SMG-1-Upf1-eRF1-eRF3 com-
plex (SURF) to the exon junction complex triggers Upf1 phosphorylation
and nonsense-mediated mRNA decay. Genes Dev. 20:355–367.
25. Kaygun, H., and W. F. Marzluff. 2005. Regulated degradation of replication-
dependent histone mRNAs requires both ATR and Upf1. Nat. Struct. Mol.
26. Kaygun, H., and W. F. Marzluff. 2005. Translation termination is involved in
histone mRNA degradation when DNA replication is inhibited. Mol. Cell.
27. Kedersha, N., and P. Anderson. 2002. Stress granules: sites of mRNA triage
that regulate mRNA stability and translatability. Biochem. Soc. Trans. 30(Pt.
28. Kedersha, N., and P. Anderson. 2009. Regulation of translation by stress
granules and processing bodies. Prog. Mol. Biol. Transl. Sci. 90:155–185.
29. Kedersha, N., S. Chen, N. Gilks, W. Li, I. J. Miller, J. Stahl, and P. Ander-
son. 2002. Evidence that ternary complex (eIF2-GTP-tRNAi
preinitiation complexes are core constituents of mammalian stress granules.
Mol. Biol. Cell 13:195–210.
30. Kedersha, N., et al. 2000. Dynamic shuttling of TIA-1 accompanies the
recruitment of mRNA to mammalian stress granules. J. Cell Biol. 151:1257–
31. Kedersha, N., et al. 2005. Stress granules and processing bodies are dynam-
ically linked sites of mRNP remodeling. J. Cell Biol. 169:871–884.
32. Kedersha, N. L., M. Gupta, W. Li, I. Miller, and P. Anderson. 1999. RNA-
binding proteins TIA-1 and TIAR link the phosphorylation of eIF-2? to the
assembly of mammalian stress granules. J. Cell Biol. 147:1431–1442.
33. Kedersha, N. L., and L. H. Rome. 1986. Isolation and characterization of a
novel ribonucleoprotein particle: large structures contain a single species of
small RNA. J. Cell Biol. 103:699–709.
34. Kim, W. J., S. H. Back, V. Kim, I. Ryu, and S. K. Jang. 2005. Sequestration
of TRAF2 into stress granules interrupts tumor necrosis factor signaling
under stress conditions. Mol. Cell. Biol. 25:2450–2462.
35. Kozlov, S., N. Gueven, K. Keating, J. Ramsay, and M. F. Lavin. 2003. ATP
activates ataxia-telangiectasia mutated (ATM) in vitro: importance of auto-
phosphorylation. J. Biol. Chem. 278:9309–9317.
36. Li, W., M. Simarro, N. Kedersha, and P. Anderson. 2004. FAST is a survival
protein that senses mitochondrial stress and modulates TIA-1-regulated
changes in protein expression. Mol. Cell. Biol. 24:10718–10732.
37. Liu, S. X., et al. 2005. Mitochondrial damage mediates genotoxicity of arse-
nic in mammalian cells. Cancer Res. 65:3236–3242.
38. Masse, I., et al. 2008. A novel role for the SMG-1 kinase in life span and
oxidative stress resistance in Caenorhabditis elegans. PLoS One 3:e3354.
39. McIlwain, D. R., et al. 2010. Smg1 is required for embryogenesis and regu-
lates diverse genes via alternative splicing coupled to nonsense-mediated
mRNA decay. Proc. Natl. Acad. Sci. U. S. A. 107:12186–12191.
40. Medghalchi, S. M., et al. 2001. Rent1, a trans-effector of nonsense-mediated
mRNA decay, is essential for mammalian embryonic viability. Hum. Mol.
41. Meslin, F., et al. 2011. hSMG-1 is a granzyme B-associated stress-responsive
protein kinase. J. Mol. Med. 89:411–421.
42. Morita, T., et al. 2007. Distant N- and C-terminal domains are required for
intrinsic kinase activity of SMG-1, a critical component of nonsense-medi-
ated mRNA decay. J. Biol. Chem. 282:7799–7808.
43. Muller, B., J. Blackburn, C. Feijoo, X. Zhao, and C. Smythe. 2007. DNA-
activated protein kinase functions in a newly observed S phase checkpoint
that links histone mRNA abundance with DNA replication. J. Cell Biol.
44. Nadezhdina, E. S., A. J. Lomakin, A. A. Shpilman, E. M. Chudinova, and
P. A. Ivanov. 2010. Microtubules govern stress granule mobility and dynam-
ics. Biochim. Biophys. Acta 1803:361–371.
45. Ohn, T., N. Kedersha, T. Hickman, S. Tisdale, and P. Anderson. 2008. A
functional RNAi screen links O-GlcNAc modification of ribosomal proteins
to stress granule and processing body assembly. Nat. Cell Biol. 10:1224–1231.
46. Oliveira, V., et al. 2008. A protective role for the human SMG-1 kinase
against tumor necrosis factor-alpha-induced apoptosis. J. Biol. Chem. 283:
47. Roberts, T. L., J. A. Dunn, M. J. Sweet, D. A. Hume, and K. J. Stacey. 2011.
The immunostimulatory activity of phosphorothioate CpG oligonucleotides
is affected by distal sequence changes. Mol. Immunol. 48:1027–1034.
48. Sarkaria, J. N., et al. 1998. Inhibition of phosphoinositide 3-kinase related
kinases by the radiosensitizing agent wortmannin. Cancer Res. 58:4375–
49. Sheth, U., and R. Parker. 2006. Targeting of aberrant mRNAs to cytoplasmic
processing bodies. Cell 125:1095–1109.
50. Shevchenko, A., M. Wilm, O. Vorm, and M. Mann. 1996. Mass spectrometric
sequencing of proteins silver-stained polyacrylamide gels. Anal. Chem. 68:
51. Shor, B., J. J. Gibbons, R. T. Abraham, and K. Yu. 2009. Targeting mTOR
globally in cancer: thinking beyond rapamycin. Cell Cycle 8:3831–3837.
52. Solomon, S., et al. 2007. Distinct structural features of caprin-1 mediate its
interaction with G3BP-1 and its induction of phosphorylation of eukaryotic
translation initiation factor 2?, entry to cytoplasmic stress granules, and
selective interaction with a subset of mRNAs. Mol. Cell. Biol. 27:2324–2342.
53. Stephens, P., et al. 2005. A screen of the complete protein kinase gene family
identifies diverse patterns of somatic mutations in human breast cancer. Nat.
54. Thomas, M. G., M. Loschi, M. A. Desbats, and G. L. Boccaccio. 2011. RNA
granules: the good, the bad, and the ugly. Cell Signal. 23:324–334.
55. Wang, D., et al. 2011. Inhibition of nonsense-mediated RNA decay by the
tumor microenvironment promotes tumorigenesis Mol. Cell. Biol. 31:3670–
56. Weischenfeldt, J., et al. 2008. NMD is essential for hematopoietic stem and
progenitor cells and for eliminating by-products of programmed DNA rear-
rangements. Genes Dev. 22:1381–1396.
57. Wong, M. P., et al. 2003. Chromosomal aberrations of primary lung adeno-
carcinomas in nonsmokers. Cancer 97:1263–1270.
58. Yamashita, A., et al. 2009. SMG-8 and SMG-9, two novel subunits of the
SMG-1 complex, regulate remodeling of the mRNA surveillance complex
during nonsense-mediated mRNA decay. Genes Dev. 23:1091–1105.
59. Yamashita, A., T. Ohnishi, I. Kashima, Y. Taya, and S. Ohno. 2001. Human
SMG-1, a novel phosphatidylinositol 3-kinase-related protein kinase, asso-
ciates with components of the mRNA surveillance complex and is involved
in the regulation of nonsense-mediated mRNA decay. Genes Dev. 15:2215–
VOL. 31, 2011 ROLE FOR hSMG-1 IN STRESS GRANULE FORMATION 4429