Journ

alof

Cell

Scie

nce

Ribonuclease/angiogenin inhibitor 1 regulates stress-induced subcellular localization of angiogenin tocontrol growth and survival

Elio Pizzo1,*, Carmen Sarcinelli1,2,*, Jinghao Sheng2, Sabato Fusco3, Fabio Formiggini3, Paolo Netti3,Wenhao Yu2, Giuseppe D’Alessio1,` and Guo-fu Hu2,`

1Department of Biology, University of Naples Federico II, Complesso Universitario di Monte S. Angelo, via Cintia, Naples 80126, Italy2Molecular Oncology Research Institute, Tufts Medical Center, 800 Washington Street, Boston, MA 02111, USA3Center for Advanced Biomaterials for Health Care Italian Institute of Technology and Interdisciplinary Research Centre on Biomaterials,University of Naples Federico II, Piazzale Tecchio 80, Naples 80126, Italy

*These authors contributed equally to this work`Authors for correspondence (guo-fu.hu@tufts.edu; dalessio@unina.it)

Accepted 20 June 2013Journal of Cell Science 126, 4308–4319� 2013. Published by The Company of Biologists Ltddoi: 10.1242/jcs.134551

SummaryAngiogenin (ANG) promotes cell growth and survival. Under growth conditions, ANG undergoes nuclear translocation and accumulates

in the nucleolus where it stimulates rRNA transcription. When cells are stressed, ANG mediates the production of tRNA-derived stress-induced small RNA (tiRNA), which reprograms protein translation into a survival mechanism. The ribonucleolytic activity of ANG isessential for both processes but how this activity is regulated is unknown. We report here that ribonuclease/angiogenin inhibitor 1

(RNH1) controls both the localization and activity of ANG. Under growth conditions, ANG is located in the nucleus and is notassociated with RNH1 so that the ribonucleolytic activity is retained to ensure rRNA transcription. Cytoplasmic ANG is associated withand inhibited by RNH1 so that random cleavage of cellular RNA is prevented. Under stress conditions, ANG is localized to the

cytoplasm and is concentrated in stress granules where it is not associated with RNH1 and thus remains enzymatically active for tiRNAproduction. By contrast, nuclear ANG is associated with RNH1 in stressed cells to ensure that the enzymatic activity is inhibited and nounnecessary rRNA is produced to save anabolic energy. Knockdown of RNH1 abolished stress-induced relocalization of ANG and

decreased cell growth and survival.

Key words: Angiogenin, RNH1, Stress granules

IntroductionANG is the fifth member of the vertebrate-specific, secreted

ribonuclease (RNase) family (Riordan, 2001). ANG expression is

upregulated in various types of human cancer (Tello-Montoliu

et al., 2006). It promotes cancer progression (Yoshioka et al.,

2006) by stimulating both tumor angiogenesis (Kishimoto et al.,

2005) and cancer cell growth (Tsuji et al., 2005). The growth-

stimulating activity of ANG is mediated by its ability to promote

ribosomal RNA (rRNA) transcription (Tsuji et al., 2005; Xu et al.,

2002). ANG undergoes nuclear translocation in proliferating

endothelial (Moroianu and Riordan, 1994), cancer (Tsuji et al.,

2005) and neuronal (Thiyagarajan et al., 2012) cells, where it

binds to the promoter region of rDNA (Xu et al., 2003) and

stimulates rRNA transcription (Xu et al., 2002). ANG-mediated

rRNA transcription is necessary for angiogenesis stimulated by a

variety of angiogenic factors (Kishimoto et al., 2005). It also

plays an important role for cancer cell proliferation in response to

both genetic and environmental insults (Ibaragi et al., 2009;

Yoshioka et al., 2006).

In contrast to its upregulation in various cancers, ANG is

downregulated in amyotrophic lateral sclerosis (ALS)

(McLaughlin et al., 2010), Parkinson’s disease (PD) (Steidinger

et al., 2011) and Alzheimer’s disease (Kim and Kim, 2012). More

importantly, loss-of-function mutations have been found in

patients with ALS and PD (Conforti et al., 2008; Gellera et al.,

2008; Greenway et al., 2006; Paubel et al., 2008; van Es et al.,

2009; Wu et al., 2007). ANG is the only ‘loss-of-function’

mutated gene identified in ALS. Most of the PD-associated

mutations are also predicted to be loss-of-function mutations.

These genetic data clearly indicate that ANG plays a role in

neuron survival, and its deficiency is a risk factor of

neurodegenerative diseases (Li and Hu, 2010).

The extension of ANG’s biological activity from cancer

progression to neuron survival coincided with the recent

discovery that ANG mediates the production of tiRNA (Emara

et al., 2010; Fu et al., 2009; Ivanov et al., 2011; Yamasaki et al.,

2009), which have been shown to suppress global protein

translation of both capped and uncapped mRNA (Ivanov et al.,

2011). However, IRES-mediated translation with weak eIF4G

binding (Baird et al., 2006), a mechanism often used by anti-

apoptosis and pro-survival genes, is not inhibited by tiRNA

(Ivanov et al., 2011). Therefore, tiRNAs reprogram protein

translation in response to stress, thereby promoting cell survival

(Thompson et al., 2008). The production of tiRNA is induced by

stress and is mediated by ANG (Yamasaki et al., 2009).

Moreover, tiRNAs are able to stimulate the formation of stress

4308 Research Article

Journ

alof

Cell

Scie

nce

granules (SGs) (Emara et al., 2010), cytoplasmic foci whereuntranslated mRNPs are transiently stored. Formation of SGs is

an important mechanism by which cells reprogram proteintranslation to survive adverse conditions (Yamasaki andAnderson, 2008).

Cellular stresses inflicted by environmental and genetic factorsare an underlying mechanism for both cancers andneurodegenerative diseases, the two pathological conditions

where ANG has been shown to play a role. For example,hypoxic and oxidative stresses are a common etiology for cancer.Oxidative stress and endoplasmic reticulum (ER) stress resulting

from accumulation of misfolded protein aggregates is a hallmarkof neurodegenerative disease. It is therefore conceivable thatANG-mediated production of tiRNAs in response to stress resultsin reprogramming of protein translation, thereby promoting cell

survival. However, a remaining question is how ANG activity iscontrolled so that it can properly stimulate rRNA transcriptionand tiRNA production, respectively, under growth and stress

conditions. We hypothesized that differential subcellularlocalizations of ANG under growth and stress conditions mightcontrol the ultimate activity of ANG in producing either rRNA or

tiRNA. The results presented in this paper show that ANG islocalized to different cellular compartments under differentconditions. Under growth conditions, ANG is mainly nuclearwith a nucleolus accumulation. Under stress conditions, ANG is

no longer localized to the nucleus, but is rather cytoplasmic andis accumulated in SGs.

The ribonucleolytic activity is essential for ANG to induceangiogenesis (Shapiro and Vallee, 1989). However, directinjection of ANG protein into the cytosol results in degradationof cellular RNA and kills the cells (Saxena et al., 1992; Saxena

et al., 1991). These results raised another important question, thatis, how the ribonucleolytic activity of ANG is controlled invarious cellular compartments so that random RNA degradation

is avoided. We hypothesized that RNH1, an abundant 50 kDaprotein (Haigis et al., 2003) that binds ANG with a Kd of ,1 fM(Lee et al., 1989), regulates the ribonucleolytic activity of ANG

in different subcellular locations and under different growthconditions. We demonstrate in this study that under growthconditions, RNH1 is associated with cytoplasmic ANG but not

with nuclear ANG so that nuclear ANG is enzymatically activebut cytoplasmic ANG is inhibited. By contrast, under stressconditions, RNH1 is associated with nuclear ANG but not withcytoplasmic ANG so that nuclear ANG is inhibited but

cytoplasmic ANG is not. Knockdown of RNH1 alters cellularlocalization of ANG and abolishes its pro-survival activity.Together, our results demonstrate that cellular activity of ANG is

controlled both by its localization and by its association withRNH1.

ResultsDifferential subcellular localization of ANG and RNH1under growth and stress conditions

The biological activity of ANG in mediating growth and thestress response is related to its ability in stimulating rRNA

transcription and tiRNA production, respectively (Li and Hu,2010; Li and Hu, 2012). Therefore, the ribonucleolytic activity ofANG is essential, and an important question is how ANG avoids

the surveillance action of RNH1 that is abundant (Haigis et al.,2003) in both cytoplasm and nucleus (Furia et al., 2011) and thatbinds ANG with femtomolar affinity (Lee et al., 1989). To

address this question, we first examined the protein levels ofANG and RNH1 in the cytoplasm and nucleus of HeLa cells

under growth and stress conditions. Immunoblot analysis(Fig. 1A) showed that under growth conditions, more ANG isdetected in the nuclear fraction than in the cytoplasmic fraction.

Oxidative stress induced a shift of ANG distribution from thenucleus to the cytoplasm. When cells were stressed with sodiumarsenite (SA), more ANG is detected in the cytoplasm than in thenucleus. Preferential localization of ANG to the nucleus and

cytoplasm under growth and stress conditions is consistent withits respective role in stimulating rRNA transcription and tiRNAproduction under these conditions.

The subcellular distribution pattern of RNH1 is opposite to thatof ANG. More RNH1 was detected in the cytoplasmic fractionthan in the nuclear fraction under growth conditions, whereas

under stress conditions, more RNH1 was detected in the nucleusthan in the cytoplasm (Fig. 1B).

Immunofluorescence (IF) was used to reveal more details ofthe converse regulation of ANG and RNH1 in the cytoplasm andnucleus under growth and stress conditions. Consistent withimmunoblot results, ANG was mainly detected in the nucleus

(Fig. 1C, indicated by arrows) when cells were cultured undernormal growth conditions. No exogenous ANG was added to thecells in these experiments so all the IF signals were generated by

endogenous ANG. Endogenous ANG was concentrated in theperinucleolar regions where rRNA processing and assembly takesplace (Nazar, 2004). ANG was also detected in the cytoplasm,

albeit not as strongly as in the nucleus. If exogenous ANG wasadded to the cells cultured under normal growth conditions, muchmore prominent and clear nucleolar accumulation of ANG wasdetected (supplementary material Fig. S1).

Under growth conditions, RNH1 was strongly detected in thenuclear plasma but not in the nucleolus (Fig. 1C, nucleoli

indicated with dashed arrows). Cytoplasmic RNH1 was alsovisible but was not as strong as in the nucleus. The merged imageshows that ANG and RNH1 are mainly colocalized in cytoplasmand nucleoplasm, but clearly not in the nucleolus. It is thus clear

that under growth conditions, at least in the nucleolus, ANG isnot associated with RNH1 and is not inhibited, so that nucleolarANG remains active as a ribonuclease for the task of stimulating

rRNA transcription (Xu et al., 2002).

Oxidative stress induced more cytoplasmic localization ofANG and more nuclear accumulation of RNH1 (Fig. 1D). The

cytoplasmic ANG displayed a more punctate staining pattern instressed cells. Two types of punctate cytoplasmic ANG stainingwere identified from the merged images: those colocalized with

RNH1 (indicated by arrows) and those free of RNH1 (indicatedby arrowheads). Prominent nucleolar staining of RNH1 wasobserved (dashed arrows), suggesting that any remaining ANG in

the nucleolus would have been inhibited by RNH1.

Taken together, these results demonstrated that subcellularlocalization of ANG and RNH1 are dependent on the growth

status of the cells and are oppositely regulated by stress. Whencells are under growth conditions, ANG is mainly in the nucleusor nucleolus where it is not colocalized with RNH1 and is

therefore not inhibited by RNH1 so that ANG remains fullyactive to stimulate rRNA transcription. When cells are stressed,the majority of ANG is not in the nucleus but RNH1 is

accumulated in the nucleolus so that the trace amounts of nuclearANG are probably inhibited by RNH1 to ensure no ANG-stimulated rRNA transcription takes place to save anabolic

RNH1 regulates ANG localization and activity 4309

Journ

alof

Cell

Scie

nce

energy and to allocate as many resources as possible for damage

repair.

Identical results were obtained with LNCaP human prostate

cancer cells (supplementary material Fig. S2). Treatment with SA

induced relocation of ANG from nucleus to cytoplasm and,

conversely, RNH1 from cytoplasm to nucleus in LNCaP cells.

Moreover, a similar pattern of relocalization of ANG and RNH1

was observed in HeLa cells treated with tunicamycin, an ER

stress inducer (supplementary material Fig. S3). These results

indicate that the opposite traffic of ANG and RNH1 between the

nucleus and cytoplasm when the environment of the cells is

shifted from growth to stress conditions is not limited to certain

types of cells or stresses. It might be a general phenomenon that

when cells are stressed, less ANG but more RNH1 is

accumulated in the nucleus. Given the fact that a major role of

nuclear ANG is to stimulate rRNA transcription, it makes sense

that when cells are in conditions unfavorable for growth, ANG

leaves the nucleus so that the rate of rRNA transcription is

reduced to avoid energy waste for cells to survive under adverse

conditions. It also makes sense for more RNH1 to accumulate in

the nucleolus to ensure that the remaining ANG in the nucleolus

is inhibited to halt rRNA transcription.

ANG is associated with RNH1 in cytoplasm under growth

conditions and in the nucleus under stress conditions

The ribonucleolytic activity of ANG is essential for its biological

activity. However, if it is not well controlled, it will randomly

degrade cellular RNAs and be detrimental to the cell. In order to

know whether the ribonucleolytic activity of cytoplasmic and

nuclear ANG is regulated by association with RNH1 under growth

and stress conditions, we performed co-immunoprecipitation (co-

IP) experiments to examine the interactions between ANG and

RNH1 in cytoplasmic and nuclear extracts of the cells cultured

under growth and SA-induced stress conditions. RNH1 was

precipitated by ANG monoclonal antibody (mAb) (Fig. 2A, lane

3). Similarly, ANG could be precipitated by RNH1 polyclonal

antibody (pAb) (Fig. 2A, lane 4). Under these conditions, no ANG

and RNH1 could be co-immunoprecipitated in the nuclear extracts

(Fig. 2A, lanes 5 to 8). When cells were stressed with SA

treatment, no association between ANG and RNH1 was detected

Fig. 1. Differential subcellular localization of ANG and RNH1 under growth and stress conditions. (A,B) Immunoblot analyses of ANG and RNH1 in

nuclear and cytoplasmic fractions of HeLa cells cultured under growth and stress conditions. HeLa cells were cultured in normal growth medium (DMEM + 10%

FBS) or treated with 0.5 mM SA at 37 C̊ for 1 hour. Cells were fractionated and the nuclear and cytoplasmic proteins (30 mg) were analyzed for ANG (A) and

RNH1 (B) by Immunoblot with affinity-purified pAb against ANG and RNH1, respectively. B23, nucleophosmin, a nuclear marker. (C,D) IF detection of ANG

and RNH1 in HeLa cells cultured under growth (C) and stress (D) conditions. ANG mAb and Alexa-Fluor-555-labeled goat anti-mouse F(ab9)2 were used to stain

ANG. RNH1 pAb and Alexa-Fluor-488-labeled goat-anti-rabbit F(ab9)2 were used to stain RNH1. Nuclei are stained with DAPI. Arrows indicate overlapping

signals of ANG and RNH1. Arrowheads indicate ANG signals non-overlapping with RNH1. Dashed arrows indicate nucleoli. Scale bars: 10 mm.

Journal of Cell Science 126 (18)4310

Journ

alof

Cell

Scie

nce

by co-IP in the cytoplasmic extracts (Fig. 2B, lanes 1 to 4) but was

apparent in the nuclear extracts (Fig. 2B, lanes 5 to 8). These

results indicate that ANG is not associated with RNH1 in the

nucleus when cells are cultured under growth conditions so that its

ribonucleolytic activity is not inhibited by RNH1 and ANG is able

to stimulate rRNA transcription thereby promoting cell growth. At

the same time, cytoplasmic ANG is associated with RNH1 and its

ribonucleolytic activity is inhibited to avoid RNA degradation,

which would be otherwise harmful to the cell. Further, these results

indicate that when cells are stressed, ANG not only moves out of

the nucleus (Fig. 1) but also the remaining nuclear ANG is

associated with RNH1 to ensure any trace amount of ANG activity

is inhibited. Under stress condition, no association between ANG

and RNH1 was detected in the cytoplasmic extracts suggesting that

ANG should be active. This is consistent with the proposed

function of cytoplasmic ANG in promoting cell survival by

mediating the production of tiRNA under stress (Emara et al.,

2010; Ivanov et al., 2011). From the viewpoint of cell survival

under stress, it is reasonable that nuclear ANG is inhibited by

RNH1 so that no unwanted rRNA transcription takes place to

apportion as much energy as possible for damage repair and cell

survival.

Stress induces redistribution of cytoplasmic ANG and

RNH1 into high molecular mass supracomplexes

Fig. 1 shows punctate cytoplasmic staining of both ANG and

RNH1 in stressed cells, suggesting that ANG and RNH1 localizes

to some cytoplasmic organelles or assembles into some

supermolecular structures. We first used gel filtration

chromatography on a Superdex G-200 column to examine the

elution profiles of ANG and RNH1 from cytoplasmic extracts of

cells cultured under growth (Fig. 3A,B) and stress (Fig. 3C,D)

conditions. Oxidative stress induced a drastic shift of absorbance

to the high molecular mass fractions, indicating formation of high

molecular mass complexes. Accompanied with the formation of

supramolecular structures, there was a shift of distribution of

ANG and RNH1 in different fractions. ANG was detected only in

fraction 7 and RNH1 was detected in fractions 6 and 7 (Fig. 3B)

under growth conditions indicating that ANG and RNH1 existed

as monomeric forms in the cytoplasm. In the fractions generated

from the cytoplasm of stressed cells, ANG was detected across

the entire spectrum from fractions 2 to 7 and RNH1 from

fractions 2 to 6, suggesting a widespread distribution of both

proteins in various structures of high molecular mass (Fig. 3D).

Stress-induced assembly of ANG and RNH1 in SGs

ANG has been shown to potentiate stress-induced formation of

SGs through production of tiRNA (Emara et al., 2010; Yamasaki

et al., 2009). However, it is unknown whether ANG itself is

located in SGs and where tiRNA is produced. The findings that

stresses induce relocation of cytoplasmic ANG and RNH1 to high

molecular mass structures (Fig. 3), and that ANG and RNH1

display punctate staining patterns in the cytoplasm under stress

(Fig. 1), led us to examine whether ANG and RNH1 are located

in SGs under stress. Double IF on SA-stressed HeLa cells with an

affinity-purified pAb specific to ANG and a mAb specific to

polyadenine binding protein (PABP), a marker of SG (Buchan

and Parker, 2009), showed colocalization of ANG and PABP

(Fig. 4A, arrows), indicating that ANG is localized to SGs. It is

also notable that some of the punctate ANG staining in the

cytoplasm did not colocalize with PABP (Fig. 4A, arrowheads),

indicating that ANG is also present in cytoplasmic organelles that

are not SGs. RNH1 was also found to be present in both SGs

(Fig. 4B, arrows) and non-SG organelles (Fig. 4B, arrowheads).

These results demonstrated that oxidative stresses induce

localization of both ANG and RNH1 in SGs. It is interesting to

note that all the PABP-containing SGs (red) overlapped with

ANG (Fig. 4A, green), but not all of them overlapped with RNH1

(Fig. 4B, dashed arrows). There were no SGs that did not contain

ANG but there are SGs that were free of RNH1, indicating that

ANG and RNH1 were colocalized in some but not all SGs.

Confocal microscopy also showed that ANG was present in all

SGs (Fig. 4C) but RNH1 was only found in some of the SGs

(Fig. 4D). In the two cells located in a selected z-section in

Fig. 4C, only 1 of the 35 countable SGs did not contain ANG

(dashed arrows). The remaining 34 SGs all contained ANG

Fig. 2. Interaction between ANG and RNH1 in the nucleus and cytoplasm under growth and stress conditions. HeLa cells were cultured under normal

growth conditions (A) and SA (0.5 mM, 1 hour)-induced oxidative stress conditions (B). Cells were fractionated and the cytoplasmic (lanes 1–4) and nuclear

fractions (lanes 5–8) were precipitated with a non-immune mouse IgG, ANG mAb, or affinity-purified RNH1 pAb. Immunoblot analyses were performed with

RNH1 pAb or ANG pAb. Input control lanes had 10% of the materials used for co-IP. The band above the specific RNH1 band is the heavy chain of the rabbit

IgG. The purity of cytoplasmic and nuclear fractions was analyzed by Immunoblot analyses of b-tubulin and PCNA from 1% of the material used for co-IP.

RNH1 regulates ANG localization and activity 4311

Journ

alof

Cell

Scie

nce

(arrows), representing a 97% ANG coverage. However, only 8 of

the 24 countable SGs in Fig. 4D contained RNH1 (arrows),

representing an RNH1 presence in 33% of SGs. The remaining

24 SGs (67%) were RNH1 free. These results suggested that

although both ANG and RNH1 are located in SGs when cells are

stressed, the extent of their SG occupancy is different. These

results suggest that some of the cytoplasmic ANG under stress

will not be associated with RNH, consistent with our finding that

ANG and RNH1 were not co-immunoprecipitated in the

cytoplasmic extracts from stressed cells (Fig. 2). Very similar

results were obtained with LNCaP cells (supplementary material

Fig. S4), suggesting that localization of ANG and RNH1 in SGs

is a general phenomenon and is not limited to a certain cell type.

Localization of ANG and RNH1 in SGs was confirmed by

their colocalization with TIA1, another SG marker (Anderson

and Kedersha, 2009). Granules that were both positive and

negative for ANG and RNH1 were observed in TIA1-positive

SGs. Again, the percentage of RNH1-positive SGs (Fig. 4F,

indicated by arrows) was lower than that of ANG-positive SGs

(Fig. 4E, indicated by arrows).

Fluorescence resonance energy transfer between

cytoplasmic ANG and RNH1 under growth and stress

conditions

Localization of both ANG and RNH1 in SGs prompted us to

examine whether they are physically associated in SGs.

Fluorescence resonance energy transfer (FRET) was used for

this purpose. When cells were not stressed (Fig. 5A), in the

marked region of interest in the cytoplasm, the intensity of green

fluorescence from the donor (ANG) was 29.367.2 units and the

intensity of the red fluorescence from the acceptor (RNH1) was

76.463.5 units, respectively. After photobleaching of red

fluorescence to 5.360.4 units (6.9% of the original), the

intensity of the donor fluorescence became 32.067.6 units,

representing an increase of 10.161.2%. Therefore, there was an

energy transfer from donor fluorophore to acceptor fluorophore

suggesting that a physical interaction existed between ANG and

RNH1 in the cytoplasm when cells were cultured under normal

growth conditions. In the selected SGs that contained both ANG

and RNH1, no energy transfer was observed (Fig. 5B). The

fluorescence intensity of the donor was 75.469.1 and 76.069.6

units, respectively, before and after photobleaching of the

acceptor, representing a mere 0.8% increase. These results

indicated that ANG and RNH1 were not physically associated in

SGs even though they were both located there. Therefore,

cytoplasmic ANG retains its ribonucleolytic activity for the

purpose of generating tiRNA to promote cell survival under

stress.

Localization of ANG and RNH1 in cells recovered from

stress

HeLa cells were stressed with 0.5 mM SA for 1 hour and then

allowed to recover in full growth medium for 3 hours.

Subcellular localization of ANG and RNH1 in recovered cells

was first examined by double IF (supplementary material Fig.

S5A) and was found to be very similar to that of the cells cultured

only in growth conditions (Fig. 1C). The most significant feature

is that ANG relocated back to the nucleus and that RNH1

disappeared from nucleolus (supplementary material Fig. S5A,

dashed arrows). There was still a trace amount of both ANG and

RNH1 in SGs (supplementary material Fig. S5B,C) but both the

number and the size of SGs was significantly reduced compared

Fig. 3. Oxidative stress induces assemble of cytosolic ANG and RNH1 into high molecular mass complexes. HeLa cells were cultured under normal growth

conditions (A,B) or under oxidative stress induced by 0.5 mM SA for 1 hour (C,D). Cytosolic fractions were extracted and subjected to gel filtration

chromatography on a Superdex G-200 column (30 cm61 cm, 25 ml). The elution profiles were recorded as milli absorbance units at 260 nm (A,C). The arrows

indicate the elution volume of size markers, which include Herceptin (148 kDa), a compact antibody (105 kDa), ovalbumin (45 kDa) and RNaseA (15 kDa).

(B,D), selected fractions from the eluates (indicated by numbers in A and C) were analyzed for ANG and RNH1 by immunoblot. Data shown are from a

representative experiment of four repeats. Identical results were obtained from each experiment.

Journal of Cell Science 126 (18)4312

Journ

alof

Cell

Scie

nce

Fig. 4. Detection of ANG and RNH1 in SGs. (A) Double IF of ANG and PABP. HeLa cells were incubated with 0.5 mM SA for 1 hour and were fixed by

methanol at 220 C̊ for 10 minutes. Cell nuclei were stained with DAPI. Arrows indicate ANG signals in SG. Arrowheads indicate staining of ANG in cytoplasmic

organelles that are not SGs. (B) Double IF of RNH1 and PABP. Arrows and arrowheads indicate RNH1 signals in SG and non-SG cytoplasmic organelles.

(C–F) Confocal image of double IF between ANG and PABP (C), RNH1 and PABP (D), ANG and TIA1 (E), and RNH1 and TIA1 (F). HeLa cells were incubated

with 0.5 mM SA for 1 hour, fixed with 4% paraformaldehyde and permeabilized with 0.5% Triton X-100 (C) and (D) or fixed in methanol at 220 C̊ (E,F). A

series of Z-section images were taken and the center panel was selected for analysis. Arrows indicate SGs that were stained both for ANG and PABP (C) or TIA1

(E), and by RNH1 and PABP (D) or TIA1 (F). Dashed arrows indicate SGs that contain only PABP (A,D) or TIA1 (E,F). Images in C and D were taken with SP5

Leica confocal microscope. Images in E and F were taken with a Zeiss LSM 410 confocal microscope. Scale bars: 10 mm.

Fig. 5. Immuno-FRET between cytoplasmic ANG and RNH1 under growth and stress conditions. HeLa cells were cultured in normal growth conditions

(A) or treated with 0.5 mM SA for 1 hour (B), fixed with 4% paraformaldehyde and permeabilized with 0.5% Triton X-100. ANG and RNH1 were detected with

mAb and pAb, respectively, and visualized with Alexa-Fluor-488- and Cy3-labeled secondary antibodies. Green and red fluorescence confocal images were taken

under a single excitation (488 nm) wavelength and were merged. The center panel of the Z-section images was used for FRET analyses. The intensity of both

green (donor, ANG) and red (acceptor, RNH1) fluorescence in the selected ROI was recorded. The red fluorescence was then bleached at 592 nm for 5 seconds,

and the green and red fluorescence was recorded again. Scale bars: 5 mm.

RNH1 regulates ANG localization and activity 4313

Journ

alof

Cell

Scie

nce

with that in the stressed cells (Fig. 4). These results indicate that

when cells were recovered from stress, SGs dissembled and

therefore ANG and RNH1 relocated both to the cytoplasm and

nucleus. Specifically, RNH1 disappeared from the nucleolus so

that the ribonucleolytic activity of ANG in the nucleolus

resumed, enabling its activity in stimulating rRNA transcription.

The fate of cytoplasmic ANG and RNH1 in recovered cells

was further studied by using a sucrose gradient. Fig. 6A shows

the sucrose gradient profiles of the cytoplasmic extracts of cells

that have been cultured in growth, stress and stress recovery

conditions. Similar to the results obtained from gel filtration

chromatography, the most significant difference between growth

and stress conditions was noticed in the high molecular mass

region between fractions 3 and 9 where the UV absorbance was

much higher in the fractions generated from stressed cells. The

profile from stress-recovered cells was very similar to that from

the cells cultured under growth conditions. Fig. 6B is thedistribution of cellular RNA (mainly rRNA) in the sucrose

gradient fractions, which marks the region (fractions 3–9) whereribosomes were eluted. It is notable that total RNA is decreasedin the stressed cells, in agreement with an early finding that stressdecreases the abundance of polyribosomes (Bevilacqua et al.,

2010). Because the size of SGs and ribosomes are similar(Souquere et al., 2009), it is likely that SGs were also distributedin these fractions. This hypothesis was confirmed by immunoblot

analysis of PABP (Fig. 6C). Under growth conditions, PABP wasdetected in a roughly equal density in every fraction, indicatingan even distribution in the cytoplasm. In stressed cells, PABP

were enriched in fractions 3–11, indicating that it was thesefractions where SGs were eluted. Both ANG and RNH1 werefound in the lower molecular mass fractions under growthconditions, confirming that cytoplasmic ANG and RNH1 were

not in supramolecular structures. Stress-induced relocation ofboth ANG and RNH1 to high molecular mass sucrose gradientfractions was confirmed. It is of significant interest to note that

ANG and RNH1 were detected in lower molecular mass fractionsin the cells recovered from stress (Fig. 6C, bottom three panels),indicating that a recovery process of cells from stress is

accompanied with a return of ANG and RNH1 to their normallocalization patterns. These results indicate that stress-inducedrelocalization of ANG and RNH1 is dynamic and is restored

when stress is lifted.

Knockdown of RNH1 impairs localization of ANG in SGsand alters cell growth and survival behavior

To understand the effect of RNH1 on localization of ANG to SGswhen cells were exposed to stress, we used lentivirus-mediatedsmall hairpin RNA (shRNA) to knock down RNH1 expression

and examined the resultant changes in ANG localization by IF.Among the five different shRNA constructs, four of themknocked down RNH1 expression efficiently as shown in Fig. 7A.

The relative intensity of RNH1 to b-tubulin in the cells infectedwith lentivirus encoding a scramble shRNA, RNH1-specificshRNA 33, 34, 35 and 37 were 100, 8.660.1, 18.264.8, 3.060.7and 12.966.9, respectively. Thus, clone 35 (Sh35) was most

efficient in knocking down RNH1 expression with a 97%decrease in RNH1 protein level, and was selected for furtherstudy. Scramble control and Sh35 shRNA transfected HeLa cells

were stressed with SA and localization of ANG was examined byIF. ANG was found to be colocalized with PABP in SGs inscramble shRNA transfected cells (Fig. 7B), not so different from

the pattern seen with untransfected HeLa cells shown in Fig. 4A.However, in RNH1-knockdown cells, SGs were formed normallybut no ANG was colocalized with PABP (Fig. 7C). Instead, ANG

was found in the nucleolus (indicated by dashed arrows). Theseresults indicate that stress-induced relocalization of ANG in SGswas impaired when RNH1 was downregulated.

Cell proliferation in RNH1-knockdown cells was significantly

reduced compared with levels in the scramble control (Fig. 8A).All four shRNA constructs inhibited cell growth and the extentsof inhibition correlated positively to the knockdown efficiency

(Fig. 7A, Fig. 8A). We next examined the sensitivity of RNH1-knockdown cells to SA stress. Fig. 8B,C shows that RNH1

knockdown decreased viability of the cells under oxidative stress.

When cells were treated with 0.5 mM SA for 4 hours, the numberof viable cells in scramble and Sh35 transfectants was 57% and26%, respectively, of that before the treatment (Fig. 8B),

Fig. 6. Dynamic localization of cytoplasmic ANG and RNH1 under

various growth conditions. HeLa cells were cultured under growth and

oxidative stress conditions (0.5 mM SA, 1 hour). In the recovery experiment,

SA was removed by medium exchange and the cells were washed three times

and cultured in growth conditions for 3 hours. Cells were fractionated and the

cytoplasmic fractions were ultracentrifuged in a sucrose gradient (60–15%).

(A) Absorbance profiles at 260 nm of the sucrose gradient fractions. (B) RNA

content in each fraction. 50 ml from every other fraction of the gradient were

separated in 1% agarose gels and stained with EB. (C) Immunoblot analyses

of ANG, RNH1 and PABP from alternate gradient fractions collected in A.

Data shown are from a representative experiment of four repeats. Identical

results were obtained from each repeat.

Journal of Cell Science 126 (18)4314

Journ

alof

Cell

Scie

nce

representing a 37% increase in cell death rate in RNH1-

knockdown cells. When the SA concentration was increased to

1 mM, cell death was obvious after 1 hour of treatment

(Fig. 8C). Again, RNH1-knockdown cells were more sensitive

than the scramble control. Viability of scramble and Sh35

transfectants was 81% and 60%, respectively (Fig. 8C). At 2 mM

SA, massive cell death occurred in both scramble and Sh35

transfectants. However, it is apparent that there were still more

viable cells in control cells (15%) than in RNH1-knockdown

cells (9%). Taken together, these results indicate that RNH1

knockdown changed the localization of ANG as well as cell

growth and survival, suggesting that a proper RNH1 level might

be essential for the cellular functions of ANG.

To examine the effect of RNH1 knockdown on cell apoptosis,

ethidium bromide (EB) and Acridine Orange (AO) staining was

used to stain apoptotic cells. AO permeates intact cells and stains

all nuclei green whereas EB enters cells only when the integrity of

plasma membrane is compromised so it stains apoptotic nuclei red

(Ribble et al., 2005). Cell apoptosis significantly increased in

RNH1-knockdown cells when they were cultured both under

growth (Fig. 8D) and stress (Fig. 8E) conditions. The percentage

of apoptotic cells in Sh35 transfectants was 11.863.7%, which is

1.9-fold that in scramble control transfectants (6.262.7%). When

the cells were treated with 1 mM SA for 1 hour, the percentage of

apoptotic cells in scramble and Sh35 transfectants was 11.364.5%

and 15.964.9%, respectively, representing a 41% increase in

SA-induced apoptosis when RNH1 was knocked down. These

results suggest that increased apoptosis might be a reason for

decreased growth and survival in RNH1-knockdown cells.

DiscussionDynamic cellular localization of ANG and RNH1

Our results show that subcellular localization of ANG is dynamic

and is dependent on the growth status of the cell. ANG is mainly

localized to the nucleus when cells are in normal growth

conditions. Oxidative and ER stress both induced a shift of ANG

localization from nucleus to cytoplasm. Differential subcellular

localization of ANG under different growth conditions might be a

mechanism of regulation of the growth and survival functions of

ANG. The growth-stimulating function of ANG is mediated by

its activity in promoting rRNA transcription. For this purpose,

ANG needs to bind to the promoter region of rDNA, which is

probably the reason why ANG is mainly in the nucleus to meet

the high metabolic demand of growing cells for rRNA. The pro-

survival function of ANG is mediated by its ability to produce

tiRNA. When cells are in adverse conditions, the protein

translation rate is decreased to save anabolic energy for

survival. The production of tiRNA meets this purpose because

tiRNAs suppress global protein translation but do not alter IRES-

mediated translations of anti-apoptotic mRNAs. It therefore

makes sense for ANG to leave nucleus when cells are under stress

to avoid waste of resources in producing unnecessary rRNA.

Fig. 7. Knockdown of RNH1 alters subcellular localization of ANG under stress. HeLa cells were infected with pLKO.1 lentiviral particles encoding RNH1-

specific shRNA and scramble control. Infected cells were selected in the presence of 1.5 mg/ml puromycin. (A) RNH1 levels in scramble control and specific

shRNA lentivirus-infected HeLa cells. Left panel, Immunoblots. Right panel, ImageJ analyses of relative band intensity of RNH1 over b-tubulin.

(B,C) Subcellular localization of ANG in RNH1-knockdown cells under oxidative stress. Cells infected with scramble control (B) and Sh35 (C) lentivirus were

subjected to 0.5 mM SA for 1 hour and stained for ANG and PABP. Arrows indicate colocalization of ANG and PABP in control cells. Arrowheads indicate

nucleoli staining of ANG in RNH1-knockdown cells. Scale bars: 10 mm. Similar results were obtained in cells infected with Sh33 and Sh34.

RNH1 regulates ANG localization and activity 4315

Journ

alof

Cell

Scie

nce

A surprising finding is that RNH1 was found to be mainly

located in the nucleus under growth conditions. RNH1 was

previously considered to be a cytoplasmic protein (Haigis et al.,

2003) even though it had also been detected in the nucleus (Furia

et al., 2011). The role of nuclear RNH1 under growth conditions

is currently unclear. However, it is not associated with ANG as

shown by co-IP experiments. Thus, normal function of nuclear

ANG in stimulating rRNA transcription is not inhibited by RNH1

under growth conditions. At the same time, cytoplasmic ANG is

associated with RNH1, as shown by both co-IP and FRET

experiments. RNH1 binds to ANG with a sub-femtomolar affinity

(Lee et al., 1989) and has been shown to inhibit both enzymatic

and angiogenic activity of ANG. We can reasonably assume that

when ANG is bound by RNH1, its activity will be completely

inhibited. Therefore, under growth conditions, the small amount

of ANG in the cytoplasm is most likely inhibited by RNH1 so

that no unfavorable RNA degradation occurs in the cytoplasm to

maintain a proper healthy status of the cell.

Even more RNH1 was found in the nucleus when cells were

stressed. More significantly, prominent nucleolar localization of

RNH1 was noticed. It is reasonable for nuclear RNH1 to be

located in nucleolus under stress conditions so that any trace

amount of ANG remained in the nucleolus will be bound and

inhibited by RNH1. Indeed, co-IP results showed that ANG and

RNH1 were associated in the nuclear fractions extracted from

stressed cells. Stresses not only reduce the amount of cytoplasmic

RNH1, but also disable its interaction with ANG in cytoplasm.

From a functional point of view, it is reasonable for RNH1 to

dissociate from ANG in cytoplasm under stress so that ANG

regains its ribonucleolytic activity for the purpose of tiRNA

production. RNH1 is sensitive to oxidation, attributable to its Cys

residues. RNH1 contains 32 free Cys residues and no disulfide

bonds, and loses activity in the absence of reducing agents

(Blackburn et al., 1977). Treatment of RNH1–RNase complexes

with p-hydroxymercuribenzoate rapidly dissociates the complex,

releasing fully active RNase. Oxidation or derivatization of Cys

residues, none of which forms important contacts with ANG

(Papageorgiou et al., 1997), drastically alters the 3D structure of

RNH1 (Fominaya and Hofsteenge, 1992), which may lead to

dissociation of ANG from RNH1–ANG complex. In any event,

ANG dissociates from RNH1 in cytoplasm under stress,

suggesting that cytoplasm is the likely place where tiRNA is

produced. These results demonstrated that interacting with RNH1

fine-tunes the regulatory activities of ANG in stimulating cell

growth or in promoting cell survival.

RNH1 regulates SG localization of ANG under stress

Stress induces traffic of ANG not only from the nucleus to the

cytoplasm but also from low molecular mass fractions to high

molecular supramolecular structures within the cytoplasm. One

class of these supramolecular structures is the SG. It is significant

that ANG is located in SGs in stressed cells and that this

localization is regulated by RNH1. ANG was detected in almost

every SG induced by SA. RNH1 was also found in some but not

Fig. 8. Knockdown of RNH1 in HeLa cells decrease growth and increases sensitivity to stress. (A) Cell growth. Stable scramble and RNH1 shRNA

transfectants were cultured in DMEM + 10% FBS in the presence of 0.5 mg/ml ANG. Cell numbers were determined with a Coulter counter. (B) Time course of

cell survival. Cells were cultured in growth medium for 48 hours, and then subjected to treatment with 0.5 mM SA for the indicated time. Cell numbers were

determined by MTT assay. (C) SA dose response of cell survival. Cells were cultured in growth medium for 48 hours and treated with different concentrations of

SA for 1 hour. Cell numbers were determined by MTT assay. Data shown in A–C are means 6 s.d. of triplicates of a representative experiment. (D) Cell apoptosis

under growth conditions. Cells were cultured in DMEM + 10% FBS and stained with EB and AO. Both apoptotic (red) and live (green) cells were counted.

(E) SA-induced apoptosis. Cells were treated with 1 mM SA for 1 hour and stained with EB and AO. The numbers shown in D–E are means 6 s.d. of the

percentage of apoptotic cells counted in five microscopic fields.

Journal of Cell Science 126 (18)4316

Journ

alof

Cell

Scie

nce

all SGs. So a subset of SGs contained ANG but was RNH1 free.Moreover, in those SGs that contained both ANG and RNH1,

ANG was not associated with RNH1. Therefore, cytoplasmicANG that is localized in SGs would be enzymatically active. SGsare primarily composed of the stalled 48S preinitiationcomplexes containing mRNA bound to small ribosome subunits

and the initiation factors (Anderson and Kedersha, 2008). SGsalso contain other proteins with diverse functions includingRNA-binding proteins, RNA helicases, nucleases, kinases and

signaling molecules (Anderson and Kedersha, 2009). It is ofparticular interest to note that SGs also contain RNA-inducedsilencing complexes. It is therefore conceivable that the function

of ANG located in SGs may be integrated with microRNA-induced translational silencing mechanism thus potentially beinvolved in diverse cellular pathways. No matter what role ANGplays in SGs, it is unlikely that SGs is the place where ANG

cleaves tRNA. Knockdown of RNH1 drastically reducedlocalization of ANG in SGs. tiRNA production would beexpected to be decreased in RNH1-knockdown cells if it were

SGs where ANG produced tiRNA. But it has been shown thattiRNA production is increased when the RNH1 level isdownregulated (Yamasaki et al., 2009).

ANG was also localized in other classes of superamolecularstructure that are not SGs. The nature of these structures areunknown at present but they are probably not P-bodies because

staining with P-body markers did not show colocalization withANG. Localization of ANG to this type of structure wassignificantly increased in RNH1-knockdown cells when theywere subjected to stress. One possible candidate for this

structure(s) is the polysome or the 60S subunit where ANG hasbeen detected. If the polysome is the place where tiRNA isproduced, it will be consistent with the recent finding that tiRNA

production is higher when protein synthesis is active and whentRNAs transit more frequently between the ribosome-bound andthe aminoacyltRNA-synthetase-bound states (Saikia et al., 2012).

In any event, these results demonstrate that both the cellularlocalization and function of ANG are regulated by the RNH1level.

RNH1 regulates cell proliferation and survival

RNH1 is one of the most abundant cellular proteins accountingfor 0.08% of total cytosolic protein content (Haigis et al., 2003).

The biochemical properties of RNH1 have been welldocumented, including characterization of high-resolution X-ray structures of free RNH1 (Kobe and Deisenhofer, 1993) and in

complex with RNaseA (Kobe and Deisenhofer, 1996), ANG(Papageorgiou et al., 1997) and RNase1 (Johnson et al., 2007).RNH1 is composed of seven leucine-rich repeats and a conservedstructure domain that are often involved in protein–protein

interactions (Kobe and Deisenhofer, 1995) and that have beenfound in many proteins with diverse cellular functions rangingfrom cell-cycle regulation to DNA repair to immune regulation

(Bella et al., 2008; de Wit et al., 2011). Compared with the well-defined biochemical properties of RNH1, its physiological andpathological roles are evolving. RNH1 was first thought to serve

as cellular ‘sentry’ (Haigis et al., 2003) to protect cells fromdamage caused by non-cytosolic RNase that gain entry into thecytoplasm. Indeed, extensive efforts have been put to design

cytotoxic RNase that do not bind RNH1 for therapeutic purposesin cancer treatment (Lee and Raines, 2008; Rutkoski and Raines,2008).

We have found that knockdown of RNH1 inhibited cellproliferation in a dose-dependent manner. One of the reasons for

decreased cell growth in RNH1-knockdown cells can beattributed to increased apoptosis. The percentage of apoptoticcells doubled when RNH1 was knocked down. Knockdown of

RNH1 also significantly decreased cell survival under stressconditions, accompanied with an increase in cell apoptosis.Decreased survival of RNH1-knockdown cells against oxidative

stress is probably related to abnormal cellular localization ofANG. The fact that ANG is localized in the nucleolus but not in

SGs when RNH1-knockdown cells were subjected to SA stressmight be the reason for decreased survival. rRNA transcription isan energetically costly process. From the cell survival viewpoint,

it is certainly counterproductive for ANG to be in the nucleolus tocontinually produce rRNA when cells are stressed. Failure ofANG to localize to SGs in stressed cells could also cause

decreased survival of RNH1-knockdown cells. Thus, a major roleof RNH1 is to regulate cellular localization of ANG, thereby

controlling cell growth and survival. The reported antioxidant(Cui et al., 2003) and redox homeostatic (Monti et al., 2007)effects might also contribute to the regulatory function of RNH1

in cell survival. RNH1 has also been shown to mediate pri-miR-21 processing, thereby contributing to cancer progression (Kimet al., 2011). In view of the facts that ANG is a RNase and that

ANG, RNH1 and the RISC complex are all found in SGs, it istempting to speculate that ANG and RNH1 play a role in microRNA biogenesis and that SGs are an additional or alternative

place where micro RNAs are generated or metabolized.

Materials and MethodsCell cultures and treatment

HeLa cells were maintained in DMEM supplemented with 10% FBS. LNCaP cellswere maintained in RPMI 1640 supplemented with 10% FBS. To induce oxidativestress, cells were treated with 0.5 mM SA for 1 hour or as otherwise indicated. Toinduce ER stress, cells were treated with 5 mg/ml tunicamycin for 24 hours. In therecovery experiments, medium containing SA was removed and cells were washedwith DMEM three times and cultured in growth medium for 3 hours.

Cell extracts and co-IP

Cells were detached by trypsin-EDTA and resuspended in 10 mM HEPES, pH 8.0,containing 10 mM KCl, 1.5 mM MgCl2, 0.5 mM DTT, and 16 proteinaseinhibitors cocktail and incubated on ice for 10 minutes. Cells were lysed by addingNP-40 to a final concentration of 0.1%. Cytoplasmic fraction was obtained bycentrifugation at 1000g for 5 minutes. The pellet was dissolved in RIPA buffer andwas designated as the nuclear fraction. The purity of the cytoplasmic and nuclearfractions was examined by immunoblot analyses of b-tubulin and PCNA,respectively. For co-IP experiments, the cytoplasmic and nuclear fractions from46106 cells were diluted in 1 ml of 10 mM HEPES containing 0.1% NP-40. Afraction of 50 ml was taken and used as input control. The remaining materialswere divided into three equal fractions, incubated with 5 mg of non-immune mouseIgG, ANG mAb 26-2F, or affinity-purified RNH1 pAb R127 at 4 C̊. Five ml fromeach sample was taken for b-tubulin and PCNA analysis. The remaining solutionwas mixed with 50 ml of 50% Protein A/G-Sepharose by rotating at 4 C̊ for2 hours. The mixture was centrifuged, washed and analysed by Immunoblot forANG and RNH1 with R113 and R127, respectively.

Gel filtration chromatography

The cytoplasmic fraction from 66106 HeLa cells was diluted in 200 ml of PBS andapplied to a Superdex G-200 column (30 cm61 cm, 25 ml) equilibrated in PBS.The eluate was monitored at 260 nm, collected in 0.25 ml fractions and analyzedfor ANG and RNH1 contents by Immunoblot.

Sucrose gradient ultracentrifugation

The cytoplasmic fractions of the cells were layered onto 60–15% sucrose gradients(10 mM HEPES, pH 7.4, 5 mM MgCl2 and 300 mM KCl). After centrifugation at38,000 r.p.m. overnight at 4 C̊ in a Beckman SW 41 rotor, the gradient fractionswere collected from the bottom of the tubes using a peristaltic pump and monitoredby UV absorbance at 260 nm. RNA contents were analyzed by agarose gelelectrophoresis. ANG and RNH1 were analyzed by Immunoblot.

RNH1 regulates ANG localization and activity 4317

Journ

alof

Cell

Scie

nce

Immunofluorescence and confocal microscopy

Cells were cultured on coverslips and were fixed in methanol at 220 C̊ for10 minutes, blocked with 3% BSA in 10 mM Tris-HCl, pH 8.0, 150 mM NaCland 0.1% Tween 20. The antibodies used were ANG mAb 26-2F (5 mg/ml),affinity-purified ANG pAb R113 (2 mg/ml), affinity-purified RNH1 pAb R127(2 mg/ml), PABP mAb 10E10 (Abcam, cat #ab6125, 1 mg/ml) and TIA1 mAb(Abcam, cat #ab2712, 1 mg/ml). The secondary antibodies used were Alexa-Fluor-488-conjugated goat anti-rabbit or Alexa-Fluor-555-conjugated goat anti-mouseF(ab9)2 (1:1000 dilution). Confocal microscopy was performed with a confocallaser-scanner microscope SP5 Leica and by Zeiss LSM 410. For SP5 Leica, thelambda of the argon ion laser and HeNe laser was set at 488 nm and 546 nm,respectively. Fluorescence emission was revealed by band pass 500–530 and 560–650, respectively, for Alexa Fluor 488 and Alexa Fluor 546. For Zeiss LSM 410,ArKr Laser (488/568/647) was used. Band pass 510–525 and 590 and 610 wasused for Alexa Fluor 488 and Alexa Fluor 555.

FRET

The acceptor photo-bleaching method was used. The donor (Alexa Fluor 488) wasexcited at 488 nm and detected at 500–530 nm. The acceptor (Cy3) bleaching wasperformed at 592 nm. FRET was measured by the increase of Alexa Fluor 488fluorescence intensity after Cy3 photo-bleaching. Measurements were performedon ROI in cytoplasm (n510) under growth and in SG under stress (n57)conditions. To ensure reproducibility and reliability of Alexa Fluor 488fluorescence measurements, Cy3 was photo-bleached to ,10% of its initialfluorescence. Efficiency of FRET was calculated as E5(IDA2ID)/ID where ID andIDA are fluorescence intensities before and after photo-bleaching.

RNH1 knockdown

A set of human RNH1-specific shRNA cloned in pLKO.1 lentiviral vector waspurchased from Open Biosystems. Lentiviral particles were packaged in HEK293Tcells co-transfected with shRNA inserted pLKO.1 (5.8 mg), envelope plasmidpMD2.G (1.8 mg) and packaging plasmid psPAX (4.4 mg). HeLa cells wereinfected with lentivirus in the presence of 10 mg/ml polybrene for 48 hours.Puromycin-resistant cells were selected and the level of RNH1 was examined byImmunoblot analysis.

EB and AO staining of apoptotic cells

Apoptotic cells were identified by EB and AO staining as described (Ribble et al.,2005). Cells before or after SA treatment (1 mM for 1 hour) were trypsinized,pelleted and washed with 4 C̊ PBS. The cells were resuspended in 100 ml of 4 C̊PBS and mixed with 5 ml of the EB/AO dye mixture (100 mg/ml each of AO andEB in PBS) at 37 C̊ for 20 minutes. Stained cells were placed on a microscopeslide and covered with coverslips.

AcknowledgementsWe thank Daria M. Monti, Rita Del Giudice, Giuseppina Fusco andHailing Yang for helpful discussions and technical assistance.

Author contributionsE.P., C.S., J.S., S.F., F.F., P.N. and E.Y. performed experiments.E.P., G.D. and G.-f.H. conceived ideas, analyzed and interpretedresults. G.-f.H. wrote the manuscript.

FundingThis work was supported in part by Italian Ministry of University andby the National Institutes of Health [grant numbers R01 NS065237and R01 CA105241 to G.-f.H.]. Deposited in PMC for release after12 months.

Supplementary material available online at

http://jcs.biologists.org/lookup/suppl/doi:10.1242/jcs.134551/-/DC1

ReferencesAnderson, P. and Kedersha, N. (2008). Stress granules: the Tao of RNA triage. Trends

Biochem. Sci. 33, 141-150.

Anderson, P. and Kedersha, N. (2009). Stress granules. Curr. Biol. 19, R397-R398.

Baird, S. D., Turcotte, M., Korneluk, R. G. and Holcik, M. (2006). Searching forIRES. RNA 12, 1755-1785.

Bella, J., Hindle, K. L., McEwan, P. A. and Lovell, S. C. (2008). The leucine-richrepeat structure. Cell. Mol. Life Sci. 65, 2307-2333.

Bevilacqua, E., Wang, X., Majumder, M., Gaccioli, F., Yuan, C. L., Wang, C., Zhu,

X., Jordan, L. E., Scheuner, D., Kaufman, R. J. et al. (2010). eIF2alphaphosphorylation tips the balance to apoptosis during osmotic stress. J. Biol. Chem.

285, 17098-17111.

Blackburn, P., Wilson, G. and Moore, S. (1977). Ribonuclease inhibitor from humanplacenta. Purification and properties. J. Biol. Chem. 252, 5904-5910.

Buchan, J. R. and Parker, R. (2009). Eukaryotic stress granules: the ins and outs oftranslation. Mol. Cell 36, 932-941.

Conforti, F. L., Sprovieri, T., Mazzei, R., Ungaro, C., La Bella, V., Tessitore, A.,

Patitucci, A., Magariello, A., Gabriele, A. L., Tedeschi, G. et al. (2008). A novelAngiogenin gene mutation in a sporadic patient with amyotrophic lateral sclerosisfrom southern Italy. Neuromuscul. Disord. 18, 68-70.

Cui, X. Y., Fu, P. F., Pan, D. N., Zhao, Y., Zhao, J. and Zhao, B. C. (2003). Theantioxidant effects of ribonuclease inhibitor. Free Radic. Res. 37, 1079-1085.

de Wit, J., Hong, W., Luo, L. and Ghosh, A. (2011). Role of leucine-rich repeatproteins in the development and function of neural circuits. Annu. Rev. Cell Dev. Biol.

27, 697-729.

Emara, M. M., Ivanov, P., Hickman, T., Dawra, N., Tisdale, S., Kedersha, N., Hu,

G. F. and Anderson, P. (2010). Angiogenin-induced tRNA-derived stress-inducedRNAs promote stress-induced stress granule assembly. J. Biol. Chem. 285, 10959-10968.

Fominaya, J. M. and Hofsteenge, J. (1992). Inactivation of ribonuclease inhibitor bythiol-disulfide exchange. J. Biol. Chem. 267, 24655-24660.

Fu, H., Feng, J., Liu, Q., Sun, F., Tie, Y., Zhu, J., Xing, R., Sun, Z. and Zheng,X. (2009). Stress induces tRNA cleavage by angiogenin in mammalian cells. FEBS

Lett. 583, 437-442.

Furia, A., Moscato, M., Calı̀, G., Pizzo, E., Confalone, E., Amoroso, M. R., Esposito,

F., Nitsch, L. and D’Alessio, G. (2011). The ribonuclease/angiogenin inhibitor isalso present in mitochondria and nuclei. FEBS Lett. 585, 613-617.

Gellera, C., Colombrita, C., Ticozzi, N., Castellotti, B., Bragato, C., Ratti, A.,Taroni, F. and Silani, V. (2008). Identification of new ANG gene mutations in alarge cohort of Italian patients with amyotrophic lateral sclerosis. Neurogenetics 9,33-40.

Greenway, M. J., Andersen, P. M., Russ, C., Ennis, S., Cashman, S., Donaghy, C.,Patterson, V., Swingler, R., Kieran, D., Prehn, J. et al. (2006). ANG mutationssegregate with familial and ‘sporadic’ amyotrophic lateral sclerosis. Nat. Genet. 38,411-413.

Haigis, M. C., Kurten, E. L. and Raines, R. T. (2003). Ribonuclease inhibitor as anintracellular sentry. Nucleic Acids Res. 31, 1024-1032.

Ibaragi, S., Yoshioka, N., Kishikawa, H., Hu, J. K., Sadow, P. M., Li, M. and Hu,G. F. (2009). Angiogenin-stimulated rRNA transcription is essential for initiation andsurvival of AKT-induced prostate intraepithelial neoplasia. Mol. Cancer Res. 7, 415-424.

Ivanov, P., Emara, M. M., Villen, J., Gygi, S. P. and Anderson, P. (2011).Angiogenin-induced tRNA fragments inhibit translation initiation. Mol. Cell 43, 613-623.

Johnson, R. J., McCoy, J. G., Bingman, C. A., Phillips, G. N., Jr and Raines,

R. T. (2007). Inhibition of human pancreatic ribonuclease by the human ribonucleaseinhibitor protein. J. Mol. Biol. 368, 434-449.

Kim, Y. N. and Kim, H. (2012). Decreased serum angiogenin level in Alzheimer’sdisease. Prog. Neuropsychopharmacol. Biol. Psychiatry 38, 116-120.

Kim, Y. J., Park, S. J., Choi, E. Y., Kim, S., Kwak, H. J., Yoo, B. C., Yoo, H., Lee,S. H., Kim, D., Park, J. B. et al. (2011). PTEN modulates miR-21 processing viaRNA-regulatory protein RNH1. PLoS ONE 6, e28308.

Kishimoto, K., Liu, S., Tsuji, T., Olson, K. A. and Hu, G. F. (2005). Endogenousangiogenin in endothelial cells is a general requirement for cell proliferation andangiogenesis. Oncogene 24, 445-456.

Kobe, B. and Deisenhofer, J. (1993). Crystal structure of porcine ribonucleaseinhibitor, a protein with leucine-rich repeats. Nature 366, 751-756.

Kobe, B. and Deisenhofer, J. (1995). A structural basis of the interactions betweenleucine-rich repeats and protein ligands. Nature 374, 183-186.

Kobe, B. and Deisenhofer, J. (1996). Mechanism of ribonuclease inhibition byribonuclease inhibitor protein based on the crystal structure of its complex withribonuclease A. J. Mol. Biol. 264, 1028-1043.

Lee, J. E. and Raines, R. T. (2008). Ribonucleases as novel chemotherapeutics: theranpirnase example. BioDrugs 22, 53-58.

Lee, F. S., Shapiro, R. and Vallee, B. L. (1989). Tight-binding inhibition of angiogeninand ribonuclease A by placental ribonuclease inhibitor. Biochemistry 28, 225-230.

Li, S. and Hu, G. F. (2010). Angiogenin-mediated rRNA transcription in cancer andneurodegeneration. Int. J. Biochem Mol. Biol. 1, 26-35.

Li, S. and Hu, G. F. (2012). Emerging role of angiogenin in stress response and cellsurvival under adverse conditions. J. Cell. Physiol. 227, 2822-2826.

McLaughlin, R. L., Phukan, J., McCormack, W., Lynch, D. S., Greenway, M.,Cronin, S., Saunders, J., Slowik, A., Tomik, B., Andersen, P. M. et al. (2010).Angiogenin levels and ANG genotypes: dysregulation in amyotrophic lateralsclerosis. PLoS ONE 5, e15402.

Monti, D. M., Montesano Gesualdi, N., Matousek, J., Esposito, F. and D’Alessio,

G. (2007). The cytosolic ribonuclease inhibitor contributes to intracellular redoxhomeostasis. FEBS Lett. 581, 930-934.

Moroianu, J. and Riordan, J. F. (1994). Nuclear translocation of angiogenin inproliferating endothelial cells is essential to its angiogenic activity. Proc. Natl. Acad.

Sci. USA 91, 1677-1681.

Nazar, R. N. (2004). Ribosomal RNA processing and ribosome biogenesis ineukaryotes. IUBMB Life 56, 457-465.

Papageorgiou, A. C., Shapiro, R. and Acharya, K. R. (1997). Molecular recognitionof human angiogenin by placental ribonuclease inhibitor—an X-ray crystallographicstudy at 2.0 A resolution. EMBO J. 16, 5162-5177.

Journal of Cell Science 126 (18)4318

Journ

alof

Cell

Scie

nce

Paubel, A., Violette, J., Amy, M., Praline, J., Meininger, V., Camu, W., Corcia, P.,Andres, C. R., Vourc’h, P.; French Amyotrophic Lateral Sclerosis (ALS) Study

Group (2008). Mutations of the ANG gene in French patients with sporadicamyotrophic lateral sclerosis. Arch. Neurol. 65, 1333-1336.

Ribble, D., Goldstein, N. B., Norris, D. A. and Shellman, Y. G. (2005). A simpletechnique for quantifying apoptosis in 96-well plates. BMC Biotechnol. 5, 12.

Riordan, J. F. (2001). Angiogenin. Methods Enzymol. 341, 263-273.Rutkoski, T. J. and Raines, R. T. (2008). Evasion of ribonuclease inhibitor as a

determinant of ribonuclease cytotoxicity. Curr. Pharm. Biotechnol. 9, 185-199.Saikia, M., Krokowski, D., Guan, B. J., Ivanov, P., Parisien, M., Hu, G. F.,

Anderson, P., Pan, T. and Hatzoglou, M. (2012). Genome-wide identification andquantitative analysis of cleaved tRNA fragments induced by cellular stress. J. Biol.

Chem. 287, 42708-42725.Saxena, S. K., Rybak, S. M., Winkler, G., Meade, H. M., McGray, P., Youle,

R. J. and Ackerman, E. J. (1991). Comparison of RNases and toxins upon injectioninto Xenopus oocytes. J. Biol. Chem. 266, 21208-21214.

Saxena, S. K., Rybak, S. M., Davey, R. T., Jr, Youle, R. J. and Ackerman,

E. J. (1992). Angiogenin is a cytotoxic, tRNA-specific ribonuclease in the RNase Asuperfamily. J. Biol. Chem. 267, 21982-21986.

Shapiro, R. and Vallee, B. L. (1989). Site-directed mutagenesis of histidine-13 andhistidine-114 of human angiogenin. Alanine derivatives inhibit angiogenin-inducedangiogenesis. Biochemistry 28, 7401-7408.

Souquere, S., Mollet, S., Kress, M., Dautry, F., Pierron, G. and Weil, D. (2009).Unravelling the ultrastructure of stress granules and associated P-bodies in humancells. J. Cell Sci. 122, 3619-3626.

Steidinger, T. U., Standaert, D. G. and Yacoubian, T. A. (2011). A neuroprotectiverole for angiogenin in models of Parkinson’s disease. J. Neurochem. 116, 334-341.

Tello-Montoliu, A., Patel, J. V. and Lip, G. Y. (2006). Angiogenin: a review of thepathophysiology and potential clinical applications. J. Thromb. Haemost. 4, 1864-1874.

Thiyagarajan, N., Ferguson, R., Subramanian, V. and Acharya, K. R. (2012).

Structural and molecular insights into the mechanism of action of human angiogenin-

ALS variants in neurons. Nat. Commun. 3, 1121.

Thompson, D. M., Lu, C., Green, P. J. and Parker, R. (2008). tRNA cleavage is a

conserved response to oxidative stress in eukaryotes. RNA 14, 2095-2103.

Tsuji, T., Sun, Y., Kishimoto, K., Olson, K. A., Liu, S., Hirukawa, S. and Hu,

G. F. (2005). Angiogenin is translocated to the nucleus of HeLa cells and is involved

in ribosomal RNA transcription and cell proliferation. Cancer Res. 65, 1352-1360.

van Es, M. A., Diekstra, F. P., Veldink, J. H., Baas, F., Bourque, P. R., Schelhaas,

H. J., Strengman, E., Hennekam, E. A., Lindhout, D., Ophoff, R. A. et al. (2009).

A case of ALS-FTD in a large FALS pedigree with a K17I ANG mutation. Neurology

72, 287-288.

Wu, D., Yu, W., Kishikawa, H., Folkerth, R. D., Iafrate, A. J., Shen, Y., Xin, W.,

Sims, K. and Hu, G. F. (2007). Angiogenin loss-of-function mutations in

amyotrophic lateral sclerosis. Ann. Neurol. 62, 609-617.

Xu, Z. P., Tsuji, T., Riordan, J. F. and Hu, G. F. (2002). The nuclear function of

angiogenin in endothelial cells is related to rRNA production. Biochem. Biophys. Res.

Commun. 294, 287-292.

Xu, Z. P., Tsuji, T., Riordan, J. F. and Hu, G. F. (2003). Identification and

characterization of an angiogenin-binding DNA sequence that stimulates luciferase

reporter gene expression. Biochemistry 42, 121-128.

Yamasaki, S. and Anderson, P. (2008). Reprogramming mRNA translation during

stress. Curr. Opin. Cell Biol. 20, 222-226.

Yamasaki, S., Ivanov, P., Hu, G. F. and Anderson, P. (2009). Angiogenin cleaves

tRNA and promotes stress-induced translational repression. J. Cell Biol. 185, 35-42.

Yoshioka, N., Wang, L., Kishimoto, K., Tsuji, T. and Hu, G. F. (2006). A therapeutic

target for prostate cancer based on angiogenin-stimulated angiogenesis and cancer

cell proliferation. Proc. Natl. Acad. Sci. USA 103, 14519-14524.

RNH1 regulates ANG localization and activity 4319