ADAR1 is required for hematopoietic progenitor cellsurvival via RNA editingRichard XuFenga,b, Matthew J. Boyera,b, Hongmei Shenc, Yanxin Lia,b, Hui Yua,b, Yindai Gaod, Qiong Yanga,e,Qingde Wanga,e,1, and Tao Chenga,b,d,1

aUniversity of Pittsburgh Cancer Institute, bDepartment of Radiation Oncology and eDepartment of Medicine, Division of Hematology and Oncology, cDepartmentof Surgery, Thomas E. Starzl Transplantation Institute, University of Pittsburgh School of Medicine, Pittsburgh, PA 15213; and dInstitute of Hematology, State KeyLaboratory for Experimental Hematology, Chinese Academy of Medical Sciences and Peking Union Medical College, Tianjin 300020, China

Edited by Darwin J. Prockop, Tulane University, New Orleans, LA, and approved August 21, 2009 (received for review March 27, 2009)

Adenosine Deaminase Acting on RNA 1 (ADAR1) is an RNA-editingenzyme that converts adenosine to inosine, following RNA transcrip-tion. ADAR1’s essential role in embryonic development, especiallywithin the hematopoietic lineage, has been demonstrated in knock-out mice. However, a specific role for ADAR1 in adult hematopoieticprogenitor cells (HPCs) remains elusive. In this report, we show thatADAR1 is required for survival of differentiating HPCs as opposed tomore primitive cells in adult mice by multiple strategies targetingfloxed ADAR1 for deletion by Cre recombinase. As a consequence,ADAR1-deficient hematopoietic stem cells (HSCs) were incapable ofreconstituting irradiated recipients although being phenotypicallypresent in the recipient bone marrow. While an effect on HSCs cannotbe completely ruled out, the preferential effect of ADAR1 absence onHPCs over more primitive hematopoietic cells was consistent with theincreased expression of ADAR1 within HPCs, as well as the inability ofADAR1-deficient HPCs to form differentiated colonies and increasedapoptotic fraction during ex vivo culture. Moreover, we have ob-tained direct evidence that ADAR1 functions in HPCs via an RNA-editing dependent mechanism. Therefore, ADAR1 plays an essentialrole in adult hematopoiesis through its RNA editing activity in HPCs.

apoptosis � hematopoietic stem cells

H ierarchical hematopoiesis is a highly regulated process in whichhematopoietic stem cells (HSCs) symmetrically or asymmetri-

cally differentiate into hematopoietic progenitor cells (HPCs) thatare directly responsible for the production of all lineages within theblood (1). The balance between HSCs and HPCs depends on thedevelopmental stage, age, as well as physiological and pathologicalconditions of the organism. Distinct molecular programs in HSCsas opposed to HPCs are the key to maintenance of homeostasiswithin the hematopoietic cascade and their disruption may con-tribute to pathological states, such as leukemogenesis. However,few molecules have been clearly demonstrated to have distinct rolesin HSCs versus HPCs.

RNA editing recodes RNA molecules thereby posttranscription-ally regulating gene function in eukaryotic cells (2, 3). As theenzymes responsible for one type of RNA editing, the adenosinedeaminases acting on RNA (ADAR) convert adenosine (A) resi-dues to inosine (I) specifically on double-stranded RNAs therebygenerating RNA molecules not encoded in the genome (4). Of thethree members of the ADAR family in mammals (5), disruption ofADAR1 in mice leads to embryonic lethality, likely due to defectivehematopoiesis in the fetal liver (6–8). Because ADAR1�/� em-bryos die at 11–12 days post coitus, it was not possible to define therole of ADAR1 during and after definitive hematopoiesis in thesemice. A recent study confirmed the previous finding of dysregulatedhematopoiesis within the fetal liver, as well as adult bone marrowby ADAR1-deficient cells (9). However, the cellular mechanismunderlying this phenomenon remains elusive, in particular the roleof ADAR1 in HPCs as opposed to HSCs. By conditional deletionof floxed ADAR1 alleles we have demonstrated the inability ofADAR1-deficient HSCs to reconstitute irradiated recipient bonemarrow despite their engraftment in the bone marrow. ADAR1-

deficient HPCs failed to form differentiated colonies ex vivo, mostlikely due to increased apoptosis within these cells and a lack ofADAR1’s RNA editing activity. HPCs are therefore dependent onADAR1 for their survival and function during adult hematopoiesis.

ResultsConditional Deletion of ADAR1 in Hematopoietic Cells. To determinethe efficiency and functional consequence of ADAR1 deletion byCre recombinase, hematopoietic cells isolated from 16-week oldmice harboring floxed ADAR1 alleles (ADAR1lox/lox) (8) weretransduced with a Murine Stem Cell Virus (MSCV) (10) carryingCre and GFP (MSCV-Cre) or GFP alone (MSCV) (Fig. 1 B and C).PCR confirmed efficient deletion of ADAR1lox/lox alleles in hema-topoietic cells transduced with MSCV-Cre (ADAR1�/�) (Fig. 1D).Moreover, 82% of single sorted HSC-enriched GFP�Lineage�c-Kit�Sca-1� (LKS�) cells isolated by micromanipulation were foundto be ADAR1�/� as early as 24 to 30 h after MSCV-Cre transduc-tion (Fig. 1E). To functionally validate ADAR1 deletion in hema-topoietic cells, in vitro proliferation of transduced Lin� bonemarrow cells was measured. While the total cell number in theADAR1lox/lox group increased 5-fold over 1 week of culture,ADAR1�/� cells decreased in number by 75% within the first 4 daysof culture (Fig. 1F) similar to the results observed in previousstudies (6, 9). Thus retroviral delivery of Cre recombinase effi-ciently deleted ADAR1 in hematopoietic cells.

Lack of Hematopoietic Reconstitution but Preservation of PhenotypicHSCs in ADAR1�/� Transplant Recipients. To study the specific roleof ADAR1 in HSCs versus HPCs, the in vivo reconstitutingability of ADAR1�/� cells was determined by transplantationinto irradiated NOD.Cg-Prkdcscid IL2rgtm1Wjl/SzJ (NOD/SCID-�cnull) (11) mice due to the mixed background of the ADAR1lox/

lox mice and unavailability of a congenic strain. One hundredthousand to 1.5 � 105 GFP� ADAR1�/� or ADAR1lox/lox bonemarrow cells were transplanted into sublethally irradiated (3.5Gy) NOD/SCID-�cnull recipients and multilineage engraftmentin the peripheral blood was monitored for up to 3 months.NOD/SCID-�cnull mice transplanted with ADAR1lox/lox cellstransduced with MSCV or wild-type cells transduced withMSCV-Cre were used as controls. While control cells expressingADAR1 were able to reconstitute up to 90% of the peripheralblood in transplant recipients (Fig. 2 A and B), engraftment ofADAR1�/� cells was less than 1% in all host mice (Fig. 2 A).Engraftment levels in the thymus and spleen were consistent

Author contributions: R.X., H.S., Q.W., and T.C. designed research; R.X., M.J.B., H.S., Y.L.,H.Y., Y.G., and Q.Y. performed research; H.S. contributed new reagents/analytic tools; R.X.,Q.W., and T.C. analyzed data; and R.X., M.J.B., Q.W., and T.C. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

1To whom correspondence may be addressed: chengt@pumc.edu.cn orwangq2@upmc.edu.

This article contains supporting information online at www.pnas.org/cgi/content/full/0903324106/DCSupplemental.

www.pnas.org�cgi�doi�10.1073�pnas.0903324106 PNAS � October 20, 2009 � vol. 106 � no. 42 � 17763–17768

CELL

BIO

LOG

Y

with that of the peripheral blood (Table S1). Flow cytometricanalysis of the blood (Fig. 2C) and flow and histological studiesof the spleen (Fig. S1) showed an absence of ADAR1�/� cells inthe periphery. Bone marrow engraftment was significantlyhigher in control recipients compared to ADAR1�/� recipients,in which the donor-derived cells were less than 0.1% (Fig. 2 Dand E and Table S1). A defect in HSC homing was not observedas CFSE-labeled ADAR1�/� cells were present in the bonemarrow 1 day following transplantation (Fig. S2).

Interestingly, within this small population of donor-derived cellsin the bone marrow of ADAR1�/� recipients, the percentage ofHSC-enriched LKS� cells was more than 40-fold higher than thatin control recipients (Fig. 2 F and G). However, the absolutenumber of LKS� cells was not statistically different between the twogroups (Fig. 2H). A similar lack of hematopoietic reconstitutiondespite the presence of phenotypically-defined HSCs in the bone

marrow was observed when HSC-enriched ADAR1�/� LKS� cellswere transplanted into irradiated recipients (Fig. S3).

Given that the LKS phenotype becomes less reliable for HSCidentification in transplant recipients (12) and retroviral transductioncanhaveanegative impactonHSCs(13),SLAMmarkers (14–16)wereused to quantify long term repopulating HSCs in mice transplantedwith cells carrying floxed ADAR1 alleles and Cre driven by an estrogenresponsive promoter (ER-Cre-ADARlox/lox). Fifteen hundred LKS�

cells heterogenous for CD45 isoform expression (CD45.1�/CD45.2�)from ER-Cre-ADAR1lox/lox or ER-Cre-ADAR1lox/wild mice as a con-trol were sorted and transplanted into 10-Gy irradiated C57/BL6recipients together with 100,000 competitor cells (CD45.2). Threemonths after transplantation, floxed ADAR1 alleles were deleted byoral administration of tamoxifen. One month later, peripheral bloodengraftment in ER-Cre-ADAR1lox/lox recipients was significantly de-creased compared to mice transplanted with ER-Cre-ADAR1lox/wild

cells (Fig. S4) similar to the engraftment observed with ADAR1lox/lox

hematopoietic cells transduced with MSCV-Cre (Fig. 2A). Six monthsafter ADAR1 deletion, the recipients were killed and a 22 to 240-foldincrease in the frequency of SLAM (Lin�CD48�CD150�) donor-derivedHSCs in thebonemarrowwasobservedeventhoughtheoverallbone marrow engraftment was �1% in ER-Cre-ADAR1lox/lox recipi-ents (Fig. 2 I and J). Consistent with ADAR1�/� transplant recipients(Fig. 2H), an almost identical absolute number of donor-derivedSLAM HSCs in the bone marrow of ER-Cre-ADAR1lox/lox and ER-Cre-ADAR1lox/wt recipient groups was found (Fig. 2K). Together, thisin vivo data from distinct gene deletion strategies and HSC phenotypesstrongly suggest that ADAR1-deficient HSCs engrafted and weresustained in the bone marrow although the production or maintenanceof mature blood cells in the recipients was disrupted.

A Selective Effect of ADAR1 Absence on HPCs Versus More PrimitiveCells. To understand the observed preferential dependence ofHSC progeny rather than HSCs themselves on ADAR1 follow-ing transplantation, the expression of ADAR1 in hematopoieticcells at different stages of maturation was determined by real-time RT-PCR. ADAR1 expression was relatively low in moreprimitive HSCs (CD34�LKS�) and mature bone marrow pop-ulations as compared to progenitor cell types (Fig. 3A). In-creased expression of ADAR1 at the protein level in HPCs byintracellular staining analyzed by flow cytometry (Fig. S5)further suggested a preferential role of ADAR1 within HPCs.

To define this association functionally, sorted Lin�c-Kit� bonemarrow cells were transduced with MSCV-Cre or MSCV and singleHSC enriched GFP�LKS� cells were sorted and deposited intoTerasaki plates by micromanipulation. Similar proliferative rateswere observed in ADAR1�/� and ADAR1lox/lox LKS� cells withinthe first 72 h of culture (Table 1). In contrast, there was limitedgrowth of the HSC-depleted and HPC-enriched Lin�c-Kit�Sca-1�

(LKS�) cells (17) upon ADAR1 deletion during 3 days of liquidculture (Fig. 3B), a result similar to that found with the previousdata derived from liquid culture of total GFP� cells followingtransduction (Fig. 1F). Assessment of the in vitro proliferation ofER-Cre-ADAR1lox/lox cells following tamoxifen induction of Creexpression confirmed the selective adverse effect of ADAR1absence on LKS� cells over more primitive LKS� cells (Fig. S6).

Furthermore, an equal number of sorted ADAR1�/� orADAR1lox/lox LKS� HPCs were seeded into semisolid methylcel-lulose medium, and the number of colonies examined after 7 daysof culture. ADAR1�/� LKS� cells yielded a strikingly lower numberof colonies, as compared to ADAR1lox/lox LKS� cells (Fig. 3C).Most colonies formed in the ADAR1�/� group were tiny and of anatypical morphology (Fig. 3D). Those atypical colonies werepicked, cytospun, and stained with May-Grunward/Giemsa. Thecells compromising the colony were also small and atypical inmorphology, as compared to the multilineage differentiated cells inthe ADAR1lox/lox group (Fig. 3E). The cells appeared to be dead,

E11E10E9E8E7E4 E5E3 E6

loxP

E10E9E8 51E41E7EE4 E5E3 E6 E12 E13

loxP loxP

Cre

E11

LTR GFP LTRPGKMSCV

LTR Cre GFP LTRIRESMSCV-Cre

Floxed

Deleted

MSCV MSCV-Cre

48h 48h72h 72h

Day 2 Day 3

Transplantation

Day 4

Enriched Lin-population

Day 0 Day 1 In vitro culture

Sorted

3 months

7 daysRecovery in fresh medium

1st and 2nd RetroviralTransduction

Pre-stimulatedwith cytokines

FloxedDeleted

Posit

iveco

ntro

lsNeg

ative

cont

rol

Indi

vidu

alCe

lls

2 day 3 day 4 day 7 day

ADAR1lox/lox

ADAR1 /

GFP

+C

ells

(x10

5 )

ADAR1 /ADAR1lox/lox

A

B

F

E

D

C

Fig. 1. Conditional gene deletion of ADAR1 in adult hematopoietic cells. (A)Relative positions of the Lox/P elements within the ADAR1 catalytic domains areshown before and after Cre recombination. (B) Diagram of the MSCV andMSCV-Cre vectors. LTR: long terminal repeat; �: packaging signal; PGK: phospho-glycerate kinase promoter; IRES: internal ribosome entry site; GFP: green fluo-rescent protein. (C) Retroviral transduction procedure and experimental design.(D) Confirmation of ADAR1 gene deletion by Cre in whole cell populations byPCR; floxed and deleted ADAR1 alleles (8) are indicated. (E) A representative PCRresult for ADAR1 deletion within single GFP�LKS� cells 24–30 h following Cretransduction.Oneofthree independentexperimentswithsimilar results is shownhere. (F) Impairedgrowthofhematopoietic cells invitro intheabsenceofADAR1.The GFP� cell number in liquid culture (IMDM containing 10% FBS, SCF, Flt3ligandandTPO)wasmonitored, startingat48hafter transductionand lastingfor7 days. Data shown are mean � SD in triplicate wells from one of four experi-ments. Cell morphology at the seventh day of culture is shown under the curve.

17764 � www.pnas.org�cgi�doi�10.1073�pnas.0903324106 XuFeng et al.

which suggests that the progenitor cells were able to proliferate atthe beginning and then underwent apoptosis during culture.

Increased Apoptosis of Differentiating HPCs Following ADAR1 Dele-tion. Our previous studies demonstrated that apoptosis is theprimary cellular mechanism underlying the embryonic lethalityof ADAR1�/� mice (6, 8). To determine whether apoptosiscould explain the decrease in colony formation upon ADAR1deletion (Fig. 3C) and absence of peripheral engraftment inADAR1�/� reconstituted mice, ADAR1lox/lox bone marrow cellswere transduced with MSCV or MSCV-Cre and cultured for 4days. The number of Lin� cells diminished rapidly in the absenceof ADAR1, whereas it continuously expanded in the ADAR1lox/

lox group during culture (Fig. 3F). In contrast, the absolutenumber of more primitive Lin�ScaI� ADAR1�/� cells was

relatively steady during the culture period (Fig. 3F) and thepercentage of Lin�ScaI� cells was not significantly differentbetween the ADARlox/lox and ADAR1�/� groups (Fig. 3G).Apoptosis was measured at the end of 4 days of culture by aTerminal uridine deoxynucleotidyl transferase Nick End Label-ing (TUNEL) assay. As compared to ADAR1lox/lox cells,ADAR1�/� Lin� cells exhibited a 10-fold increase in apoptosis,while more primitive Lin� cells only showed a slight increase inthe percentage of apoptotic cells (Fig. 3 H and I). An increasedrate of apoptosis in the differentiated cells was confirmed byAnnexin V staining (Fig. S7). Furthermore, senescence wasunlikely involved due to equivalent �-galactosidase staining inboth Lin- and Lin� populations of cells (Fig. S8). Therefore, anincrease in apoptosis rather than senescence was largely respon-sible for the decrease of differentiating HPCs following ADAR1

ADAR1 Control

100

10

1

0.1

0.01

0.001

Perip

hera

l Blo

od

Engr

aftm

ent (

%)

A B

Mul

tilin

eage

Engr

aftm

ent i

n C

ontr

ol (%

)

CD3 B220 CD11b

8070605040302010

CD45.1

GFP

ADAR1lox/lox ADAR1 /C

ADAR1 Control

E FD100

1

0.01BM

Eng

raftm

ent (

%)

Mul

tilin

eage

Engr

aftm

ent

in C

ontr

ol (%

)

CD3 B220 CD11b

100

80

60

40

20

0

G**

ADAR1 Control

20

10

0

Don

or-D

eriv

edLK

S+C

ells

(%)

ADAR1 Control

2

1

0

Abs

olut

e Yi

eld

of

Don

or-D

eriv

ed L

KS+

Cel

ls (x

104 )

H I

ADAR1 ADAR1 wt

50

40

30

20

10

0

Abs

olut

e Yi

eld

of D

onor

-

deriv

ed S

LAM

cel

ls (x

103 )KJ

**

5

4

3

2

1

0

Don

or-d

eriv

edSL

AM

cel

ls (%

)

ADAR1 ADAR1 wt

Fig. 2. Elimination of hematopoietic reconstitution and preservation of phenotypic hematopoietic stem cells in transplant recipients upon ADAR1 deletion. (A) Totalpercentage engraftment of GFP� cells in the peripheral blood 9–12 weeks following transplantation. There is a significant difference between ADAR1�/� and control groups(P � 0.001, by Mann-Whitney, n � 9–13). The data were pooled from three independent experiments. (B) Multilineage distribution of donor-derived cells that was onlydetectable in thecontrolgroupdueto lowoverall engraftmentofADAR1�/� cells. (C)Representativeflowcytometricanalysisofperipheralbloodengraftment inADAR1lox/lox

and ADAR1�/ � recipients. The donor-derived cells are CD45.1�GFP� positive, as indicated in the upper square. (D) Percent bone marrow engraftment in mice killed 13 weeksafter transplantation on a log scale. n � 9 and 11 for ADAR1�/� and control groups respectively (P � 0.001 by Mann-Whitney test). (E) Multilineage analysis of donor derivedcells in the bone marrow for the control recipients. (F) Increased percentage of primitive cells in ADAR1�/� transplant recipients. Flow cytometric plots of Lin� gateddonor-derivedLKS� cells, indicated intheoverlaidsquare, fromcontrolandADAR1�/� recipients13weeksafter transplantation.Thenumbershownis thepercentageofLKS�

cells in thetotaldonor-derivedcellpopulation. (G)AveragepercentageofLKS� cellswithin thedonor-derivedcells (P�0.01,n�6–7). (H) TheabsoluteyieldofdonorderivedLKS�cells intransplantrecipients.ThereisnostatisticaldifferencebetweenthetwogroupsbyStudent’sttest.(I) IndependentassessmentforthepreservationofHSCsfollowingtransplantation by SLAM markers. Representative flow cytometric plots show long-term (6 month) engraftment of ADAR1�/wt and ADAR1�/� HSCs defined by the SLAMmarkers (Lin�CD48�CD150�). The number in the square indicates the percentage of HSCs within the donor derived Lin� cells. (J) The frequency of SLAM HSCs within thedonor-derived cell population (P � 0.01, n � 3–4). (K) Absolute yield of SLAM HSCs in recipient mice (P � 0.05, n � 3–4).

XuFeng et al. PNAS � October 20, 2009 � vol. 106 � no. 42 � 17765

CELL

BIO

LOG

Y

deletion during culture and likely underlies the decreased colonyformation and in vivo engraftment of ADAR1-deficient cells.

ADAR1 Functions in Hematopoietic Cells via Its RNA Editing Domain.ADAR1 consists of a catalytic RNA editing domain, as well as RNAand Z-DNA binding domains, which are believed to mediate different

functions of ADAR1 (8). To determine whether the RNA editingactivity of ADAR1 is necessary for its function in hematopoietic cells,the ability of an MSCV-delivered wild type or mutant ADAR1 lackingRNA editing activity (Fig. 4A) to rescue the colony forming ability ofER-Cre-ADAR1�/� HPCs was determined (Fig. 4B). Following theinduction of Cre expression with 30 �M 4-hydroxytomoxifen, trans-

Rel

ativ

eEx

pres

sion

**

**** *

LT-HSC ST-HSC CMP GMP MEP CLP CD11b B220 CD3

2.0

1.6

1.2

0.8

0.4

0

A

Fold

incr

e ase

***

ADAR1 lox/lox

6

5

4

3

2

1

0

B C

CF C

s/1

000

cells

***

ADAR1 ADAR1 lox/lox

100

80

60

40

20

0

40x 100x

AD

AR

1A

DA

R1

AD

AR

1lo

x/lo

x

D

1000x

AD

AR

1lo

x/lo

x

E

F

Cel

lNum

ber(

x104

)

ADAR1lox/lox

2 3 4

250

200

150

100

50

0

Days in Culture

Cel

lNum

ber(

x104

)

Lin-Sca-1+

Lin+

ADAR1∆ /∆

2 3 4Days in Culture

60

50

40

30

20

10

0

MSCVMSCV-Cre

Lin-

Sca1

+ce

lls(%

)

ADAR1lox/lox

ADAR1∆ /∆

2 3 4

Days in Culture

30

25

20

15

10

5

0

TUNEL DAPI MergedLin+ cells

AD

AR

1lox

/lox

(8/2

89)

AD

AR

1∆/∆

AD

AR

1∆/∆

(88/

258)

G I

HTUNEL DAPI Merged

Lin- cells

AD

AR

1lox

/lox

(11/

222)

(21/

224)

Apo

ptot

icR

atio

(%)

Lin+ Lin-

4035302520151050

Fig. 3. Preferential impact of ADAR1 deletion on differentiating hematopoietic progenitor cells. (A)The gene expression pattern of ADAR1 in hematopoieticcells at different stages. The mRNA levels of ADAR1 were examined using real-time RT-PCR in different HSC/HPC subsets and mature hematopoietic cellpopulations. LT-HSCs, long-term hematopoietic stem cells; ST-HSCs, short-term hematopoietic stem cells; CMPs, common myeloid progenitors; CLPs, commonlymphoid progenitors; GMPs, granulocyte-monocyte progenitors; MEPs, megakaryocyte-erythrocyte progenitor; CD11b, peripheral blood myeloid cells; B220,peripheral blood B cells; CD3, peripheral blood T cells. Data shown are summarized from 6 experiments. The expression level of ADAR1 in LT-HSC is significantlylower than that in ST-HSC and different HPC populations (*, P � 0.05; **, P � 0.01). (B) The proliferation of LKS� HPCs in liquid culture within 3.5 days (threepooled independent experiments; ***, P � 0.001). (C) The colony forming ability of LKS� HPCs following 7 days of culture (three pooled independentexperiments, ***, P � 0.001). (D) The morphology of colonies examined in panel C. (E) May-Grunwald/Giemsa stained cells from the individual colonies. (F) Theexpansion of Lin�ScaI� and Lin� cell populations in the presence or absence of ADAR1 during 4 days of in vitro culture. The data shown are from one of twoindependent experiments with similar results. (G) The percentage of GFP�Lin�ScaI� cells within ADAR1lox/lox and ADAR1�/� groups. (H) TUNEL staining of sortedLin� and Lin� cells at 48 h after transduction. Apoptotic cells are labeled by red fluorescence and the DAPI-stained nucleus is shown in blue. The numbers in theparentheses to the left of the images indicate the positive cell number vs. total cell number counted in four fields. (I) Percentage of TUNEL� cells shown in panelH. More than 200 cells in four different fields were counted on each slide.

17766 � www.pnas.org�cgi�doi�10.1073�pnas.0903324106 XuFeng et al.

duced wild-type ADAR1 (p150) rescued more than half of the colonyforming ability of HPCs. There were significantly fewer coloniesgenerated from cells transduced with ADAR1 (E/A) harboring a pointmutation in its RNA editing catalytic site. The morphology of coloniesgenerated from cells without ADAR1 or expressing E/A was similar tothose generated by ADAR1�/� (compare Figs. 3D and 4D). Thisdemonstrates that the role of ADAR1 in HPCs is largely dependent onits RNA editing function.

DiscussionOur current study demonstrates a relatively selective effect ofADAR1 deletion in HPCs as compared to HSCs. Unlike somehematopoietic transcription factors such as MLL, SCL, andRUNX1 (18–20), ADAR1 is required for adult, but not primitiveembryonic hematopoiesis and is more critical for differentiatingHPCs, as opposed to more primitive HSCs or mature blood cells.

This data, taken together with the fact that the growth of neitherembryonic stem (ES) cells (7, 8) nor differentiated fibroblastic cells,(8, 21) appear to require ADAR1 (Fig. S9) suggests that the survivalof differentiating progenitor cells such as HPCs may be moredependent on ADAR1 than other cell types. While we haveevidence indicating that increased apoptosis may underlie theselective effect of ADAR1 deletion on differentiating HPCs, wecould not formally exclude direct effects of ADAR1 on otherfunctions of HSCs such as self renewal. However, given the equiv-alence of HSC numbers following transplantation of ADAR1lox/lox

and ADAR1�/� cells (Fig. 2 H and K) and efficient homing ofADAR1�/� cells (Fig. S2), it seems unlikely that ADAR1 plays anindispensable role in HSCs.

This work contrasts a recently published paper by Hartner et al.(9) claiming that ADAR1 is an essential regulator of HSC main-tenance. In Hartner’s work (9), there was no definitive evidence forthe effect of ADAR1 on HSCs perhaps due to the fact that theauthors failed to demonstrate successful deletion of ADAR1 in theanalyzed cell populations in the conditional gene-deletion model,thereby jeopardizing their conclusions regarding the cellular mech-anism of ADAR1. In fact, increased HSC abundance in ADAR1-null fetal liver shown in their study would support our current claimthat ADAR1 is more essential in HPCs than HSCs (9). In contrast,we have direct evidence for efficient gene deletion in the cellpopulation (Fig. 1D) as well as at the single cell level (Fig. 1E).Based on this important prerequisite, we have multiple lines ofcompelling evidence consistently demonstrating a selective effect ofADAR1 on differentiating progenitor cells over more primitivecells, that is dependent on its RNA editing function.

As one of the RNA processing mechanisms, adenosine to inosineRNA editing regulates gene activity through modification ofmRNA (22, 23) noncoding RNA (24, 25), and regulatory smallRNAs (e.g., microRNAs) (21, 26–28). The RNAs that code criticalhematopoietic regulators, or the regulatory RNAs themselves, mayrequire ADAR1-mediated editing before their physiological ac-tions in these cells. Therefore, defining the targets of ADAR1 in

Table 1. Cell survival and division rates in single stem cellculture

�1 cell �1 division �3 divisions

24 h 48 h 72 h 24 h 48 h 72 h in 72 h

Experiment oneADAR�/� 57 27 22 15 17 18 4ADARlox/lox 40 28 20 15 19 19 7ADARwt/wt 55 23 16 16 18 18 5

Experiment twoADAR�/� 80 27 26 32 32 34 6ADARlox/lox 81 37 25 41 44 45 10

Single sorted GFP�Lin-Sca-1� cells were picked and deposited into Terasakiplates by micromanipulation, and cell division was monitored daily for up to 1week. Initially, 144 cells were input at specific time point for each group. Thenumber in the table indicates the wells containing live cells (first column) oraccumulative wells with cell divisions. There was no statistical significance amonggroups in terms of survival and division rates according to Chi-Square test(P � 0.05).

ADAR1-p150

ADAR1-E/A

Day 0

Lin- enrichment and pre-stimulation

Day 1

Retroviral Transduction with different form of ADAR1 vectors

Day 2 Day 3 Day 5

Input for CFC culture

Sort GFP+

cells

Count in 10 days

Induction of Cre expression by 30 uM of 4-HT

B

A

***

***

No 4-HT Plus 4-HT

60

50

40

30

20

10

0CFC

s/30

0 0ce

lls

MSCV MSCV p150 E/A

C

Plus

4-H

TN

o4-

HT

MSCV p150 E/A

D

Fig. 4. ADAR1 functions in hematopoietic cells through its RNA editing activity. (A) Confirmation of the cDNA sequence of wild-type (MSCV-p150) orpoint-mutated (MSCV-E/A) ADAR1. The red arrow points to the A to C mutation within the catalytic domain. (B) Experimental design of ADAR1 rescue followinginduction of ADAR1 deletion by tamoxifen. (C) The number of colonies yielded from MSCV-p150 or MSCV-E/A transduced ER-Cre-ADAR1�/� cells. MSCV: MSCVvector only transduced group; p150: MSCV-p150 transduced group; and E/A: MSCV-E/A transduced group; 4-HT: 4-hydroxytomoxifen (***, P � 0.001 by Student’st test). (D) The morphology of colonies counted in panel C.

XuFeng et al. PNAS � October 20, 2009 � vol. 106 � no. 42 � 17767

CELL

BIO

LOG

Y

differentiating hematopoietic cells will likely reveal a class ofmolecules that are critical for hematopoietic regeneration andperhaps also relevant for studies concerning leukemia or otherhematopoietic disorders.

Materials and MethodsMice. ADAR1lox/lox mice in a C57BL/6 and FVB mixed background were generatedin our previous study (8). B6.Cg-Tg(CAG-cre/Esr1)5Amc/J (ER-Cre) mice were pur-chased from Jackson Laboratory and crossed to our ADAR1lox/lox mice which werebred inhouse. Excision of floxed ADAR1 alleles in ER-Cre-ADAR1lox/lox mice wasaccomplished by oral administration of 0.2 mg/g tamoxifen in 0.2 to 0.3 mLvegetable oil daily for 3 days. NOD.Cg-PrkdcscidIL2rgtm1Wjl/SzJ (NOD/SCID-�cnull) mice were bred in the animal facility at University of Pittsburgh CancerInstitute. All mice were handled in accordance to institutional guidelines foranimal care.

Retroviral Production. The Murine Stem Cell Virus retroviral vector expressingboth Cre recombinase and GFP (MSCV-Cre) was a gift from B. Lu (University ofPittsburgh). The control vector expressing GFP (MSCV) alone was generouslyprovided by G. Sauvageau (Universite de Montreal). The ADAR1-p150 and p150-E/A point mutation cDNAs were cloned into MSCV vector and the mutationconfirmed by sequencing. The virus packaging cell line, 293T, was maintained inDulbecco’s Modified Eagle’s Medium (DMEM) (GIBCO Life Technologies) supple-mented with 10% heat-inactivated FBS (HyClone). To produce virus, 6 � 106 293Tcells were transfected with 2.5 �g of each of the packaging vectors pKat andVSV-Gaswellas10�gofMSCVplasmidusingLipofectamine2000(Invitrogen)perthe manufacturer’s instructions. The virus containing supernatant was collectedand filtered with a 0.45 �m filter to remove the packaging cell debris at 48 h and72 h after transfection. The retrovirus titer determined on NIH 3T3 cells was 5 �106 to 1 � 107 IU/mL.

Retroviral Transduction of Murine Hematopoietic Cells. Lineage� (Lin�) bonemarrow cells from 16-week-old ADAR1lox/lox mice were enriched using theMACS™ Streptavidin Kit and separation column (Milteny Biotec) per themanufacturer’s instructions. The cells were prestimulated with 100 ng/mLmSCF, 50 ng/mL Flt-3L, and 25 ng/mL TPO (Peprotech) in Iscove’s ModifiedDulbecco’s Medium (IMDM, Invitrogen) with 10% FBS for 24 h at 37 °C, 5%CO2. Cells were then plated in a 24-well plate precoated with Retronectin(takara Bio Inc.) at 1 to 5 � 105 cells/well with 80% viral supernatant (MOI �10 to 20) and 20% fresh medium supplemented with cytokines and 4 �g/mLpolybrene (Sigma). The plate was then centrifuged at 1,700 rpm for 30 min,and cultured for 12–14 h at 32 °C, 5% CO2. The cells were cultured in freshmedium at 37 °C for 8–10 h before a second round of transduction. Thetransduction of ER-Cre cells is described in Fig. 4B.

Flow Cytometry and Cell Sorting. Cultured and freshly harvested bone marrownucleated cells were stained with lineage (CD3, CD4, CD8, CD11b, CD45R, Gr-1,and Ter119), Sca-1, and/or CD45.1, CD45.2 (BD PharMingen) surface markersand examined by flow cytometry as previously described (17).

CFC Assay. GFP�Lin�Sca-1� cells were sorted 24 h after transduction, resus-pended in semisolid methylcellulose medium M3434 (StemCell Technologies),and plated in triplicate in 24-well plates at 4,000 cells/well. The cells were culturedat 37 °C, 5% CO2 for up to 14 days and colonies counted after 7–11 days in culture.

TUNEL Staining. Forty-eight hours after transduction, GFP�Lin� and GFP�Lin�

cell populations were sorted and 10,000—40,000 cells were immediatelyattached to glass slides by cytospin. Slides were stained with the In Situ CellDeath Detection Kit, TMR Red (Roche), followed by nuclear staining with 5�g/�L DAPI (Sigma) per the manufacturer’s instructions. Fluorescent imageswere taken immediately after staining and positive and total cell numberswere counted for more than 3 fields, or 200 to 300 cells, per slide.

Hematopoietic Cell and HSC Transplantation. Six hours before transplantation, 6-to 8-week-old recipient NOD/SCID-�cnull mice were sublethally (3.5 Gy, 82 cGy/min, 137Cs � radiator) irradiated. 1.0 to 1.5 � 105 transduced GFP�Lin� orGFP�Lin�Sca-1� cells were sorted and injected into the tail veins of recipients.Recipient mice were supplied with sterile food and acidified water. Multilineageengraftment in peripheral blood was monitored monthly for 3 months by flowcytometry.

Real-Time RT-PCR. Five thousand hematopoietic cells from 6- to 8-week-oldC57/Bl6 mice were directly sorted into cell lysis buffer and the RNA was extractedusing Absolute RNA extraction kits (STRATAGENE) according to the manufactur-er’s instruction. cDNA was synthesized using SuperScript III (Invitrogen) andreal-time RT-PCR was performed using Dynamo SYBR green qPCR kit (Finnzymes)on a PTC-200 thermo cycler (MJ Research) as previously described (6).

May-Grunwald/Giemsa Staining. Cytospin slides were stained with May-Grunwald and Giemsa (Sigma) according to the manufacturer’s instructions.

ACKNOWLEDGMENTS. We thank Dr. Kazuko Nishikura for providing theADAR1lox/lox mouse strain; Drs. Binfeng Lu and Guy Sauvageau for providingthe MSCV retroviral vectors; Drs. Beyong-Chel Lee, Fred Moolten, Sid Kar, andWilliam Chambers for their comments on this manuscript; and E. MichaelMeyer for his help in flow cytometry work. This work was supported byNational Institutes of Health Grant RO1-HL70561 and Tianjin National ScienceFoundation (07JCZD JC10600) (to T.C.); the Hillman Foundation (to Q.W.);Scholar Awards from the Leukemia & Lymphoma Society (to T.C.), aChangjiang Scholarship from the Ministry of Education of China (to T.C.), andthe National Science Foundation of China (30825017) (to T.C.).

1. Cheng T (2008) Toward ‘SMART’ stem cells. Gene Ther 15:67–73.2. Benne R, et al. (1986) Major transcript of the frameshifted coxII gene from trypanosome

mitochondriacontains fournucleotides thatarenotencoded intheDNA.Cell46:819–826.3. Benne R (1996) RNA editing: How a message is changed. Curr OpinGen Dev 6:221–231.4. Bass BL (2002) RNA editing by adenosine deaminases that act on RNA. Ann Rev

Biochem 71:817–846.5. Nishikura K (2006) Editor meets silencer: Crosstalk between RNA editing and RNA

interference. Nat Rev 7:919–931.6. Wang Q, Khillan J, Gadue P, Nishikura K (2000) Requirement of the RNA editing

deaminase ADAR1 gene for embryonic erythropoiesis. Science 290:1765–1768.7. Hartner JC, et al. (2004) Liver disintegration in the mouse embryo caused by deficiency

in the RNA-editing enzyme ADAR1. J Bio Chem 279:4894–4902.8. Wang Q, et al. (2004) Stress-induced apoptosis associated with null mutation of ADAR1

RNA editing deaminase gene. J Bio Chem 279:4952–4961.9. Hartner JC, Walkley CR, Lu J, Orkin SH (2009) ADAR1 is essential for the maintenance

of hematopoiesis and suppression of interferon signaling. Nat Immunol 10:109–115.10. Stier S, Cheng T, Dombkowski D, Carlesso N, Scadden DT (2002) Notch1 activation

increases hematopoietic stem cell self-renewal in vivo and favors lymphoid overmyeloid lineage outcome. Blood 99:2369–2378.

11. Ito M, et al. (2002) NOD/SCID/gamma(c) (null) mouse: An excellent recipient mousemodel for engraftment of human cells. Blood 100:3175–3182.

12. Spangrude GJ, Brooks DM, Tumas DB (1995) Long-term repopulation of irradiated micewith limiting numbers of purified hematopoietic stem cells: In vivo expansion of stemcell phenotype but not function. Blood 85:1006–1016.

13. Kittler EL, et al. (1997) Cytokine-facilitated transduction leads to low-level engraft-ment in nonablated hosts. Blood 90:865–872.

14. Kiel MJ, Yilmaz OH, Iwashita T, Terhorst C, Morrison SJ (2005) SLAM family receptorsdistinguish hematopoietic stem and progenitor cells and reveal endothelial niches forstem cells. Cell 121:1109–1121.

15. Yilmaz OH, Kiel MJ, Morrison SJ (2006) SLAM family markers are conserved amonghematopoietic stem cells from old and reconstituted mice and markedly increase theirpurity. Blood 107:924–930.

16. Kim I, He S, Yilmaz OH, Kiel MJ, Morrison SJ (2006) Enhanced purification of fetal liverhematopoietic stem cells using SLAM family receptors. Blood 108:737–744.

17. Yuan Y, Shen H, Franklin DS, Scadden DT, Cheng T (2004) In vivo self-renewing divisionsof haematopoietic stem cells are increased in the absence of the early G1-phaseinhibitor, p18INK4C. Nat Cell Bio 6:436–442.

18. Ernst P, Wang J, & Korsmeyer SJ (2002) The role of MLL in hematopoiesis and leukemia.Curr Opin Hema 9:282–287.

19. Mucenski ML, et al. (1991) A functional c-myb gene is required for normal murine fetalhepatic hematopoiesis. Cell 65:677–689.

20. Rosenbauer F, Tenen DG (2007) Transcription factors in myeloid development: Balanc-ing differentiation with transformation. Nat Rev Imm 7:105–117.

21. Yang W, et al. (2005) ADAR1 RNA deaminase limits short interfering RNA efficacy inmammalian cells. J Bio Chem 280:3946–3953.

22. Seeburg PH, Hartner J (2003) Regulation of ion channel/neurotransmitter receptorfunction by RNA editing. Curr Opin Neurobio 13:279–283.

23. Burns CM, et al. (1997) Regulation of serotonin-2C receptor G-protein coupling by RNAediting. Nature 387:303–308.

24. Levanon EY, et al. (2004) Systematic identification of abundant A-to-I editing sites inthe human transcriptome. Nat Biotech 22:1001–1005.

25. Athanasiadis A, Rich A, Maas S (2004) Widespread A-to-I RNA editing of Alu-containingmRNAs in the human transcriptome. PLoS biology 2:e391.

26. Yang W, et al. (2006) Modulation of microRNA processing and expression through RNAediting by ADAR deaminases. Nat Struct Mol Bio 13:13–21.

27. Blow MJ, et al. (2006) RNA editing of human microRNAs. Gen Bio 7:R27.28. Kawahara Y, et al. (2007) Redirection of silencing targets by adenosine-to-inosine

editing of miRNAs. Science 315:1137–1140.

17768 � www.pnas.org�cgi�doi�10.1073�pnas.0903324106 XuFeng et al.