p21-mediated RNR2 repression restricts HIV-1replication in macrophages by inhibiting dNTPbiosynthesis pathwayAwatef Alloucha,1,2, Annie Davida, Sarah M. Amieb, Hichem Lahouassac, Loïc Chartierd, Florence Margottin-Goguetc,e,f,Françoise Barré-Sinoussia,g, Baek Kimb,h, Asier Sáez-Cirióna, and Gianfranco Pancinoa,g,1

aUnité de Régulation des Infections Rétrovirales, Institut Pasteur, 75015 Paris, France; bCenter for Drug Discovery, Department of Pediatrics, Emory University,Atlanta, GA 30322; cInstitut National de la Santé et de la Recherche Médicale, Unité 1016, Institut Cochin, 75014 Paris, France; dUnité de Recherche etd’Expertise Epidémiologie des Maladies Emergentes, Institut Pasteur, 75015 Paris, France; eCentre National de la Recherche Scientifique, Unité Mixte deRecherche 8104, 75014 Paris, France; fUniversité Paris Descartes, 75006 Paris, France; gInstitut National de la Santé et de la Recherche Médicale, 75013 Paris,France; and hCollege of Pharmacy, Kyung Hee University, Seoul 130-701, South Korea

Edited* by Stephen P. Goff, Columbia University College of Physicians and Surgeons, New York, NY, and approved September 13, 2013 (received forreview April 10, 2013)

Macrophages are a major target cell for HIV-1, and their infectioncontributes to HIV pathogenesis. We have previously shown thatthe cyclin-dependent kinase inhibitor p21 inhibits the replication ofHIV-1 and other primate lentiviruses in human monocyte-derivedmacrophages by impairing reverse transcription of the viral ge-nome. In the attempt to understand the p21-mediated restrictionmechanisms, we found that p21 impairs HIV-1 and simian immuno-deficiency virus (SIV)mac reverse transcription in macrophages byreducing the intracellular deoxyribonucleotide (dNTP) pool to levelsbelow those required for viral cDNA synthesis by a SAM domain andHD domain-containing protein 1 (SAMHD1)-independent pathway.We found that p21 blocks dNTP biosynthesis by down-regulatingthe expression of the RNR2 subunit of ribonucleotide reductase, anenzyme essential for the reduction of ribonucleotides to dNTP. p21inhibits RNR2 transcription by repressing E2F1 transcription factor,its transcriptional activator. Our findings unravel a cellular path-way that restricts HIV-1 and other primate lentiviruses by affect-ing dNTP synthesis, thereby pointing to new potential cellulartargets for anti-HIV therapeutic strategies.

Macrophage infection by HIV-1 contributes to viral patho-genesis and progression to AIDS (1). Because of their

longevity and relative resistance to the cytopathogenic effect ofHIV-1, macrophages contribute to the establishment of viralreservoirs and are able to transmit the virus to other target cells(2–4). HIV-1 infection of macrophages also affects their functionsin innate and adaptive responses to pathogens (5). However,macrophage susceptibility to HIV-1 infection varies according totheir differentiation status and tissue localization (6, 7). Envi-ronmental stimuli such as cytokines, bacterial products, and im-mune complexes also influence macrophage permissiveness toHIV-1 infection (8). HIV-1 replication in macrophages is regu-lated by cellular factors that can either enhance or inhibit varioussteps of the viral replicative cycle (9). In particular, HIV-1 reversetranscription is much slower in monocyte-derived macrophages(MDMs) than in activated T cells. This difference is attributedmainly to the limited availability of nucleotide precursors in thesenondividing cells, in which the average deoxynucleoside triphos-phate (dNTP) concentration (∼26 nM) is about 200 times lowerthan in PHA-activated T cells (10–12). The dNTP triphospho-hydrolase SAM domain and HD domain-containing protein 1(SAMHD1) decreases dNTP levels by hydrolyzing dNTP intodeoxynucleosides (dNs) and triphosphates, thereby limiting HIV-1reverse transcription in dendritic cells, macrophages, and resting Tcells (13–20). Contrary to HIV-1, SIV and HIV-2 have evolveda protein, Vpx, that relieves SAMHD1-mediated restriction bytargeting SAMHD1 for proteasome-dependent degradation (21–23). In addition to SAMHD1, other cellular factors, in particular

antiviral pathways induced by type I IFN, affect early postentrysteps of HIV-1 replication in macrophages (24–27).We have previously shown that p21Waf1/Cip1 (referred to here-

after as p21) restricts HIV replication in MDMs (28). p53-inde-pendent p21 induction by immune complex aggregation of FcγRsleads to a strong reduction in reverse transcripts of HIV-1, HIV-2,SIVmac, and SIVagm in MDMs whereas siRNA-mediated p21depletion rescues viral replication (28). A further minor block ofviral integration by p21 has been also observed (28) that would bea consequence of the major block of reverse transcription leadingto the formation of incomplete reverse transcripts unable to inte-grate properly. The mechanism underlying p21-mediated HIV-1inhibition in macrophages remains to be elucidated. In particular,we did not detect p21 interactions with HIV-1 proteins involvedin reverse transcription/integration, including Viral protein R,matrix, reverse transcriptase, and Integrase (28). Although p21causes a preintegrative HIV-1 blockade in human hematopoieticstem cells (29), it has been reported to either inhibit or enhanceHIV-1 transcription in MDMs and T cells (30–32). However, wedid not observe a significant impact of FcγR-mediated activationon HIV-1 transcription in MDMs (33).

Significance

Macrophages, with CD4+ T lymphocytes, are a major cell targetfor HIV-1 infection. We have previously reported that the in-duction of a cellular protein, the cyclin-dependent kinase p21,inhibits HIV-1 replication in macrophages. We now show thatp21 impairs the reverse transcription of HIV-1 and other pri-mate lentiviruses, including the simian immunodeficiency virus(SIV)mac, by blocking the synthesis of cellular deoxynucleo-tides (dNTP) that are used by retroviral reverse transcriptasefor viral DNA synthesis. p21 represses the expression of a keyenzyme of the dNTP biosynthesis pathway, the RNR2 subunitof the ribonucleotide reductase. Our findings point to newpotential cellular targets for antiretroviral strategies.

Author contributions: A.A. and G.P. designed research; A.A., A.D., S.M.A., B.K., and A.S.-C.performed research; H.L. and F.M.-G. contributed new reagents/analytic tools; A.A., A.D.,S.M.A., H.L., L.C., F.M.-G., F.B.-S., B.K., A.S.-C., and G.P. analyzed data; and A.A. and G.P.wrote the paper.

The authors declare no conflict of interest.

*This Direct Submission article had a prearranged editor.1To whom correspondence may be addressed. E-mail: a.allouch@sns.it or gianfranco.pancino@pasteur.fr.

2Present address: Institut National de la Santé et de la Recherche Médicale U1030, InstitutGustave Roussy, F-94805 Villejuif, France.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1306719110/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1306719110 PNAS | Published online September 30, 2013 | E3997–E4006

MICRO

BIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

Nov

embe

r 3,

202

0

p21 belongs to the Cip/Kip family of cyclin-dependent kinaseinhibitors (CDKIs) (34, 35). As a cell-cycle inhibitor, p21 playsa critical role in the control of cell growth, but it also plays a rolein monocyte differentiation and survival (36–38). p21 influencescell-cycle progression through cyclin/CDK inhibition, binding toPCNA, and control of E2F transcription and activity (39, 40).Although p21 is not a transcription factor, it may exert indirecteffects on cellular gene expression. p21 induction in a fibrosar-coma cell line regulates the transcription of several genes, in-cluding genes involved in DNA replication, segregation, andrepair, either repressing or enhancing their expression (41).Despite the fact that p21 is strongly expressed in terminallydifferentiated, nonreplicating MDMs, there is scarce informationon its impact on gene expression in these cells.Here, we addressed the mechanisms underlying p21-mediated

restriction of the reverse transcription of HIV-1 and related len-tiviruses in macrophages. Specifically, we investigated whether p21might affect the dNTP pool in macrophages by interfering withgenes involved in dNTP metabolic pathways.

Resultsp21 Reduces dNTP Concentrations in MDMs. We have previouslydemonstrated that induction of p21 expression in MDMs throughFcγR aggregation by immobilized Ig (IVIg) inhibits HIV-1 in-fection and also MDM transduction by a VSV-G pseudotypedNL4.3-Luc HIV-1 virus (HIV-1.luc/VSVG) containing a lucifer-ase reporter gene (28). Single-cell bioluminescence (42) showedthat HIV-1 gene expression in infected living MDMs was almostabrogated in IVIg-coated wells with a significant (P < 0.001) re-duction of the number of luciferase-positive cells (N = 0.6 ± 0.8positive cells) in comparison with unstimulated (US) MDMs (N=8.3 ± 2.2 positive cells) (Fig. 1A). The overall effect is reflected bya decrease (>95%) of luciferase activity in the lysates of IVIg-activated MDMs (IVIg) (Fig. 1B). This last measure, more easilyquantifiable, was used in subsequent experiments.We have previously shown that p21 blocks HIV-1 replication

mainly by inhibiting the reverse transcription step (28). Exoge-nous deoxynucleosides (dNs) accelerate HIV-1 reverse tran-scription in MDMs by serving as dNTP precursors and therebyincreasing the intracellular dNTP pool, which is normally a lim-iting factor (11, 43). Exogenous dNs are converted into dNTPsthrough the dNTP biosynthesis salvage pathway, and it has beenreported that p21 can repress genes that encode enzymes in-volved in dNTP biosynthesis (41).We therefore examined whether adding dNs to MDM cultures

counteracted p21 inhibition of viral reverse transcription. Unsti-mulated and IVIg-stimulated MDMs were infected with HIV-1.luc/VSV-G and then treated with increasing amounts of pyrimi-dine and purine dNs (0.25 mM–1 mM). HIV-1 replication wasinhibited by 70% in IVIg-stimulated MDMs in comparison withunstimulated MDMs (Fig. 1C). The addition of dNs increasedHIV-1 replication in unstimulated MDMs and abrogated HIV-1restriction in IVIg stimulated MDM in a concentration-dependentmanner (Fig. 1C), suggesting that p21 inhibits HIV-1 replicationby limiting dNTP pools available for reverse transcription.We then investigated whether p21 expression modulates the

intracellular dNTP pool. Intracellular dNTP concentrations werequantified in unstimulated and in IVIg-stimulated MDMs inwhich p21 had been silenced by transfection with a small in-terfering RNA (siRNA) targeting the p21 gene (sip21). Controlcells were transfected with a nontargeting siRNA pool (siCTL).Intracellular dNTP concentrations were measured with a highlysensitive single-nucleotide incorporation assay (11, 19). Bothabsolute and fold increase dNTP concentrations were deter-mined for the siCTL or sip21 RNA treated MDMs from threedifferent donors. A representative experiment for donor 1 isdepicted in Fig. S1, and the quantification of dATP and dTTPconcentrations for the three donors is presented in Fig. 1D. As

expected, dNTP levels in unstimulated and IVIg-stimulatedMDMs treated with control siRNA (siCTL) were very low (Fig.1D and Fig. S1C) because of the low cellular dNTP level inmacrophages, below or close to the linearity range of quantifi-cation. After p21 knockdown (sip21), dATP and dTTP intra-cellular concentrations increased in MDMs from the threedonors (4.8-, 5.4-, and 4.3-fold for dATP and 1.5-, 2.2-, and 2.6-fold for dTTP in donors 1, 2, and 3, respectively) (Fig. 1D). Insip21-knockdown IVIg-stimulated MDMs, dATP concentrationsincreased of 3.6-, 2.6-, and 1.7-fold for donor 1, 2, and 3, re-spectively, and dTTP concentrations increased 1.9- and 2.8-foldin donors 1 and 3 and remained unchanged in donor 2 (Fig. 1D).These results indicate that p21 interferes with dNTP metabolismin MDMs, decreasing the dNTP pool.

p21 Inhibits the Expression of Ribonucleotide Reductase Subunit R2(RNR2). We then examined whether p21-mediated down-regula-tion of dNTP levels in MDMs was due to interference with dNTPbiosynthesis pathways. We measured the expression of proteinsinvolved in de novo dNTP biosynthesis [RNR1, ribonucleotidereductase subunit R2 (RNR2), and p53R2] by Western blot(WB) in unstimulated and IVIg-stimulated MDM-silenced ornot for p21. The expression of PCNA, a p21-interacting protein,and of GAPDH, a housekeeping protein, was used to normalizeeach sample. As shown in Fig. 2A, IVIg stimulation increasedp21 protein levels, and p21 knockdown (sip21) reduced p21protein levels relative to unstimulated and IVIg-stimulatedMDMs transfected with control siRNA (siCTL). RNR1 andp53R2 expression levels remained unchanged whether p21 levelsincreased or decreased (Fig. 2A). RNR2 protein was depletedupon p21 induction and strongly increased after p21 knockdown(sip21) (Fig. 2A). These data indicated that p21 down-regulatesRNR2 protein expression in MDMs.To determine whether p21 regulates RNR2 expression at the

transcriptional level, RNR2 and p21 mRNAs were quantified byquantitative real-time PCR (qRT-PCR) in the same conditionsas those described above. Results with MDMs from five differentdonors showed that RNR2 mRNA levels fell significantly fol-lowing IVIg stimulation (by 60–90%; P = 0.005), in parallel withthe increase in p21 mRNA levels (Fig. 2B). In contrast, p53R2mRNA levels remained unchanged upon IVIg stimulation (Fig.S2), consistently with the lack of modulation of p53R2 proteinlevels (Fig. 2A). RNR2 mRNA levels increased significantly (5-to 30-fold; P = 0.005) in unstimulated MDMs depleted of p21(sip21)(Fig. 2B). The RNR2 mRNA levels were rescued andincreased by 2- to 15-fold in the IVIg-activated MDMs silencedfor p21. The variations in p21 and RNR2 expressions corre-sponded closely to the effects of p21 induction/depletion onHIV-1 replication in all donor cells tested (Fig. 2C). HIV-1.luc/VSVG replication was inhibited in IVIg-activated MDMs (highp21 and low RNR2 levels) and enhanced in p21-knockdownMDMs (low p21 and high RNR2 levels) (Fig. 2C). These re-sults suggested that p21 inhibits HIV-1 reverse transcriptionby inhibiting RNR2 expression and thereby blocking dNTPbiosynthesis.

RNR2 Depletion Is Sufficient to Inhibit HIV-1 Replication at the ReverseTranscription Step. We detected both RNR2 and p53R2 expres-sions in MDMs, but only RNR2 expression was regulated by p21.We therefore investigated the specific role of RNR2 in p21-mediated inhibition of HIV-1 in macrophages. To evaluate therole of RNR2 in HIV-1 replication in MDMs, RNR2 was de-pleted by transfecting MDMs with a pool of four siRNAs tar-geting the RNR2 gene. Control cells were transfected with a poolof nontargeting siRNAs (siCTL). As shown in Fig. 3A, RNR2siRNAs (siRNR2) depleted the protein in MDMs as efficientlyas IVIg stimulation. Given the high sequence homology of theRNR2 gene with the p53R2 gene (44), p53R2 protein expression

E3998 | www.pnas.org/cgi/doi/10.1073/pnas.1306719110 Allouch et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 3,

202

0

was also checked by WB, which showed that RNR2 siRNAs didnot affect p53R2 expression (Fig. 3A). RNR2 siRNAs reducedRNR2 mRNA levels in unstimulated MDMs by 65–98% inMDMs from seven different donors (P = 0.0008) (Fig. 3B). Incontrast, p21 mRNA levels were not significantly affected byRNR2 knockdown (Fig. 3B). RNR2 knockdown significantlyinhibited viral replication in unstimulated HIV-1.luc/VSVG-

infected cells from all seven donors (−50% to −98% versuscontrol unstimulated MDMs; P = 0.0008) (Fig. 3C), indicatingthat RNR2 is required for HIV-1 replication in MDMs. Inparallel experiments, IVIg activation of MDMs inhibited HIV-1replication to a similar extent (64–94%) (Fig. 3C). HIV-1 latereverse transcription products (LRTs) were quantified by qRT-PCR in MDMs from six of the seven donors. RNR2 knockdown

Fig. 1. Exogenous deoxynucleosides (dNs) rescue p21-mediated restriction of HIV-1 and p21 knockdown increases dNTP levels in MDMs. (A and B) Unsti-mulated (US) and IVIg-activated MDMs were infected with HIV-1.luc/VSVG 2 h after IVIg stimulation. HIV-1 replication was quantified at 5 d postinfection inliving cells by bioluminescence imaging of the infected living MDMs (integration, 80 s) (A). The numbers of positive cells indicated above the images aremeans ± SD of 9 and 13 random fields for US and IVIg MDMs, respectively. In parallel samples, HIV-1 replication was quantified by measuring luciferaseactivity in lysates of infected MDMs (B). (C) Unstimulated (US) and IVIg-activated MDMs were infected with HIV-1.luc/VSVG 2 h after IVIg stimulation and werethen treated with increasing amounts of dNs (0.25–1 mM) for 24 h. HIV-1 replication was determined by measuring luciferase activity 72 h postinfection (hpi).The data are means ± SD of triplicate wells. (D) Unstimulated (US) and IVIg activated MDMs from three donors were p21-silenced by transfection with p21siRNAs (sip21) whereas control MDMs were transfected with a pool of nontargeting siRNAs (siCTL). At 24 h postactivation and 22 h postsilencing, MDMs werelysed for the quantification of intracellular dATP and dTTP using a single nucleotide primer extension gel analysis (11). The graphs represent the mean ofduplicate absolute values in fmoles/million cells of intracellular dATP and dTTP in p21 knockdown (sip21), unstimulated (US), and IVIg-activated MDMs fromthree different donors.

Allouch et al. PNAS | Published online September 30, 2013 | E3999

MICRO

BIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

Nov

embe

r 3,

202

0

reduced LRT levels by 80–93% in unstimulated MDMs (P =0.002) (Fig. 3D), consistent with luciferase inhibition (Fig. 3C). Asimilar reduction in LRTs was observed in IVIg-activated MDMs(70–80%) (Fig. 3D). These results demonstrate that the sup-pression of RNR2 by p21 induction or by siRNA-mediatedknockdown inhibits HIV-1 reverse transcription and indicatethat RNR2 is an essential HIV-1 cofactor that promotes reversetranscription in MDMs.However, p21 might interfere with other molecules involved in

dNTP metabolism, such as kinases of the salvage pathway, whichmay contribute to p21 inhibition of reverse transcription. Wereasoned that, in this case, the effect of RNR2 silencing on HIVreverse transcription could be modulated in cells depleted of p21.We then performed double knockdown experiments of p21 andRNR2 in MDMs and analyzed the effects on HIV-1 infection.Unstimulated and IVIg-activated MDMs were silenced for

p21 alone or for both p21 and RNR2 by transfection with specificsiRNAs (sip21 and siRNR2) before infection with HIV-1.luc/VSVG. Control MDMs were transfected with equal amounts ofnontargeting siRNAs. Fig. 3 E and F shows results obtained withMDMs from two different donors in which RNR2 could besuccessfully silenced in p21-knockdown cells. As already shown,induction of p21 expression in IVIg-treated MDMs coincidedwith repression of RNR2 transcription (Fig. 3E). Reciprocally,p21 knockdown up-regulated RNR2 expression in both unsti-mulated and IVIg-activated MDMs (Fig. 3E). Silencing of p21alone enhanced or rescued HIV-1 reverse transcription in unsti-mulated and IVIg-activated MDMs, respectively (Fig. 3F) whereasp21 and RNR2 cosilencing (sip21 + siRNR2) did not rescue HIV-1reverse transcripts either in unstimulated or in IVIg-activatedMDMs (Figs. 3F). Collectively, these data confirm that RNR2

suppression is sufficient to account for the inhibition of HIV-1reverse transcription mediated by p21.

p21-Mediated Restriction Remains Active in the Absence of SAMHD1.SAMHD1 was recently demonstrated to restrict HIV-1 rep-lication in macrophages by degrading the already existing in-tracellular pool of dNTPs (13, 14, 19). Moreover, SAMHD1has been shown to be degraded by HIV-2 and SIV Vpx, whichconfers these two viruses the ability to escape the SAMHD1restriction in macrophages and dendritic cells. Although we havepreviously demonstrated that p21 inhibits HIV-2 and SIV (28,33), we wanted to verify whether SAMHD1 is involved in p21-mediated restriction of HIV-1 described above. We first con-trolled the expression of SAMHD1 by WB in unstimulated andIVIg-activated MDMs. As shown in Fig. 4A, the induction of p21protein in IVIg-stimulated MDMs did not affect SAMHD1protein levels compared with unstimulated MDMs. We thenverified that SIVmac251, which expresses Vpx, induces thedegradation of SAMHD1 after infection of MDMs (Fig. 4B). Incontrast, p21 expression was not affected (Fig. 4B). Furthermore,we confirmed that SIVmac251 coinfection enhances HIV-1 in-fection, as previously reported (21) (Fig. 4C). We then examinedwhether, in the presence of Vpx that degrades SAMHD1, theinduction of p21 by IVIg and the knockdown of RNR2 suppressSIV replication in MDMs. Unstimulated and IVIg-activatedMDMs, silenced or not for RNR2 by transfection of RNR2siRNAs (siRNR2), were infected with the same amounts ofSIVmac251 used in Fig. 4 B and C. SIVmac251 replication, de-termined by quantifying p27 antigen 7 d postinfection, wasinhibited by 80% in IVIg-activated MDMs (P= 0.046) (Fig. 4D),confirming our previous results (28). Importantly, SIVmac251 re-

Fig. 2. p21 inhibits the expression of the metabolic dNTP enzyme RNR2. (A) Unstimulated (US) and IVIg-activated MDMs were transfected with p21 siRNA(sip21) and a pool of nontargeting control siRNAs (siCTL). At 24 h after IVIg stimulation and 22 h after siRNA transfection, MDMs were analyzed by Westernblot (WB) for p21, RNR2, RNR1, p53R2, PCNA, and GAPDH protein expression, using the indicated antibodies. (B) p21 and RNR2 mRNA from unstimulated andIVIg-activated control (siCTL) and p21-silenced (sip21) MDMs were quantified by qRT-PCR. The data are the fold change with respect to control cells(unstimulated siCTL) ± SE of MDMs from five donors. (C) In parallel, unstimulated and IVIg-activated siCTL and sip21 MDMs were infected with HIV-1.luc/VSVG. Viral replication was determined by measuring luciferase activity 72 hpi. The data are the fold change with respect to control MDMs (unstimulatedsiCTL) ± SE for MDMs from six donors.

E4000 | www.pnas.org/cgi/doi/10.1073/pnas.1306719110 Allouch et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 3,

202

0

plication was inhibited to the same extent by RNR2-knockdown(Fig. 4D). As shown in Fig. 4E, the amount of SIVmac251 LRTs,quantified by qRT-PCR, was strongly reduced (94%) in RNR2-knockdown MDMs. These results indicate that SIVmac251 in-fection is inhibited at the level of reverse transcription by thesuppression of RNR2 through p21 induction or siRNA-mediatedknockdown, despite its ability to degrade SAMHD1. Thus, theinhibition of both HIV-1 and SIVmac is caused by a blockadeof dNTP synthesis by p21 that is independent of SAMHD1dNTP catabolic pathway.

p21 Represses RNR2 Transcription by Down-Regulating E2F1 Indepen-dently from RB. Induction of RNR2 transcription depends on thetranscription factor E2F1 in fibroblast cell lines (45–47). To de-termine whether p21 represses RNR2 gene expression in MDMsby regulating E2F1, we first studied the effects of p21 inductionby IVIg stimulation and of siRNA-mediated p21 knockdown onE2F1 protein and mRNA levels by using WB and qRT-PCR,respectively. E2F1 protein was depleted in IVIg MDMs (Fig. 5A).Importantly, p21 knockdown strongly enhanced the E2F1 proteinlevel, indicating that p21 inhibits E2F1 expression (Fig. 5A) in

Fig. 3. RNR2 depletion is sufficient to inhibit HIV-1 replication at the reverse transcription step. (A) Unstimulated (US) and IVIg activated MDMs weretransfected with a pool of four siRNAs against the RNR2 gene (siRNR2) or with a pool of nontargeting control siRNAs (siCTL). At 24 h after IVIg stimulation and22 h after siRNA transfection, MDMs were analyzed by WB with antibodies against RNR2, p53R2, and actin. (B) RNR2 and p21 mRNAs in unstimulated (US) andIVIg-activated siCTL and siRNR2 MDMs were quantified by qRT-PCR. The data are the fold change with respect to control MDMs (siCTL unstimulated) ± SE forMDMs from seven donors. (C) Unstimulated and IVIg-activated siCTL and siRNR2 MDMs were infected with HIV-1.luc/VSVG. Viral replication was determinedby measuring luciferase activity 72 hpi. The data are the fold change with respect to control MDMs (siCTL unstimulated) ± SE for MDMs from seven donors. (D)Unstimulated and IVIg-activated siCTL and siRNR2 MDMs were infected with HIV-1.luc/VSVG. Reverse transcription was measured by quantifying late reversetranscripts (LRTs) at 72 hpi by means of qRT-PCR. The data are the fold change with respect to control MDMs (siCTL unstimulated) ± SE for MDMs from sixdonors. (E) Unstimulated and IVIg-treated MDMs from two donors were transfected with siRNA against the p21 gene (sip21) or cotransfected simultaneouslywith sip21 and a pool of four siRNAs against the RNR2 gene (sip21+siRNR2). Control MDMs were transfected with a pool of nontargeting siRNAs (siCTL). At24 h after IVIg stimulation and 22 h after siRNA transfection, p21 and RNR2 mRNAs were quantified by qRT-PCR. The data are the mean fold change withrespect to control MDMs (unstimulated siCTL) in the two donors. Cyan and pink circles represent the values for each donor. (F) Unstimulated and IVIg-treatedMDMs (siCTL, sip21, and sip21+siRNR2) at 24 h post IVIg activation and 22 h post siRNA transfection were infected with HIV-1.luc/VSVG. HIV-1 reversetranscription was determined by quantifying LRTs at 72 hpi by qRT-PCR. The data are the mean fold change with respect to control MDMs (unstimulatedsiCTL) in the two donors. Cyan and pink circles represent the values for each donor.

Allouch et al. PNAS | Published online September 30, 2013 | E4001

MICRO

BIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

Nov

embe

r 3,

202

0

MDMs from four different donors, and E2F1 mRNA levels fellby 65–80% after IVIg stimulation (P= 0.014) (similarly to RNR2mRNA) while increasing 2.15- to 8.00-fold after p21 depletion(P = 0.014) (Fig. 5B). p21 knockdown increased E2F1 mRNAlevels 1.33- to 4.37-fold (P = 0.014) in IVIg-activated MDMs(Fig. 5B). These data show that p21 represses E2F1 transcription.To assess the effect of E2F1 on RNR2 transcription, we silencedE2F1 by transfecting MDMs with four specific siRNAs (siE2F1)and then measured RNR2 and p21 mRNA levels by qRT-PCR.In E2F1-silenced MDMs (siE2F1) from four different donors,E2F1 mRNA levels were reduced by 50–80% compared withcontrol cells transfected with nontargeting siRNAs (siCTL) (P =0.014) (Fig. 5C). Importantly, E2F1 knockdown reduced RNR2mRNA levels by 30–70% (P = 0.014) without affecting p21mRNA levels, indicating that E2F1 specifically transactivatesRNR2 transcription in MDMs (Fig. 5C). Finally, we examinedwhether E2F1 knockdown, like RNR2 knockdown, inhibitedHIV-1 replication. Replication of HIV-1.luc/VSVG in MDMsfrom four different donors was indeed inhibited, by 73–92%, afterE2F1 silencing (P = 0.014) (Fig. 5D). In MDMs from three ofthese four donors, E2F1 knockdown inhibited HIV-1 reversetranscription (LRTs) by 56–87% (P = 0.03) (Fig. 5E), indicatingthat E2F1 acts as a cellular cofactor that promotes HIV-1 reversetranscription in MDMs by transactivating RNR2 expression.p21 has been shown to inhibit E2F1 transactivation activity via

the Retinoblastoma protein (RB). Indeed p21 by inhibitingcyclin-dependent kinases (CDKs) maintains RB in an activated(unphosphorylated) state able to retain E2F1 and block itstransactivation activity (48, 49). However, some reports showedthat p21 binds E2F1 through the N-terminal region and is able toinhibit its transactivation activity independently from RB (40,50). To assess the role of RB in E2F1 suppression by p21 inMDMs, we first investigated the expression of RB and found thatRB mRNA levels, as measured by qRT-PCR, were unaffected by

p21 up-regulation or knockdown in MDMs from three donors(Fig. S3A). We then knockdowned RB by specific siRNAs(siRB1) transfection. Transfection with RB siRNAs efficientlysilenced RB in both unstimulated MDMs (70%) and IVIg-acti-vated MDMs (85%), as shown by WB (Fig. S3B) and mRNAquantification (Fig. S3C). In contrast to p21 knockdown, RBsilencing increased neither E2F1 nor RNR2 gene expression(Fig. S3C), indicating that RB does not affect the transcription ofRNR2 in MDMs. In keeping with these results, RB knockdown,contrary to p21 knockdown, failed to enhance HIV-1 replicationin unstimulated MDMs or to rescue HIV-1 replication in IVIg-activated MDMs (Fig. S3D). These findings ruled out a relevantcontribution of RB to either p21 suppression of E2F1 and p21-mediated restriction of HIV-1 replication.

DiscussionMacrophages are among the cells targeted by HIV-1 and playa crucial role in the establishment of viral reservoirs and diseaseprogression. We have previously shown that p21 inhibits theaccumulation of reverse transcripts of HIV-1 and related lenti-viruses in macrophages (28). Here, we examined the mechanismunderlying this p21-mediated HIV-1 restriction. We show thatp21 reduces the size of cellular dNTP pools required for HIVreverse transcription by suppressing the expression of RNR2, anenzyme indispensable for dNTP biosynthesis, through the down-regulation of the transcription factor E2F1. Therefore, the mech-anism of p21-mediated lentiviral restriction affects the synthesis ofdNTP and differs from the SAMHD1 catabolic pathway.Lentivirus reverse transcription is much slower in macro-

phages than in activated CD4+ T cells (10), due in part to thelow cellular dNTP levels, which create a rate-limiting step in viralcDNA synthesis (11). Although lentiviral reverse transcriptasehas acquired the ability to catalyze DNA synthesis in the pres-ence of limited dNTP pools (11), dNTP levels must still be above

Fig. 4. p21-mediated restriction remains active in the absence of SAMHD1. (A) Unstimulated (US) and IVIg-activated MDMs were controlled by WB for p21,SAMHD1, and actin expressions at 24 h post IVIg activation using the indicated antibodies. (B and C) MDMs were infected with HIV-1.luc/VSVG or coinfectedwith HIV-1.luc/VSVG and SIVmac251 (+SIV). MDMs were controlled for SAMHD1, p21, and actin expressions at 24 hpi (B) and for HIV-1 replication by mea-suring luciferase activity at 72 hpi (C). The data are means ± SD of triplicate wells. (D) Unstimulated and IVIg-activated MDMs from one donor were silencedfor RNR2 by transfecting four specific siRNAs (siRNR2) and infected with SIVmac251 at 22 h post siRNAs transfection and 24 h post IVIg activation. ControlMDMs were transfected with nontargeting control siRNA (siCTL). Viral replication was measured by quantifying p27 antigen 7 d postinfection. The data aremeans ± SD of triplicate wells. These data are representative of experiments with MDMs from three donors. (E) SIV LRTs were determined by qRT-PCR at 48 h pi.

E4002 | www.pnas.org/cgi/doi/10.1073/pnas.1306719110 Allouch et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 3,

202

0

a critical threshold. Indeed, dNTP degradation by SAMHD1impairs HIV-1 replication in noncycling myeloid cells and restingCD4+ T cells by reducing dNTPs to levels below those requiredfor efficient reverse transcription (13–16, 19, 20). SIV and HIV-2counteract SAMHD1 by the expression of Vpx, which targetsSAMHD1 for proteosomal degradation, thus allowing theseviruses to escape SAMHD1-mediated restriction in macrophagesand dendritic cells (13, 14). We show here that p21 restricts HIV-1reverse transcription by blocking the production of dNTP throughthe suppression of RNR2 expression, a key enzyme required forthe de novo pathway of dNTP synthesis. Importantly, this mech-anism was also active against lentiviruses that express Vpx, such asSIVmac, despite Vpx-induced degradation of SAMHD1 in infec-ted cells.

Interestingly, it has been reported that SAMHD1 phosphor-ylation by CDK1 prevents its antiviral activity without affectingits dNTP triphosphohydrolase activity (51). Because CDK1 isa probable target of p21, p21 might modulate SAMHD1 phos-phorylation. In this case, p21 would have a double-negative ef-fect on HIV-1 replication in macrophages, by blocking the dNTPsynthesis on one side and by enhancing SAMHD1 antiviral ac-tivity by preventing its phosphorylation on the other side. Fur-ther studies are needed to assess the potential role of p21 inSAMHD1 phosphorylation and antiviral activity.In mammalian cells, dNTP synthesis and degradation path-

ways are finely tuned to ensure a balance of the four dNTP pools(52). The predominant pathway of dNTP synthesis requires thereduction, by RNR, of ribonucleoside diphosphates (rNDP) to

Fig. 5. p21 represses RNR2 transcription by down-regulating E2F1. (A) Unstimulated (US) and IVIg-activated MDMs were transfected with p21 siRNA(sip21) and a pool of nontargeting control siRNAs(siCTL). At 24 h after IVIg stimulation and 22 h aftersiRNA transfection, MDMs were analyzed by WB forE2F1, p21, and actin protein expression. (B) p21,E2F1, and RNR2 mRNAs from siCTL and sip21(unstimulated or IVIg-treated) MDMs were quanti-fied by qRT-PCR 24 h after IVIg stimulation and 22 hafter siRNA transfection. The data are the foldchange with respect to control cells (unstimulatedsiCTL) ± SE for MDMs from four donors. (C) Unsti-mulated and IVIg-activated MDMs were transfectedwith a pool of four siRNAs against the E2F1 gene(siE2F1) or with control siRNAs (siCTL). At 24 h afterIVIg stimulation and 22 h after siRNA transfection,E2F1, RNR2, and p21 mRNAs were quantified byqRT-PCR. The data are the fold change with respectto control MDMs (siCTL unstimulated) ± SE for MDMsfrom four donors. (D) siCTL and siE2F1 (unstimulatedor IVIg-treated) MDMs were infected with HIV-1.luc/VSVG. Viral replication was quantified by measuringluciferase activity at 72 hpi. The data are the foldchange with respect to control MDMs (siCTL unsti-mulated) ± SE for MDMs from four donors. (E) HIV-1reverse transcription in siCTL and siE2F1 (unstimu-lated or IVIg) MDMs infected with HIV-1.luc/VSVGwas measured by quantifying LRTs by qRT-PCR 72hpi. The data are the fold change with respect tocontrol MDMs (siCTL unstimulated) ± SE for MDMsfrom three donors.

Allouch et al. PNAS | Published online September 30, 2013 | E4003

MICRO

BIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

Nov

embe

r 3,

202

0

their corresponding deoxynucleoside diphosphates (dNDP), whichare then phosphorylated to form the DNA precursor dNTP (53).In mammalian cells, RNR consists of two nonidentical subunits.The large subunit, RNR1, possesses binding sites for catalytic andregulatory activities whereas the small subunit, RNR2, providesa tyrosyl-free radical essential for nucleotide reduction. Anotherhomologous small subunit, p53R2, also exists (44). Regulation ofRNR activity has so far been studied in cell lines (mostly ofmurine origin) and not in terminally differentiated primary cellssuch as macrophages. In cycling cells, RNR2 is degraded duringlate mitosis, and quiescent fibroblast cell lines contain p53R2 butlittle or no RNR2 (54–56). Thus, it has been suggested thatp53R2 serves as a partner of RNR1 in the RNR complex tosupply noncycling cells that lack RNR2 with dNTP for DNArepair. However, p53R2 expression in response to DNA damagein mammalian cells did not lead to dNTP pool expansion (55).Moreover, recent reports suggest that, in quiescent cells, p53R2is required primarily for mitochondrial DNA replication andrepair (57, 58). Therefore, RNR regulation in noncycling cells isnot yet fully elucidated. Surprisingly, we readily detected RNR2mRNA and protein in terminally differentiated MDMs. Thepresence of RNR2, albeit at a low level, suggests that this subunitdoes contribute to the synthesis, by the RNR complex, of dNTPsrequired for DNA repair in human macrophages. As for thesource of dNTP used by lentiviral reverse transcriptase in mac-rophages, we found, importantly, that RNR2 suppression (by p21induction or siRNA knockdown) was sufficient to block HIV-1and SIVmac251 reverse transcription. This result indicates thatRNR2 plays a major role in providing dNTPs for HIV-1 reversetranscription in macrophages although p53R2 is also expressedin these cells. Elsewhere, increased RNR2 expression in differ-entiated macrophages infected with Leishmania has been foundto enhance HIV-1 replication (59).Our findings indicate that p21 regulates RNR2 expression at

the transcriptional level in macrophages. These data concur withresults from previous studies using a p21-inducible fibrosarcomacell line (41). However, contrary to these latter cells, we foundthat p21 did not modulate RNR1 expression in macrophages. Asp21 does not appear to contain a DNA-binding domain, it mustinteract indirectly with the RNR2 promoter, through anothertranscription factor or cofactor. The RNR2 promoter containsE2F elements (60), and RNR2 transcription in fibroblast celllines is dependent on the transcription factor E2F1 (45–47). Wefound that p21 depletion strongly induced both E2F1 and RNR2expressions in macrophages. Importantly, siRNA-mediated de-pletion of E2F1 down-regulated RNR2 expression and blockedHIV-1 replication by inhibiting the reverse transcription step.Similar effects were observed upon p21 induction by FcγR ag-gregation. Although it has been reported that RNR1 is also acti-vated by E2F1 in cell lines (45), RNR1 expression in macrophageswas not modulated upon p21 induction or knockdown. p21 cancontrol E2F transcriptional activity through an RB-dependentpathway (48, 49) or by direct interaction with E2F1 (40, 50). RB-sensitive E2F activity is induced during the G1 phase in pro-liferating cells (50, 61). p21, by inhibiting CDK2 kinase, maintainsRB in the active (hypophosphorylated) form, which negativelyregulates the activity of E2F transcription factors during thetransition through G1 in cycling cells (48, 49). In addition, RB hasbeen reported to inhibit the expression of dNTP metabolicenzymes, including RNR2, in cell lines (62). However, thesestudies are not directly transposable to primary macrophages. Wefound that RB silencing did not significantly affect either RNR2expression or HIV-1 replication in macrophages. Thus, RB doesnot appear to be involved in p21-mediated RNR2 regulation inmacrophages or to play a determining role in p21-mediated HIV-1restriction. Together, these data strongly support the followingmechanism of lentiviral restriction in macrophages (Fig. 6): p21represses E2F1 transcription, possibly by blocking E2F-binding

sites in the E2F1 promoter and, therefore, E2F1 autoactivation(63, 64). The repression of E2F1 leads to down-regulation ofRNR2 gene expression, which blocks de novo dNTP biosynthesis,thus impairing reverse transcription. In addition to impairing HIVcDNA synthesis, dNTP depletion following RNR2 down-regula-tion may also favor misincorporation of cellular rNMPs that areefficiently incorporated into HIV-1 DNA in macrophages (65, 66).Our results suggest that HIV-1 cDNA synthesis could be in-

hibited by targeting RNR2, either directly or indirectly through p21or E2F1. Interestingly, a transient reduction in residual plasma vi-remia and viral reservoirs has been reported in patients treated withhigh-dose IVIg in addition to effective highly active antiretroviraltherapy (HAART) (67). It is conceivable that IVIg-mediated p21induction in macrophages may contribute to this effect. Likewise,the ribonucleoside reductase inhibitor, hydroxyurea (HU) wasshown to have a synergistic effect with conventional HAART onsuppressing HIV (68, 69). However, HU inhibits both RNR2 andp53R2 subunits of RNR (70). The identification of new moleculesthat specifically inhibit RNR2 might lead to an improved anti–HIV-1 activity.

Materials and MethodsMonocyte-Derived Macrophages. Buffy coats from healthy donor blood wereobtained from the French blood bank (Etablissement Français du Sang) aspart of a convention with the Pasteur Institute. In accordance with Frenchlaw, written informed consent to use the cells for clinical research wasobtained from each donor. Monocytes were isolated from buffy coats andwere differentiated into macrophages by using human AB serum in mac-rophage medium, as previously described (28, 33). For experiments, MDMswere harvested and suspended in macrophage medium containing 10% (vol/vol) heat-inactivated FBS, yielding 91–96% of CD14+ cells that also expressedFcγRs (CD16, CD32, and CD64), differentiation markers (C11b and CD71), andM2 macrophage polarization markers (CD163 and CD206). For Fcƴ receptorstimulation, 96-well, 24-well, and 12-well plates were precoated for 2 h at37 °C with respectively 100 μL, 250 μL, and 500 μL of 1 mg/mL human IgG fortherapeutic use (IVIg) (Endobuline; Baxter) in PBS. We have previouslyshown that immobilized IVIg and preformed immune complexes have thesame impact on HIV-1 replication in MDMs (33). MDMs (1 × 106/mL ofmacrophage medium + 10% FBS) were seeded in uncoated or IVIg-pre-coated wells (6.5 × 104 MDMs per well of 96-well plates, 2.5 × 105 MDMs perwell of 24-well plates, and 0.5 × 106 MDMs per well of 12-well plates).Unstimulated and IVIg-activated MDMs were allowed to attach to the platesfor 2 h at 37 °C before infection or siRNA transfection.

Fig. 6. Model of p21-mediated restriction of HIV-1 and related lentivirusesin macrophages. p21 inhibits E2F1 expression. The RNR2 transactivation isconsequently inhibited. The RNR2 down-regulation decreases the dNTPsbiosynthesis. The HIV-1 cDNA synthesis is blocked due the unavailability ofdNTPs for reverse transcription (RT) catalysis.

E4004 | www.pnas.org/cgi/doi/10.1073/pnas.1306719110 Allouch et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 3,

202

0

siRNA Transfection. Small interfering RNAs (siRNAs) were all purchased fromDharmacon. The siRNA against the p21 gene (sip21) was on-target plus siRNAn.12. The siRNAs against the RNR2 (siRNR2), E2F1 (siE2F1), and RB (siRB) genesare siGenome smart pool, each composed of a pool of four siRNAs. ControlsiRNAs were a pool of four on-target plus nontargeting siRNAs. The siRNAsused in knockdown experiments had the following sequences: sip21, 5′ AGACCA GCA UGA CAG AUU U 3′; siRNR2, 1–5′ GCA CUC UAA UGA AGC AAU A3′, 2–5′ GAA GAG UAG GGA GUA U 3′, 3–5′ GAG UAG AGA ACC CAU UUG A3′, and 4–5′ GGC UCA GCU UGG UCG ACA A 3′; siE2F1, 1–5′ GAA CAG GGCCAC UGA CUC U 3′, 2–5′ UGG ACC ACC UGA UGA AUA U 3′, 3–5′ CCC AGGAGG UCA CUU CUG A 3′, and 4–5′ GGC UGG ACC UGG AAA CUG A 3′; andsiRB, 1–3′ GAA AGG ACA UGU GAA CUU A 5′, 2–3′GAA GAA GUA UGA UGUAUU G 5′, 3–3′ GAA AUG ACU UCU ACU CGA A 5′, and 4–3′ GGU UCA ACUACG CGU GUA A 5′.

siRNAs transfection was performed using INTERFERin kits (PolyplusTransfection). Various amounts of siRNAs were prediluted in 1 mL of Opti-MEM, to which 20 μL of INTERFERin was added, and the transfection mix wasleft at room temperature for 10 min. Transfection mix volumes of 32 μL, 125μL, and 250 μL, respectively, were added to 6.5 × 104, 2.5 × 105, and 0.5 × 106

MDMs at final concentrations of 50 nM sip21, siRNR2, or siE2F1 siRNAs, and25 nM siRB1 siRNAs. Equal amounts of on-target plus nontargeting siRNAswere added to control MDMs. MDMs were then incubated at 37 °C for 22 h.The medium was replaced with fresh macrophage medium supplementedwith 10% FBS before infection. Cell lysates were assayed for protein ex-pression by Western blot and for mRNA expression by real-time quantitativePCR (RT-qPCR) to determine the efficiency of gene knockdown at the time ofinfection. Western blot analysis of RB-knockdown MDMs was performed44 h post-siRNA transfection.

Cell Infections.NL4.3-Luc HIV-1 (pseudotypedwith VSV-G HIV-1.luc/VSVG)wasproduced and quantified as previously described (28, 33). Replication-com-petent SIVmac251 was produced in CEM × 174 T-cell line. Knockdown cellswere infected 24 h after IVIg stimulation and 22 h after siRNA transfection.MDMs were infected with 4–5 ng of HIV-1.luc/VSVG p24 and/or 300 ng ofSIVmac251 p27 per 1 × 106 cells during 1 h of spinoculation (1,200 × g atroom temperature) followed by 1 h at 37 °C. The cells were then washedwith PBS and incubated at 37 °C in fresh macrophage medium supple-mented with 10% FBS for 24 h to control SAMHD1 degradation by WB bySIVmac251, for 72 h to measure HIV-1 replication (luciferase activity) and LRTproducts (qRT-PCR). For SIVmac251 replication, the culture supernatantswere harvested 7 d postinfection and quantified for p27 antigen content byusing an SIV p27 ELISA kit (Zeptometrix Corp). SIV LRT quantification wasperformed 48 h postinfection. For viral cDNA detection by qRT-PCR, viralstocks were pretreated with DNase I (Roche Diagnostics) for 1 h at roomtemperature. Unstimulated and IVIg-activated MDMs (2.5 × 105 per well of24-well plate) were treated with the indicated concentrations of deoxy-nucleosides (dNs) immediately after infection in 1 mL total volume for 24 hat 37 °C. The deoxynucleosides consisted of a mixture of dA (D8668), dC(D0776), dG (D0901), and dT (T1895), all from Sigma Aldrich.

Quantitative Real-Time PCR. Total RNA from 2.5 × 105 to 0.5 × 106 MDMs wasextracted with the RNeasy kit (Qiagen) and treated with DNase according tothe manufacturer’s instructions. RNA was transcribed with SuperScript II RT(Invitrogen). PCR amplification of cDNAs was carried out as previouslydescribed (26) on an ABI Prism 7000 Sequence Detection System using theTaqman gene expression premade mix (Applied biosystems): GAPDH,Hs99999905_m1; RNR2, Hs00357247_g1; p21, Hs00355782; RB, Hs01078066_m1;and E2F1, Hs00153451_m1. The data were analyzed with the cycle threshold (CT)method, and the amount of target mRNA in each sample was normalized toGAPDH mRNA as an endogenous reference.

For the quantification of HIV-1 late reverse transcripts (LRTs), MDM DNAwas extractedwith the DNeasy Tissue kit (Qiagen) 72 h postinfection. qRT-PCRanalysis of LRTs, based on quantification of U5-Gag viral DNA, was carried on

an ABI Prism 7000 as previously described (28, 33). SIVmac251 LRTs werequantified as previously described (71). LRT amounts were normalized to anendogenous reference gene (albumin).

dNTP Quantification. p21-knockdown (sip21) or control (siCTL) MDMs (3 × 106)were washed with cold PBS and lysed, 22 h after siRNA transfection, in 2 mLof 60% (vol/vol) aqueous methanol. Lysates were held at 95 °C for 7 min,clarified by centrifugation at 16,000 × g for 15 min at 4 °C, and dried undervacuum. Extracted dNTPs were then resuspended and quantified by a primerextension assay as previously described (11). The dNTP assay linearity range isbetween 2% and 32% of the primer extension.

Bioluminescence Quantification in Living Cells. The technique for quantifica-tion of bioluminescence in infected cells has been described in detail else-where (42). Briefly, the system includes a fully automated invertedmicroscope (200M; Carl Zeiss) and is housed in a light-tight dark box. Low-level light emission was collected using an Imaging Photon Detector (IPD 3;Photek Ltd.), and brighteld images were obtained with a CCD camera(Coolsnap HQ; Roper Scientic). Observations were made with Plan-Neouar10× (air, NA = 0.5), 25× (oil, NA = 0.8) objectives (Carl Zeiss). The data ac-quisition software allowed superposition of photon events with brightfieldimages and selective quantification of photons in delimited surfaces. The 105

MDMs were plated after infection on glass-bottomed dishes (MatTek),washed, and cultured in complete medium for 5 d before analysis. Theanalyses were conducted in PBS plus 5% FCS or complete phenol red freemedium at room temperature in the presence of endotoxin-free beetleD-luciferin (Promega) (final concentration, 1 mM).

Western Blot and Antibodies. MDMs (0.5 × 106) were lysed in 100 μL of lysisbuffer (50 mM Tris·HCl, pH 7.4, 150 mM NaCl, 1 mM EDTA, and 1% TritonX-100) containing a protease and phosphatase inhibitor mixture (Roche).Cell extracts (30–50 μg) were hybridized with the primary antibodies, fol-lowed by secondary horseradish peroxidase anti-goat, anti-rabbit (Sigma), oranti-mouse antibodies (Cell Signaling). Proteins were revealed on Hyperfilms(Amersham) by using the SuperSignal West picochemiluminescent substrate(Pierce). The primary antibodies—anti-PCNA (PC10), anti-RNR2 (N-18), anti-E2F1 (KH95), anti-RB (IF8), and anti-RNR1 (H-300)—were all from Santa Cruzand were all used at 1:200 dilution in 1% skimmed milk in PBS-0.05% Tween.The primary plolyclonal anti-p53R2 (1:500), monoclonal anti-SAMHD1 (1:2,000),and monoclonal anti-GAPDH (1:1,000) antibodies were from Abcam. Themonoclonal anti-p21 (1:1,000) was from Millipore, and anti-actin (1:5,000) wasfrom Sigma.

Statistical Analysis. Fold change is calculated as the ratio of the mean valuefrom the “treated” samples (IVIg activated, knockdowned MDMs) to themean value from the control sample (unstimulated and unsilenced MDMs).The mean value is obtained from the three replicate values. Differencesamong groups were analyzed with the Kruskal–Wallis nonparametric test.When a significant difference was detected, two-by-two comparisons wereperformed with the Mann–Whitney test. Data were analyzed with STATAsoftware version 12.0 (Stata Corporation). Experiments with less than threevalues were not subjected to statistical analysis.

Note Added in Proof.During the submission process of this work, Diaz-Griffero’sand Benkirane’s laboratories reported that the anti–HIV-1 activity of SAMHD1is modulated by its phosphorylation by Cyclin A2/CDK1 (51, 72).

ACKNOWLEDGMENTS. We thank Beatrice Jacquelin (Institut Pasteur) forproviding the SIVmac251 viral stock. This study was funded by Sidaction,France, and National Institutes of Health (NIH) Grant AI049781 (to B.K.). A.A.received a postdoctoral fellowship from Sidaction, and S.M.A. was supportedby NIH Training Grant GM068411.

1. Verani A, Gras G, Pancino G (2005) Macrophages and HIV-1: Dangerous liaisons. Mol

Immunol 42(2):195–212.2. Aquaro S, et al. (2002) Long-term survival and virus production in human primary

macrophages infected by human immunodeficiency virus. J Med Virol 68(4):479–488.3. Orenstein JM, Fox C, Wahl SM (1997) Macrophages as a source of HIV during op-

portunistic infections. Science 276(5320):1857–1861.4. Sharova N, Swingler C, Sharkey M, Stevenson M (2005) Macrophages archive HIV-1

virions for dissemination in trans. EMBO J 24(13):2481–2489.5. Kedzierska K, et al. (2003) Defective phagocytosis by human monocyte/macrophages

following HIV-1 infection: Underlying mechanisms and modulation by adjunctive

cytokine therapy. J Clin Virol 26(2):247–263.

6. Cassol E, Cassetta L, Alfano M, Poli G (2010) Macrophage polarization and HIV-1 in-

fection. J Leukoc Biol 87(4):599–608.7. Shen R, et al. (2009) Macrophages in vaginal but not intestinal mucosa are monocyte-like

and permissive to human immunodeficiency virus type 1 infection. J Virol 83(7):3258–3267.8. Herbein G, Coaquette A, Perez-Bercoff D, Pancino G (2002) Macrophage activation and

HIV infection: Can the Trojan horse turn into a fortress? Curr Mol Med 2(8):723–738.9. Bergamaschi A, Pancino G (2010) Host hindrance to HIV-1 replication in monocytes

and macrophages. Retrovirology 7:31.10. O’Brien WA, et al. (1994) Kinetics of human immunodeficiency virus type 1 reverse

transcription in blood mononuclear phagocytes are slowed by limitations of nucleo-

tide precursors. J Virol 68(2):1258–1263.

Allouch et al. PNAS | Published online September 30, 2013 | E4005

MICRO

BIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

Nov

embe

r 3,

202

0

11. Diamond TL, et al. (2004) Macrophage tropism of HIV-1 depends on efficient cellulardNTP utilization by reverse transcriptase. J Biol Chem 279(49):51545–51553.

12. Triques K, Stevenson M (2004) Characterization of restrictions to human immuno-deficiency virus type 1 infection of monocytes. J Virol 78(10):5523–5527.

13. Hrecka K, et al. (2011) Vpx relieves inhibition of HIV-1 infection of macrophagesmediated by the SAMHD1 protein. Nature 474(7353):658–661.

14. Laguette N, et al. (2011) SAMHD1 is the dendritic- and myeloid-cell-specific HIV-1restriction factor counteracted by Vpx. Nature 474(7353):654–657.

15. Baldauf HM, et al. (2012) SAMHD1 restricts HIV-1 infection in resting CD4(+) T cells.Nat Med 18(11):1682–1687.

16. Descours B, et al. (2012) SAMHD1 restricts HIV-1 reverse transcription in quiescent CD4(+) T-cells. Retrovirology 9:87.

17. Goldstone DC, et al. (2011) HIV-1 restriction factor SAMHD1 is a deoxynucleosidetriphosphate triphosphohydrolase. Nature 480(7377):379–382.

18. Powell RD, Holland PJ, Hollis T, Perrino FW (2011) Aicardi-Goutieres syndrome geneand HIV-1 restriction factor SAMHD1 is a dGTP-regulated deoxynucleotide triphos-phohydrolase. J Biol Chem 286(51):43596–43600.

19. Lahouassa H, et al. (2012) SAMHD1 restricts the replication of human immunodefi-ciency virus type 1 by depleting the intracellular pool of deoxynucleoside triphos-phates. Nat Immunol 13(3):223–228.

20. Kim B, Nguyen LA, Daddacha W, Hollenbaugh JA (2012) Tight interplay amongSAMHD1 protein level, cellular dNTP levels, and HIV-1 proviral DNA synthesis kineticsin human primary monocyte-derived macrophages. J Biol Chem 287(26):21570–21574.

21. Sharova N, et al. (2008) Primate lentiviral Vpx commandeers DDB1 to counteracta macrophage restriction. PLoS Pathog 4(5):e1000057.

22. Srivastava S, et al. (2008) Lentiviral Vpx accessory factor targets VprBP/DCAF1 sub-strate adaptor for cullin 4 E3 ubiquitin ligase to enable macrophage infection. PLoSPathog 4(5):e1000059.

23. Bergamaschi A, et al. (2009) The human immunodeficiency virus type 2 Vpx proteinusurps the CUL4A-DDB1 DCAF1 ubiquitin ligase to overcome a postentry block inmacrophage infection. J Virol 83(10):4854–4860.

24. Peng G, et al. (2007) Myeloid differentiation and susceptibility to HIV-1 are linked toAPOBEC3 expression. Blood 110(1):393–400.

25. Berger G, et al. (2011) APOBEC3A is a specific inhibitor of the early phases of HIV-1infection in myeloid cells. PLoS Pathog 7(9):e1002221.

26. Goujon C, et al. (2013) Evidence for IFNα-induced, SAMHD1-independent inhibitors ofearly HIV-1 infection. Retrovirology 10(1):23.

27. Dragin L, et al. (2013) Interferon block to HIV-1 transduction in macrophages despiteSAMHD1 degradation and high deoxynucleoside triphosphates supply. Retrovirology10(30 (11 March 2013)):30.

28. Bergamaschi A, et al. (2009) The CDK inhibitor p21Cip1/WAF1 is induced by Fcgam-maR activation and restricts the replication of human immunodeficiency virus type 1and related primate lentiviruses in human macrophages. J Virol 83(23):12253–12265.

29. Zhang J, Scadden DT, Crumpacker CS (2007) Primitive hematopoietic cells resist HIV-1infection via p21. J Clin Invest 117(2):473–481.

30. Vázquez N, et al. (2005) Human immunodeficiency virus type 1-induced macrophagegene expression includes the p21 gene, a target for viral regulation. J Virol 79(7):4479–4491.

31. Cherrier T, et al. (2009) p21(WAF1) gene promoter is epigenetically silenced by CTIP2and SUV39H1. Oncogene 28(38):3380–3389.

32. Chen H, et al. (2011) CD4+ T cells from elite controllers resist HIV-1 infection by se-lective upregulation of p21. J Clin Invest 121(4):1549–1560.

33. David A, et al. (2006) The engagement of activating FcgammaRs inhibits primatelentivirus replication in human macrophages. J Immunol 177(9):6291–6300.

34. Harper JW, Adami GR, Wei N, Keyomarsi K, Elledge SJ (1993) The p21 Cdk-interactingprotein Cip1 is a potent inhibitor of G1 cyclin-dependent kinases. Cell 75(4):805–816.

35. Xiong Y, et al. (1993) p21 is a universal inhibitor of cyclin kinases. Nature 366(6456):701–704.

36. Liu M, Iavarone A, Freedman LP (1996) Transcriptional activation of the human p21(WAF1/CIP1) gene by retinoic acid receptor: Correlation with retinoid induction ofU937 cell differentiation. J Biol Chem 271(49):31723–31728.

37. Xaus J, et al. (1999) Interferon gamma induces the expression of p21waf-1 and arrestsmacrophage cell cycle, preventing induction of apoptosis. Immunity 11(1):103–113.

38. Asada M, et al. (1999) Apoptosis inhibitory activity of cytoplasmic p21(Cip1/WAF1) inmonocytic differentiation. EMBO J 18(5):1223–1234.

39. Sherr CJ, Roberts JM (1995) Inhibitors of mammalian G1 cyclin-dependent kinases.Genes Dev 9(10):1149–1163.

40. Delavaine L, La Thangue NB (1999) Control of E2F activity by p21Waf1/Cip1. Onco-gene 18(39):5381–5392.

41. Chang BD, et al. (2000) Effects of p21Waf1/Cip1/Sdi1 on cellular gene expression:Implications for carcinogenesis, senescence, and age-related diseases. Proc Natl AcadSci USA 97(8):4291–4296.

42. Sáez-Cirión A, Nicola MA, Pancino G, Shorte SL (2006) Quantitative real-time analysisof HIV-1 gene expression dynamics in single living primary cells. Biotechnol J 1(6):682–689.

43. Collin M, Gordon S (1994) The kinetics of human immunodeficiency virus reversetranscription are slower in primary human macrophages than in a lymphoid cell line.Virology 200(1):114–120.

44. Tanaka H, et al. (2000) A ribonucleotide reductase gene involved in a p53-dependentcell-cycle checkpoint for DNA damage. Nature 404(6773):42–49.

45. DeGregori J, Kowalik T, Nevins JR (1995) Cellular targets for activation by the E2F1transcription factor include DNA synthesis- and G1/S-regulatory genes. Mol Cell Biol15(8):4215–4224.

46. DeGregori J, Leone G, Miron A, Jakoi L, Nevins JR (1997) Distinct roles for E2F proteinsin cell growth control and apoptosis. Proc Natl Acad Sci USA 94(14):7245–7250.

47. Ishida S, et al. (2001) Role for E2F in control of both DNA replication and mitoticfunctions as revealed from DNA microarray analysis. Mol Cell Biol 21(14):4684–4699.

48. Dotto GP (2000) p21(WAF1/Cip1): More than a break to the cell cycle? Biochim Bio-phys Acta 1471(1):M43–M56.

49. Perkins ND (2002) Not just a CDK inhibitor: Regulation of transcription by p21(WAF1/CIP1/SDI1). Cell Cycle 1(1):39–41.

50. Dimri GP, Nakanishi M, Desprez PY, Smith JR, Campisi J (1996) Inhibition of E2F ac-tivity by the cyclin-dependent protein kinase inhibitor p21 in cells expressing orlacking a functional retinoblastoma protein. Mol Cell Biol 16(6):2987–2997.

51. White TE, et al. (2013) The retroviral restriction ability of SAMHD1, but not its de-oxynucleotide triphosphohydrolase activity, is regulated by phosphorylation. CellHost Microbe 13(4):441–451.

52. Rampazzo C, et al. (2010) Regulation by degradation, a cellular defense against de-oxyribonucleotide pool imbalances. Mutat Res 703(1):2–10.

53. Nordlund P, Reichard P (2006) Ribonucleotide reductases. Annu Rev Biochem 75:681–706.

54. Chabes AL, Pfleger CM, Kirschner MW, Thelander L (2003) Mouse ribonucleotide re-ductase R2 protein: A new target for anaphase-promoting complex-Cdh1-mediatedproteolysis. Proc Natl Acad Sci USA 100(7):3925–3929.

55. Håkansson P, Hofer A, Thelander L (2006) Regulation of mammalian ribonucleotidereduction and dNTP pools after DNA damage and in resting cells. J Biol Chem 281(12):7834–7841.

56. Björklund S, Skog S, Tribukait B, Thelander L (1990) S-phase-specific expression ofmammalian ribonucleotide reductase R1 and R2 subunit mRNAs. Biochemistry 29(23):5452–5458.

57. Bourdon A, et al. (2007) Mutation of RRM2B, encoding p53-controlled ribonucleotidereductase (p53R2), causes severe mitochondrial DNA depletion. Nat Genet 39(6):776–780.

58. Pontarin G, Ferraro P, Bee L, Reichard P, Bianchi V (2012) Mammalian ribonucleotidereductase subunit p53R2 is required for mitochondrial DNA replication and DNA re-pair in quiescent cells. Proc Natl Acad Sci USA 109(33):13302–13307.

59. Mock DJ, et al. (2012) Leishmania induces survival, proliferation and elevated cellulardNTP levels in human monocytes promoting acceleration of HIV co-infection. PLoSPathog 8(4):e1002635.

60. Chabouté ME, Clément B, Sekine M, Philipps G, Chaubet-Gigot N (2000) Cell cycleregulation of the tobacco ribonucleotide reductase small subunit gene is mediated byE2F-like elements. Plant Cell 12(10):1987–2000.

61. Verona R, et al. (1997) E2F activity is regulated by cell cycle-dependent changes insubcellular localization. Mol Cell Biol 17(12):7268–7282.

62. Angus SP, et al. (2002) Retinoblastoma tumor suppressor targets dNTP metabolism toregulate DNA replication. J Biol Chem 277(46):44376–44384.

63. Johnson DG, Ohtani K, Nevins JR (1994) Autoregulatory control of E2F1 expression inresponse to positive and negative regulators of cell cycle progression. Genes Dev8(13):1514–1525.

64. Neuman E, Flemington EK, Sellers WR, Kaelin WG, Jr. (1994) Transcription of the E2F-1gene is rendered cell cycle dependent by E2F DNA-binding sites within its promoter.Mol Cell Biol 14(10):6607–6615.

65. Ferraro P, Franzolin E, Pontarin G, Reichard P, Bianchi V (2010) Quantitation of cel-lular deoxynucleoside triphosphates. Nucleic Acids Res 38(6):e85.

66. Kennedy EM, Amie SM, Bambara RA, Kim B (2012) Frequent incorporation of ribo-nucleotides during HIV-1 reverse transcription and their attenuated repair in mac-rophages. J Biol Chem 287(17):14280–14288.

67. Lindkvist A, et al. (2009) Reduction of the HIV-1 reservoir in resting CD4+ T-lympho-cytes by high dosage intravenous immunoglobulin treatment: a proof-of-conceptstudy. AIDS Res Ther 6:15.

68. Gao WY, Cara A, Gallo RC, Lori F (1993) Low levels of deoxynucleotides in peripheralblood lymphocytes: A strategy to inhibit human immunodeficiency virus type 1 rep-lication. Proc Natl Acad Sci USA 90(19):8925–8928.

69. Lori F, et al. (1999) Treatment of human immunodeficiency virus infection with hy-droxyurea, didanosine, and a protease inhibitor before seroconversion is associatedwith normalized immune parameters and limited viral reservoir. J Infect Dis 180(6):1827–1832.

70. Shao J, Zhou B, Chu B, Yen Y (2006) Ribonucleotide reductase inhibitors and futuredrug design. Curr Cancer Drug Targets 6(5):409–431.

71. Hofmann-Lehmann R, et al. (2000) Sensitive and robust one-tube real-time reversetranscriptase-polymerase chain reaction to quantify SIV RNA load: comparison of one-versus two-enzyme systems. AIDS Res Hum Retroviruses 16(13):1247–1257.

72. Cribier A, Descours B, Valadão AL, Laguette N, Benkirane M (2013) Phosphorylation ofSAMHD1 by cyclin A2/CDK1 regulates its restriction activity toward HIV-1. Cell Rep3(4):1036–1043.

E4006 | www.pnas.org/cgi/doi/10.1073/pnas.1306719110 Allouch et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 3,

202

0