Single-Cell Imaging of HIV-1 Provirus (SCIP)Cristina Di Primioa, Valentina Querciolia, Awatef Allouchb, Rik Gijsbersc, Frauke Christc, Zeger Debyserc, Daniele Arosiod,and Anna Ceresetoe,1

aLaboratory of Neurobiology, Scuola Normale Superiore, 56124 Pisa, Italy; bUnité de Régulation des Infections Retrovirales, Institut Pasteur, 75724 Paris,France; cLaboratory of Molecular Virology and Gene Therapy, KU Leuven, 3000 Leuven, Flanders, Belgium; dIstituto di Biofisica, Consiglio Nazionale delleRicerche, 38123 Trento, Italy; and eLaboratory of Molecular Virology, Centre for Integrative Biology, 38123 Trento, Italy

Edited* by Stephen P. Goff, Columbia University College of Physicians and Surgeons, New York, NY, and approved February 21, 2013 (received for reviewSeptember 21, 2012)

Recent advances in fluorescence microscopy provided tools for theinvestigation and the analysis of the viral replication steps in thecellular context. In the HIV field, the current visualization systemssuccessfully achieve the fluorescent labeling of the viral envelopeand proteins, but not the genome. Here, we developed a systemable to visualize the proviral DNA of HIV-1 through immunofluores-cence detection of repair foci for DNA double-strand breaks specif-ically induced in the viral genome by the heterologous expression ofthe I-SceI endonuclease. The system for Single-Cell Imaging of HIV-1Provirus, named SCIP, provides the possibility to individually trackintegrated-viral DNAwithin the nuclei of infected cells. In particular,SCIP allowed us to perform a topological analysis of integrated viralDNA revealing that HIV-1 preferentially integrates in the chromatinlocalized at the periphery of the nuclei.

yH2AX | retargeting | CBX-LEDGF

Technical developments in imaging-based techniques havegreatly improved our understanding of HIV–host cell inter-

actions. HIV-1 virions labeled with fluorophores were pivotal inshedding light onto multiple aspects of the virus–host interplayduring all steps of HIV-1 replication cycle (1–13). Nevertheless,few optical approaches have been so far developed to visualizeviral particles within the nuclear compartment (14, 15), whichlimits our comprehension of the interaction between HIV-1 andthe nuclear architecture.Moreover, the existing detection tools arebased on the visualization of the viral protein complexes or en-velope but not of the viral DNA with the only exception of thefluorescence in situ hybridization (FISH) technique. Even thoughFISH is a powerful technique, it is not very sensitive for HIV-1detection and moreover disrupts the native architecture of thenuclear compartment as it requires harsh denaturation conditions.In addition, this technique does not allow the discrimination be-tween integrated and nonintegrated viral DNA (16, 17). Here wedescribe a fluorescent approach to visualize HIV-1 DNA in thenuclear compartment of infected cells. We exploited a site-specificgenome engineering technique that represents one of the mostpromising approaches to detect specific genome regions in modi-fied organism (18) allowing for their spatial localization in the cell(19, 20). This technique couples endogenous repair pathways, in-duced by rare cutting endonuclease, with immunofluorescenceanalysis. Rare cutting endonucleases, such as the yeast-homingendonuclease I-SceI, specifically cuts target sequences that cannotbe found in the mammalian genome. By engineering DNA tocontain the I-SceI cleavage site, it is thus possible to induce en-dogenous repair mechanism for double-strand breaks (DSBs) atspecific genomic positions. DSB repair leads to the formation ofdistinct subnuclear structures that are generally referred to as“foci” (21). The first sensor of theDSB is the histoneH2AX, whichbecomes massively phosphorylated at serine 139 (γ-H2AX). Fociof DNA repair are thus visible through immunofluorescence byusing specific γ-H2AX antibodies. Here, by inserting an I-SceI sitein the HIV-1 genome, we show that individual proviral DNA canbe efficiently detected at the level of a single infected cell [Single-Cell Imaging of HIV-1 Provirus (SCIP)]. The power of SCIP is thetemporal and spatial analysis of HIV-1 in individual nuclei of

infected cells, a great benefit over currently used techniques. A 3Dtopological analysis, performed by SCIP, demonstrated thatintegrated viral DNA localizes at the periphery of the nuclei re-vealing important insights in the nuclear biology of HIV-1.

ResultsConstruction of the HIV–I-SceI Reporter System and Viral DNACleavage Validation. DNA DSB leads to the phosphorylation ofH2AX molecules and the formation of γ-H2AX foci, which can bevisualized by immunofluorescence (IF) technique. The ability ofthe rare cutting endonuclease I-SceI to induce specific DSB wasexploited to detect HIV-1 DNA in infected cells. In specific, weengineeredHIV-1 viral particles by inserting an 18-bp target site forthe endonuclease I-SceI into viral transfer vectors (pHR-CMVGFP-I-SceI, pHR-CMVΔGFP-I-SceI; Fig. 1A); the cleavage activity ofthe enzyme on viral DNA was then verified by recombinant I-SceIdigestion (Fig. S1).The system was initially evaluated in U2OS clones obtained by

stable transfection of the transfer vector (pHR-CMVGFP-I-SceI)and subsequent clonal selection. Representative stable cellclones, U2OS-6 and U2OS-8, containing an average of 6 and 22integrated DNA copies, respectively, as quantified by quantitativePCR (qPCR), were transfected with the I-SceI endonucleaseencoding plasmid (pCBASce) and immunostained with specificanti–γ-H2AX antibodies to detect nuclear foci by confocal mi-croscopy. Fig. 1B shows that γ-H2AX foci are clearly detectable innuclei of U2OS-6 and U2OS-8 cells expressing the I-SceI endo-nuclease, although few background signals, likely generated byspontaneous DNA strand breaks, are detected in nontransfectedcells. Moreover, cells positive for γ-H2AX foci were also ex-pressing the viral reporter gene (GFP; Fig. 1B). Quantificationanalysis of the γ-H2AX foci revealed that the total number ofγ-H2AX foci per nucleus closely correlates with the integratedviral DNA copies measured by qPCR (Fig. S2 A and B).To further prove that the γ-H2AX foci detected by IF corre-

spond to the viral DNA containing the I-SceI site we performedchromatin immunoprecipitation experiments (ChIP) using γ-H2AXantibodies. Chromatin immunoprecipitates from U2OS-8 cells ex-pressing or not expressing the I-SceI endonuclease (Fig. S2C) wereamplified by real-time PCR with primer pairs covering the im-mediate proximity of the I-SceI site (primer pairs A and B) andwith primers targeting the viral DNA (MH531-532) (22) (Fig.1C). We observed an enrichment of γ-H2AX (10–30-fold) withprimers specific for the I-SceI region and viral sequences; noenrichment was obtained in control samples (Fig. 1C).

Author contributions: C.D.P., V.Q., F.C., Z.D., D.A., and A.C. designed research; C.D.P., V.Q.,and A.A. performed research; R.G. and A.C. contributed new reagents/analytic tools; D.A.analyzed data; and C.D.P., V.Q., and A.C. wrote the paper.

The authors declare no conflict of interest.

*This Direct Submission article had a prearranged editor.1To whom correspondence should be addressed. E-mail: cereseto@science.unitn.it.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1216254110/-/DCSupplemental.

5636–5641 | PNAS | April 2, 2013 | vol. 110 | no. 14 www.pnas.org/cgi/doi/10.1073/pnas.1216254110

Dow

nloa

ded

by g

uest

on

Janu

ary

31, 2

021

Similar results were obtained using stable clones infectedwith HIV-1 particles containing the I-SceI restriction site (HIV-CMVGFP-I-SceI) and pseudotyped with the vesicular stomatitisvirus G (VSV-G) envelope (Fig. S2 D and E). Taken together,our data proved that the I-SceI target site–mediated modifica-tion of the HIV-I genome allows visualizing viral DNA at thenuclear level.

Visualization of Viral DNA in Infected Cells. The HIV-I-SceI re-porter system was then evaluated in the context of viral infectionusing HIV-CMVGFP-I-SceI viral particles pseudotyped with theVSV-G envelope, in U2OS cells transfected with the pCBASceplasmid. Cells were immunostained with anti–γ-H2AX anti-bodies at 24 and 48 h postinfection and analyzed by 3D confocalmicroscopy. As shown in Fig. 2A, γ-H2AX foci were detected ininfected cells expressing both the GFP reporter gene and theI-SceI endonuclease (Fig. 2A, Top), and background signalscould be detected in infected (GFP positive) cells in the absenceof the I-SceI endonuclease (Fig. 2A, Middle). When infectingcells with or without I-SceI endonuclease with the control viralvector HIV-CMVGFP, only background signals derived fromspontaneously formed DSBs was observed (Fig. S3).Because, in the early replication steps, the viral DNA species

exist as nonintegrated or integrated proviral DNA, we investigatedthe ability of the HIV-I-SceI visualization system to detect the twodifferent forms. To this aim we infected cells with an integration

defective HIV-1 virus (HIV-CMV-GFP-I-SceID116A) bearing theD116A mutation in the integrase active site causing a catalyticdefect in integrase. Following infection this mutant virus produceseither linear or circularized (2-LTR) nonintegrated DNA. Resultsin Fig. 2A, Bottom show that even though infected cells produceGFP from nonintegrated forms, no γ-H2AX foci could be detec-ted. To better describe this observation the foci formation in cellsinfected with HIV-CMV-GFP-I-SceI wild-type or integration-defective (D116A) viruses was quantified. The quantification anal-ysis was set up by comparing three approaches in cells infected with

A

B C

Fig. 1. Detection of γ-H2AX foci associated with HIV-1 DNA in U2OS cells.(A) Schematic representation of viral constructs and mechanism of γ-H2AXfoci formation. The plasmid pHR-CMVGFP was modified by inserting theI-SceI target site to produce either the HIV-CMVGFP-I-SceI virions expressingGFP in infected cells or the HIV-CMVΔGFP-I-SceI GFP deleted. The endonu-clease I-SceI (blue spot) exogenously expressed by transfecting infected cellscleaves the I-SceI site producing DNA DSBs. H2AX molecules at the DSB sitebecome phosphorylated forming γ-H2AX foci (red spots) detectable by im-munofluorescence. (B) U2OS-6 and U2OS-8 stable cell clones containing pHR-CMVGFP-I-SceI and expressing or not expressing the I-SceI endonuclease(pCBASce) by transient transfection were immunostained with anti–γ-H2AXantibodies: γ-H2AX foci (in red) viral GFP expression (in green). Scale bars,10 μm. (C) Chromatin immunoprecipitation (ChIP) analysis of viral DNA as-sociated with γ-H2AX. (Upper) Schematic representation of the positions ofthe primers used for real-time PCR quantification. (Lower) The γ-H2AX ChIPanalysis in the U2OS-8 cell clone expressing or not expressing the I-SceI en-donuclease. Relative enrichment represents the enrichment of γ-H2AXcompared with an IgG control (normalized with a PCR internal control toa locus other than the viral DNA). Error bars represent SDs from at least twoindependent experiments.

A

B C

D

Fig. 2. Visualization of viral DNA in infected cells. (A) Detection of γ-H2AXfoci and GFP in cells expressing (Top) or lacking (Middle) the exogenous I-SceIendonuclease. Cells were infected by integration competent virus or by in-tegration defective virus (D116A) (Bottom). Scale bars, 10 μm. (B) Quantifi-cation of γ-H2AX foci number at 24 and 48 hpi in cells infected with 4 RTUof pHR-CMVGFP-I-SceI or pHR-CMVGFP-I-SceID116A (means ±SEM from threeindependent experiments after background subtraction). At least 150cells per experiment have been analyzed; [Pwt vs. D116A = 1.332e-15 at 24 hpi,Pwt vs. D116A = 4.692e-13 at 48 hpi, Kolmogorov–Smirnov test (KS test)]. (SeeMaterials and Methods for background correction.) (C) Quantification of thenumber of γ-H2AX foci at 48 h and 13 d postinfection. Cells were infectedwith0.004 RTU (white bars), 0.4 RTU (gray bars), or 4 RTU (black bars) (means± SEMfrom at least two independent experiments after background subtraction); atleast 200 cells per experiment have been analyzed. (D) Quantification ofγ-H2AX foci number per cell with respect to percentage of GFP positive cellsat the three different RTU after background subtraction; (P0.04RTU vs. bg =0.04542, P0.4RTU vs. bg = 2.7e-5, P4RTU vs. bg = 2.9e-7, KS test).

Di Primio et al. PNAS | April 2, 2013 | vol. 110 | no. 14 | 5637

MICRO

BIOLO

GY

Dow

nloa

ded

by g

uest

on

Janu

ary

31, 2

021

increasing amounts of HIV-CMVGFP-I-SceI. The first approach,2D quantification, consisted in counting the number of γ-H2AX fociin the optical section taken in the center of each cell nucleus (Fig.S4A). The second approach measured the total γ-H2AX foci fluo-rescence intensity in the central optical section of each nucleus (Fig.S4B). The third approach, 3D quantification, consisted in the au-tomated counting of the number of γ-H2AX foci within the entirenuclear volume and was based on the nuclear 3D reconstructionfrom the Z stacks, by using dedicated imaging software (ImarisBITPLANE Scientific Software, ImageJ NIH). These threeapproaches showed analogous fold increase at increasing viraltiter values (Fig. S4B), thus proving the validity HIV-I-SceI re-porter system to detect HIV-1 genomes in infected cells. In thefurther study we applied the 2D quantification approach. We ob-served that in noninfected control cells the background number ofγ-H2AX foci varied among individual experiments (Fig. S4C). Forthis reason a rigorous procedure was set up for the evaluation ofthe background number of foci. Background number of foci wasquantified in noninfected cells expressing the I-SceI enzyme andsubtracted in each individual experiment from the number of focidetected in cells infected with the HIV-CMVGFP-I-SceI (exem-plified in Fig. S4 C–E and Materials and Methods).The foci quantification in cells infected with HIV-CMV-GFP-

I-SceI wild-type or integration-defective (D116A) viruses wasperformed at 24 and 48 h postinfection. We observed that cellsinfected with wild-type virus were positive for γ-H2AX foci atboth time points, but only background values of γ-H2AX fociwere detected by using the HIV-CMV-I-SceID116A virus (Fig. 2B

and Fig. S5). Therefore, we can conclude that the HIV-I-SceIsystem specifically detects HIV-1 DNA integrated into the hostgenome of individual infected cells; accordingly this method wasnamed Single-Cell Imaging of HIV-1 Provirus.To further explore the correlation of γ-H2AX foci formation

with amounts of viral genomes in time and to establish the lowestmultiplicity of infection that can be used in SCIP, cells wereinfected with increasing amounts of HIV-CMVGFP-I-SceI (0.04,0.4, and 4 RT units (RTU) as measured in ref. 23) and analyzedat 48 h and 13 d postinfection. To avoid continuous foci for-mation at both time points the I-SceI endonuclease was trans-fected at 24 h before immunostaining. The number of foci pernucleus detected at both 48 h and 13 d increased consistentlywith increased viral titers, thus indicating their correlation withthe amounts of integrated DNA and their stability from the timeof infection (Fig. 2C and Fig. S5). The analysis performed bymeasuring the number of foci above background in cells positivefor GFP, revealed that as few as 15% of infected cells (GFPpositive) can be detected through SCIP (Fig. 2D). However, toobtain maximum sensitivity the following experiments were per-formed using 4 RTU leading to almost 95% infectivity (Fig. 2D).Finally, to verify the specificity of the γ-H2AX foci signals

detected in HIV-CMVGFP-I-SceI infected cells, coimmunos-taining was performed with Rad51 or 53BP1 antibodies recog-nizing DNA repair focus hallmark proteins. Signal colocalizationwas observed between γ-H2AX and Rad51 or 53BP1 foci, thusconfirming the specificity of the γ-H2AX foci (Fig. S6A).

A

B

D

C

Fig. 3. SCIP analysis in transportin-SR2 knockdowncells and in anti-retroviral drug treated cells. (A) Im-munofluorescence detection of γ-H2AX foci (yellowspots) in cells transfected with fluorescent siRNA (bluespots) targeting TRN-SR2 mRNA (siRNA-SR2) or a mis-match control siRNA (siRNA-MM) and infected with4RTU of HIV-CMVΔGFP-I-SceI. HIV-CMVΔGFP-I-SceIvirus was used to avoid the crosstalk between fluo-rophores. Nuclear middle z stacks are shown. Scalebars, 10 μm. (B) Quantification of γ-H2AX foci in cellstransfected by siRNA-MM (black bars) and cells trans-fected by siRNA-SR2 (white bars) after backgroundsubtraction; (p MM+ISceI vs. SR2+ISceI = 7.946e-12, KS test).(C) Immunofluorescence detection of γ-H2AX foci(red spots) in infected cells (4 RTU) either untreated(ctrl) or treated with antiretroviral drugs as indicated.(Right) The γ-H2AX foci signal merged with theGFP signal. (D) Quantification of γ-H2AX foci ininfected cells either untreated (ctrl-black bar) andin or treated as indicated (white bars) (means ±SEMfrom at least two independent experiments afterbackground subtraction). Drugs concentrations areindicated. (Put vs. 100nM AZT = 9.555e-05, Put vs. 1uM =4.485e-07, Put vs. 10uM= 4.384e-08, Put vs. 2uM nev= 0.0003,Put vs. 10uM nev = 8.572e-13, Put vs. 100nM ral = 3.466e-11, Put vs. 500nM ral < 2.2e-16, Put vs. 1uM ral < 2.2e-16,Put vs. 100nM CX = 0.0004, Put vs. 1uM CX = 3.046e-13,Put vs. 10uM CX = 6.122e-10, KS test). (See SI Materialsand Methods for background correction.)

5638 | www.pnas.org/cgi/doi/10.1073/pnas.1216254110 Di Primio et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

31, 2

021

These results demonstrate that SCIP is a technique to simul-taneously analyze HIV-1 expression and integration in individualcells. This single-cell investigation tool, as opposed to the con-ventional bulk approaches, is relevant to monitor the expression/integration profiles within a population of infected cells.

SCIP Analysis after Inhibition of Retroviral Replication. To furthertest the SCIP technique as a tool to analyze HIV-1 proviral DNAat single-cell level, infections were performed in different con-ditions, where viral replication was inhibited by cofactor knock-down or antivirals.Transportin-SR2 (TRN-SR2, TNPO3) is a cellular import

factor of SR-rich proteins that is involved in HIV-1 entry in thenuclei of infected cells (24–31). In fact, after TRN-SR2 knock-down HIV-1 nuclear entry is inhibited and, consequently, theamount of integrated DNA is reduced. Cells depleted for TRN-SR2 (using siRNA-SR2) and expressing the I-SceI endonucleasewere infected with the HIV-I-ΔGFP-SceI virus (4 RTU) andquantitatively analyzed for the amounts of γ-H2AX foci. Asshown in Fig. 3 A and B and Fig. S5, the number of γ-H2AX focidropped to background levels in TRN-SR2 knockdown cells thusindicating a lack of viral integration. Because alterations ofTRN-SR2 levels may interfere with H2AX phosphorylation, theformation of foci was verified in TRN-SR2 knockdown cellstreated with neocarzinostatin (NCS), a strong inducer of DSBsfoci formation. In these conditions H2AX molecules remainphosphorylated proving normal foci formation after depletion ofTRN-SR2 (Fig. S6B).To further test SCIP as an efficient tool for the analysis of HIV

replication steps up to integration, infections were performed withthe HIV-CMVGFP-I-SceI vector (4 RTU) in the presence of re-verse transcriptase inhibitors (AZT and nevirapine) or an integraseinhibitor (raltegravir) used in antiretroviral therapy. In addition,a recently developed antiviral compound, the LEDGIN CX05045,which inhibitsHIV-1 integration by disrupting the binding betweenintegrase and LEDGF/p75, was also used (32, 33).The quantitative analysis of these experiments demonstrates

that both reverse transcriptase and integrase inhibitors decreasedthe number γ-H2AX of foci (Fig. 3 C, Middle and D; and Fig. S5).The distinct activity of these drugs inHIV-1 replication is indicatedby the differential expression of GFP. In fact, GFP expression wasgreatly reduced in cells treated with reverse transcriptase inhib-itors (nevirapine and AZT) as reported both by microscopy (Fig.3C, Left) and by FACS analysis (Fig. S7) at 36 h postinfection.However, cells treated with integrase inhibitors (raltegravir andCX05045) show unaltered levels of GFP expression (Fig. 3C, Left;and Fig. S7) produced by nonintegrated viral DNA.Therefore, SCIP has the power to discriminate whether the

absence of integration following infection (no γ-H2AX foci) isgenerated by specific inhibition of viral integration (GFP positivecells) or by upstream viral impairments (GFP negative cells).

Integrated Viral DNA Localizes at the Nuclear Periphery. HIV-1preintegration complexes (PICs) preferentially localize at thenuclear periphery, suggesting that integration may preferentiallyoccur at the border of the nuclear compartment (14). To verifythis hypothesis we analyzed the localization of the integrationspots through SCIP. After measurement of the individual nuclearradius and the γ-H2AX foci radial positions, the γ-H2AX focidistances from the nuclear lamin were established. The analysis inU2OS cells revealed that 48 h postinfection in U2OS cells, 55% ofthe integrated viral DNA positioned within 1.5 μm of distancefrom the lamin (Fig. 4A and B, black bars). The same analysis wasperformed in CEMss T cells, a lymphoid cell line, thus morephysiologically relevant in HIV-1 infection. We observed that62% of HIV-1 proviruses localized in the periphery of the nu-cleus (Fig. 4 C and D) and were mostly detected within 0.5 μmdistance from the lamin. Therefore, compared with U2OS, cells

the peripheral positioning was even more pronounced. The viralarrangement along the inner surface of the nuclear lamin of theCEMss-infected cells clearly emerged from the z sectioning of theinfected cells (Fig. 4E).To verify that no bias was introduced by the SCIP approach on

the nuclear topology, the analysis was also performed in U2OSand CEMss cells treated with NCS to induce random DNADSBs. As shown in Fig. 4 B–D (red bars) NCS-induced γ-H2AXfoci do not show any preferential nuclear localization by dis-tributing into the whole nuclear compartment. Finally, γ-H2AXfoci generated by spontaneous DSBs in U2OS and CEMss cellsshows no preferential distribution within the nucleus in the ab-sence of infection (Fig. S8 A and B).

E

C

A B

D

Fig. 4. Analysis of the proviral DNA nuclear localization. (A) Immunofluo-rescence images of γ-H2AX foci (in red) and nuclear lamin (in blue) fromU2OS cells infected with 4 RTU of HIV-CMVGFP-I-SceI. Nuclear middle z stacksare shown. Scale bars, 10 μm. (B) Distribution of the distances of γ-H2AX focifrom the nuclear lamin. Black bars represent the frequency counts of γ-H2AXfoci in cells infected with HIV-CMVGFP-I-SceI, red bars represent the fre-quency counts of γ-H2AX foci induced by NCS (P = 3.273e-10, KS test). (C)Immunofluorescence images of γ-H2AX foci (in red) and nuclear lamin (inblue) from CEMss T cells infected with 4 RTU of HIV-CMVGFP-I-SceI. Nuclearmiddle z stacks are shown. Scale bars, 10 μm. (D) Distribution of the distancesof γ-H2AX foci from the nuclear lamin. Black bars represent the frequencycounts of γ-H2AX foci in cells infected with HIV-CMVGFP-I-SceI, red barsrepresent the frequency counts of γ-H2AX foci induced by NCS (P = 0.0004,KS test). (E) Z sectioning of an infected CEMss nucleus. Scale bars, 10 μm.

Di Primio et al. PNAS | April 2, 2013 | vol. 110 | no. 14 | 5639

MICRO

BIOLO

GY

Dow

nloa

ded

by g

uest

on

Janu

ary

31, 2

021

To investigate over time the behavior of theHIV-1 provirus in thenuclear compartment, the same topological analysis was also per-formed 13 d postinfection. HIV-1 proviruses distribution changedremarkably over the explored time interval (Fig. 5A). Moreover,according to the classical nuclear topology analysis where the nucleuscan be divided into three concentric zones of equal surface area(34, 35), the outcome of the SCIP analysis indicated a positioning ofHIV-1 proviruses at the nuclear periphery: at 48 h postinfection,the majority of the proviral DNA localized in the outer rim of thenuclear compartment and at day 13 the virus was randomly dis-tributed in the three nuclear zones defined (Fig. 5B).We next applied SCIP to cells where HIV-1 integration is

retargeted toward heterochromatin and intergenic regions, thusaltering the physiological lentiviral integration preference towardgene-rich units (36). This experimental system was obtained byexpressing in target cells a chimera LEDGF/p75 engineered tocontain an alternative chromatin-binding domain, CBX1, at its Nterminus (CBX-LEDGF325–530). The topological analysis of viralassociated γ-H2AX foci in CBX1 hybrid cell lines revealed thatproviral DNA is randomly distributed in the nuclei, showing nopreferential localization toward the nuclear periphery (Fig. 6 Aand B; see Fig. S8 C and D for background foci distribution).Therefore, SCIP analysis performed in infected cells showed

that HIV-1 preferentially integrate at the nuclear border in normalconditions, and random distribution is observed in cells engineeredto retarget integration toward noncanonical spots of the chroma-tin. Finally, proviral DNA localization is dynamic because repo-sitioning could be observed at later time points from infection.

DiscussionHere we report SCIP, an imaging technique to efficiently visualizeHIV-1 DNA integrated into the host genome of individual cells.The possibility to analyze viral integrated DNA within structurallyintact nuclei of individual cells allowed studying the topology ofHIV-1 in the nuclear compartment. This analysis revealed that at48 h postinfection, HIV-1 is preferentially localized in the nuclearperiphery. The observed spatial distribution is consistent with theone obtained by analyzing fluorescently labeled HIV-1 proteincomplexes at the nuclear level (14). It is interesting to note thatthe same analysis performed at day 13 from infection revealeda relocalization of proviral DNA toward the center of the nucleuswith a random distribution in the nuclear compartment. The lo-calization of PICs and provirus in the periphery of the nucleus atearly time points strongly suggests that the virus integrates soonafter its transition through the nuclear envelope. Nevertheless,once the HIV-1 genome is stably integrated into the host pe-ripheral chromatin, its radial distribution substantially changesover time probably owing to a rearrangement of the chromosometerritory where the integration occurred. We observed that in cellsengineered to retarget integration toward nonphysiological in-tegration spots (CBX-LEDGF325–530), proviral DNA does notshow any preferential nuclear distribution. These observationsclearly indicate that in addition to a linear preference of HIV-1toward gene-rich regions of the chromatin (36), the virus alsoacquires a specific spatial orientation within the nuclearcompartment.It is remarkable that SCIP does not detect linear or circular

nonintegrated viral DNA. This could be due to the conformationof the viral cDNA within the structure of the preintegrationcomplex, where the I-SceI target site would not be accessible toendonucleases because masked by viral and cellular factors (37).An alternative explanation could be that nonintegrated forms ofviral DNA may not be adequately chromatinized to determineγ-H2AX accumulation at I-SceI sites.

B

A

Fig. 5. Analysis of the proviral DNA nuclear localization at different timepoints after infection. (A) Distribution of distances of proviruses from thenuclear lamin at 48 h (black bars) and 13 d postinfection (red bars) (P= 5.76e-08,KS test). (B) Data are represented in bar graphs as the percentage of γ-H2AX fociin the three concentric zones of equal surface area. The peripheral zone is inblack, the middle zone in light gray, and the inner zone in gray.

B

A

Fig. 6. Analysis of the proviral DNA retargeting. (A) Distribution of thedistances of HIV-1 γ-H2AX foci from the nuclear lamin in HeLa control cells(black bars) and in cells expressing CBX-LEDGF325–530 (blue bars) (P = 2.147e-06, KS test). (B) Data are represented in bar graphs as the percentage ofγ-H2AX foci in the three concentric zones of equal surface area as in Fig. 5B.

5640 | www.pnas.org/cgi/doi/10.1073/pnas.1216254110 Di Primio et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

31, 2

021

Nevertheless, the presence of nonintegrated viral forms can beindirectly reported in γ-H2AX negative cells by the presence ofthe GFP expressed from the nonintegrated viruses. Therefore, bycombining GFP and SCIP it is possible to determine three dif-ferent infectivity conditions: (i) HIV-1 infection leading to com-pleted integration revealed by cells positive for both γ-H2AX andGFP signals, (ii) viral DNA reaches the nuclei of the cells butintegration is impaired revealed by cells negative for γ-H2AX andpositive for GFP owing to the expression of the nonintegratedDNA, and (iii) absence of viral DNA in the nuclei as a result ofreplication impairments upstream from nuclear import, revealedby the lack of both GFP and γ-H2AX signals.An additional important feature of SCIP is the possibility to

analyze individual infected cells without culture clonal selection,as opposed to the bulk analysis performed with conventionalapproaches. This advancement of the SCIP technique revealedthat cells are heterogeneously infected showing different densi-ties of HIV-1 genomes per cell (Figs. S4 and S5). This obser-vation is consistent with former reports showing that clonesderived from the same infection display variable numbers ofproviruses (38, 39).The power of SCIP to monitor expression /integration in in-

dividual cells might be key to investigate the still obscure mech-anisms leading to HIV-1 latency (40). Taken together, these data

strongly demonstrated that SCIP is a robust and clear-cut readoutfor understanding HIV-1 replication in details such as cellularconditions or factors affecting HIV-1 nuclear entry and the to-pology of integration at the individual cell level. The technicaladvantages and the multiple endpoints detection provided by thissystem are expected to contribute as a powerful tool also for large-scale analysis. In fact, high content analysis of γ-H2AX foci wasreported (41). Similarly, by coupling SCIP to a high-throughputfluorescence microscopy platform, the technique here reportedmay become a systematic qualitative and quantitative screen forHIV-1 potential cofactors and antiretroviral drugs.

Materials and MethodsFor a detailed description, please refer to SI Materials and Methods. Itincludes detailed procedures for confocal microscopy imaging acquisitionand analysis. It also includes description of constructs engineering, virusproduction, cell culture conditions, and immunofluorescence procedures.

ACKNOWLEDGMENTS. The authors are grateful to the laboratory of W. Thysand M. Giacca for valuable discussion and to A. Calvello for technicalassistance. This work was supported by grants from the European UnionSeventh Framework Programme (THINC, Health-2008-201032), by the IstitutoSuperiore di Sanità Italian AIDS Program (Grant 40H90), by the ProvinciaAutonoma di Trento (COFUND Project, Team 2009 – Incoming), and by FIRB2008 Futuro in Ricerca (RBFR08HSWG).

1. Campbell EM, Perez O, Anderson JL, Hope TJ (2008) Visualization of a proteasome-independent intermediate during restriction of HIV-1 by rhesus TRIM5alpha. J CellBiol 180(3):549–561.

2. Campbell EM, Perez O, Melar M, Hope TJ (2007) Labeling HIV-1 virions with twofluorescent proteins allows identification of virions that have productively enteredthe target cell. Virology 360(2):286–293.

3. Jha NK, et al. (2011) Imaging single retrovirus entry through alternative receptorisoforms and intermediates of virus-endosome fusion. PLoS Pathog 7(1):e1001260.

4. Jouvenet N, Bieniasz PD, Simon SM (2008) Imaging the biogenesis of individual HIV-1virions in live cells. Nature 454(7201):236–240.

5. Jouvenet N, Simon SM, Bieniasz PD (2009) Imaging the interaction of HIV-1 genomesand Gag during assembly of individual viral particles. Proc Natl Acad Sci USA 106(45):19114–19119.

6. Lelek M, et al. (2012) Superresolution imaging of HIV in infected cells with FlAsH-PALM. Proc Natl Acad Sci USA 109(22):8564–8569.

7. McDonald D, et al. (2002) Visualization of the intracellular behavior of HIV in livingcells. J Cell Biol 159(3):441–452.

8. Melikyan GB, Barnard RJ, Abrahamyan LG, Mothes W, Young JA (2005) Imaging in-dividual retroviral fusion events: From hemifusion to pore formation and growth.Proc Natl Acad Sci USA 102(24):8728–8733.

9. Miyauchi K, Kim Y, Latinovic O, Morozov V, Melikyan GB (2009) HIV enters cells viaendocytosis and dynamin-dependent fusion with endosomes. Cell 137(3):433–444.

10. Resh MD (2005) Intracellular trafficking of HIV-1 Gag: How Gag interacts with cellmembranes and makes viral particles. AIDS Rev 7(2):84–91.

11. Rudner L, et al. (2005) Dynamic fluorescent imaging of human immunodeficiencyvirus type 1 gag in live cells by biarsenical labeling. J Virol 79(7):4055–4065.

12. Steffens CM, Hope TJ (2004) Mobility of the human immunodeficiency virus (HIV)receptor CD4 and coreceptor CCR5 in living cells: Implications for HIV fusion and entryevents. J Virol 78(17):9573–9578.

13. Steffens CM, Hope TJ (2003) Localization of CD4 and CCR5 in living cells. J Virol 77(8):4985–4991.

14. Albanese A, Arosio D, Terreni M, Cereseto A (2008) HIV-1 pre-integration complexesselectively target decondensed chromatin in the nuclear periphery. PLoS ONE 3(6):e2413.

15. Arhel N, et al. (2006) Quantitative four-dimensional tracking of cytoplasmic and nu-clear HIV-1 complexes. Nat Methods 3(10):817–824.

16. Bell P, Montaner LJ, Maul GG (2001) Accumulation and intranuclear distribution ofunintegrated human immunodeficiency virus type 1 DNA. J Virol 75(16):7683–7691.

17. Nagel J, et al. (2012) Stably integrated and expressed retroviral sequences can in-fluence nuclear location and chromatin condensation of the integration locus.Chromosoma 121(4):353–367.

18. Porteus M (2011) Seeing the light: Integrating genome engineering with double-strand break repair. Nat Methods 8(8):628–630.

19. Bekker-Jensen S, et al. (2006) Spatial organization of the mammalian genome sur-veillance machinery in response to DNA strand breaks. J Cell Biol 173(2):195–206.

20. Soutoglou E, et al. (2007) Positional stability of single double-strand breaks inmammalian cells. Nat Cell Biol 9(6):675–682.

21. Nagy Z, Soutoglou E (2009) DNA repair: Easy to visualize, difficult to elucidate. TrendsCell Biol 19(11):617–629.

22. Butler SL, Hansen MS, Bushman FD (2001) A quantitative assay for HIV DNA in-tegration in vivo. Nat Med 7(5):631–634.

23. Pizzato M, et al. (2009) A one-step SYBR Green I-based product-enhanced reversetranscriptase assay for the quantitation of retroviruses in cell culture supernatants.J Virol Methods 156(1-2):1–7.

24. Ocwieja KE, et al. (2011) HIV integration targeting: A pathway involving Transportin-3 and the nuclear pore protein RanBP2. PLoS Pathog 7(3):e1001313.

25. Zhou L, et al. (2011) Transportin 3 promotes a nuclear maturation step required forefficient HIV-1 integration. PLoS Pathog 7(8):e1002194.

26. Brass AL, et al. (2008) Identification of host proteins required for HIV infectionthrough a functional genomic screen. Science 319(5865):921–926.

27. Christ F, et al. (2008) Transportin-SR2 imports HIV into the nucleus. Curr Biol 18(16):1192–1202.

28. König R, et al. (2008) Global analysis of host-pathogen interactions that regulateearly-stage HIV-1 replication. Cell 135(1):49–60.

29. Krishnan L, et al. (2010) The requirement for cellular transportin 3 (TNPO3 or TRN-SR2) during infection maps to human immunodeficiency virus type 1 capsid and notintegrase. J Virol 84(1):397–406.

30. Lai MC, Lin RI, Huang SY, Tsai CW, Tarn WY (2000) A human importin-beta familyprotein, transportin-SR2, interacts with the phosphorylated RS domain of SR proteins.J Biol Chem 275(11):7950–7957.

31. Thys W, et al. (2011) Interplay between HIV entry and transportin-SR2 dependency.Retrovirology 8:7.

32. Christ F, et al. (2012) Small-molecule inhibitors of the LEDGF/p75 binding site of in-tegrase block HIV replication and modulate integrase multimerization. AntimicrobAgents Chemother 56(8):4365–4374.

33. Christ F, et al. (2010) Rational design of small-molecule inhibitors of the LEDGF/p75-integrase interaction and HIV replication. Nat Chem Biol 6(6):442–448.

34. Alfredsson-Timmins J, Henningson F, Bjerling P (2007) The Clr4 methyltransferasedetermines the subnuclear localization of the mating-type region in fission yeast.J Cell Sci 120(Pt 11):1935–1943.

35. Bystricky K, et al. (2009) Regulation of nuclear positioning and dynamics of the silentmating type loci by the yeast Ku70/Ku80 complex. Mol Cell Biol 29(3):835–848.

36. Gijsbers R, et al. (2010) LEDGF hybrids efficiently retarget lentiviral integration intoheterochromatin. Mol Ther 18(3):552–560.

37. Miller MD, Farnet CM, Bushman FD (1997) Human immunodeficiency virus type 1preintegration complexes: Studies of organization and composition. J Virol 71(7):5382–5390.

38. Kustikova OS, et al. (2003) Dose finding with retroviral vectors: Correlation of ret-roviral vector copy numbers in single cells with gene transfer efficiency in a cellpopulation. Blood 102(12):3934–3937.

39. Zielske SP, Lingas KT, Li Y, Gerson SL (2004) Limited lentiviral transgene expressionwith increasing copy number in an MGMT selection model: Lack of copy numberselection by drug treatment. Mol Ther 9(6):923–931.

40. Margolis DM (2010) Mechanisms of HIV latency: An emerging picture of complexity.Curr HIV/AIDS Rep 7(1):37–43.

41. Kim S, Jun DH, Kim HJ, Jeong KC, Lee CH (2011) Development of a high-contentscreening method for chemicals modulating DNA damage response. J Biomol Screen16(2):259–265.

Di Primio et al. PNAS | April 2, 2013 | vol. 110 | no. 14 | 5641

MICRO

BIOLO

GY

Dow

nloa

ded

by g

uest

on

Janu

ary

31, 2

021