ARTICLE

Down-Regulation of CD81 Tetraspanin in HumanCells Producing Retroviral-Based Particles:Tailoring Vector Composition

A.F. Rodrigues,1,2 M.R. Guerreiro,1,2 V.M. Santiago,1 C. Dalba,3 D. Klatzmann,4

P.M. Alves,1,2 M.J.T. Carrondo,1,2,5 A.S. Coroadinha1,2

1Instituto de Biologia Experimental e Tecnologica (IBET), Apartado 12, P-2781-901 Oeiras,

Portugal; telephone: þ351-21-4469520; fax: þ351-21-4421161; e-mail: avalente@itqb.unl.pt2Instituto de Tecnologia Quımica e Biologica, Universidade Nova de Lisboa (ITQB-UNL),

Apartado 12, P-2781-901 Oeiras, Portugal3EPIXIS SA, F-75013 Paris, France4AP-HP, Hopital Pitie-Salpetriere, Biotherapy, F-75013 Paris, France5FCT-UNL, P-2829-516 Caparica, Portugal

Received 6 February 2011; revision received 20 April 2011; accepted 1 June 2011

Published online 8 June 2011 in Wiley Online Library (wileyonlinelibrary.com). DOI 10.1002/bit.23231

ABSTRACT: Retroviral-derived biopharmaceuticals (RV)target numerous therapeutic applications, from gene thera-py to virus-like particle (rVLP)-based vaccines. Duringparticle formation, beside the pseudotyped envelope pro-teins, RV can incorporate proteins derived from the virusproducer cells (VPC). This may be detrimental by reducingthe amounts of the pseudotyped envelope and/or by incor-porating protein capable of inducing immune responseswhen non-human VPC are used. Manipulating the reper-toire of VPC proteins integrated onto the vector structure isan underexplored territory and should provide valuableinsights on potential targets to improve vector pharmaco-kinetic and pharmacodynamic properties. In this work,human HEK 293 cells producing retrovirus-like particles(rVLPs) and infectious RV vectors were used to prove theconcept of customizing RV composition by manipulatingcellular protein content. The tetraspanin CD81 was chosensince it is significantly incorporated in the RV membrane,conferring to the vector significant immunogenicity whenused in mice. RNA interference-mediated by shRNAlentiviral vector transduction was efficiently used to silenceCD81 expression (up to 99%) and the rVLPs produced byknocked-down cells lack CD81. Silenced clones wereanalyzed for cell proliferation, morphological changes, sus-ceptibility to oxidative stress conditions, and rVLP produc-tivities. The results showed that the down-regulation of VPCproteins requires close monitoring for possible side effectson cellular production performance. Yet, they confirm thatit is possible to change the composition of host-derived

immunogens in RV by altering cellular protein content withno detriment for vector productivity and titers. This con-stitutes an important manipulation tool in vaccinology—byexploiting the potential adjuvant effect of VPC proteins orusing them as fusion agents to other proteins of interest to beexposed on the vector membrane—and in gene therapy, byreducing the immunogenicity of RV-based vector and en-hancing in vivo half-life. Such tools can also be applied tolentiviral or other enveloped viral vectors.

Biotechnol. Bioeng. 2011;108: 2623–2633.

� 2011 Wiley Periodicals, Inc.

KEYWORDS: retrovirus-like particles; retroviral vector;host proteins; shRNA; CD81; immunotoxicity; vaccinesand gene therapy

Introduction

Retroviral (RV)-based particles have been extensively used forbiotechnological applications ranging from gene therapy torVLP-based vaccines or cancer therapy (Dalba et al., 2005,2007; Edelstein et al., 2007). However, vector improvementshave primarily addressed the manipulation of virus-encodedgenes, little attention being paid to manipulating producer cellgenes for enhanced therapeutic performance or vector titer.

Retrovirus-like particles (rVLPs) derived from murineleukemia virus (MLV) emerged as an attractive platform forrecombinant vaccine development capable of inducing strongimmune responses both humoral and cellular, including theproduction of neutralizing antibodies (Bellier et al., 2006,2009). Furthermore, they provide a stable structure for antigenpresentation with a very simple and safe design (Dalba et al.,

Correspondence to: A.S. Coroadinha

Contract grant sponsor: European Commission (FP6)

Contract grant number: LSHB-CT2006-018933

Contract grant sponsor: Fundacao para a Ciencia e a Tecnologia (FCT)—Portugal

Contract grant number: PTDC/EBB-BIO/102649/2008; PTDC/EBB-BIO/100491/2008;

SFRH/BD/48393/2008

Additional supporting information may be found in the online version of this article.

� 2011 Wiley Periodicals, Inc. Biotechnology and Bioengineering, Vol. 108, No. 11, November, 2011 2623

2007). As enveloped particles, retrovirus acquire their lipidmembrane from the producer cell; during this process,considerable amounts of host proteins are incorporated ontothe viral membrane (Hammarstedt et al., 2000; Segura et al.,2008). These proteins are not restricted to those of the cellularmembrane (Ott, 1997; Segura et al., 2008); neither themechanisms of incorporation nor their potential role invirions are well understood (Ott, 2002). However, suchproteins can hinder virus efficiency when used as therapeuticvectors (Arthur et al., 1995). Some of the host proteinsexposed on the RV membrane are strongly immunogenic,reducing gene therapy vectors half-life in vivo, or skewing theimmune response of rVLP-based vaccines.

CD81, a membrane protein belonging to the tetraspaninsuperfamily, plays an important role in the immune systemand is known to influence a variety of other biologicalprocesses such as cell adhesion, migration, morphology,proliferation, and signal transduction (Maecker et al., 1997;Levy et al., 1998). Recently, CD81 has been identified as oneof the host cell proteins majorly incorporated into MLV-derived vectors (Segura et al., 2008). Furthermore, CD81present on MLV viral particles produced in human 293 cellsis highly immunogenic, inducing strongly neutralizingantibodies in mice (unpublished results). Therefore, theknock-down of CD81 is an attractive strategy to study theeffects of its incorporation in the immunotoxic profile ofRV-derived vectors.

RNA interference (RNAi) mediated by double-strandedRNA (dsRNA) has been a popular choice for knock-downstudies in basic functional genomics (Shan, 2010). Due to itsinherent simplicity and high efficiency, it also offersswiftness and flexibility to the cell line development field.Nevertheless, it is known that dsRNA-mediated RNAi canresult in several side consequences such as off-target effectsdue to the redundancy of short nucleotide sequences in thehuman transcriptome (Jackson et al., 2003) and interferon(IFN) activation (Bridge et al., 2003). Thus, the use of RNAias a tool in cell line development requires monitoring ofproduct quality and yields. Particularly, in the case of RVpackaging cells, IFN-associated antiviral response can resultnot only in reduced cell growth and viability but also inlower vector productivity.

In this work, we investigated the possibility of altering thecomposition of host immunogens incorporated in rVLPsproduced by a human cell line. Tetraspanin CD81 down-regulation via shRNA was targeted since it is largely integratedin RVmembrane, conferring a strong immunogenic characterto the vector in mice. This work highlights VPC proteinmanipulation as a potential tool to change RV composition.

Materials and Methods

Cell Lines and Culture Media

293rVLP is a HEK 293 (ATCC CRL-1573)-based cell lineestablished by stable transfection of pCeB plasmid, a pCRIP-derived vector (Danos and Mulligan, 1988) driving MLV

Gag-pol expression by the MLV 50LTR as described inCosset et al. (1995). 293 FLEX 18 is a HEK 293-derived cellline producing MLV-based vector expressing Gibbon ApeLeukemia Virus (GaLV) ecotropic envelope and harboring aLacZ reporter gene and established as described inCoroadinha et al. (2006a). These cells were used to establishthe shCD81 cell populations, their derivative clones and thenon-target shRNA control, by shRNA lentiviral vectortransduction. 293T cells (ATCC CRL-11268) were used forproducing and titrating shRNA lentiviral vector stocks andTe671 cell line (ATCC CCL-136) was used as the target cellto titrate infectious RV particles. All cells were maintained inDulbecco’s modified Eagle’s medium (DMEM) (Gibco,Paisley, UK) supplemented with 10% (v/v) foetal bovineserum (FBS) (Gibco).

Plasmids and Lentiviral Vector Production

Short hairpin (sh) RNA lentiviral vectors based on pLKO.1-puro (Moffat et al., 2006) containing a puromycin resistancegene and targeting five different sequences on the humanCD81 gene and a non-target shRNA vector were purchasedfrom Sigma Mission1RNAi (Sigma, St. Louis, MO).shCD81 hairpin sequences were TRCN0000057608 toTRCN0000057612, corresponding to the first version ofthe shRNA library from The RNAi Consortium (TRC 1)(Broad Institute, Cambridge, MA). For a matter ofsimplicity these sequences were designated shCD81 8 to12. The non-target control used was Sigma Mission1RNAiCat. No SHC002. For each of the shRNA, lentiviral vectorstocks were produced by transient co-transfection of 293Tcells with psPAX2 and pMD2.G [kindly provided by DidierTrono from the Swiss Federal Institute of Technology(EPFL) through Addgene plasmid repository (Cambridge,MA)], providing the packaging and the envelope functions,respectively.

Cell Transduction, Selection, and Cloning

For the establishment of the shRNA populations, 293rVLPand 293 FLEX 18 cells were seeded in six-well plates at1� 105 cells/cm2; 24 h later, culture medium was replaced by600mL of viral supernatant prepared in fresh culture mediumcontaining 8mg/mL of polybrene and diluted to achieve a lowmultiplicity of infection (MOI< 1). Identical MOIs were usedfor the different shRNA, including the non-target control.After 4 h of incubation at 378C to promote virus adsorption,1.5mL of fresh medium was added to each well. 48 h post-infection, the culture supernatant was exchanged forpuromycin-containing medium (1.5mg/mL) and cells werecultured for 21 additional days, exchanging culturemedium ateach 2 days, establishing the shCD81 8 to 12 cell populations.This period was previously confirmed to eliminate all non-transduced cells. For further enhancement of CD81 down-regulation, shCD81 10 cell population derived from 293rVLPwas re-infected with shCD81 9 and 12 lentiviral-transducingvectors, as described above, with an MOI of 1 for each vector,

2624 Biotechnology and Bioengineering, Vol. 108, No. 11, November, 2011

followed by 14 days of puromycin selection, establishing theshCD81 9þ 10þ 12 cell population. This population wascloned by limiting dilution in a total of 240 wells in 96-wellplates, seeded at a density of 0.5 cells/well and 30 clones wereisolated.

rVLP Production and Western Blotting

For rVLP production, 293 rVLP, shCD81 cell populations orclones were seeded at a density of 8� 104 cells/cm2 in four150 cm2 t-flasks (Nunc, Roskilde, Denmark) and culturedfor 3 days until approximately 80% of cellular confluencewas reached. Culture medium was replaced by fresh one andrVLPs produced during the subsequent 24 h were harvested,filtered through 0.45mm for clarification and concentratedby ultracentrifugation as described elsewhere (Coroadinhaet al., 2006b). CD81 in viral suspensions was detectedby Western blotting, after separation in a 4–12% (w/v)acrylamide NuPAGE gradient pre-cast gel (Invitrogen,Paisley, UK) using MOPS running buffer. Each well wasloaded with the same amount of total protein, assessed byBCA Protein Assay Kit (Thermo Scientific, Rockford, IL)following the manufacturer’s instructions. Non-reducedsamples were resolved for 80min at a constant voltage of125V and transferred into a PVDF membrane for 90min,which was then incubated with a mouse monoclonalantibody against human CD81 (Epixis SA, Paris, France) orwith a rat monoclonal antibody against MLV Gag producedby the hybridoma R187 (ATCC CRL-1912). Detection wasperformed with the corresponding secondary antibody,conjugated with alkaline phosphatase, revealed with 1-stepNBT/BCIP reagent (Thermo Scientific).

Infectious Vector Production and Titration

For infectious vector production, 293 FLEX 18, thecorresponding non-target control and 293 FLEX shCD8110 cell population were seeded in six-well plates at a densityof 8� 104 cells/cm2. Samples of viral supernatants werecollected with 24-h intervals during the cell growth, filteredthrough 0.45mm for clarification, aliquoted and stored at�858C until further analysis. For each culture, triplicateswere performed. For the infectious particle titration, Te671were used as target cells. Infectious procedure and viral titerdetermination were performed by serial dilution infectionassay, as previously described in Rodrigues et al. (2009).

Flow Cytometry for CD81 and GaLV Detection

The presence of CD81 on the cellular surface was detected byflow cytometry. Briefly, 1–10� 105 total cells were collectedby centrifugation (300 g, 10min, 48C) during differentculture times and incubated in phosphate-buffered saline(PBS) with 4% (v/v) of FBS containing a mouse monoclonalantibody against human CD81 (Epixis), during 1 h at 48C.For each cell line or population, an isotype control was alsoanalyzed by replacing the primary antibody with an IgGisotype antibody (Invitrogen). After incubation, cells were

washed twice by successive centrifugation, ressuspended incold PBS, and incubated in PBS with 4% (v/v) of FBScontaining the secondary antibody, Alexa Fluor1 488 dyeconjugated (Invitrogen) for 30min at 48C. A last double-washing step was performed and the percentage of cellsstaining positive for the CD81 presence was determined witha CYFlow space flow cytometer (Partec, Munster,Germany). The detection of GaLV on cell surface for 293FLEX 18 cells and shRNA cell population was performedwith a procedure similar to that described above, usingrabbit anti-serum against the GaLV envelope (Eurogentec,Seraing, Belgium) as primary antibody, a biotinylatedpolyclonal goat anti-rabbit (BD Biosciences, San Jose, CA)as secondary antibody and a streptavidin–phycoerythrinconjugate (BD Biosciences) as signal amplifier. Theacquisition and analysis of the results was performed withFlowMax1 (Partec) software.

Oxidative Stress Induction and Cell ViabilityAssessment

The susceptibility to oxidative stress injury was assessed byanalyzing tert-butyl hydroperoxide induced death rates.Cells were inoculated at a density of 8� 104 cells/cm2 in 24-well plates and cultured for 3 days until approximately 80%of confluence was reached. At the 3rd day, tert-butylhydroperoxide (Sigma) was added to increasing concentra-tions ranging from 1.3 to 374.4mM; 18 h later, both culturesupernatant and attached cells were harvested and ressus-pended in culture medium containing propidium iodide(PI) (Invitrogen) to a final concentration of 1mg/mL. Cellviability was quantified by negative PI staining by flowcytometry (Partec), and normalized to the control (0mMtert-butyl hydroperoxide). Triplicates were performed. IC50

values were obtained by non-linear fitting of the experi-mental data in the form of ‘‘normalized cell viability versuslog10[tert-butyl hydroperoxide]’’ using Graphpad Prism(Graphpad Software, Inc., La Jolla, CA), considering a 95%confidence interval.

Reverse Transcriptase Activity

To determine MLV reverse transcriptase (RT) activity theRetroSys C-Type RT activity kit (Innovagen, Lund, Sweden)was used, according to the manufacturers’ instructions. Allreactions were linear with respect to incubation time andamount of virus added. Each measurement was performedin duplicate.

Calculations

To determine cell specific growth rate and RT productivitiesthe Boltzmann equation was considered:

dP

dt¼ mPC (1)

Rodrigues et al.: Manipulation of Retroviral-Based Particles Composition 2625

Biotechnology and Bioengineering

where P represents either viable cell concentration or RTactivity, C is the viable cell concentration, t is the time andmP is the cell specific rate of each parameter in study. Theconfidence interval considered was 95%.

Results

CD81 Down-Regulation via shRNA Lentiviral VectorTransduction

To evaluate the possibility of reducing CD81 incorporationin rVLPs, lentiviral vector stocks coding for five RNAi shorthairpin sequences, were produced in 293T cells and used toinfect 293 rVLP cells. As negative control, a non-targetsequence (ntshRNA) was used, corresponding to a shRNAthat does not target any known sequence in the humangenome, but activates the RNAi pathway. Rather than non-transduced cells, non-target controls have been regarded asthe best negative control for RNAi studies, to distinguish thephysiological effects induced by RNAi pathway activationfrom those derived from gene silencing (Editorial, 2003;Scacheri et al., 2004). Also, a low multiplicity of infectionwas used to avoid RNAi cytotoxicity and prevent possiblehindering effects on rVLPs production (Fish and Kruithof,2004). Each cell population, shCD81 8 to 12, as well as thenon-target control, were analyzed for the presence of CD81by flow cytometry (Fig. 1); the percentage of CD81 negativecells was chosen as a measure of CD81 silencing efficiency(Supplementary Table I). CD81 down-regulation efficiencyvaried considerably among the five shRNA sequences,ranging from negligible in the case of shCD81 11 to a verystrong silencing effect in shCD81 10 cell population, withmore than 90% of the cells negative for CD81 (Fig. 1).ntshRNA population stained almost totally positive (94%)for CD81 confirming to be a suitable negative control. Theparental cell line, 293rVLP, presented identical values tontshRNA population (data not shown).

With a considerable silencing effect, the shCD81 10 cellpopulation (93% of CD81 negative cells) was used to assessthe efficiency of CD81 elimination from the rVLPs. Thus,rVLPs produced by shCD81 10 cells as well as ntshRNA and293rVLP parental cell line were concentrated and purified byultracentrifugation and analyzed for CD81 presence byWestern blotting (Fig. 2). Neither RNAi pathway activation(ntshRNA cells) nor CD81 down-regulation seemed tointerfere with the structural protein content of the viralparticles, since the Western blotting profiles revealed by a-MLV Gag primary antibody were identical to the non-transduced cells (Fig. 2B). When probed with an a-hCD81primary antibody, however, an evident reduction of CD81incorporation into the rVLPs produced by the shCD81 10cell population was observed (Fig. 2C). CD81 was detectedin two distinct bands (20 and 30 kDa). The presence of thesecond band (30 kDa) evidences complex formation withfurther proteins, possibly other tetraspanins as CD9, existingin the lipid membrane of the viral particle. This profile was

also observed in cellular extracts (data not shown) and hasbeen previously described by other authors (Segura et al.,2008).

The strong knock-down of CD81 in shCD81 10 cellpopulation with the consequent reduced incorporation ofthe protein on viral particles supported the hypothesis ofcomplete elimination of CD81 from rVLPs if produced by atotally silenced cell population. To increase CD81 down-regulation, the shCD81 10 cell population was doublyinfected with shCD81 9 and 12 (Fig. 1). Although no furtherselection could be achieved by infecting with the shCD81 9and 12 lentiviral vectors, since the resistance marker was thesame, a higher CD81 down-regulation effect in thispopulation was observed, when compared to the shCD8110. Only 1.5% of the transduced cells were positive for CD81which, considering the background values for the ntshRNAcontrol population (approximately 6%), suggested a totallyCD81 silenced population.

rVLP shCD81 Clones: CD81 Expression Dynamics andPhysiological Implications

dsRNA-based knock-down can result in lower proliferationrates due to dsRNA-meditated IFN activation or when thesilenced protein is relevant for cytostasis. Consequently, thesilenced phenotype in cell populations is typically unstablein the long-term owing to the proliferative advantage of cellswith less dsRNA load (Fish and Kruithof, 2004), anobservation that our results further substantiated (datanot shown). To circumvent unstable silencing, the shCD819þ 10þ 12 cell population was cloned by limiting dilutionand 30 clones were isolated (Table I). These cells were namedrVLP shCD81. The clones were markedly distinct regardingCD81 expression profile and also growth proliferationpatterns (Table I). Six of the clones, although surviving tothe cloning step, were never able to proliferate to colonizethe whole growth surface area; among the remaining 24clones, seven have shown substantially reduced proliferationrates (indicated with superscript letter ’a’ in Table I) andsome of them exhibited different morphological character-istics (Fig. 3). The doubling time for these clones was foundto be higher than 55 h, with a maximum value for clone #29reaching 70 h, which was drastically higher than the average35 h for the parental cell line. These cells have also shownconsiderably reduced recovery rates after freezing. CD81down-regulation profile, ranged from almost no silencingeffect in the case of rVLP shCD81 #20 to the totally silencedclones rVLP shCD81 #19, #29, and #30. It is also noteworthythat CD81 down-regulation profile was found to fluctuate5–10% when assessed at different culture times, particularlyfor clones with �90% of CD81 silencing.

To further the investigate the cellular response to CD81down-regulation and/or RNAi pathway activation, fiverVLP shCD81 clones were analyzed for cell growth, CD81expression dynamics in different growth phases, and deathsusceptibility under oxidative stress conditions. According

2626 Biotechnology and Bioengineering, Vol. 108, No. 11, November, 2011

to Table I, clones #14, #19, #22, and #30 were chosen asrepresentatives of the highest CD81 silencing (�85%) andclone #9 as a representative of intermediate CD81 silencing.Cell population ntshRNA was used as negative control forspecifically distinguishing CD81 silencing from RNAiactivation pathway effects; results were analyzed comparingto the non-transduced parental cell line, 293rVLP. With the

exception of clone #14, all the silenced clones have shownlower growth rates (Fig. 4B). This effect, however, seemed tobe attributed to CD81 down-regulation, as it was notdetected in the ntshRNA control. On the other hand, lowerhalf maximal inhibitory concentration values (IC50) underoxidative stress were observed for dsRNA-tranduced cells,including the ntshRNA population, suggesting that the

Figure 1. Flow cytometry analysis of human CD81 protein in 293rVLP cells transduced with CD81 short hairpin RNA sequences. For the populations shCD81 8 to 12, 293 rVLP

cells were infected with MOI � 0.1 and selected for 21 days in puromycin supplemented medium. The population shCD81 9þ 10þ 12, was obtained by infecting shCD81 10

population with shCD81 9 and 12 lentiviral vectors with MOI�1 for each vector, followed by 14 days of selection. For each cell population, the corresponding isotype control is

represented. ntshRNA: non-target short hairpin RNA, control vector, that activates RNAi pathway but does not target any human sequence; shCD81 8 to 12: silencing vectors,

targeting different sequences in the human CD81 mRNA as described in Material and Methods section.

Rodrigues et al.: Manipulation of Retroviral-Based Particles Composition 2627

Biotechnology and Bioengineering

activation of RNAi pathway played a role in injury induceddeath susceptibility profile of the silenced cells (Fig. 4C).Clone #9 was the exception. CD81 expression dynamics alsovaried considerable during culture time (Fig. 4A andTable II), matching our previous observations on down-regulation fluctuations of cells when analyzed at differentdays. The profile in Figure 4A suggests an oscillatory pattern,with the highest CD81 silencing occurring in the windowrange between the middle and the late exponential phase ofcell growth. Clone #14 showed consistently higher levels ofCD81 down-regulation, up to 99%, being the lowest value of� 90% achieved at 40 h of culture time (Fig. 4A andTable II). Therefore, this clone was selected to confirm theefficiency of CD81 removal in rVLPs produced by silencedcells.

Production of rVLPs With rVLP shCD81 Cell Lines

Reverse transcriptase in the viral supernatant was quantified,as a measurement of Gag-pol expression, to assess rVLPproductivity levels among the different shCD81 clones(Fig. 5B). Three out of the five clones presented lower RTproductivities, whereas clones #14 and #22 had shownidentical values when compared either with the non-targetsilencing control or with the non-transduced parental cellline. This suggested that CD81 down-regulation does nothave direct implications in viral particles production,although it may have an indirect effect, possibly mediatedby other physiological intervenients.

Finally, rVLP shCD81 #14, was used to produce rVLPs,which were concentrated and analyzed for the presence of

Figure 2. Analysis of CD81 incorporation on MLV-derived rVLPs. A: LDS-PAGE and (B and C) Western blotting profile of rVLPs produced by the parental cell line (293 rVLP) (1),

the non-targeting short hairpin RNA (ntshRNA) (2), and the 93% CD81 silenced (shCD81 10) cell population (3). VLPs were concentrated by ultracentifugation as described in

Material and Methods section and non-reduced samples were run in a 4–12% polyacrylamide gel in Bis–Tris buffer using MOPS running buffer. Identical amounts of protein were

loaded per lane. Primary antibodies in Western blotting: (B) a-MLV Gag (P30 capsid protein, �30 kDa) (C) a-human CD81 protein (�19 kDa). [Color figure can be seen in the online

version of this article, available at http://wileyonlinelibrary.com/bit]

Table I. CD81 down-regulation in rVLP shCD81 clones.

rVLP shCD81 clone CD81 silenced cells (%)

1 75

2 90a

3 59

4 17

5 42a

6 48

7 16

8 21

9 62a

10 62

11 b

12 9a

13 39

14 85

15 16

16 82

17 21

18 86

19 94a

20 3

21 b

22 90

23 b

24 b

25 b

26 72

27 b

28 89

29 97a

30 95a

aClones with evident lower proliferation ratebClones that, although surviving the limiting dilution cloning step, were

not able to proliferate.

2628 Biotechnology and Bioengineering, Vol. 108, No. 11, November, 2011

CD81 by Western blotting (Fig. 5B). rVLPs produced byrVLP shCD81 #14 were shown to be absent of host CD81further confirming the preliminary results obtained withshCD81 10 cell population (Fig. 2) and demonstrating thatCD81 can be eliminated from the viral membrane, whenproduced by silenced cells.

Infectious Vector Production With CD81 Silenced Cells

Although the main focus of the present study was evaluatingthe potential of reducing the incorporation of CD81 inretroVLPs for vaccination purposes, we further analyzed theeffects of CD81 down-regulation in 293 FLEX cells

Figure 3. Morphology changes induced by CD81 down-regulation. A: parental cell line, 293rVLP, (B) non-targeting control control, ntshRNA, (C–G) silenced CD81 clones, rVLP

shCD81 #9, #14, #19, #22, and #30, respectively.

Rodrigues et al.: Manipulation of Retroviral-Based Particles Composition 2629

Biotechnology and Bioengineering

producing infectious RV particles, to evaluate the impact ofsilencing CD81 in infectious vector productivity. Theprofiles of CD81 down-regulation in 293 FLEX cells wereidentical to those in 293rVLPs transduced cells for thedifferent shCD81 sequences, although the silencing effectswere slightly weaker (data not shown). Again, shCD81 10sequence yielded the strongest CD81 silencing profile, witharound 70% of CD81 negative cells. Thus, this population aswell as 293 FLEX cells and the corresponding non-targetcontrol were analyzed for infectious vector production.

CD81 and envelope expression on the cell surface were alsoassessed in the middle/late exponential phase of cell growthwhich, according to the previous results on CD81 expressiondynamics (Fig. 4A and Table II), corresponded to themaximum CD81 silencing. The results are summarized inTable III. CD81 down-regulation resulted in 1.6-foldincrease in envelope expression on the cell surface whencompared both with the parental cell line (P¼ 0.005) andwith the non-target control (P¼ 0.008). Yet, no significantincreases in infectious vector productivity could beassociated to CD81 silencing and/or increased GaLVincorporation.

Discussion

293rVLP and 293 FLEX cells producing MLV rVLPs andinfectious RV vector particles, respectively, were used toevaluate the strategy of customizing vector composition bymanipulating cellular protein content. CD81 tetraspaninwas chosen as a case study, given that it is significantlyincorporated into RV particles, conferring them a strongimmunogenic character in mice. No direct correlation wasobtained between CD81 down-regulation and rVLPproduction, evidenced either in terms of Gag proteinsprofile detected by Western blotting (Figs. 2 and 5B) or byRT activity quantification (Fig. 5A). Also no directcorrelation was observed between CD81 reduction andinfectious particles titers for 293 FLEX. These results suggestthat, although naturally incorporated on RV, CD81 is notessential for the assembly or secretion of fully functionalparticles. The same set of results (Figs. 2 and 5), showed thatRNAi pathway activation did not result in evident adverseeffects on rVLP production as the ntshRNA controlpresented RT level and Gag proteins profile detected byWestern blotting identical to the ones of non-transducedcells. This suggests that IFN response occurred at lowextension, if so, further supporting the use of low MOIs asan effective strategy to avoid strong IFN activation (Bridgeet al., 2003; Fish and Kruithof, 2004).

Figure 4. Characterization of rVLP shCD81 cells. A: CD81 expression profile

during culture time detected by flow cytometry, (B) specific cellular growth rates, and

(C) half maximal inhibitory concentration values (IC50) of rVLP shCD81 clones in

response to increasing concentrations of tert-butyl hydroperoxide in the culture

medium. Error bars in (B) and (C) correspond to 95% confidence interval as defined

in Material and Methods section. ntshRNA: non-targeting short hairpin RNA; 293 rVLP:

parental cell line; rVLP shCD81: selected clones silenced for CD81 protein.

Table II. CD81 expression profile during culture time.

CD81 silenced cells (%)

OscillationMaximum Minimum Average

293 rVLP 1.6 0.3 0.8 0.5

ntshRNA 3.8 0.4 2.1 1.5

rVLP ShCD81 #9 95.2 76.2 86.4 8.9

rVLP ShCD81 #14 99.1 92.8 97.1 2.5

rVLP ShCD81 #19 98.0 84.7 92.9 5.8

rVLP ShCD81 #22 96.8 72.3 88.9 10.4

rVLP ShCD81 #30 92.5 65.0 82.3 10.7

Culture samples were analyzed for CD81 expression every 24 h by flowcytometry, as described in Material and Methods section. Oscillationcorresponds to the standard deviation (n¼ 5) and is a measurement ofCD81 silencing stability during cell growth. Maximum values of rVLPshCD81 clones were obtained in the late exponential phase.

2630 Biotechnology and Bioengineering, Vol. 108, No. 11, November, 2011

Although no direct correlation effect in rVLP outputcould be noticed by the down-regulation of CD81, a set ofcellular changes was observed which required furtherattention in the context of rVLP production. With theexception of clone shCD81 #14, all the clones with morethan 90% of silencing showed reduced proliferation patterns(Table II and Fig. 4A). For some of those, a 2-fold increase inthe cellular doubling time (average of 35 h in the parentalcell line to 70 h in clone shCD81 #29) was observed. Indeed,

CD81 has been previously described to be implicated inproliferation and cytostasis of various cell types, includinglymphocytes (Oren et al., 1990), fibroblasts (Toledo et al.,2004), epithelial cells (Geisert et al., 2002), and astrocytes(Geisert et al., 1996). Therefore, it is not surprising that aconsiderable number of clones have shown alteredproliferation patterns. When plotting RV productivity(RT) as a function of cell growth, a nearly linear tendencyis obtained (R2¼ 0.85) pointing out that reduced cellproliferation due to CD81 down-regulation should becarefully evaluated in the context of vector production, as itmay result in reduced productivities.

In addition to disturbed growth patterns, altered cellmorphology was the most obvious change in some of theshCD81 clones (Fig. 3). Such changes had already beenreported by the use of anti-CD81 antibodies (Geisert et al.,1996); also, CD81 is known to interact with severalcytoskeleton proteins including actin (Penas et al., 2000;Crotta et al., 2006). CD81 silenced cells could then beexpected to evidence morphologic alterations. Nevertheless,not all the highly silenced clones showed markedmorphologic changes. This suggests that absolute levels ofCD81 silencing are not the only reason for altered cellmorphology. Considering vector production, cytoskeletonproteins are known to participate in retrovirus particlestrafficking and recently, actin has been demonstrated todrive MLV assembly and budding (Naghavi and Goff, 2007;Gladnikoff et al., 2009). Hence, the effects of CD81 down-regulation may result in lower rVLP output by shiftingcytoskeleton properties and/or functionality. Indeed, RTquantification did not show consistent Gag-pol reductionamong silenced clones (Fig. 5) but clones with evidentlyaltered morphology (Fig. 3) were the ones showing lower RTproductivities, particularly shCD81 #9. This supports thehypothesis that cellular changes (e.g., in cytoskeleton)induced by CD81 knock-down, but not CD81 knock-downper se, can result in lower rVLP productivities. These resultsalso recommended precaution in the choice of VPCproteins to be manipulated. In fact, many of the cellularproteins incorporated on the viral particles are thought tohave direct or indirect roles upon the virus life cycle (Ott,

Figure 5. Characterization of rVLPs produced with rVLP shCD81 cells.

A: Specific MLV reverse transcriptase (RT) productivity in rVLP shCD81 clones.

B,C: Western blotting profile of rVLPs produced by (1) the parental cell line and (2)

the highest CD81 silenced clone (rVLP shCD81 #14); (3) corresponds to a bovine serum

albumin negative control. Error bars in (A) correspond to 95% confidence interval as

defined in Material and Methods section. Primary antibodies in Western blotting:

a-MLV Gag (P30 capsid protein, �30 kDa) and a-human CD81 protein (�19 kDa).

ntshRNA: non-targeting short hairpin RNA; 293 rVLP: parental cell line; rVLP shCD81:

selected clones silenced for CD81 protein.

Table III. CD81 down-regulation in 293 FLEX producing infectious

retroviral vector.

CD81 silenced

cells (%)

Relative GaLV

expression

Infectious particles

productivity

(I.P./cell h)

293 FLEX 0.28 1.0� 0.2 0.018� 0.001

ntshRNA 0.18 0.98� 0.2 0.024� 0.002

ShCD81_10 70.61 1.6� 0.1 0.022� 0.002

GaLV expression on cell surface is relative to non-sh RNA transducedcells (293 FLEX); errors correspond to standard deviation (n¼ 3). GaLVexpression in shCD81 cells is significantly different from 293 FLEX(P¼ 0.005) and from ntshRNA cells (P¼ 0.008), given by a non-pairedtwo-tailed t-test using three independent cultures of each cell population.

The errors in infectious particles productivities correspond to standarddeviation (n¼ 3).

Rodrigues et al.: Manipulation of Retroviral-Based Particles Composition 2631

Biotechnology and Bioengineering

2002). Considering so, 293 FLEX cells producing infectiousRV vector were also silenced for CD81 expression, in orderto evaluate possible side effects on vector infectivity (TableIII). The results showed that not only no reduction ininfectious vector productivity occurs due to CD81 silencingbut also that it has a slight beneficial effect in envelopeexpression on cellular surface. However, it was not obviousif increased envelope incorporation also occurred for theproduced RV particles since this was not translated in amajor increase in infectious vector titer in the culturesupernatant. If so, and despite no major differences ininfectious vector productivities, it can be of relevance forinstance to increase down-stream yields, by increasing thebinding strength of the viral particles to the purificationmatrixes and/or to reduce infectivity lost due to envelopeshedding.

We investigated cell death susceptibility in rVLP-shCD81clones, due to the use of RNAi as a knock-down tool.The CD81 silenced clones analyzed revealed to be moresusceptible to the oxidative agent tert-butyl hydroperoxide(Fig. 4C), presenting evidently lower IC50 values, withthe exception of clone shCD81#9. However, the percentageof dead cells in the negative control (0mM tert-butylhydroperoxide) for this clone was much lower (74%) thanthe remaining ones (all above 90%); since the percentage ofviable cells was calculated relatively to the negative control,this could have possibly induced an artificially lessaccentuated viable cell decrease profile and, therefore, ahigher IC50 value. Decreased IC50 was also observed for thenon-target control, suggesting that this could be a sideconsequence either mediated by the RNA-triggered IFN-associated antiviral pathways or by other off-target effect(Kaufman, 1999; Scacheri et al., 2004). Also, the fact thatntshRNA cells showed the lowest IC50 value seems toindicate that CD81 knock-down did not play any additionalrole in increased cell death susceptibility. Also, it wasinteresting to notice the oscillation pattern of CD81 down-regulation during cell growth (Fig. 4). Indeed, thetranscription activity of RNA polymerase III, responsiblefor shRNAs expression, is known to be cell cycle dependentand coupled to growth control mechanisms (Dumay-Odelotet al., 2010), thus hindering stable and tight control of thesilenced phenotype. Therefore, considering both increasedcell susceptibility to stressful conditions and the fluctuationpatterns of the silenced phenotype, the use of RNAi mightbe reconsidered for the purpose of stable cell lineestablishment. However, it is evidently faster, simpler andmore cost-effective than other techniques such as knock-out, and therefore undeniably useful for preliminary stages.Additionally, for certain vital proteins, a knock-out may notbe possible requiring a compromise between cell survivaland down-regulation effects upon vector composition/properties.

In this work, it is demonstrated the possibility of changingthe composition of VPC proteins incorporated into rVLPsand infectious RV vectors. rVLPs used in this study arecurrently being studied as displaying platform in vaccine

development and therefore CD81 was chosen since itsstrongly immunogenic in mice. Yet, the results hereinpresent support the manipulation of cellular proteincontent, CD81 or others, as promising tool in future effortsto control RV composition. Immunotoxic proteins can beused as potential adjuvant in vaccinology, whereas the non-toxic ones can be potential fusion agents to other proteins ofinterest to be exposed on the vector membrane.Additionally, for gene therapy purposes, the removal ofimmunotoxic proteins in the producer cell may improve theefficacy of retro, lenti, or other enveloped virus vectors byreducing their in vivo immunogenic potential.

The authors acknowledge the financial support received from the

European Commission (LSHB-CT2006-018933) and from Fundacao

para a Ciencia e a Tecnologia (FCT)—Portugal (PTDC/EBB-BIO/

102649/2008 and PTDC/EBB-BIO/100491/2008). A.F. Rodrigues ac-

knowledge FCT for her PhD grant (SFRH/BD/48393/2008). The

authors also acknowledge Dr. Helena Vieira for help and fruitful

discussions on the cell death induction experiments and Dr. Rute

Castro for critical revision of this article.

References

Arthur LO, Bess JW, Jr, Urban RG, Strominger JL, Morton WR, Mann DL,

Henderson LE, Benveniste RE. 1995. Macaques immunized with HLA-

DR are protected from challenge with simian immunodeficiency virus.

J Virol 69(5):3117–3124.

Bellier B, Dalba C, Clerc B, Desjardins D, Drury R, Cosset FL, Collins M,

KlatzmannD. 2006. DNA vaccines encoding retrovirus-based virus-like

particles induce efficient immune responses without adjuvant. Vaccine

24(14):2643–2655.

Bellier B, Huret C,MiyalouM, Desjardins D, FrenkielMP, Despres P, Tangy

F, Dalba C, Klatzmann D. 2009. DNA vaccines expressing retrovirus-

like particles are efficient immunogens to induce neutralizing anti-

bodies. Vaccine 27(42):5772–5780.

Bridge AJ, Pebernard S, Ducraux A, Nicoulaz AL, Iggo R. 2003. Induction of

an interferon response by RNAi vectors in mammalian cells. Nat Genet

34(3):263–264.

Coroadinha AS, Schucht R, Gama-Norton L, Wirth D, Hauser H, Carrondo

MJ. 2006a. The use of recombinase mediated cassette exchange in

retroviral vector producer cell lines: Predictability and efficiency by

transgene exchange. J Biotechnol 124(2):457–468.

Coroadinha AS, Silva AC, Pires E, Coelho A, Alves PM, Carrondo MJ.

2006b. Effect of osmotic pressure on the production of retroviral

vectors: Enhancement in vector stability. Biotechnol Bioeng 94(2):

322–329.

Cosset FL, Takeuchi Y, Battini JL, Weiss RA, Collins MK. 1995. High-titer

packaging cells producing recombinant retroviruses resistant to human

serum. J Virol 69(12):7430–7436.

Crotta S, Ronconi V, Ulivieri C, Baldari CT, Valiante NM, Abrignani S,

Wack A. 2006. Cytoskeleton rearrangement induced by tetraspanin

engagement modulates the activation of T and NK cells. Eur J Immunol

36(4):919–929.

Dalba C, Bellier B, Kasahara N, Klatzmann D. 2007. Replication-competent

vectors and empty virus-like particles: New retroviral vector designs for

cancer gene therapy or vaccines. Mol Ther 15(3):457–466.

Dalba C, Klatzmann D, Logg CR, Kasahara N. 2005. Beyond oncolytic

virotherapy: Replication-competent retrovirus vectors for selective and

stable transduction of tumors. Curr Gene Ther 5(6):655–667.

Danos O, Mulligan RC. 1988. Safe and efficient generation of recombinant

retroviruses with amphotropic and ecotropic host ranges. Proc Natl

Acad Sci USA 85(17):6460–6464.

2632 Biotechnology and Bioengineering, Vol. 108, No. 11, November, 2011

Dumay-Odelot H, Durrieu-Gaillard S, Da Silva D, Roeder RG, Teichmann

M. 2010. Cell growth- and differentiation-dependent regulation of

RNA polymerase III transcription. Cell Cycle 9(18):3687–3699.

Edelstein ML, Abedi MR, Wixon J. 2007. Gene therapy clinical trials

worldwide to 2007—An update. J Gene Med 9(10):833–842.

Editorial. 2003. Whither RNAi? Nat Cell Biol 5(6):489–490.

Fish RJ, Kruithof EK. 2004. Short-term cytotoxic effects and long-term

instability of RNAi delivered using lentiviral vectors. BMCMol Biol 5:9.

Geisert EE, Jr, Abel HJ, Fan L, Geisert GR. 2002. Retinal pigment epithelium

of the rat express CD81, the target of the anti-proliferative antibody

(TAPA). Invest Ophthalmol Vis Sci 43(1):274–280.

Geisert EE, Jr, Yang L, IrwinMH. 1996. Astrocyte growth, reactivity, and the

target of the antiproliferative antibody, TAPA. J Neurosci 16(17):5478–

5487.

Gladnikoff M, Shimoni E, Gov NS, Rousso I. 2009. Retroviral assembly and

budding occur through an actin-driven mechanism. Biophys J

97(9):2419–2428.

Hammarstedt M, Wallengren K, Pedersen KW, Roos N, Garoff H. 2000.

Minimal exclusion of plasma membrane proteins during retroviral

envelope formation. Proc Natl Acad Sci USA 97(13):7527–7532.

Jackson AL, Bartz SR, Schelter J, Kobayashi SV, Burchard J, Mao M, Li B,

Cavet G, Linsley PS. 2003. Expression profiling reveals off-target gene

regulation by RNAi. Nat Biotechnol 21(6):635–637.

Kaufman RJ. 1999. Double-stranded RNA-activated protein kinase med-

iates virus-induced apoptosis: A new role for an old actor. Proc Natl

Acad Sci USA 96(21):11693–11695.

Levy S, Todd SC, Maecker HT. 1998. CD81 (TAPA-1): A molecule involved

in signal transduction and cell adhesion in the immune system. Annu

Rev Immunol 16:89–109.

Maecker HT, Todd SC, Levy S. 1997. The tetraspanin superfamily:

Molecular facilitators. Faseb J 11(6):428–442.

Moffat J, Grueneberg DA, Yang X, Kim SY, Kloepfer AM, Hinkle G, Piqani

B, Eisenhaure TM, Luo B, Grenier JK, Carpenter AE, Foo SY, Stewart

SA, Stockwell BR, Hacohen N, Hahn WC, Lander ES, Sabatini DM,

Root DE. 2006. A lentiviral RNAi library for human and mouse

genes applied to an arrayed viral high-content screen. Cell 124(6):

1283–1298.

Naghavi MH, Goff SP. 2007. Retroviral proteins that interact with the host

cell cytoskeleton. Curr Opin Immunol 19(4):402–407.

Oren R, Takahashi S, Doss C, Levy R, Levy S. 1990. TAPA-1, the target of an

antiproliferative antibody, defines a new family of transmembrane

proteins. Mol Cell Biol 10(8):4007–4015.

Ott DE. 1997. Cellular proteins in HIV virions. Rev Med Virol 7(3):167–

180.

Ott DE. 2002. Potential roles of cellular proteins in HIV-1. Rev Med Virol

12(6):359–374.

Penas PF, Garcia-Diez A, Sanchez-Madrid F, Yanez-Mo M. 2000.

Tetraspanins are localized at motility-related structures and involved

in normal human keratinocyte wound healing migration. J Invest

Dermatol 114(6):1126–1135.

Rodrigues AF, CarmoM, Alves PM, Coroadinha AS. 2009. Retroviral vector

production under serum deprivation: The role of lipids. Biotechnol

Bioeng 104(6):1171–1181.

Scacheri PC, Rozenblatt-Rosen O, Caplen NJ, Wolfsberg TG, Umayam L,

Lee JC, Hughes CM, Shanmugam KS, Bhattacharjee A, Meyerson M,

Collins FS, 2004. Short interfering RNAs can induce unexpected and

divergent changes in the levels of untargeted proteins in mammalian

cells. Proc Natl Acad Sci USA 101(7):1892–1897.

Segura MM, Garnier A, Di Falco MR, Whissell G, Meneses-Acosta A,

Arcand N, Kamen A. 2008. Identification of host proteins associated

with retroviral vector particles by proteomic analysis of highly purified

vector preparations. J Virol 82(3):1107–1117.

Shan G. 2010. RNA interference as a gene knockdown technique. Int J

Biochem Cell Biol MicroRNAs Dev Dis 42(8):1243–1251.

Toledo MS, Suzuki E, Handa K, Hakomori S. 2004. Cell growth regulation

through GM3-enriched microdomain (glycosynapse) in human lung

embryonal fibroblast WI38 and its oncogenic transformant VA13. J

Biol Chem 279(33):34655–34664.

Rodrigues et al.: Manipulation of Retroviral-Based Particles Composition 2633

Biotechnology and Bioengineering