Gene Therapy (2012) 19, 1058 --1064 & 2012 Macmillan Publishers Limited All rights reserved 0969-7128/12 www.nature.com/gt

ORIGINAL ARTICLE A competitive growth for the detection of subtle effects of gene transduction on cell proliferation

JJM Eekels1, AO Pasternak1, AM Schut1, D Geerts2, RE Jeeninga1 and B Berkhout1

RNA interference (RNAi) is a sequence-specific gene silencing mechanism with therapeutic potential against many human pathogens. To obtain a durable therapeutic effect, stable transduction of target cells with for instance a lentiviral vector that expresses a short hairpin (shRNA) inducer of the RNAi pathway is necessary. Apart from the intended therapeutic effect, this treatment can induce negative effects on cell proliferation via off-target effects. A careful evaluation of the transduced cells is required to develop a safe gene therapy approach. Stably transduced cells are usually selected by expression of the enhanced green fluorescent protein (GFP) marker. In this study we show that the mixed transduction culture, containing both transduced GFP þ and untransduced GFPÀ cells, can simply be passaged to score the GFP þ /GFPÀ ratio by longitudinal flow cytometric analysis as a measure of the negative impact of the RNAi treatment on the cellular proliferation rate. We show that this assay is sensitive, easy to use and internally controlled for assessing subtle effects on cell proliferation of lentiviral transduction and transgene expression.

Gene Therapy (2012) 19, 1058--1064; doi:10.1038/gt.2011.191; published online 24 November 2011 Keywords: lentiviral vector; transduction; cell growth; RNA interference

INTRODUCTION that functions as cofactor for virus replication can affect cell RNA interference (RNAi) is an evolutionary conserved mechanism growth. Caution is particularly warranted in case of stable gene induced by double-stranded RNA, leading to sequence-specific knockdown, as subtle differences in cell growth rates can have gene silencing at the post-transcriptional level.1 RNAi can be a major impact on biological assays in diverse in vitro settings. induced transiently by transfecting target cells with small inter- The use of a stably integrating retroviral or lentiviral vector can fering RNAs (siRNAs) or stably by intracellular expression of short cause additional adverse effects due to genome integration, hairpin RNAs (shRNAs). Both strategies can be exploited for thera- for example, disruption of the regulation of local cellular genes. peutic purposes against a wide variety of diseases or microbial Obviously, in vivo RNAi and in fact any transgene application pathogens such as viral infections.2--6 Both viral RNA and cellular should put an emphasis on addressing such putative adverse transcripts that encode cofactors necessary for viral replication can effects on cell physiology and cell growth. be targeted.7 Especially for the attack on viruses that cause a Gross cell growth defects can be monitored using a variety of persistent infection the preferred method is the generation of assays. Staining with and other dyes, such as 4’,6- stable knockdown cells, in which constitutive expression of the diamidino-2-phenylindole or Hoechst 33342, can distinguish antiviral shRNAs is achieved by vector transduction. In particular, between live and dead cells. To calculate the cell doubling time, lentiviral vectors have been very successful because they are able the increase in number of cells can be measured over time, either to transduce many different cell types, both actively dividing, and by direct counting of the cells (manually or automatically) or quiescent cells, in which the viral genome is stably integrated in indirectly by reporter luciferase assays that measure the ATP the host cell DNA. A marker, which usually encodes for antibiotic content of metabolically active cells. However, there is a need for resistance or a fluorescent marker protein like the green fluo- a simple assay to score for more subtle cell growth effects. rescent protein (GFP), can be used to select the transduced cells. We explored the option to maintain the mixed culture obtained An important consideration when developing gene therapeutic after gene transduction that consists of transduced GFP þ and strategies based on RNAi is the potential impact on the physiology non-transduced GFPÀ cells. The GFP þ /GFPÀ ratio can simply be and viability of the transduced cells. RNAi can have adverse effects evaluated by Fluorescence-activated cell sorting (FACS) analysis on cellular properties and affect cell growth for several reasons. over time, which allows one to score minor cell growth defects as First, off-target effects of the shRNA on an unspecified mRNA with a gradual loss of the percentage GFP þ cells. This assay is based on partial sequence complementarity can affect cell growth in an the competitive cell growth between transduced and non-trans- unpredictable manner.8 Second, overexpression of shRNA mole- duced cells. We compared this competitive cell growth (CCG) cules can saturate components of the RNAi pathway, and thus assay to established methods of counting cells (both by FACS and disturb normal cellular gene regulation by microRNAs.9,10 Third, with a hemacytometer), the MTT assay and a commercially available shRNA overexpression can trigger innate immune responses such ATP bioluminescence assay. We demonstrate that the CCG assay as the interferon cascade.11 --13 Fourth, targeting of a cellular protein provides a sensitive, internally controlled and simple procedure.

1Laboratory of Experimental Virology, Department of Medical Microbiology, Center for Infection and Immunity Amsterdam, Academic Medical Center of the University of Amsterdam, Amsterdam, The Netherlands and 2Department of Human Genetics, Academic Medical Center of the University of Amsterdam, Amsterdam, The Netherlands. Correspondence: Dr B Berkhout, Laboratory of Experimental Virology, Department of Medical Microbiology, Academic Medical Center of the University of Amsterdam, Meibergdreef 15, Amsterdam 1105 AZ, The Netherlands. E-mail: [email protected] Received 31 March 2011; revised 14 September 2011; accepted 21 September 2011; published online 24 November 2011 Competitive cell growth assay JJM Eekels et al 1059 RESULTS in the transduced cells (Figure 2a, upper panels). These results also Design of the competitive cell growth assay indicate that integration of the lentiviral vector into the host cell To establish an easy and sensitive assay to monitor changes in genome (SHC1) and the induction of the RNAi pathway (SHC2) proliferation of transduced cells we developed the CCG assay. This have no effect on the cell proliferation capacity in this setting. assay is based on the assumption that direct competition between However, a steady reduction in the percentage of GFP-expressing transduced and untransduced cells will be the most sensitive and cells was observed over time when a vector encoding a shRNA controlled way to compare growth differences. In fact, both against specific targets was used. The reduction was small in the untransduced and transduced cells will be present in the culture case of the least toxic shRNA1 and increased when more toxic after transduction and this crude transduction mixture can thus shRNAs were expressed (Figure 2a, lower panels). This reduction be used directly in the CCG assay. This provides a direct the was observed with all three lentiviral vector inputs, thus no careful advantage over other proliferation assays, for which selection of titration is needed to score toxic effects. the transduced cells is necessary before one can determine the When one wants to determine the quantitative differences, the impact on cell growth (Figure 1). growth defect of the GFP þ cells can be calculated based on the To critically test the competitive CCG assay we used several known doubling time of SupT1 cells, the GFP þ /GFPÀ cell ratio at shRNA-expressing lentiviral vectors. Two control lentiviral vectors the start and end of the experiment and the time period. For each were used; SHC1 lacking a shRNA cassette (empty vector) and cell line and for each lentiviral vector input, we calculated the SHC2 expressing a scrambled shRNA molecule that has no percentage growth rate defect based on a doubling time of 1.1 identifiable mRNA target in human cells. Three shRNA constructs days for SupT1 cells. This doubling time is the average of values were selected that target specific cellular mRNAs and that were scored in the four alternative cell proliferation assays (Figure 5b, previously demonstrated---based on visual inspection of long-term lower panels). The most prominent differences were observed for cultures---to have a small effect on cell proliferation (shRNA1), an shRNA3 with 10 ml lentiviral vector input. The shRNA3 cells intermediate effect (shRNA2) or a severe effect (shRNA3). We proliferated 21±6% slower than the SupT1 control cells in the currently do not know whether this toxicity is due to downregulation same culture (Figure 2b). For shRNA1 and shRNA2 also significant of the specific mRNA target or due to an unspecific off-target effect. differences were measured when compared with SupT1 cells, with However, the impact on cell physiology was reproducible, which 11±1% and 11±0.5% reduction in growth rate, respectively. With made them excellent candidates to validate the CCG assay. a lower lentiviral vector input, the shRNA-expressing cells again Lentiviral vectors encoding these shRNAs were generated and exhibited a reduced proliferation rate, but this difference is only used to stably transduce the SupT1 T-cell line. A fixed amount of significant for the most toxic shRNA3. No growth rate differences SupT1 cells was transduced with 0.1, 1 or 10 ml of the lentivirus were observed between SupT1 and the two control cell lines SHC1 stocks. The titer of all vectors used in this study was comparable and SHC2 (Figure 2b). (data not shown). These cultures were maintained as normal The CCG assay is compatible with high-throughput screening. SupT1 cell lines (1 in 10 split twice weekly) over a period of 26 When many transduced cultures have to be analyzed in parallel, days. Two days post transduction one usually determines the per- one simply maintains the transduction mixtures in a 24- or even a centage of transduced, GFP-positive cells, which can be FACS- 96-wells format. Many flow cytometers are equipped with a plate sorted for further experimentation. However, the CCG assay is reader, thus allowing the screening of hundreds of cultures within performed on the raw transduction mixture of transduced and an hour. For instance, we analyzed 19 shRNA-expression lentiviral untransduced cells. This cell mixture is simply maintained and vectors, including the controls SHC1 and SHC2 at a low (1 ml) and periodically scored for the percentage of GFP-positive cells. high (100 ml) transduction level (Figure 3). The shRNAs are shown The percentage of GFP-expressing cells in the control cell lines in increasing order of , shRNA4 being the least SHC1 and SHC2 was stable over time, indicating a lack of toxicity cytotoxic with a 4% decrease in proliferation rate compared with

CCG assay

GFP selection Expansion Cell proliferation assays

Limiting dilution Expansion

Figure 1. The CCG assay versus other cell proliferation assays. The CCG assay follows the percentage of transduced GFP þ cells over time, and is performed on the crude transduction mixture, providing fast and internally controlled results. Other cell proliferation assays requires the selection of pure GFP þ transduced cells, for example, by FACS sorting and subsequent expansion. In some cases, one even needs to generate clonal cell lines by limiting dilution and by expanding the clonal cells before proliferation assays can be performed.

& 2012 Macmillan Publishers Limited Gene Therapy (2012) 1058 --1064 Competitive cell growth assay JJM Eekels et al 1060

Figure 2. Calibration of the CCG assay. (a) SupT1 cells were transduced with 0.1 (multiplicity of infection between 0.01 and 0.04), 1 (multiplicity of infection between 0.1 and 0.4) and 10 ml (multiplicity of infection between 1 and 4) of lentiviral vector. Cells were cultured and passed as normal twice weekly and samples were taken for FACS analysis. Control cell lines SHC1 (empty vector) and SHC2 (expressing a scrambled shRNA) are shown in the upper panels and show a stable percentage of GFP-expressing cells (y axis) over time (x axis). Lentiviral vectors expressing shRNAs against three cellular factors (lower panels) show a gradual decrease in the number of GFP þ cells. The experiment was performed twice and one representative experiment is shown. (b) The percentage difference in growth rate between SupT1 and GFP þ cells was calculated based on two experiments. No significant differences were measured in the control cell lines SHC1 and SHC2 when compared with SupT1. All shRNA-expressing cell lines showed a slower proliferation rate with 10 ml lentivirus input.

SupT1, and shRNA20 showing the biggest decline in proliferation The CCG assay was performed with a mix of the sorted cells in a rate; growing 23% slower than SupT1. These cell lines were 1:1 ratio with untransduced SupT1 cells. A sample was analyzed analyzed once every week, with a maximal assay time of 1 h. daily by flow cytometry to determine the decrease in GFP per- centage (Figure 5a). The percentage of GFP þ cells remained constant in the control cultures SHC1 and SHC2, but decreased Decrease in percentage of GFP-expressing cells is due to outgrowth of non-transduced cells due to shRNA1, shRNA2 and shRNA3 expression. The percentage difference in growth rates was compared with SHC1 (Figure 5b, We reasoned that the decrease in percentage of GFP-expressing upper panel). The difference in growth rates was significantly cells results from outgrowth of untransduced cells, but it cannot higher for shRNA1, shRNA2 and shRNA3, with shRNA1 being the formally be excluded that GFP-positive cells loose transgene least toxic, with a 9±2% slower growth rate. A 34±5% slower expression over time. Such transgene silencing has been reported growth rate was measured for shRNA3 when compared with SHC1 on many occasions.14,15 Although the polymerase III-driven shRNA mixed culture. cassettes seem to be rather sturdy,16 the polymerase II-driven GFP Of the four alternative proliferation assays, the easiest way to unit may be affected. To distinguish between these two pos- measure cell proliferation in culture is by counting the number sibilities, a new batch of cells was transduced with shRNA3 using a of cells over time. This can be performed manually, using a high lentiviral vector input. After sorting, this culture was 99.8% GFP-positive, but also in this culture the percentage of GFP- hemacytometer in combination with trypan blue staining of dead positive cells decreased after 40 days. Clonal transduced cells were cells in the culture sample (‘manual counting’ in Figure 5a). generated by limiting dilution on day 48 post transduction and Although easy to use, this is a laborious method that is not ap- the parental culture was frozen. The clonal cells and the thawed plicable for high-throughput screening. We also determined the parental culture were analyzed for GFP expression over a period number of live cells by flow cytometry (‘automated counting’ in of 13 days, which corresponds to 68--99 days post transduction Figure 5a). A fixed volume (100 ml) of was analyzed for the clonal cells and days 61--74 for the parental culture. daily. This method allows one to distinguish live from dead cells FACS analysis on day 13 of the experiment indicated the expected and the increase in number of cells provides a measure of cell loss of GFP-positive cells in the parental culture, but no loss of proliferation. Furthermore, the procedure is suitable for high- GFP expression was observed in the two clonal cells (Figure 4). throughput experiments. We therefore conclude that the decrease in percentage of Some proliferation assays measure the increase in metabolic GFP-expressing cells in the CCG assay is due to outgrowth of activity due to an increase in cell number. The most well-known 17 untransduced over transduced cells. test is the MTT assay. The yellow dye MTT is conversed to purple formazan crystals when added to the cell culture and the absorbance can be quantified with a spectrophotometer. This Comparison of the CCG assay with other well-established methods conversion only takes place when reductase enzymes are active, A variety of methods can be used to measure the impact of a gene thus only when live cells are present in the culture. The more live therapy treatment on cell proliferation. Here we compared the cells in a culture, the more purple formazan is produced, resulting new CCG assay with several well-established methods. Cells in higher absorbance values over time (‘MTT’ in Figure 5a). We also transduced with 100 ml lentiviral vector were FACS sorted for GFP tested an ATP bioluminescence assay that measures the ATP expression and subsequently expanded. This directly reveals an content of a sample, which is an indicator for metabolically active advantage of the CCG assay, in which these extra steps are not cells.18 A sample is taken daily and mixed with the assay reagent required (Figure 1). Five different cell proliferation assays, including containing luciferase, and the light signal is measured in a the CCG test, were performed in parallel over a period of 5 days. luminometer. The luciferase signal is proportional to the ATP

Gene Therapy (2012) 1058 --1064 & 2012 Macmillan Publishers Limited Competitive cell growth assay JJM Eekels et al 1061

Figure 3. High-throughput analysis of GFP-positive cell lines. (a) SupT1 cells were transduced with 1 or 100 ml lentiviral vector, expressing shRNAs against different cellular mRNAs or with the control lentiviral vectors SHC1 and SHC2. The cells were passaged normally and weekly samples were taken for FACS analysis. (b) The percentage of GFP-positive cells (y axis) per shRNA is shown, with the 1-ml transduced samples in gray and the 100-ml transduced samples in black bars. Cell lines were followed for 33 days post transduction (x axis). present in the sample and thus the amount of metabolically active DISCUSSION cells (‘ATP’ in Figure 5a). In this study we describe a new method to assess small changes in A single representative curve is shown for each of the four the proliferation rate of cells treated with a viral vector harboring assays (Figure 5a, lower panels). To compare the proliferation a fluorescent selection marker, for example, GFP. This method, assays, the doubling time of each cell line was calculated based on the CCG assay, is based on scoring the differences in proliferation two experiments and normalized to SupT1, of which the doubling rates between transduced and untransduced cells in the same culture. time was set at 1 (Figure 5b). The doubling times of the cell lines As transduced cells express GFP in addition to a (therapeutic) show a clear trend where shRNA3 is causing a cell growth delay, transgene, the loss in percentage of GFP-positive cells over time is but these differences were not significant except for the manual a direct measure of the cell proliferation defect of transduced cells counting method. No significant differences in cell growth could compared with untransduced cells. be scored for the weak and intermediate toxic shRNA1 and The CCG assay has some advantages over other well-established shRNA2. Thus, the CCG assay is clearly most sensitive as significant cell proliferation assay. First of all, the raw transduction mixture cytopathic effects were reproducibly measured for all three can be tested directly, that is, no extra steps of selection and shRNAs. expansion of transduced cells are necessary. Importantly, the CCG

& 2012 Macmillan Publishers Limited Gene Therapy (2012) 1058 --1064 Competitive cell growth assay JJM Eekels et al 1062 Day 1 significantly detected in the CCG assay, and although all tested proliferation assays show similar shRNA toxicity, only manual Day 13 counting revealed a significant toxic effect of the most toxic Mixture shRNA3. In the CCG assay even smaller growth defects can be observed. Although some of these values may not be statistically significant, the fact that the percentage of GFP þ cells decreases over time is a strong indication of a minor proliferation defect. For some of the shRNA cassettes we have been able to correlate cell growth problems as measured in vitro with the CCG test with in vivo results in humanized mouse models (unpublished results). The CCG assay is not limited to human T-cell lines or to shRNA research. Other cell types can be used, both suspension and adherent cells and even primary cells, at least when a certain level

2 3 4 5 of proliferation is observed in 5 days. In case of primary cells, one 10 10 10 10 should keep in mind that a culture of primary cells often consists of different sub-populations, each with a different transducibility. Clonal 1 If GFP measurement over time can be combined with FACS analysis of cell-type-specific markers, one could even monitor growth of the different cell types in a single assay. It remains important to include proper controls, such as an empty lentiviral vector. In this study we scored for the impact of shRNA expression

Count on cell growth, but any transduction event can be assessed in the CCG assay, for instance cellular problems due to vector integration or transgene expression. Not only a decrease in cell proliferation can be measured but also an increase in percentage of GFP þ cells over time, which would indicate an increased cell proliferation 102 103 104 105 rate. This may reveal oncogenic changes in the transduced cells. In exceptional cases the shRNA could have an off-target impact on the GFP reporter, which could be misinterpreted as cellular Clonal 2 toxicity. However, such an effect will cause a reduced GFP intensity, which was not observed for the clonal cells (Figure 4). An increase in percentage of GFP þ cells may be observed when the transgene provides a selective advantage under the culture conditions applied. One should be cautious when the transgene can have an effect on untransduced bystander cells in the mixed culture. We used lentiviral vectors to stably transduce the shRNA cassette, but other vector systems, integrating (retroviral) or non- integrating (e.g., adenovirus or adeno-associated virus), are compatible with the transient CCG test. Therefore, we propose 102 103 104 105 that the CCG assay is a welcome addition to the existing array of GFP signal proliferation assays, especially when one is interested in scoring subtle effects on cell growth due to stable transgene expression. Figure 4. Decreased GFP expression is due to outgrowth of untransduced cells. Clonal GFP þ cells were generated from the mixed cell culture expressing shRNA3. After the generation of clonal MATERIALS AND METHODS cells via limiting dilution, the percentage of GFP-expressing cells DNA constructs was monitored over a time period of 13 days; light gray line (day 1) and black line (day 13). (a) The percentage of GFP-expressing cells Lentiviral vector pLKO.1 constructs expressing shRNA were from the decreases in the original mixture of transduced and untransduced MISSION TRC-Hs 1.0 library,19 including the negative control constructs culture. (b, c) Two GFP þ clonal cell lines demonstrate a constant SHC001 and SHC002 (renamed as SHC1 and SHC2), were obtained as high level of GFP transgene expression. bacterial clones from Sigma-Aldrich (St Louis, MO, USA). Plasmid DNA was extracted using the Nucleobond Midiprep columns according to the manufacturer’s instructions (Macherey-Nagel, Du¨ren, Germany). assay is internally controlled as it is performed with a mixture of Sequences for shRNA1, 2 and 3 are, respectively, as follows: untransduced and transduced cells. The CCG assay is user-friendly, 50-CCGGCATCAAACCATTCCTTCTGTACTCGAGTACAGAAGGAATGGTT as no precise ratio of transduced over untransduced cells is TGATGTTTTTG-30, required. There is an obvious lower limit of the assay, below B5% 0 þ þ 5 -CCGGCCCTGCCAAACAAGCTAATATCTCGAGATATTAGCTTGTTTGG GFP cells, to allow a reliable measurement of an altered GFP / 0 À CAGGGTTTTT-3 GFP ratio (data not shown). Too high transduction efficiencies 50-CCGGGAAACTGTCAAGCAAGGCGTTCTCGAGAACGCCTTGCTTGAC may be dangerous in the sense that multiple vector integrations AGTTTCTTTTT-30. may occur, which will result in increased shRNA expression levels. Indeed, such increased toxicity was scored for high-level trans- The pLKO.1 constructs from the MISSION TRC-Hs 1.0 library contain a duction of the shRNAs and even the SHC2 scrambled shRNA con- puromycin selection marker and this was replaced by the GFP gene that trol exerted some toxic effects when overexpressed. We scored was amplified from the lentiviral vector JS1 (ref. 20) by PCR (forward such effects when over 95% of the cells were transduced. primer: 50-GAATTCACCGGTCGCCACCAT-30, reverse primer: 50-ACTAGTGTC A major advantage of the CCG assay is its extreme sensitivity. GACCCCGGGCTC-30). The GFP--PCR product was cloned into pCR2.1-TOPO We measured a significant decrease in cellular proliferation rates TA cloning vector using the TOPO TA-cloning kit (Invitrogen, Carlsbad, CA, when toxic shRNAs are expressed, compared with the control cell USA), according to the supplier’s protocol. The BamHI--KpnI fragment from lines SHC1 and SHC2. A 10% reduction in proliferation rate can be this vector, encompassing the complete GFP gene, was subsequently

Gene Therapy (2012) 1058 --1064 & 2012 Macmillan Publishers Limited Competitive cell growth assay JJM Eekels et al 1063

Figure 5. Comparison of the CCG assay with other cell proliferation assay. (a) Sorted GFP þ cells were mixed with SupT1 and the percentage GFP þ cells was measured daily over a period of 5 days. Although in SHC1 and SHC2 mixed culture the percentage is constant, in the shRNA1, shRNA2 and shRNA3 mixtures the GFP percentage decreases (upper panel). Sorted GFP þ cells were used in four proliferation assays; manual counting, automated counting, MTT assay and an ATP-bioluminescence assay. For every cell line, growth curves were obtained for each specific assay. One representative experiment for each assay is shown. (b) For the CCG assay the percentage decrease in proliferation rate compared with SupT1 is shown (upper panel). Statistical analysis shows significant slower proliferation rate in the case of shRNA1, shRNA2 and shRNA3 when compared with the SHC1 mixed culture. With the growth curves of two independent experimentstherelativedoublingtimesforeachcelllinewascalculated. Doubling time of SupT1 was set at 1. When compared with SupT1 in the manual counting assay, shRNA3 has a significant increased doubling time. Results are from two independent experiments, performed in triplicates (manual and automated counting) or in sixfolds (MTT and ATP bioluminescence assays). *Po0.1, **Po0.01 and ***Po0.001. cloned into pLKO.1, cut with the same enzymes to replace the puromycin- were seeded in a 24-wells plate (1 Â 105 cells per well). Lentiviral vector resistant marker. GFP expression is under the control of the pGK promoter (0.1, 1, 10 or 100 ml) was added and incubated overnight. Excess virus was that was already present in pLKO.1 construct The resulting plasmid was washed away on the second day. named pLKO.1--GFP. All constructs were verified by restriction enzyme analysis and GFP expression. Generation of clonal transduced cell lines Cell culture For the generation of clonal cell lines, the lentiviral-transduced cells were serially diluted (10-fold) in conditioned medium (filter-sterilized super- Human embryonic kidney 293T (HEK293T) adherent cells were grown in natant from a 3-day culture of SupT1 cells, culture medium RPMI þ 10% Dulbecco’s modified Eagle’s medium (Invitrogen) supplemented with 10% fetal calf serum). For every dilution, a 96-wells plate was seeded and fetal calf serum, 100 U mlÀ1 penicillin and 100 mgmlÀ1 streptomycin. Human incubated for 2 weeks in a humidified chamber at 37 1C and 5% CO . When SupT1 T cells were grown in suspension in RPMI medium (Invitrogen) 2 a plate showed cell growth in equal to or less than 30 of the 96 wells, the supplemented with 10% fetal calf serum, 100 U mlÀ1 penicillin and 100 mgmlÀ1 positive cultures were considered to represent clonal cell lines, based on streptomycin. Cell lines were cultured in a humidified chamber at 37 1C Poisson distribution. and 5% CO2.

Lentiviral vector production and transduction Lentiviral vectors were produced as previously described.16 In short, Cell samples for analysis of GFP expression were spun down at 1500 g for HEK293T cells were co-transfected with shRNA-expressing construct and 4 min and the cell pellet was gently resuspended in 200 ml FACS solution the packaging plasmids pRSV-Rev, pMDLg/pRRE and pVSV-G using (phosphate-buffered saline þ 2% FCS). Flow cytometry analyses were Lipofectamine 2000 (Invitrogen). One day after transfection the medium performed on a FACS Canto cytofluorometer (BD Biosciences, Franklin was refreshed and the following day the supernatant was collected. The Lakes, NJ, USA). Live cells were discriminated from cell debris and dead virus-containing supernatant was centrifuged, filtered (0.45 mm) and aliquots cells based on physical parameters (forward- and side-light scatter). were stored at À80 1C. A sample was taken for CA-p24 enzyme-linked Fluorescence background levels were set with untransduced and immunosorbent assay to monitor lentiviral particle production. SupT1 cells unstained cells.

& 2012 Macmillan Publishers Limited Gene Therapy (2012) 1058 --1064 Competitive cell growth assay JJM Eekels et al 1064 Cell proliferation assays ACKNOWLEDGEMENTS For the different proliferation assays, cells were FACS-sorted based on GFP This research was supported by the Dutch AIDS fund (grant 2006006 and 2007028). expression. The sorted cell lines and SupT1 cells were seeded in 96-wells We thank Berend Hooibrink for FACS sorting, Stephan Heynen for CA-p24 enzyme- plates at a density of 1  105 cells per ml in 100 ml per well. The pro- linked immunosorbent assay experiments and Rene´e van der Sluis and Dave Speijer liferation of these cell lines was tested in four proliferation assays, which for useful discussions. We also thank the Belgian Federal Government for financial were performed in triplicates for automated and manual counting and in support through the Inter-University Attraction Pole grant P6/41. sixfold for the MTT and luciferase assays. In parallel the CCG assay was performed by mixing the sorted cultures with untransduced SupT1 cells REFERENCES and by following the GFP percentage over a period of 5 days. The CellTiter-Glo Luminescent Cell Viability Assay (Promega, Madison, 1 Fire A, Xu S, Montgomery MK, Kostas SA, Driver SE, Mello CC. Potent and specific genetic interference by double-stranded RNA in Caenorhabditis elegans. Nature WI, USA) was used as follows. A 100-ml sample was taken daily for each cell 1998; 391: 806 --811. 1 line in sixfolds over a period of 5 days and stored at À20 C until 2 Liu YP, Haasnoot J, ter BO, Berkhout B, Konstantinova P. Inhibition of HIV-1 by measurement following manufacturer’s instructions. In brief, 20 ml sample multiple siRNAs expressed from a single microRNA polycistron. Nucleic Acids Res was mixed 1:1 with CellTiter-Glo reagent in a white half-well 96-wells plate 2008; 36: 2811 --2824. for 10 min. The luciferase signal, which is directly proportional to the 3 Zhiqiang W, Yaowu Y, Fan Y, Jian Y, Yongfeng H, Lina Z et al. Effective siRNAs amount of metabolically active cells in the sample, was measured on a inhibit the replication of novel influenza A (H1N1) virus. Antiviral Res 2010; Glomax luminometer (Promega). 85: 559 --561. For automated cell counting, daily samples were taken. Cells clusters in 4 Pan Q, Henry SD, Metselaar HJ, Scholte B, Kwekkeboom J, Tilanus HW et al. the culture were disrupted by gently pipetting up and down and 100 mlof Combined antiviral activity of interferon-alpha and RNA interference directed against hepatitis C without affecting vector delivery and gene silencing. J Mol Med cell suspension was transferred to a FACS tube containing 200 ml FACS 2009; 87: 713 --722. buffer. Each sample was counted for 20 s with identical FACS settings. The 5 Eckstein A, Grossl T, Geisler A, Wang X, Pinkert S, Pozzuto T et al. Inhibition of increase in number of live cells over time in the sample is a direct measure adenovirus infections by siRNA-mediated silencing of early and late adenoviral of cell proliferation. For manual counting, a 10 ml sample was taken daily gene functions. Antiviral Res 2010; 88:86--94. and mixed 1:1 with trypan blue. A total volume of 10 ml was loaded into a 6 Liu YP, von Eije KJ, Schopman NC, Westerink JT, ter BO, Haasnoot J et al. hemacytometer (Hycor, Garden Grove, CA, USA) and cells were counted in Combinatorial RNAi against HIV-1 using extended short hairpin RNAs. Mol Ther nine squares (0.1 mm3) to determine the number of cells per ml. 2009; 17: 1712 --1723. The MTT assay was performed as described earlier.17 Each day for 5 days 7 Eekels JJ, Geerts D, Jeeninga RE, Berkhout B. Long-term inhibition of HIV-1 30 ml MTT was added to the wells of a single plate for 4 h at 37 1C. The wells replication with RNA interference against cellular co-factors. Antiviral Res 2011; 89:43--53. on the edge of the plate were excluded to prevent variation due to 8 Jackson AL, Bartz SR, Schelter J, Kobayashi SV, Burchard J, Mao M et al. Expres- evaporation and wells filled only with medium served as a negative sion profiling reveals off-target gene regulation by RNAi. Nat Biotechnol 2003; control. After incubation 100 ml was carefully removed from each well and 21: 635 --637. 100 ml dissolving solution (0.1% Triton-X, 4 mM HCl in isopropanol) was 9 Grimm D, Streetz KL, Jopling CL, Storm TA, Pandey K, Davis CR et al. Fatality in added to dissolve the formazan crystals. Formazan absorbance was mea- mice due to oversaturation of cellular microRNA/short hairpin RNA pathways. sured with an ELx808 microplate reader (Biotek Instruments, Winooski, VT, Nature 2006; 441: 537 --541. USA) at 550 nm and background levels were scored at 650 nm. 10 Khan AA, Betel D, Miller ML, Sander C, Leslie CS, Marks DS. Transfection of small RNAs globally perturbs gene regulation by endogenous microRNAs. Nat Biotechnol 2009; 27: 549 --555. Calculations and statistical analysis 11 Judge AD, Sood V, Shaw JR, Fang D, McClintock K, MacLachlan I. Sequence- To calculate the defect in cell growth rate measured in the CCG assay, with dependent stimulation of the mammalian innate immune response by synthetic the assumption that the decrease in the ratio of GFP þ to GFPÀ cells over siRNA. Nat Biotechnol 2005; 23: 457 --462. time is exponential, the following formula was used: 12 Hornung V, Guenthner-Biller M, Bourquin C, Ablasser A, Schlee M, Uematsu S et al. Sequence-specific potent induction of IFN-alpha by short interfering RNA in Gd ¼ (Td(GFP þ )ÀTd(GFPÀ))/Td(GFPÀ)  100% ¼ÀTd(GFPÀ)  log2(y)/ plasmacytoid dendritic cells through TLR7. Nat Med 2005; 11: 263 --270. (x þ Td(GFPÀ)  log2(y))  100%, 13 Sioud M. Induction of inflammatory cytokines and interferon responses by double-stranded and single-stranded siRNAs is sequence-dependent and requires where Gd is the calculated growth defect, Td(GFP þ ) and Td(GFPÀ) are the endosomal localization. J Mol Biol 2005; 348: 1079 --1090. doubling times in days of the GFP þ and GFPÀ cells, respectively, x is the 14 He J, Yang Q, Chang LJ. Dynamic DNA methylation and histone modifications number of days over which the decrease in GFP percentage is measured, contribute to lentiviral transgene silencing in murine embryonic carcinoma cells. and y is calculated as %GFP þ /%GFPÀ at the time point x divided by the J Virol 2005; 79: 13497 --13508. %GFP þ /%GFPÀ at the time point 0. 15 Ellis J. Silencing and variegation of gammaretrovirus and lentivirus vectors. Hum CCG experiments were performed on at least two occasions, each in Gene Ther 2005; 16: 1241 --1246. 16 ter Brake O, Konstantinova P, Ceylan M, Berkhout B. Silencing of HIV-1 with RNA triplicates. Cell counting with FACS and a hemacytometer were also interference: a multiple shRNA approach. Mol Ther 2006; 14: 883 --892. performed in triplicate and in sixfolds for the MTT and ATP biolumines- 17 Mosmann T. Rapid colorimetric assay for cellular growth and survival: application cence assays. All data were corrected for variation between experiments to proliferation and cytotoxicity assays. J Immunol Methods 1983; 65: 55 --63. 21 using Factor Correction. For cell counting, MTT and ATP bioluminescence 18 Crouch SP, Kozlowski R, Slater KJ, Fletcher J. The use of ATP bioluminescence assays, doubling times were calculated based on exponential growth curve as a measure of cell proliferation and cytotoxicity. J Immunol Methods 1993; fitting of the cell growth curves and represented as mean doubling 160: 81 --88. time±s.d. To test whether doubling times were significantly different from 19 Root DE, Hacohen N, Hahn WC, Lander ES, Sabatini DM. Genome-scale loss- the control culture (SupT1), one-way analysis of variance was performed of-function screening with a lentiviral RNAi library. Nat Methods 2006; 3: with 95% confidence interval. For the CCG assay, a Student’s t-test was 715 --719. 20 Seppen J, Rijnberg M, Cooreman MP, Oude Elferink RP. Lentiviral vectors for performed on the percentage difference in doubling time, and compared efficient transduction of isolated primary quiescent hepatocytes. J Hepatol 2002; with SHC1, again with 95% confidence interval. 36: 459 --465. 21 Ruijter JM, Thygesen HH, Schoneveld OJ, Das AT, Berkhout B, Lamers WH. Factor correction as a tool to eliminate between-session variation in replicate CONFLICT OF INTEREST experiments: application to molecular biology and retrovirology. Retrovirology The authors declare no conflict of interest. 2006; 3:2.

Supplementary Information accompanies the paper on Gene Therapy website (http://www.nature.com/gt)

Gene Therapy (2012) 1058 --1064 & 2012 Macmillan Publishers Limited