NF1 Transcription Factors Act As Potent Genetic Insulators for Integrating Gene Transfer Vectors

ENABLING TECHNOLOGIES CTF/NF1 transcription factors act as potent genetic insulators for integrating gene transfer vectors

A Gaussin1,2, U Modlich3, C Bauche4, NJ Niederla¨nder1, A Schambach3, C Duros4, A Artus4, C Baum3, O Cohen-Haguenauer4,5 and N Mermod1

Gene transfer-based therapeutic approaches have greatly benefited from the ability of some viral vectors to efficiently integrate within the cell genome and ensure persistent transmission of newly acquired transgenes to the target cell progeny. However, integration of provirus has been associated with epigenetic repercussions that may influence the expression of both the transgene and cellular genes close to vector integration loci. The exploitation of genetic insulator elements may overcome both issues through their ability to act as barriers that limit transgene silencing and/or as enhancer-blockers preventing the activation of endogenous genes by the vector enhancer. We established quantitative plasmid-based assay systems to screen enhancer- blocker and barrier genetic elements. Short synthetic insulators that bind to nuclear factor-I protein family transcription factors were identified to exert both enhancer-blocker and barrier functions, and were compared to binding sites for the insulator protein CTCF (CCCTC-binding factor). Gamma-retroviral vectors enclosing these insulator elements were produced at titers similar to their non-insulated counterparts and proved to be less genotoxic in an in vitro immortalization assay, yielding lower activation of Evi1 oncogene expression and reduced clonal expansion of bone marrow cells. Gene Therapy (2012) 19, 15–24; doi:10.1038/gt.2011.70; published online 12 May 2011

Keywords: oncogene; insulators; CTF/NF1; CTCF; genome integration; genotoxicity

INTRODUCTION the ability of the enhancer to activate other promoters.14 Insulators Ex vivo retrovirus-mediated gene transfer into hematopoietic progeni- with boundary properties set the borders of neighboring chromatin tor cells proved to be an efficient therapeutic strategy for a substantial domains. Thus, boundaries prevent the propagation of condensed number of patients suffering from severe combined immunodefi- chromatin structures that silence expression and counteract the effects ciency.1 However, the appearance of leukemia cases raised questions of chromosomal position on transgene expression.15 Particular insu- on the safety of this approach.2 The ability of retroviral vector genomic lators are able to act both as enhancer-blocker and boundary, such as integration to activate the expression of endogenous oncogenes the well-characterized 1.2-kb chicken b-globin hypersensitive site-4 uncovered a new type of genotoxicity stemming from the nonspecific (cHS4)-containing insulator.11,16 activation of cellular genes by viral regulatory elements.3,4 In addition, Previous attempts to incorporate insulators into recombinant viral it is now well established that integrating viral vectors have a tropism vectors intended for gene therapy yielded reduced transgene expres- for integration into particular chromosomal regions, prompting their sion variegation and limited chromosomal position effect.17–21 Most insertion in or near active transcriptional units.5–7 Integrating vectors of these studies were performed on gamma-retroviral and lentiviral used for gene therapy usually carry strong enhancers, which are able vectors with the cHS4 insulator in various mouse and human cell to provide high and persistent transgene expression through their types.22 Implementation of the full-length, 1.2-kb element in gamma- capacity to withstand silencing. However, these regulatory elements retroviral vectors was associated with reduced genotoxicity, but also are also more prone to promote the dysregulation of genes near the with important constraints concerning vector design and titer of site of integration of the vector. Extensive studies have shown that the infectious particles, mainly owing to the substantial size of this enhancer sequences are the major cause of cell transformation, making sequence.18,23,24 A shorter sub-portion of the cHS4, namely the 250- the design of viral vectors for life-long therapy approaches even more bp insulator core element, is more suitable with viral vectors’ biology, challenging.8–10 but it failed to reproduce the activity of the full-length element.9,20,25,26 Insulators mark the boundaries of chromatin domains and limit the The enhancer-blocking function of the cHS4 has been attributed to range of action of enhancers and silencers.11 They are characterized by the CCCTC-binding factor (CTCF), an 11-zinc-finger DNA-binding at least one of the following properties: Enhancer blocking and/or protein highly conserved in vertebrates. CTCF was implicated in boundary.12,13 An insulator with enhancer-blocking properties is able diverse regulatory functions, including transcriptional activation/ to specifically block communication between an enhancer and a repression, insulation and imprinting.27 CTCF organizes higher- promoter when interposed. However, enhancer-blockers do not alter order chromatin structures and was associated with the formation

1Institute of Biotechnology, University of Lausanne, Lausanne, Switzerland; 2Doctoral program in Biotechnology and Bioengineering, Ecole Polytechnique Fe´ de´rale de Lausanne, Lausanne, Switzerland; 3Department of Experimental Hematology, Hanover Medical School, Hanover, Germany; 4Laboratory of Biotechnology and Applied Pharmacogenetics, UMR 8113, Ecole Normale Supe´rieure de Cachan, Cachan, France and 5Universite´ Paris-Diderot & Hopital Saint-Louis, Paris Cedex 10, France Correspondence: Professor N Mermod, Institute of Biotechnology, University of Lausanne, Station 6, EPFL-FSB-LBTM, Lausanne 1015, Switzerland. E-mail: [email protected] or Dr O Cohen-Haguenauer, Ecole Normale Supe´rieure de Cachan, 61 avenue du President Wilson, F-94235 Cachan Cedex, France. E-mail: [email protected] Received 23 August 2010; revised 15 February 2011; accepted 2 March 2011; published online 12 May 2011 CTF/NF1 as insulators for gene transfer A Gaussin et al 16

of chromatin loops that may mediate its enhancer-blocking func- lines. The presence of the 5¢HS2 LCR significantly increased the tion.28 The boundary activity of the cHS4 element derives from the occurrence of G418-resistant colonies in K562 and HeLa cells, whereas combined effect of the upstream transcription factors 1 and 2 (USF1 the cHS4 insulator was able to block the LCR-mediated upregulation and USF2).29 In addition, the cHS4 insulator sequence was shown to of the selection gene when interposed between the enhancer and the be highly concentrated in the nuclear matrix fraction, suggesting that promoter. The cHS4 insulator decreased the number of G418-resistant it may be involved in the topological organization of the genome.30 colonies nearly four-fold and 10-fold in K562 and HeLa, respectively, The CAAT box-binding transcription factor/nuclear factor-1 (NF1, and fully prevented LCR-mediated upregulation to levels comparable also called CTF/NF1) consists of a family of widely expressed tran- to those observed from the g-globin promoter without LCR and scription factors that possess a barrier function.31,32 This family insulator (Figure 1b). comprises NF1-A, NF1-B, NF1-C and NF1-X subtypes that share The ability of the cHS4 insulator to block activation from the the same DNA-binding domain.33,34 Of note, the NF1-C regulatory potent enhancer present on the Friend-murine leukemia virus LTR domain prevents the propagation of repressive chromatin structures was then similarly assessed in HeLa cells. Substitution of the b-globin that stem from the telomere, and thereby prevent transgene silencing LCR by the Fr-MuLV LTR in either orientation strongly increased the at mammalian and yeast cells telomeres.31,32 NF1-C regulates DNA occurrence of resistant colonies (Figure 1c). The Fr-MuLV LTR proved replication by promoting the recruitment of the DNA polymerase to to promote much stronger reporter gene activation than the b-globin the adenovirus and SV40 origins of replication, and is also involved in LCR in this cell type. Nevertheless, the cHS4 was able to decrease the promoter regulation.35–37 Thus, CTF/NF1 proteins were described as growth of resistant colonies nearly eight-fold when interposed between barrier elements that act to control DNA transcription and replication, the enhancer and the promoter of the reporter construct, yielding yet a potential enhancer-blocking function had not been assessed. levels similar to those obtained in the absence of an enhancer. Previous identifications of insulator elements for gene therapy viral Previous studies have indicated that inclusion of the HS4 element vectors have resulted in impaired vector production and/or poor and/or its core may repress the expression of nearby reporter genes, insulating efficacy. We developed a quantitative procedure to screen which may complicate the interpretation of insulator function insulator elements and assess their ability to block gene activation by assays.20,38 We therefore designed a two-reporter gene assay imple- the strong Friend-murine leukemia virus enhancer-containing long menting the accuracy of the former assay, but in which polar terminal repeat (Fr-MuLV LTR). We designed novel synthetic elements insulating activities can be distinguished from enhancer inhibition, acting as binding sites for CTCF and CTF/NF1 proteins, and found global gene silencing effects or gene copy number variations that they mediate potent enhancer-blocking activities, resulting in a (Figure 1a, construct-2). Changes in the expression of the reporter commensurate reduction of genotoxicity when implemented in viral and reference genes were assessed by cytofluorometry from the green gene therapy vectors. fluorescent protein (GFP) and blue fluorescent protein (BFP) fluorescence profiles of single cells within populations of transiently trans- RESULTS fected HeLa cells. Interposition of a copy of the cHS4 insulator between Positive assessment of cHS4 variants in quantitative enhancer- the cytomegalovirus (CMV) minimal promoter driving the GFP gene blocking insulator activity assay and the LTR enhancer, for example, at position 1 of Figure 1a, led to a Using an enhancer-blocking insulator assay based on a neomycin significant decrease in the expression of GFP but not BFP, as illustrated (G418) resistance-conferring plasmid (construct-1 of Figure 1a), we in Figure 1d. Single-cell imaging of the relative GFP and BFP levels assessed the capability of the full-length 1.2-kb cHS4 element to showed a homogeneous decrease of GFP expression relative to BFP insulate a g-globin promoter/neo reporter gene from activation by (Figure 1e). When interposed, cHS4 induced a significant decrease in the mouse 5¢HS2 locus control region (LCR) in several human cell GFP expression down to 20% of the enhancer-activated level

Figure 1 Insulator/enhancer-blocker assay of cHS4 variants in cellular systems. (a.1) Constructs allowing semi-quantitative determination of enhancer- blocking activity. The reporter gene (neo) is driven by the g-globin promoter under the control of either the b-globin LCR element or the Fr-MuLV enhancer- containing LTR (in both orientations). Expression of neo is assessed by the number of neomycin-resistant colonies obtained after stable transfections. The insulated neo gene is flanked by two copies of the 1.2-kb cHS4 insulator (interposed and external positions, referred to as 1 and 2, respectively), whereas its non-insulated counterpart is flanked by a single external cHS4 element. (a.2) Constructs generated for quantitative enhancer-blocker assay. A BFP reference gene is placed under the control of the promoter and the enhancer-containing Fr-MuLV LTR. A reporter GFP gene is expressed from the minimal CMV promoter and the LTR enhancer. The insulated GFP gene is flanked by two copies of the 1.2-kb cHS4, to prevent activation of the minimal CMV promoter by the LTR enhancer. GFP fluorescence is normalized to that of BFP, which serves as internal reference for potential variations in transfection efficiency and transgene expression in each analyzed cell. The cHS4 core, or a spacer, replaces the full-length cHS4 depending on experiments. (b, c) Percentage of neomycin-resistant colonies counted 3 weeks after transfection and G418 selection of HeLa (black bars) and K562 (dashed bars) cells. Presence of an enhancer (Enh) from either the b-globin LCR element (LCR) or the Fr-MuLV LTR in one orientation (LTR) or in the inverted orientation (LTRinv), and/or of an interposed cHS4 insulator, is as depicted in panel a.1. The number of resistant colonies obtained in the absence of the interposed copy of the cHS4 was set to 100%. (d) Cytofluorometric analysis of the activity of the cHS4 insulator using quantitative assay in transiently transfected HeLa cells. The panels show the data of the same two representative cell populations obtained 48 h after transfection of constructs containing (blue) or not containing (red) an interposed copy of the cHS4 (as depicted in panel a.2). GFP expression of BFP-positive cells, or BFP expression of the total cell population, is shown at the left- and right-hand sides, respectively. The black profiles correspond to the background fluorescence of non-transfectedcellsas control. (e) Relationship between GFP and BFP fluorescence at individual cell levels. Cells transiently transfected with a construct containing or not containing an interposed cHS4 insulator, as analyzed in panel d, are depicted by blue and red dots, respectively. (f) Quantitative analysis of the enhancer- blocking activity of the cHS4 insulator. GFP fluorescence values were determined 48 h after HeLa cell transfection and were normalized to BFP expression for each analyzed cell. Normalized fluorescence values are plotted for cell populations transfected with constructs containing the indicated insulator, whereas D ins refers to the construct with an external cHS4 but without an interposed insulator sequence. Fluorescence value ratios were normalized to the ratio obtained with a construct lacking both copies of the insulator. Elements interposed between the enhancer and the promoter driving GFP expression and their respective sizes are as indicated. Spacers refer to portions of coding sequences used as negative controls. P-value was determined by two-tailed t-test. cHS4, chicken b-globin hypersensitive site-4; Fr-MuLV, Friend-murine leukemia virus; GFP, green fluorescent protein; LCR, locus control region; LTR, long terminal repeat.

Gene Therapy CTF/NF1 as insulators for gene transfer AGaussinet al 17

(Figure 1f). Only a small proportion of this effect (approximately shorter in size (270 bp), this element showed at least half the insulator 20%) may be attributed to the increased distance between the activity of the 1.2-kb cHS4. A high insulating activity was also shown enhancer and the promoter driving the GFP, as interposition of a when the CTCF nat 6x was embedded within an inactivated LTR, thus 1.2-kb-long neutral fragment had little effect on the GFP/BFP ﬂuor- mimicking the context in which the insulator would be in a retroviral or escence ratio (Supplementary Figure 1). Insertion of a single copy of lentiviral vector (Figure 2b). Another CTCF-binding element, compris- the 250-bp cHS4 core did not result in a signiﬁcant insulator effect in ing six repeats of the consensus binding site,39 was also evaluated (CTCF this assay, in agreement with previous studies showing that one copy cons; Figure 2a and Supplementary Table 1). Linkers were added of the cHS4 core does not mediate a high insulating activity.9,20,25,26 between each binding site to make up for the size of a CTCF footprint. This consensus element showed comparable activity with native binding Optimized CTCF and CTF/NF1 binding sites show potent sites. Of relevance, doubling the number of consensus binding sites, to enhancer-blocking activities reach 12 consecutive elements, fully reproduced the insulation effect of We designed a composite element, based on Bell et al.,22 containing theentire1.2-kbcHS4element(Figure2b). multiple copies of CTCF binding sites from both the cHS4 insulator To characterize the enhancer-blocking activity of CTF/NF1, we and the BEAD-1 element from the human T-cell receptor a/d locus evaluated a panel of binding sites for CTF/NF1 proteins in the double (CTCF nat 6x; Figure 2a and Supplementary Table 1). While much reporter-assay system. The binding sites’ derivatives were generated to

LTR cHS4 cHS4 LCR γ NEO 1 (1) (2)

cHS4 cHS4 2 BFP LTR CMV GFP (1) (2) cHS4 core spacer

cHS4 cHS4 Enh (int) Enh (int) HeLa LTRinv - - - K562 LTRinv cHS4 LTR - - cHS4 LTR cHS4 - - LCR cHS4 - cHS4 LCR cHS4 LCR - LCR - 0 20406080100 0 50 100 150 200 250 Percent neomycin resistant colonies

100 100 104 cHS4-insulated 80 80 3 Non-insulated 10 60 60 102 40 40 101

% Maximum 20 20 BFP expression 0 0 100 0 1 2 3 4 0 1 2 3 4 10 10 10 10 10 10 10 10 10 10 100 101 102 103 104 GFP expression BFP expression GFP expression

cHS4 core - 250bp cHS4(int) - 1200bp spacer - 250bp p<0.001 spacer - 1200bp Δ ins 0 0.2 0.4 0.6 0.8 1 Mean (GFP/BFP per cell)

Gene Therapy CTF/NF1 as insulators for gene transfer A Gaussin et al 18

Figure 2 Quantitative analysis of the enhancer-blocking activity of synthetic CTCF and CTF/NF1 binding sites. (a) Sequence and pairwise alignment of the different types of CTCF and CTF/NF1 binding sites were synthesized and assessed. The conserved nucleotides between the sequences are highlighted in red and a star indicates their position. BEAD-A and FII indicates CTCF binding sites in the chicken cHS4 insulator and in the human T-cell receptor locus, respectively, whereas adeno indicates the adenovirus type-II origin CTF/NF1-binding sequence. (b, c) Quantitative analysis of the enhancer-blocking activity of multimerized CTCF and CTF/NF1 binding sites. Transfection of HeLa cells, determination of GFP-to-BFP fluorescence ratio and normalization to the values obtained without any insulator are explained as described in Figure 1F. The number and type of adenovirus CTF/NF1 binding sites (adeno), or consensus binding sites (cons), and the spacing between adjacent binding sites are as specified. (d) Quantitative analysis of the enhancer-blocking activity of CTCF and CTF/NF1 binding sites in stable transfections. The mean GFP expression normalized to BFP expression per cell is plotted for each population 2–3 weeks after selection of transfected cells with the constructs depicted in Figure 1a.2. Fluorescence ratios were normalized to the values obtained with construct lacking both insulator copies. The elements interposed between the enhancer and the promoter of GFP are as indicated. P-values of two-tailed t-tests are shown. cHS4, chicken b-globin hypersensitive site-4; CTCF, CCCTC-binding factor; CTF/NF1, CAAT box-binding transcription factor/nuclear factor-1; GFP, green fluorescent protein.

alter the nature of the last nucleotide of the binding site, to be either a most potent enhancer-blockers, even embedded within an LTR. thymidine, mimicking the native CTF/NF1 binding site from the Decreasing the number of repeats did not lead to a significant loss adenovirus type-II origin of replication (referred to as adeno), or an of insulating activity. A single 20-bp binding site still mediated half adenine, to fit the consensus CTF/NF1 binding site (referred to as of the insulating effect of the full-length cHS4, like 7, 4 or 3 binding cons). The length of the spacing between two adjacent binding sites sites (Figure 2c, and data not shown). Even though the 10-bp spacing was also altered, with either 5 or 10 base pairs, orienting sites on should provide sufficient length to accommodate all directly contacted similar or opposite sides of the DNA double helix (Figure 2a and nucleotides within the binding sites,40 the spacing of 5 bp provided Supplementary Table 2). Spacers of 5 bp appeared to provide the most the best insulating activity for all of the tested CTF/NF1-binding favorable configuration, as binding sites interspaced with 5 bp are the sequences.

Gene Therapy CTF/NF1 as insulators for gene transfer AGaussinet al 19

CTF/NF1 binding sites act as effective intra-chromosomal boundary elements CTF/NF1 binding sites have been shown to function as barrier elements that can prevent the propagation of repressive histone modifications that stem from the telomere, and thereby prevent the silencing of telomeric genes.16,32 Nevertheless, whether these binding sites may also function as barrier elements upon transgene integration at internal chromosomal loci has not been assessed. The CTF/NF1 adeno (5-bp spacing) or CTCF nat binding sites were sub-cloned on each side of an SV40 promoter/GFP gene cassette (Figure 4a), to address the potential barrier properties of these sequences at random chromosomal locations. A multiple cloning site (MCS) spacer element replaced insulators as negative control, whereas a matrix attachment region (MAR) element that potently abrogates silencing effects was used as positive control.41 GFP fluorescence profiles were assessed for each construct on stable polyclonal populations pooling hundreds of Figure 3 CTF/NF1 proteins mediate the enhancer-blocking activity of their independent cell clones and thus distinct integration loci.41 Three GFP cognate DNA binding sites. (a) The enhancer-blocking properties of the CTF/ expression profiles delineated sub-populations termed M1, M2 and NF1 binding sites were quantitated as described in the legend of Figure 1f M3, which designate, respectively, low, medium, and high GFP in comparison with that of a 250-bp DNA control spacer. HeLa cells were expression ranges (Figure 4b). transfected with siRNA targeting CTF/NF1 (controls: mock transfection or scrambled siRNA) and subsequently transfected with the insulator- In the presence of the MAR element, most of the cells expressed containing constructs harboring either a neutral spacer of 250 bp or seven GFP and their distribution within the population of GFP-positive cells binding sites for CTF/NF1 (adeno, 5-bp spacing). FACS analyses were was 65% in M2 and 35% in M3 (Figure 4c). This distribution may be performed 48 h after transfection. The average GFP-to-BFP fluorescence explained by a barrier activity of MARs that shields the transgene from ratio was determined and plotted as described in the legend of Figure 1f. silencing at the site of integration in the host cell chromosomes. The Fluorescence ratios were normalized to those obtained from the mock construct containing a neutral MCS sequence showed only about 2% transfection of the siRNA and the transfection of the DNA construct of high-GFP-expressing cells (M3). In comparison, the GFP expression containing the 250-bp spacer. The P-value of two-tailed t-test is indicated. (b) Western blot analysis of cell extracts from populations analyzed in panel for the CTF/NF1 construct was improved relative to the MCS control, a. The immunoblot was performed using antibodies specific for CTF/NFI and with only one-third of the cells in M1, 60% in M2 and nearly 10% in GAPDH as loading control. CTF/NF1, CAAT box-binding transcription factor/ M3. The M3 cell populations for the MAR and CTF/NFI were more nuclear factor-1; FACS, fluorescence-activated cell sorting; Fr-MuLV, Friend- prominent than the MCS control (Figure 4b). A putative enhancer murine leukemia virus; GFP, green fluorescent protein; siRNA, short function of the tested CTF/NFI-binding elements was excluded by interfering RNA. inserting them upstream from the minimal CMV promoter, in the absence of an enhancer, which did not alter positively or negatively the expression of the reporter gene as compared with the unfused minimal promoter alone (data not shown). These data strongly To ascertain that CTCF as well as CTF/NF1 binding sites may also suggest that CTF/NF1 binding sites possesses barrier properties at show an enhancer-blocking activity in the context of a native internal chromosomal positions. Populations generated from MCS, chromatin structure, stable cell transfections were analyzed. The CTF/NF1 and MAR constructs contained around 50, 60 and 65% of insulating window of the full-length cHS4 was reduced 2.5-fold of M2 cells, respectively. However, the CTCF construct showed only 10% the reporter gene expression, whereas the cHS4 core did not show any of cells in the M2 sub-population, whereas the majority of the cells significant activity (Figure 2d). The elements composed of seven CTF were either expressing at low levels or did not show detectable GFP sites (adeno) and six CTCF native sites showed insulating effects that fluorescence. Thus, flanking the transgene with CTCF binding sites were comparable to those observed in transient transfections. The was deleterious for long-term gene expression, suggesting that CTCF binding sites for CTCF (nat) and CTF/NF1 (adeno, 5 bp) were can exert a silencing activity, in agreement with previous observations combined and assessed using the double-reporter assay system. This of a repressive function mediated by the cHS4 CTCF binding site.20 composite element conserved significant insulator activity, but no The potential delay of transgene silencing over time was evaluated synergistic effect could be observed as compared with the insulating through a 30-day time-course analysis of GFP expression in stably activity of each element tested separately (data not shown). transfected polyclonal pools (Figure 4d). Each global expression In order to validate CTF/NF1 as responsible for the CTF/NF1 pattern was conserved over time, although a slight shift toward binding site-mediated enhancer-blocking activity, cells were co-trans- lower fluorescence was generally observed after 16 days. Overall, this fected with a short interfering RNA (siRNA) targeting all CTF/NF1 indicated that MAR- and CTF/NF1-mediated silencing protection isoforms. The insulator assay was performed with constructs contain- effects are stable and can withstand cell division. ing either a neutral spacer of 250 bp or the most active combination of the CTF/NF1 binding sites. The enhancer-blocking activity of CTF/ Gain in safety of retroviral vectors flanked with CTCF and CTF/NFI NF1 was observed with mock-transfected cells or with cells transfected binding sites with a scrambled siRNA sequence (Figure 3a). However, the insulator We assessed whether CTCF and CTF/NF1 may shield off the retroviral activity was entirely lost upon an 80% knock-down of CTF/NF1 vector enhancer from activating the expression of cellular genes and/or protein levels with the specific siRNA, demonstrating the role of the mediating clonal cell proliferation. To this end, we used a gamma- CTF/NF1 transcription factor as enhancer-blocker insulators in mam- retroviral self-inactivating (SIN) vector (SRS.SF), which contains the malian cells (Figures 3a and b). spleen focus-forming virus enhancer/promoter as internal promoter.42

Gene Therapy CTF/NF1 as insulators for gene transfer A Gaussin et al 20

INS SV40 GFP INS M1 M2 M3 100%

M1 M2 100 80%

80 M3 60%

60 • MCS • CTF/NFI 40% 40 • CTCF % of max • MAR 20% 20 Cell percentage in sub-populations 0 0% 100 101 102 103 104 MCS CTF/NFI CTCF MAR GFP expression

MCS CTFNFI MAR CTCF 100 • 16 d. 80 • 20 d. • 27 d. 60 • 30 d. 40 % of max 20

0 100 101 102 103 104 100 101 102 103 104 100 101 102 103 104 100 101 102 103 104

GFP expression

Figure 4 CTF/NF1 binding sites dampen adverse chromosomal position effects. (a) A schematic representation of the insulated GFP transgene construct. GFP expression was driven by an SV40 promoter and the effect of elements inserted on both sides of the transgene was evaluated in stable transfections of HeLa cells. (b) Results of representative FACS analysis for GFP expression of HeLa cell populations stably transfected with constructs described in panel a (16 days after transfection). The GFP transgene was flanked by either a MCS, or seven binding sites for CTF/NF1 (adeno, 5-bp spacing), or six binding sites for CTCF (nat) or the 1–68 MAR element. The profile of non-transfected cells is depicted in gray. The population of GFP-positive cells, that is, the total cell population excluding non-expressing cells, was divided in three sub-populations as following: M1 designates cells expressing low levels GFP, whereas M2 and M3 designate cells with medium or high ranges of GFP levels, respectively. (c) Relative distribution of each sub-population of cells according to GFP expression levels. The M1, M2 and M3 sub-populations are defined as described in panel b. The results are expressed as percentage of cells in the designated sub-population relative to the population of GFP-positive cells (excluding non-expressing cells). (d) Time-course FACS analysis of GFP transgene expression when flanked with the designated insulators in stably transfected HeLa cells. Results of FACS analysis were acquired 16, 20, 27 and 30 days after transfection, under constant antibiotic selection. CTF/NF1, CAAT box-binding transcription factor/nuclear factor-1; FACS, fluorescence-activated cell sorting; GFP, green fluorescent protein; MAR, matrix attachment region; MCS, multiple cloning site.

This vector was previously shown to trigger insertional transformation mutant clones typically contain a vector insertion within the first events in both in vitro immortalization assays (IVIM assay) and intron of the Evi1 gene that results in the insertional upregulation of transplanted mice.7,9 Insulators were inserted to replace the deleted Evi1 transcription. The IVIM assay measures the replating frequency enhancer-promoter in the U3 region of the 3¢LTR, ensuring their of mutant clones within the transduced culture (‘clonal fitness’), as presence in both LTRs after reverse transcription and integration well as the incidence of mutation events. The SRS.SF vector was shown (Figure 5a). Inclusion of CTCF and CTF/NF1 binding sites had little to be transforming in every culture tested (incidence of 2Â10À5), with effect on the titers obtained from this gamma-retroviral vector, which a replating frequency/copy number of B0.0035 (mean of n¼10). Both remained above 107 transducing units per milliliter. In both cases, CTCF and CTF/NF1 insulators reduced the number of replating o20% titer reduction was noted when compared with the non- clones (Figure 5b) and were able to reduce the replating frequency/ insulated construct. While we previously observed a weak effect of copy number 5 and 4-fold, respectively, when compared with the non- the 250-bp cHS4 core insulator on transgene expression in SC-1 insulated vector (SRS.SF CTCF versus SRS.SF P¼0.055; SRS.SF CTF/ fibroblasts transduced at a low multiplicity of infection,9 the present NF1 versus SRS.SF P¼0.043; n¼7 each; Wilcoxon two-sample test; insulated vectors decreased GFP transgene expression 50% compared Figure 5c). The lower replating frequency was paralleled by the lower with the control vector. The CTCF and the CTF/NF1 insulators had a Evi1 expression levels observed in the presence of the insulated vectors similar attenuating effect on the expression of the internal promoter- (Figure 5d). driven eGFP. The composite insulators and LTRs were reverse- We assessed whether the reduced replating frequency and Evi1 transcribed and integrated in an intact form in the target cells expression reactivation observed in the presence of the insulators (Supplementary Figure 2). might have stemmed from an altered integration frequency around the Insulator activity was assessed using IVIM assay based on the Evi1 gene locus rather than from active insulation. We performed the in vitro selection of insertional mutant clones that gain a proliferative assessment by selecting clones that arose in the IVIM assays and by advantage after stable transduction by retroviral vectors. Immortalized mapping vector integration sites. Among the 35 insulated retroviral

Gene Therapy CTF/NF1 as insulators for gene transfer AGaussinet al 21

DISCUSSION Designing new generations of gene transfer viral vectors is a promi- sing avenue to achieve safer gene therapy. Implementation of genetic insulator elements in retroviral vectors is intended to allow the transgene cassette to behave as an autonomously regulated expression unit once integrated in the host cell genome. When flanking the transgene cassette, insulators may be beneficial in two ways: (i) Enhancer-blockers would limit the range of action of the viral vector enhancer on nearby cellular genes, thus decreasing the risk of insertional activation, and (ii) barrier elements would stop the spreading of silent chromatin, to ensure long-term transgene expression, and counteract detrimental integration site position effects.29 This study outlines a standardized screening procedure, which evaluates the enhancer-blocking activity of insulator elements. Unlike approaches based on mRNA levels or reporter protein secretion, this assay can be used to efficiently process large cell populations, provid- ing a quantitative measurement of the insulating activity at a single- cell resolution. Such quantitative assays parallel recently described assays of the barrier function of insulators specifically integrated at mammalian cell telomeres.43 Our screening evaluated a collection of novel insulating sequences comprising optimized binding sites for CTCF and CTF/NF1 insulator proteins. A 472-bp element comprising 12 CTCF binding sites reconstituted the enhancer-blocking activity of the full-length 1.2-kb cHS4. Moreover, a single copy of the cHS4 core showed little activity under experimental conditions assessing specifically its enhancer- Figure 5 CTF/NF1 and CTCF binding sites curtail retroviral vector geno- blocking function, consistent with recent results where the 250-bp toxicity. (a) Vector architecture of the gamma-retroviral SIN vector SRS.SF.eGFP.pre shown as provirus. It contains a splice-competent leader cHS4 core did not recapitulate the insulating function of the full- 26 region and a post-transcriptional regulatory element (PRE) of the woodchuck length element. hepatitis virus. The U3 region is almost completely deleted, leaving only the The binding sites for the CTF/NF1 transcription factor family integrase attachment sites intact. eGFP is driven by the enhancer/promoter showed, even from a single copy, significant enhancer-blocking elements derived from the potent spleen focus-forming virus SF enhancer/ activity. Binding sites interspaced by 5 bp showed the most potent promoter. In the insulated vectors, the insulator sequences were inserted enhancer-blocking activity. This may result from lower steric hin- into the U3 region of the vector’s LTRs. (b, c) Introduction of insulator drance effects between adjacent CTF/NF1 sites lying on opposite sides sequences into the LTRs of the SRS.SF.eGFP.pre vectors reduce the transformation potential. The replating frequencies of LinÀ cells corrected to of the DNA double helix. Here, insulation could be entirely attributed the mean retroviral vector copy number as measured in the DNA of mass to CTF/NF1 proteins upon knockdown assays, thus establishing cultures were plotted for insulated vectors and for the parental un-insulated a previously unknown enhancer-blocking activity for this family of SRS.SF gamma-retroviral vector. The insulators implemented in retroviral transcriptional regulators. vectors are six copies of CTCF binding sites (nat) or seven copies of CTF/ The compatibility of the insulator size with retroviral vectors was NF1 binding sites (adeno, 5-bp spacing). The median replating frequency/ B considered, as insertion of large elements in the 3¢LTR reduces vector copy number of the SRS.SF.eGFP.pre.CTCF vector was reduced 5-fold, 24 and that of the SRS.SF.eGFP.pre.CTF/NF1 B4-fold. The data points shown titers and impairs transduction efficiency. Therefore, insulator ele- for the SRS.SF.eGFP.pre vector contain those generated in this study ments of varying sizes were designed to fit the LTR of retroviral and/or (black dots) and in previously published data (gray dots; Modlich et al., lentiviral vectors, without affecting viral vector efficacy. Whereas 2009). The horizontal lines indicate the respective medians of the insulator potency correlated well with their variation in length, populations. (d) Quantitative real-time PCR analysis of Evi1 mRNA observed results were clearly distinct from sole distance effects, as expression levels in mass cultures of vector-transduced lineage-negative interposition of nonspecific spacer DNA fragments between the bone marrow cells on the day of replating. Evi1 expression was scored in four independent experiments (a–d) comparing CTCF and CTF/NFI binding enhancer and the promoter presented no variation. We conclude site insulated vectors to that of a non-insulated control (SRS.SF.eGFP.pre). that elements as short as 20 bp remain efficient for the mediation of Basal Evi1 expression in expanded and un-transduced cells (mock) was set significant enhancer-blocking function. to 1. CTCF, CCCTC-binding factor; CTF/NF1, CAAT box-binding transcription Derivatives of CTF/NF1- and CTCF-binding insulator sequences factor/nuclear factor-1; GFP, green fluorescent protein; LTR, long terminal yielded reduced genotoxicity when inserted in SIN gamma-retroviral repeat; SIN, self-inactivating. vectors, without significantly altering virus titers. These insulators led to a five-fold reduction of retroviral vector genotoxicity in an IVIM assay. Furthermore, the decreased occurrence of clonal cell prolifera- vector integrations sites that were mapped, five (14%) were in tion correlated well with the 5- to 10-fold lower expression of Evi1 proximity to the Mds1/Evi1 locus (Supplementary Table 3). These noted in presence of the CTF/NF1 insulator. This implies that the data indicated that the decrease in Evi1 expression reported in enhancer-blocking activity obtained with a plasmid-based assay was Figure 5d, thus, does not stem from a lack of integration of the preserved for integrating a retroviral vector. Here, our results intro- insulated vectors close to Evi1. Overall, we therefore conclude that the duce novel insulator elements suitable for implementation in viral CTCF and CTF/NFI elements acted as insulators that potently reduced vectors, endorsing previous reports on insulator elements able to the genotoxic outcome of the gamma-retroviral vector. reduce insertional genotoxicity.9,26,44,45

Gene Therapy CTF/NF1 as insulators for gene transfer A Gaussin et al 22

The barrier activity of the novel insulating elements was probed in enhancer, and the EBFP gene was PCR-amplified from the pEBFP-N1 plasmid the context of random transgene chromosomal integration upon (Clontech Laboratories, Inc., Mountain View, CA, USA). Both reporters were stable transfection. Flanking the transgene with CTCF binding sites sub-cloned in pBS2-SKP (Stratagene, La Jolla, CA, USA). The Fr-MuLV LTR led to a decrease in its expression, stably propagated over cell division. was inserted upstream from the EBFP gene such that transcription from the Prior studies on cHS4 insulators showed that deletion of the CTCF LTR is directed toward EBFP, and a copy of the 1.2-kb cHS4 was inserted downstream from the eGFP gene. Insulator sequences were inserted between binding sites was associated with loss of enhancer-blocking activity, 22,46 the Fr-MuLV LTR and the minimal CMV promoter driving eGFP expression. but did not alter the barrier function of the element. However, The 250-bp cHS4 core was PCR-amplified from the full-length cHS4 (GenBank earlier work on the natural cHS4 locus could not readily assess a accession number: U78775.2; amplification from position 1–250). A series of potential silencing effect of CTCF in addition to its enhancer-blocking neutral DNA spacers of various lengths were PCR-amplified from the mouse activity.27 Analysis of CTCF in several genomic contexts brought utrophin cDNA (GenBank accession number: BC062163.1; amplifications from forward its key function in the organization of the chromatin position 355 to positions 605 or 1555). The plasmid construct shown in architecture. Although current models of CTCF action rely on a Figure 4a was based on a plasmid previously described by Girod et al.41 looping mechanism, CTCF might also orchestrate genome architec- Binding sites for CTCF and CTF/NF1 were obtained by annealing com- ture through epigenetic chromatin modifications such as recruitment plementary oligonucleotides. Native CTCF binding sites refer to alternate 22 of chromatin-modifying proteins. It has been implicated in imprinting combinations of three binding sites from each of BEAD-A and FII sequences. Consensus CTCF binding sites correspond to direct repeats of the consensus as well as repression and activation of transcription, suggesting a binding motif.39 CTF/NF1 binding sites from the adenovirus type-II origin of versatile mode of action upon a biological context. As such, imple- replication were isolated from pNF7CAT.51 The consensus CTF/NF1 binding menting CTCF-binding synthetic sequences in viral vectors that site was obtained from SELEX-SAGE experiments.40 Spacer sequences separat- integrate at multiple and potentially random loci in the genome ing adjacent CTCF or CTF/NF1 binding sites were randomly chosen in order to could yield effects unanticipated from its activity at the cHS4 or at limit the occurrence of repetitive DNA sequences. The GenBank accession imprinted loci.47 Large-scale analyses of CTCF-insulated and non- number of the 1–68 MAR is EF694965. insulated vectors consequences will be required to address these issues. Transgenes flanked with binding sites for CTF/NF1 appeared to be Cell culture and transfection assays protected from silencing effects when integrated at random chromo- DNA transfection of K562 cells was performed as previously described,49 and somal loci. This observation is consistent with previous studies where resistant colonies were counted after 3 weeks of selection for G418 resistance. CTF/NF1 proteins act as barrier elements that block the propagation HeLa cells were transfected using FuGENE 6 (Roche Diagnostics, Basel, of silent chromatin structures, and thus protect transgenes from Switzerland) according to the manufacturer’s recommendations. Equimolar silencing effects mediated by repressive chromatin structures.31,32,43 amounts of the various plasmids were transfected in each experiment (using Hence, CTF/NF1 binding sites act both as enhancer-blockers and as the pBS2-SKP backbone plasmid as carrier). To obtain stable populations, the barrier insulator elements. Maintenance of a euchromatic status of the reporter constructs were co-transfected with a puromycin resistance-encoding plasmid (pPUR; Clontech) at a molar ratio of 10:1 and cells were grown in provirus should favor the production of retroviral vectors. This could Dulbecco’s modified Eagle’s medium containing 10% fetal bovine serum and also be advantageous when using tissue-specific promoters to drive 0.5 mgmlÀ1 puromycin (all from Gibco, Carlsbad, CA, USA). Fluorescence transgene expression, as these are often weaker than promoters of analyses were performed using the FACS Cyan (Dakocytomation, Copenhagen, viral origin, which may lead to the silencing of the therapeutic gene Denmark). Data analysis of the double-reporter assay consisted of normalizing over time. GFP fluorescence to BFP fluorescence for each cell and averaging these values The tropism of retroviral vectors for specific genomic regions such over the total cell population of several hundreds of independent clones. This as certain proto-oncogenes still remains a major issue for gene therapy method ensured to discriminate variations in GFP expression owing to the safety,6,7 despite reassuring reports of long-term follow-up in gene insulator effect from differences in expression levels resulting from variability in therapy-treated adenosine deaminase patients.48 Using insulators to transfection efficiency. For the boundary assay, cells whose fluorescence profile counteract position effects and the occurrence of poor expression of overlapped with the profile of non-transfected cells were considered as non- expressing cells. some integrated vectors may favor therapeutic outcome from lower 24 HeLa cells were transfected with 50 nmoles of an siRNA targeting the mRNA multiplicities of infection and reduced vector integration events, of all CTF/NF1 isoforms (sc-43561; Santa Cruz Biotechnology, Santa Cruz, CA, which should further reduce the risk of both activating and inactivat- USA) or with a non-targeting control (scrambled siRNA, sc-37007; Santa Cruz ing integration events. Biotechnology) using Oligofectamine (Invitrogen, Carlsbad, CA, USA) accord- Ensuring complete insulation of provirus sequences in large popu- ing to the manufacturer’s instructions. The cells were transfected with the lations of cells may be difficult to demonstrate, as insulator function double-reporter construct 24 h later and fluorescence-activated cell sorting depends potentially on the chromatin state at the site of integration (FACS) analyses were performed after 48 h. and/or on the expression of the insulator protein in the targeted cells or tissues. In that respect, the ubiquitous expression of CTF/NF1 Western blot analysis family members is a favorable feature, and their use should further Western blotting was performed following standard protocols: Protein extracts improve already safer SIN lentiviral vectors. from a defined number of cells were separated by denaturing PAGE and transferred to a nitrocellulose membrane (Schleicher und Schuell, Dassel, Germany), and incubated with the primary antibodies anti-NF1 (H-300; Santa MATERIALS AND METHODS Cruz Biotechnology; dilution 1:200) applied overnight or anti-glyceraldehyde-3- Plasmid vectors and insulator sequences phosphate dehydrogenase (GAPDH) (sc-32233; Santa Cruz Biotechnology) The plasmid constructs shown in Figure 1a.1 were constructed from the pJC5-4 applied2hafterblockingthemembranein 5% dried-milk in phosphate-buffered plasmid provided by Dr Gary Felsenfeld,49 where the 5¢HS2 LCR was saline. After incubation with a goat anti-rabbit horseradish peroxidase-coupled substituted by the Friend-murine leukemia virus LTR (Fr-MuLV, FB29 strain; secondary antibody (Sigma-Aldrich, St Louis, MO, USA) or a goat anti-mouse see reference Cohen-Haguenauer et al.50), either in its 5¢–3¢ native orientation horseradish peroxidase-coupled secondary antibody (Jackson Laboratory, Bar or in the inverted orientation. The plasmid construct shown in Figure 1a.2 was Harbor, ME, USA), the membrane was subjected to immunodetection by obtained as follows: The eGFP gene expressed from a minimal CMV promoter enhanced chemiluminescence (Amersham, Munich, Germany). Band intensity was PCR-amplified from a pcDNA3-EGFP plasmid excluding the CMV was quantified using the ImageJ software (http://rsb.info.nih.gov/ij/).

Gene Therapy CTF/NF1 as insulators for gene transfer AGaussinet al 23

Retroviral vectors 5 Wu X, Li Y, Crise B, Burgess SM. Transcription start regions in the human genome are The gamma-retroviral SIN vector has been described previously.52 The insu- favored targets for MLV integration. Science 2003; 300: 1749–1751. lator sequences were inserted into the NheIsiteofthe3¢ DU3 region, which is 6 Metais JY, Dunbar CE. The MDS1-EVI1 gene complex as a retrovirus integration site: impact on behavior of hematopoietic cells and implications for gene therapy. Mol Ther copied into the 5¢LTR after reverse transcription, and thus results in a design 2008; 16: 439–449. flanking the introduced expression cassette. Gamma-retroviral supernatant 7 Modlich U, Schambach A, Brugman MH, Wicke DC, Knoess S, Li Z et al. Leukemia production was performed using 293T cells as previously described, with co- induction after a single retroviral vector insertion in Evi1 or Prdm16. Leukemia 2008; expression of ecotropic envelope proteins.42,52 Cells were maintained in 22: 1519–1528. 8 Modlich U, Navarro S, Zychlinski D, Maetzig T, Knoess S, Brugman MH et al. Dulbecco’s modified Eagle’s medium supplemented with 10% fetal calf serum, Insertional transformation of hematopoietic cells by self-inactivating lentiviral and À1 100 U ml penicillin/streptomycin and 2 mM glutamine. Viral titers, deter- gammaretroviral vectors. Mol Ther 2009; 17: 1919–1928. mined on SC-1 cells by flow cytometry, were in the range of 5Â106 to 9 Zychlinski D, Schambach A, Modlich U, Maetzig T, Meyer J, Grassman E et al. 2Â107 IUmlÀ1 in un-concentrated supernatants. Physiological promoters reduce the genotoxic risk of integrating gene vectors. Mol Ther 2008; 16: 718–725. 10 Maruggi G, Porcellini S, Facchini G, Perna SK, Cattoglio C, Sartori D et al. Transcrip- tional enhancers induce insertional gene deregulation independently from the vector In vitro immortalization assay type and design. Mol Ther 2009; 17: 851–856. À Lineage-negative (Lin ) bone marrow cells of untreated C57BL6/J mice 11 West AG, Gaszner M, Felsenfeld G. Insulators: many functions, many mechanisms. (Charles River Laboratories, Wilmington, MA, USA) were obtained and Genes Dev 2002; 16: 271–288. transduced as previously described.53 DNA samples for real-time PCR analysis 12 Bell AC, West AG, Felsenfeld G. Insulators and boundaries: versatile regulatory (copy numbers) and flow cytometry (FACSCalibur; Becton-Dickinson, Franklin elements in the eukaryotic. Science 2001; 291: 447–450. 13 Gaussin A, Mermod N. Chromatin insulators and prospective application for gene Lakes, NJ, USA) were obtained 4 days after the last transduction. Quantitative therapy. Gene Ther Rev 2009; 1. PCR was performed using the Applied Biosystems 7300 Real-Time PCR 14 Kellum R, Schedl P. A group of scs elements function as domain boundaries in an System (Foster City, CA, USA), using the Quantitect SYBR Green kit enhancer-blocking assay. Mol Cell Biol 1992; 12: 2424–2431. (Qiagen, Hilden, Germany), as described previously.8 The IVIM assay 15 Sun FL, Elgin SC. Putting boundaries on silence. Cell 1999; 99:459–462. 16 Fourel G, Magdinier F, Gilson E. Insulator dynamics and the setting of chromatin 54 was performed as described previously. After retroviral transduction, bone domains. Bioessays 2004; 26: 523–532. marrow cells were expanded for 14 days, plated onto 96-well plates at 17 Emery DW, Yannaki E, Tubb J, Nishino T, Li Q, Stamatoyannopoulos G. Development of a concentration of 100 cells per well and positive wells were counted. virus vectors for gene therapy of beta chain hemoglobinopathies: flanking with a The frequency of replating cells was calculated based on Poisson statistics chromatin insulator reduces gamma-globin gene silencing in vivo. Blood 2002; 100: 2012–2019. and differences in replating frequency/copy number between cultures were 18 Yannaki E, Tubb J, Aker M, Stamatoyannopoulos G, Emery DW. Topological constraints evaluated by Wilcoxon two-sample test. governing the use of the chicken HS4 chromatin insulator in oncoretrovirus vectors. Mol Ther 2002; 5: 589–598. 19 Yao S, Osborne CS, Bharadwaj RR, Pasceri P, Sukonnik T, Pannell D et al. Retrovirus CONFLICT OF INTEREST silencer blocking by the cHS4 insulator is CTCF independent. Nucleic Acids Res 2003; The authors declare no conflict of interest. 31: 5317–5323. 20 Aker M, Tubb J, Groth AC, Bukovsky AA, Bell AC, Felsenfeld G et al. Extended core sequences from the cHS4 insulator are necessary for protecting retroviral vectors from silencing position effects. Hum Gene Ther 2007; 18: 333–343. ACKNOWLEDGEMENTS 21 Nishino T, Tubb J, Emery DW. Partial correction of murine beta-thalassemia with a This work was performed with support from EC-DG research (FP6-NoE), gammaretrovirus vector for human gamma-globin. Blood Cells Mol Dis 2006; 37:1–7. CLINIGENE (LSHB-CT-2006-018933 to NM, OCH and CB) and the 22 Bell AC, West AG, Felsenfeld G. The protein CTCF is required for the enhancer blocking University of Lausanne (NM). UM and CB also received support from the activity of vertebrate insulators. Cell 1999; 98: 387–396. 23 Li CL, Xiong D, Stamatoyannopoulos G, Emery DW. Genomic and functional assays Food and Drug Administration (project FDA-09-1056211-KP) and the German demonstrate reduced gammaretroviral vector genotoxicity associated with use of the Ministry of Research and Education (BMBF project iGene, 01GU0813). cHS4 chromatin insulator. Mol Ther 2009; 17:716–724. The kind gift of plasmids by Dr G Felsenfeld is acknowledged. We thank 24 Urbinati F, Arumugam P, Higashimoto T, Perumbeti A, Mitts K, Xia P et al. Mechanism Julien DeRoyer (LBAP, ENSC); Johanna Krause (Department of Experimental of reduction in titers from lentivirus vectors carrying large inserts in the 3¢LTR. Mol Ther ´ ´ 2009; 17: 1527–1536. Hematology, Hannover); Dr Jeremy Catinot and Pieric Doriot for 25 Modin C, Pedersen FS, Duch M. Lack of shielding of primer binding site silencer- technical help. mediated repression of an internal promoter in a retrovirus vector by the putative insulators scs, BEAD-1, and HS4. JVirol2000; 74: 11697–11707. AUTHOR CONTRIBUTION 26 Arumugam PI, Urbinati F, Velu CS, Higashimoto T, Grimes HL, Malik P. The 3¢ region of the chicken hypersensitive site-4 insulator has properties similar to its core and is AG designed and performed experiments on the identification and required for full insulator activity. PLoS One 2009; 4:e6995. assay of insulators, analyzed data and wrote the paper with NN under 27 Phillips JE, Corces VG. CTCF: master weaver of the genome. Cell 2009; 137: 1194–1211. the supervision of NM; C Bauche constructed insulators and insulated 28 Splinter E, Heath H, Kooren J, Palstra RJ, Klous P, Grosveld F et al. CTCF mediates long-range chromatin looping and local histone modification in the beta-globin locus. hybrid LTRs; CD and AA performed experiments on the implementa- Genes Dev 2006; 20: 2349–2354. tion of insulators in viral vectors, and analyzed data under the 29 Gaszner M, Felsenfeld G. Insulators: exploiting transcriptional and epigenetic mechanisms. Nat Rev Genet 2006; 7: 703–713. supervision of OCH who initiated the program with NM; and UM 30 Yusufzai TM, Felsenfeld G. The 5¢-HS4 chicken beta-globin insulator is a CTCF- and AS performed experiments to assay the cytotoxicity of insulated dependent nuclear matrix-associated element. Proc Natl Acad Sci USA 2004; 101: viral vectors, and analyzed data under the supervision of C Baum. 8620–8624. 31 Pankiewicz R, Karlen Y, Imhof MO, Mermod N. Reversal of the silencing of tetracycline- controlled genes requires the coordinate action of distinctly acting transcription factors. JGeneMed2005; 7: 117–132. 32 Ferrari S, Simmen KC, Dusserre Y, Muller K, Fourel G, Gilson E et al. Chromatin domain boundaries delimited by a histone-binding protein in yeast. JBiolChem2004; 279: 1 Cavazzana-Calvo M, Hacein-Bey S, de Saint Basile G, Gross F, Yvon E, Nusbaum P et al. 55520–55530. Gene therapy of human severe combined immunodeficiency (SCID)-X1 disease. 33 Rupp RA, Kruse U, Multhaup G, Gobel U, Beyreuther K, Sippel AE. Chicken NFI/ Science 2000; 288:669–672. TGGCA proteins are encoded by at least three independent genes: NFI-A, NFI-B 2 Hacein-Bey-Abina S, Garrigue A, Wang GP, Soulier J, Lim A, Morillon E et al. Insertional and NFI-C with homologues in mammalian genomes. Nucleic Acids Res 1990; 18: oncogenesis in 4 patients after retrovirus-mediated gene therapy of SCID-X1. JClin 2607–2616. Invest 2008; 118: 3132–3142. 34 Kruse U, Qian F, Sippel AE. Identification of a fourth nuclear factor I gene in chicken by 3 Baum C, von Kalle C, Staal FJ, Li Z, Fehse B, Schmidt M et al. Chance or necessity? cDNA cloning: NFI-X. Nucleic Acids Res 1991; 19:6641. Insertional mutagenesis in gene therapy and its consequences. Mol Ther 2004; 35 Cohen RB, Sheffery M, Kim CG. Partial purification of a nuclear protein that binds to 9:5–13. the CCAAT box of the mouse alpha 1-globin gene. MolCellBiol1986; 6: 821–832. 4 Thrasher AJ, Gaspar HB, Baum C, Modlich U, Schambach A, Candotti F et al. Gene 36 Alevizopoulos A, Mermod N. Antagonistic regulation of a proline-rich transcription therapy: X-SCID transgene leukaemogenicity. Nature 2006; 443: E5–E6; discussion factor by transforming growth factor beta and tumor necrosis factor alpha. JBiolChem E6–E7. 1996; 271: 29672–29681.

Gene Therapy CTF/NF1 as insulators for gene transfer A Gaussin et al 24

37 Muller K, Mermod N. The histone-interacting domain of nuclear factor I activates 47 D’Apolito D, Baiamonte E, Bagliesi M, Di Marzo R, Calzolari R, Ferro L et al. The sea simian virus 40 DNA replication in vivo. JBiolChem2000; 275: 1645–1650. urchin sns5 insulator protects retroviral vectors from chromosomal position effects by 38 Petrykowska HM, Vockley CM, Elnitski L. Detection and characterization of silencers maintaining active chromatin structure. Mol Ther 2009; 17: 1434–1441. and enhancer-blockers in the greater CFTR locus. Genome Res 2008; 18: 1238–1246. 48 Cassani B, Montini E, Maruggi G, Ambrosi A, Mirolo M, Selleri S et al. Integration 39 Kim TH, Abdullaev ZK, Smith AD, Ching KA, Loukinov DI, Green RD et al. Analysis of of retroviral vectors induces minor changes in the transcriptional activity of the vertebrate insulator protein CTCF-binding sites in the human genome. Cell 2007; T cells from ADA-SCID patients treated with gene therapy. Blood 2009; 114: 128: 1231–1245. 3546–3556. 40 Roulet E, Busso S, Camargo AA, Simpson AJ, Mermod N, Bucher P. High-throughput 49 Chung JH, Whiteley M, Felsenfeld G. A 5¢ element of the chicken beta-globin domain SELEX SAGE method for quantitative modeling of transcription-factor binding sites. Nat Biotechnol 2002; 20: 831–835. serves as an insulator in human erythroid cells and protects against position effect in 41 Girod PA, Nguyen DQ, Calabrese D, Puttini S, Grandjean M, Martinet D et al. Genome- Drosophila. Cell 1993; 74: 505–514. wide prediction of matrix attachment regions that increase gene expression in mam- 50 Cohen-Haguenauer O, Restrepo LM, Masset M, Bayer J, Dal Cortivo L, Marolleau JP malian cells. Nat Methods 2007; 4:747–753. et al. Efficient transduction of hemopoietic CD34+ progenitors of human origin using 42 Schambach A, Bohne J, Chandra S, Will E, Margison GP, Williams DA et al. Equal an original retroviral vector derived from Fr-MuLV-FB29: in vitro assessment. Hum Gene potency of gammaretroviral and lentiviral SIN vectors for expression of O6-methylgua- Ther 1998; 9: 207–216. nine-DNA methyltransferase in hematopoietic cells. Mol Ther 2006; 13: 391–400. 51 Tarapore P, Richmond C, Zheng G, Cohen SB, Kelder B, Kopchick J et al. DNA binding 43 Esnault G, Majocchi S, Martinet D, Besuchet-Schmutz N, Beckmann JS, Mermod N. and transcriptional activation by the Ski oncoprotein mediated by interaction with NFI. Transcription factor CTF1 acts as a chromatin domain boundary that shields human Nucleic Acids Res 1997; 25: 3895–3903. telomeric genes from silencing. Mol Cell Biol 2009; 29: 2409–2418. 52 Schambach A, Mueller D, Galla M, Verstegen MM, Wagemaker G, Loew R et al. 44 Evans-Galea MV, Wielgosz MM, Hanawa H, Srivastava DK, Nienhuis AW. Suppression of Overcoming promoter competition in packaging cells improves production of self- clonal dominance in cultured human lymphoid cells by addition of the cHS4 insulator inactivating retroviral vectors. Gene Therapy 2006; 13: 1524–1533. to a lentiviral vector. Mol Ther 2007; 15: 801–809. 53 Li Z, Schwieger M, Lange C, Kraunus J, Sun H, van den Akker E et al. Predictable and 45 Ramezani A, Hawley TS, Hawley RG. Combinatorial incorporation of enhancer-blocking components of the chicken beta-globin 5¢HS4 and human T-cell receptor alpha/delta efficient retroviral gene transfer into murine bone marrow repopulating cells using a BEAD-1 insulators in self-inactivating retroviral vectors reduces their genotoxic poten- defined vector dose. Exp Hematol 2003; 31: 1206–1214. tial. Stem Cells 2008; 26: 3257–3266. 54 Modlich U, Bohne J, Schmidt M, von Kalle C, Knoss S, Schambach A et al. Cell-culture 46 Chung JH, Bell AC, Felsenfeld G. Characterization of the chicken beta-globin insulator. assays reveal the importance of retroviral vector design for insertional genotoxicity. Proc Natl Acad Sci USA 1997; 94: 575–580. Blood 2006; 108: 2545–2553.

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

Gene Therapy