Self-inactivating retroviral vectors with improved RNA processing

RESEARCH ARTICLE

Self-inactivating retroviral vectors with improvedRNA processing

J Kraunus1,2,3, DHS Schaumann1,6, J Meyer3, U Modlich3, B Fehse2, G Brandenburg4, D von Laer4,

H Klump3, A Schambach3, J Bohne3 and C Baum3,5

1Department of Cell & Virus Genetics, Heinrich-Pette-Institute, Hamburg, Germany; 2Bone Marrow Transplantation, UniversityHospital Eppendorf, Hamburg, Germany; 3Department of Hematology, Hemostaseology and Oncology, Hannover Medical School,Germany; 4Infectiology, Georg-Speyer-Haus, Frankfurt, Germany; and 5Division of Experimental Hematology, Cincinnati Children’sHospital, Cincinnati, OH, USA

Three RNA features have been identified that elevateretroviral transgene expression: an intron in the 5 0 untrans-lated region (5 0UTR), the absence of aberrant translationalstart codons and the presence of the post-transcriptionalregulatory element (PRE) of the woodchuck hepatitis virus inthe 3 0UTR. To include such elements into self-inactivating(SIN) vectors with potentially improved safety, we excisedthe strong retroviral promoter from the U3 region of the 3 0

long terminal repeat (LTR) and inserted it either downstreamor upstream of the retroviral RNA packaging signal (C). Thelatter concept is new and allows the use of an intron in the5 0UTR, taking advantage of retroviral splice sites surround-ing C. Three LTR and four SIN vectors were compared toaddress the impact of RNA elements on titer, splice

regulation and transgene expression. Although titers of SINvectors were about 20-fold lower than those of their LTRcounterparts, inclusion of the PRE allowed production ofmore than 106 infectious units per ml without further vectoroptimizations. In comparison with state-of-the-art LTRvectors, the intron-containing SIN vectors showed greatlyimproved splicing. With regard to transgene expression, theintron-containing SIN vectors largely matched or evenexceeded the LTR counterparts in all cell types investigated(embryonic carcinoma cells, fibroblasts, primary T cells andhematopoietic progenitor cells).Gene Therapy advance online publication, 16 September 2004;doi:10.1038/sj.gt.3302309

Keywords: mouse leukemia virus; vector safety; hematopoiesis

Introduction

Retroviruses and related retrotransposable elements areunique tools for the stable delivery of transgenes intotarget cells, because they can be inserted at defined copynumbers and without rearrangement of neighboringDNA.1,2 Retroviral RNA is transcribed under the controlof enhancer-promoter sequences that are located in theU3 region of the long terminal repeats (LTRs). Thisconfiguration is required for serial infection. However,the terminal duplication of transcriptional control re-gions may increase the risk of insertional activationof neighboring cellular genes, either by long-distanceenhancer interactions or by providing new promoteractivity from the 30LTR.1,2 These concerns have beenreinforced by the first findings of leukemic complicationsof retroviral transgene delivery into repopulating hema-topoietic cells.1–5 Enhanced safety may be achieved withself-inactivating (SIN) vectors, in which the retroviralpromoter is deleted from the LTR. Instead, another viral

or cellular promoter is placed internally, downstream ofthe RNA packaging signal (C) and thus immediatelyupstream of the cDNA.6 However, despite the use of SINvectors for more than 15 years, the impact of this designon gene expression and vector safety has not beenstudied in great detail. Such an investigation can only beperformed if the internal promoter of the SIN vector isidentical with the one used in LTR vectors.

A drawback of many SIN vectors constructed on thebasis of murine leukemia viruses (MLV) or lentivirusessuch as the human immunodeficiency virus (HIV) isan unfavourable configuration of untranslated regions(UTR). Thus, RNA processing may be suboptimal,leading to a substantial loss of transgene expression.Optimal RNA processing is expected to increase thesafety of retroviral transgene technology, as it shouldreduce the danger of insertional mutagenesis by allowingthe use of weaker enhancer-promoter elements and byinducing optimal splicing and termination of thetranscript.

In the wild-type, LTR-controlled retroviral configura-tion, the 50UTR contains R and U5 of the 50LTR, theprimer binding site (PBS), the splice donor (SD) and C.The PBS initiates minus-strand synthesis during reversetranscription. C and neighbouring elements mediatedimerization and inclusion of the proviral RNA in vectorparticles. As we have shown previously, the SD and C,Received 12 November 2003; accepted 8 April 2004

Correspondence: Dr C Baum, Experimental Cell Therapy, Department ofHematology, Hemostaseology and Oncology, Hannover Medical School,Carl-Neuberg-Strabe 1, Hannover D-30625, Germany6Current address: Molecular Biology, Deutsches Rheumaforschungszen-trum, Berlin, Germany

Gene Therapy (2004), 1–11& 2004 Nature Publishing Group All rights reserved 0969-7128/04 $30.00

www.nature.com/gt

but not the viral coding sequences, are required forefficient packaging of MLV vectors.7 An artificial spliceacceptor (SA) can be introduced downstream of C andupstream of the cDNA of interest to obtain an alter-natively spliced intron in the 50UTR.7

MLVs and derived vectors generate both singlyspliced and unspliced RNA. The unspliced RNArepresents the genomic RNA incorporated into virions.In all replication-competent retroviruses, it also serves asthe mRNA of the gag-pol gene. The env gene is expressedfrom a spliced message, which represents up to 50% ofMLV transcripts. Vectors utilizing retroviral splice sitesupstream of the cDNA of interest typically maintain thisratio of balanced splicing.7–9 Introns in the 50UTR arefound in many cellular genes, improving transcriptionalelongation and nuclear export of RNA.10,11 In someretroviruses, constitutive RNA export elements havebeen identified within the viral coding sequences thatfacilitate the expression of unspliced RNA.12–14 ForMLVs, the sequences regulating alternative splicing andmediating nuclear export of unspliced RNA are un-known.12,14 Incomplete splicing may be due to a slightlydegenerated splice acceptor consensus upstream of env.Moreover, the first 28 nucleotides of the R region form astem loop (RSL) that supports the cytoplasmic accumu-lation of unspliced MLV RNA.15 To promote correcttranslation of the unspliced retroviral transcript, we havepreviously removed aberrant translational start codons,7

which are present in the majority of retroviral vectorsused so far in clinical applications.4,16,17

RNA export and translation of retroviral vectors mayalso be improved when incorporating the post-transcrip-tional regulatory element (PRE) of the woodchuckhepatitis virus into the 30UTR. This PRE promotesretroviral RNA expression in many but not all celltypes,9,18,19 in part by linking RNA to the CRM-1-dependent nuclear export pathway that is typically usedfor protein export but also exploited by the Rev-dependent export of HIV.20

In the present study, we developed a series of SIN andLTR vectors with identical promoters and found that theLTR configuration is superior for achieving high titersbut not required for strong transgene expression.Introducing a novel vector design, we demonstrate thatSIN vectors can be constructed with a favorable intronin the 50UTR, that these vectors show improved RNAsplicing compared to LTR vectors, and that the inclusionof the PRE increases the titer of these vectors to levelsthat allow the transduction of primary hematopoieticcells without further concentration procedures.

Results

Construction of retroviral vectorsAll retroviral vectors used in this study lacked viralcoding sequences and aberrant translational start codonsthat potentially interfere with the function of the insertedsequences. The modular vector assembly allowed preciseexchanges of cis-active elements. All constructs carriedidentical enhancer-promoter sequences from the U3region of spleen focus-forming virus (SFFVp), whichwe have shown to be a good choice for driving highlevels of transgene expression in hematopoietic cells.21

The activity of the SFFVp enhancer-promoter is excellent

in fibroblasts and hematopoietic cells, followed byT-lymphocytes, and is strongly reduced in primitiveembryonic cells.22 The 50UTR originates from the murineembryonic stem cell virus and contains a PBS that doesnot suppress LTR-mediated transcription in primitive,multipotent cells such as embryonic carcinoma (EC)cells, a feature that may also be relevant for somemultipotent stem cells of adult tissues.21,23 All vectorsused in the present study expressed the enhanced greenfluorescent protein (EGFP) from the position of theretroviral gag gene, thus mimicking the wild-typeconfiguration of the retroviral 50UTR.7 Three versions ofLTR vectors were compared,9 either lacking both intronand PRE (SF110), or containing an intron in the 50UTR(SF91), or containing an intron in the 50UTR plus the PREin the 30UTR (SF91P) (Figure 1a).

Corresponding SIN vectors with similar RNA mod-ules and an identical retroviral enhancer-promoter weregenerated by deleting the entire enhancer-promotercontained in the U3 region of the plasmid’s 30LTR,leaving just the first 22 bp (upstream of enhancer) andthe last 14 bp (downstream of TATA box) of the U3region. To be used for internal initiation of transcription,the same retroviral enhancer-promoter fragment plus thelast 14 bp of U3 and the adjacent RSL (designated as U3rfragment) was placed into internal positions of the SIN

Figure 1 Schematic comparison of the proviral vector configurations, aspresent after retroviral transduction of target cells. Full lines indicateuntranscribed genomic sequences, cylinders represent components of thetranscript produced in retrovirally transduced cells. The intron (SD-C-SA) is represented by a smaller cylinder because it is removed in a largeproportion of transcripts. The dotted line indicates deletions. (a) LTRvectors. (b) SIN vectors. The U3r module inserted downstream of the PBSlacks the first 22 nucleotides of U3, but contains the RSL for the expressionof unspliced RNA. SD, major splice donor; SA, splice acceptor; C,packaging and dimerization signal; PRE, post-transcriptional regulatoryelement of the woodchuck hepatitis virus; cDNA, EGFP coding region.

RNA processing of SIN vectorsJ Kraunus et al

2

Gene Therapy

backbone. We chose U3r and not a pure U3 fragmentlacking RSL because unpublished data confirmed thatthe RSL improved U3-mediated transgene expression,in line with previous observations.15 U3r was insertedimmediately 50 of the EGFP cDNA (vector SinSF), as inconventional SIN vectors. As a new strategy, weintroduced U3r into a SalI-site which we had generatedby site-directed mutagenesis 18 bp 30 of the PBS (+182with respect to the cap site) and 24 bp 50 of the SD. In thisconfiguration, the 50UTR contained C with or withoutflanking retroviral splice sites (vectors SinSF91 orSinSF110, respectively). The addition of the PRE intothe 30UTR of SinSF91 resulted in SinSF91P (Figure 1b).

Retroviral transduction of target cellsRetroviral vector plasmids were transiently transfectedinto Phoenix-gp packaging cells for the production ofreplication-defective vector stocks, using cotransfectedplasmids encoding either ecotropic, gibbon ape leukemiavirus (GALV) or vesicular stomatitis virus (VSV) envel-ope glycoproteins.24,25 The vectors were first evaluated inmurine SC1 fibroblasts, HT1080 human fibrosarcomacells, EL4 mouse lymphoma cells and F9 mouse EC cells.Target cells were transduced with a defined multiplicityof infection (MOI) to obtain equal transgene dosage,largely limited to one copy per cell. This is important, aswe have demonstrated in previous work a clear increaseof transgene expression level and clonal variability ofgene expression when introducing more than one copyinto target cells.24,25 Protein expression was determinedby flow cytometry of EGFP fluorescence in single cells ofunselected polyclonal mass cultures.

Cell type dependence of LTR vector expressionIn all SIN vectors except SinSF, the retroviral enhancer-promoter was placed a short distance downstream ofthe PBS, to maintain the highest degree of similarity withthe 50UTR of an LTR-controlled vector. To compare LTRdriven with internal initiation of transcription, we startedwith vectors lacking RNA control elements such as splicesites or the PRE. When comparing SF110 and SinSF110,fibroblasts showed a significantly increased expressionof the LTR configuration (about 50–100% higher activityof SF110 in SC1 and HT1080 cells). In contrast, thecorresponding SinSF110 vector with the internal promo-ter was slightly superior in F9 EC cells (ratio SF110/SinSF110¼ 0.7, P¼ 0.13, Table 1). The reduced activity ofLTR vectors in F9 cells would be compatible with earlierobservations that linked attenuation and shut-off ofretroviral gene expression in primitive embryonic cellsto the presence of tissue-specific repressor elements inthe retroviral enhancer.23,26 The better performance of the

LTR configuration in ‘permissive’ cells such as SC1 orHT1080 may as well be explained by the terminalduplication of the retroviral enhancer. These dataindicated that the duplication of the enhancer-promoterin the LTR (and the deletion of these sequences from theLTR in SIN vectors) has a cell-type-dependent impact onthe transcriptional potency of retroviral vectors. There-fore, subsequent investigations involved a series ofdifferent cell types, with a stronger emphasis on somaticcells that are more relevant for human gene therapy thanEC cells.

Splicing to improve SIN vector performance:psintron vector conceptTo determine the optimal configuration of the 50UTR,several configurations of SIN vectors were compared.The absence of C from the 50UTR (vector SinSF, Figure1b) doubled the expression levels (Po 0.05), irrespectiveof the cell type investigated (Table 1, ratio SinSF/SinSF110). The unspliced C region present in SinSF110may thus inhibit retroviral RNA export and/or transla-tion, possibly as a result of its complex secondarystructure.

We next investigated whether encompassing C bysplice signals would improve the performance of SINvectors. Taking advantage of the necessity for C inretroviral replication would select full-length transgenescontaining the intron in particle formation.7 For biosafetyreasons, this concept appears to be superior to theproposed use of cytoplasmic RNA viruses in theproduction of intron-containing retroviral vectors.27

Consequently, we placed the U3r fragment down-stream of the PBS and thus outside the LTR, butupstream of the SD and C (vector SinSF91, Figure 1b).This region is the only reasonable target for promoterinsertion when attempting to generate SIN vectorscontaining a retroviral intron. However, it was unclearas to whether this region would tolerate the insertion oflarger sequences, especially when considering the highlyconserved primary and secondary structure of theretroviral leader region. The 50UTR was terminated bya minimal retroviral SA mediating balanced splicing inLTR-controlled vectors.7 The novel design for SINvectors with C within the intron was named psintron(vectors SinSF91 and SinSF91P, Figure 1b).

In retroviral packaging cells, genomic RNA of psintronvectors is driven from the MLV enhancer-promoterlocated in the plasmid’s 50LTR. Interestingly, despitethe potential for promoter competition with the neigh-boring internal MLV promoter, the psintron designallowed the production of infectious vector particleswith useful titers (although not as efficiently as the LTR

Table 1 Influence of promoter position, absence of C in the 50UTR and the retroviral intron on transgene expression

Element Ratio F9, N¼ 5 EL4, N¼ 3 SC1, N¼ 16–37 HT1080, N¼ 4–8

Promoter in LTR or 30 of PBS SF110/SinSF110 0.8 1.0 1.4* 1.9*Absence of C in 50UTR SinSF/SinSF110 1.9* 2.4* 1.9* 1.8*Splicing of C in SIN vectors SinSF91/SinSF110 2.8* 4.6* 2.9* 2.9*

Calculations based on mean values of at least four (up to 37) independent determinations per construct for F9, SC1 and HT1080, and threedeterminations for EL4 cells. Standard deviations were o20%.*P-valueo0.05 (Student’s t-test, two-sided).


3

Gene Therapy

vectors, see below). The absence of U5 (containing theattachment sites for the retroviral integrase) and PBS(strictly required for reverse transcription) from theinternal psintron transcript guaranteed that these vectorscannot be mobilized following the first round of retro-viral transduction. Accordingly, retroviral packagingcells transduced with psintron vectors did not produceinfectious particles (data not shown).

To address the impact of the retroviral intron on EGFPexpression, the mean green fluorescence of unselectedmass cultures transduced with SinSF91 was determinedby flow cytometry. The SinSF110 vector lacking splicesites served as a control (same MOI). It should be notedthat the EGFP coding sequences are not dependent onsplicing for mRNA export, while some cellular se-quences such as b-globin are.9 With EGFP, splicingsignificantly increased expression by a factor of 2.8 (F9cells) to almost 5 (EL4 cells, Table 1, ratio SinSF91/SinSF110). Splicing of the retroviral intron was confirmedby Northern blot and RT-PCR analysis (below). Takentogether, these data provide proof of concept for thedesign of MLV-based SIN vectors with a functionalintron in the 50UTR.

These data are in line with our findings using LTRvectors, in which splicing increases EGFP expression upto five-fold both in cell lines and in primary hemato-poietic cells7,9 (and data not shown). Thus, LTR vectorscontaining the engineered intron used here have beendemonstrated to express several transgenes in primaryhematopoietic and lymphoid cells to potentially ther-apeutic levels.28–31

Differentiation dependence of the PREInclusion of the PRE in the 30UTR (vector SinSF91P,Figure 1) further elevated EGFP levels in cell lines(Btwo- to three-fold, Figure 2a and b, Po0.05). Theenhancing effect of the PRE was independent of thevector configuration (psintron or LTR) and occurred inF9 cells as well as in fibroblasts (Figure 2a).

To address the kinetics of transgene expression,polyclonal cultures of unsorted cells were analyzed atregular intervals within 14 days post-transduction. Theproportion of transduced cells remained constant. Withinthe observation period of 14 days in vitro, attenuation orshut-off of gene expression over time was not observedwith any of the vectors. A comparison of all vectors inunselected cultures of SC1 cells or undifferentiated F9cells revealed that SinSF91P, the psintron vector with thePRE in the 30UTR, mediated the best protein expressionlevels of all SIN vectors tested (Figure 2a). According togreen fluorescence, SinSF91P was four times strongerthan the conventional SIN vector (SinSF, Po0.05). Thehistograms demonstrated that EGFP expressionmediated by either SF91P or SinSF91P was well abovebackground levels in both cell types (F9 and SC1), with asimilar variability between independent unselected cells(Figure 2b). The FACS histograms also indicated thatretroviral vector expression was generally attenuated inF9 cells compared with SC1 cells, in line with earlierfindings.23,26 Both in F9 cells and in fibroblasts, SinSF91Peven showed slightly higher activity than the corre-sponding LTR vector (SF91P), although this did notreach statistical significance. Nevertheless, this standsin contrast to the results achieved with SIN vectors

lacking RNA processing modules (ratio SF110/SinSF110,Table 1).

However, in an independent study we have observedthat the PRE reduced rather than increased the expres-sion of SIN vectors in primary hematopoietic cells.19

These results were achieved with vectors in which thePRE was introduced 30 of the EGFP cDNA in the SinSFbackbone (which lacks an intron).19 We obtained similarresults when comparing the LTR vectors SF91 and SF91Pin primary murine hematopoietic cells (Figure 2c). Thecells were transduced with identical MOI of both vectors,resulting in the transduction of o15% of target cells(range 1–14%). After transduction, cells were cultured for

Figure 2 Differentation dependence of the PRE module. (a) F9 and SC1cells polyclonal unselected cultures transduced with equal MOI of thevectors shown. EGFP expression was determined at regular intervalswithin 14 days. The relative mean fluorescence intensity (RFI) wasnormalized for the value obtained with SinSF at day 2. The mean values oftwo to six independent experiments, standard deviations were o30%. ThePRE vectors SF91P (LTR) and SinSF91P show the highest expression.Representative histograms are shown in (b), revealing that vector studieswere performed under conditions of low MOI. (c) The PRE attenuatestransgene expression in primary murine bone marrow cells. Experimentswere performed with the LTR vectors SF91P and SF91 at equal MOI.


4

Gene Therapy

8 days in vitro. Combining the data of four experiments,we observed a B50% reduction of transgene expressionin the presence of the PRE (Po0.05 for all lineages at day8). Under these conditions, the attenuating effect of thePRE in primary hematopoietic cells was independentof the differentiation status (Figure 2c). Nevertheless, wepreferred to introduce the PRE in MLV SIN vectors,because we observed a striking effect on the yield ofinfectious particles produced in retroviral packaging cells.

PRE is important for high titers of psintron vectorsAn important practical aspect of viral vectors is the titerof replication-defective particles harvested from safety-modified packaging cells. To obtain infectious retroviralparticles, vector transcripts starting with the R region ofthe 50LTR and maintaining C must be generated from theplasmid’s 50LTR promoter in the transfected packagingcells. However, transcripts originating in packaging cellsfrom the internal U3r promoter are not able to generateinfectious virus, due to the lack of essential motifs (suchas U5 and PBS) located between +33 and +182 of theretroviral 50UTR.

For identical plasmids, titers of cell-free retroviralvector stocks harvested after independent transfectionsof packaging cells were highly reproducible, but theyvaried strikingly with respect to vector design (Figure 3).Titers were determined by limiting dilution infection ofconstant numbers of target cells (SC1) and flow cyto-metric analyses of transduction frequencies 3 days aftervector exposure. When producing ecotropic particles, thebest LTR vector (SF91P) generated more than 5� 107

transducing units per ml cell-free supernatant. Psintronvectors (SinSF91, SinSF91P) had at least 20-fold lowertiters than their LTR counterparts (SF91 and SF91P,respectively). In psintron vectors SinSF91 and SinSF91Pand the splice-defective control SinSF110, the internalU3r promoter inserted between PBS and SD is expectedto compete with the upstream positioned 50LTR promo-ter of the plasmid, thus lowering infectious titers (seeDiscussion for alternative hypotheses).

Interestingly, the presence of splice sites surroundingthe C region increased rather than decreased infectioustiters, confirming our earlier data obtained with LTRvectors.7 This result may be independent of splicing,possibly due to improved folding of the wild-type leaderand/or more stringent interaction with gag nucleocapsiddomains.

The PRE improved titers with a much stronger effectin the SIN configuration (10� increase) than in the LTRcontext (2� increase). The comparison of the psintronvector (SinSF91) with the conventional SIN vector (SinSF)revealed that the insertion of the promoter between PBSand SD reduced the output of infectious particles 5- to10-fold (Figure 3). Still, a reasonable yield (above 106

particles/ml) was achieved with psintron vectors whenintroducing the PRE (SinSF91P).

We also tested whether introducing the PRE into theresidual U3 region of the psintron vector (instead ofupstream of the residual 30LTR) would result in a betterexpression. However, vector titers were suboptimal(average yield 1.5� 106/ml versus 3� 106/ml withSinSF91P), and the expression levels were also slightlyreduced compared with SinSF91P (range 85–95%, n¼ 7,SC1 fibroblasts).

Finally, we used real-time PCR to determine thetransgene copy number in SC1 cells transduced withan MOI of up to 0.3, using either SF91P or its SINcounterpart, SinSF91P. The real-time PCR showed verysimilar threshold cycles for both vectors (SF91P:30.870.2; SinSF91P: 30.370.1), indicating that the re-duced titers of the psintron vector could not be explainedby an increased rate of silencing.

In summary, the PRE was required to producesufficient titers of the psintron vectors, thus avoidingthe need of particle concentration when targetingprimary cells. It should be noted that, despite thesignificant differences in vector yields, all expressionanalyses described here were performed using normal-ized titers, ruling out the possibility that results werecaused by different gene dosages.24,25

Improved splicing of the retroviral intron in psintronvectorsTo investigate to what extent EGFP expression levelswere dependent on RNA expression levels and splicing,total cellular RNA was prepared from sorted butuncloned populations of SC1 cells transduced with thevectors of interest and analyzed in Northern blots. Nosplicing of the constructs was observed in the absence ofthe retroviral splice sites (Figure 4a, lanes 2, 3 and 6).Similar results were obtained by RT-PCR using RNAfrom transduced EL-4 cells (Figure 4b).

A surprising and important finding of these experi-ments was that vectors containing an identical retroviralintron (SD-C-SA) but different sequences upstream of SDdiffered with respect to the expression of unsplicedRNA. Complete splicing of the intron was observed inSinSF91 (psintron), as opposed to balanced splicing inthe LTR-driven vector, SF91 (Figure 4a, compare lanes 4and 7). Similar findings were made when the PRE wasintroduced into the 30UTR (Figure 4a, compare lanes 5and 8), although a faint larger band indicating unsplicedRNA was observed with vector SinSF91P (lane 5). Theamount of unspliced RNA was thus strongly reducedwhen compared with the LTR vector SF91P (lane 8).Prolonged exposure of the Northern blot did not revealaberrant cryptic splicing activity with any of the vectorsshown (data not shown). In line with improved splicing,the increase in RNA levels mediated by the splice siteswas stronger in the context of the SIN vectors (SinSF91

Figure 3 Reasonable titers can be obtained with the psintron configura-tion, although the yield of infectious units is lower than with LTR vectors.Ecotropic vector titers 48 h after transfection in packaging cells, meanvalues and standard deviations of at least three independent experiments.ETU, EGFP transfer units. Error bars indicate standard deviations.


5

Gene Therapy

versus SinSF110, Figure 4c) than with the LTR vectors(SF91 versus SF110, Figure 4c).

As described above, the major sequence differencebetween the 50UTRs of SF91 (alternative splicing withequimolar transcript ratio) and SinSF91 (complete spli-cing) was the absence of nucleotides located between +33and +182 in the latter. RNAs of both vectors containedthe RSL sequence, previously found to be required forthe accumulation of unspliced RNA.15 Our data thereforesuggest that the RSL region cooperates with sequenceslocated between +33 and +182, either to promote exportof the unspliced transcript or to impede splicing ofthe retroviral transcript. Moreover, the quantificationof the hybridization signals revealed that the presence ofC reduced overall RNA levels (compare SinSF andSinSF110, Figure 4c). Taken together, these data indicatepreviously unknown functions of retroviral leaderelements in RNA splicing and export. The precise natureof the sequences and mechanisms involved is subject ofan ongoing investigation.

Improved protein expression of SIN vectorsRNA levels and EGFP fluorescence were generally ingood agreement within the series of SIN vectors (Figure4c). In contrast, all LTR vectors (SF110, SF91, SF91P)showed a relatively lower EGFP expression per RNA,

which was likely due to insufficient expression and/orexport of unspliced LTR transcripts. In line with thishypothesis, presence of the PRE partially compensatedthe less efficient protein expression of LTR vectors. Thesedata showed that although LTR vectors tend to producemore RNA, protein expression levels achieved withpsintron vectors were at least equivalent. We concludethat a reduced (in comparison with LTR vectors)transcriptional activity of SIN vectors can be compen-sated by improved RNA processing.

The vector-dependent differences in protein expres-sion levels observed with the sorted mass cultures(Figure 4c) were similar to those of unsorted cells (Figure2a). It appears safe to conclude from the above findingsthat psintron vectors are capable of mediating similarexpression levels as their LTR counterparts whilesimultaneously reducing the load of potentially ‘danger-ous’ enhancer sequences by 50% and completely avoid-ing the presence of a promoter in the 30LTR.

Expression of SinSF91P in primary hematopoieticand lymphoid cellsTo demonstrate the utility of psintron vectors for genetransfer into primary hematopoietic cells, we transducedmurine bone marrow cells using unconcentrated

Figure 4 Psintron vectors differ from corresponding LTR vectors in splicing pattern and EGFP expression efficiency. (a) Northern blot of vector transcriptsin mass cultures of transduced SC1 fibroblasts, hybridized using an EGFP probe, boxed signals representing the b-actin control for RNA load. (b) Correctsplicing of the vectors was confirmed using RT-PCR with RNA from transduced uncloned EL4 cells. The forward primer annealed to the 30-end of theinternal U3r fragment, the reverse primer to the 50-end of EGFP. The products corresponding to the spliced transcript of SinSF91 vectors, the unsplicedtranscript of SinSF vectors and the unspliced transcript of SinSF110 vectors were 285, 240 and 660 bp, respectively. The size of the PCR product for SF110was 790 bp. Lanes 1, 2, 4, 5 and 7 show data from vectors harboring putative nuclear scaffold attachment regions in the LTR, which did not affect splicing.Gels of control PCRs performed on RNA preparations without prior RT were negative (not shown). (c) Quantitative comparison of RNA and protein levels.Vector RNA levels were quantified using a phoshpoimager and normalized for the b-actin RNA load. EGFP levels were derived from the mean fluorescenceintensity measured by flow cytometry. Data obtained with the SinSF vector were set to 1.


6

Gene Therapy

supernatants and cultured cells for 8 days post-transduc-tion. As in all other experiments described above, wecompared SF91P and SinSF91P with low MOIs to achievepolyclonal mass cultures with gene expression frequen-cies below 30%. This results in single copy insertions inthe majority of EGFP+ cells, as shown in detail in ourearlier studies.24,25,32 SinSF91P almost matched theexpression level of SF91P in primary murine hemato-poietic cells, irrespective of the differentiation status ofthe cells (Figure 5a and b).

Ecotropic supernatants were also used to transduceprimary murine T cells. The average transductionefficiency was 20% for SF91P (n¼ 2) and 14% forSinSF91P (n¼ 4). Cells were analyzed for 8 days inculture, showing strong expression of both vectors(Figure 5c). Interestingly, SinSF91P tended to mediatehigher (B30%) transgene expression than SF91P (Figure5d). For the transduction of human primary T cellsderived from peripheral blood leukocytes, particlespseudotyped with the envelope protein of GALV wereproduced. The titer of the SinSF91P vector did not exceed106/ml in this production process, resulting in a lowgene transfer efficiency into human T cells (1–2%). Still,a side-by-side comparison with SF91P was possible,showing similar results as with murine T cells. In CD4+helper T cells as well as in CD8+ cytotoxic T cells,SinSF91P mediated B25% stronger EGFP expression incomparison with SF91P (n¼ 2 transductions for eachvector, mean fluorescence levels varied by less than 10%in independent cultures transduced with the samevector).

Discussion

This study introduces a novel configuration of intron-containing SIN retroviral vectors with promising expres-sion features in embryonic and somatic cells andpotentially increased biosafety. We also show for thefirst time that foreign sequences can be inserted betweenthe PBS and the retroviral SD, opening further perspec-tives for the development of retroviral (includinglentiviral) vectors.

We started with the observation that SIN vectors withinternal MLV promoters showed better expression thancorresponding LTR vectors in undifferentiated EC cells,but not in fibroblasts. An intact LTR probably increasesthe susceptibility to cell-type-specific transcriptionalactivators and repressors by duplicating the retroviraltarget sequences. The likelihood of attracting activatorsmay thus be increased in differentiated cells, but not inundifferentiated EC cells that lack key regulators of theretroviral enhancer such as AML1/PEBP.33,34 Likewise,the LTR configuration may increase the susceptibilityto transcriptional repressors in EC cells, which expressseveral negative regulators of MLV transcription.23,26

Superior persistence of gene expression in differentiatingembryonic stem cells has been demonstrated for lentivir-al vectors with an internal promoter in comparison toretroviral LTR vectors.35–37 Our data, in line withprevious observations of Stewart et al,38 suggest ahypothesis that improved performance in primitiveembryonic cells is mainly dependent on the use of anappropriate promoter in an internal position of a retro-viral vector.

Figure 5 Potent expression of SinSF91P in primary hematopoietic cellsand T-lymphocytes. (a) Murine lineage-depleted bone marrow cells weretransduced using an MOI of 1 for each vector. Transduced cells werecultured for 8 days as described in Materials and methods. Flow cytometrywas performed on the populations defined by antibodies against surfacemarkers c-Kit, Sca1, CD11b, Gr1 and a lineage (Lin) cocktail comprisingCD11b, Gr1, Ter119, B220 and CD3. The EGFP expression was verysimilar with both vectors, irrespective of the differentiation status. Thehistograms shown in (b) underline that SinSF91P mediated similar EGFPexpression as SF91P in undifferentiated as well as differentiatinghematopoietic cells. In murine T cells analyzed up to 8 days post-transduction (c) as well as human T cells analyzed up to 9 days post-transduction (d), SinSF91P expressed EGFP to slightly higher levels thanthe LTR vector SF91P. MeanF, mean fluorescence intensity of EGFP.


7

Gene Therapy

To increase the potential of SIN vectors, we focused onthe improvement of RNA processing. This approach maybe as relevant for both biosafety and efficiency as theinclusion of chromatin modulating elements,39 becauseimproved RNA processing reduces the need for strongenhancers. We have shown that an intron in the 50UTRprovides a robust increase in transgene expression,while the PRE acts in a more differentiation-dependentmanner. The PRE thus even decreased transgene expres-sion in primary hematopoietic cells, at least in the contextof the vectors studied here that express high levels oftranscript in these cells. Similarly, the PRE has beenfound to reduce expression in primary hematopoieticcells when used in MLV SIN vectors that did not containan intron.19 This may indicate a relative deficiency ofcellular cofactors such as CRM-1 which seem to berequired for PRE activity.20 In packaging cells andnonhematopoietic target cells studied here, the PREenhanced transgene expression. Given the requirementof the PRE to obtain high-titer supernatants of MLV SINvectors and its relatively mild attenuation of transgeneexpression in primary hematopoietic cells, we wouldrecommend the inclusion of this element in MLV SINvectors. The preferred position appears to be upstream ofthe 30 SIN LTR, but inclusion into the SIN LTR alsoworked without a substantial loss of activity.

Our study establishes that an intron can be introducedin SIN vectors when placing the promoter in a novelposition upstream of the SD and thus a short distancedownstream of the PBS. The presence of C within theintron guaranteed its transfer to target cells. HIV vectorsmay be able to transfer intron-containing transgenes,possibly due to splice inhibitory features of the Rev/RREinteraction occurring in packaging cells.40 However, thefrequency of intron loss has not been carefully investi-gated. The MLV vector technology lacks such a condi-tional splice inhibitory mechanism. The psintron strategycompensates this deficiency and may be more reliablethan the Rev/RRE interaction in transducing the intronto target cells. The psintron concept is therefore also ofinterest for the construction of lentiviral vectors, espe-cially when their packaging process is rendered Revindependent.41

The expression of SIN vectors was enhanced by theintron in the 50UTR, which could be explained by thepositive feedback of splicing factors with transcriptionalelongation and the improved nuclear export of splicedRNA.10,11 The psintron architecture should be of parti-cular interest when attempting to express cDNAs such asb-globin, which are even more dependent on splicing fornuclear export.9

A welcome consequence of the psintron design wasthe virtually complete splicing of the retroviral intron.This may prevent interaction of the retroviral SD withdownstream splice acceptor sequences of cellular genesand thus further reduce the mutagenic potential ofrandom vector insertions. Improved splicing also ele-vates expression and may thus allow the use of weakerpromoters without loss of efficiency. Somewhat surpris-ingly, improved splicing occurred in our constructsdespite the use of retroviral splice consensus motifs,which are considered to be suboptimal.12,14 Our studythus points to the presence of yet unknown retroviralRNA processing signals, which may either have a spliceinhibitory activity or promote the nuclear export of

unspliced retroviral RNA. As indicated by our ongoingexperiments, these sequences are likely located upstreamof the position in which we introduced the promoter ofpsintron vectors (in the region comprising the 30 half ofR, the complete U5, the PBS and 17 bp 30 of the PBS).

While SIN (including psintron) vectors offer valuablesafety features and should be suitable to incorporatepromoters for inducible or differentiation-specific ex-pression, their major disadvantage is related to vectorproduction. Cloned producer cells with high titers forclinical scale vector production are not easily obtainedwith such a vector configuration. Several reasons mayhave contributed to the reduced yield of infectiousparticles with the vectors developed in the presentstudy: (i) the known negative effect of U3 deletions onthe titer of MLV vectors,6 a problem not encounteredwith HIV-derived vectors;40 (ii) promoter interference42

between the 50LTR and the nearby located internalpromoter of psintron vectors, leading to reduced expres-sion of genomic RNA in transfected packaging cells; (iii)interference of the internal U3r with the functions ofneighboring domains of C or PBS; and (iv) suboptimalprocessing during later steps of reverse transcription.Considering that titer optimization was not the focusof the present study, the results obtained with the firstgeneration of psintron vectors were encouraging, inparticular, with the use of the PRE (SinSF91P). Of note,both titers and expression levels were higher than thoseachieved with the intronless ‘conventional’ SIN vector,SinSF. This allowed a reasonable transduction efficiencyin primary hematopoietic cells without the need forconcentrating vector supernatants.

Finally, it remains to be tested whether the SINconfiguration of MLV vectors affects termination andpolyadenylation of the retroviral RNA, as previouslydescribed for HIV-based SIN vectors.43 Since by far themost frequent mechanism for insertional proto-oncogeneactivation is long-distance enhancer interaction,44 thereduction of enhancer load in the SIN configurationalready promises a major step towards increased safety.This may be further enhanced when incorporatingweaker enhancer-promoter elements than the one usedhere, given that therapeutic levels of transgene expres-sion can still be achieved.

In addition to the implications for retroviral vectorexpression and safety, our observation that the smallspace between the PBS and the SD can accommodateforeign sequences extends the flexibility in the design ofretroviral (including lentiviral) vectors. In an LTR-controlled backbone, this position could be used tointroduce a coding sequence or a transcriptional en-hancer. In SIN vectors, this position may be used forinsertion of any type of chromatin-organizing domain(such as matrix attachment region, insulator, enhancer),recognition sites for sequence-specific recombinases orpossibly even complete transgene cassettes (including apromoter). Fine-tuning of the distance to the PBS and theSD may result in further improved vector performance.

Materials and methods

PlasmidsVector plasmids pSF110, pSF91 and pSF91P wereconstructed using the modular assembly described


8

Gene Therapy

earlier.9 A SalI restriction site was introduced at position+179 (relative to the cap site) of the retroviral untrans-lated leader region by PCR (primer leaSal+ 50-GGAGACCCCTGCCCGTCGACCACCGACCCCCCCG-30, mu-tating bases underlined), to allow insertion of foreignsequences. For the construction of the SIN vectors, theplasmids 30LTR was deleted using two PCR products.The first was generated using oligonucleotide U3NheI(50-ATCGCTAGCGGTGGGGTCTTTCATTCCCCCCT-30)and M13forward (50-TGACCGGCAGCAAAATG-30), andcleaved with BamHI, cutting after the EGFP cDNA. Thesecond product was obtained using primers U3EcoRI(50-ATCGAATTCACAACCCCTCACTCGGCGCGCCA-30)and M13reverse (50-GGAAACAGCTATGACCATG-30),and cleaved with XhoI, cutting after the end of the30LTR. Ligation of the cleaved PCR products generatedan EcoRV site between NheI and EcoRI, all correspondingto the primer sequences. Thus, only the first 22 bp andlast 14 bp of the U3 region were left, excluding the entireenhancer/promoter. The U3-RSL fragment was ampli-fied and tailed with novel restriction sites using primersXhoEagSF65 (50-CTCGAGCGGCCGCTAGCTGCAGTAACGC-30) and SalEagSF63 (50-GTCGACCGGCCGACTCAGTCAATCGG-30). For insertion into the NotI sitelocated downstream of C, EagI was used, resulting inpSinSF. For insertion into the SalI site located betweenthe PBS and the SD, the product was cut using XhoI andSalI, resulting in pSinSF110. By exchanging the sequencesbetween SalI and NotI, a defined leader fragmentdistinguished by the presence of splice sites wasintroduced, resulting in plasmid pSinSF91. The PREelement was obtained from plasmid pSF91EGFP-gPRE,9

and introduced either downstream of EGFP (plasmidpSinSF91P) or into the SIN LTR (plasmid pSinSF91P2)after a partial digest with EcoRI. All PCRs for cloningwere performed using proofreading Pfx polymerase(Invitrogen, Karlsruhe, Germany) in combination withTaq polymerase (Qiagen, Hilden, Germany).

Cells and transfectionsPhoenix-gp packaging cells and human HT1080 fibro-sarcoma cells were grown in Dulbecco’s modified Eaglemedium, mouse SC1 fibroblasts (ATCC # CRL-1404), andmouse F9 EC cells (ATCC # CRL-1720) in minimalessential medium, mouse EL4 lymphoma cells (ATCC #TIB-39) in RPMI. The media were supplemented with10% FCS, 2 mM glutamine, 1 mM sodium pyruvate andwith 100 U/ml penicillin/streptomycin. The day beforetransfection of Phoenix-gp cells,45 5� 106 cells wereplated in a 9 cm plate. For transfection, the mediumwas exchanged and 25� mM chloroquine (Sigma) wasadded. Retroviral vector DNA (6 mg) was transfectedusing the calcium phosphate precipitation method. Togenerate VSV-G pseudotyped particles, 10 mg M57 (MLVgag/pol expression plasmid) and 2 mg M75 (VSV-Gexpression plasmid) were transfected in addition. Forthe generation of ecotropic or GALV-pseudotypedparticles, an expression vector for the ecotropic envelopeprotein or the GALV glycoprotein, respectively, was usedinstead of M75.46,47 The medium was exchanged after 6–12 h. In some experiments, equal transfection efficiencywas controlled by flow cytometry of transfected cells.Supernatants containing the viral particles were collected24–84 h after transfection, filtered through a 0.22 mm filterand used to transduce 1�105 target cells. Transduction

was assisted by adding 4 mg/ml protamine sulfate andcentrifugation for 60 min at 400 g and 25–321C. Vectortiters were determined by transducing predefinednumbers of SC1 or HT1080 target cells with serialdilutions of supernatant and analyzing the percentageof positive cells by flow cytometry. Further analysis waslimited to those experiments where less than 30% oftarget cells were productively transduced. After trans-duction, cells were grown for another 2–3 days beforeanalysis. Sorted populations of EGFP-expressing cellswere grown for at least 5 weeks before analysis.

Primary hematopoietic cells and lymphocytesBone marrow cells from C57Bl/6 mice were collectedfrom femurs and tibias and suspended in IMDM plus10% FCS. Lineage marker negative cells were enrichedusing the mouse lineage panel (BD Pharmingen, Heidel-berg, Germany) and MACS column (Miltenyi, Bergisch-Gladbach, Germany) separation following the manufac-turer’s instructions, as described.48 The flow throughfraction (Lin�) was cultured in IMDM supplementedwith 20% FCS, 2 mM glutamine, 100 U/ml penicillin/streptomycin, 50 ng/ml mSCF, 100 mg/ml hFLT3L,100 ng/ml hIL11 and 20 ng/ml mIL3.48 At the thirdday of culture, cells were transduced with cell-freeretroviral particles at an MOI of 1.48 Flow cytometrywas performed one and 8 days after gene transfer usinglineage antibodies (BD Pharmingen) as indicated inResults.

Human mononuclear blood cells were isolated at day1 from healthy donor buffy coats by centrifugation(900 g; 20 min) on a Ficoll gradient (Biochrom, Berlin,Germany). T-lymphocytes were activated (using anti-CD3 and anti-CD28 antibodies) and cultured in X-Vivo10containing 8% autologous serum (in the presence of300 U/ml IL-2) essentially as described.49 The firsttransduction was carried with the five vectors SF91,SF91P, SinSF91, SinSF91P and, for control, SF11EGFPrev24

and with noninfectious supernatant (mock control) in themorning of day 4 using our centrifugation protocol.49 Asecond infection cycle was performed in the evening ofthe same day for the low titer constructs SinSF91 andSinSF91P. All procedures were repeated on day 5. On day6, only SinSF91 and SinSF91P were used for repeatedtransduction cycles. Overall, we performed six transduc-tion cycles with the low-titer and 2 with the high-titerconstructs. All cells were kept at 371C and 5% CO2 in ahumidified atmosphere. Transduction efficiency andtransgene expression were measured in the evening ofday 8, on days 11 and 15 on a FACScan instrument usingCellQuest software (Becton Dickinson, San Jose, CA,USA).

Flow cytometryAt least 106 cells were harvested and washed in PBS. Todetermine the expression of the EGFP transgene, cellswere analyzed in a FACScalibur (Becton Dickinson,Heidelberg, Germany) using CellQuest software (BectonDickinson). In the analysis of immortalized cells, a gatewas set on a homogeneous cell population, as deter-mined by scatter characteristics. A marker was set tocalculate the percentage and mean fluorescence intensityof positive cells. Mass cultures of EGFP-expressing F9or SC1 cells were sorted using a MoFlo instrument(Cytomation, Freiburg, Germany).


9

Gene Therapy

Northern blot and RT-PCRTotal RNA preparation was performed using the RNAInstapure reagent (Eurogentec, Brussels, Belgium). ForNorthern blot, 15 mg of each RNA were separated at 2 V/cm in 1% agarose gels, after denaturing RNA sampleswith glyoxal and DMSO. Subsequently RNAs weretransferred onto Biodyne B membrane (0.45 mm, Pall)by capillary transfer and UV crosslinked (Stratalinker,Stratagene). Specific probes used for hybridizationcorresponded to the cDNAs of EGFP or chicken b-actin.Probes (25 ng) were radiolabeled using the Prime it IIkit (Stratagene) to an activity of at least 106 cpm/mlhybridization solution and separated from unincorpo-rated nucleotides on spin columns. Filters were washed,sealed and exposed to X-ray films (Kodak X-omat-AR) orquantified by phosphoimager (Fuji, Dusseldorf, Ger-many) analysis. RNAs of transduced, unsorted EL4 cellswere reverse transcribed using Superscript II RT (GibcoBRL, Karlsruhe, Germany) and oligo(dT)15 primer(Boehringer Mannheim, Germany). Specific productswere obtained by PCR using primers SD5-1 (50-CCGTATTCCCAATAAAGCCT-30) and EGFPreverse (50-GGTCAGCTTGCCGTAGGTGGC-30).

Real-time PCRGenomic DNA was purified using the QIAamp DNABlood Mini Kit (Qiagen, Hilden, Germany). Quantitativereal-time PCR was performed using the QuantiTect SYBRGreen PCR Kit (Qiagen) and the iCycler real-time PCRdetection system (Bio-Rad Laboratories, Muenchen,Germany) according to the manufacturer’s instructions.Primers 50GFP RT (TATATCATGGCCCGACAAGCA)and 30GFP RT (ACTGGGTGCTCAGGTAGTG) amplifieda 162 bp EGFP DNA fragment.19 DNA samples weremeasured in triplicate.

Acknowledgements

This work was supported by grants from EuropeanCommunity (QLK3-2001-01265, QLRT-2001-00427) andthe DFG (BA 1837/4-1). We thank Andreas Rimek forflow-cytometry sorting of the EGFP+ mass cultures oftransduced cells and Cordula Gruttner for her technicalassistance.

References

1 Baum C et al. Side effects of retroviral gene transfer intohematopoietic stem cells. Blood 2003; 101: 2099–2114.

2 Thomas CE et al. Progress and problems with the use of viralvectors for gene therapy. Nat Rev Genet 2003; 4: 346–358.

3 Hacein-Bey-Abina S et al. LMO-2 associated clonal lympho-proliferations in two patients after gene therapy for SCID-X1.Science 2003; 302: 415–419.

4 Kohn D et al. American society of gene therapy (ASGT) ad hocsubcommittee on retroviral-mediated gene transfer tohematopoietic stem cells. Mol Ther 2003; 8: 180–187.

5 Kohn DB et al. Occurrence of leukaemia following gene therapyof X-linked SCID. Nat Rev Cancer 2003; 3: 477–488.

6 Yu SF et al. Self-inactivating retroviral vectors designed fortransfer of whole genes into mammalian cells. Proc Natl Acad SciUSA 1986; 83: 3194–3198.

7 Hildinger M et al. Design of 50 untranslated sequences inretroviral vectors developed for medical use. J Virol 1999; 73:

4083–4089.8 Kim SH et al. Construction of retroviral vectors with improved

safety, gene expression, and versatility. J Virol 1998; 72: 994–1004.9 Schambach A et al. Context dependence of different modules for

posttranscriptional enhancement of gene expression fromretroviral vectors. Mol Ther 2000; 2: 435–445.

10 Fong YW, Zhou Q. Stimulatory effect of splicing factors ontranscriptional elongation. Nature 2001; 414: 929–933.

11 Reed R, Hurt E. A conserved mRNA export machinery coupledto pre-mRNA splicing. Cell 2002; 108: 523–531.

12 Hammerskjold ML. Constitutive transport element-mediatednuclear export. Curr Top Microbiol Immunol 2001; 259: 77–93.

13 Wodrich H et al. A new RNA element located in the codingregion of a murine endogenous retrovirus can functionallyreplace the rev/rev-responsive element system in humanimmunodeficiency virus type 1 gag expression. J Virol 2001; 75:

10670–10682.14 Wodrich H, Krausslich HG. Nucleocytoplasmic RNA

transport in retroviral replication. Results Probl Cell Differ 2001;

34: 197–217.15 Trubetskoy AM et al. R region sequences in the long terminal

repeat of a murine retrovirus specifically increase expression ofunspliced RNAs. J Virol 1999; 73: 3477–3483.

16 Aiuti A et al. Correction of ADA-SCID by stem cell gene therapycombined with nonmyeloablative conditioning. Science 2002;296: 2410–2413.

17 Hacein-Bey-Abina S et al. Sustained correction of X-linked severecombined immunodeficiency by ex vivo gene therapy. N Engl JMed 2002; 346: 1185–1193.

18 Zufferey R et al. Woodchuck hepatitis virus posttranscriptionalregulatory element enhances expression of transgenes delivered

by retroviral vectors. J Virol 1999; 73: 2886–2892.19 Werner M et al. B cell-specific transgene expression using a self-

inactivating retroviral vector with human CD19 promoter andviral posttranscriptional regulatory element. Gene Therapy 2004;

11: 992–1000.20 Popa I et al. CRM1-dependent function of a cis-acting RNA

export element. Mol Cell Biol 2002; 22: 2057–2067.21 Baum C et al. Novel retroviral vectors for efficient expression of

the multidrug-resistance (mdr-1) gene in early hemopoietic cells.J Virol 1995; 69: 7541–7547.

22 Baum C et al. Retroviral vector-mediated gene expression inhematopoietic cells. Curr Opin Mol Ther 1999; 1: 605–612.

23 Grez M et al. Embryonic stem cell virus, a recombinant murineretrovirus with expression in embryonic stem cells. Proc NatlAcad Sci USA 1990; 87: 9202–9206.

24 Wahlers A et al. Influence of multiplicity of infection and proteinstability on retroviral vector-mediated gene expression inhematopoietic cells. Gene Therapy 2001; 8: 477–486.

25 Kustikova O et al. Dose finding with retroviral vectors:correlation of retroviral vector copy numbers in single cellswith gene transfer efficiency in a cell population. Blood 2003; 110:2099–2104.

26 Challita PM et al. Multiple modifications in cis elements of thelong terminal repeat of retroviral vectors lead to increasedexpression and decreased DNA methylation in embryoniccarcinoma cells. J Virol 1995; 69: 748–755.

27 Li KJ, Garoff H. Packaging of intron-containing genes intoretrovirus vectors by alphavirus vectors. Proc Natl Acad Sci USA1998; 95: 3650–3654.

28 Engels B et al. Retroviral vectors for high transgene expression inT lymphocytes. Hum Gene Ther 2003; 14: 1155–1168.

29 Knipper R et al. Improved post-transcriptional processing of anMDR1 retrovirus elevates expression of multidrug resistancein primary human hematopoietic cells. Gene Therapy 2001; 8:

239–246.


10

Gene Therapy

30 Schiedlmeier B et al. High-level ectopic hoxb4 expression confersa profound in vivo competitive growth advantage on humancord blood cd34+ cells, but impairs lymphomyeloiddifferentiation. Blood 2003; 101: 1759–1768.

31 Egelhofer M et al. Inhibition of HIV-1 entry in cells expressinggp41-derived peptides. J Virol 2004; 78: 568–575.

32 Wahlers A et al. In vivo analysis of retroviral enhancer mutationsin hematopoietic cells: SP1/EGR1 and ETS/GATA motifscontribute to long terminal repeat specificity. J Virol 2002; 76:303–312.

33 Baum C et al. The potent enhancer activity of SFFVp inhematopoietic cells is governed by a binding site for sp1 inthe upstream control region and by a unique enhancercore creating an exclusive target for PEBP/CBF. J Virol 1997;71: 6323–6331.

34 Sun W et al. Transactivation of the Moloney murine leukemiavirus and T-cell receptor beta-chain enhancers by CBF and ETSrequires intact binding sites for both proteins. J Virol 1995; 69:4941–4949.

35 Hamaguchi I et al. Lentivirus vector gene expression during EScell-derived hematopoietic development in vitro. J Virol 2000; 74:10778–10784.

36 Lois C et al. Germline transmission and tissue-specificexpression of transgenes delivered by lentiviral vectors. Science2002; 295: 868–872.

37 Pfeifer A et al. Transgenesis by lentiviral vectors: lack of genesilencing in mammalian embryonic stem cells andpreimplantation embryos. Proc Natl Acad Sci USA 2002; 99:2140–2145.

38 Stewart CL et al. Expression of retroviral vectors in transgenicmice obtained by embryo infection. EMBO J 1987; 6: 383–388.

39 Ramezani A et al. Performance- and safety-enhanced lentiviralvectors containing the human interferon-beta scaffoldattachment region and the chicken beta-globin insulator. Blood2003; 101: 4717–4724.

40 Ailles LE, Naldini L. HIV-1-derived lentiviral vectors. Curr TopMicrobiol Immunol 2002; 261: 31–52.

41 Wagner R et al. Rev-independent expression of synthetic gag-polgenes of human immunodeficiency virus type 1 and simianimmunodeficiency virus: implications for the safety of lentiviralvectors. Hum Gene Ther 2000; 11: 2403–2413.

42 Emerman M, Temin HM. Genes with promoters in retrovirusvectors can be independently suppressed by an epigeneticmechanism. Cell 1984; 39: 449–467.

43 Zaiss AK et al. RNA 30 readthrough of oncoretrovirus andlentivirus: Implications for vector safety and efficacy. J Virol 2002;76: 7209–7219.

44 Akagi K et al. RTCGD: retroviral tagged cancer gene database.Nucleic Acids Res 2004; 32: D523–D527.

45 Kinsella TM, Nolan GP. Episomal vectors rapidly and stablyproduce high-titer recombinant retrovirus. Hum Gene Ther 1996;7: 1405–1413.

46 Morita S et al. Plat-e: an efficient and stable system for transientpackaging of retroviruses. Gene Therapy 2000; 7: 1063–1070.

47 Fehse B et al. A novel ‘sort-suicide’ fusion gene vector for t cellmanipulation. Gene Therapy 2002; 9: 1633–1638.

48 Li Z et al. Predictable and efficient retroviral gene transfer intomurine bone marrow repopulating cells using a defined vectordose. Exp Hematol 2003; 31: 1206–1214.

49 Ayuk F et al. Establishment of an optimised gene transferprotocol for human primary T lymphocytes according to clinicalrequirements. Gene Therapy 1999; 6: 1788–1792.


11

Gene Therapy

Documents

Self-inactivating retroviral vectors with improved RNA processing