HUMAN GENE THERAPY 11:771– 775 (March 20, 2000)Mary Ann Liebert, Inc.

Brief Report

Retroviral Preparations Derived from PA317 Packaging CellsContain Inhibitors That Copurify with Viral Particles and

Are Devoid of Viral Vector RNA

JURGEN SEPPEN, SIMON BARRY, GREGORY M. LAM, NAGARAJAN RAMESH, and WILLIAM R.A. OSBORNE

ABSTRACT

Obtaining high expression levels of a therapeutic gene in target cells could be achieved by integrating multi-ple copies of a recombinant retrovirus. However, we observed that cells retrovirally infected at high multi-plicities of infection (MOIs) carried only single or double integrated proviral copies, suggesting that maxi-mum retroviral transduction was achieved at relatively low MOIs. The same results were obtained whenpurified virus, free of most medium components, was used. Retroviral infection was shown to be inhibited bysupernatants of other viral producer cell lines, and this inhibition could be removed by a centrifugation stepthat also removed more than 90% of infectious virus. Quantitative-competitive PCR of retroviral prepara-tions showed that the amount of retroviral vector RNA present was similar to the amount expected on thebasis of virus titers. Our data suggest that retroviral preparations derived from PA317 packaging cells con-tain inhibitors that copurify with retroviruses and do not contain viral vector RNA. We postulate that theseinhibitor particles cannot achieve a productive infection but interfere with transduction of the target cells byinfectious virions. This study might define an important criterion for the selection of more effective packag-ing cell lines.

771

INTRODUCTION

S IN CE RETR O VIR USES infect and integrate in dividing cells theyare ideally suited for gene therapy applications that involve

the isolation, ex vivo transduction, and expansion of target cellssuch as fibroblasts, hematopoietic stem cells, smooth musclecells, and lymphocytes. For an effective therapy high levels oftransduced gene expression are essential and this can beachieved by integrating multiple copies of the virus per cell.However, several studies have shown that when cells are in-fected at increasing multiplicities of infection (MOIs), the ex-pression levels reach a plateau at relatively low amounts of virus(Kahn et al., 1992; Le Doux et al., 1996). Titers of retroviralpreparations are typically performed at high dilutions and lowMOIs and this is not necessarily a good predictor of transduc-

tion efficiency at high MOIs (Forestell et al., 1995). As thenumber of amphotropic receptor molecules per target cell hasbeen estimated to be 8 3 104 (Battini et al., 1996) an inhibit-ing factor may be responsible for these observations. Since in-hibitors in the original preparation are diluted as well, such titersmay not be representative of infectivity at low dilutions andhigh MOIs. These observations are in agreement with studiesshowing that simultaneous transduction occurs infrequentlywhen cells are coinfected with two different retroviral vectors(Walker et al., 1996; MacNeill et al., 1999). These studies pro-vided evidence that retroviral preparations may contain in-hibitors of infection. Therefore, to understand further the in-ability to obtain high copy number in target cells when usingvirus at high MOIs, we investigated retroviruses prepared fromPA317 packaging cell lines.

Department of Pediatrics, University of Washington School of Medicine, Seattle, WA 98195.

MATERIALS AND METHODS

Cell culture and transduction

The LNFZ retroviral vector has the neo gene under the con-trol of the viral long terminal repeat (LTR) and b -galactosidase( b -Gal) cloned 3 9 to the foot-and-mouth disease virus internalribosome entry site (Ramesh et al., 1996). PA317 packagingcells were used for virus production (Miller and Rosman, 1989)and three different PA317 LNFZ producer clones were studied.Titers of LNFZ varied between 5 3 105 and 5 3 106 CFU/mland were free of replication-comp etent virus as determined bymarker rescue assay. Producer lines were cultured in Dul-becco’s Modified Eagle’s medium (DMEM) (Ramesh et al.,

1996). Afer 24 hr of culture virus was collected from conflu-ent plates of producer cells, filtered through a 0.45- m m poresize filter, and used immediately. Virus titers were perform edby seeding 4 3 105 3T3 cells in 6-cm-diamete r plates. The nextday diluted virus was added in 4 ml of DMEM with Polybrene(4 m g/ml) and cells were split 1:10 immediately after infection.After culture in the presence of G418 (1 mg/ml) for 6 days, thecolonies were fixed, stained, and counted (Ramesh et al., 1996).

Infections of cells for quantitative determination of b -Galactivity were perform ed under the same conditions in the samevolume of medium, except that different amounts of undilutedvirus were used and 2 3 105 cells were seeded. The media werechanged the day after infection and on day 4 the cells were har-vested with 150 m l of 0.5% Triton X-100, 1 mM EDTA in phos-phate-buffered saline (PBS), centrifuged, and b -Gal activity ofclear supernatant was determined using o-nitrophenyl- b -D -galactopyranoside as substrate at pH 7.5 (Maniatis et al., 1989)and protein assayed. The b -Gal activity in transduced cell sam-ples was determ ined by subtracting background b -Gal activityof untransduced cells, which was less than 1% of that of trans-duced cells. Retrovirus was purified by centrifugation (Seppenet al., 1997). Briefly, virus was collected and spun at 3000 3g, at 4°C for 20 hr. For the inhibition experiments, each platecontained 2 ml of LNFZ supernatant and 2 ml of fresh culturemedium or 2 ml of supernatant from 3T3 cells, or PA317 cellsor PA317 viral producer lines. These supernatants were col-lected for the same time and in the same cell-to-medium vol-ume ratio as the LNFZ supernatants. To remove virus, viral su-pernatants were centrifuged for 30 min at 15,000 3 g at roomtemperature. Titers for all virus batches were determined in par-allel with each transduction experiment. All viral infectionswere performed in duplicate, as were the b -Gal and protein de-terminations. The PA317 viral producer lines used for inhibi-

SEPPEN ET AL.772

FIG. 1. Number of integrations in 3T3 cells transduced withLNFZ at an MOI of 12. Chromosom al DNA isolated fromG418-resistant clones was digested with HindIII and analyzedby Southern blotting (Maniatis et al., 1989). Of 13 clones ana-lyzed, 8 have 1 integrated provirus and 5 have 2 integratedproviruses. The size of the molecular mass markers is indicatedin kilobases.

FIG. 2. Kinetics of retroviral infection at various MOIs. Relationship between specific b -Gal activity in 3T3 cells transducedwith LNFZ at various MOIs. Similar data were obtained when LNFZ, purified from most medium components by a mild, low-speed centrifugation procedure, was used. A representative experim ent using duplicate data points is shown.

tion and titer experiments all transferred LXSN-based viruses(Miller and Rosman, 1989) containing different cDNA inserts.

PCR protocols

Two primers, LXSN- c -F (TGGCCAGCAACTTATCT-GTG) and LXSN-c -R (GGGTGATGAGGTCTCGGTTA),were designed to bracket a 579-bp stretch, nucleotides830–1409, in the packaging region of LXSN (Miller and Ros-man, 1989). To construct a competitor, an 81-bp deletion (nu-cleotides 904–985) in the packaging region of LXSN was gen-erated by partial digestion with AatII, complete digestion withSpeI, blunting of the recessed ends with T4 DNA polym erase,and self-ligation. The LXSN-c -F and LXSN-c -R primersbracket 498 bp in the deleted vector. Viral RNA was isolatedby mixing 100 m l of viral supernatants with 1 ml of Tri reagent(Molecular Research Center, Cincinnati, OH) according to themanufacturer instructions. The isolated RNA was dissolved in11 m l of water and cDNA was generated with the LXSN-c -Rprimer and Superscript II reverse transcriptase (Life Technolo-gies, Gaithersburg, MD). For each PCR, 1 m l of cDNA wasused. The PCR was performed with 10 pmol of each primer,0.2 mM dNTPs, 2 units of Taq polymerase, 1 m l of cDNA, ina total volume of 50 m l according to the following program ;94°C for 10 sec, and 45 cycles of 94°C for 30 sec, 55°C for 15sec, and 72°C for 30 sec, followed by a 10-min extension at72°C. The viral vector RNA titer of LNFZ was determined withhalf-log increm ents of competitor; in the other cases log incre-ments were used. The PCR products were separated on a 1.5%agarose gel and stained with ethidium bromide.

RESULTS AND DISCUSSION

Southern analysis of chromosomal DNA from 3T3 cellstransduced with the LNFZ vector at an MOI of 12 showed thatof 13 clones analyzed, 8 had single and 5 had double integra-tion events (Fig. 1). As HindIII has only one recognition sitein the provirus, the size of the restriction fragment dependedon the random occurrence of a second HindIII site in the flank-ing chromosomal DNA. These DNA data were confirm ed atthe protein level by infecting 3T3 cells with LNFZ virus at dif-ferent MOIs. We determined specific b -Gal activity in thetransduced cells as a measure of integration number. Experi-ments were also performed with virus that had been purified by

a mild, slow-speed centrifugation (Seppen et al., 1997). Withpurified or crude virus preparations the maximum b -Gal activ-ity was obtained at relatively low MOIs, confirm ing that onlya few integration events per cell had occurred (Fig. 2). Similardata were obtained by counting the percentage of b -Gal-posi-tive cells after infection of 3T3 and primary rat smooth musclecells with LNFZ at different MOIs (data not shown).

These experiments suggested an inhibitory factor was pre-sent in retroviral preparations produced by PA317 packagingcells and that this factor copurified with infectious retrovirus.To confirm this, we infected 3T3 cells with LNFZ at MOIsgreater than 5 (which is saturating), in the presence of super-natants from 3T3 cells, PA317 cells, different PA317 virus pro-ducer cells, and centrifugation-dep leted viral supernatants. Onlythe presence of PA317 or PA317 viral supernatants significantlyinhibited infection of cells with the LNFZ virus (p , 0.05 asdetermined by ANOVA) and removal of virus by centrifuga-tion removed this inhibition (Table 1). The b -Gal expressionkinetics observed with purified or crude virus preparations aresimilar (Fig. 2). The maximum b -Gal activity was obtained atrelatively low MOIs, confirming that only a few integrationsper cell occur. The same results were obtained when a semi-quantitative assay was used, counting the percentage of b -Galpositive cells after infection of target cells with LNFZ at dif-ferent MOIs. These experiments were performed in 3T3 and

INHIBITORS OF RETROVIRAL TRANSDUCTION 773

TA BLE 1. INH IB ITIO N O F RETR O VIR AL INFEC TIO N OF 3T3 CELLS a

Relative b -Gal activityInhibiting supernatant (%)

LNFZ 1 medium 100LNFZ 1 3T3 supernatant 86 6 20b

LNFZ 1 PA317 supernatant 62 6 9b1LNFZ 1 viral supernatant 66 6 25b

LNFZ 1 centrifuged viral supernatant 88 6 21b

aThe results presented are the average of four independentexperiments. Only PA317 and viral supernatants inhibitedLNFZ transduction significantly.

bp , 0.05.

TABLE 2. CO RRELA TIO N OF BIO LO G IC AL A ND RNA TITER S

OF DIFFER EN T VIR AL PREP ARATIO NSa

Viral Colony-formin g Viral vector RNAproducer units/ml molecules/ ml

1 6 3 105 2 3 106

2 5 3 106 2 3 107

3 4 3 105 2 3 105

4 3 3 106 2 3 106

5 1 3 107 2 3 107

6 1 3 107 2 3 107

7 1 3 106 2 3 106

8 1 3 107 2 3 107

9 6 3 106 2 3 107

LNFZ 3 3 106 2 3 107

aTen different viral producers were titered by a colony-forming assay. cDNA was generated from viral RNA and theRNA titer was determined by competitive PCR.

FIG. 3. Competitive PCR to determine viral vector RNA titerof LNFZ virus. The amount of competitor DNA added is indi-cated above the lanes; –C, no competitor added; –T, no tem-plate added. The size of the full-length product from viral RNAis 579 bp, and the size of the competitor is 498 bp. In this ex-periment equal amounts of product are present with the addi-tion of 5 3 104 and 1 3 104 competitor molecules.

primary rat smooth muscle cells (results not shown). We ob-served no difference in titer after diluting virus in supernatantfrom 3T3 cells or fresh culture medium (data not shown), sug-gesting the absence of an inhibiting factor such as proteogly-cans (Le Doux et al., 1996) in 3T3 cells. Using competitivePCR we measured the amount of viral vector RNA present inviral preparations and compared this with the biological titer(Fig. 3). The biological and RNA titers of 10 different viralpreparations were in excellent agreement (Table 2), indicatingthe absence of large amounts of viral RNA containing inhibitor.A correlation between biological titer and viral RNA assay hasbeen shown previously (Morgan et al., 1990; Murdoch et al.,1997; Onodera et al., 1997; Quinn and Trevor, 1997).

As expected, retroviral infection was inhibited in the pres-ence of competing viral supernatant. Since the additional viruscan simply compete for cellular infection, this finding does notprove that inhibitors are present in viral supernatants. The in-hibiting properties of the competing retroviral preparation canbe removed by a centrifugation step that also removes infec-tious virus, suggesting that the inhibiting factor has propertiessimilar to those of infectious retrovirus. This complements ourfinding that a slow-speed centrifugation that concentrates in-fectious retrovirus concentrates the inhibiting factor as well. Su-pernatant of PA317 cells contains viral particles and our ex-periments confirm previous studies (Davis et al., 1997;MacNeill et al., 1999) showing that these particles can inhibitinfection.

Although estimates place the number of amphotropic virusreceptors at 8 3 104 per 3T3 cell, not all are involved in virusbinding (Battini et al., 1996). Even if a retrovirus uses multi-ple receptors for entry, the ratio of noninfectious particles ver-sus infectious particles must be high to account for the observedinhibition. Indeed, several studies have shown that only 0.1 to2% of murine retroviral particles in a preparation are infectious(Rein et al., 1978; Andersen and Nexo, 1983). The half-life ofa retrovirus at 37°C is estimated to be between 4 and 9 hr(Forestell et al., 1995; Kaptein et al., 1997) and our viral su-pernatants were collected for 24 hr. Therefore, although a pro-portion of the noninfectious particles may originate from inac-tivated virus, this may not account for the large amount ofnoninfectious particles that must be present to explain our ob-servations. A fraction of the noninfectious particles will con-tain endogenous retroelements such as VL30 (Hatzoglou et al.,1990; Chakraborty et al., 1994; Patience et al., 1998). As onlyone in seven viral particles in a murine retroviral preparationcontain VL30 sequences (Patience et al., 1998), it is unlikelythat viral particles containing retroelements are responsible forthe observed inhibition.

Retroviruses generated from PA317 packaging cells havebeen successfully used in numerous laboratory studies andmany clinical trials. An important property of more effectivepackaging lines would be the reduction of inhibitor secretionand this study suggests some experim ental approaches that canbe used to screen for packaging lines to meet these criteria.

ACKNOWLEDGMENTS

The research described in this article was supported by NIHGrants DK50686, DK53384, and DK43727.

REFERENCES

ANDERSEN, K.B., and NEXO, B.A. (1983). Entry of murine retro-virus into mouse fibroblasts. Virology 125, 85–98.

BATTINI, J.L., RODRIGUES, P., MULLER, R., DANOS, O., andHEARD, J.M. (1996). Receptor-binding properties of a purified frag-ment of the 4070A amphotropic murine leukemia virus envelope gly-coprotein. J. Virol. 70, 4387–4393.

CHAKRABORTY, A.K., ZINK, M.A., and HODGSON, C.P. (1994).Transmission of endogenous VL30 retrotransposons by helper cellsused in gene therapy. Cancer Gene Ther. 1, 113–118.

DAVIS, J.L., WITT, R.M., GROSS, P.R., HOKANSON, C.A., JUN-GLES, S., COHEN, L.K., DANOS, O., and SPRATT, S.K. (1997).Retroviral particles produced from a stable human-derived packag-ing cell line transduce target cells with very high efficiencies. Hum.Gene Ther. 8, 1459–1467.

FORESTELL, S.P., BOHNLEIN, E., and RIGG, R.J. (1995). Retrovi-ral end-point titer is not predictive of gene transfer efficiency: Im-plications for vector production. Gene Ther. 2, 723–730.

HATZOGLOU, M., HODGSON, C.P., MULARO, F., and HANSON,R.W. (1990). Efficient packaging of a specific VL30 retroelementby psi 2 cells which produce MoMLV recombinant retroviruses.Hum. Gene Ther. 1, 385–397.

KAHN, M.L., LEE, S.W., and DICHEK, D.A. (1992). Optimization ofretroviral vector-mediate d gene transfer into endothelial cells in vitro.Circ. Res. 71, 1508–1517.

KAPTEIN, L.C., GREIJER, A.E., VALERIO, D., and VANBEUSECHEM, V.W. (1997). Optimized conditions for the produc-tion of recombinant amphotropic retroviral vector preparations. GeneTher. 4, 172–176.

LE DOUX, J.M., MORGAN, J.R., SNOW, R.G., and YARMUSH,M.L. (1996). Proteoglycans secreted by packaging cell lines inhibitretrovirus infection. J. Virol. 70, 6468–6473.

MACNEILL, E.C., HANENBERG, H., POLLOK, K.E., VAN DERLOO, J.C., BIERHUIZEN, M.F., WAGEMAKER, G., and WIL-LIAMS, D.A. (1999). Simultaneous infection with retrovirusespseudotyped with different envelope proteins bypasses viral recep-tor interference associated with colocalization of gp70 and target cellson fibronectin CH-296. J. Virol. 73, 3960–3967.

MANIATIS, T., FRITSCH, E.F., and SAMBROOK, J. (1989). Mole-cular Cloning: A Laboratory Manual , 2nd Ed. (Cold Spring HarborLaboratory Press, Cold Spring Harbor, NY).

MILLER, A.D., and ROSMAN, G.J. (1989). Improved retroviral vec-tors for gene transfer and expression. BioTechniques 7, 980–990.

MORGAN, R.A., CORNETTA, K., and ANDERSON, W.F. (1990).Applications of the polymerase chain reaction in retroviral-mediatedgene transfer and the analysis of gene-marked human TIL cells. Hum.Gene Ther. 1, 135–149.

MURDOCH, B., PEREIRA, D.S., WU, X., DICK, J.E., and ELLIS, J.(1997). A rapid screening procedure for the identification of high-titer retrovirus packaging clones. Gene Ther. 4, 744–749.

ONODERA, M., YACHIE, A., NELSON, D.M., WELCHLIN, H.,MORGAN, R.A., and BLAESE, R.M. (1997). A simple and reliablemethod for screening retroviral producer clones without selectablemarkers. Hum. Gene Ther. 8, 1189–1194.

PATIENCE, C., TAKEUCHI, Y., COSSET, F.L., and WEISS, R.A.(1998). Packaging of endogenous retroviral sequences in retroviralvectors produced by murine and human packaging cells. J. Virol. 72,2671–2676.

QUINN, T.P., and TREVOR, K.T. (1997). Rapid quantitation of re-combinant retrovirus produced by packaging cell clones. BioTech-niques 23, 1038–1044.

RAMESH, N., KIM, S.-T., WEI, M.Q., KHALIGHI, M., and OS-BORNE, W.R.A. (1996). High-titer bicistronic vectors employingfoot-and-mouth disease virus internal ribosome entry site. NucleicAcids Res. 24, 2697–2700.

SEPPEN ET AL.774

REIN, A.L., GERWIN, B.I., BASSIN, R.H., SCHWARM, L., andSCHIDLOVSKY, G. (1978). A replication-defe ctive variant ofMoloney murine leukemia virus. I. Biological characterization . J. Vi-rol. 25, 146–156.

SEPPEN, J., KIMMEL, R.J., and OSBORNE, W.R.A. (1997). Serum-free production, concentration and purification of recombinant retro-viruses. BioTechniques 23, 788–790.

WALKER, P.S., HENGGE, U.R., UDEY, M.C., AKSENTIJEVICH,I., and VOGEL, J.C. (1996). Viral interference during simultaneoustransduction with two independent helper-free retroviral vectors.Hum. Gene Ther. 7, 1131–1138.

Address reprint requests to:Dr. William R.A. OsborneDepartment of Pediatrics

RR244, Box 356320University of Washington School of Medicine

Seattle, WA 98195

E-mail: wosborne@u. washington.edu

Received for publication October 20, 1999; accepted after re-vision December 27, 1999.

INHIBITORS OF RETROVIRAL TRANSDUCTION 775