Proc. Nati. Acad. Sci. USAVol. 88, pp. 1330-1334, February 1991Cell Biology

Genetically modified skin fibroblasts persist long aftertransplantation but gradually inactivate introduced genes

(gene therapy/retroviral vectors/adenosne deaminase/neomycin phosphotransferase/severe combined immunodeficiency)

THEO D. PALMER*, GuY J. ROSMAN*, WILLIAM R. A. OSBORNEt, AND A. DUSTY MILLER**Program in Molecular Medicine, Fred Hutchinson Cancer Research Center, Seattle, WA 98104; and tDepartment of Pediatrics, University of Washington,Seattle, WA 98195

Communicated by E. Donnall Thomas, November 14, 1990

ABSTRACT Genetically engineered fibroblasts have beensuccessfully used to produce therapeutic proteins in animals,but sustained production of the proteins has not been achieved.This limits the potential of fibroblast-mediated gene therapy inhumans. We have studied the phenomenon of decreased pro-duction in rats by using retroviral vectors carrying genesencoding human adenosine deaminase and neomycin phospho-transferase. While transplanted skin fibroblasts containingvector sequences persisted at constant levels for at least 8.5 mo,vector expression decreased by >1500-fold after 1 mo. Cellularor antibody-mediated immune responses were not detected intransplanted animals, and expression could not be restored infibroblasts recultivated from the grafts. This phenomenon isreminiscent of sequence-specific gene inactivation observed inother cell types. Because genetic manipulation and expressionof foreign proteins did not affect survival of the transplantedcells, effective long-term therapy may be possible with the useof alternative gene regulatory elements.

Although many somatic cell types are potential gene therapytargets for treatment of genetic or acquired disease (1), skinfibroblasts are attractive because they are easy to obtain andtransplant and can be rapidly grown to large numbers inculture. Immortalized fibroblasts make a variety of secretoryproducts after genetic modification and transplantation intoanimals (2-7), but these are ultimately not appropriate modelsfor gene therapy because these cell lines grow uncontrollablyand often form tumors in recipient animals. Primary fibro-blasts, however, could be used for human therapy. In ani-mals, primary embryo or skin fibroblasts produce clottingfactor IX at systemic levels that approach therapeutic utility(5, 6), but production of the protein for >1 mo has not beenachieved. The gradual decrease in protein levels might becaused by immune response against the foreign protein, poorsurvival of the transplanted cells, or inactivation of thetransferred genes.To address these issues, we have studied the transfer of a

human adenosine deaminase (hADA)-encoding gene intoskin fibroblasts of inbred rats. By using hADA in place offactor IX, we could avoid several complicating problemsassociated with production of a foreign secreted protein:hADA should not be immunogenic because it is an intracel-lular protein; hADA is localized at the site of the graft, anddetection is not dependent on systemic distribution; andfinally, sensitive assays make even low-level hADA readilydetectable above endogenous rat adenosine deaminase(ADA). By careful examination of transplanted tissues, weshow here that expression of hADA had no effect on cellsurvival, but that transplanted cells gradually inactivateretrovirally transferred genes.

MATERIALS AND METHODSPrimary Cell Culture. Primary skin fibroblasts were iso-

lated by standard methods (8) from either human foreskin orforearm biopsies, or from inbred Fischer 344 rats. Cells weregrown at 370C in a 10%o CO2 atmosphere with Dulbecco'smodified Eagle's medium (DMEM) containing 5 g of glucoseper liter, 10%o (human cells) or 15% (rat cells) fetal bovineserum, and amphotericin B, penicillin, and streptomycinantibiotics. To avoid potential complications of in vitro aging,human and rat cell cultures were used soon after establish-ment of the primary culture. In general, human fibroblastswere used within 3 to 20 population doublings of the primaryculture, and rat fibroblasts were used within 3 to 5 populationdoublings.

Vectors. Except for pLASAN, all vectors have been de-scribed (9). To construct pLASAN, a deletion was introducedinto the gene for neomycin phosphotransferase (neo) ofpLASN by removing 400 base pairs (bp) between Fsp I andRsr I. High-titer helper-free virus stocks were produced withPA317 amphotropic packaging cells (10), as described (11).Virus titers measured on NIH 3T3 cells ranged from 106 to 8x 106 colony-forming units (cfu) per ml. As noted in previousstudies, the titers were generally 10-fold lower on normaldiploid fibroblast strains from either humans or rats (6, 9, 12,13).

Transplantation. All fibroblast transplants were performedbetween inbred Fischer 344 rats cared for in accordance withinstitutional guidelines. Skin fibroblasts were isolated, cul-tured, and cast in collagen matrices as described (6). Fordermal equivalent transplants, the matrix was placed in acircular full-thickness skin wound and protected with a burndressing consisting of a collagen/condroitin sulfate matrixbound to a silicone backing (Integra, Marion Merrell Dow)similar to that described by Yannas et al. (14). The dressingwas glued at the edges to the surrounding skin with cy-anoacrylate glue (Nexaband Liquid, CRX Medical, Raleigh,NC). The graft was covered with nonadherent dressing, andthe rat was wrapped with an elastic bandage. Subcutaneousampicillin was given 1 day before transplantation and con-tinued for 10 days.

Polymerase Chain Reaction (PCR). Genomic DNA (250 ng)was amplified in a total reaction volume of 55 ILI (0.01%gelatin/50 mM KCI/10 mM Tris, pH 8.5/2 mM MgCl2/0.1mM each of dNTP/0.1 AM [a-32P]dTT1P or (a-32P]dCTP/1.5units of Taq polymerase/125 ng each of Neol and Neo5primers). Primer sequences were as follows; Neol, 5'-CAAGATGGATTGCACGCAGG-3'; and Neo5, 5'-CCCGC-TCAGAAGAACTCGTC-3', respectively. In addition, 5 fg ofplasmid pLASAN (Fig. 1) was added to all samples as an

Abbreviations: neo, neomycin phosphotransferase; ADA, adenosinedeaminase; hADA, human ADA; LTR, long terminal repeat; SV40,simian virus 40; cfu, colony-forming units; PNP, purine nucleosidephosphorylase; MoMLV, Moloney murine leukemia virus; PCR,polymerase chain reaction.

1330

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Dow

nloa

ded

by g

uest

on

Oct

ober

22,

202

0

Proc. Natl. Acad. Sci. USA 88 (1991) 1331

internal control. To guarantee that each sample received anequal amount of pLASAN, an aliquot of plasmid was addedto a complete reaction mixture, and then the mixture wasdivided equally between DNA samples. Amplification con-sisted of 25-30 cycles of 950C for 2 min followed by 720C for5 min. The reaction products were resolved in 5% nondena-turing acrylamide gels, which were then dried onto filterpaper and exposed to film at -700C. Two PCR products wereexpected, a full-length neo band (neo) from the integratedLASN vector and a shorter band (Aneo) from pLASAN.Band intensities (arbitrary units) were determined by com-puter analysis of a digitized video image (Biosystems imageanalysis system). Briefly, bands were defined as inclusivegroups of pixels with optical densities above background. Anindividual band intensity is the sum optical density for allpixels within a band. To avoid problems with film saturationor reciprocity failure, several exposures of each gel wereanalyzed. The intensity of the Aneo sequence amplified frompLASAN was used to correct for differences in amplificationefficiency between samples (see Fig. 4).

Cytotoxic Lymphocyte Assays. Lymphocytes from proxi-mal lymph nodes were used to lyse 51Cr-labeled fibroblasts.Target fibroblasts were loaded with 51Cr by mixing an equalvolume of 5"Cr (5 mCi per ml in saline; 1 Ci = 37 GBq) witha single-cell suspension of fibroblasts (107 cells per ml) inDMEM/20% fetal bovine serum. The mixture was incubatedfor 1 hr at 370C. The cells were then washed three times andresuspended in RPMI 1640/10%o heat-inactivated fetal bovineserum. Specific cell lysis was measured by mixing 4 x 103fibroblasts with different numbers of lymphocytes in V-bot-tom 96-well dishes. The cell mixtures were pelleted andincubated at 370C for 4 hr. Spontaneous release of 51Cr wasmeasured on fibroblasts alone, and 1% Triton X-100 lysateswere used to measure maximal release. Specific lysis wascalculated as follows: % specific lysis = (experimental cpm- spontaneous cpm)/(maximum cpm - spontaneous cpm) x100.

RESULTSExpression of hADA in Cultured Fibroblasts. Several ret-

roviral vectors encoding hADA (Fig. 1) were tested forhADA expression in primary diploid fibroblasts from humansand rats and in NIH 3T3 immortalized mouse fibroblasts(Table 1). Primary isolates of skin fibroblasts or secondarypassages of a primary isolate were infected with limitingamounts of virus to ensure single-copy integration and werethen treated with G418. Bulk populations of G418-resistantcells were then assayed for hADA. All three retroviralvectors produced significant amounts of human ADA, butLASN and LNCA were most efficient in human fibroblasts.

LNSAr ADA

A B

FIG. 2. Outer (A) and inner (B) surfaces of a dermal equivalenttissue at 8.5 mo. (Magnification is the same in both images.)

Enzyme activities in human fibroblasts infected with LNCAvector were so high that only -5 x 107 cells (a numberfeasible to transplant) would provide enough ADA to correctthe biochemical defect in humans suffering from ADA defi-ciency (see ref. 13 for method of estimation). To examinelong-term production of hADA in an animal model, theLASN vector, which was the most active vector in ratfibroblasts, was used to introduce hADA into primary skinfibroblasts of inbred Fischer 344 rats.hADA Production and Cell Survival After Transplantation.

Genetically modified primary rat skin fibroblasts were cast incollagen matrices and used as dermal equivalent grafts. Thegrafts were easily recognizable for over 8 mo, having ahairless upper surface (Fig. 2A) and a vascularized but palelower surface (Fig. 2B). The grafted tissues were removed atvarious times after transplantation and cut into pieces forADA and vector DNA analysis, and for reinitiation of fibro-blast cultures. The pattern of hADA production was similarto that reported for factor IX (6); while hADA was clearlydetected early, activities dropped to undetectable levels at=1 mo (Fig. 3, Table 2). From the limit of detection forhADA(=1% of endogenous rat ADA) and the percentage of donorcells in the tissue sample at 1 mo (30%, Table 2), we estimatethe drop from 50-fold above endogenous ADA levels (Table1) to <1% of endogenous levels represents >1500-fold re-duction in hADA levels.The decrease in hADA levels directly contrasts with the

continued presence of vectorDNA in the transplanted tissue.PCR analysis of DNA extracted from tissues shows vectorsequences to persist at undiminished levels for at least 8.5 mo(Fig. 4, Table 2). Although PCR analysis is inherently vari-able, appropriate internal controls and replicate assays (Fig.4) allow rough quantitation of vector sequences. For exam-

TimeO 2Days 2

RatADA

(A)nIS- M HLTR

9v+ neo A ADA (A)nLNCA

AD LRIADA e- neo (A)n

H _I~~~~Isv I&\\~f\\\\: H TRI

ADA r--Aneo vSVy LTR

HumanADA

I-r pBR322

1 kb

FIG. 1. ADA vectors. Retroviral vectors LNSA, LNCA, andLASN use three different promoters to express ADA; the simianvirus 40 (SV40) early promoter (SV), the cytomegalovirus immediateearly promoter (CMV), and the Moloney murine leukemia virus(MoMLV) promoter in the long terminal repeat (LTR). pLASAN wasused as a plasmid control for PCR reactions. Arrows, transcriptioninitiation sites; (A)n, polyadenylylation sites; O', packaging signal.

H C C C D D D S S D D S S D D D S S S

FIG. 3. Human and rat ADA in transplanted tissues. Equalamounts of total ADA were analyzed by starch gel electrophoresis asdescribed (16). Although not shown, loading of equal amounts oftotal protein showed rat ADA to remain relatively constant in rat cellsamples. Lanes: H, human T-cell sample; C, cultured cells beforetransplantation; D, dermal equivalent transplants; S, subcutaneoustransplants of fibroblast-containing collagen matrices. The subcuta-neous transplants behaved like the dermal equivalent transplants butwere much more difficult to localize at long times after transplanta-tion and are not further discussed.

-N +LASN [}

pLASAN }

Cell Biology: Palmer et al.

I

Dow

nloa

ded

by g

uest

on

Oct

ober

22,

202

0

Proc. Natl. Acad. Sci. USA 88 (1991)

Table 1. ADA activity in cultured fibroblasts

ADA activity, gimol per hr per mg of protein

No LNSA LNCA LASNCells vector vector vector vector

Normal human fibroblasts 0.9 ± 0.4 8.2 + 2.1 120 + 55 80 ± 45ADA- human fibroblasts <0.01 6.2 ± 2.6 120 ± 20 80 ± 40Rat fibroblasts 0.9 ± 0.1 6.6 ± 2.3 4.9 ± 0.2 50 ± 25NIH 3T3 Cells 2.2 ± 0.2 4.8 ± 1.3 5.9 + 1.8 9.5 ± 4.7

Cells were infected with the indicated vectors and selected for G418 resistance. ADA was determinedas described (15) for polyclonal (>105) populations, and the values given are means + SDs for 2-12separate primary strains.

ple, three transplants harvested at 8.5 mo had mean ± SDvalues (n = 3 or 4) of 61 ± 40, 63 ± 33, and 55 ± 28%vector-positive cells. This result yields an average value of60% vector-positive cells at 8.5 mo (Table 2). Clearly, manycells contain vector sequences after 8.5 mo. Thus vector-infected cells persisted long-term, but expression of thetransferred hADA gene was severely decreased.Immune Responses to Transplanted Fibroblasts. Immune

responses against vector-expressing fibroblasts might haveselected for rare nonexpressing cells. To test this, five ratsreceived dermal equivalent transplants of LASN-expressingcells. One month later the grafts were removed and shown tocontain but not express vector sequences. To boost potentialimmune responses, the animals received s.c. injections of 107additional cells when the grafts were removed. Ten days laterlymphocytes from lymph nodes, spleens, and peripheralblood were assayed for specific lysis of LASN-expressingtarget fibroblasts. Plasmid from the rats was also tested forantibodies against the LASN-expressing fibroblasts by add-ing plasma and purified complement to the fibroblasts andscoring cell lysis by trypan blue dye exclusion. There was no

increase in antibody-mediated cell lysis after transplantation(data not shown) or in cell lysis by lymphocytes fromtransplanted animals compared with that obtained by usingnontransplanted animals (Fig. 5). In contrast, control animalsinjected with human fibroblasts generated strong cytolyticresponses against the human cells.

Control DNA DNA from DE TissueVector Copies per Genome at 8.5 Months

1.0 0.1 0.01 0.001 FRi FR2 FR3 FRI FR2

neo

A neo

3.

Band neo 71.6 17.3 1.2 0.6 15.1 15.3 31.2 15.3 22.2

Intensity Aneo 6.1 7.6 5.5 4.0 0.9 1.1 2.3 2.3 2.7

neo/Aneo 11.7 2.3 0.22 0.15 16.8 13.9 13.6 6.7 8.2

% vector positive 82 13 1 0.6 120 100 95 45 55

FIG. 4. PCR analysis of transplanted cells. PCR was used ratherthan Southern analysis because of the limited amount of DNArecovered from each tissue sample. (Left) Amplified products fromcontrol DNA containing known amounts of integrated LASN vector.(Right) Amplified products from three dermal equivalent (DE) tissuesexplanted at 8.5 mo (replicate assays are shown for transplants FR1and FR2). Each sample contains the same amount (5 fg) to pLASANplasmid, which produces a shorter neo-specific band (Aneo) thanproduced from integrated vector sequences (neo). Ratios of neolAneo band intensities for control DNAs were plotted against vectorcopies per genome on logarithmic graph paper; an estimated best-fitline was drawn through plotted points. Control and experimentalneo/Aneo values were then converted to% vector containing cells byusing this plot. Although background in this assay falls between 1%and 0.1%, sensitivity can be enhanced by increasing number ofamplification cycles.

Although strong immune responses were not detected,low-level selection against cells expressing hADA or neo

could have caused the gradual decrease in expression. Todetermine whether vector expression inhibited the survival oftransplanted cells, naive or vector-expressing fibroblastsfrom female rats were transplanted onto male hosts. At 2 moafter transplant, cultures were initiated from the graftedtissues, and metaphase cells from each culture were exam-ined for female or male karyotype (17). The average fractionof female cells (XX karyotypes/total karyotypes) from du-plicate transplants were 0.41 ± 0.10 for naive cells and 0.34± 0.06 for vector-expressing cells. Twenty-three to 41 meta-phase figures were karyotyped from each culture, and thevalues given are the means ± SDs for two transplants of eachcell type. This result shows that genetic manipulation andexpression of vector-encoded proteins has no significanteffect on the survival of the transplanted fibroblasts andsuggests that vector down-regulation was the major cause ofdecreased hADA expression.

Vector Expression in Cells Cultured from Skin Grafts.Fibroblasts were easily cultured from the dermal equivalentgrafts and were found positive for vector DNA, although atsomewhat lower levels than seen in the graft (Table 2). Thepresence of vector sequences in culture shows that vector-containing cells in the grafts were viable. Expression ofhADA in the cultured fibroblasts decreased as a function oftime that the fibroblasts had been in animals (Table 2), similarto the decrease seen in grafts. We also measured ability of thecultured fibroblasts to form colonies in G418 as a measure ofSV40 promoter-directed neo gene expression (Table 2). Cul-tures initiated from grafts that were removed soon aftertransplantation showed a high proportion of G418-resistantcells, whereas those removed at 8.5 mo after transplantationshowed no G418-resistant cells (<0.001%). These resultsshow that decreased expression was not from the quiescent(i.e., nonreplicative) state of the fibroblasts in vivo because

50scnCa)% 40-

-J

C) 30~0 20-

0) 200O1 0-

--

10 100 10 100

Ratio of Lymphocytes to Target Cells

FIG. 5. Cytotoxic T lymphocyte assays. Lymph node lympho-cytes from transplanted animals were mixed with '1Cr-loaded fibro-blasts, and 51Cr release was used to measure specific cytolyticactivity. (A) Lymphocytes were harvested from animals transplantedwith target cells. (B) Lytic activity of lymphocytes from nontrans-planted animals on same targets. o, Human fibroblasts; M. LASN-infected rat fibroblasts; A, uninfected rat fibroblasts.

A Immunized B Naive

1..f 4

1332 Cell Biology: Palmer et al.

1

Dow

nloa

ded

by g

uest

on

Oct

ober

22,

202

0

Proc. Natl. Acad. Sci. USA 88 (1991) 1333

Table 2. ADA vector persistence and expression in transplanted rat fibroblasts

Transplanted tissue Cells cultured from tissue

Cells Vector Vector G418rtransplanted Time positive, % ADA/PNP positive, % ADA/PNP colonies, %

LASN infected 0 100 3.6 602 days 40 1.9 30 3.2 462 weeks 30 0.34 13 0.87 81 mo 30 0.23 9 0.13 0.038.5 mo 60 0.13 6 0.14 <0.001

VectorDNA was analyzed by PCR. ADA and purine nucleoside phosphorylase (PNP) activities weredetermined as described (15). ADA activities were normalized to PNP activities to correct for variabilityin tissue protein extraction. ADA/PNP in normal rat skin is 0.17 ± 0.04, and in cultured rat fibroblaststhe ratio is 0.17 ± 0.03. ADA/PNP in uninfected control transplants was indistinguishable from thoseof normal skin or cultured cells. Percentage of G418r colonies = (colonies that grew in G418)/(coloniesthat grew in the absence of G418) x 100. Plating efficiency of G418r cells in G418 is 20-70% lower thanthat without G418; thus, even a cell population grown in G418 does not give 100lo G418F colonies (e.g.,day 0). Values are the means of two or three separate explants for each period.

growth of the vector-infected cells in culture did not restorevector expression and suggest that suppression was causedby stable inactivation ofvector sequences. Both the MoMLVpromoter and the SV40 early promoter were subject tosuppression in vivo.Vector Expression in Continually Cultured Cells. Expres-

sion of the LASN vector was monitored in cultured cells thatwere not transplanted. LASN-infected, G418-selected ratfibroblasts were cultured in a replicating state by sequentialsubculture without G418. In parallel, a mixture of infectedand uninfected cells was passaged as a confluent monolayerto mimic the quiescent state oftransplanted cells in vivo. Thepercentage of G418-resistant cells was monitored periodi-cally as a measure of vector expression. In both replicatingand quiescent cultures, the proportion of G418-resistant cellsvaried no more than 3-fold over 2 mo (data not shown). Thiscontrasts with the >1000-fold decrease in G418 resistanceseen after 1 mo in animals (Table 2). Thus, suppression ofexpression from the transduced genes was only seen intransplanted cells.

Vector RNA in Transplanted Cells. To determine whetherdown-regulation of hADA and neo were mediated at a

transcriptional or translational level, RNA was isolated fromcells before transplantation and from cultured cells recoveredfrom grafts that had been left on animals for 1 mo (Fig. 6).Although the cultures established from grafts contained be-tween 4% and 11% vector-positive cells (Table 3), neitherLTR- nor SV4O-promoted transcripts were detectable. Al-

+ - A B

28S--

18S-

c

H1 -

FIG. 6. RNA from transplanted cells. Tissue was removed fromthree animals 1 mo after transplant, and cellular RNA was preparedfrom cells cultured from grafts. Ten micrograms ofRNA was run ineach lane of an agarose gel and then transferred to nylon filters.Filters were hybridized with 32P-labeled neo probe and then rehy-bridized with histone Hi° probe to show equal RNA loading in eachlane. Full-length LTR messages migratejust below the 28S ribosomalband, whereas the shorter SV40 message migrates at -18S. Lanes:A, B, and C, RNA from cells grown from grafted tissue; +, RNAfrom hADA-expressing nontransplanted fibroblasts; -, RNA fromuninfected fibroblasts.

though decreased message stability could explain this result(i.e., cells recovered from grafted tissue are altered in a waythat renders vector messages differentially unstable), it ismore likely that message accumulation is decreased bytranscriptional mechanisms.

Reactivation of Vector Sequences. DNA methylation cansuppress provirus expression, an effect that can be partiallyreversed by growth of cells in 5-azacytidine (18, 19). In someinstances, proviruses can also be reactivated to a limitedextent by using 5-bromodeoxyuridine (20). To determinewhether similar mechanisms were responsible for vectorinactivation in fibroblasts, cells cultured from dermal equiv-alent grafts were treated with 5-azacytidine (3 ,uM, 48 hr) orwith 5-bromodeoxyuridine (20 ,ug/ml, 48 hr). Activation wasmonitored by determining the number of cells able to formcolonies in G418 (Table 3). One of the three cultures (B)shows a slight increase in G418-resistant colonies after treat-ment with 5-azacytidine similar to the levels of reactivationseen in previous work (18, 20)-i.e., 5-60 reactivated cellsper 106 treated cells. Although 5-azacytidine had no measur-able effect in the other two cultures, the measurements aretoo close to the limits of detection to show a definitive lackof response. Treatment of cells with 5-bromodeoxyuridinehad no detectable effect in any cultures isolated from grafts.Two of the cultures (A and B) exhibited small increases in

G418-resistant colony formation after treatment with sodiumbutyrate (5 mM, 48 hr). Sodium butyrate induces hyperacety-lation of histones, and the resulting increase in expressionfrom some genes is thought to occur through modification ofchromatin structure (21). This result, along with the smallamount of reactivation seen with 5-azacytidine, suggests thatde novo methylation and chromatin restructuring may playsmall roles in vector inactivation, but the inability of thesedrugs to reactivate the vector sequences in most cells sug-gests that the major cause of inactivation remains unex-plained.

DISCUSSIONThese studies show profound suppression oftransduced geneexpression in transplanted fibroblasts. One month after trans-plantation, vector-encoded hADA expression was down>1500-fold. The effect was not seen during long-term cultureof the cells in vitro, and the suppression was not reversed byrecultivation of the transplanted cells. In contrast, the en-dogenous rat ADA gene was active continuously in thetransplanted cells (Fig. 3). The vector-infected cells usedwere polyclonal (>105 infected clones) populations, and nineseparately infected populations of cells were used, suggestingthat suppression was independent of the virus integrationsite. A lack of measurable immune responses to transplanted

Cell Biology: Palmer et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

22,

202

0

Proc. Natl. Acad. Sci. USA 88 (1991)

Table 3. G418 resistance of cells isolated from grafts after treatment with 5-azacytidine,bromodeoxyuridine, or sodium butyrate

Vector- G418r colonies per 106 total coloniespositive

Culture cells, % Untreated 5-AzaC BrdUrd Butyrate

A 11 0.7 <4.0 <4.2 3.9B 7 0.6 36 <4.7 17C 4 <1.1 <17 <36 <3.7- <0.01 <0.7 <13 <19 <4.7+ 100 3.2 x 105 2.2 x 105 4.9 x 105 5.7 x 105Cells cultured from grafts that had remained on animals for 1 mo (A, B, C) were treated with the

indicated chemicals and assayed for colony formation with and without G418. -, Uninfectedfibroblasts; +, infected G418-resistant fibroblasts; numbers preceded by < indicate the sensitivity ofthe assay when no G418-resistant colonies were scored. BrdUrd, bromodeoxyuridine; 5-AzaC,5-azacytidine.

cells and the high proportion of donor cells in the graft at alltimes suggests that selection for rare nonexpressing cells wasunlikely. This was also supported by sex-mismatched trans-plants, where the proportion of female cells isolated frommale hosts was the same regardless of whether or not thedonor cells expressed hADA at the time of transplant.Although there were no overt signs of immune response to

the grafted tissue, histological analysis showed minor infil-tration consistent with normal wound healing. Possibly, theprocess of wound healing produces inhibitory cytokines thatcontribute to decreased expression. Regardless of the actualmechanism involved, the absence of selection against vector-expressing cells suggests that suppression is a result of generegulation and not of cell death.The LASN vector has also been used to infect mouse

hematopoietic stem cells (22, 23). Although evidence existsfor inactivation in lymphoid cells, hADA was clearly presentin blood of most transplanted mice for >5 mo, in some miceat levels equivalent to those ofendogenous mouse ADA. Thisresult demonstrates that the LASN vector can achieve long-term in vivo expression in some instances and that suppres-sion occurs via a mechanism specific for certain cell types,including fibroblasts.

Perhaps regulatory elements within the LASN vector canbe modified to provide expression in fibroblasts. This ap-proach has been successfully used in murine embryonalcarcinoma cells and embryo stem cells where MoMLVexpression is strongly suppressed (18-20). MoMLV regula-tory sequences are not functional in these cell types (24), andmutations within the MoMLV enhancer and adjacent se-quences have been used to map the sequences that limitexpression (25-28). With this information, several new vec-tors have been derived from MoMLV that are expressed inembryonal cells. Because vectors that exhibit suppression ofexpression in vivo are all based on MoMLV (5, 6), similarmanipulation of the MoMLV enhancer may yield vectorscapable of long-term expression in transplanted fibroblasts.These experiments point to significant problems that must besolved before gene therapy with fibroblasts can be attemptedand provide an interesting model for studying regulation ofgene expression in vivo.

Sequence information for Neol and Neo5 primers was kindlyprovided by Richard Morgan. We thank Eileen Bryant and the FredHutchinson Cancer Research Center Department ofCytogenetics forhelp with cytogenetic analysis. This work was supported by GrantsAG00057, HL41212, A119565, and DK38531 awarded by the Na-tional Institutes of Health.

1. Miller, A. D. (1990) Blood 76, 271-278.

2. Garver, R. I., Jr., Chytil, A., Courtney, M. & Crystal, R. G.(1987) Science 237, 762-764.

3. Selden, R. F., Skoskiewicz, M. J., Russell, P. S. & Goodman,H. M. (1987) N. Engl. J. Med. 317, 1067-1076.

4. Rosenberg, M. B., Friedmann, T., Robertson, R. C., Tuszyn-ski, M., Wolff, J. A., Breakefield, X. 0. & Gage, F. H. (1988)Science 242, 1575-1578.

5. St. Louis, D. & Verma, I. M. (1988) Proc. Nat!. Acad. Sci.USA 85, 3150-3154.

6. Palmer, T. D., Thompson, A. R. & Miller, A. D. (1989) Blood73, 438-445.

7. Wolff, J. A., Fischer, L. J., Xu, L., Jinnah, H. A., Langlais,P. J., luvone, P. M., O'Malley, K. L., Rosenberg, M. B.,Shimohama, S. & Friedmann, T. (1989) Proc. Nat!. Acad. Sci.USA 86, 9011-9014.

8. Ham, R. G. (1980) Methods Cell Biol. 21a, 255-276.9. Hock, R. A., Miller, A. D. & Osborne, W. R. A. (1989) Blood

74, 876-881.10. Miller, A. D. & Buttimore, C. (1986) Mol. Cell. Biol. 6,

2895-2902.11. Miller, A. D., Trauber, D. R. & Buttimore, C. (1986) Somatic

Cell Mol. Genet. 12, 175-183.12. Osborne, W. R. A. & Miller, A. D. (1988) Proc. Nat!. Acad.

Sci. USA 85, 6851-6855.13. Palmer, T. D., Hock, R. A., Osborne, W. R. A. & Miller,

A. D. (1987) Proc. Nat!. Acad. Sci. USA 84, 1055-1059.14. Yannas, I. V., Burke, J. F., Orgill, D.-P. & Skrabut, E. M.

(1982) Science 215, 174-176.15. Osborne, W. R. A. (1980) J. Biol. Chem. 255, 7089-7092.16. Osborne, W. R. A. & Spencer, N. (1973) Biochem. J. 133,

117-123.17. Zieverink, W. D. & Moloney, W. C. (1965) Proc. Soc. Exp.

Biol. Med. 119, 370-376.18. Stewart, C. L., Stuhlmann, H., Jahner, D. & Jaenisch, R.

(1982) Proc. Natl. Acad. Sci. USA 79, 4098-4102.19. Jahner, D., Sthulmann, H., Stewart, C. L., Harbers, K.,

Lohler, J., Simon, I. & Jaenisch, R. (1982) Nature (London)298, 623-628.

20. Niwa, O., Yokota, Y., Ishida, H. & Sugahara, T. (1983) Cell 32,1105-1113.

21. Candido, E. P. M., Reeves, R. & Davie, J. R. (1978) Cell 14,105-113.

22. Kaleko, M., Garcia, J. V., Osborne, W. R. A. & Miller, A. D.(1990) Blood 75, 1733-1741.

23. Osborne, W. R. A., Hock, R. A., Kaleko, M. & Miller, A. D.(1990) Hum. Gene Ther. 1, 31-41.

24. Linney, E., Davis, B., Overhauser, J., Chao, E. & Fan, H.(1984) Nature (London) 308, 470-472.

25. Franz, T., Hilberg, F., Seliger, B., Stocking, C. & Ostertag, W.(1986) Proc. Nat!. Acad. Sci. USA 83, 3292-32%.

26. Barklis, E., Mulligan, R. C. & Jaenisch, R. (1986) Cell 47,391-399.

27. Hilberg, F., Stocking, C., Ostertag, W. & Gretz, M. (1987)Proc. Nat!. Acad. Sci. USA 84, 5232-5236.

28. Weiher, H., Barklis, E., Ostertag, W. & Jaenisch, R. (1987) J.Virol. 61, 2742-2746.

1334 Cell Biology: Palmer et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

22,

202

0