Proc. Natl. Acad. Sci. USAVol. 93, pp. 10371-10376, September 1996Medical Sciences

Clonal analysis of stably transduced human epidermal stem cellsin culture

(keratinocyte/gene therapy/epidermis)

MONICA B. MATHOR*t, GIULIANA FERRARIt, ELENA DELLAMBRA*t, MICHELE CILLI*, FuLvIo MAVILIOt,RANIERI CANCEDDA*, AND MICHELE DE LUCA*t§*Istituto Nazionale per la Ricerca sul Cancro and Centro di Biotecnologie Avanzate, 16100 Genoa, Italy; tIDI, Istituto Dermopatico dell'Immacolata, 00100Rome, Italy; and tThe Telethon Gene Therapy Program, DIBIT-H.S. Raffaele, 20100 Milan, Italy

Communicated by Howard Green, Harvard Medical School, Boston, MA, June 12 1996 (received for review January 20, 1996)

ABSTRACT We have transduced normal human kerati-nocytes with retroviral constructs expressing a bacterial.8-galactosidase (,B-gal) gene or a human interleukin-6 (hIL-6)cDNA under control of a long terminal repeat. Efficiency ofgene transfer averaged approximately 50%Yo and 95% of clono-genic keratinocytes for ,B-gal and hIL-6, respectively. Bothgenes were stably integrated and expressed for more than 150generations. Clonal analysis showed that both holoclones andtheir transient amplifying progeny expressed the transgenepermanently. Southern blot analysis on isolated clonesshowed that many keratinocyte stem cells integrated multipleproviral copies in their genome and that the synthesis of theexogenous gene product in vitro was proportional to thenumber of proviral integrations. When cohesive epidermalsheets prepared from stem cells transduced with hIL-6 weregrafted on athymic animals, the serum levels of hIL-6 werestrictly proportional to the rate of secretion in vitro andtherefore to the number of proviral integrations. The possi-bility of specifying the level of transgene expression and itspermanence in a homogeneous clone of stem cell origin opensnew perspectives in the long-term treatment of genetic disor-ders.

Cultured human keratinocytes generate cohesive sheets ofepithelium (1), which maintain the characteristics of theoriginal donor site (2, 3) and retain stem cells, namely cells thatupon division replace their own number and also give rise tocells that differentiate further into one or more specializedtypes (4-6). Cultured epithelial sheets are routinely used tomake autologous grafts for patients suffering from large skinand mucosal defects; the engrafted stem cells assure thepersistence of the regenerated epidermis during the patient'slifetime (7-9). Epidermal keratinocytes synthesize and secreteseveral gene products that might accomplish autocrine, para-crine, and even endocrine functions (10). Indeed, it has beenunambiguously demonstrated that keratinocyte-derivedpolypeptides cross the basement membrane and are secretedinto the bloodstream (11, 12). Basal keratinocytes are alsosources of adhesion molecules and matrix proteins, which formthe anchoring structures of the epidermis, and some of theseproteins are affected in genetic skin diseases (13). Therefore,epidermal cells are very attractive targets for the gene therapyof skin diseases, as well as for systemic delivery of recombinantproteins for the treatment of a number of genetic disorders.Ex vivo transduced mammalian epidermal cells, including

human keratinocytes, have been successfully generated (14-16). Protein production in vitro and/or secretion into thecirculation by keratinocytes grafted onto animals has beenshown for exogenous growth hormone (17-20), neomycinphosphotransferase (21), steroid sulfatase (22), factor IX (23),

al-anti-trypsin (24), 13-chorionic gonadotropin (15), and ,B-ga-lactosidase (13-gal) (19, 24-26) after transduction with repli-cation-deficient retroviral or adenoviral vectors. However,none of these approaches led to permanent transgene expres-sion in normal human keratinocytes either in vitro or in vivo(14). Because grafts derived from a transduced human epi-dermal keratinocyte cell line were able to sustain secretion ofthe transgene product for a relatively long period in vivo (18),the most likely explanation for the short-term transgeneactivity in normal human keratinocytes, either in culture orafter transplantation on athymic animals, is the inefficienttransduction of a stem cell population able to sustain thelong-term turnover of the transduced cells.Using clonal analysis, we show here that (i) the cells with the

greatest growth potential in keratinocyte cultures can be stablytransduced by retroviral constructs, (ii) the transgene expres-sion lasts for the entire life-span of the culture (more than 150cell generations), (iii) keratinocyte stem cells integrate eitherone or multiple proviral copies in their genome, (iv) thesynthesis of the exogenous protein is proportional to thenumber of proviral integrations, and (v) when grafted onathymic animals, epidermal sheets generated from transducedclones secrete the exogenous protein over the short term in theblood circulation at levels strictly proportional to the rate ofsecretion in vitro and therefore to the number of proviralintegrations.

MATERIALS AND METHODSCell Culture. Normal human keratinocytes were obtained

from healthy donors and cultivated on a lethally irradiatedfeeder-layer of 3T3-J2 cells (1) in Dulbecco-Vogt Eagle'smedium and Ham's F-12 media (3:1 mixture). Supplementsand cell passages were as described in ref. 27. For infectionexperiments only secondary keratinocytes were used.

Swiss mouse 3T3-J2 cells (1), GP+E-86 ecotropic packagingcells (28), and GP+env Aml2 amphotropic packaging cells(29) were grown in Dulbecco-Vogt Eagle's medium supple-mented with 10% calf serum (3T3-J2) or 10% fetal calf serum(GP+E-86 and GP+env Aml2).

Generation of the High-Titer LBSN and LIL6SN RetroviralVectors. LBSN and LIL6SN were constructed by cloning,respectively, a 3.5-kb fragment containing the full-length Esch-erichia coli 13-gal cDNA and a 706-bp human IL-6 EcoRI/BanIfragment into the EcoRI/HpaI sites of LXSN retroviral vector(30), as described (31). The Am12/LBSN and Aml2/LIL6SNproducer cell lines were generated by the transinfection pro-tocol (see also ref. 32). Briefly, plasmid DNA was transfected

Abbreviations: ,3-gal, ,B-galactosidase; IL-6, interleukin 6; hIL-6, hu-man IL-6.§To whom reprint requests should be addressed at: Laboratory ofTissue Engineering, IDI, Istituto Dermopatico dell' Immacolata, Viadei Castelli Romani 83/85, 00040 Pomezia (Rome), Italy.

10371

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

10372 Medical Sciences: Mathor et al.

into the GP+E-86 ecotropic packaging cell line (28) bystandard calcium phosphate coprecipitation. Forty-eight hoursafter transfection, supernatant was harvested and used toinfect the amphotropic packaging cell line GP+env Aml2 (29)for 16 hr in the presence of 8 ,ug/ml Polybrene. Infected Aml2cells were clonally selected in HXM medium (GIBCO) (29)supplemented with 10% FCS (HyClone), and containing 0.8mg/ml G418 and 0.2 mg/ml hygromycin B (Sigma). Singlecolonies were screened for human interleukin 6 (hIL-6) pro-duction by standard radioimmunoassay using an antibodyhighly specific for hIL-6 (Quantikine, R & D Systems, Min-neapolis) and for viral titer. The resulting producer cell linesshowed a viral titer of 106 colony-forming units (cfu)/ml and3 x 105 cfu/ml, respectively.

Infection ofHuman Keratinocytes. Subconfluent secondarycultures were trypsinized and seeded (5 x 103 cells/cm2) ontoa feeder-layer (2.3 x 104 cell/cm2) composed of lethallyirradiated 3T3-J2 cells and producer GP+env Am12 cells (a 1:2mixture). Epidermal growth factor (10 ng/ml) was added atplating. After 3 days of cultivation in regular medium, cellswere collected and plated onto a regular 3T3-J2 feeder-layer.Subconfluent cultures were then passaged at least once beforeadditional analysis. For colony-forming efficiency assays, sub-confluent cultures were trypsinized, counted, and 100 cellswere plated onto regular feeder-layers. Twelve days later,colonies were assayed for (3-gal activity (31) and then stainedwith Rhodamine B (5). hIL-6 production by transduced kera-tinocytes was assayed by radioimmunoassay (see above).

Clonal Analysis and DNA Analysis. Transduced keratino-cytes obtained from a 7-day culture (1-2 passages aftertransduction) were trypsinized and single cells, isolated underthe microscope, were plated onto a regular feeder-layer in amulti-well plate. After 7 days of cultivation, plates werecarefully scored for the presence of a colony under a micro-scope. A sample of culture medium was taken from wellscontaining a colony and hIL-6 was measured as above. Pro-ductive colonies were photographed under a Zeiss Axiovertmicroscope, and their areas were measured using a comput-erized image analyzing system. Each productive clone was thentransferred by trypsinization to two indicator dishes. One dishwas used for serial propagation and further analysis. Thesecond dish was fixed 9-12 days later and stained withRhodamine B for classification of clonal type, as described byBarrandon and Green (5). Analysis of integrated proviralgenomes was performed by Southern blot analysis as described(31).

Grafting. Cohesive epidermal sheets were generated fromclones transduced with the hIL-6 cDNA. Confluent cultureswere detached from culture vessels by using Dispase II (1),mounted on a vaseline gauze and grafted basal side up onto theinner surface of a rectangular flap of the dorsal skin of athymicnude mice, following the everted-flap technique described byBarrandon et al. (33). At regular intervals, mice were bled fromthe tail and sacrificed for histological examination. Sera wereanalyzed for hIL-6 as above.

RESULTS,8-Gal Transfer. Experiments aimed at determining condi-

tions allowing stable gene transfer in normal human kerati-nocytes were initially performed with a bacterial (3-gal genebecause of the easiness of its detection, also in isolatedkeratinocyte colonies.

Infections were performed by coculturing keratinocyteswith lethally irradiated 3T3-J2 cells and producer GP+envAml2 cells. Coculture was followed by the transfer of thekeratinocytes on a regular 3T3-J2 feeder-layer at clonal density(50-1000 cells per plate), so that each colony was formed bya single cell and could be scored as ,B-gal+ or 3-gal-. Thehighest transduction efficiency ("50%) was obtained when

keratinocytes were seeded for 3 days on a feeder-layer com-posed of a 1:2 mixture of 3T3-J2 and GP+env Aml2 cells. Inpreliminary experiments, we found that exposure to G418reduces the growth potential of transduced epidermal cells(ref. 34, Y. Barrandon, personal communication, and data notshown). Therefore, we avoided G418 selection to analyzetransduced cells during serial cultivation and to evaluate theirlifetime in culture.

Serial propagation of keratinocytes is sustained by themaintenance of stem cells (5-7). Senescence occurs when thelast stem cell present in culture loses its "stem-ness" andgenerates terminal colonies. Therefore, if stem cells werestably transduced, and if the integration and expression of the(3-gal gene did not affect the multiplicative capacity of the cells,one would expect that, during serial cultivation, the initiallymixed population of (3-gal+ and (3-gal- colonies will eventuallygenerate only ,8-gal+ or (3-gal- colonies. Fig. 1 shows that thisis indeed the case. The KE1-LBSN strain was composed of,B-gall and (3-gal- colonies (Fig. 1 A and B). The percentageof blue colonies was constant during the first 2-3 months ofcultivation (Fig. 1D). When senescence approached, and thelast stem cell initiated its clonal evolution, 100% of thecolonies became ,B-gal+ (Fig. 1 C and D). The same experimentwas performed on other three strains of (3-gal transducedkeratinocytes. As expected, one strain eventually generatedonly (3-gall colonies, whereas two strains generated only(3-gal- colonies (not shown).As shown in Fig. 24 (o), (3-gal-transduced keratinocytes

underwent approximately 150 doublings before losing prolif-erative ability (as for the parental untransduced cells, (r),suggesting that integration and expression of the transgene didnot affect the growth potential of the stem cells.

hIL-6 Transfer. To evaluate the capacity of transducedkeratinocytes to secrete an exogenous protein, cells wereinfected with defective retrovirus carrying an hIL-6 cDNA.Clonal analysis (see below) indicated that hIL-6 transductionefficiency was consistently higher than that obtained for 3-gal,because it was always over 95%. hIL-6-transduced keratino-cytes underwent approximately 150 doublings before senes-cence, a lifetime not different from that of (-gal-transducedcells and uninfected cells (Fig. 2A, *). As shown in Fig. 2B (i),mass cultures of transduced keratinocytes secreted hIL-6 at anaverage rate of approximately 350 ng/106 cells/day. hIL-6secretion into the culture medium remained constant duringthe entire lifetime of the culture (Fig. 2B, *), at a level morethan four orders of magnitude higher than background (Fig.2B, o), further suggesting that the epidermal stem cells wereindeed expressing the transgene.

Clonal Analysis of Stem Cells Expressing the Transgene.Three clonal types of keratinocyte with different capacities formultiplication have been identified (7). The holoclone isgenerated by the epidermal stem cell (5, 6). The paraclone isgenerated by a transient amplifying cell (5). The meroclone isan intermediate type of cell and is a reservoir of transientamplifying cells (5, 6). The transition from holoclone tomeroclone to paraclone is a unidirectional process occurringduring natural aging, as well as during serial cultivation (6).The direct demonstration of stem cell transduction was

obtained by clonal analysis of hIL-6-transduced keratinocytes(K160-LIL6SN strain). Single cells were isolated under directvision and plated into multi-well plates. Seven days later,medium from a well containing a single colony was assayed. Itwas found that hIL-6 was produced by 96.3% of the colonies.Colonies were photographed, trypsinized, and clonal analysiswas performed as described in Material and Methods (see alsoref. 5).

Fig. 3 A and C shows colonies generated by the clonesKIL-23 (A) and KIL-A15 (C) 7 days after their isolation. Theirperimeters are smooth and regular and their surface areas are0.80 mm2 and 0.86 mm2, respectively. One-half of the KIL-23

Proc. Natl. Acad. Sci. USA 93 (1996)

Proc. Natl. Acad. Sci. USA 93 (1996) 10373

A

100-

;> 80-

60-........

. >40-

C 20-g!4'

B

D

20 40 60 80 100 120Time (days)

FIG. 1. 3-gal expression during clonal evolution. (-gal transduced keratinocytes (KE1-LBSN strain) were serially cultivated. Colony formingefficiency and ,8-gal staining, followed by rhodamine staining, were performed at each passage. All keratinocyte colonies are stained red byrhodamine. The blue color of 13-gal in colonies of transduced cells partially masks the rhodamine staining. Assays are shown at (A) the 8th passage,(B) the 14th passage, and (C) the 18th passage. (D) Percentage of P3-gal+ colonies during serial cultivation. The arrow indicates keratinocyte infection.Each point corresponds to a cell passage.

colony and one-quarter of the KIL-A15 colony generated 876and 477 large and smooth daughter colonies, respectively (Fig.3 B and D). Because the estimated number of cells in akeratinocyte colony is approximately 2000 per mm2 (5), vir-tually 100% of KIL-23- and KIL-A15-derived cells were clo-nogenic (5). The percentage of aborted colonies (calculated asin ref. 5) was 4.6 and 2.2, respectively. Serial cultivation wascarried out for more than 4 months, and both clones producedmore than 150 generations (Fig. 3E, blue circles and red

A B160

10008140

100.120

100 -10.o~~~80 0~~~~

60 ~0.1

lme (days)

FIG. 2. Serial propagation of transduced keratinocytes. (A) Con-trol untransduced (O), 13-gal transduced KE1-LBSN strain (0), andhIL-6 transduced K160-LIL6SN strain (U) keratinocytes were seriallycultivated for more than 4 months and underwent approximately 150doublings, before losing all proliferative ability. (B) Transducedkeratinocytes (-) secreted hIL-6 at an average rate of approximately350 ng per 106 cells per day, and secretion remained constant duringthe entire lifetime of the culture. Arrows indicate keratinocyte infec-tion. Each point corresponds to a cell passage. These experiments wereperformed on three different keratinocyte strains and comparableresults were obtained.

triangles). These data fulfill the criteria for holoclones arisingfrom stem cells.During their lifetime cultures generated by KIL-23 (blue

circles) and KIL-A15 (red triangles) stem cells secreted hIL-6at a rate of approximately 800 ng and 180 ng per 106 cells perday, respectively, values higher and lower than that for a masspopulation (Fig. 3F). Comparable data were obtained fromother holoclones analyzed (Table 1). These data unambigu-ously show that epidermal stem cells can be stably transduced.Clonal analysis showed that meroclones can be stably trans-duced as well. Fig. 3E (A) shows a "good" transduced mero-clone (secreting hIL-6 at a rate of 200 ng/106 cells/day), whichbecame senescent and converted completely to paraclone-forming cells after approximately 80 generations. Other trans-

Table 1. Proviral integrations and hIL-6 production in clones

hIL-6 secretion

CloneKIL-B17KIL-Q22KIL-A15KIL-FlKIL-K13KIL-AlOKIL-017KIL-R14KIL-23KIL-D4KIL-C8

No. of proviralintegrations*t

1233344578

15

In culturemedium,tt

ng/106 cells/day21028218015045052244947881011401014

In serum ofgrafted mice,*t

pg/ml16.0NDND10.8NDNDNDND59.176.569.6

ND, not determined.*Each value is the average of three grafted mice.thIL-6 production in vitro vs. proviral integrations: r2 = 0.7001.thIL-6 serum levels in mice vs. hIL-6 production in vitro: r2 = 0.9944.

Medical Sciences: Mathor et al.

10374 Medical Sciences: Mathor et al.

C

1-k

p. 1.t

_.j...t..s

q -r

A -

E

0 40 60 80 100 120 140

Time (days)

1000

-

-J00

0,01

Proc. Natl. Acad. Sci. USA 93 (1996)

_ B_ K__ _.

____ \__ __

_._

__ t$-

_K ............ i_ __ erf__ _rr_ _r-X FEV11_fS_ r

__rn._

F

40 60 80 100 120 140

Time (days)

FIG. 3. Isolation of stably transduced stem cell clones. Single hIL-6-transduced keratinocytes, isolated under direct vision, were plated onto a regularfeeder-layer in a multi-well plate. Clones that were found to synthesize and secrete hIL-6 were photographed and their areas were measured. Each clonewas then transferred to two indicator dishes. One dish was used for serial propagation and further analysis. The second dish was fixed 9-12 days laterand stained with Rhodamine B for classification of clonal type (5). Two clones are shown here. After 7 days of cultivation, KIL-23 (A) and KIL-A15 (C)clones generated smooth and regular colonies, covering a surface of 0.80 mm2 (A) and 0.86 mm2 (C), respectively. The estimated number of cells (seeref. 5) was 1600 and 1700, respectively. One-half of the KIL-23 colony and one-quarter of the KIL-A15 colony generated 876 (B) and 477 (D) large andsmooth daughter colonies, respectively, indicating that virtually 100% of cells were clonogenic. Indicator dishes were stained at day 9 (B) and day 10 (D).The percentage of aborted colonies (5) was 4.6 and 2.2, respectively. (E) Serial cultivation and resultant cell doublings of parental mass culture (blacksquares), KIL-23 (blue circles), and KILHA15 stem cell (red triangles) are shown. The arrow indicates the time of infection. Cell generations producedby a transduced "good" meroclone (clone KIL-C69) are shown for comparison (open triangles). (F) Transduced KIL-23 (blue circles) and KIL-A15 (redtriangles) stem cells secreted hIL-6 at an average rate of approximately 800 and 180 ng per 106 cells per day, respectively, and secretion remained constantduring their lifetime. The parental mass culture is indicated by the black squares. Measurements began at about 40 days (arrow) after the processing ofthe skin biopsy, the preceding interval being devoted to the infection of the mass culture, the isolation of clones, and their growth to suitably largepopulations. Comparable data were obtained from other holoclones analyzed (see also Table 1).

duced meroclones became senescent after only 20-30 gener-ations (not shown).

Analysis of Integrated Proviral Genomes in Mass Culturesand in Clones Producing hIL-6. Integration of the LIL6SN

vector in keratinocyte cultures was analyzed by Southernhybridization of the genomic DNA extracted from 5 x iO to2 x 106 cells, using a Neo-specific probe. The DNAs fromkeratinocyte cultures at passages 7-17 were digested with

Co)0)c

~0C1)00c

I I 0.ft

Proc. Natl. Acad. Sci. USA 93 (1996) 10375

HindIll, which cuts the proviral genome only once. Hybrid-ization to a Neo-specific probe allowed us to detect 3' inte-gration sites (Fig. 4D). DNAs obtained from the earlierpassages showed a smear-like pattern, probably resulting fromnumerous integrated proviruses in a heterogeneous popula-tion of transduced cells. During several cultivations this patternprogressively evolved into one containing only a few discretebands (Fig. 4A). This showed that the polyclonal population oftransduced keratinocytes was progressively restricted to cellsderiving from fewer transduced clonogenic cells, and ulti-mately from a single clone responsible for maintenance of theculture. Southern hybridization performed on clonal culturesderiving from single cell isolates, such as is shown for cloneJUL-23 (Fig. 4B), gave a distinct pattern of integrated bandsthat was maintained without modification throughout theculture life of the clone.As shown in Fig. 4, cloned keratinocytes possessed different

numbers of integrated proviral copies in their genome (KIL-23in B, KIL-B17 and KIL-Fl in C). For each of a larger numberof clones, the number of integrants was compared with the rateof hIL-6 secretion in the culture medium. As shown in Table1, there was a very good correlation between the number ofproviral integrations (from 1 to 15) and the hIL-6 secretionrate (150-1140 ng/106 cells/day).

Transplantation onto Athymic Mice of Epidermal SheetsGenerated from Transduced Clones. Cohesive epidermalsheets were generated from clones producing either low (JUL-B17 and JUL-Fl), intermediate (JUL-23), or high (JUL-C8 andJUL-C4) levels of hIL-6 and grafted onto athymic mice asdescribed in Materials and Methods. The size of the graft wasapproximately 1 cm2. Evaluation of hIL-6 content in themouse serum was performed 3 days after grafting. As shownin Table 1, the serum levels of hIL-6 (10.8-76.5 pg/ml) werestrictly proportional to the rate of secretion in vitro (150-1140ng/106 cells/day), and to the number of proviral integrations.

A B

Comparable circulating hIL-6 levels were maintained up to 7days after grafting, the longest period during which the graftmaintained a healthy stratified epithelium.

DISCUSSIONThe long-term expression of a transgene in an epidermalpopulation can be expected only if the transgene is carried bythe stem cells. In culture, holoclones have the greatest growthpotential. Because epidermis generated in humans from suchcultures appears to last indefinitely and is also composed ofspecialized differentiated cells, holoclone-formers appear tobe equivalent to epidermal stem cells. The transient amplifyingcell population, which arises from stem cells and will eventuallygenerate terminally differentiating progeny (5, 6), proliferatesonly for a limited period of time and represents the largestgroup of dividing cells. Clonal analysis allows the identificationof stem cells and transient amplifying cells in human epidermisand hair matrix (5, 6, 35).

Recently, difficulties in obtaining stable transgene expres-sion by human keratinocytes had been attributed to theunsuccessfull targeting of stem cells (14). We show here thatholoclones can indeed be stably transduced by retroviruses andthat the transgene expression lasts for the entire life-span ofthe culture. The question of whether epidermal stem cells canbe transduced has been raised in the past by Garlick et al. (25).In their experiments, expression of the transgene was detectedfor approximately 30 doublings (25). This might suggest thatthe targets of the transduction were mainly transient amplify-ing cells and meroclones. Similarly, Fenjves et al. (36) reportedretrovirus-driven expression of factor IX for five cell passages.Alternatively, their use of G418 selection may have damagedstem cells or reduced growth potential. In this respect, it isworth noting that Flowers and colleague (21) reported long-term (130 days) retrovirus-driven expression of neomycin

C

p7 p11 p14 pl7 f....... ....... .. ...,i,,,.,,,5t 5 _:¢.'..'.:,> 1__1 5.*X

.:.4::'. - r, ,,-.: iF_ . .w .....................#* :. :,::.

:t.!! _

$:..: :. ,:5 t"iCa t4::.! ... ..*.te.F.',..tw $,......* : : : :.- . . . .* 4 5 _i!':'.''.''.'.' .....* .. ::,.:: .: . v ., ..... : .

:' :': ..: ::. . . ^ :.: _i;,', ..... .. ...

H3I

S

KIL-23

p10 p13 p15

H3 S

INb_ _ ~~~~~~~~~~~~~~~~~...... . ....,:.:-5.2

'..''' '','-','..'~~~~~..'.t.'',..'.'.,...:.

-4.2

_ E _ ~~~~~-3.5

_ : ::: _ ~~~2.0-1.95-1.5

11

H3 S H3

DS H3 S

LIL6SNN

FIG. 4. Study of integrated proviruses by Southern blot analysis. Southern blot analysis of LIL6SN retroviral vector integration in genomic DNAfrom bulk cultures of human keratinocytes at passages (p) 7, 11, 14, and 17 after transduction (A); from culture of clones KIL-23 at passages 10,13, and 15 (B); and from cultures of clones KIL-B17 and KIL-Fl at passage 10 (C). The LIL6SN producer cell line, containing two copies ofintegrated provirus, is shown as a control (C). DNA (5 ,ug) were digested with HindlIl (H3) and hybridized to a Neo-specific probe. An SstI (S)digest of DNA (5 ,ug) from the LIL6SN-producing Aml2 clone (Aml2/LIL6SN) is shown as a control for the size of the intact integrated provirus(3.4 kb) in all panels. The size marker is a Aphage HindIII digest (in kb). (D) Schematic map of the LIL6SN provirus. Solid boxes indicate the virallong terminal repeats, open boxes the hIL-6 and neomycin phosphotransferase (NeoR) cDNAs, and arrowhead-shaped box the simian virus 40 earlypromoter and origin of replication (SV). Location of the SstI (S) and HindIII (H3) restriction sites is indicated.

Medical Sciences: Mathor et al.

I

r%L

2

10376 Medical Sciences: Mathor et al.

phosphotransferase in canine keratinocytes. The very smallproportion of the keratinocytes retaining G418 resistence aftergrafting may be attributed to stem cells.

Southern blot analysis showed multiplicity of random pro-viral integration in single epidermal stem cells, as previouslyshown for human myogenic cells (31, 37). The rate of secretionof the exogenous protein by cultures generated by single cloneswas proportional to the number of integrations per progenitorcell. More importantly, when grafted on athymic animals,epidermal sheets generated from transduced clones secretedthe exogenous protein in the blood circulation at levels strictlyproportional to the rate of secretion in vitro and hence to thenumber of proviral integrations. Therefore, by using trans-duced keratinocyte stem cell clones, the circulating levels ofthe transgene product can be modulated by selecting a clonecarrying a determined number of proviral integrations.The possibility of specifying the level of transgene expres-

sion and its permanence in an homogeneous clone of stem cellorigin opens new perspectives in the long-term treatment ofgenetic disorders.

We thank Prof. Howard Green for providing us with 3T3-J2 cellsand Prof. Adriano Tito Franzi for helping us with histological evalu-ations. We thank Barbara Minuto and Claudia Rossi for expertassistance. The financial support of Telethon-Italy (grants N° A.08and A.41) is gratefully acknowledged. This work was also supported byConsiglio Nazionale delle Ricerche, Associazione Italiana per laRicerca sul Cancro, and Istituto Superiore di Sanita. (Progetto sosti-tuzioni funzionali, organi artificiali e trapianti d'organo).

1. Green, H., Kehinde, 0. & Thomas, J. (1979) Proc. Natl. Acad. Sci.USA 76, 5665-5668.

2. Compton, C. C., Gill, J., Bradford, D. A. M., Regauer, S., Gallico,G. G. & O'Connor, N. E. (1989) Lab. Invest. 60, 600-612.

3. De Luca, M. E., Albanese, M., Megna, R., Cancedda, R., Man-giante, P. E., Cadoni, A. & Franzi, A. T. (1990) Transplantation50, 454-459.

4. Lajtha, L. G. (1979) Differentiation 14, 23-34.5. Barrandon, Y. & Green, H. (1987) Proc. Natl. Acad. Sci. USA 84,

2302-2306.6. Barrandon, Y. (1993) Dev. Biol. 4, 209-215.7. Gallico, G. G., O'Connor, N. E., Compton, C. C., Kehinde, C. &

Green, H. (1984) N. Engl. J. Med. 311, 448-451.8. De Luca, M., Albanese, E., Bondanza, S., Megna, M., Ugozzoli,

L., Molina, F., Cancedda, R., Santi, P. L., Bormioli, L., Stella, M.& Magliacani, G. (1989) Burns 15, 303-309.

9. Romagnoli, G., De Luca, M., Faranda, F., Bandelloni, R., Franzi,A. T., Cataliotti, F. & Cancedda, R. (1990) N. Engl. J. Med. 323,527-530.

10. Boyce, S. T. (1994) J. Invest. Dermatol. 102, 8-10.11. Fenjves, E. S., Gordon, D. A., Pershing, L. K., Williams, D. L. &

Taichman, L. B. (1989) Proc. Nati. Acad. Sci. USA 86,8803-8807.12. Alexander, M. Y., Bidichandani, S. I., Cousins, F. M., Robinson,

C. J. M., Duffie, E. & Akhrust, R. J. (1995) Hum. Mol. Gen. 4,993-999.

13. Eady, R. A. J. & Dunnill, M. G. S. (1994) Arch. Dermatol. Res.287, 2-9.

14. Fenjves, E. S. (1994) J. Invest. Dermatol. 103, 70S-75S.15. Krueger, G. G., Morgan, J. R., Jorgensen, C. M., Schmidt, L., Li,

H. L., Kwan, M. K, Boyce, S. T., Wiley, H. S., Kaplan, J. &Petersen, M. J. (1994) J. Invest. Dermatol. 103, 76S-84S.

16. Greenhalgh, D. A., Rothnagel, J. A. & Roop, D. R. (1994)J. Invest. Dermatol. 103, 63S-69S.

17. Morgan, J. R., Barrandon, Y., Green, H. & Mulligan, R. C.(1987) Science 237, 1476-1479.

18. Teumer, J., Lindahl, A. & Green, H. (1990) FASEB J. 4,3245-3250.

19. Vogt, P. M., Thompson, S., Andree, C., Liu, P., Breuing, K.,Hatzis, D., Brown, H., Mulligan, R. C. & Eriksson, E. (1994)Proc. Natl. Acad. Sci. USA 91, 9307-9311.

20. Jensen, U. B., Jensen, T. G., Jensen, P. K. A., Rygaard, J., Han-sen, B. S., Fogh, J., Kolvraa, S. & Bolund, L. (1994) J. Invest.Dermatol. 103, 391-394.

21. Flowers, M. E. D., Stockschlaeder, M. A. R., Schuening, F. G.,Niederwieser, D., Hackman, R., Miller, A. D. & Storb, R. (1990)Proc. Natl. Acad. Sci. USA 87, 2349-2353.

22. Jensen, T. G., Birk-Jensen, U., Jensen, P. K. A., Ibsen, H. H.,Brandrup, F., Ballabio, A. & Bolund, L. (1993) Exp. Cell Res. 209,392-397.

23. Gerrard, A. J., Hudson, D. L., Brownlee, G. G. & Watt, F. M.(1993) Nat. Genet. 3, 180-183.

24. Setoguchi, Y. S., Jaffe, H. A., Danel, C. & Crystal, R. G. (1994)J. Invest. Dermatol. 102, 415-421.

25. Garlick, J. A., Katz, A. B., Fenjves, E. S. & Taichman, L. B.(1991) J. Invest. Dermatol. 97, 824-829.

26. Stockschlader, M. A. R., Schuening, F. G., Graham, T. C. &Storb, R. (1994) Gene Ther. 1, 317-322.

27. Zambruno, G., Marchisio, P. C., Marconi, A., Vaschieri, C.,Melchiori, A., Giannetti, A. & De Luca, M. (1995) J. Cell Biol.129, 853-865.

28. Markowitz, D., Goff, S. & Bank, A. (1988) J. Virol. 62, 1120-1124.

29. Markowitz, D., Goff, S. & Bank, A. (1988) Virology 167,400-406.30. Miller, A. D. & Rosman, G. J. (1989) BioTechniques 7, 980-990.31. Salvatori, G., Ferrari, G., Mezzogiorno, A., Servidei, S., Coletta,

M., Tonali, P., Giavazzi, R., Cossu, G. & Mavilio, F. (1993) Hum.Gene Ther. 4, 713-723.

32. Miller, A. D., Miller, D. G., Garcia, J. V. & Lynch, C. M. (1993)Methods Enzymol. 217, 581-599.

33. Barrandon, Y., Li, V. & Green, H. (1988)J. Invest. Dermatol. 91,315-318.

34. Stockschlader, M. A. R., Storb, R., Osborne, W. R. & Miller,A. D. (1991) Hum. Gene Ther. 2, 33-39.

35. Rochat, A., Kobayashi, K. & Barrandon, Y. (1994) Cell 76,1063-1073.

36. Fenjves, E. S., Yao, S. N., Kurachi, K. & Taichman, L. B. (1996)J. Invest. Dermatol. 106, 576-578.

37. Ferrari, G., Rossini, S., Giavazzi, R., Maggioni, D., Nobili, N.,Soldati, M., Ungers, G., Mavilio, F., Gilboa, E. & Bordignon, C.(1991) Science 251, 1363-1366.

Proc. Natl. Acad. Sci. USA 93 (1996)