Gene Expression from Adeno-associated Virus Vectors in Airway Epithelial Cells Terence R. Flotte, Rikki Solow, Roland A. Owens, Sandra Afione, Pamela L. Zeitlin, and Barrie J. Carter Laboratory of Molecular and Cellular Biology, National Institute of Diabetes, Digestive and Kidney Diseases, National Institutes of Health, Bethesda, and Department of Pediatrics, Johns Hopkins School of Medicine, Baltimore, Maryland

Lung diseases such as cystic fibrosis (CF) might be treated by gene therapy using viral vectors delivered to the airway. One potential vector is the defective human parvovirus, adeno-associated virus (AAV). We examined the AAV ps transcription promoter for gene expression in immortalized cell lines derived from the airway (IB3-1) or pancreas (CFPAC-I) of CF patients. AAV vectors expressing the prokaryotic genes cat (pAAVpscat) or neo (pAAVpsneo) from the ps promoter were evaluated after introduction into IB3-1 or CFPAC-I cells by lipofection. In transientassays in both cell lines, the cat gene was expressed 5- to 1O-fold more efficiently from the ps promoter than from a simian virus 40 early gene promoter (pSVcat). IB3-1 cells were transformed stably to geneticin resistance by pAAVpsneo at a 5-fold higher efficiencythan by an SVneo vector. The AAV inverted terminal repeat (ITR) region immediately upstream of the ps promoter appears to have an enhancer effect and the promoter also contains a CREB site which confers a response to forskolin. In IB3-1 cells, expression of the cat gene from a Ps promoter was decreased about 5-fold by deletion of both the upstream ITR and the CREB site. The AAVpsneo vector was also packaged into AAV particles and used to infect IB3-1 cells as a transducing virus. Under these conditions, 60 to 70% of the cells could be stably transformed to geneticin resistance. Thus, AAV transducing vectors appear to be a highly efficient delivery system for stable integration and expression of genes in cultured airway epithelial cells.

The airway epithelium is the critical site of cellular dysfunction for cystic fibrosis (CF), the most common lethal genetic disease in North America (I). It may be possible to treat this disease by delivering a functional cystic fibrosis transmembrane regulator (CFTR) gene to the airway surface. The CF defect has now been complemented by introduction and expression of the CFTR gene in human cells in culture (2-4). This potentially provides a valuable tool for both experimental and therapeutic applications. Any gene delivery system for treatment of CF patients will require the capacity to transfer the gene to a high proportion of the cells and maintain efficient levels of expression over prolonged periods. One approach to gene therapy is to use viruses that are adapted to airway epithelial cells (5). Of the many viruses (Received in original form January 13, 1992 and in revised form April 9, 1992) Address correspondence to: Barrie J. Carter, Ph.D., Laboratory of Molecular and Cellular Biology, NIDDK, Building 8, Room 304, National Institutes of Health, Bethesda, MD 20892. Abbreviations: adeno-associated virus, AAV; adenovirus type 5, Ad 5; cyclic adenosine monophosphate, cAMP; cystic fibrosis, CF; cystic fibrosis transmembrane regulator, CFTR; cytomegalovirus, CMV; cAMP response element, CRE; inverted terminal repeat, ITR; Rous sarcoma virus, RSV. Am. J. Respir. CeU Mol. BioI. Vol. 7. pp. 349-356, 1992

that may infect the human respiratory epithelium, most do not undergo stable DNA integration, and almost all are pathogenic to the host. One exception is adeno-associated virus type 2 (AAV2), which is a defective human parvovirus that normally replicates productively only in the presence of helper functions provided by a co-infecting adenovirus or herpesvirus (6). Generally, in the absence of helper virus, AAV2 infection results in high-frequency, stable integration of its genome into the host cell at a specific site on human chromosome 19 (7-9). Further, AAV has not been implicated as the etiologic agent of any disease, and integration of an AAV provirus has no apparent affect on cell growth, morphology, or differentiation. The AAV genome contains two genes, rep and cap, and inverted terminal repeat (ITR) sequences (10, 11). AAV genomes inserted into bacterial plasrnids are infectious when transfected into adenovirus-infected cells, and this has allowed the study of AAV genetics and development of AAV vectors. The ITR sequences, which are required in cis, are replication origins and may also contain signals for encapsidation of viral genomes. The rep and cap genes are required only in trans and function for replication and encapsidation, respectively, of the viral genome. AAV vectors have been constructed in AAV plasmids by substitution of foreign DNA for either or both of the cap and

350

AMERICAN JOURNAL OF RESPIRATORY CELL AND MOLECULAR BIOLOGY VOL. 7 1992

the rep genes while leaving the ITRs intact. These vectors can be packaged into AAV particles by co-transfection of the vector plasmid and a second packaging plasmid containing the rep and cap genes into adenovirus-infected cells. Vectors packaged in this way have been shown to be useful for transducing reporter genes, such as the neomycin phosphotransferase (neo) gene into several mammalian cell lines (12-16). AAV vectors have also been used for transient or stable expression of reporter genes after direct transfection into target cells (13, 17, 18). These studies used the AAV p40 promoter, which normally is used for cap gene expression. Also, co-transfection of two AAV vector plasmids expressing the a and {3 subunits of human thyroid-stimulating hormone led to efficient production of biologically active human thyrotropin (19). This suggests that AAV vectors and AAV transcription promoters may also be useful in non-viralbased gene delivery systems such as liposomes or receptormediated endocytosis. We have performed preliminary studies to determine if AAV2 vectors are likely to be useful for stable, high-level expression of genes introduced into a CF airway epithelial cell line by either transfection or virus particle-mediated transduction. The size of the CFTR coding region is close to the packaging limit for AAV transduction vectors and, therefore, construction of an AAV-CFTR vector will require very compact transcription regulatory elements. We have analyzed the suitability of the AAV ps promoter for expression because this promoter forms a convenient cassette with the AAV left-hand ITR and is one of the smallest promoters available. We report here that the AAV2 ps promoter was efficient for expression of reporter genes in both stable and transient assays. Also, expression from the ps promoter was moderately responsive to activation by forskolin. Further, an AAVpsneo vector packaged into AAV transducing particles resulted in stable expression of geneticin resistance in up to 70 % of airway epithelial cells.

Plasmids Plasmids used for transient transfection assays were constructed by standard methods (24). The plasmids pR01472, pROl, pROlO3, pR0l2, and pROll all contain varying amounts of the left-hand terminus of the AAV2 genome immediately upstream of the chloramphenicol acetyltransferase (cat) gene and express CAT activity under the control of the AAV ps promoter (Figure 1). Only pR01472 contains an intact copy of the AAV2 left-hand ITR. pROl contains a portion of the ITR. All of the plasmids contain some or all of the AAV2 ps promoter sequences. pROll lacks a consensus cyclic

pR01472

pRO]

pR0103

pR012

Materials and Methods Cells The IB3-l cell line (IB3 cells) is an immortalized human bronchial epithelial cell line derived from a CF patient (20, 21). These cells retain characteristics of epithelial cells and are deficient in protein kinase A-mediated activation of chloride conductance, which is diagnostic of the CF genetic defect. IB3 cells were grown in serum-free LHC-8 medium (Biofluids, Inc., Gaithersburg, MD) with added endothelial cell growth supplement (15 j.tglml) in culture flasks or dishes coated with collagen (150 j.tg/ml), fibronectin (10 j.tglml), and bovine serum albumin (lO j.tg/ml). CFPAC-l (CFPAC) cells derived from a pancreatic adenocarcinoma taken from a CF patient (21) were grown in Iscove's modified Dulbecco's medium supplemented with 10% fetal calf serum. The 29331 cells, originally derived from human embryonic kidney cells transformed with the adenovirus type 5 EtA and ElB genes (23), were grown in Eagle's minimal essential medium with lO% fetal calf serum and were used for packaging AAV2 vectors into virus particles (13). All three cell types were grown at 37° C in 5% CO2 •

pROll

190

Figure 1. Structure of AAVpsCAT expression plasmids. pR01472 contains the first 321 nucleotides of AAV2 DNA inserted immediately upstream of the cat coding sequence. The AAV and cat sequences are inserted in a pBR322-derived plasmid, but for simplicity the pBR sequences are not shown. The other plasmids were derived from pROI472 by deleting various regions of the AAV sequence extending rightwards from nucleotide 1 as indicated. The AAVnucleotide sequence is numbered according to Srivastava and associates (46), as indicated by the vertical arrows. The diagram is not drawn to scale. Regions of the AAV sequence are indicated as follows: inverted terminal repeat/replication origin extending from nucleotide 1 to 145 (ITR), a CREB/ATF consensus site (CREB), an MLTFIUSF binding site (MLTF), an adenovirus ElA protein responsive site (ElA) , a TFIID binding site (TATA), and the start site for transcription (horizontal arrow).

Flotte, Solow, Owens et al.: AAV Vectors in Airway Cells

351

adenosine monophosphate (cAMP) response element, which is a binding site for a protein (CREB) that mediates transcriptional activation by cAMP via protein kinase A. The AAV-cat plasmids were constructed as follows: The parental plasmid, pAV2, contains the entire 4,68 l-nucleotide sequence of AAV2inserted in a pBR322 derived plasmid via a polylinker and BglII linkers (25). From this a plasmid, pTSl, was derived by deleting AAV nucleotides 2397 to 3987 and inserting an 873-bp cat gene coding sequence (17, 26) at the HindIII site (AAV nucleotide 1882). A plasmid, pYT45, was then derived from pTSl by successive deletion of an SphI/PvuII fragment (pBR322 nucleotides 561 to 2065) followed by deletion of AAV nucleotides 1910 to 4162), and finally deletion of AAV nucleotides 268 to 1882. This placed the cat sequence in pYT45 under control of the AAV ps promoter. pR0l472 was derived from pYT45 by first deleting an SnaBI/Ndel fragment (AAV nucleotide 4498 to pBR322 nucleotide 2295) to yield pR045. This removed the righthand AAVITR but retained the AAV polyadenylation (polyA) site downstream of the cat gene. pROI472 was then constructed by insertion of a synthetic double-stranded oligonucleotide into the HindIII site of pR045. The oligonucleotide consisted of AAV nucleotides 266 to 321 flanked by HindIII overhangs such that only the 5' end of the insert had a complete HindIII site after ligation. Proper insertion was confirmed by sequencing. The final construct, pR0l472 , contains AAV nucleotides 1 to 321 upstream of the cat gene (except that nucleotides 264 and 265 are changed from the wild type sequence CC to TT) and AAV nucleotides 1882 to 1910 and 4162 to 4492 (containing the polyA signal) downstream. pR01, pROI03, pR0l2, and pROll were derived by cleavage ofpR01472 at the BamHI site in the polylinker upstream of the ITR, digestion with Ba131, and ligation of BamHI linkers. The DNA was then cleaved with BamHI and HindIII, and BamHI/HindIII fragments of70 to 250 nucleotides were isolated on a 2 % agarose gel and ligated into the large BamHI/HindIII fragment of pR0l472. The extent of the deletions were confirmed by restriction cleavage and DNA sequencing. The plasmid pSV2cat contains the SV40 promoter upstream of the cat gene (27). Similarly, the pSV2neo plasmid contains a neo gene driven from the SV40 early promoter. The pAAVSVneo plasmid was constructed by inserting the SV40 promoter and the neo gene sequences from pSV2neo into the large XbaI fragment of pSub201(+) between the AAV ITRs (28). pAAVpsneo contains the AAV2 ITRs derived from pAV2, as well as the neo gene coding sequence expressed from the ps promoter and a synthetic oligonucleotide of 60 bp containing a polyA signal (29). It was derived by cleaving pYT45 with HindIII and SnaBI to delete both the cat sequence and the endogenous AAV polyA site by insertion of the neo sequence from pSV2neo and the synthetic polyA site. The packaging plasmid pAAV/Ad (16) contains the two AAV genes, rep and cap, necessary for replication and packaging of AAV2 vectors but is deleted for the AAV ITRs. Therefore, the AAV genome in pAAV/Ad cannot replicate.

Transfection of AAV2 Vectors DNA transfection in IB3-1 and CFPAC-1 cells was performed in 10- or 35-mm dishes using lipofection with cationic liposomes. A total of 30 p,g of BRL- Lipofectin reagent (Life Sciences Inc., Gaithersburg, MD) was used for each 5 to 6 p,g of DNA transfected. Lipofectin and DNA were mixed in 1.0 m1 of LHC-8 serum-free medium and added to cells (5 X lOS in 35-mm wells) already covered with 0.5 ml of medium. Cells were exposed to DNA for 4 h, rinsed with phosphate-buffered saline, and then covered with fresh medium. CAT Assays-Forskolin Stimulation Cells used for transient expression of cat vectors were harvested at 48 h after lipofection, lysed by three cycles of freezing and thawing, and assayed for CAT activity (17). CAT activity in CFPAC-1 cells was normalized for the amount of protein extract used in the assay. In IB3-1 cells, protein amounts were not used for normalizing because of the contribution of the collagen/fibronectin/bovine serum albumin substrate protein. In each case, values are expressed as a percent of maximal CAT activity in that cell line, normalized to 100% for each set of experiments. In IB3-1 cells, the maximal activity was 27% acetylation per 106 cells per 90 min of incubation in the CATassay. In CFPAC-1 cells, the maximal activity was 36 % acetylation per mg protein per 90 min. For testing of cAMP induction of expression, cells were incubated with forskolin at a final concentration of 20 p,Mfor 4 h immediately after transfection. Cells were then processed as with the other CAT assays. Packaging of AAV2-neo Vectors Packaging of AAV2vectors was accomplished by first infecting 293-31 cells (grown to semiconfluence in 100-mm dishes) with adenovirus type 5 (Ad5) (at a multiplicity of 5 to 10 infectious units/cell), and then co-transfecting the vector plasmid, pAAVpsneo (1 Jlg) and the packaging pAAV/Ad (5 p,g) using the CaP04 transfection procedure (16, 17). Medium was replaced 2 h before transfection, and Ad5 was inoculated into the medium 1 h before transfection. The medium was changed 4 h after transfection. Cells were grown for 3 to 4 days, then harvested by gently scraping into the medium followed by low-speed (4,000 rpm) centrifugation and resuspension in 10 mM Tris-HCl buffer, pH 8.0. Cells were lysed by freezing and thawing 3 times, and the virus was concentrated and purified using CsCl density gradient ultracentrifugation (30). Fractions taken for transduction assays were then dialyzed against l x SSC 3 times for 1 h at room temperature and heat-treated at 60° C for 15 min to inactivate any possible residual adenovirus. The titer of the vector preparation was determined by DNA slot-blot hybridization (12, 16). AAV2 Particle-mediated Transduction Virus particle-mediated neo transduction of IB3-1 CF bronchial epithelial cells was accomplished by infecting 103 to 4 x 104 cells in individual wells of a 24-well dish with a known number of AAVpsneo vector particles per cell. The

352

AMERICAN JOURNAL OF RESPIRATORY CELL AND MOLECULAR BIOLOGY VOL. 7 1992

cells were then selected for expression of geneticin resistance as described below. Geneticin Selection Cells used for stable neo expression were trypsinized and replated at a 1:10 dilution 24 to 48 h after transfection with plasmid DNA or infection with transducing vector particles, and geneticin sulfate was added at 72 h at a concentration of 120 I-tg/rnl. This concentration of geneticin was chosen based on a minimal lethal dose titration of IB3 cells, which are extremely sensitive to this antibiotic. Geneticin-resistant (gen') colonies were counted at 14 to 16 days after beginning selection. Transduction frequency (%) was calculated as the number of colonies (gen') obtained in the presence of geneticin divided by the number of colonies obtained in the absence of geneticin multiplied by 100. Assays were performed in triplicate.

TABLE I

Activity of M Vp5cat vectors in CF epithelial cell lines Cell Linet Plasmid"

pR0l472 pROl pRO 103 pR0l2 pRO 11 pSV2cat

IB3

100.0 58.8 55.0 72.3 41.3 8.8

± 14.8 ± 23.3 ± 21.6t ± 11.8§ ± 5.411 ± 6.411·'

CFPAC

64.2 ± 6.9 100.0 ± 25.9 76.7 ± 11.3 84.4 ± 8.7 58.6 ± 14.1 30.7 ± 4.6H

" Cells were transfected with 5 ",g of the indicated plasmid using lipofection. ± SEM as described in MATERIALS AND METHODS. The data represent the mean of four determinations for each plasmid in each cell line. t p < 0.05 versus pROI472 according to Student's t test. § P < 0.05 versus pRO I I. II P < 0.01 versus pROI472. , P < 0.01 versus pRO I I. t CAT activity is expressed as percent maximal CAT activity

Results DNA Transfection of AAV2 Vectors Activity of the MV Ps promoter in transient assays. To test the efficiency of the AAV ps promoter in human epithelial cells having the CF defect, we constructed several AAVp-eat plasmids. The plasmid pR0l472 contains the cat gene coding sequence positioned immediately downstream of AAV nucleotides 1 to 321 (Figure 1). This region of AAV includes several notable features including the AAV replication origin (ITR) and the TATA box of the ps promoter at nucleotide 255. Nucleotides 204 to 213 constitute a binding site for the MLTF (USF) transcription factor, and nucleotides 217 to 236 comprise a 10-bp repeat that constitutes a novel response element for the adenovirus transcription factor E1A (31). A previous report indicated that an AAV promoter consisting of nucleotides 190 to 310 had only minimal activity in HeLa cells unless activated by the E1A protein (31). In contrast, an AAV promoter comprising the nucelotides 145 to 310 had significant activity in HeLa cells in the absence ofE1A (32). These differences may reflect the presence between nucleotides 160 and 180, of the sequence GlGACGlGAATTACGTCATAG, which has homology to the cAMP response element (CRE) and the binding site for the CREB/ATF transcription factor family (33-37). Several additional plasmids were derived from pR01472 that had the ITR or CRE deleted. The AAVpscat plasmids were tested for CAT expression in transient assays after transfection into IB3 cells and CFPAC cells (Table 1). pR0l472, containing both the ITR and the entire AAV Ps promoter, had the highest activity in IB3 cells. Its activity was higher than that of pROlO3 or pR0l2, both of which lack the ITR. pROl, which was deleted for 83 nucleotides of the ITR, also had lower activity than pR0l472 in IB3 cells. pROl1, lacking both the ITR and the CRE site, had significantlylower activity than either pR0l472 or pR012, which retained the CRE site. Furthermore, the activity ofthe complete AAV ps promoter in pR01472 was more than 10fold higher than that of the SV40 early promoter in IB3 cells. Even the minimal ps promoter in pROll was nearly 5-fold more active than the SV40 promoter. In CFPAC cells (Table 1), pR01472 also showed significantly higher activity than pSV2cat. The other AAVpscat

plasmids, except for pR01, behaved similarly to their relative activities in IB3 cells. Deletion of the left half of the ITR in pROl suggests that this region has a modest inhibitory effect in CFPAC cells but a modest enhancer effect in IB3 cells. Any ps promoter to be used in an AAV vector that is packaged into AAV particles will have an AAV ITR present. Therefore, it is convenient happenstance that this region has an activator effect on expression, at least in IB3 cells. Even in CFPAC cells, although this region may be slightly inhibitory, it still permitted significant activity. To determine if the CRE might function in the context of the complete AAV ps promoter, we tested the effect of forskolin-induced elevation of the level of cAMP on the subsequent expression of CAT (Figure 2). In IB3 cells, there was a 2-fold increase of CAT activity from pR0l472 in response to forskolin but there was no effect on pR01l, which was deleted for the CRE site or on pSV2cat. In CFPAC cells, there was almost a 4-fold induction from pR0l472, a minor effect on pROl1, and again no effect on pSV2cat. These results suggest that, in the context of the entire left-hand terminus of AAV2, forskolin can mediate increased expression from the AAV ps promoter. This is consistent with the presence of the CRE. Activity ofthe MV Ps promoter for neo gene expression. The above experiments demonstrated the efficiency of the AAV ps promoter for gene expression in IB3 cells in a transient assay. To test this promoter for longer term, stable expression, we constructed a pAAVpsneo vector. This plasmid contains a neo gene coding sequence inserted immediately downstream from the ps promoter and left-hand ITR, analogous to the cat plasmid R01472. In addition, pAAVpsneo contains the right-hand AAV ITR downstream of the polyA site. This was compared with pSVneo, which contains neo expressed from an SV40 early gene promoter and pAAVSVneo. In this latter plasmid, the SV40 promoter and neo gene cassette are flanked by left and right AAV ITRs. These plasmids were transfected into IB3-1 cells using Lipofectin and the cells were subsequently selected in the antibiotic geneticin. The efficiency of geneticin-resistant colony formation achieved by directly transfecting IB3 cells with the various

Flotte, Solow, Owens et al.: AAV Vectors in Airway Cells

A

KO

Figure 2. Expression of CAT activity in transfected IB3-l cells (A) or CFPAC-1 cells (B). CAT activity was assayed 48 h after lipofection with the indicated plasmids and after treatment of the cells either with (hatched bars) or without (open bars) 20 JLM forskolin. Activity is expressed as percent acetylation of the substrate.

Z c

70

60

60

50

50

40

40

30

30

20

20

10

10

1=

-e

...l

~

'"

u -e

. ~

;; 1=

u < f-o -e

w

pROII

AAV ps- and SV-neo plasmids is shown in Table 2. Large colonies were defined as those> 0.5 to 1 mm in diameter. Smaller colonies were seen at approximately 3 times greater frequency. pSV2neo produced geneticin-resistant colonies at a frequency of 0.8 X 10- 4 colonies per cell in the original transfection. Both pAAVpsneo and pAAVSVneo transfections showed a 7- to 8-fold increase in the number of colonies produced. This increased neo expression appeared more dependent on the presence of the AAV ITR sequence than on the promoter. This suggests that in IB3 cells the AAV ITR can function as an enhancer for the SV40 promoter. A similar observation has been reported in human HeLa cells (31). These experiments show that the AAV p- promoter was relatively efficient for stable expression of a selective marker in the IB3 cell line.

TABLE 2

Stable expression of neo in IB3-I cells Large Coloniest

pBR322 (control) pSV2neo PAAVSVneo pAAVpsneo

SO

70

pROI472

Plasmid*

353

Total Coloniest

Number

Frequency!

Number

Frequency

0 4 35 30

0 0.008 0.07 0.06

0 24

0 0.048 0.20 0.19

101

93

* Triplicate wells of 5 x lOS m3-l cells were transfected with 5 /tg of the indicated plasmid using lipofection and selection for gen' colonies as described in MATERIALS AND METHODS. t Only colonies> 0.5 mm in diameter were counted. t Includes COlonies < 0.5 mm. § Frequency is expressed as the percentage of cells plated yielding gen' corrected for a colony-forming efficiency of 10%.

pSV2CAT

pR0141f

pROII

pSV2CAT

Gene Transfer by AAVneo Transducing Particles The above experiments showed that the AAV ps promoter in the presence of the AAV ITR was an efficient promoter for transient gene expression in IB3 cells, as well as for stable expression of the neo gene. However, delivery of plasmid DNA by lipofection is generally not a very efficient system for stable integration and expression of DNA in mammalian cells. In contrast, AAV vector DNA can be packaged into AAV particles, and infection of cells with these AAV transducing particles is a much more efficient delivery system in several human cell lines. To test this in human CF airway epithelial cells, we first packaged the AAVpsneo vector into AAV transducing particles by co-transfecting adenovirusinfected 293 cells with the pAAVpsneo DNA, together with pAAV/Ad DNA to provide complementing rep and cap functions. The AAVpsneo transducing particles were concentrated from celllysates by CsCI equilibrium density gradient centrifugation. The AAVneotransducing particles were used to infect IB3 cells that were subsequently placed under geneticin selection. As shown in Figure 3, expression of the neo gene was highly efficient, and up to 60 to 70% of the cells were rendered geneticin resistant. Cells that were not infected with the AAVneo virus did not yield geneticin-resistant colonies, and cells that were infected with the AAVneo virus but not selected for geneticin resistance had a colony-forming ability at least90% of that for uninfected cells (not shown). This indicated that infection with the AAVneo particles was not toxic. In the experiments of Figure 3, 60 to 70% of the cells were transduced to geneticin resistance with an input of about 850 transducing particles per cell. For infectivity as-

354

AMERICAN JOURNAL OF RESPIRATORY CELL AND MOLECULAR BIOLOGY VOL. 7 1992

zw (.9

'J'.

>zw

u

isw

a: u,

z o

B

::J

o z

l/l

«a: f-

TRANSDUCING PARTICLES PER CELL

Figure 3. Expression of geneticin resistance in m3 cells by AAVneo transducing vector. m3 cells were infected with AAVneotransducing virus particles as indicated and selected for geneticin resistance as described in MATERIALS AND METHODS. Open and closed symbols show two independent experiments, and each point is the mean oftriplicate dishes of cells. The transduction frequency (percentage of cells rendered geneticin resistant) was calculated as described in MATERIALS AND METHODS.

says of wild-type AAV particles, which measure replication of AAV, one infectious unit is equivalent to about 40 to 100 AAV particles per cell, i.e., the particle-to-infectivity ratio of AAV is about 40 to 100 (38, 39). In the standard infectivity assay, a multiplicity of one infectious unit per cell is taken as the number of particles per cell required to infect 63 % of the cells (30). Thus, the results of Figure 3 suggest that in terms of particle ratios, the transducing titer of the AAV vector may be about 8-fold lower than the infectivity titer of wild-type AAV. Nevertheless, the vector was capable of stably expressing a foreign gene in the bulk of the cells.

Discussion Expression of genes in differentiated epithelial cell lines has already been accomplished by several investigators. Lipofection has been used to express CAT from the Rous sarcoma virus (RSV) LTR, human cytomegalovirus (CMV) IE and SV40 early promoters (20), and from the mouse mammary tumor virus LTR (40). Adenovirus vectors have been used to achieve high levels of expression of the at-antitrypsin gene, in vivo as well as in vitro (5). The expression systems previously used for the CF gene include Moloney murine leukemia virus-based vectors (2), vaccinia virus-based vectors (3), and baculovirus vectors (41). Stable high-level expression in an intact differentiated airway epithelium may still be difficult to achieve. For instance, baculovirus vectors do not function in mammalian cells, vaccinia vectors result in cell death, and retrovirus vectors may not function in airway cells where only a small percentage of cells are actively

dividing. The development of AAV vectors for stable highlevel expression in airway epithelia is, therefore, particularly relevant. Our present results suggest that AAV vectors can provide high-level, stable gene expression in a differentiated CF airway cell line. Previous studies showed that the SV40 promoter was as efficient as the RSV LTR or the CMV promoter for expression of the cat gene in IB3 cells using a lipofection-mediated transient assay (20). The transient assays with AAVp5cat vectors (Table 1) provided evidence for significantly higher activity (10- to 12-fold) of the P5 promoter compared with an SV40 early promoter in IB3 cells. These assays also suggested that the region upstream of AAV nucleotide 190 plays a significant role in the activity of the P5 promoter in IB3 cells. The region from AAV nucleotide 145 to 190 includes a consensus CREB/ATF motif, and this appears to be reflected in the responsiveness of the promoter to elevation of cAMP by forskolin. AAV nucleotides 1 to 145 constitute the ITR, and this region appears to contribute significantly to the activity of the P5 promoter in IB3 cells. This region appears also to have a more general enhancer function as suggested by the stable expression experiments (Table 2) using the neo vectors in which the activity of the SVneo vector was increased about 5-fold to the level of the AAVp5neo vector by addition of the AAV terminal region. Our studies may be consistent with studies by two other groups who analyzed the activity of an AAV P5 promoter in human HeLa cells. In one study, an AAV promoter containing only AAV nucleotides 190 to 310 had very low CAT expression in HeLa cells unless stimulated by expression of the adenovirus EIA transcriptional activator (31). In contrast, an AAV promoter containing nucleotides 144 to 310 had significant CAT activity in HeLa cells even in the absence of EIA (32). One difference between these observations that may be reconciled by our observation is the presence of the CREB motif between nucleotides 144 and 190. In the latter study (32), the AAV terminal region was also observed to have modest enhancer activity for expression from an SV40 promoter positioned immediately downstream, which is also consistent with our results in IB3-1 cells. More recently, the EIA response element in the AAV P5 promoter was shown to comprise a binding site for a transcriptional regulatory protein termed YY1, which acts as a repressor in the absence ofEIA (42). However, the function of YYl is complex, and three groups have now independently identified this factor as YYl (29,42),0 (43), or NF-El (44) and have shown that it acts as an activator or repressor depending on the promoter context (45). We have not evaluated the function of this YYl site directly in IB3 cells. Nevertheless, in the context of the entire AAV left-hand terminus present in our P5 vectors, our results suggest that the region upstream of the EIA (YY1) site may be more dominant in establishing the overall activity of the promoter at least in this IB3 CF airway epithelial cell line. In the context of the entire AAV genome, the P5 promoter is strongly autoregulated by the AAV rep gene product in transient assays (32). In cells having integrated latent AAV provirus, expression from any of the AAV promoters has not been detected. It is unclear if this reflects simple autoregulation by rep or effects of the conformation of the integration

Flotte, Solow, Owens et al.: AAV Vectors in Airway Cells

355

site. However, our evidence indicates that the ps promoter, like the P19 and P40 promoters (13, 18), is not intrinsically inactive when integrated. The apparent functionality of the ps promoter CREB site in IB3 cells is also of interest. This may provide a means of increasing expression of AAV vector-gene products in airway cells. After successful transfer of AAV vectors into cells, expression might be increased by using agents that increase intracellular cAMP, such as {3-adrenergic agonists. Transduction of 60 to 70 % of the cells which we achieved with the AAV vector in IB3 cells compares well with an earlier report of transduction of 70 % of the cells by an AAVSVneo transducing vector in human Detroit 6 cells (16). In both studies, a multiplicity of up to 800 to 1,000 particles was required to reach this level. The AAVSVneo transducing vector used in the Detroit 6 cells (16) was similar in structure to the pAAVSVneo vector we used in the experiments of Table 2, in that it contained an SVneo cassette inserted between AAV ITRs. The similar efficiency of the ps and SVneo viral vectors in transduction of cells may be accounted for partly by the enhancer effect of the AAV ITRs. These studies show that AAV vectors can yield very high transduction efficiencies. Our experiments show that in a CF airway epithelial cell line, the AAV ps promoter was efficient in the transducing vector. This may be particularly important for designing AAV vectors to deliver the CFTR gene because the coding region of this gene is at the packaging limit (about 4.5 kb) for an AAV vector, and it will thus be necessary to use a very compact promoter. In this respect, the Ps promoter provides a compact cassette with the AAV ITR that must be included in any AAV transducing vector. This has the added advantage that, at least in airway cells, the ITR may provide an enhancer effect for the promoter. It remains to be determined if efficient expression in 60 to 70 % of the cells of the respiratory tract in vivo will be achievable. In any case, the AAV2-airway cell expression system may have other useful applications including functional analysis of CFTR in vitro. It is also possible that AAV vectors may prove useful for expression of other physiologically important genes (antioxidants, antiproteases, or tumor suppressor proteins) in the respiratory or gastrointestinal tract. This could be helpful in understanding or treating a variety of disease processes in those cell types.

5. Rosenfeld, M. A., W. Siegfried, K. Yoshimura et al. 1991. Adenovirusmediated transfer of a recombinant alpha I-antitrypsin gene to the lung epithelium in vivo. Science 252:431-434. 6. Carter, B. J. 1990. The growth cycle of adeno-associated virus. In Handbook ofParvoviruses. P. Tijssen, editor. Vol. I. CRC Press, Boca Raton, FL. 155-168. 7. Kotin, R. M., M. Siniscalo, R. J. Samulski et al. 1990. Site-specific integration of adeno-associated virus. Proc. Natl. Acad. Sci. USA 87: 2211-2215. 8. Samulski, R. J., X. Zhu, X. Xiao et al. 1991. Targeted integration of adeno-associated virus (AAV) into human chromosome 19. EMBO J. 10:3941-3950. 9. Kotin, R. M., J. C. Menninger, D. C. Ward, and K. I. Berns. 1991. Mapping and direct visualization of a region-specific viral integration site on chromosome 19q13-qter. Genomics 10:831-834. 10. Hermonat, P. L., M. A. Labow, R. Wright, K. I. Berns, and N. Muzyczka. 1984. Genetics of adeno-associated virus: isolation and prelintinary characterization of adeno-associated virus type-2 mutants. J. Viral. 51: 329-339. II. Tratschin, J.-D., I. L. Miller, and B. J. Carter. 1984. Genetic analysis of adeno-associated virus: properties of deletion mutants constructed in vitro and evidence for an adeno-associated virus replication function. J. Viral. 51:611-619. 12. Hermonat, P. L., and N. Muzyczka. 1984. Use of adeno-associated virus as a mammalian DNA cloning vector. Proc. Natl. Acad. Sci. USA 81:6466-6470. 13. Tratschin, J.-D., I. L. Miller, M. G. Sntith, and B. J. Carter. 1985. An adeno-associated virus vector for high frequency integration, expression and rescue of genes in mammalian cells. Mol. Cell. Bioi. 5:3251-3260. 14. Lebkowski.T. S., M. M. McNally, T. B. Okarma, andL. B. Lerch. 1988. Adeno-associated virus: a vector system for efficient introduction and integration of DNA into a variety of cell types. Mol. Cell. Bioi. 8:39883996. 15. McLaughlin, S. K., P. Collis, and N. Muzyczka. 1988. Adeno-associated virus general transduction vectors: analysis of proviral structures. J. Viral. 62:1963-1973. 16. Samulski, R. J., L.-S. Chang, and T. Shenk. 1989. Helper free stocks of recombinant adeno-associated viruses: normal integration does not require viral gene expression. J. Viral. 63:3822-3828. 17. Tratschin, J.-D., M. P. H. West, T. Sandbank, and B. J. Carter. 1984. A human parvovirus, adeno-associated virus, as a eucaryotic vector: transient expression and encapsidation of the procaryotic gene for chloramphenicol acetyltransferase. Mol. Cell. Bioi. 4:2072-2081. 18. Mendelson, E., M. G. Sntith, and B. J. Carter. 1988. Expression and rescue of a non-selected marker from an integrated AAV vector. Virology 166:154-165. 19. Wondisford, F. E., S. J. Usala, G. S. De Cherney et al. 1988. Cloning of the human thyrotropin beta-subunit gene and transient expression of biologically active human thyrotropin after gene transfection. Mol. Endocrinol. 2:32-39. 20. Luo, L., P. L. Zeitlin, W. B. Guggino, andR. W. Craig. 1989. Gene transfer by lipofection in rabbit and human secretory epithelial cells. Pflugers Arch. 415: 198-203. 21. Zeitlin, P. L., L. Lu, 1. Rhim etal. 1991. A cystic fibrosis bronchial epithelial cell line: immortalization by adeno-12-SV40 infection. Am. J. Respir. Cell Mol. Bioi. 4:313-319. 22. Schoumacher, R. A., J. Ram, M. C. Iannuzzi et al. 1990. A cystic fibrosis pancreatic adenocarcinoma cell line. Proc. Natl. Acad. Sci. USA 87: 4012-4016. 23. Graham, F. L., J. Sntiley, W. C. Russell, and N. Naiva. 1977. Characteristics of a human cell line transformed by adenovirus DNA. J. Gen. Viral. 36:59-72. 24. Sambrook, J., E. F. Fritsch, and T. Maniatis. 1989. Molecular Cloning: A Laboratory Manual. Cold Spring Harbor Laboratory, Cold Spring Harbor, NY. 25. Laughlin, C. A., J.-D. Tratschin, H. C. Coon, and B. J. Carter. 1983. Cloning of infectious adeno-associated virus genomes in bacterial piasntids. Gene 23:65-73. 26. Tratschin, J.-D., J. Tal, and B. J. Carter. 1986. Negative and positive regulation in trans of gene expression from adeno-associated virus vectors in mammalian cells by a viral rep gene product. Mol. Cell. Bioi. 6:28842894. 27. Southern, P., and P. Berg. 1982. Transformation of mammalian cells to antibiotic resistance with a bacterial gene under control of the SV40 early promoter region. J. Mol. Appl. Genet. 1:327-341. 28. Samulski, R. J., L.-S. Chang, and T. Shenk. 1987. A recombinantplasntid from which an infectious adeno-associated virus genome can be excised in vitro and its use to study viral DNA replication. J. Viral. 63:38223828. 29. Proudfoot, N. 1991. Poly(A) signals. Cell 64:671-674. 30. Carter, B. J., C. A. Laughlin, L. M. de la Maza, and M. Myers. 1979. Adeno-associated virus auto-interference. Virology 92:499-462. 31. Chang, L.-S., Y. Shi, and T. Shenk. 1989. Adeno-associated virus p5 pro-

Acknowledgments: We thank J. Tal for constructing pYT45 and R. Frizzell for providing CFPAC-I cells. This work was funded in part by a Cystic Fibrosis Foundation Grant to B. Carter. T. Flotte was supported by a Cystic Fibrosis Foundation Johns Hopkins-NIH Fellowship in the Endwood Division of Respiratory Disease.

References 1. Boat, T., M. J. Welsh, and A. Beaudet. 1989. Cystic fibrosis. In The Metabolic Basis of Inherited Disease. C. Scriver, A. Beaudet, W. Sly, and D. Valle, editors. McGraw-Hill, New York. 2649-2860. 2. Drumm, M. L., H. A. Pope, W. H. Cliff et al. 1990. Correction of the cystic fibrosis defect in vitro by retrovirus-mediated gene transfer. Cell 62:1227-1233. 3. Rich, D. P., M. P. Anderson, R. J. Gregory et al. 1990. Expression of cystic fibrosis transmembrane conductance regulator corrects defective chloride channel regulation in cystic fibrosis airway epithelial cells. Nature 347:358-363. 4. Riordan, J. R., J. M. Rommens, B. Kerem et al. 1989. Identification of the cystic fibrosis gene: cloning and characterization of complementary DNA. Science 245: 1066-1073.

356

32. 33. 34. 35. 36. 37. 38. 39.

AMERICAN JOURNAL OF RESPIRATORY CELL AND MOLECULAR BIOLOGY VOL. 7 1992

moter contains an adenovirus EIA-inducible element and a binding site for the major late transcription factor. J. Virol. 63:3479-3488. Beaton, A., P. Palumbo, and K. I. Berns. 1989. Expression from the adeno-associated virus p5 and pl9 promoters is negatively regulated in trans by the rep protein. J. Virol. 63:4450-4454. Hai, T., F. Liu, E. A. Allegretto, M. Karin, and M. R. Green. 1988. A family of immunologically related transcription factors that includes multiple forms of ATF and AP-1. Genes Dev. 2:1216-1226. Lin, Y.-S., andM. R. Green. 1988. Interaction ofacommon cellular transcription factor, ATF, with regulatory elements in both EIA- and cyclic AMP-inducible promoters. Proc. Natl. Acad. Sci. USA 85:3396-3400. Montminy, M., and L. M. Bilezkjian. 1987. Binding of a nuclear protein to the cyclic-AMP response element of the somastatin gene. Nature 328:175-178. Montminy, M., G. A. Gonzalez, and K. K. Yamamato. 1990. Regulation of cAMP-inducible genes by CREB. TINS 13:184-188. Sassone-Corsi, P. 1988. Cyclic AMP induction of adenovirus promoters involves sequences required for EIA trans-activation. Proc. Natl. Acad. Sci. USA 85:7192-7196. De la Maza, L. M., and B. J. Carter. 1980. Heavy and light particles of adeno-associated virus. J. Virol. 33:1129-1137. Laughlin, C. A., M. W. Myers, D. L. Risin, and B. J. Carter. 1979. Defective-interfering particles of the human parvovirus adeno-associated virus. Virology 94: 162-174.

40. Hazinski, T. A., P. A. Ladd, andC. A. DeMatteo. 1991. Localization and induced expression of fusion genes in the rat lung. Am. J. Respir. Cell Mol. Bioi. 4:206-209. 41. Kartner, N., J. W. Hanrahan, T. J. Jensen et al. 1991. Expression of the cystic fibrosis gene in non-epithelial invertebrate cells produces a regulated anion conductance. Cell 64:681-691. 42. Shi, Y., E. Seto, L.-S. Chang, and T. Shenk. 1991. Transcriptional repression by YY I, a human Gl.l-Kruppel-related protein, and relief of repression by adenovirus EIA protein. Cell 67:377-388. 43. Hariharan, N., D. E. Kelley, and R. P. Perry. 1991. 0, a transcription factor that binds to downstream elements in several polymerase II promoters, is a functionally versatile zinc finger protein. Proc. Natl. Acad. Sci. USA 88:9799-9803. 44. Park, K., and M. L. Atchison. 1991.Isolation ofa candidate repressor/activator, NF-EI (YY-I, 0), that binds to the immunoglobulin K 3' enhancer and the immunoglobulin heavy-chain/LEI site. Proc. Natl. Acad. Sci. USA 88:9804-9808. 45. Seto, E., Y. Shi, and T. Shenk. 1991. YYI is an activator sequence binding protein that directs and activates transcription in vitro. Nature 354:241245. 46. Srivastava, A., E. W. Lusby, and K. I. Berns. 1983. Nucleotide sequence and genome organization of the adeno-associated virus 2 genome. J. Viral. 45:555-564.