viRoiocy178,419

428(1990)

Retrovirus

Vector-Targeted

Inducible

JOHN F. ENGELHARDT, OnCOlOgY

Center

MERRILLJ.

and *Department School Keceived

Expression KELLUM,

FATINA B&AT,

of Molecular B/ology of Medicine, Bait/more, February

of Human ,&interferon AND

PAULA M PITHA’,*

and GenetIcs, The Johns Hopkins Maryland 2 7205

5, 1990; acceptedJune

Gene to B-Cells

Unwerstty

1, 7990

We have introduced the human b-interferon gene with its promoter region into murine B-cell and fibroblast cell lines via a Moloney murine leukemia virus (M-MuLV) vector and have studied the inducible expression of the o-interferon gene as a function of the various retroviral vector designs. By deleting the enhancer within the 3’ viral long terminal repeat (LTR), inserting the human @-interferon gene, and varying placement of the immunoglobulin heavy chain enhancer, we were able to construct vectors which yielded proviruses with various cell type-specific regulation. One of the vectors (pT154) led to a greater than 21 -fold increase in &interferon protein synthesis after viral infection in the two B-cell lines analyzed, while no inducibility was seen in the fibroblast cells. The data show that inducible p-interferon expression within a MuLV vector was highly dependent on the absence of the viral enhancer region in the long terminal repeat and the orientation of the &interferon gene within the proviral transcriptional unit; the insertion of the immunoglobulin enhancer elevated both constitutive and (or) inducible expression of p-interferon in B-cells but inhibited constitutive expression of this gene in fibroblasts. Q 1990Academic press. hc

INTRODUCTION The targeted expression of inducible cytokine genes such as interferon to tumor cells of hematopoietic lineage containing deletions of both cy- and P-interferon genes (Diaz et al., 1988), or to cells that are chronically infected by viruses [such as human immunodeficiency virus (HIV), hepatitis-B virus (HBV), or Epstein-Barr virus (EBV)] which are often not able to express the endogenous interferon gene efficiently (Pitha eta/., 1988; Bednarik et al., 1989), could possibly prove to be a very effective alternative for interferon therapy(Pitha, 1987). This approach would provide both high and continuous interferon synthesis in the target cells that is often required for the suppression of the virus infection or its antitumor effect and could eliminate the side effects which accompany the administration of high levels of an exogenous interferon. To evaluate the feasibility of such a gene targeting strategy, we attempted to elicit inducible, tissue-specific expression of the human pinterferon gene inserted within a murine leukemia virus (MuLV)-based retroviral vector. Retroviral vectors have been used in numerous cases to efficiently and stably transfer genes into the recipient cellular genomes of a variety of cell types both in vitro and in vivo (Williams er al., 1986; Armentano et a/., 1987; Mclvor et a/., 1987). However, in a majority of these studies, the retroviral promoter in the long terminal repeat (LTR) was utilized to express the internally inserted gene (Temin, 1986; Morgan et al., 1987; ’ To whom)

requests

for reprints

should

be addressed.

Schwarzbauereral., 1987; St. Louis and Verma, 1988). There are two main limitations to this approach: First, the viral LTR has often been found to be transcriptionally inactivated by methylation (Jahner eta/., 1982), and second, the expression via the viral LTR is unregulated. In this study, we wanted to conserve the regulated expression of the o-interferon gene, but impose on it Bcell-specific transcription. In so doing, we asked two basic questions: (1) can expression of o-interferon (under the direction of its own promoter) be correctly regulated and inducible when inserted into a retroviral vector and (2) is it possible to achieve a cell type-specific inducibility of the human p-interferon gene by creating a dimeric enhancer/promoter regulatory unit within the retroviral vector? To achieve a cell type-specific expression, we chose the murine immunoglobulin heavy chain enhancer (IgEnh) [shown to be a tissue-specrfic transcriptional enhancer in lymphocytes (Banerji et a/., 1983; Gillies et al., 1983)] as a prototype and examined whether the insertion of the IgEnh 5’ or 3’ of the interferon gene would lead to a cell type-specific inducible expression of human P-interferon wrthin munne B-cells. There are relatively few examples of inducible promoters used within retroviral vectors (Selden er al., 1987; Cone et a/., 1987; Hatzoglou et a/., 1988) and, from these, only the @-globin promoter was found to be inducible in a cell type-specific manner. The results of this study show that the inducible and ttssue-specrfrc expressron of @-Interferon wrthtn the retroviral vector used was greatly influenced by the dele-

420

ENGELHARDT

tion of the LTR enhancer, as well as by the orientation of both the p-interferon gene and IgEnh within the vector. Although the insertion of the IgEnh increased constitutive expression of P-interferon in B-cells, the inducibility in different cell types was affected by the position of IgEnh with respect to the interferon gene. MATERIALS

AND

METHODS

Cell lines The I+-2 cell line, NIH/3T3GV mouse fibroblasts [a subline of NIH/3T3GV (Jainchill eta/., 1969; Andersson et al., 1979)], and SC-1 murine fibroblast (Hartley and Rowe, 1975) were maintained in Dulbecco’s modified Eagle’s medium (GIBCO, Grand Island, NY) with 10% fetal calf serum. A20 murine B-cells (Kim et al., 1979; Glimcher et a/., 1982) and M 12.4.5 murine hybridoma cells (Hamano et a/., 1982) were maintained in OPTIMEM (GIBCO) with 5% fetal calf serum. Construction

of retroviral

vectors

The manipulations used in the cloning of various restriction fragments were performed as described by Maniatis et al. (1982). The pL.J (or Dol) (Korman et al., 1987) and Zipneosv(x) (Cepko et al., 1984) vectors were obtained from Dr. R. C. Mulligan. Zipneoenhwas constructed from zipneosv(x) by deleting the 160bp PvulIIXbal viral enhancer fragment from the 3’ LTR. The plJ-modified vectors, pUenh+ and pUenh-, were constructed by cloning the SVneo ClaIIBamHI fragment of pU into the zipneosv(x) backbone. The human P-interferon EcoRIIHindlII genomic fragment, which contains the endogenous promoter, coding region, and polyadenylation site, was cloned into the BamHl site of the pUenh+ (pU) and pllenhvectors in both orientations to construct pB8, pB3, pB6, and pB1 (Fig. 1). The EcoRIIXbal fragment of the genomic murine immunoglobulin heavy chain enhancer (Gillies et al., 1983) was undirectionally cloned 5’ and 3’ to the ,&interferon genes within pB1 to give pT159 and pT149 as well as within pB3 to give pT154 and pT166 (Fig. 1). $2 packaging

cell lines

Plasmid DNA containing the retroviral constructs was purified by CsCl banding and transfected into the q-2 ecotropic packaging cell line (Mann et a/., 1983) by calcium phosphate precipitation as previously described (Bisat et a/., 1988). Briefly, cells were seeded at a density of 5 X 1 O5 tells/60-mm plate and after 24 hr, plasmid DNA (10 pg) was added as a calcium phosphate precipitate. After a 16-hr incubation, cells were washed with serum-free DMEM, treated with 15% glycerol for 60 set, washed twice with phosphate

ET AL.

buffer saline (PBS), and incubated for 24 hr in DMEM supplemented with 5 mM sodium butyrate. Forty-eight hours after glycerol shock cells were trypsinized and split 1: 10 and grown in medium containing 1 mg/ml of G418 (GIBCO). Resistant clones were selected after 10 to 14 days and individual colonies were grown into cultures and tested for virus production. Virus carrying a selectable neo gene was harvested from a confluent monolayer of rc/-2 cells following a 16-hr exposure to fresh medium and was titered on NIH/3T3GV cells. Infections were carried out in 1 ml DMEM supplemented with 10% fetal calf serum and 10 Fg/ml of polybrene (Sigma, St. Louis, MO)for 2 hrwith intermittent rocking. Following infection, 4 ml of fresh media were added and cells were incubated for an additional 16 hr. Cells were then trypsinized, split 1: 10, and incubated in medium containing 1 mg/ml G418. Drug-resistant colonies were counted 12 days later. Production

of proviral

cell lines

Supernatants from q-2 cells (pools with greater than 200 colonies) containing the various retroviral constructs were used to infect A20, Ml 2.4.5, NIH/3T3GV, and SC-l cells at an m.o.i. of less than 0.001. Infections were carried out as described for titration of J/-2 cell line viral stocks on NIH/3T3GV. Plates containing greater than 200 colonies from a single infection were combined to give infected cell line pools. The absence of recombinant virus production by infected cell lines was confirmed by assaying supernatants from infected NIH/3T3GV and A20 cells for the production of virus carrying the neomycin gene by reinfecting NIH/3T3GV cells and by assaying reverse transcriptase activity of the supernatants (Adachi et a/., 1986). Interferon

induction

NIH/3T3GV and SC-l cells were seeded at 5 X lo5 tells/35-mm dish 24 hr before inductions and then infected with Newcastle disease virus (NDV) (LaSota strain) at an m.o.i. of 10 for 4 hr in 1 ml serum-free medium. Following infection 1 ml of DMEM with 10% fetal calf serum was added and cultures were incubated for an additional 20 hr. Supernatants were collected, spun at 3000 g for 15 min, acidified to pH 2.0, stored at 4“ for 5 days to destroy NDV, and then neutralized and stored at -70” until assayed for interferon. A20 and M 12.4.5 B-cell cultures were similarly induced except that they were seeded at 5 X 1O6 tells/35-mm dishes in 1.5 ml OPTI-MEM without fetal calf serum and infected with an m.o.i. of 2. Following infection, 0.5 ml OPTIMEM with 15% fetal calf serum was added to each culture. All inductions were done in triplicate.

TARGETING

Interferon

OF THE

INTERFERON

assays

Medium from induced cells was serially diluted in DMEM with 3% fetal calf serum and biologically active interferon was assayed in triplicate by the inhibition of the cytopathic effect of encephalomyocarditis (EMC) virus on L-cells (murine interferon) and vesicular stomatitis virus (VSV) on human primary fibroblasts (human interferon) (Finter, 1969). The level of interferon antiviral activity in each sample was quantitated by comparison with standards of mouse a- and P-interferon (Lee Biomolecular, San Diego, CA) and human p-interferon (Triton Biosciences Inc., Alameda, CA). Induction

and assay of interferon

mRNA

RNA inductions were carried out as inductions for interferon protein except that cells were harvested 8 hr after NDV infection or 6 hr after treatment with medium containing cycloheximide (5 pg/ml). Following induction, cells were washed twice with phosphate-buffered saline (PBS) and total RNA was prepared by the guanidinium isothiocyanate method (Raj and Pitha, 1983). Northern hybridization was done as previously described (Raj and Pitha, 1981) using riboprobes to the coding region of human and murine P-interferon and random-primed DNA probes for actin, P-interferon, LTR, and neomycin. All probes were used at a concentration of 1 x lo6 cpm/ml; hybridizations with riboprobes were done for 16 hr at 65” and with DNA probes for 36 hr at 32”. The relative levels of mRNAs were determined by densitometric scanning of the autoradiogram. Intensities were normalized to a constant level of endogenous actin or endogenous mouse interferon mRNA. RESULTS Generation of virus producing $2 packaging cell lines To obtain inducible expression of the P-interferon gene, we deleted the enhancer region from the 3’ LTR of MuLV-based retroviral vector; since the inactive LTR would not allow efficient expression of the neomycin gene (used as a marker for transfection), we used the pU vector in which the expression of neomycin was directed by the SV40 promoter region. Vector constructs shown in Fig. 1 were transfected into the #-2 packaging cell line and then transfected cells were grown in G418 for 14 days. Virus containing the recombinant genome was harvested from a pool (more than 200 colonies) of transfected clones and titrated by conferring neomycin resistance to NIH/3T3GV cells. Viral titers obtained from the pool of transfected cells are shown in Table 1. Only pools of transfected #-2 cells

GENE

421

EXPRESSION

were used for the preparation of viral stocks to avoid the selection of a clone which might contain a rearrangement of the P-interferon gene within the vector and a consequent alteration In the inducible expression. Synthesis of human B-interferon protein fibroblasts and B-cells containing the human B-interferon gene

in murine

Murine fibroblast (NIH/3T3GV and SC-l cells) and Bcells (A20 and Ml 2.4.5) were infected with recombinant virus at a m.o.i. less than 0.001, and the infected cells were selected by the resistance to G418. The infected cells did not produce any neomycin-contalning recombinant virus by criteria described under Materials and Methods. Southern analysis of the genomic DNA isolated from infected NIH/3T3GV and A20 individual clones revealed the presence of a single integration site and insertion of a single copy of the human P-interferon gene (data not shown). For the study of the interferon induction by NDV infection In these cell lines, we used pools of infected cells, rather than single clones, to avoid both clonal cell variability in interferon inducibility and the position effect of retroviral integration on expression. While the fibroblast lines (NIH/3T3GV and SC-l cells) produced only murine @-interferon, the M 12.4.5 and A20 cells produced both murine (Y- and p-interferons (unpublished results). The results in Fig. 1 show the differences in the levels of the induced endogenous murine interferon synthesized in the four cell lines tested. The NIH/3T3GV cells were a more efficient producer of the endogenous murine interferon than A20 or SC-1 cells. Highest levels of mouse interferon could be detected in induced M. 12.4.5 cells; however, the majority of this interferon was identified as an (Y type (unpublished results). With the exception of the A20 lines, the variability in the levels of endogenous interferon induced was not significant. The levels of endogenous interferon induced in one of the infected A20 lines (pB3) was low; we believe that this difference may represent a clonal expansion of a poorly inducible clone during the G418 selection. None of the lines tested produced detectable levels of mouse interferon, constitutively. Constructs

containing

human &interferon

gene

Analysis of the relative levels of human P-interferon synthesized in the A20 and NIH/3T3GV cells containing the human p-interferon gene inserted either in the pU or pLJenh- vector (pB8, pB3, pB1, and pB6) showed a low constitutive synthesis of human p-interferon protein (Fig. 1). Since the human &interferon promoter region is very tightly regulated and not expressed in human cells in vitro without induction, we

ENGELHARDT

422

ET AL. Interter~miIUlml) A20

Rebovirel

Human f-IFN

Cor~olructs

LTR

‘V’

in Pro viral

Cew

H2.4.5

NIH3T3

SC-I

Mouse IEM

Human E!a

Mouse EM

Human Elm

3.6(0.9)

420(120)

220(52) 420(37)

62500(O)