WO1985005629A1

WO1985005629A1 - Ltr vectors, methods of preparation and use

Info

Publication number: WO1985005629A1
Application number: PCT/US1985/000986
Authority: WO
Inventors: William A. Haseltine; Craig A. Rosen; Jack R. Lenz; Daniel W. Celander
Original assignee: Dana-Farber Cancer Institute
Priority date: 1984-05-25
Filing date: 1985-05-24
Publication date: 1985-12-19
Also published as: AU4404785A; EP0181930A1; JPS61502932A

Abstract

Process for the tissue specific expression of heterologous genes using modified LTR vectors. Any retroviral LTR can be synthetically modified in its enhancer region using the methods described herein and tissue specificity can thereby be achieved. Also described is an assay useful in proving or predicting tissue specificity of the modified LTR vector. This invention is also directed to novel tissue specific LTR vectors. Long terminal repeats (LTRs) of Moloney sacoma virus (MSV) have been shown useful in activating the expression of a normally quiescent cellular DNA sequence by Vande Woude et al., in U.S. Patent No. 4,405,712. The present invention demonstrates the use of modified mammalian LTR vectors for obtaining selectable tissue expression of heterologous genes.

Description

LTR VECTORS, M ETHODS OF PREPARATION AND USE

FI ELD OF T H E INVENTION

The present invention relates to the use of viral vectors in recombinant DNA manipulation and in therapeutic and analytical proced ures. More particularly, it relates to the use of modif ied mammalian long terminal repeat (LTR) vectors for obtaining selectable tissue exp ression of heterologous genes.

BACKGROUND OF TH E INVENTION

M uch effort has been spent over the years in attempting to understand the mode of action of viruses, pa rticularly that of retroviruses. Questions for which answers have been sought include the reasons that certain of such viruses preferentially inf ect and/or replicate in certain types of cells as opposed to other types of cells.

Conventional wisdom has held that the ability of the virus to infect and/or replicate in ce rtain types of cells has been controlled by a segment of the viral genome related to the production of the envelope of the virus (i.e. , the envelope (env) gene) . Although there has been some suggestion that the LTR segments of the viral genome may be involved, (Des Groseillers, L. , a normally quiescent cellular DNA sequence by Vande Woude et al., U. S. Patent No. 4,405,712. See also Gluz man, et al., (ed.) , Enh ancers and Euk a ryotic Gene

Exp ression. Cold Spring Harbor Laboratory (1983) . However, such systems have not taken advantage of the tissue tropic enhance r regions of LTRs.

SUMMARY OF THE INVENTION

This invention is directed to a process for the tissue specific exp ression of heterologous genes using modified LTR vectors. Any retroviral LTR can be synthetically modified in its enhancer region using the methods described herein and tissue specificity can thereby be achieved. Also described is an assay useful in analyzing or predicting tissue specificity of LTR vectors. This invention is also directed to novel tissue specific LTR vectors.

In accordance with the present invention, tissue specific vectors are constructed, using a tissue specific enhance r(s) operatively positioned in the same sequence with a heterologous DNA segment corresponding to the polypeptide of interest, as well as a stop codon and polyadenylation sequence downstream (3 ') from that gene. The vector should also contain a replication origin, as is known in the art.

The vector contains at least the segment of an enhancer which determines the tissue specificity of that enhancer, hereinafter referred to as the "tissue specificity determinant." The vector preferably contains a complete viral enhancer which preferably includes the tissue specific determinant for the desired tissue. The promoter contained in the vector can be any of the known promoters which function to permit expression of the desired product in the host of choice. Preferably the promoter is a viral promoter from the same class of virus in the enhancer. The preferred class of virus is retrovirus, and the preferred viruses for use in conjunction with the invention are the Akv, SL3-3, and F riend viruses.

The term "tissue specific" as used in this disclosure and claims, means that the vector operates to produce a g reater amount of desired product in the targeted tissue than it does in other tissues under normal culture conditions. Tissue specific vectors may produce 1.5 to 1,000 or more times as much expression product in the target tissue as in other tissues.

The tissue specific determinant can be homologous, meaning it came from the same virus as the promoter, or heterologous, in which case it is not from the same virus as the promoter. Heterologous tissue specific determinants can be excised f rom other viral systems, or can be synthesized using known techniques. Tissue specific determinants which are specific to the target tissue can be identified by assay techniques, where vectors encoding an indicator or marker compound, e.g., chloramphenicol acetyl transf erase (CAT) , an indicator which can be easily quantified as described below, to determine which vectors are effective in the tissue.

If desired, enhancer(s) from tissue specific vectors can be compared in DNA sequence to the enhancers which are not specific to the target tissue to determine its DNA sequence of the tissue specific determinant. Thereafter, an enhancer which contains the tissue specific determinant may be utilized in the desired vector containing the gene which it is desired to express, and the resulting tissue specific vectors utilized to express the desired product in the tissue of choice.

Tissue specific expression is highly advantageous in the production of large quantities of desired product from the chosen tissue. It minimizes expression in other tissues. Such techniques can be used to characterize or identify tissue, to foster growth of some tissues over others, to minimize the amount of contaminants produced, and for other purposes.

The vectors of the present invention can be in the form of plasmids, or viral vectors, such as those produced in accordance with the procedure described by Mann et al., Cell, 33: 153 (1983) .

The vectors constructed in accordance with this invention can also be utilized in vivo, e.g., for the treatment of diseases caused by genetic or other defects wherein certain types of cells produce superabundant or damaging amounts of particular materials. In the latter case, such excess materials can be eradicated for example, by expressing the gene which codes for an enzyme or other material which reacts with the overproduced material. Other methods of genetic treatment utilizing the present invention are described below.

Where the tissue specific vectors of the present invention are utilized in vivo, the vectors used are preferably viral vectors as described above. DES CRIPTION OF THE FIGURES

Figures 1A, 1B, and 1C illustrate the isogenic constructs made when the U3 region of Akv (pAU3CAT) and SL3-3 (pU3CAT) were placed 5' to the chloramphenicol acetyl transf erase (CAT) gene;

Figu res 2(a) - (d) illustrate the dramatic difference in the CAT activity, seen between the Akv and SL3-3 , LTR sequences; and

Figure 3 is a compa rison of the seq uence organization within the tandem repeat regions of the Akv (Van Beveren, C., et al., J. Virol., 41: 542 - 556 (1982) , SL3-3 (Lenz, J. , et al., J. Virol., 47: 317 - 328 (1983) and Fr-MuLV57 (Olif f, A.I., et al., J. Virol., 33: 475 - 486 (1980) ) LTRs. The sequence of a single 99 bp repeat element of Akv is shown at the top. The solid lines below represent identical sequences present in the 72 bp repeats and 66 bp repeats of SL3-3 and Fr-MuLV57 respectively. The number after the b racketed region indicates how many times their sequence is repeated. Deletions are indicated by the delta sy mbol and insertions are indicated by an up arrow. The actual sequences in the asterisked areas are shown in (b) . These sequences are not present in the repeat region of Akv.

DETAILED DESCRIPTION OF TH E PRE FERRED EMBODIM ENTS

To demonstrate the region of the viral genome which confers upon the virus the ability to potentiate disease, a series of recombinants were constructed between infectious proviruses SL3-3 and Akv. SL3-3 is an N tropic, ecotropic, potent leukemogenic virus isolated f rom a spontaneous T cell lymph oma of an AKR mouse. Although SL3-3 resembles the endogenous, N tropic Akv virus in both viral genome structure and replication properties in tissue culture, SL3-3 is capable of inducing T cell leukemias in a wide variety of mouse strains, whereas Akv is non-leukemogenic.

Virus produced upon transfection of NIH 3T3 with the SL3-3 and Akv recombinant proviruses was tested for the ability to induce disease. One such recombinant, a virus that contains all of the Akv structural genes flanked by SL3-3 LTR sequences, induced T cell leukemias in several strains of mice. Nucleotide sequence analysis of the SL3-3 LTR revealed that the only difference distinguishing the LTR of the SL3 from that of the avirulent Akv virus, was located within the repeat element present in the U3 region, (with the exception of a single nucleotide change near the 5 ' end of the LTR) . This repeat region, known to function as an enhancer element in related viruses, may permit viral gene expression to occur in the appropriate cellular context. For this reason the ability of the LTR sequences of SL3-3 and Akv to function as transcriptional elements in several cell types were tested.

Inoculation of susceptible strains of mice with the SL3-3 strain of murine leukemia viruses induces T cell lymph omas, whereas injection of the Akv strain does not (see, Lenz, J. , et al., J. Virol., 43: 943-951

(1982) ; Pedersen, F. S., et al., Natu re, 292: 167-170 (1981) ; and Lenz, J., et al., J. Vi rol . , 47: 317-328 (1983) ) . Recombinants viruses that contain the long terminal repeat (LTR) of the SL3-3 virus and the g ag, pol and env genes of the Akv virus also induce T-cell leukemias (see, Lenz, J., et al., Nature. 308: 467-470 (1984)) , demonstrating that the major determinant of leudemogenicity is located within the LTR region. Recombinant viruses that possess the LTR region of SL3-3 also replicate to high titers in thymocytes, but not in marrow or spleen cells of infected animals. The avirulent Akv virus itself does not replicate to high titers in any of these tissues.

The cell type specificity of leukemias induced by viruses containing different LTR seq uences is due in part to the ability of the virus tp replicate in the approp riate cellular environment (see, DesG roseillers, L. , et al., PNAS USA, 80: 4203-4207 (1983) ; Chatis, P. A. , et al., PNAS USA, 80 : 4408- 4411 (1983) ) .

One explanation of the role of the LTR in determination of both cell tropism and leuk emogenic potential is that the LTR encodes tissue permissive transcriptional elements. It has been discovered that there are differences in the transcriptional activity of the SL3-3 and Akv LTR sequences in different murine cell types, and that the sequences present in the LTR of SL3-3 exhibit significantly enhanced transcriptional activity in T cells as compared to the cor responding region of the Akv LTR. The results suggest that transcriptional elements are primary determinants of cell tropism and of leukemogenicity of these viruses.

To assess the ability of the Akv and SL3-3 LTR sequences to serve as promoter elements in different cellular environments, isogenic constructions (see, Maniatis, T. , et al., "Molecular Cloning, A Laboratory Manual" Cold Spring Harbor Laboratory. (1982) ) were made in which the U3 regions of Akv (pAU3 CAT) and SL3-3 (pSU3CAT) were placed 5' to the chloramphenicol acetyl transferase (CAT) gene (see. Figu re la and Gorman, C. M., et al., Mol. Cell. Biol. , 2: 1044-1051 (1982) ) . These plasmids were introduced into cultured cells using either the calcium phosphate (see, Wigler, M. , etal., Cell, 14: 725-731 (1978) ) or the DEAE-dextran methods (see, Stafford, J. et al., Natu re, 306: 77-79 (1983) ) . After a transient expression period of 44-50 hours, cell extracts were prepared, and the level of CAT enzymatic activity was measured (Gorman et al., s upra) .

Previous investigators have demonstrated that for constructions such as those used here, for which the sequence surrounding the RNA start sites are identical, and which differ only in sequences remotely located to the start site (Lenz et al., Natu re, supra) , the level of indicator gene activity cor responds directly to the transcription rate (see for example, Keller, J. M. , et al., Cell, 36: 381-389 (1984) ; He rrera- Estrella, L. , et al., Natu re, 310: 115-120 (1984); McKnight, S. L. , et al., Cell, 25: 385-398 (1981) ; and Spandidos, D. A. et al., EMB O J. , 2 : 1193-1199 (1983) ) .

To determine the level of non-specific and specific expression of the CAT gene in murine cells respectively, parallel transfection experiments were conducted in which cultured cells were exposed to plasmids that contained either no eukaryotic promoter element (pSVOCAT) (Gorman et al., supra) or the 3 '-LTR of Rous sarcoma virus (pRSVCAT) (see, Gorman, C., et al., PNAS USA, 77: 6777-6781 (1982) ) located 5' to the CAT gene.

Various cell lines differ in their ability to take up and express transfected DNA. In each expe riment the CAT activity directed by the plasmids that contained murine LTR sequences was normalized to that of plasmids that contained the 3 ' LTR of Rous sarcoma virus (RSV) . The transcriptional elements in RSV 3' LTR have been shown to function in a wide variety of cell types and have been utilized as relatively neutral reference promoters in similar studies (Gorman, et al., supra; and M. D. Walker, et al., Nat ure, 306: 557 (1983)) . The level of CAT activity following transf ection of these plasmids into murine fibroblast cells is shown in Figure 2 and Table 1. In NIH 3T3 cells, the level of CAT gene expression of the plasmid containing the Akv LTR sequences is th ree times that of the plasmid containing the corresponding region of the SL3-3 virus. A similar result was obtained using mouse L cells, another murine fibroblast cell line. F rom these results we conclude that in some murine fibroblasts, the Akv LTR seq uences are more potent than those of SL3-3 in promoting CAT gene exp ression.

A similar set of experiments was done using murine lymphoid cell lines. In marked contrast to the results observed using fibroblasts, the level of CAT activity directed by the plasmid containing the SL3-3 LTR sequences is much greater than that containing the corresponding Akv sequences in all the T cell lines tested (Table 1, Figu re 2 B) . The SL3-3/Akv ratio of transcriptional element activities ranges f rom 0.3 to 0.6 in fibroblasts and f rom 6 to 20 in T lymphocytes.

The preferred exp ression in T cell lines of the PSU3CAT plasmid relative to that of the pAU3CAT is evidently not a consequence of virus infection since the SL3-3/Akv ratio of transcriptional element activ ity in SL3-3-infected L691 cells (designated as L691.SL3-3) is similar to that found in uninfected L691 cells. Preference for SL3-3 LTR seq uences is also observed in primary cell preparations f reshly eluted f rom the thymus of NFS/N mice (Table 1) . Evidently, the transcriptional preference of SL3-3 LTR sequences in the established T cell lines is a general property of T cells. The preferential expression of pSU3CAT is not observed upon transfection of the murine B cell line, M12 (Table 1) , suggesting that preferential expression of pSU3 CAT is not a general property of all lymphoid cells.

The only differences in LTR sequence between SL3-3 and Akv are located within the tandemly duplicated region located 221 to 443 bp from the RNA start site (Lenz et al., Natu re, s upra) , a region shown to include enhancer element functions of other murine retroviruses (see, Spandidos et al., supra; Blair, D. G., et al.,PNAS USA, 77: 3504-3508 (1980) ; Blair, D. G., et al.,

Science, 212: 941-43 (1981) ; Levinson, B., et al.,

Natu re, 295: 568-572 (1982) ; Laimins, L. A. , et al., P NAS USA, 79: 6453-6457 (1982) ; Jolly, D. J. , et al., Nucleic Acids Res.. 11: 1855-1872 (1983) ; Berg, P. E. , et al., Mol. Cell. Biol. , 3: 1246-1254 (1983) ; Kriegler, M. et al., Mol. Cell. Biol., 3: 325-339

(1983) ; Wood, T. G. , et al., J. Virol., 46: 726-736

(1983) ; Laimins, L. A., et al., J. Vi rol., 49: 183-189

(1984) ; Srinivasan, A., et al., Science. 223: 286-289

(1984) ; Linney, E., et al., Natu re. 308: 470-472

(1984) , Cocoran, L. M. , et al., Cell, 37: 113-122

(1984)) .

This observation raises the possibility that sequences remotely located f rom the start site function as primary determinants of the efficiency of promoter element utilization in different cell types. To test this possibility, we measured the ability of the portion of the U3 region harboring the tandemly repeated element of SL3-3 and Akv LTR sequences to function as transcriptional enhancer elements in murine T cells.

The portion of the SV40 early region promoter that governs start site specificity (see, Benoist, C., et al., Nature. 290: 304-310 (1981) and Fromm, M., et al., J. Mol. Appl. Genet., 1: 457-481 (1982)) was positioned 5' to the CAT gene (designated as pSVIXCAT-AB3) , and the region harboring the tandem repeat element of either SL3-3 or Akv was linked in cis, in both orientations, at a site about 2 k ilobases from the SV40 early start sites (Figure 1b) . These plasmids were transfected into murine T cells (the AKSL3 cell line) , and the levels of CAT activity were determined relative to pRSVCAT and pSVIXCAT-AB3) (Figu re 2C) , Table 2) . The level of CAT activ ity directed by the plasmid that contains the SL3-3 repeat region was more than 30 times that of the pSVlXCAT-AB3 plasmid that lacks murine LTR seq uences. In contrast to the marked increase in CAT activity observed for the plasmids containing the SL3-3 sequences, the level of CAT activity for those plasmids containing the Akv tandem repeat region sequences is only increased two-fold relative to pSVIXCAT- AB3. Similar results were obtained using another murine T cell line, the SL3B cell line (data not shown) .

These results demonstrate that in T cells the region harboring a tandemly repeated element of the SL3-3 LTR behaves as a transcriptional enhancer element (see, Banerji, J. , et al., Cell, 27: 299-308 (1981) and Wasylyk, B., et al., Cell, 32: 503-514 (1983⁾) ; i.e. , it potentiates increased activ ity f rom a remotely located heterospecific promoter element in an orientation- independent manner. Thus, the enhanced transcriptional activity of the SL3-3 LTR sequences relative to Akv sequences in murine T cells can be attributed to differences in enhancer element function encoded by sequences remotely located in the U3 region.

The increased rate of CAT gene transcription directed by the SL3-3 LTR element in T cells, relative to that of Akv sequences, permits an explanation of the T cell tropism and leukemogenic potential of the SL3-3 LTR containing viruses. Replication of retroviruses proceeds via transcription from integrated proviruses. Tissue specific differences in the rate of transcription of the magnitude observed here, 10 to 60-fold, may have a profound effect on the extent of viral replication, as such differences might be expected to be magnified as a power function on successive rounds of replication. Consonant with this model, recent studies indicate that a correlation exists between the cell specificity of viral replication and the type of leukemia induced by an isogenic set of viruses that differ only in the LTR regions (see, DesGroseillers et al., sup ra and Chatis et al., supra) . Thus, it is likely that viral replication, via transcription, in the appropriate cellular context may play a major role in the induction of a tissue specific disease by these viruses.

Finally, recent studies suggest that murine leukemia viruses may induce disease by proviral integration adjacent to common loci (see, Cocoran et al., supra; Tsichlis, P. N. , et al., Natu re. 302:

445-449 (1983) ; Steffen, D. , PNAS USA, 81: 2097-2101 (1984) ; and Cuypers, H. T., et al., Cell, 37: 141-150 (1984)) , some of which harbor known oncogenes (Cocoran et al., su pra and Steffen, supra) . Analysis of these integration sites indicates that these loci may be transcriptionally activated by the adjacent viral LTR enhancer element (Cocoran et al., supra; Tsichlis fit al., s upra; Steffen, s upra and Cuypers et al., supra).

The present inventors speculate that an additional, if not simultaneous, characteristic of the LTR of a T cell leukemogenic virus. is to provide strong enhancer element function sufficient to transcriptionally activate adjacent cellular loci. In NIH 3T3 cells, the level of CAT gene expression on the plasmid that contained the Akv LTR sequence is three times that observed by the plasmid that contains the corresponding region of the SL3-3 virus. A similar result was obtained using mouse L cells, another murine fibroblast cell line.

A similar set of experiments was done using murine lymphoid cell lines. Several T cell lines established from murine T lymphomas were used as recipients and include ; SL3, a cell line derived from a premature thymic T cell tumor of an AKR mouse that had been injected with the SL3-3 virus (see, Hayes, E.F., et aI., J. Natl. Cancer Inst., 69: 1077 - 1082 (1982); and Hayes, E.F., et al., Cancer Res., 37: 726 - 730 (1977)); AKSL3, a cell line established from a spontaneous tumor of an AKR mouse, the cell line from which the SL3-3 virus was isolated (see, Nowinski, R.C., et al., J. Virol., 27: 13 - 18 (1978); Laimins, L.A., et al., PNAS U SA, 79: 6543 - 6547 (1982); Oskarsson, M., et al., Science. 207: 1224 - 1226 (1980); and Kriegler, M., et al., Molec. Cell. Biol.. 3: 325 - 339 (1983)); and an uninfected murine T cell line, L691 (see, McGrath, M.S., et al., in Contemporary Topics in Immunobiology. Vol.II, Warner, N. L., ed., pp 157 - 184 (1980); and Nowinski, R.C., et al., Virology. 81: 363 - 370 (1977)); as well as L691 infected in vitro with the SL3-3 virus (L691.SL3-3). The murine B cell line M12 was also used as a recipient.

In marked contrast to the results observed using fibroblasts, the level of CAT activity directed by the plasmid that contains the SL3-3 LTR sequences is much higher than that which contains the corresponding Akv sequences in all the T cell lines tested. The ratio of activities in SL3-3/Akv ranges f rom 0.3 to 0.6 in fibroblasts and f rom 6 to 20 in T lymphocytes. These results demonstrate that there is a dramatic difference in the function of the SL3-3 and Akv LTR sequences in T cells as compared to fibroblasts.

Another substantial advantage obtained fromt he present invention results from the fact that this tissue specific determinant does notdepend on the production of any virally encoded gene products, which further supports the utility of these elements in efficient and compact vector design. Levels of CAT activity following transf ection of NIH 3T3 fibroblasts with these plasmids is summarized in Table 1. In parallel experiments NIH 3T3 cells were also exposed to the plasmid that contains the enhancer promoter region of SV40, (pSV2CAT) and the LTR region of Rous sarcoma virus (pRSVCAT) , that were located 5' to the CAT gene. The level, of activity of these two plasmids is roughly comparable in all cell types used, and activ ity of the viruses that contain the murine LTR. sequences are expressed relative to the level CAT activity obtained in transfection experiments with pRSVCAT and pSV2 CAT plasmids.

Each element's activity is normalized relative to the absolute activity of the 3 ' LTR of RSV in pRSVCAT within each cell line. Absolute activity for each plasmid in every cell line was calculated f rom the slope of the line in the initial velocity of the CAT enzymic assay for cell extracts containing 400 fg of protein as depicted in Figure 2. The absolute activ ity in any given cell line transfected with each plasmid is expressed in terms of percent conversion/hour in the in vitro assays, and may be obtained quantitatively by multiplying the relative value in the Table by the absolute activity value of the same cell line transfected with pRSVCAT.

The absolute activity of pRSVCAT in each cell line is given parenthetically: NIH3T3 (2.06%/hr) , L (0.64%/hr) , M12 (2.5%/hr) , AKSL3 (4.8%/hr) , SL3 B (2.04%/hr) , L691 (0.84%/hr) , AND L691.SL3-3 (2.37%/hr) .

These results demonstrate that in NIH 3T3 fibroblasts the level of CAT activity of the plasmids that contain the Akv LTR sequences is approximately twice as great as that which contains U3 sequences of SL3-3. These experiments show that in the fibroblasts the Akv LTR sequences are more potent than those of SL3 in promoting CAT transcription, and that in T cells the SL3 LTR sequence is much more potent than Akv.

Recent evidence indicates that one step in leukemogenesis induced by non-acute murine and avian retroviruses is transcriptional activation of cellular oncogenes, via integration of a nearby provirus. Available evidence indicates that integration of provirus occurs at many possible sites and that the process may be random. Conseq uently, it is likely that many rounds of replication are required for activation of a cellular oncogene. As described herein, a good correlation between the site of viral replication and the type of leukemia induced by an isogenic set of virus that differ only in the LTR regions has been found. Therefore, it is proposed that both the T cell tropism and the leukemogenic potential of the non- acute leukemia viruses is a property of transcriptional enhancer elements that determine the ability of the virus to replicate efficiently in differentiated cell types.

Upon inj ection of susceptible strains of mice, murine leukemia viruses exhibit characteristic tissue specificity, latency period, and disease phenotypes. Recent findings demonstrate that the ability to induce T cell leukemia is associated with the non-coding long terminal repeats (LTRs) of these viruses (Lenz, supra; Chatis, P. A. , et al., PNAS U SA. 80: 4408 (1983) ; and Des Groseillers, supra) . To test if the LTR sequences might determine the preferred organ site of replication and disease phenotype, a series of isogenic recombinant viruses that differ only in the LTR regions were constructed. Viruses used included the Akv virus (Rowe, W. P., et al., Cold Spring Harbor Symp. Quant. Biol.. 44: 1265-1268 (1979); and Hays, E. F., et al., Cancer Res., 37: 726-730 (1977)); an a virulent strain that is endogenous to the AKR mouse, the thymic. leukemia virus, SL3-3, (Pedersen, F. S., et al., Nature (London, 292: 167-170 (1981)), and the erythro leukemia virus. Friend helper virus (Fr-MuLV) (Oliff, A. I., et al., J. Virol., 35: 924-936 (1980)). As described herein, the ability of these viruses to replicate efficiently in the thymus or spleen and to induce thymic or erythro leukemias is determined by the LTR sequences. Additionally, it has been discovered that recombinant MCF viruses are not detected in animals infected with these viruses.

Cloned infectious proviruses used in these experiments were the a virulent Akv virus, SL3-3, (Lenz, supra.) and a non-defective component of the Friend virus complex, Fr-MuLV 57. Substantial sequence differences were discovered in the U3 region of these viruses (Figure 3). To examine how these differences might effect viral replication these in specific tissues, the series of recombinant proviruses pictured in Figure 1 was constructed. The S13 LTR - Akv genome recombinant, RECAS115, has previously been shown to induce chronic leukemia in AKR, CBA, C2H, SJL and NFS strains of mice (Lenz, supra).

Recombinant proviruses in which the LTR sequences of Fr-MuLV57 replaced the corresponding Akv LTR or SL3-3 sequences were also constructed (Figure 1). Infectious virus was obtained from each of these recombinant proviruses by transfection of NIH 3T3 cells with proviral DNA (Lenz, supra; Lenz, supra (1982)). To determine the site of replication of viruses newborn NFS mice were inj ected with 1 X 10⁴ infectious units as determined by an XC plaque assay (Gross, L., P roc. Soc. Exp. Biol. Med. , 94: 767 - 771 (1957)) . The NFS strain of mouse was selected, as these mice do not express detectable levels of endogenous MuLV.

The ability of virus to replicate in thymus, spleen, and marrow cells was determined by flushing the cells from the organ, and measuring the number of infected cells using an infectious center, XC assay. Table 2 shown that the titers of Akv virus remained low in the thymus, spleen and marrow of animals infected with Akv for the duration of the experiments, 20-40 weeks. In contrast, the virus titers in the thymus of animals inoculated with RECAS115 were high in the range of 1-5 X 10⁵ by 6 weeks post- inoculation. However, the titers were low in the spleen and marrow, generally less than 5 X 10 ² throughout the duration of the experiment.

To determine whethe r the infected thymocytes represent a mature cortical population several animals were injected with dextamethasone one day prior to removal of the thymus. This treatment greatly reduces the number of infected thymocytes indicating that almost all the inf ected thymocytes are of cortical origin (see. Table 3) .

The phenotype of the cells was also examined. These studies demonstrated that the majority of the cells infected by viruses that contain the SL3-3 LTR sequences were adherent neither to nylon wool nor to plastic, were theta antigen positive, IgG negative, Ly1+, and Lt2-. About 20% of infected cells in the spleens of these animals were theta negative and, IgG positive, indicating that B cells, or B cell precursor might also be infected. Previous studies show viruses that induce T cell leukemia also infect B cells at a low but measurable frequency.

The organotropism of the Friend recombinants differs dramatically from those with Akv or the S13-3 LTR recombinants (Table 4) . These viruses replicate to high titers in the spleen but not the thymus. Viral titers are at least two orders of magnitude greater in the spleens, and about two orders of magnitude less in the thymus relative to Akv and RECAS115.

Expressed as XC+ infectious centers/10⁷ cells.

Some of the animals that were inoculated with the Friend-recombinant viruses, developed erythro leukemia. The spleenic lympohocytes obtained from these animals stained positively for non-specific esterase and negatively with Sudan black.

Same mice as listed in Table 4. Infectious centers (XC+) after treatments.

Recombinants between inoculated viruses and cellular sequences that encode MCF envelope genes are often detected in the preleukemic and leukemic stages of disease induced by both thymic and erythro leukemia viruses (Teich, N.M., et al., Cell, 12: 937 - 982 (1977) ; Coffin, J., Endogeneous Viruses in RNA Tumor Viruses. Cold Spring Harbor Laboratory (1982) ; Famulari, N.G., et al., J. Vi rol.. 40: 971 - 976 (1981)) . To determine whether MCF viruses were present in animals inoculated with the viruses described here, infected thymocytes or spleen cells were tested for the presence of MCF viruses. The ability of the infected cells to produce foci in mink cells (Hartley, J. W. et al., P roc. Natl. Acad. Sci .

USA, 74: 789-792 (1977) ) , to induce synctia in mink cells when first plated on SCI cells and then UV irradiated (the UV - mink test) was examined. Binding of an M CF specific recombinant antibody by the infected cells was also examined. No M CF-like viruses were detected using these tests. No M CF-like glycoproteins or viruses were present in the infected cells even in animals that had fully dev eloped thymic or erythro leukemias and in which a high proportion of cells produced infectious virus.

The results presented here demonstrate that the ability of viruses to replicate efficiently in either thymus or spleen of NFS mice is a property of the LTR and is not determined by the gag, pol or env genes of the virus.

These results are in agreement with previous studies that indicate that the LTR regions of MuLV viruses code the major tissue tropic and disease phenotype (Lenz, s upra, (1983) ; Chatis, supra, (1983) ; Des G roseillers, s upra. The disease phenotypes of the Gross A virus SL3-3 is determined by the LTR sequence (Des Groseillers, supra) . Recombinant virus that contain the LTR sequences of the thymic leukemia viruses of Moloney murine leukemia virus and the coding sequences of F riend virus induce a T cell leukemia characteristic of Mo-MuLV (Chatis, supra. whereas the reciprocal recombinant induces disease typical of the Friend virus.

The ability of the virus to replicate to high titers in specific tissues as shown herein to be a property of the LTR sequences, is a prerequisite for disease induction by the non-acute retroviruses. To date mice inj ected with SF24, the Fr-SL3 recombinant virus, have not developed thymic disease. This in itself confirms that the coding sequences of SL3-3 without the appropriate LTR are not sufficient to induce thymic disease. However, only a minor proportion of these mice have developed the erythroblastosis which is characteristic of Fr-MuLV inf ection.

The tissue specificity of replication of the viruses examined herein is not attributable to formation of M FC recombinants, as no such recombinants were formed upon infection of NFS mice for either of these viruses. For the same reason, it is not believ ed that induction of the thymic leukemia of erythroleukemia requires formation of M FCs.

The sequence, of the LTR of Akv, SL3-3, Fr-MuLV and Mo-MuLV are shown in Figure 3. As noted prev iously, with the exception of a single point mutation, the only difference in the sequence of the Akv and SL3-3 viruses are located in the tandem repeat sequence of the U3 region of the LTR (Lenz, supra) . The sequence of the U3 region of Fr-MuLV differs from that of Akv and SL3-3 in a similarly placed tandem repeat region. The sequence of the LTR of Fr-MuLV also differs from that of the Akv and SL3-3 virus in the region 3' to the tandem repeat elements. However, in this region of the LTR, the sequence of the Friend LTR is very similar to that of the thy motropic Mo- MuLV (Koch, 1983) . Thus, it is that arrangement of sequences within the tandem repeat which determines the tissue tropism of the virus. The tandem repeat elements of Mo-M uLV, Akv, and SL3-3 hav e been shown to function as transcriptional enhance r elements capable of di recting high levels of t ranscription of heterologous promoters (Lev inson, B., et al., Natu re, 295: 568-572 (1982) ; Laimins, su pra; Oskarsson, M. , et al., Science. 297 : 1224-1226 (1980) ; and Kreigler, M. , et al., Molec. Cell. B iol., 3 : 325-339 (1983) ) . Differences in the ability of tandem repeat elements to promote transcription and thereby promote viral replication in specific, differentiated tissues therefore determines both the tissue tropism and pathogenesis of these viruses.

The present invention is further illustrated by the following examples. These example are prov ided to aid in the understanding of the invention and are not to be constr ued as a limitation thereof.

EXAMPL E 1

Construction of Plasmids Used to Assess Promoter Element Function of M u rine LTR Sequences

The plasmid subclones harboring the 5 '-LTR of Akv and SL3-3 (pLTR-A4 and pLTR-S3, respectively) were provided by J. Lenz. The Pstl-Aval fragment of the LTRs comprise the U3 and a small f raction of the R regions, extending from -443 to +31, relativ e to the viral RNA start site (designated as +1) for both viruses. This fragment was isolated f rom each LTR εubclone where. the natural Aval site was converted to a Hindlll site using Hindlll synthetic linkers, and the natural PstI site was converted to a blunt end.

The plasmid that served as the recipient vector of the LTR fragments contains the entire CAT gene transcription unit (Gorman et al., Mol. Cell. Biol ., su pra) (CAT coding sequences followed by the simian virus 40 (SV40) t- intron and polyadenylation signals) under the control of a SV40 early region promoter mutant that carries a deletion of nearly the entire 72 bp tandem repeat element (designated as pSVIXCAT, see infra) . The SV40 early region promoter, bounded by a synthetic Xhol site (5') and a natural Hindlll site (3') , was removed f rom pSVIXCAT by successive digestion with Xhol, T4 DNA polymerase, and Hindlll enzymes in a fashion whereby the Xhol site was converted to a blund end. The LTR fragments were ligated to the vector fragment containing the CAT gene transcription unit to generate plasmids in which the CAT gene is under the control of the transcriptional elements in the U3 region of Akv (pAU3CAT) or SL3-3 (pSU3 CAT) . The Akv and SL3-3 U3 regions are identical except for the presence of a single base insertion at nucleotide position 65 (designated by an asterisk) in SL3-3 relative to Akv and for the sequences within the repetitive elements (designated by arrows) (see, Lenz, et al., Natu re, supra) . The relative positions of

CAAT, TATA, and initiation (designated as +1) sequences are indicated. The location and boundary of LTR sequences used in , pAU3 CAT and pSD3CAT are shown relative to a typical retroviral LTR. EXAMPL E 2

Construction of Plasmids Used to Assess Enhance r Element Activ ity of Mu rine LTR Sequences.

A plasmid that ca rries the CAT gene under the control of the SVB40 early region promoter (designated as pSV2 CAT (Gorman et al., Molec. Cell. Biol., su pra) was digested with AccI and SphI to remove the majority of the 72 bp repeat region. The natu ral Acl and SphI sites were converted to Xhol sites using T4 DN A polymerase and an Xhol synthetic leader. Following digestion with Xhol and recirculariz ation in the presence of DNA ligase, the resulting plasmid (pSVIXCAT) contains the CAT gene unde r the control of SV40 promoter sequences extending f rom -127 to +65 (relative to the 5 ' most early start site) (Benoist et al., su pra) . The structure of this promoter deletion mutant is shown relative to the entire SV40 early region promoter found in pSV2 CAT. The repetitive elements (designated by arrows) , the origin of replication (ori) , the TATA element (denoted at A/T) and early start site(s) (designated as +1) are illustrated. The SV40 promoter seq uences present in pSVIXCAT are deficient in enhancer element activity; however, this activity can be restored by linkage of sequences containing enhancer activity to pSVIXCAT at a variety of sites relative to the SV40 early start site(s) (unpublished observations) .

For this study, a variant of SVIXCAT (designated as pSVIXCAT-AB3) was constructed in which the natural Apal site (see, Fiers, W. , et al., Nature. 273: 113-120 (1978) ) located greater than 2.1 kilobases from the SV40 early start site(s) was converted to a Bglll site using T4 DNA polymerase and a synthetic Bglll linker. The LTR subclones, pLTR-A4 and pLTR-S3, were digested with Sau3A and EcoRI, and a 620 bp Sau3A f ragment containing 337 bp of U3 region sequences and 283 bp of Sau3A fragment D of pBR322 was isolated. The U3 region sequences present in the 620 bp Sau3A f ragment consists of sequences extending f rom -443 to -106 (relative to the viral LTR start site in Figure la) ., and contains the entire tandem repeat region of each virus. The 620 bp Sau 3A f ragment f rom each LTR subclone was inserted into pSVIXCAT-AB3 in either orientation at the unique Bglll site. All manipulations were done according to standard recombinant DNA techniques (M aniatis et al ., su pra) .

EXAMPLE 3

CAT Activity Directed By Viral LTR Sequences in different Murine Cells

Representative CAT assays for plasmids containing either the Akv U3 region abbreviated Au3; SL3-3 U3 region, abbreviated SU3; the 3'-LTR of Rous sarcoma virus, abbreviated RSV; or no eukaryotic promoter element, abbreviated SVO, linked 5' to the CAT gene transcription unit transfected in parallel into cultured cells is shown in Figs. 1(a) -1(c) for NIH3T3 fibroblasts (a) and AKSL3 T cells (b). Representative CAT assay of enhancer element function for plasmids containing either the SV40 promoter sequences lacking enhancer linked sequences, abbreviated SVIX-AB3; the Akv repeat region inserted at the Bglll site 2 kilobases away from the SV40 early start site(s) in the same and opposite transcriptional sense as CAT gene transcription, abbreviated AE(+) and AE(-), respectively; the SL3-3 repeat region inserted at the Bglll site and in the same and opposite transcriptional sense as CAT gene transcription, abbreviated SE(+) and SE(-), respectively; and the 3' LTR of Rous sarcoma virus, abbreviated RSV, linked to the CAT gene and transfected in parallel into cultured cells is illustrated for AKSL3 T cells (c). Results are expressed as percentage conversion of 14C-chloramphenicol to 14C-chloramphenicol acetate by extracts containing 400ug protein in (a) and (b) and extracts containing 600 ug protein in (c). The insets show typical auto-radiograms of CAT assays at 30 minutes (a), 80 minutes (b), and 30 minutes (c). The upper two spots correspond to the two isomers of 14C-chloramphenicol monoacetate while the spot is the un reacted substrate, 14 C-chloramphenicol.

Each murine cell line was transf ected with an equivalent amount of each plasmid DNA per transfection cocktail. The amount of DNA used for each experiment was within the liner range of optimal CAT activity. The NIH3T3 cell transf ections were carried out as described previously (Wigler, et al., su pra) . The L cell and lymphoid cell transf ections were done as described previously (Stafford et al., sup ra) . After a transient expression period (44 hours in promoter studies as in (a) and (b) ; 50 hou rs in enhancer studies as in (c)) , the cells were washed with cold, lxPB S and resuspended in 200 ul of 0.25 M Tris- HCl, pH 7.9. Cell extracts were prepared by 3 cycles f reeze (-70° C) - thaw (37° C) treatment and centrifugation.

In vitro CAT enzymic assays were performed according to published procedu res (Gorman et al., Mol. Cell. Biol., su pra) except that a final concentration of 2.4 mM acetyl coenzyme A was used in each extract assay. Analysis and quantitation of in v itro enzymatic assay was done according to previously described procedures (ibid.) .

In Table 8, the CAT activity of each plasmid is normalized relative to the absolute activity of the 3'- LTR of RSV in pRSVAT within each cell line. Absolute activity for each plasmid in every cell line was calculated from the slope of the line in the initial velocity of CAT enzymic assay for cell extracts containing 400 ug of protein. The absolute activity in any given cell line transfected with each plasmid is expressed in terms of % conversion/hour in the in vitro assays, and may be obtained quantitatively by multiplying the relative value in Table 8 by the absolute activity value of the same cell line transfected with pRSVCAT. The absolute activity of pRSVCAT in each cell line is given parenthetically: NIH3T3 (2.06%/hr) , L (0.64%/h4) , M12 (2.5%/hr) , AKSL3 (4.87%/hr) , SL3 B (2.04%/hr) , L691 (0.84%/hr) , L691.SL3-3 (2.37%/hr) and NFS/N thymus cell (0.03%/hr) .

These data reflect an average of at least two independent transfection experiments, and CAT assays performed using extracts containing 400 ug of protein. The range of activities around a particular value varied not more than 30% of that value.

**Several T cell lines established f rom murine T lymphomas were used as recipients and include: AKSL3, a cell line established f rom a spontaneous tumor of an AKR mouse (the cell line from which the SL3-3 virus was isolated) (see, Nowinski, R. C, et al., J . Virol . , 27: 13-18 (1978) ) ; SL3 B, a cell line derived from a thymic T cell tumor of an AXR mouse that had been injected with the SL3 virus (see. Hays, E. F., et al., J. Natl. Cancer Institute, 69: 1077-1082

(1982)) ; the uninfefcted murine T cell line, L691 (see, McGrath, M. S., et al., in: Contempora ry Topics in Immunobiology. Vol. II, ed. N. L. Warner, Plenum Press, New York, 157-184 (1980) ); L691.SL3-3, L691 cells infected in vitro with SL3-3 virus and freshly eluted primary thymus cells isolated f rom NFS/N mice (see, Now inski, R. C , et al., Vi rology. 81: 363-370 (1977)) . The murine B cell line, M12, was also used as a recipient.

As shown in Table 9, the CAT activity of each plasmid is normaliz ed relative to the absolute activity of the 3 '-LTR in RSVCAT within the T cell line AKSL3. Absolute activ ity was determined as described in Table 8. Results shown here are an average of 3 transf ection experiments and CAT assay s performed using extracts containing 600 ug of protein. The range of activ ities around a particular value varied not more than 30% of that value.

Fold-enhancement for plasmids containing viral repeat regions was determined relative to PSVIXCVAT-AB3, the plasmid lacking a viral repeat region.

EXAMPL E 4

Structu re of Recombinant Viral Genomes

a) Figu re 1 indicates the viral sequences which are present in the recombinant plasmids. b) The solid lines depict the regions of the SL3-3 LTR that is present in the respective recombinant genomes. Regions corresponding to the SL3-3 LTR are shown in black and cross-hatching respectively. The strategy taken to construct the recombinant genomes PR- 8+ RE CAS 115 has been previously described (Lenz, (1982) , su pra) . RE CAS 115 contains the coding sequences of Akv virus and the LTR plus 75 nucleotides of the untranslated leader sequence of SL3-3 virus. The recombinant genomes SF24 and AF226 contain the Fr-MuLV U3 region plus part of H and the coding sequence of SL3-3 and Akv virus respectively. An aproximately 450 bp Pstl-Kpnl fragment encompassing the U3 region and part of R was removed from the Akv (LTRΛ4) and SL3 LTR (LTRS3) εubclones. A similar Psti-Kpnl fragment encompassing the corresponding region of Fr-MuLV57 was isolated from a proviral clone (Oliff, et al., sup ra) and ligated to the pLTR-S3 and pLTRA4 εubclones. The resulting εubclones were cleaved with PVUI- EcoRI and ligated to restriction f ragments of the Akv and SL3 genome that lack the LTR sequences. Structure of the recombinant genomes was confirmed by digestion with restriction enzymes unique to the individual LTRs. Recombinant genomes containing a single LTR were linearized at their PstI sites and ligated to form concatamers containing two LTRs. Transf ection of recombinant proviral DNA into NIH 3T3 cells to yield infectious virus has been previously described (Lenz, et al., (1983) , su pra ; and Lenz, et al., (1982) , su pra) .

EXAMPL E 5

Figu re 1C illustrates a comparison of the sequence organization within the tandem repeat regions of the Akv (Van Beveren, et al., su pra) , SL3-3 (Lenz, et al., (1983) , supra and Fr-MuLV57 (Koch, (1983) , supra) LTRs. The sequence of a single 99 bp repeat element of Akv is shown at the top. The solid lines below represent identical sequences present in the 72 bp by repeats and 66 bp repeats of SL3-3 and Fr-MuLV57 respectively. The number after the bracketed region indicates how many times their sequence is repeated. The actual sequences in the askerisked areas are shown in (b) . These sequences are not present in the repeat region of Akv. The LTR vectors of the present invention can be employed in gene therapy or gene repair. For example, in one embodiment, an infectious viral vector carrying the tissue specific enhancer is inserted within the 3 ' region of the genome of the present invention between the env gene and the LTR. An animal can then be inf ected with a virus that contains this modified LTR vector suitable for targeting to a particular tissue. Inf ection unde r these conditions would lead to a pref erential infection on a massive level, in the specific tissue controlled by the selection of the enhance r. For example, the gene can be pref erentially directed into matu re thy mocytes, or eryth roblasts, using SL3 or Friend virus respectively. Thus th is concept can be employed to (1) determine, using a CAT assay, other tissue specific enhancers; and (2) for the bu ilding of appropriate tissue specif ic LTR vectors.

An other utility of the LTR vectors of the present invention is in cancer therapy . Each cancer cell, or each general type of cancer, has different transcriptional properties which are recognizable by different enhance r elements. Using the general approach outlined herein, enhancer seq uences for particular types of cancer cells (as determined using the CAT assay) can be made. Once the appropriate enhance r has been identif ied, two approaches can be used. First, an infectious route wherein the LTR with an engineered enhancer is built onto a virus which carries a gene, or is built onto a non- replicating infector virus. In either case, the viruses could be used to infect patients suffering the particular cancer, and because of the tissue specific enhancer, the viruses would preferentially infect the tumor cells. Alternatively, the cancerous cells can be removed f rom the patient, e.g./ bone marrow, thereafter infected with the virus, and replaced into the patient. Depending upon the nature of the gene in the LTR vector, either a lethal product can be expressed, which would kill the tumor cells, or a protein could be expressed, which in coating the cell surface, could provide a recognition site for a monoclonal or polyclonal antibody. For example, hemagluten flu or ricin A chain genes could be inserted into the vectors of the present invention, either in a replicating or non- replicating cell.

A further expression of utility for the tissue specific LTRs of the present invention, or portions thereof (especially the tissue specific determinants) represents the first pharmaceutical use of these species. For example, the viral LTRs may be combined with an appropriate pharmaceutical carrier for delivery of the tissue specific entity to a mammal. Moreover, the tissue specific vectors may be employed in the delivery of antitumor or like pharmaceutical agents, such as those described in the Physician' s Desk Reference. 38th ed.. Medical Economics Co. , Oradell, NJ (1984) , the disclosure of which is incorporated herein by reference.

As experiments have established, it is not necessεary to have the vector used code for any viral proteins in order to obtain the tissue specific benefits of the present invention. Nor is it necessary or advantageous to utilize infected tissue in order to express the proteins of interest using the tissue specific enhancers. Heterologous enhancers can be used to advantage with non-viral promoters to provied tissue specific production in uninf ected mammalian tissue. Tissue specific viral enhancers may be provided and the tissues for which they are specific identified, by providing virus with the enhancer sequence to be tested, preferably also providing the viral genome with a detectable marker, infecting a mammal with such virus, and analyzing the various mammalian tissues for presence of the virus and the marker product (s) .

It is evident that those skilled in the art, given the benefit of the foregoing disclosure, may make numerous other uses and modifications of, and departures from the specific embodiments described herein without departing f rom the inventive concepts. Consequently, the invention is to be construed as embracing each and every novel combination of features present in or possessed by the products and techniques herein described, and limited solely by the scope and spirit of the appended claims.

Claims

WHAT IS CLAIME D IS:

1. A tissue specific vector, comprising a tissue specificity determinant, a promoter, and a heterologous DNA fragment which codes for the expression of a selected product in a selected tissue, the tissue specificity determinant being specific for the selected tissue.

2. The tissue specific vector of claim 1, wherein the vector contains at least one viral enhancer, and the tissue specificity determinant is part of a viral enhancer.

3. The tissue specific vector of claim 1, wherein the tissue specificity determinant is heterologous to the promoter.

4. The tissue specific vector of claim 1, wherein the tissue specificity determinant is one derived from Akv, SL3-3, or Friend virus.

5. The tissue specific vector of claim 1, wherein the ratio of product produced at standard expression conditions using the vector in the selected tissue to product produced in other tissue is at least 1.5 to 1.

6. The tissue specific vector of claim 5, wherein the ratio is at least 5 to 1.

7. The tissue specific vector of claim 6, wherein the ratio is at least 10 to 1.

8. A tissue specificity determinant comprising a DNA sequence which renders expression vectors specific to a selected tissue.

9. A method of expressing a product in a selected tissue, comprising joining a DNA sequence which codes for expression of the product to an expression vector containing a tissue specific determinant, transfecting the vector into the selected tissue, and expressing the product.

10. A method of obtaining tissue specific enhancers which are specific to a selected tissue, comprising ligating test enhancers into a vector which contains a DNA sequence which codes for expression of an indicator material, expressing the vector in the selected tissue, and screening for those vectors which produce expression of the indicator material in the selected tissue.

11. The method of claim 10, wherein the indicator material is chloramphenicol acetyltransferase.

12. A pharmaceutical composition comprising a tissue specificity determinant and a pharmaceutically acceptable carrier.