Myod Regulatory Region

The invention relates to genetic engineering involving recombinant DNA technology, and particularly to expression vectors for tissue-specific transcription of genes in muscle cells.

The myoD gene converts many differentiated cell types into muscle cells. MyoD protein is a member of the basic-helix-loop-helix (bHLH) family of proteins; this 68-amino acid domain in MyoD is necessary and sufficient for myogenesis. The MyoD protein is sufficient to orchestrate the coordinated expression of most, if not all, of the skeletal myogenic program in cell types derived from all three germ layers of the embryo. As such, the regulation of Myo D expression and protein activity creates a nodal point, or master switch, that integrates the genetic and environmental influences on a cell. The expression and activity of the MyoD protein is regulated, in part, by interactions with members of a large family of proteins related to MyoD by sequence homology, the helix-loop-helix (HLH) proteins. MyoD binds cooperatively to muscle-specific enhancers and activates transcription. The helix-loop-helix (HLH) motif is responsible for dimerization, and, depending upon its dimerization partner, MyoD activity can be controlled.

MyoD is a nuclear protein, 318 amino acids in length, that binds to many muscle-specific enhancers. Although it is phosphorylated, the function of the phosphorylation has not been defined. The protein and RNA turn over rapidly. MyoD binds to a consensus DNA binding sequence that includes a CAN NTG sequence (N represents A, T, G, or C) present in most muscle-specific enhancers (1; see the appended Citations).

To date, four different, but related, genes have been reportedly cloned from muscle cells: myoD, encoding the MyoD protein; myogenin; myf-5; and the gene encoding MRF4/Herculin/Myf6 (2-8). Each is sufficient to activate the myogenic program when expressed in a non-muscle cell line, and together they encode the MyoD family of myogenic regulatory proteins.

MyoD alone is sufficient to activate the myogenic program without the a priori cooperation of other muscle-specific factors. Myogenic regulatory genes are capable of interacting with and suppressing tissue-specific regulatory factors in other cell types, such as fibroblasts, even when those cells would not ordinarily express myogenic genes. The MyoD protein seems to activate myogenesis by directly binding to the control regions of muscle-specific genes (1).

In in vitro gel shift assays MyoD protein binds DNA and paired MyoD protein binding sites are reportedly essential parts of several characterized enhancers of muscle-specific genes (1, 9-11). MyoD reportedly binds cooperatively to the paired sites in the muscle creatine kinase (MCK) enhancer, and deletion analysis suggests that the NH.sub.2 -terminal 50 amino acids are necessary for this cooperativity (9). In enhancers that contain two MyoD protein binding sites apparently both are necessary for full enhancer activity, arguing for the functional importance of cooperative binding. In contrast, some enhancers, such as that for alpha-cardiac actin, contain a single identified MyoD binding site (12). In this case the adjacent CArG and SP1 binding sites can repotedly functionally substitute for the second MyoD binding site (12), because the enhancer activity is dependent on the presence of all three sites.

In vitro DNA binding assays suggest that heterodimerization of the MyoD family of myogenic bHLH proteins with an E2A protein may be crucial for functional activity. MyoD and related myogenic bHLH proteins have a relatively weak binding affinity to the CANNTG sites in the MCK enhancer, whereas the heterodimer formed with an E2A protein binds with much higher affinity (13-15). That is, MyoD can bind DNA as a homodimer, but it binds to its target sequence more than ten times better (16) as a heterodimer with one of the bHLH proteins encoded by the E2A gene, first identified by Murre et al. (15). The E2A gene product is ubiquitously expressed and encodes Coy differential splicing), for at least three different bHLH proteins, E12, EA7, and ITF1 (25,26). Analysis of the binding site preferences for MyoD and E2A proteins show that each has a slightly different preferred half-site, giving homodimers and heterodimers slightly different preferred binding sequences (18). Mutations of the MyoD basic region, when compared to wild-type MyoD, show that some mutants bind with similar or better affinity to the MyoD binding sites in the MCK enhancer but do not activate transcription (15), arguing that the NH.sub.2 -activation domain of MyoD is usually "hidden" from the transcriptional machinery and that activation of the region can be regulated independent of DNA binding. Methylation interference indicates that the basic region of the bHLH proteins interacts with the major groove of the DNA double helix (1). In order to activate muscle-specific genes, MyoD protein complexes must bind to two or more upstream sites (9,13,16). As assayed by dissociation rate, MyoD binds cooperatively, mediated by its NH.sub.2 -terminal acidic domain, which also harbors the activation function.

Domain-swapping experiments with achaete-scute showed that the helix 1, loop, or helix 2 regions could be exchanged into MyoD without loss in the ability of the chimeric MyoD to form dimers with E2A, to bind to muscle-specific enhancers, or to activate myogenesis in C3H 10T1/2 cells. However, swapping of the basic region of achaete-scute into MyoD resulted in chimeric protein that could bind specifically to muscle-specific enhancer DNA and dimerize with E2A, but it failed to induce muscle-specific differentiation or trans-activation of muscle-specific reporter genes. The same result occurs when the E2A basic region is swapped into MyoD, and in this case the E2A basic region differs from MyoD in only four of ten positions which are conserved among the MyoD family. Changing just one of these E2A residues back to the corresponding MyoD amino acid (an alanine) restores DNA binding, but myogenic activity still fails to occur. Changing the second amino acid, back to a threonine, restores myogenic activation (19). These results show that the specificity for myogenic activation resides in the basic region of MyoD and have led to the suggestion that activation requires properties of both the DNA sequence and the basic region of the MyoD protein.

Negative and positive (20) mechanisms for regulating MyoD have been postulated. One possible mechanism of negatively regulating the activity of the myogenic bHLH proteins such as MyoD is by decreasing the available pool of E2A proteins through dimerization with the HLH protein Id (21). This protein shares the HLH domain but lacks an adjacent basic region, thereby rendering it incapable of binding DNA, but capable of binding other HLH proteins. It has been demonstrated in vitro that Id dimerizes with the E2A protein products with a relatively high affinity. That is, Id binds about five-fold better to E2A than to MyoD, and when added to MyoD-E2A-DNA complexes at 37.degree. C. Id rapidly causes dissociation of the complexes. Thus, by competing for HLH dimerization regions Id can prevent the formation of MyoD:E2A protein oligomers. Because Id lacks a region equivalent to the DNA binding domain of the bHLH proteins, Id-containing hetero-oligomers do not bind to core CANNTG sequence. For protein complexes containing Id, the choice is probably on-or-off; for complexes of E2A or MyoD homodimers or heterodimers, the results can be altered DNA binding specificity. In many cell lineages, including muscle, when the cells are induced to differentiate the level of Id mRNA declines (21). MyoD expression also decreases when cells are transformed with transforming agents (22, and citations therein).

MyoD protein is expressed only in skeletal muscle. Cardiac and smooth muscle, which express many of the same muscle-specific structural genes as skeletal muscle, reportedly do not express MyoD protein (2,23). Factors that influence growth regulate expression of MyoD protein, but MyoD protein itself can also inhibit cell growth (24). MyoD protein inhibits the cell cycle directly (24). MyoD protein is capable of positively regulating its own expression, and "autoactivation" mechanisms have also been postulated. The nucleotide sequence of the regulatory regions of the MyoD gene has not been identified, until now.

Transfection of LTR-myoD constructs into non-muscle cells can inhibit cell growth (24). MyoD inhibits the cell cycle directly and when introduced into cells prevents entry of the cells into S phase (24). The myoD gene is capable of activating previously silent muscle-specific genes when introduced into a variety of differentiated cell types with a viral long terminal repeat (LTR) used to promote constitutive transcription (2,25,26,29). Because MyoD molecules with mutations in the basic (DNA binding) region still cause growth arrest, it would seem that MyoD protein can halt the cell cycle independent of its ability to induce myogenesis. In certain cell types, the entire program for muscle differentiation seems to be activated (26). The range of cell types converted to muscle by the myoD gene includes a number of fibroblast cell lines, adipocytes, melanoma cells, a hepatoma cell line, neuroblastoma cells, osteosarcoma cells, and P19 teratocarcinoma cells, as well as primary cultures of chondrocytes, smooth muscle, retinal pigment, fibroblasts, and brain cells. In primary fibroblasts, MyoD expression is repressed by transacting factors (27), and during development myoD transcription may require, among other things, the removal of negative control.

Cotransfection of a MyoD expression vector with a muscle-specific expression vector carrying a single MyoD binding site results in no activity. Cotransfection with a vector carrying two sites lead to high levels of activity; however, if a third plasmid carrying a single site is also included and is in excess, inhibition occurs. This suggests that plasmids carrying a single site will compete specifically for the activation of a vector with two sites. This suggests (as noted above) that binding to a single DNA site is not sufficient to accomplish transcriptional activation of a muscle-specific gene and that, like other activator proteins (28), MyoD must occupy two or more sites before transcription is initiated.

The apparent simplicity of the observed MyoD myogenic effect in transfected cells belies a rather sophisticated interplay between the cooperative interactions of at least MyoD, E2A, and Id proteins, the multiple DNA regulatory sites that bind these factors, and cell type-specific and differentiation-specific transcriptional activation factors in cells that bind at other regulatory sites. The varied and complex effects of oncogenes, anti-oncogenes, and growth factors on MyoD expression and activity may also contribute to the failure of some cells to activate myogenic genes after introduction of MyoD (25).

Despite numerous attempts to identify the regulatory regions controlling expression of MyoD, these sequences have remained elusive. Without delineation of the myoD regulatory regions and definition of their specific interactions with MyoD, it has not been possible to engineer a vector that will reproduce the endogenous natural control of MyoD expression in a cell, or to select for cells expressing MyoD, or to utilize the myoD regulatory region to control other genes of interest in a tissue-specific manner in skeletal muscle cells or precursors to those cells. It would be highly desirable to have a source of isolated DNA capable of faithfully reproducing the natural mechanisms of regulation of MyoD expression in a cell, because MyoD expression would be useful for engineering transplantable skeletal muscle cells, or for targeting gene expression in a tissue-specific manner to skeletal muscle.

The invention provides nucleotide sequences of the myoD regulatory region containing both proximal promoter and distal enhancer regulatory regions and their constituent regulatory elements. The myoD regulatory region alone is sufficient to faithfully reproduce the natural mechanisms of regulation of MyoD expression in a cell. It is now possible to select for MyoD-expressing cells, e.g., by using a selectable marker regulated by the myoD regulatory region. By selecting MyoD expressing cells it is now possible to genetically engineer pre-muscle cells that can be selected and substantially purified for transplantation. Aspects of the invention also provide the regulatory region of the myoD gene for controlling other genes of interest in a tissue-specific manner in both myoblasts and early myotubes.

FIGS. 1A, 1B, and 1C together show the nucleotide sequence of the proximal regulatory region (PRR) of the myoD gene (SEQ ID NO:1);

FIG. 2 schematically depicts the organization of the proximal regulatory region (PRR) of the myoD gene;

FIG. 3 schematically represents the deletion mutants constructed to identify the proximal regulatory region (PRR) of the myoD gene;

FIG. 4A and 4B together show the nucleotide sequence of the distal regulatory region (DRR) of the myroD gene;

FIG. 5 schematically represents the deletion mutants constructed to identify the distal regulatory region (DRR) of the myoD gene;

FIG. 6 presents data that shows the activity of myoD regulatory region in directing expression of a gene interest in a cell;

FIGS. 7A, 7B, 7C, 7D, 7E, 7F, and 7G together show SEQ ID NO:3, comprising the LTR, .PSI., HisD, neo, and pBR322 regions of the myoD retroviral expression vector pLHDMDN.531;

FIGS. 8A, 8B, 8C, 8D, 8E, 8F, and 8G together show SEQ ID NO:4, comprising the LTR, .PSI., HisD regulatory, neo, and pBR322 regions of the myoD retroviral expression vector pLHDMDN.NSA; and

FIG. 9 shows schematiclaly the organization of the myoD retroviral expression vectors LHDMDN.531 and LHDMDN.NSA.

Initial attempts to identify the activity of the myoD regulatory region by transient transfection of conspecific mouse cells were unsuccessful. However, surprisingly, chicken myocytes were responsive to mouse myoD proximally regulatory region constructs containing promoter elements in transient transfection assays, and initial characterization of proximal regulatory regions was accomplished using these cells. In experiments to determine why transfection of mouse myoD regulatory regions in mouse cells were nonfunctional, the location of the upstream distal enhancer regulatory region of the myoD gene was identified. It was discovered that the failure to earlier demonstrate these distal enhancer regions was due to their inability to function in cells unless they are integrated in a stable manner into the genome of the recipient cell.

The identification of the proximal and distal regulatory regions provides sequences that normally control the tissue-specific expression of MyoD in muscle cells, and thereby myogenesis. The specific DNA regulatory sites of myoD were identified in the 5' flanking sequences of the myoD gene in a region from -7000 to +140 relative to the major transcriptional start of the cDNA.

A plasmid vector (pKCAT2) containing the 7 kb myoD genomic 5' regulatory region has been deposited with the American Type Culture Collection (ATCC), Rockville, Md., and granted accession No. 75091.

Retrovital vector "LHDMDN.531" containing 2,244 bp of myoD genomic 5' regulatory region (i.e., including both the PRR and DRR regions, below) has been deposited with the ATCC, and has been granted Accession No. 75092.

The following terms as used herein are intended to mean the following:

"MyoD regulatory region" as used herein describes a region of 5' myoD genomic DNA nucleotide sequence from -7000 to +140 relative to the major transcriptional start of the cDNA and inclusive of the PRR and DRR.

"PRR" is used to mean the myoD proximal regulatory region shown from positions 1 to 1620 in FIG. 1A, and inclusive of the regulatory elements from positions 590 to 900 in FIG. 1A.

"DRR" the used to mean the myoD distal regulatory regions shown from positions 1 to 1200 in FIG. 4A, and inclusive of the regulatory elements from positions 421 to 1200 in FIG. 4A .

"Capable of hybridizing under stringent conditions" means annealing to one of the myoD regulatory regions (shown in FIGS. 1A-C or 4A or 4B), or its complementary strand, under standard conditions, e.g., high temperature and/or low salt content which tend to disfavor hybridization. A suitable protocol (involving 0.1.times.SSC, 68.degree. C. for 2 hours) is described in Maniatis, T., et al., Molecular Cloning: A Laboratory Manual, Cold Spring Harbor Laboratory, 1982, at pages 387-389.

"bHLH" means polypeptides having the helix-loop-helix motif and a basic DNA binding region as described in the Background section, above.

"bHLH dimer binding site" means a myoD regulatory region nucleic acid capable of binding a bHLH oligomer consisting of two or more bHLH polypeptides in a complex, and wherein the complex is capable of binding to the bHLH dimer binding site in the nucleic acid. Representative examples of bHLH genes include, but are not limited to, myoD, myogenin, myf-5, and the gene encoding MRF4/Herculin/Myf6. Representative bHLH dimer binding sites include, but are not limited to, MyoD binding sequences (e.g., CANNTG), and also, including the bHLH dimer binding sites in PRR at positions 595 to 606, positions 666 to 675, and positions 875 to 883 shown in FIG. 1A; and, in the DRR, at positions 265 to 270, positions 486 to 491, positions 608 to 613, and positions 741 to 746 shown in FIG. 4A.

"SP1 site" means a myoD regulatory region nucleic acid capable of binding a protein that binds to an SP1 site having a nucleotide sequence with GGGCGG. Representative SP1 binding sites include, but are not limited to, sites in PRR at positions 740 to 745 shown in FIG. 1A; and, in DRR, at positions 291 to 297 shown in FIG. 4A.

"M-CAT site" means myoD regulatory region nucleic acid capable of binding a protein that binds to an M-CAT site having a nucleotide sequence with CATTCCT. Representative M-CAT binding sites include, but are not limited to, sites in PRR at positions 751 to 757 shown in FIG. 1A; and, in DRR, at positions 163 to 169 shown in FIG. 4A.

"CAAT box" means a myoD regulatory region sequence motif, such as CAAT box sequences (e.g., AACCAAT and GGCCAAT) that are often incorporated into a "CArG box" in other muscle promoters. The "CArG box" means a CC(A/T).sub.6 GG sequence that binds SRF and other factors. In myoD the CAAT box is part of a CC(A/T).sub.4 GG motif, that is similar to the CArG box motif. A representative example of a CAAT box sequence in the myoD regulatory region is found in the PRR at positions 758 to 763 shown in FIG. 1A; and a representative example of a CArG box is found in the DRR at positions 467 to 476 shown in FIG. 4A.

"ATAAATA region" means a myoD regulatory region sequence capable of hybridizing under stringent conditions to the sequence at positions 821 to 830 shown in FIG. 1A. "TATA element" and "TATA box" are also used interchangeably to refer to this element of the PRR.

"MEF binding site" means a myoD regulatory region nucleic acid capable of binding a protein that binds to an MEF site. Representative MEF binding sites include, but are not limited to, sites in the DRR at positions 312 to 321 and at positions 1097 to 1105 shown in FIG. 4A.

"AP1 binding site" means a myoD regulatory region nucleic acid capable of binding a protein that binds to an AP1 site. A representative AP1 binding site is found in the DRR at positions 254 to 260 shown in FIG. 4A.

"Operably-linked" is used to refer to linkage of a gene of interest to a myoD regulatory region nucleic acid such that the regulatory element(s) in the myoD regulatory region are capable of regulating downstream transcription of the gene of interest in a manner similar to the regulation of MyoD protein expression in skeletal muscle cells.

"Subtractive deletion" is used to mean the process of removing portions of nucleic acid sequence from a DNA or RNA, and subsequently testing the functional activity of the deleted sequence, so that regions exerting regulatory control or regions contributing to functional activity of a protein are identified.

The invention provides DNA or RNA expression vectors containing a myoD regulatory region for introducing a gene of interest into a ceil, such that the gene of interest is regulated in a tissue-specific manner in muscle cells. Useful vectors for these purposes include infectious DNA and RNA vectors. Representative vectors include retroviral, vaccinia, adenovirus, CMV vectors, and other such genetically engineered viral vectors, as well as plasmid vectors useful in transfecting cells. The expression vectors are constructed with a myoD regulatory region operably linked to a gene of interest, and transcription of the gene of interest is controlled by the MyoD regulatory region when the vector is introduced into the cell. When the vector DNA is integrated in the host-cell DNA, transcription of the vector DNA is regulated in a muscle-specific manner by the myoD regulatory region sequence(s). That expression of the gene of interest is so regulated, i.e., in a muscle-specific manner, can be determined by introducing the vector into muscle and non-muscle cells and observing the transcription or translation of the gene of interest in the muscle cell, such as by monitoring mRNA, protein, or protein functional activity, e.g., enzymatic activity or drug resistance. Useful muscle cells in which muscle-specific expression will occur include cells such as pre-muscle cells, myocytes, myoblasts, and also cells that have been genetically-engineered to incorporate at least one myoD gene into their genome so that they express MyoD protein. Cells transducer or transfected with expression vectors containing the myoD-regulatory-region are therapeutically and commercially useful, e.g., for (i) construction of tissue-specific gene transfer vectors; (ii)production of genetically-engineered proteins, i.e., by tissue-specific expression a gene of interest; (iii) muscle transplantation; (iv) expression of genes that regulate growth of myoblasts; and (v) expression of mutant forms of MyoD that modulate muscle differentiation. The expression vectors containing myoD regulatory region have the advantage that transcription of the gene of interest will occur in pre-muscle cells, myoblasts, and the like, but when the cells mature into myotubes (i.e., skeletal muscle) the expression of MyoD and the gene of interest will be down-regulated.

The invention provides myoD regulatory region DNA and RNA for expression of genes that encode marker proteins in replicating myoblasts and pre-muscle cells. Representative examples of marker proteins include marker enzymes (e.g., .beta.-galactosidase, alkaline phosphatase, and the like), drug resistance markers (e.g., neomycin phosphotransferase, hygromycin, methotrexate, and the like), and antigenie and adhesion markers (e.g., such as can be sorted, panned, or identified by immunochemical or in vitro methods). These markers will be recognized as those elements conferring a phenotype that can be positively selected for, or a genotype that confers a selective growth advantage to the cells expressing the gene of interest, in this case either in vitro or in vivo. Positive selection of such myoblast and pre-muscle cells thus provides a way of obtaining substantially pure cell cultures of myoblasts, myocytes, pre-muscle cells, and the like. To be active the gene of interest is preferably operably linked to the myoD regulatory DNA or RNA so that the regulatory region elements exert effects on the gene of interest in a manner similar to those exerted on the myoD gene when it is residing in the genome of a muscle cell.

The invention also provides myoD-regulatory expression vectors for inducing a muscle phenotype in a non-muscle cell. In this case, the MyoD protein coding sequence and the myoD regulatory region are both introduced into the non-muscle cell, such that transcription of the myoD coding region is controlled by the myoD regulatory region. Expression of MyoD in turn activates muscle-specific genes in the cell and directs the cell to become a muscle cell. Fibroblasts are representative of a cell type that may be so induced to become a muscle cell.

Rhabdomyosarcoma (RMS) tumors are tumors of skeletal muscle and represent one of the most common such solid tumors of childhood. All RMS are thought to express MyoD mRNA and protein, and, as such, MyoD expression is beginning to be used to classify solid tumors. While there have been real advances in the chemotherapeutic treatment of RMS tumors, a significant percentage remain refractory to chemotherapy. It is now possible, however, using myoD regulatory DNA and RNA in chimeric expression vectors to genetically alter RMS cells to broaden their sensitivity to chemotherapeutic agents. For example, such tumors can be genetically engineered to convert a prodrug to an active drug. A representative example is provided by genetically engineered tumor cells expressing a Herpes Simplex Virus thymidine kinase gene (HSV-tk). Such cells become sensitive to the prodrugs Gancyclovir and Acyclovir, i.e., guanosine analogs, because these prodrugs are substrates specific for the HSV-TK enzyme, and enzymatic conversion of the prodrug to an active drug takes place in these cells. Thus, the cytotoxicity of the drug, e.g., converted Gancyclovir and Acyclovir, is limited to cells expressing HSV-TK. While this general approach has been employed previously, there has been a general problem in targeting the expression of the HSV-tk gene in only tumor cells, and without such targeting significant toxicity could result. However, since expression of the myoD gene is restricted to skeletal muscle cells, and retrovital integration is restricted to replicating cells, the subject retroviral vectors, in which a gene of interest, in this case a therapeutic gene, is placed under the control of muscle-specific myoD regulatory DNA and RNA, will only be expressed in proliferating muscle cells and sarcoma cells. Representative examples of such therapeutic genes include genes encoding enzymes that convert pro-drugs to drugs, genes encoding toxins (e.g., ricin A chain and the like) and metabolic poisons, and genes encoding antiproliferative drugs. Thus, the provision of myoD regulatory DNA and RNA solves the problem of tissue-specific gene expression required for gene therapy in RMS cancer by targeting expression of the therapeutic gene to only tumor cells. Retroviral vectors are just one representative example of genetically engineered expression vectors that are useful in these therapeutic protocols.

The myoD PRR and DRR regulatory regions are useful as promoters and enhancers to regulate the expression of the HSV-tk gene, and other genes of interest. The results presented below demonstrate that myoD regulatory DNA and RNA can regulate expression of non-muscle genes, i.e., CAT, or neomycin drug resistance, in muscle cells transformed or transduced with expression vectors. The subject myoD regulatory DNA and RNA is useful in retroviral (and other) expression vectors for restricting expression of therapeutic genes to replicating skeletal muscle precursor cells and tumor cells in therapeutic protocols. Since the retroviral vector construct containing the myoD regulatory DNA and RNA will only express in replicating muscle cells and tumor cells, the other non-tumor cells of the host will be immune from the effects of the HSV-tk gene. In non-damaged muscle the fused myotubes and satellite cells are not replicating, and thus the predominant effects of the retroviral gene transfer will be limited to the tumor cells. This approach has particular merits for targeting tumor cells in RMS, e.g., for therapeutic intervention with Acyclovir and Gancyclovir as an adjunct to conventional chemotherapy. However, it is understood that genes other than HSV-tk and drugs other than Acyclovir or Gancyclovir will also be useful in similar protocols.

myoD regulatory DNA and RNA is also useful for enriching primary cell cultures to obtain substantially purified cultures of myoblasts. In this case, expression vectors are constructed wherein a marker gene (e.g., drug resistance or an enzymatic marker such as betagalactosidase) is placed under the regulatory control of the myoD regulatory DNA and RNA. When these vectors are introduced into myocytes (e.g., by transfection or transduction), and the cells are selected for the marker gene (e.g., drug resistance), only muscle cells, i.e., cells with endogenous or engineered MyoD protein expression, will be selected. This, method provides differential selection and elimination of cells unable to express the selectable gene, i.e., non-myoblast cells. This method of positive selection for muscle cells is especially useful in isolating and substantially purifying sources of genetically engineering cells, e.g., myoblasts, prior to transplantation.

Transplantation of allogeneic skeletal myoblasts shows promise already, without these selection methods, as an effective therapy for several neuromuscular degenerative disorders. However, success in the present therapeutic protocols often depends upon myoblasts expressing foreign histocompatibility antigens within a narrow range of limits, i.e., tissues with a good "match" of antigens between the donor and recipient. While it is fortunate that myoblasts express a relatively limited range of histocompatibility antigens, and are thus relatively weak immunogens, the cultures of cells which are transplanted are often heterogeneous and contain non-muscle cells (e.g., fibroblasts and tissue macrophages) that are highly immunogenic and capable of inducing an unfavorable immune response in the recipient, leading to elimination of the transplanted muscle cells. Currently, several techniques are used in an attempt to purify muscle cells for transplantation. These include growth of the heterogeneous cell cultures under selective growth conditions, panning techniques with specific substrates that favor attachment of muscle cells, and antibody-mediated panning or FAGS sorting. Antibody-mediated sorting methods have proven capable of providing purification relatively pure cultures of muscle cells; however, these techniques suffer from the additional disadvantage that the transplantable cells contain a foreign antibody on their cell surfaces. The foreign antibody may act both as an immunogen, i.e., inducing an immune response, and as an immune opsonic stimulant, i.e., stimulating an immune response to antigens of the skeletal muscle cell.

Selecting muscle cells based on the expression of marker genes under the control of the myoD regulatory region DNA and RNA offers significant advantages in efficiency and purity for muscle cell cultures useful in transplantation. Using positive selection methods is possible to substantially enrich populations of muscle cells that have a purity preferred for transplantation, i.e., contaminating immunogenic cells are removed and the resultant muscle populations have lower immunogenicity and thus improved survival in the recipient.

In such transplantation protocols the most desirable cell for use is the very early stage replicating myoblast. It is important, therefore, to use muscle promoter and enhancer regulatory sequences that are active as transcriptional control elements operative at an early stage in muscle differentiation. Muscle enhancer and promoter regions that have previously been described, i.e., from genes other than myoD, are not suitable for use because they direct expression of genes only in differentiated muscle cells. Thus, the subject myoD regulatory sequences offer the best regulatory elements for directing the expression of marker genes in these transplantable cell cultures that include such stages in muscle differentiation. The myoD regulatory sequence also offers the advantage that when the transplanted myoblasts have become established in the host, and differentiated into myotubes, MyoD expression is down-regulated. Thus, the invention offers the advantage of providing transient expression of a gene of interest in circumstances where this is desirable, or where it is desirable to have gene expression during development, (e.g., when treating developmental disorders, genetic diseases, and the like), or where it is desirable to have a gene that is actively expressed when muscle cells are proliferating and silent when they are not proliferating.

Fibroblasts may actively suppress transcription of the myoD gene, and this may account for the lack of expression of this gene in some non-muscle cells. Deletional analysis, similar to that described below, in conjunction with transfection and transduction can identify the myoD negative regulatory region(s) responsible for suppression of MyoD expression in such non-muscle cells. These suppressor region nucleotide sequences can be separated from the myoD regulatory regions, or replaced by positively acting regulatory regions, and the cells convened from non-muscle cells to muscle cells.

Agricultural uses are particularly contemplated for the myoD regulatory region. For example, expression of growth factor genes (e.g., encoding insulin, insulin-like growth factor, and the like) is useful when regulated by the myoD regulatory region for stimulating exuberant production of skeletal muscle in domestic animals. By using the myoD regulatory region to drive expression of such growth factors during the muscular development of a young animal, the number of muscle cells in the animal is increased relative to a non-treated animal. Since MyoD is normally down-regulated in myotubes, this intervention allows expression of the growth factor gene only during the developmental period, and leaves mature tissue free from the growth factor gene expression. As the increased number of myoblast mature into myotubes, a more rapid acquisition of muscle mass and/or larger muscle mass results than in untreated animals. This phenotypic result is commercially desirable in domestic animals, particularly in food animals such as beef and poultry.

Myotubes contain multiple nuclei, but it is known that some genes regulated by MyoD and Myogenin are expressed only in the nucleus that is under the synaptic cleft. The myoD regulatory region expression vectors pursuant to this disclosure can be used to determine which of the sub-synaptic nuclei in a myotube selectively express MyoD protein. Selective expression of MyoD by a nucleus allows the myoD regulatory regions vectors of this disclosure to be used for targeting gene expression, i.e., to a particular nucleus in a myotube (e.g., to provide a source of a protein of interest at a specific site within a myotube such as at a synaptic cleft).

The invention provides expression vectors useful for identifying nuclear regulatory proteins, e.g., transcription regulatory factors, that bind to the regulatory elements delineated in the proximal and distal portions of the disclosed myoD regulatory region. In a representative example of a positive selection method, myoD regulatory DNA or RNA is placed upstream from a marker gene, e.g., an enzyme such as .beta.-galactosidase, and the construct is transfected or transduced into a series of mutant host cells, e.g., treated with a mutagenic agent causing DNA deletions, substitutions, or replacement. In this case, absence of expression of the marker gene in a host cell indicates the existence of a defective nuclear regulatory factor binding to the myoD regulatory region. Test assays and cell systems are thereby provided for cloning nuclear regulatory factors that influence the functional activity of the myoD regulatory region. In another example involving a negative selection method, a drug resistance marker gene is placed downstream from a myoD regulatory region. In this case, absence of expression of the marker gene in the host cell results in the death of the host cell, and by using replica plating techniques mutants that have altered nuclear regulatory factors influencing the disclosed myoD regulatory region are readily identified. Other representative assays useful for identifying nuclear factors binding myoD regulatory DNA is provided below in Example 4.

It will be recognized by those skilled in the art that hybridization of nucleic acid with the disclosed mouse myoD regulatory region(s), or construction of test oligonucleotide probes based on the disclosed MyoD regulatory sequences, permissibility employing the test assays and schematic experimental approach of subtractive deletion and evaluation of marker gene expression and myogenesis outlined below in the Examples, provides the necessary information to readily identify corresponding MyoD regulatory regions in other animal species.

To isolate the myoD gene, a mouse liver cosmid library was screened with a MyoD cDNA. Seven cosmids were isolated that hybridized to the MyoD cDNA and restriction analysis demonstrated that these represented five distinct, overlapping cosmids that encompassed approximately 60 kb. The cDNA hybridization was limited to a region of about 3 kb. Sequence from this region confirmed the identity of the gene as myoD and demonstrated the presence of two introns (positions 1680 to 2125 and 2205 to 2523, FIGS. 1A-C), the first of which begins after the HLH domain. The sequence in FIG. 1A-C, entitled MYOD GENOMIC, also shows the nucleotide sequence from approximately position -840 (i.e., upstream from the major transcriptional start site) to position +2790. The genomic sequence is distinguished from the cDNA by showing three exons, two introns, and a short sequence of the 3' untranslated cDNA sequence (positions 3427 to 3443, FIG. 1C) that probably represents a divergence in the genomic sequence as a result of genetic polymorphism in this region between the C3 H strain cells from which the cDNA was cloned and the A/J strain from which the genomic cosmid library was prepared, i.e., rather than a short intron.

RNAse protection assays on myoblast and myotube total RNA were used to determine the site of transcription initiation. The major fraction of the RNA extended approximately 40 nucleotides 5-prime of the previously characterized cDNA, and a minor fraction extended an additional 20 nucleotides 5-prime. These two putative transcription initiation sites were also obtained using primer extension assays. There was no apparent difference in the utilization of the two start sites in myoblasts or myotubes, nor was there a significant difference between azacytidine derived myoblasts and the C2 myoblast cell line.

C2 myoblast and myotube DNA were analyzed for hypersensitive sites using DNAse digestion. Sites were present in the promoter region and approximately 6 kb upstream of the start site. In addition, 3-prime of the start site, sites were present near the polyadenylation site and approximately 2 kb further 3-prime. No difference was noted between myotubes and myoblasts.

These studies identified a muscle-specific promoter region between positions 732 and 772 (as depicted in FIG. 1A) which together with the region between position 802 and 837 (FIG. 1A) was sufficient for induction of muscle-specific transcription. The regulatory region contains five interesting motifs: namely,

1. approximately 20 nucleotides upstream of the major start site is an AT-rich region similar to a TATA box, i.e., at positions 823 to 829, FIG. 1A, *1;

2. an inverted CAAT box, i.e., at positions 758 to 763, FIG. 1A, *2;

3. a near match of an M-CAT binding sequence, i.e., at positions 751 to 757, FIG. 1A, *3;

4. at least one SP1 site, i.e., at positions 740 to 745 as shown in FIG. 1A, 4, and 5. The MyoD binding sequence, i.e., at positions 677 to 682, FIG. 1A, the position of these sites in relation to the major start site for transcription is depicted in FIG. 2: namely, the inverted CAAT sequence ("C1"); a six-out-of-seven match to the M-CAT binding sequence ("M"); two potential SP1 binding sites ("S"); and the CANNTG core sequence ("A") creating a binding site for HLH proteins, including MyoD. A near direct repeat of a sequence surrounding the site at position -190 (FIG. 2, "A", nucleotides 666-675 of SEQ ID NO: 1) is found at position -260 (FIG. 2, "B", nucleotides 596-606 of SEQ ID NO: 1). Gel retardation assays using synthetic oligonucleotides demonstrated that the "A" site at -190 (FIG. 2) had a similar affinity for MyoD:E12 heterodimers as the MEF1 binding site (the "right" site) from the MCK enhancer. The "B" site at -260 (FIG. 2) also binds MyoD:E12 heterodimers, but with a lower affinity.

Portions of the myoD gene 5-prime to the point of translation initiation were cloned upstream of the bacterial reporter gene chloramphenicol acetyltransferase (CAT). These constructs were transfected into mouse muscle cell lines (both the C2 cell line and azacytidine derived myoblast lines) and into primary chick muscle cultures. Chimeric constructs were prepared using the regions upstream of the major transcriptional start site to drive the CAT reporter gene.

A deletion series of mutants was constructed (as depicted in FIG. 3) to study the functional activity of these upstream regions in mouse and chicken cells. (The "B", "A", "S", "M" and "C" sites are schematically depicted in FIG. 2, above.) In transient transfection assays using the complete upstream putative regulatory region of the myoD gene, mouse myocyte cell lines did not reliably demonstrate CAT activity above background levels. Surprisingly, and in distinct contrast, transient transfection of the regulatory region constructs into primary chicken myoblast cells did demonstrate high transcription activities in these cells (as determined by CAT activity). Studies of the deletion series of mutants in transient transfection of primary chick myocytes (CM) (using chick fibroblasts (CF) as a control) demonstrated the region between -270 and -50 conferred high levels of activity (FIG. 3, "CM"), whereas additional upstream sequence of up to 7 kb did not enhance this level of activity. (Note that the data presented in FIG. 3 for percentage expression of CAT in these cells have been normalized to 100%, relative to the -270 bp regulatory region construct and corrected for relative transfection efficiency.) This activity was specific to muscle cells in that the constructs were inactive in chick dermal fibroblasts. RNAse protection demonstrated that the constructs faithfully replicated the start sites of the endogenous gene in mouse muscle cells.

Additional deletions were made in the -270 bp regulatory region construct and these were again assayed for activity in primary chick muscle cultures and chick fibroblasts. Of interest is that the two MyoD binding sites (i.e., "A", "B", FIG. 2) can be deleted without any measurable loss of activity. In fact, a region of only 40 nucleotides in addition to the region of transcriptional initiation was both necessary and sufficient for nearly full activity as compared to the larger constructs. These 40 nucleotides contain the upstream SP1 site, i.e., "S" -110, FIG. 2, the CAAT box, i.e., "C1", FIG. 2, and the 6/7 match for the M-CAT sequence, i.e., "M", FIG. 2. Both the -110 SP1 site and the CAAT box were necessary for activity, since deletions disrupting either resulted in less than 10% wild-type activity.

To investigate the possible functional significance of the region containing the MyoD binding sites, i.e., "A+ and "B" , FIG. 2), the region between -260 and -90 was cloned upstream of the TK promoter driving a CAT reporter gene. Co-transfection of myoD constructs (or control constructs) with these CAT-reporter constructs into 10T1/2 fibroblasts with either MyoD demonstrated that MyoD led to a 5-10-fold trans-activation of the constructs containing the potential "A" and "B" MyoD binding sites (FIG. 2), supporting the notion that the binding sites may have a function in the auto-regulation of the myoD gene.

Gel retardation assays were utilized to search for tissue-specific factors which might interact with the proximal regulatory region of myoD. Double strand synthetic oligonucleotides were used in the assays with nuclear extracts from cultured chick muscle or chick dermal fibroblasts.

An oligonucleotide containing the upstream SP1 site (-110 "S" , FIG. 2) interacted with nuclear proteins in both dermal fibroblasts, skeletal muscle and skeletal muscle substituted with BrdU. In contrast, only skeletal muscle nuclear extracts contain a factor(s) that interacted with oligonucleotides containing the M-CAT ("M", FIG. 2) and CAAT ("C1", FIG. 2) sequences. Competition with an oligonucleotide containing a consensus M-CAT binding site, but no CAAT site, indicates that while some of the minor species of faster mobility may be specific to M-CAT binding factors, the predominant slow migrating species is not an M-CAT binding factor, and therefore is likely to be a CAAT binding protein(s). Of particular interest is the absence of the CAAT binding factor(s) in chick skeletal muscle substituted with BrdU. Together with the transient assays that show decreased activity of the proximal region constructs, the data are consistent with the notion that BrdU suppresses the expression or activity of a CAAT binding factor present in chick skeletal muscle that is necessary for the transcription of myoD. Thus, the combined results identify trans-acting transcription regulatory factors in chick skeletal muscle nuclear extracts that bind to at least the -110 SP1 site ("S", FIG. 2), the CAAT site ("C1", FIG. 2), the M-CAT site ("M", FIG. 2), and MyoD binding sites ("C2", "A", and "B", FIG. 2).

Nuclear extracts from the murine myoblast cell line C2C12 and the murine fibroblast cell line C3H10T1/2 (10T1/2) were also used in similar gel retardation assays. Both cell lines contained factors that interacted with the SP1 and CAAT containing oligonucleotides. The presence of the CAAT binding factor in the 10T1/2 cells may be relevant to the ability of this cell line to initiate MyoD expression either by treatment with the demethylating agent 5-azacytidine or by transfer of a myoD containing chromosome from a primary fibroblast.

A distal regulatory enhancer region (DRR) was identified that confers myogenic activity in mouse myoblasts and myotubes. Since it is known that MyoD RNA and protein are expressed in replicating myoblasts as well as in myotubes, but in transient assays with mouse muscle myoblast lines we were unable to consistently demonstrate functional activity of our myoD-CAT reporter constructs in these cells (noted above), the possibility was investigated that other regulatory regions may exist. The regulatory region conferring muscle specific activity of the myoD-CAT reporter gene in the mouse cells was localized using deletion mutants to an element between -7 kb and -4 kb (FIGS. 4A-B). This element was functional only following stable integration but was not functional in transient assays. The sequence entitled "KX FRAG MYOD ENHANCER" is 1978 nucleotides spanning the region between an Apa I site and a Xho I site at approximately -5870 and -3892 nucleotides upstream from the major start site of myoD gene transcription. The results of additional deletion analysis demonstrate that the necessary region of the enhancer lies between approximately -5390 and - 4670, i.e., between positions 421 and 1200 of the sequence represented in FIG. 4A (see FIG. 4A). However, it is noted that this larger regulatory region also includes other nucleotide sequence motifs that will most likely provide important additional regulatory elements. These regulatory sequence motifs represent:

1. two sequences approximating MEF2 binding sites, i.e., *7; *12; positions 312 to 321 and 1097 to 1105, FIG. 4A;

2. sequences representing consensus binding sites for helix-loop-helix regions in regulatory trans-acting bHLH proteins such as MyoD, i.e., *5; *9, *10, *11; positions 265 to 270, 486 to 491, 608 to 613, and 741 to 746, FIG. 4A;

3. an M-CAT binding site similar to that described in FIG. 2A, i.e., *3; positions 163 to 169, FIG. 4A;

4. an AP1 binding site, i.e., *4; positions 254 to 260, FIG. 4A;

5. an SP1 binding site similar to that described in FIG. 2A, i.e., *6; positions 291 to 297, FIG. 4A; and,

6. a CArG Box or SRF site, i.e., *8; positions 467 to 476, FIG. 4A.

An analysis of myogenesis (myotube, "Tubes"; and myoblast, "Blasts") in mouse myocyte C2C12 cells shows that both the DRR enhancer and the PRR promoter regions are required to achieve expression (i.e., "C1", FIG. 2). The results with CAAT are similar to the results obtained with primary chick muscle cultures (above), wherein deletion of the CAAT binding sequence in the proximal regulatory region (PRR) resulted in a significant decrease in CAT activity expressed from a DRR-containing myoD-CAT construct.

In contrast to the myogenic activity of the myoD-CAT constructs in myoblasts and myotubes, i.e., described above, stable integrants of these constructs in non-myoblastic 10T1/2 cells were inactive, similar to the inactivity observed for the endogenous 10T1/2 myoD gene in these cells. Since it is known that the endogenous 10T1/2 myoD gene can be activated either by treatment with 5-azacytidine or by forced expression of a transfected MyoD cDNA, this method of auto-activation was tested to see how this auto-activation of myogenesis affected expression from the regulatory sequences. To investigate the regulatory effects operative during azacytidine-induced autoactivation, a polyclone of 10T1/2 cells that had been transfected with the 7 kb myoD-CAT DRR reporter construct was converted to myoblasts either by treatment with azacytidine or infection with a MyoD expressing retrovirus, and then CAT expression was evaluated. In the case of both azacytidine and retrovirus infection, the 7 kb myoD-CAT construct was activated coincident with the activation of the endogenous myoD gene. We conclude from these experiments that the 7 kb myoD-CAT DRR construct contains the requisite nucleotide sequence and elements sufficient for auto-regulation, since in each case the CAT activity was coincident with the expression of the endogenous myoD gene; however, the results do not allow us to distinguish between auto-activation and de novo activation.

The experiments above demonstrate that the distal and proximal regulatory elements, i.e., DRR and PRR, confer myoblast-specific activity on a reporter construct and that these regulatory regions alone appear sufficient for a) expression of myogenic activity in murine myoblasts and myotubes and b) auto-regulation in 10T1/2 cells, i.e., the ability of the cells to up-regulate MyoD expression in the presence of environmental challenges such as 5'azacytidine or exogenous MyoD expression. However, the inactivity of the 7 kb myoD-CAT DRR constructs in mouse non-myoblastic 10T1/2 cells, i.e., prior to the activation of the endogenous gene, was considered. It was reasoned that either unknown cis- or other transacting elements might account for "triggering" of auto-activation, or alternatively, triggering might be quantitative, i.e., with small amounts of MyoD (or MyoD mRNA) being produced which feed-back and induce additional MyoD expression. In the latter case, our CAT reporter gene constructs might not be capable of autoregulation because the transcribed product was CAT instead of MyoD. If this were the case, then transfection of 10T1/2 cells with myoD genomic fragments that contained both the DRR and PRR driving the MyoD coding region might be sufficient to convert them to myoblasts without the requirement for exogenous environmental agents. Therefore, 10T1/2 cells were cotransfected with a) the selectable marker for neomycin resistance driven by a viral LTR and b) a genomic fragment of myoD extending from -12.0 kb to several hundred nucleotides past the poly-A site; this region included the entire DRR and PRR plus the coding region with the presumptive poly-A site. A control vector was also made to test the functional integrity of the coding region and poly-A site of the genomic constructs by cloning the MSV-LTR upstream of the MyoD coding region. Clones of stable transfectants were selected in G418 and switched to differentiation medium to score for myogenesis. While the MSV-LTR-driven genomic sequence yielded muscle colonies at a frequency similar to transfection with the LTR-driven cDNA, the genomic colonies had only a few scattered muscle cells. This finding would be consistent with either the absence of the appropriate trans-acting regulatory factor(s) or the lack of all the necessary cis acting regulatory elements. Similar experiments were therefore conducted wherein 10T1/2 cells were transfected with the entire cosmids from which the genomic clones were derived. Similar results were recorded. These genomic cosmids contained as much as 20 kb both 5-prime and 3-prime to the gene, suggesting that any putative cis regulatory elements sufficient for activity in 10T1/2 cells would have to be either (i) very distant from the gene (i.e., >20 kb), (ii) deleted from the genomic cosmid constructs, e.g., by rearrangements, or (iii) silenced by a cis modification, e.g., following transfection, or (iv) represented as a general polymorphism-conferring function in the C3H strain but not in A/J strain of mouse. Thus, while not ruling out the presence of additional cis regulatory elements, these combined results support the notion that the major regulatory elements controlling myoD gene expression are trans-acting regulatory elements which exert their effects upon elements in the PRR and DRR 5' regions of the myoD gene.

In light of the above, the 40 nucleotide region of the myoD PRR that is necessary for muscle-specific activity contains the SP1, M-CAT, and CAAT binding motifs. SP1 sites are reportedly found in other muscle-specific regulatory regions, e.g., the alpha cardiac actin promoter, and are thought to represent general activation transcription factors without cell type specificity. The gel shifts with SP1 sequences from the myoD PRR (above) are consistent with the interpretation that a ubiquitously expressed factor binds to this site.

The M-CAT binding sequence was reported in the Troponin I promoter. In that case two intact M-CAT sites are necessary for expression of this gene in muscle cells. In gel retardation assays using an M-CAT oligo the results suggest that only a minor fraction of the shifted material can be attributed to M-CAT binding.

CAAT boxes are reportedly found in the promoters of many genes and are often incorporated into a CArG box in muscle promoters. The CArG box is a CC(A/T).sub.6 GG sequence that binds SRF and other factors. While the CAAT box in the myoD PRR is not incorporated into a CArG box, it is part of a CC(A/T).sub.4 GG motif, which, while similar to a CArG box, has not been described as having any functional relevance in other muscle genes. The CAAT box binding protein CEBP has been reported to be important for the tissue-specific expression of genes in adipocytes. The suppression of CAAT box binding activity in BrdU substituted chick myoblasts (above) raises the possibility that BrdU inhibits MyoD transcription and myogenesis in chick cells by inhibiting the transcription of this factor. Since CAAT box binding proteins are reportedly necessary for tissue-specific transcription in other cells types, it is possible that BrdU may inhibit the expression CAAT binding proteins that are involved in the transcriptional activation of regulatory genes. BrdU does not inhibit CAAT binding factors in mouse cell lines and therefore locating the region of the myoD regulatory sequences that respond to BrdU in mouse cells will define an important regulatory domain.

The fourth sequence motif present in the PRR is the TATA box. The sequence of the TATA box is conserved in the Xenopus and chicken myoD gene. This strong conservation of sequence makes it plausible that the TATA element may confer some muscle specificity, e.g., such as has been reported for the myoglobin and chicken cardiac actin genes.

The fifth region of interest is the MyoD binding sites upstream from the PRR. While deletion of these sites does not affect gene expression in a transient assay in primary chick muscle cells, the ability of the sites to act as a regulatory region responsive to trans-activation by MyoD can be demonstrated when they are joined to the TK promoter or the MCK promoter. It is likely that these MyoD binding sites modulate activity in a manner that is not necessary in chick muscle cells.

The DRR is necessary in addition to the PRR for muscle-specific activity in mouse muscle cell lines. Overlapping deletions of the 5-prime region of the myoD gene were useful in mapping a necessary region of the DRR between approximately -5390 bp and -4670 bp. An unusual feature of the DRR is the requirement for integration to achieve enhancer activity. While this requirement for integration is also reportedly true for the hemoglobin LAR, in neither case is there an understanding of the role integration plays in regulating the activity of these regions. At this time it is only possible to speculate on why the myoD DRR is relatively inactive in transient assays: i) The factors interacting with the DRR may be in very limited supply and transient assays may titrate-out the available factors. This interpretation presupposes the independent binding of two or more factors. ii) The DRR may be insufficient for activity by itself, but integration near a second necessary element may permit activity. Since approximately 50% of stable colonies show activity, any such putative additional regulatory element would have to be frequently represented in the genome. iii) Chromatin structure may be necessary to expose regulatory domains. iv) It may be necessary to remove bacterial methylation patterns that could interfere with factor binding.

A second curious feature of the DRR is that it is required for PRR activity in mouse muscle, but not in chick muscle. In part, this may have to do with a higher abundance of activation factors in chick, but it could also reflect a different strategy of regulating MyoD expression. Since both C2C12 myoblasts and 10T1/2 cells contain abundant amounts of SP1 and CAAT binding factors, but do not activate transcription without the DRR, it is reasonable to conclude that simple abundance of the identified factors is not the limiting step in transcriptional activation.

Neither the myoD-CAT constructs nor the myoD genomic sequences demonstrate significant activity when transfected into 10T1/2 cells. In most aspects, the activity of the transfected CAT constructs closely match the activity of the endogenous gene. Both the myoD-CAT constructs and the endogenous gene are inactive in 10T1/2 cells, but can be activated either by expression of an exogenous MyoD or by treatment with 5-azacytidine, although it is possible that the latter might be activating the CAT construct through the activity of the endogenous MyoD. Another similarity between the myoD-CAT constructs and the endogenous myoD gene is that they are both activated by MyoD in the presence of either high serum or TGF-Beta. This is in sharp contrast to other defined muscle-specific enhancers for which serum and TGF-Beta reportedly act to inhibit the activity of MyoD.

The specificity of the myoD regulatory regions was investigated in mouse C2 muscle cells or 10T1/2 fibroblasts transfected with either the -7 kb myoD-CAT regulatory construct or a -4 kb myoD-CAT regulatory construct. Stable integrants were selected and approximately 1000 such clones were pooled and analyzed for CAT activity. The results presented in Table 1, below, show substantially background levels of activity in fibroblasts and significant activity in muscle cells.

The isolated myoD gene regulatory DNA sequences encompassing the PRR and DRR sequences were utilized to construct two retrovital expression vectors, termed "LHDMDN.NSA" (i.e., containing 1445 bp of myoD regulatory sequence) and "LHDMDN.531" (i.e., containing 2244 bp of myoD regulatory sequence). Both vectors have the same myoD regulatory sequence, both contain the PRR and DRR sequences, and both were found to have the same functional activity. For purposes of simplicity, only the "LHDMDN.531" vector has been deposited with the American Type Culture Collection, Rockville, Md.

The construction of retroviral vectors containing myoD regulatory RNA will now be described.

The test vectors, LHDMDN.531 and LHDMDN.NSA, consisting of a standard Moloney Murine Leukemia Virus (MoMLV) backbone that carry two selectable markers: namely, (1) the enzyme Histidinol Dehydrogenase (hisD; "HD"), conferring drug resistance to L-histidinol; and, (2) the enzyme Neomycin Phosphotransferase (neo; "N"), conferring resistance to the neomycin analogue G418. The hisD gene in the construct is under the regulatory control of the promoter region in the MoLV long terminal repeat sequences (LTR; "L"), while the neogene was placed under the control of the myoD PRR and DRR regulatory sequences containing the 5' enhancer and promoter regions of the myoD gene (ME)P; "MD"). The LTR promoter is operative in directing expression of downstream genes in a wide variety of cell types, while MDP directs preferential expression of downstream genes in early replicating myoblasts. The arrangement of these two marker genes in the construct allows ubiquitous expression of HisD, i.e., as a selectable marker in both muscle cells and non-muscle cells, e.g., in fibroblasts. The MDP driven neogene is used as a muscle-specific selectable marker. (Materials and methods of these constructions, infections of cells, and assays for marker enzymes in cells are described below in the section entitled "Materials and Methods".)

Although retroviral particles will efficiently deliver vector sequences into the genome of a target cell, a drug-resistance phenotype is only induced if the selectable marker gene is expressed at a sufficiently high level following integration. The relative titer of a virus preparation depends in part on the level of expression achieved in successfully infected target cells. Thus, if exceedingly high levels of drug are used, none of the infected cells will express sufficient amounts of the drug resistance gene marker to counter the high selective pressure. In the case of the LHDMN vectors, expression of the hisD gene from the LTR should provide sufficient gene product in both muscle and non-muscle cells to exhibit reasonably high titers under L-histidinol selection. On the other hand, the MDP driven neogene should express less neogene product in non-muscle cells than in muscle cells. This should allow for the preferential selection of muscle cells under sufficiently stringent selective conditions.

To test whether the LHDMDN vectors expressed the MDP driven neo gene in a muscle specific manner, C2C12 mouse muscle cells (C2) and NIH3T3 mouse fibroblasts (3T3) were infected with limiting dilutions of LHDMDN virus. Both the 531 and NSA forms of LHDMDN were used and the infected cells were selected in L-histidinol or in G418. Table 2 shows the relative titers of four representative virus producer clones and compares the levels of infection in muscle and non-muscle cells (ratio of the C2 viral titers to the 3T3 viral titers).

Under the selective conditions used here (i.e., 8 mM L-histidinol; 1.5 mg/ml G418), a clear tissue-specific differential expression of G418 resistance can be seen with 200-300-fold higher titers in muscle cells than in fibroblasts (ratio C2/3T3). For comparison, L-histidinol selection of myoblasts or fibroblasts shows at most a 7.5 fold difference between muscle cells and fibroblasts (ratio C2/3T3). This suggests that the MDP expression elements were able to provide muscle-specific enhanced gene expression in a cell and further that the elements were preserved in an operative fashion while integrated in the genome in a retroviral context.

To more closely examine the difference in neo expression in muscle cells and fibroblasts, the specific activity of neomycin phosphotransferase (NPT) was determined for both the C2 and 3T3 cells infected with the LHDMDN.NSA vector. As a positive control for NFT activity, cells were infected with the LHDSN vector an selected in parallel. LHDSN is identical to LHDMDN except that the neogene expression is controlled by the strong SV40 promoter which is active in both myoblasts and fibroblasts. Table 3 lists the activity of NPT found in bulk populations of LHDMDN.NSA- or LHDSN-infected C2 and 3T3 cells which were selected in L-histidinol.

Comparison of SV40 mediated NPT activity with the activity of MDP expressed neo shows that the MDP enhancer-promoter produces nearly as much NFT activity as SV40 in C2 cells (i.e., 80% of SV40 values) while in 3T3 fibroblasts MDP produces about one-third of the NPT activity of SV40. When comparing MDP directed NPT activity in C2 vs. 3T3 cells, a two-fold difference is seen, again showing the cellular-specificity of the regulatory effects of the MDP elements. It is worthy of note that although both comparisons of NFT activity clearly show that the MDP elements are preferentially expressed in muscle cells, the differences with retroviral constructs were actually lower than expected based on transfection analyses (above) wherein differences of 20- to 50-fold between expression in C2 muscle cells and 10T1/2 fibroblasts were common observed using NSA MDP. One possible explanation for this effect of retroviral vectors is that the strong enhancer elements in the viral LTR act at a distance to promote expression from the MDP promoter in non-muscle cells, i.e., in the controls.

PA317 cells (a derivative fibroblast cell line) were transduced with EMC11.S, a plasmid containing an RSV promoter driving expression of MyoD protein coding sequences. When exogenous MyoD protein is expressed in fibroblasts it induces expression of a number of myoD-regulatable proteins in the fibroblasts and thus yields cells that are permissive for MDP enhancer-promoter directed gene expression.

A clone of myoD transducer PA317 cells which expressed high levels of MyoD protein, PAMD308, was chosen for use in further studies. To test that the PAMD308 cells were permissive for MDP-directed expression, LHDMDN virus, (i.e., wherein the viral LTR drives ubiquitous expression of histidinol resistance but the myoD regulatory DNA drives expression of muscle-specific expression of neomycin-resistance), was used to infect PA317 cells and PAMD308 cells. The number of L-histidinol was compared with the number of G418 resistant colonies. Table 4 shows the ratio of L-histidinol or G418 resistant colonies in PAMD308 vs. PA317 cells. Little difference is seen between the cell lines with L-histidinol selection (1-3.4-fold higher titer on PAMD308 cells). In contrast, selection with G418 shows a 100-500-fold increase in infection rate when myoD protein is present. This demonstrates an enhanced expression level of the MDP enhancer-promoter in the presence of MyoD protein, and shows that in the presence of MyoD there is selection for muscle cells, i.e., for cells that are resistant to G418 because of their expression of MyoD.

pLHDMDN was constructed by removing the SV40 promoter from pLHDSN (30) and replacing it with sequences containing the myoD PRR and DRR enhancer-promoter elements. Two vectors were constructed using MDP sequences which conferred muscle specific expression in transfection studies. The two MDP elements were termed "531" and "NSA". The resulting plasmids containing each vector were termed pLHDMDN.531 and pLHDMDN.NSA.

Cell lines producing LHDMDN.531 or LHDMDN.NSA virus were isolated using standard techniques (31) involving transfection of the vector construct into PE501 cells followed by infection of PA317 cells with transiently expressed virus from the PE501 cells. Infected PA317 cells were selected in L-histidinol and the resulting drug resistant clones were tested for LHDMDN virus production.

Infection protocols were conducted essentially as described previously (31). Briefly, virus containing medium was harvested from producer cells and was used in limiting dilution analyses to infect target C2 (muscle) or 3T3 (fibroblast) cells which had been plated at a density of 5.times.10.sup.5 cells per 6 cm dish on the previous day. After 24 hrs. of exposure to virus, the target cells were treated with trypsin and 1/10th and 1/50th of the cells were replaced in the presence of the selective drug. The concentration of drugs used for selection was 4 mM L-histidinol or 1.5 mg/ml G418 (50% active) for both the C2 and 3T3 cell cultures. In later assays, the L-histidinol was increased to a concentration of 8 mM to facilitate selection of the more resistant C2 cells. Relative titers (i.e., of virus) remained the same with this increase in drug concentration.

The activity of the neogene product, neomycin phosphotransferase (NPT), was determined as follows: briefly, 5 ug total soluble cellular protein (i.e., in a cellular extract) was incubated for 30 minutes at 37.degree. C. in the following reaction mixture: 67 mM Tris-Maleate pH 7.1, 42 mM MgCl.sub.2, 400 mM NH.sub.4 Cl, 1 uM rATP/gamma.sup.32 P-rATP, 0.5 mM neomycin sulfate. The reaction was stopped by phenol extracting the reaction mixture and aliquots of the aqueous phase were spotted onto P-81 ion exchange paper. The paper was extensively washed and .sup.32 P CPM determined. NPT activity is expressed as the CPM incorporated into neomycin sulfate per ug soluble protein per minute.

1. Lassar, A. B. et al. 1989. Cell 58: 823.

2. Davis, R. L. et al. 1987. Cell 51: 987.

3. Wright, W. E. et al. 1989. Cell 56: 607.

4. Edmonson, D. G. et al. 1989. Genes Dev. 3: 628.

5. Braun, T. et al. 1989. EMBO J. 8: 701.

6. Rhodes, S. J. et al. 1989. Genes Dev. 3: 2050.

7. Miner, J. H. et al. 1990. Proc. Natl. Acad. Sci. USA 87: 1089.

8. Braun, T. et al. 1990. EMBO J. 9: 821.

9. Weintraub, H. R. 1990. Proc. Natl. Acad. Sci. USA 87: 5623.

10. Piette, J. et al. 1990. Nature 345: 353.

11. Rosenthal, N. et al. 1990. Nucleic Acids Res. 18: 6239.

12. Sartorelli, V. 1990. Genes Dev. 4: 1811.

13. Murre, C. et al. 1989 Cell 58: 537.

14. Brennan, T. J. 1990. Genes Dev. 4: 582.

15. Davis, R. L. et al. Cell 60: 733.

16. Sun, X. H. et al. 1991. 64: 459.

17. Henthorn, P. et al. 1990. Science 247: 467-470.

18. Blackwell, T. K. et al. 1990. Science 250: 1104.

19. Davis, R. & H. Weintraub unpublished observations.

20. Schafer, B. W. et al. 1990. Nature 344: 454.

21. Benezra, R. et al. 1990. Cell 61: 49-59.

22. Lassar, A. B. et al. 1989. Cell 58: 659.

23. Sasson, D. et al. 1989. Nature 341: 303.

24. Sorrentino, V. R. et al. 1990. Proc. Natl. Acad. Sci. USA 87: 8442.

25. Weintraub, H. et al. 1989. Proc. Natl. Acad. Sci. USA 86: 5434.

26. Choi, J. et al. 1990. Proc. Natl. Acad. Sci. USA 87: 7988.

27. Thayer, M. et al. 1990. Cell 63: 23.

28. Ptashne, M. et al. 1990. Nature 346: 329.

29. Lin, A. Y. et al. 1989. Gene Dev. 3: 986.

30. Stockschalecler, M. et al. 1991. Human Gene Therapy 2: 33-39.

31. Miller, A. D. et al. 1986. Mol. Cell. Biol. 6: 2895.

While the preferred embodiment of the invention has been illustrated and described, it will be appreciated that various changes can be made therein without departing from the spirit and scope of the invention.