WO2001083716A2 - Immortalized lines of endothelial brain cells and therapeutic application thereof - Google Patents

Abstract

The invention relates to optionally modified immortalized lines of endothelial brain cells of mammalians, as well as applications as preventive or curative drug, and particularly for the treatment of primary and secondary, neurological or psychiatric diseases, inclding Alzheimer's disease, Huntington's disease, Amyotrophic Lateral Sclerosis (Lou Gehrig's disease), Parkinson's disease, glioblastoma and other brain tumors, and stroke. The invention also relates to the method for preparing the cell lines. The endothelial cell lines of mammalians disclosed are comprised of immortalized endothelial brain cells presenting characteristics of differentiated endothelial brain cells, in a stable way. The cell lines comprise a nucleic acid having at least one immortalizing viral or cellular oncogene, optionally associated with at least one selection gene, and an expression vector comprising a sequence coding for polypeptide, a protein, or a viral vector, optionally associated with at least one selection gene and optionally at least one marker gene, and they are capable in vivo to integrate brain vessels of a host mammalian and produce said polypeptide, the protein or the viral vector.

Description

IMMORTALIZED LINES OF ENDOTHELIAL BRAIN CELLS AND THERAPEUTIC APPLICATIONS THEREOF

FIELD OF THE INVENTION The present invention relates to immortalized lines of mammalian endothelial brain cells, where appropriately modified, as well as to their applications as a medicinal product for preventive or curative use, and in particular for the treatment of various primary and secondary neurological or psychiatric disorders or diseases, including Alzheimer's disease, Huntington's disease, Amyotrophic Lateral Sclerosis (Lou Gehrig's disease), Parkinson's disease, glioblastoma and other brain tumors, and stroke.

BACKGROUND INFORMATION For some years, new methods of treatment of a number of neurological disorders, formerly considered to be refractory to all conventional treatments, have made use of gene therapy. These new methods are linked, in particular, to the advances made in the field of the construction of effective expression vectors and of transporters of viral and cellular transgenes, and in the characterization of target cells suitable for gene therapy of the nervous system.

Two different approaches have been proposed for carrying out the transfer of genes into the nervous system: (1) a so-called in vivo approach which focuses on the direct transfer of the genetic material to the cells in vivo, using viral and chemical agents; and (1) an ex vivo approach, which is characterized in that the gene transfer is performed in cells in culture, which are then implanted into the host body. The latter approach includes steps of molecular manipulations, of cloning and of cell implantation, so as to permit the distribution of the active substances in the host (Suhr et al, 50 Arch. Neurol. (1993)).

Many neurological disorders are associated with focused lesions or dysfunctions of the nervous system, and have hence been chosen to test these techniques. The first trials in this field have been related to neurodegenerative diseases such as Parkinson's disease, and include the intracerebral grafting of fetal neural tissue or of adrenal medullary tissue in the brain (Bjδrkland, 14(8) Trends Neurosci. 319-322 (1991)).

The use of primary nervous tissues of fetal origin for cell transplantation in human therapy is a source of numerous ethical and practical problems. An alternative to this problem is to use primary cell lines of neural origin (for example neurons, glial cells, such as astrocytes) or non-neural cell lines (for example fibroblasts, myoblasts, chromaffin cells of the adrenal medulla, hepatocytes) (Gage et al., 14 Trends Neurosci. 328-333 (1991)). Although cell lines of adrenal medulla, of fibroblasts or of myoblasts can actually release active substances in vivo, they are not normally present in the nervous system, can modify the normal function of the nervous system's blood-brain barrier and can give rise to a rejection reaction.

With the object of treating neurological or psychiatric disorders or diseases, including brain tumors, confined or otherwise to a specific region of the brain, it is necessary to have at one's disposal a cellular vector which is able to integrate completely in the nervous tissue while expressing a bioactive substance, in particular a protein, in a stable manner.

SUMMARY OF INVENTION The invention provides a cellular vector which better meets the needs encountered in practice, in particular, in that it expresses, in a stable manner and in vivo, at least one selected polypeptide or protein or viral vector; in that it is of brain origin, capable of integrating in the normal brain vascularization and brain parenchyma; and in that it is well tolerated.

The subject of the present invention is mammalian endothelial cell lines, characterized:

(a) in that they consist of immortalized endothelial brain cells displaying at least one of the following features of differentiated endothelial brain cells, in a stable manner:

( 1 ) the expression of endothelial markers,

(2) the secretion of vasoactive substances,

(3) the expression of molecules of the major histocompatibility complex (MHC),

(4) the expression of hormone receptors, and

(5) the existence of tight junctions,

(b) in that they comprise a nucleic acid comprising at least one immortalizing viral or cellular oncogene, where appropriate in combination with at least one selectable gene, and an expression vector comprising a sequence coding for a polypeptide, a protein, or a viral vector, where appropriate in combination with at least one selectable gene and where appropriate at least one reporter gene, and (c) in that they are capable in vivo of integrating in the brain vessels and brain parenchyma of a host mammal and of producing the peptide, the protein or the viral vector.

The subject of the present invention is also compositions, characterized in that they comprise at least one endothelial brain cell line according to the invention, in combination with at least one pharmaceutically acceptable vehicle. Such compositions preferably contain between 10⁴ and 10⁵ endothelial cells/μl. Such compositions may be advantageously administered via the intracranial, subcutaneous, intracerebroventricular, subdural, venous, or arterial (for example, intracarotid), intramuscular, or intrathecal route.

The subject of the present invention is also a method for obtaining a modified cell line according to the invention, which method is characterized in that:

(a) a first transfection is carried out by: ( 1 ) culturing endothelial brain cells, preferably those of brain microvessels, in a suitable culture medium supplemented with serum and with growth factors,

(2) transfection of the cells between the 2^nd and the 6^th passage with a nucleic acid comprising at least one immortalizing viral or cellular oncogene and, where appropriate, at least one selectable gene, in particular a gene coding for resistance to an antibiotic,

(3) selection of the transfected cells on a selection medium suited to the selectable gene, if necessary,

(b) a transfection of the cells obtained in (a) is then carried out with a vector containing the polypeptide sequence or protein sequence to be produced or a viral vector to be expressed. The subject of the invention is also a model for studying the integration in the brain of cells that deliver active substances to the brain, characterized in that it comprises an RBEZ cell line according to the invention. The subject of the invention is also a model for studying and identifying the biochemical and cellular systems of the blood-brain barrier in vitro, characterized in that it comprises at least one cell line according to the invention. The subject of the invention is, in addition, a method for producing a polypeptide or a protein, characterized in that it comprises the use of at least one endothelial cell line according to the invention, in a suitable bioreactor.

BRIEF DESCRIPTION OF THE DRAWINGS

Besides the foregoing arrangements, the invention also comprises other arrangements which will become apparent from the description which follows, which relates to examples of implementation of the method which is the subject of the present invention as well as to the attached drawings, wherein: FIG. 1 illustrates the in vitro analysis of the expression of the NGF transgene in

RBE/NGF cells, by in situ hybridization. FIG. 1 A is a photograph showing the NGF staining in RBE/NGF cells in culture, using a digoxigenin-labelled antisense oligonucleotide probe specific for murine NGF. FIG. IB is a photograph showing the NGF staining in uninfected control RBE4 cells. FIG. 2 illustrates the stimulation of axonal budding of PC 12 cells, obtained from the supernatant of RBE/NGF cells in vitro.

FIG. 3 also illustrates the stimulation of axonal budding of PC 12 cells, obtained from the supernatant of RBE/NGF cells in vitro.

FIG. 4 illustrates the pre-labeling of RBE4 cells before transplantation, with the nuclear stain Hoechst 33342 (bisbenzimide).

FIG. 5 illustrates the visualization of the cells pre-labeled with the Hoechst stain, after transplantation into adult rat brain. FIG. 5 A shows a general view of the region of grafting in the brain parenchyma. The asterisks symbolize the course of a blood vessel (x250). FIG. 5B and FIG. 5C show a blood vessel at high magnification, located in the region of implantation of the graft, FIG. 5B, numerous Hoechst-positive RBE4 cells integrated in a luminal (arrows) and perivascular position may be observed. In FIG. 5C, this same vessel is immunolabelled with an anti-laminin (specific marker of blood vessels) antibody (x600). FIG. 6 illustrates the analysis of the morphological and functional integration of RBEZ cells, by visualization of the expression of the nls-lacZ transgene and of the antigenic marker of integrity of the blood-brain barrier (BBB), EBA (endothelial barrier antigen). FIG. 6A, FIG. 6B, FIG. 6C, and FIG. 6D show blood vessels located away from the region of grafting, onto which RBEZ cells have migrated after transplantation. In these figures, nuclei of endothelial cells, expressing the nls-lacZ transgene, in a luminal position (arrows) (FIG. 6A, FIG. 6C) may be observed. These same vessels, in fluorescence illumination (FIG. 6B, FIG. 6D), express the SEA antigen (arrow heads). (x750 in FIG. 6A and FIG. 6B; xl500 in FIG. 6C and FIG. 6D; FIG. 6A, and FIG. 6C in transmitted interference contrast). FIG. 7 illustrates the ultrastructural analysis by electron microscopy, demonstrating the morphological and functional integration of RBEZ cells after intracerebral grafting, by visualization of the expression of the nls-/αcZ transgene. FIG. 7A shows, in Nomarski optics, the perinuclear β-galactosidase labelling in the grafted cells on a semithin section of brain (2 pm) (xl660) . On examination in electron microscopy, the cells are observed either in the parenchyma (FIG. 7B) or in a vascular position (FIG. 7C), forming blood vessels of the host. The arrow heads point to the perinuclear precipitates of X-gal, which are dense to electrons.

FIG. 8 illustrates the analysis of the morphological and functional integration of RBEZ cells after grafting into an intracerebral tumor, by visualization of the expression of the nls-lacZ trans gene. FIG. 9 illustrates the identification of the nls-tαcZ gene in tumors implanted with

RBEZ cells, by PCR.

FIG. 10 illustrates the in vivo analysis of the expression of the NGF transgene in RBE/NGF cells, three weeks after transplantation into the nucleus basalis (basal nucleus).

FIG. 11 illustrates the control brain structures used as internal control of the in situ hybridization of the NGF messenger, in vivo.

FIG. 12 illustrates the biological effect of the NGF secreted by RBE/NGF cells, three weeks after grafting, in the nucleus basalis.

FIG. 13 illustrates the biological effect of the NGF secreted by RBENGF cells, three weeks after grafting, away from the nucleus basalis. FIG. 14 illustrates the quantification of the biological effect induced by the expression of the ΝGF transgene at 3 and 8 weeks after grafting, and this is reflected in the area occupied by the p75LΝGFR immunolabeling relative to the area of the graft.

FIG. 15 is a graph showing survival curves for 9L gliosarcoma tumor model rats intracerebrally implanted with 10⁴ 9L cells mixed with 2xl0⁶ TK2 cells. Ganciclovir was i.p. injected daily at a dose of 50 mg/kg from day 0 to day 6 for the treated group (n=10). Control group (n=6) was given i.p. an equal volume of saline during the same period. FIG. 16 is a graph showing survival curves for 9L gliosarcoma tumor model rats intracerebral ly implanted with 10⁴ 9L cells mixed with 2xl0⁶ TK2 cells. Ganciclovir was injected daily at a 100 mg/kg dose for 5 treated animals. The control group (n=5) was injected with saline. FIG. 17 is a graph showing cumulative survival curves for 9L gliosarcoma tumor model rats, comparing rats intracerebrally implanted with a mixture of 10⁴ 9L cells with

3xl0⁶ TK2 cells versus rats intracerebrally implanted with 10⁴ 9L + 3xl0⁶ RBE4 cells (n=5, control group with parental cells).

FIG. 18 is a graph showing cell number at the indicated days in culture (n=2 per point). The doubling time is 20 hr and cells reach confluency between day 3 and day 4. FIG. 19 is a set of graphs showing the sensitivity of NTC-121 to G418 and

Hygromycin B. The amount of cells/well (mean of triplicates) in OD units (optical density units) is a function of the selective agent concentration in μg/ml. Points on the y axis correspond to wells containing no selective agent. Background values from wells containing only medium and WSTl was subtracted. At day 4, OD=3,000 corresponds to a confluent cell monolayer.

FIG. 20 is a also set of graphs showing the sensitivity of NTC-121 to G418 and

Hygromycin B. The amount of cells/well (mean of triplicates) in OD units (optical density units) is a function of the selective agent concentration in μg/ml. Points on the y axis correspond to wells containing no selective agent. Background values from wells containing only medium and WSTl was subtracted. At day 4, OD= 3.000 corresponds to a confluent cell monolayer.

FIG. 21 is a set of bar graphs showing the % of lysis of NTC-121 and RBE4 cells in presence of the indicated serum dilutions. The following calculation formula was used: % of lysis = [[(NC OD - sample OD) / NC OD]] x 100. 100% of RBE4 and NTC-121 cells are lysed by human (up to % dilution) and rabbit (up to 1/16 dilution) sera in 90 min. Lysis was inhibited by heat-inactivation of the sera.

FIG. 22 is an ultraviolet (UV) epifluorescence and bright field view representing effects of IL-2 on 9L tumors. (A) SVAREC cells, 7 day post-implantataion, X80. (B) SVAREC cells, 14 day post-implantataion, X320, showing graft zone where grafted cells encircle blood vessels. FIG. 23 is an ultraviolet (UN) epifluorescence and bright field view representing effects of IL-2 on 9L tumors. (A) SVAREC cells, 30 day post-implantation, XI 60, showing localization within the grey matter. (B) SVAREC cells, 30 day post-implantation, X320, showing grafted cells in blood vessel walls.

DETAILED DESCRIPTION For the purposes of the invention, "expression vector" is understood to mean any nucleic acid integrated in the genome or present in the cytoplasm, and capable of permitting the expression of the polypeptide, protein, or viral vector. For the purposes of the invention, "immortalized" is understood to mean at least extended proliferation capacity in culture.

According to an advantageous embodiment of the lines, the nucleic acid comprises at least one immortalizing oncogene that contains the neomycin resistance gene and a SV40 T oncogene.

According to another advantageous embodiment of the lines, the nucleic acid comprises at least one immortalizing oncogene contains the El A early region of the adenovirus 2 genome and the neomycin resistance gene.

According to another advantageous embodiment of the lines, the expression vector is a retroviral vector, in particular an MFG vector. Preferably, the retroviral vector is an MFG-NB vector, which is defective for replication. The vectors are described, in particular, in Mulligan et al, 81 Proc. Natl. Acad. Sci. USA 6349-6353 (1984) and Ferry et al, 88 Proc. Natl. Acad. Sci. USA 8377-8381 (1990). Preferably also, the endothelial cells are cells of brain capillaries.

Trials employing non-immortalized primary peripheral vascular endothelial cells have previously been described, but they do not constitute a suitable vector in that they do not constitute a pure, homogeneous, and sufficient source for the purpose of a reproducible application to transplantation, and in that they do not display the endothelial brain phenotype.

According to yet another advantageous embodiment of the cell lines, the sequence coding for a polypeptide or a protein is selected from the sequences coding for enzymes such as proteases; enzyme inhibitors such as protease inhibitors; cytokines; neurotransmitters; neurotrophins; growth factors; toxins; antimetabolites; neurohormones; gangliosides; antibiotics; thrombolytic factors; coagulation factors; vasodilator or vasoconstrictor factors; hypo- or hypercholesterolaemic factors; hyper- or hypoglycemic factors; or any other substance of interest.

According to the invention, the endothelial cells advantageously comprise, as immortalizing gene, the El A early region of the adenovirus 2 genome and the neomycin resistance gene, and, as vector, an MFG-NB retroviral vector containing the nls-/αcZ gene coding for β-galactosidase. This cell line has been designated RBEZ by the inventors. According to the invention, the cell line has been deposited under the No. 1-1481 dated October 10, 1994 with the Collection Nationale de Cultures de Micro-organismes [National Collection of Microorganism Cultures] held by the Institut Pasteur, 28 rue de Docteur Roux, 75724 PARIS CEDEX 15.

Also according to the invention, the endothelial cells advantageously comprise, as immortalizing gene, the El A early region of the adenovirus 2 genome and the neomycin resistance gene, and, as vector, a retroviral vector pMoMuLVisisNGF coding for murine β-NGF. This cell line has been designated RBE/NGF-4 by the inventors. According to the invention, the cell line has been deposited under the No. 1-1482 dated October 10, 1994 with the Collection Nationale de Cultures de Micro-organismes held by the Institut Pasteur, 28 rue de Docteur Roux, 75724 PARIS CEDEX 15.

The subject cultures are deposited under conditions that ensure that access to the cultures will be available during the pendency of the patent application disclosing them to one determined by the Commissioner of Patents and Trademarks to be entitled thereto under 37 C.F.R. § 1.14 and 35 U.S. C. § 122. The deposits are available as required by foreign patent laws in countries where counterparts of the subject application, or its progeny, are filed. However, the availability of a deposit does not constitute a license to practice the subject invention in derogation of patent rights granted by governmental action. Further, the subject culture deposits will be stored and made available to the public in accord with the provisions of the Budapest Treaty for the Deposit of Microorganisms, i.e., they will be stored with all the care necessary to keep them viable and uncontaminated for a period of at least 30 years after the date of deposit or for the enforceable life of any patent which may issue disclosing the cultures plus 5 years after the last request for a sample from the deposit. The depositor acknowledges the duty to replace the deposits should the depository be unable to furnish a sample when requested, due to the conditions of the deposits. All restrictions on availability to the public of the subject culture deposits will be irrevocably removed upon granting of a patent disclosing them.

Unexpectedly, endothelial cells of brain capillaries integrate well in the brain vascularization and brain parenchyma, are very well tolerated, and release in vivo, over a long period, the active substance they express. They find application in the preparation of a composition for the treatment of primary and secondary neurological or psychiatric disorders or diseases (including Alzheimer's disease, Huntington's disease, amyotrophic lateral sclerosis (Lou Gehrig's disease), Parkinson's disease, glioblastoma, and other brain tumors, and stroke) or for stimulating the growth and reproduction of livestock (poultry, sheep, cattle, pigs, horses, lagomorphs, rodents, and the like). In particular, in the basal nucleus and in the striatum, many grafted endothelial cells according to the invention adopt a vascular localization. In both implantation sites, a not insignificant number of grafted cells are not associated with the host's vascular network. This non-vascular localization does not bring about an exacerbation of cell death, even at 1 year after implantation. According to the invention, the endothelial cells may be cells of the same species as the host (allograft or homograft) or of a different species (xenograft).

Preferably, the transfection of step (b) of the method of the invention is carried out with a retroviral vector in which the sequence coding for the protein to be expressed has been incorporated beforehand. According to the invention, the methods of immortalization can be by the use of cellular oncogenes (such as myc, ras, and raj), viral oncogenes, transforming viruses, methods to reduce tumor suppressor activity (such as RB or p53), methods to increase telomerase activity, inactivation of the genes that restrict cell cycle progression (for example, p53 the CDK-4 inhibitor (CDKN2), and prohibitin) by insertional mutation. See also, Cell Immortalization, Macieira-Coelho, Ed., (Progress in Molecular and Subcellular Biology, Vol. 24, Springer Life Science, 1999).

Among the viral oncogenes that can immortalize cells are, for example, T antigens of papovaviruses (e.g., polyoma, JC, SV40), early proteins (e.g., E6, E7) of papillomaviruses, and Epstein-Barr virus (e.g. EBNA-5). In these cases, the viral proteins may interact and inactivate one or more cellular tumor suppressor proteins (e.g., Rb, p53), resulting in a significantly impaired cell cycle regulation. During the perturbed cell cycling, accumulation of mutations may occur either spontaneously or as an effect of other agents (virus, chemical, radiation) in cellular oncogenes (e.g., H-ras, K-ras; c-myc), in tumor suppressor genes (e.g., p53, Rb), or in other cellular genes.

DNA or RNA tumor viruses may mediate multiple changes that convert a normal cell into a malignant one. RNA tumor viruses usually transform cells to a malignant phenotype by integrating their own genetic material into the cellular genome and may also produce infectious progeny. There are two general patterns by which cell transformation may be accomplished: (1) the tumor virus may introduce and express a so-called transforming gene in the cells or (2) the tumor virus may alter the expression or coding capacity of preexisting cellular genes. These mechanisms are not mutually exclusive, and both may occur in the same cell.

Telomerase, the reverse transcriptase that maintains the ends of eukaryotic chromosomes, is association with human cell immortalization. Most normal cells are devoid of telomerase's enzymatic activity and lack its main protein component (human telomerase reverse transcriptase, or hTERT). In normal cells, insufficient telomerase activity and a finite store of telomeric DNA limit the number of divisions a cell can undergo before critical telomere shortening signals entry into replicative senescence, defined by a finite capacity for cell division. Ectopic expression of hTERT in primary human cells could confer endless growth in culture. The telomerase gene is neither tumor suppressor nor oncogene (see, de Lange & DePinho, 283 Science 947-949 (1999)). Preferably, the first transfection of step (a) of the method of the invention enables

RBE4 cells to be obtained, which cells are immortalized by transfection with a plasmid containing the El A early region of the adenovirus 2 genome and the neomycin resistance gene under the control of the SV40 promoter, and which are deposited under the No. I- 1142 with the Collection National de Micro-organismes (CNCM) held by the Institut Pasteur, 28 rue de Docteur Roux, 75724 PARIS CEDEX 15.

In one embodiment, the endothelial cells of the invention can be encapsulated and used to deliver neurotransmitters, according to known encapsulation technologies, including microencapsulation (see, e.g., United States patents 4,352,883; 4,353,888; and 5,084,350, herein incorporated by reference), (b) macroencapsulation (see, e.g., United States patents 5,284,761, 5,158,881, 4,976,859 and 4,968,733 and published PCT patent applications WO 92/19195, WO 95/05452, each incorporated herein by reference). If the cells are encapsulated, we prefer macroencapsulation, as described in United States patents 5,284,761; 5,158,881; 4,976,859; 4,968,733; and 5,800,828, and published PCT patent application WO 95/05452, each incorporated herein by reference. Cell number in the devices can be varied; preferably each device contains between 10 - 10⁹ cells, most preferably 10⁵ to 10⁷ cells. A large number of macroencapsulation devices may be implanted in the subject; we prefer between one to 10 devices.

In the adults, the proliferation rate of endothelial cells is typically low compared to other cell types in the body. The turnover time of these cells can exceed one thousand days. Physiological exceptions include when angiogenesis results in rapid proliferation occurs under tight regulation are found in the female reproduction system and during wound healing. The details of one or more embodiments of the invention are set forth in the accompanying description above. Although any methods and materials similar or equivalent to those described herein can be used in the practice or testing of the present invention, the preferred methods and materials are now described. Other features, objects, and advantages of the invention will be apparent from the description and from the claims. In the specification and the appended claims, the singular forms include plural referents unless the context clearly dictates otherwise. Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. All patents and publications cited in this specification are incorporated by reference. Apart from the foregoing provisions, the invention also comprises other provisions, which will become apparent from the following description referring to EXAMPLES of how to carry out the process forming the subject of the present invention, and to the attached drawings. The following EXAMPLES are presented in order to more fully illustrate the preferred embodiments of the invention. It should, nevertheless, be clearly understood that these EXAMPLES are given only by way of illustration of the subject of the invention, and in no way constitute a limitation of the latter. EXAMPLE 1 PREPARATION OF RBE4 IMMORTALIZED ENDOTHELIAL BRAIN CELLS

Endothelial cells of microvessels of rat brains (primary culture) were immortalized by transfection with the plasmid pEl A-neo, containing the sequence coding for the adenovirus El A followed by the neomycin resistance gene.

A clone designated RBE4 was obtained in this way, and its features were described, in particular, in PCT International patent application WO 93/06222 as well as in the papers published by Durieu-Trautmann et al, 155 J. Cell. Physiol. 104-111 (1993) and Roux et al, 159 J. Cell. Physiol. 101-113 (1994). This clone was deposited under the No. I-l 142 with the Collection Nationale de Cultures de Micro-organismes (CNCM).

To carry out the transfection, the calcium phosphate co-precipitation technique was used, as described in PCT International patent application WO 93/06222 and repeated below. The transfection of the cells was carried out at the 5^th passage with the above-mentioned plasmid (10 μg) containing, besides the El A early region of the adenovirus 2 genome and the neomycin resistance gene, the SV40 promoter. This transfection took place after culturing these cells in collagen-coated dishes containing an α-MEM/FlO (2/3; 1/3) medium supplemented with 10% fetal calf serum (FCS), 1 ng/ml bFGF, glutamine (2 mM), and penicillin/streptomycin. The cell line obtained possesses some of the features of primary endothelial brain cells. It possesses, in particular, an untransformed phenotype; contact inhibition; growth factor- and adhesion factor- dependent proliferation; expression of endothelial differentiation markers (antigen related to factor VIII); binding site for Griffonia simplicifolia agglutinin; and absence of tumorigenic effect in nude mice. Furthermore, these cells were stimulated by astrocytes to express the specific enzymatic markers of the blood-brain barrier, namely glutamine transferase and alkaline phosphatase.

EXAMPLE 2 PREPARATION OF RBEZ ENDOTHELIAL BRAIN CELLS.

The RBE4 cells obtained in EXAMPLE 1 were subjected to 2 passages/week on an α-MEM/FlO (1/1; Seromed, France) medium supplemented with 2 mM glutamine, 10% FCS, 1 ng/ml bFGF, and 300 :g/ml G418. The cells were plated out at a density of 10⁴ cells/cm² on collagen-coated dishes, and used between passages 30 and 60. (a) Preparation of the retroviral vector. An MFG-NB vector, which is defective for replication and contains the lacL gene, was obtained by inserting the sequence coding for E. coli β-galactosidase fused to a sequence coding for the nuclear localization sequence (nls) of 21 amino acids originating from the SV40 T antigen (Kalderon et al, 39 Cell 499-509 (1984). This vector, MFG-NB nls-/αcZ, was introduced into P-2 retrovirus-producing cells (Mulligan et al, loc. cit.) (recombinant retroviral infection of P-2) and enabled P-2-MFG-NB cell lines to be obtained. These retrovirus-producing cells were plated out in dishes at a density of 10⁶ cells per dish 10 mm in diameter in 7 ml of RPMI 1640 medium supplemented with 10% FCS. After 24 hr, a volume of 6 ml of medium containing the virus was filtered and used for infection, or alternatively stored at -80°C until used.

(b) Infection ofRBE4 endothelial cells. RBE4 cells were plated out on dishes at a density of 10⁴ cells/cm² and, after 24 hr, the virus (3 ml) was added in the presence of polybrene (10 μg/ml) for 2 hr. After a further 24 hr period in complete medium, the RBE4 cells were subcultured and reinfected under the same conditions. (c) Selection of endothelial cells expressing the trans gene. The cells expressing β-galactosidase (RBEZ cells) were sorted by FACS (fluorescent activated cell sorting; Nolan et al, 85 Proc. Natl. Acad. Sci. 2603-2607 (1988)) using fluorescein β-D-galactopyranoside (βDG) as substrate for the enzyme. According to this technique, 10⁶ RBEZ cells in 100 μl were incubated at 37°C for 5 minutes (min) in a 5 ml polystyrene tube before adding 100 μl of βDG (2 mM). After mixing, the cells were placed again at 37°C for 1 min and then on ice, and the volume was adjusted to 2 ml.

(d) Detection of the expression of the transgene by visualization of the β-galactosidase enzyme activity using a chromogenic substrate, X-gal

(1) Protocol. The enzyme activity was detected by incubating the cells at 37°C in phosphate-buffered saline (PBS) buffer containing 2 mM 5-bromo-4-chloro-3-indolyl β-D-galactopyranoside (X-gal), 20 mM potassium ferrocyanide, 2 mM potassium ferrocyanide and 2 mM MgCl .

(2) Results. The presence of a β-galactosidase enzyme activity was revealed by the formation of a blue coloration. Approximately 50-80% of RBE4 endothelial cells infected with the retrovirus stain blue in the histochemical situation, but the level of coloration varies from one cell to another. Control (uninfected) RBE4 cells were not stained under the same conditions.

(e) Properties of RBEZ cells in vitro. The RBEZ cells obtained were cultured on a collagen-coated support in an α-MEM/FlO medium supplemented with 10% FCS, 2 mM glutamine, 1 ng/ml bFGF, and 300 μg/ml G418. These cells display contact inhibition and growth factor- and adhesion factor-dependent proliferation; they express, in addition, endothelial differentiation markers.

EXAMPLE 3 PREPARATION OF RBE/NGF ENDOTHELIAL BRAIN CELLS.

The RBE4 cells obtained in EXAMPLE 1 were subjected to 2 passages/week on an α-MEM/FlO (1/1; Seromed, France) medium supplemented with 2 mM glutamine, 10% FCS, 1 ng/ml bFGF, and 300 :g/ml G418. The cells were plated out at a density of 10⁴ cells/cm² on collagen-coated dishes and used between passages 30 and 60.

(a) Preparation of the-retroviral vector. The procedure was as in EXAMPLE 2, using a retroviral vector pMoMuLVisisNGF which is deficient for replication and into which the sequence coding for mouse β-NGF is inserted (Scott et α/.,.302 Nature 538-540 (1983). This vector, introduced into P-2 producing cells, enabled P-2-MoNuLVisisNGF cell lines to be obtained.

(b) Infection ofRBE4 endothelial cells. The procedure was as in EXAMPLE 2.

(c) Selection of endothelial cells expressing the transgene by a 2-site ELISA method. Following subcloning of the infected RBE4 cells by the limiting dilution method, subclones secreting βNGF (RBE/NGF cells) were identified using a 2-site ELISA (Ladenheim et al, 60 J. Neurochem 260-266 (1993)). More specifically, an anti-βNGF monoclonal antibody designated 27/21 (0.1 mg/ml in 0.05 M carbonate buffer, pH 9.6) was applied to Costar EIA/RIA plates for 2 hours (hr) at 37°C. The plates were washed 3x with a mixture of 50 mM Tris-HCl, pH 7.4, 200 mM NaCl, 10 mM CaCl₂, 0.1% Triton X-100 and 0.05%) sodium azide, and incubated at 4°C overnight in a conditioned medium or βNGF standards (30-1000 pg/ml) in the same buffer supplemented with 1% bovine serum albumin. After washes, βNGF was detected using the same antibody conjugated to β-D-galactosidase (400 mU/ml) after incubation for 4 hr at 37°C. The chromogenic substrate was chlorophenol red β-galactopyranoside (1 mg/ml in a 100 mM HEPES, pH 7, 150 mM NaCl, 2 mM MgCl₂, 0.1% sodium azide medium). The absorbance at 570 nm was read after 2 hr at 37°C. Two highly positive subclones designated RBE/NGF-2 and RBE/NGF-4 were selected, as well as 2 less positive subclones, from 94 clones tested.

(d) Cellular detection of NGF synthesis by in situ hybridization of the nucleotide sequence (mRNA) coding for NGF. An in situ hybridization was carried out with a 48-mer antisense probe specific for the nucleotide sequence (mRNA) coding for βNGF, corresponding to positions 897-944 of the cDNA sequence of mouse βNGF (Scott et al, 302 Nature 538-540 (1983)), of the following formula: 48-mer mature NSF 5'-3' antisense sequence: 5'-CTGCTTCTCATCTGTTGTCAACGCCTTGACGAAGGTGTGAGTCGTGGT-3' (SEQ ID NO:l), so as to visualize the βNGF transcript in the infected cells in culture. These results showed a substantial expression of βNGF mRNA in the infected cells, at a level which was variable from one cell to another. FIG. 1 illustrates the in vitro analysis of the expression of the NGF transgene in RBE/NGF cells, by in situ hybridization. The immunoenzymatic visualization of the expression of the NGF transgene was carried out using a digoxigenin-labeled antisense oligonucleotide probe specific for murine NGF. Under these conditions, the mRNA/NGF probe hybrids were visualized with an anti-digoxigenin antibody coupled to alkaline phosphatase, the enzymatic reaction of which with the NBT-BCIP substrate complex produces a blackish precipitate. FIG. 1 A shows an intense signal in the RBE/NGF cells in culture, indicating a high level of expression of the NGF transgene. In FIG. IB, the absence of a positive reaction in uninfected control RBE4 cells was observed (x300 in FIG. 1A and FIG. IB; FIG. IB in phase contrast).

(e) In vitro activity of the secreted NGF. The biological activity of the NGF secreted into the supernatant of the RBE/NGF cells was demonstrated by the property of promoting a budding of axons from rat pheochromocytoma PC 12 cells. To carry out this test, the RBE/NGF cells were plated out on dishes at a density of 10⁴/cm² in dishes 100 mm in diameter, and growth was carried out for 3 to 4 days to confluence (10⁷ cells/dish). The medium was changed (10 ml) and the supernatants were collected after 24 hr. The results were illustrated in FIG. 2 and FIG. 3, which show that the 24 hr cell supernatant behaves in the same manner as purified NSF (0.1-50 ng/ml) used as internal standard, and leads to a stimulation of axonal budding in approximately 55% of the PC 12 cells. In addition, a 1:40 dilution of the supernatant displays a biological activity comparable to NSF at a concentration of 0.4 ng/ml (40% of cells bearing axons). Consequently, the capacity for secretion of biologically active NGF by the RBE/NGF cells may be estimated at 16 ng/10⁶ cells/24 hr.

FIG. 2 and FIG. 3 show, as abscissae, the NGF concentration (ng/ml) (FIG. 2) or the degree of dilution (FIG. 3; curve 1: RBE/NGF cells, and curve 2: RBE4 cells), and as ordinates, the percentage of cells bearing axons.

EXAMPLE 4 IMPLANTATION IN THE BRAIN OF RBE4, RBEZ, AND RBE-NGF CELLS (I) RBE4 cells: survival and integration.

For the characterization of the RBE4 cell lines transplanted into adult rat brain, a method of prelabeling with Hoechst bisbenzimide was used, in order to monitor the grafted cells (Gansmuller et al, 4 GLIA 580-590 (1991)). FIG. 4 illustrates the prelabeling of RBE4 cells before transplantation, with the nuclear stain Hoechst 33342 (bisbenzimide). The suspended cells were visualized in fluorescence microscopy under ultraviolet light. The fluorescence of the stain clearly defines the positively labeled cell nuclei (x270).

Three to 8 weeks after implantation of labeled RBE4 cells in different regions of the brain (gray matter and white matter), the graft had a compact appearance with a small and constant spread of some RBE4 cells around its mass. This migration took place essentially along the host's vascular network, suggesting a preferential interaction between the implanted endothelial cells and the host's vascular components.

Histological staining of the brains grafted in this way showed a minimum of signs of necrotic cells, this occurring essentially during the first week following the surgical trauma due to the transplantation. Within the graft, the cell density was homogeneous (little or no presence of pyknotic cells). The GFAP (glial fibrillary acidic protein) immunoreactivity characteristic of astrocytes was considerable from the first week after implantation, both around the graft and in the graft itself, indicating an infiltration of astrocytes into the latter. Unexpectedly, the implanted RBE4 cells migrate and integrate in the vascular environment, sometimes with a direct participation in the host's vascular network. FIG 5A-5C show the cells prelabeled with the Hoechst stain, after transplantation into adult rat brain. FIG. 5A shows a general view of the region of grafting in the brain parenchyma. The fluorescent grafted endothelial cells appear to accumulate preferentially around vascular components of the host's brain. The asterisks symbolize the course of a blood vessel (x250). FIG. 5B and FIG. 5C show a blood vessel at high magnification, located in the region of implantation of the graft. In FIG. 5B, numerous Hoechst-positive RBE4 cells integrated in a luminal (arrows) and perivascular position may be observed. In FIG. 5C, this same vessel was immunolabeled with an anti-laminin (specific marker of blood vessels) antibody (x600) (II) RBEZ cells: survival, integration and expression of the transgene.

(a) Prelabeling of cells with Hoechst bisbenzimide. See, this EXAMPLE, Section I.

(b) Preparation of cells before their implantation. Immediately before the grafting procedure, the cells were rinsed 3x in a grafting solution comprising PBS supplemented with MgCl₂ and CaCl₂ at a concentration of 1 μg/ml and glucose at a concentration of 0.1 %, so as to remove the DMEM-FCS medium.

(c) Surgery and implantation of cells. Adult male rats belonging to the Lewis strain and weighing 300 g received a graft of pre-labeled RBEZ cells, as specified in this EXAMPLE, Subsection (1), under deep anesthesia, under stereotactic conditions (Kopf® stereotactic frame, Paxinos & Watson, The Rat Brain in Stereotaxic Coordinates (Academic Press, New York, 1998)). Twenty animals received, respectively, a stereotactic implantation in the basal nucleus of RBEZ cells (right cerebral hemisphere) and of control RBE4 cells (left cerebral hemisphere).

A total of 300,000 cells suspended in a grafting solution (3 :1) was injected per site using an Exmire® 10 μl microsyringe having an external needle diameter of 0.5 mm.

(d) X-gal histochemical. visualization for light microscopy. The anaesthetized rats were sacrificed by perfusion with 150 ml of PBS and then with 300 ml of 4% PFA in a PBS solution (0.1 M, pH 7.4) at 4°C. To visualize the presence of β-galactosidase, the brains were cryoprotected and frozen by inclusion in a compound OCT® for cryostat sectioning. After sectioning, the enzyme activity of the nls-/αcZ transgene was detected by incubating the tissue at 37°C in PBS buffer containing 2 mM 5-bromo-4-chloro-3-indolyl-β-D-galactoside (X-gal, Sigma), potassium ferrocyanide (20 mM), potassium ferricyanide (20 mM) and MgCl₂ (2 mM). These reaction conditions did not give rise to staining in control animals and for the grafted unmodified RBE4 cells (cells grafted on the left-hand side). The cellular localization and the morphology in the tissue sections was increased if necessary using Nomarski optics, which permits a reliable identification of the cell type in the vascular structure. This visualization was combined, if necessary, with EBA (endothelial barrier antigen) immunohistochemistry for a cell characterization in greater depth.

FIG. 6 illustrates the analysis of the morphological and functional integration of RBEZ cells, by visualization or the expression of the nls-ZαcZ transgene and of the antigenic marker of integrity of the blood-brain barrier (BBB), EBA. FIG. 6A, 6B, 6C, and 6D show blood vessels located away from the region of grafting, onto which RBEZ cells have migrated after transplantation. In these FIGS., nuclei of endothelial cells, expressing the nls- αcZ transgene, in a luminal position (arrows) (FIG. 6A, FIG. 6C) were observed. These same vessels, in fluorescence illumination (FIG. 6B, FIG. 6D), express the EBA antigen (arrow heads), thereby indicating that the vascular insertion of the grafted cells has not impaired the integrity of the BBB (x750 in FIG. 6A and FIG. 6B; xl500 in FIG. 6C and FIG. 6D, FIG. 6A, and FIG. 6C, in transmitted interference contrast). Calculation of the percentage of grafted cells expressing the transgene was undertaken. Thus, at 1 week after grafting, 6.9 ± 0.6% of the grantee Hoechst-positive cells express β-galactosidase. Expression of the transgene was considerable up to 5 weeks after implantation, but decreases afterwards. The presence of implanted cells remains, however, detectable using the above-mentioned Hoechst stain. The absence of specific major changes in the host's immune reaction with respect to the presence of the lacZ transgene shows that these RBEZ cells were properly integrated. In addition, the RBE4 and RBEZ cells never develop tumors in vivo, since they display a great stability of their phenotype.

(5) X-gal histochemical visualization for electron microscopy. Animals (n=10) were treated for an ultrastructural analysis of the integration of the RBEZ cells by electron microscopy. In this case, the animals were perfused with PBS solution containing 2.5% of PFA and 0.5% of glutaraldehyde. The brains were removed and left in the same fixative overnight. After rinsing, they were cut with a vibratome into sections of thickness 75 p.m. For the visualization of cells expressing the lacZ gene, the substrate X-gal was used as for light microscopy, which, under the action of β-galactosidase, formed a precipitate which was dense to electrons and visible in the electron microscope. A pre-localization of the graft was performed on sections before treatment with 1% OsO . These thick sections were thereafter dehydrated and then included in Epon. The tissue blocks were thereafter sectioned with an ultramicrotome into semithin and ultrathin sections, which were counterstained or otherwise with uranyl acetate and lead citrate and then examined on a JEOL CXI 00 instrument. FIG. 7 illustrates the analysis of the morphological and functional integration of RBEZ cells by visualization of the expression of the nis-/αcZ transgene by electron microscopy. FIG. 7A shows, in Nomarski optics, the perinuclear β-galactosidase labeling in the grafted cells on a semithin section of brain (2 :m; xl660). On examination in electron microscopy, the cells were observed either in the parenchyma (FIG. 7B) or in a vascular position (FIG. 7C), forming blood vessels of the host. The arrow heads point to the perinuclear precipitates of X-gal, which were dense to electrons. Even integrated in the brain parenchyma, in the absence of direct contacts with the blood compartment, these endothelial brain cells were capable of surviving for long periods. The grafted cells appear to be metabolically active and capable of establishing specialized connections between themselves and with the cells of the host (presence of desmosomes and of tight junctions). In the vascular position, the RBEZ cells display a normal phenotype from the first week after grafting (tight junction and few pinocytotic vesicles; xl 60,000 in 7B and 7C; L: vascular lumen).

(Ill) RBEZ cells and brain tumors (a) Implantation. RBEZ cells at confluence were trypsinized and resuspended in

DMEM without serum immediately before they were implanted in the host animals. For an intracranial implantation, the RBEZ cells (5xl ⁵ cells) were injected stereotactically with a syringe (Hamilton, gauge 26, with a beveled end), into the caudate nucleus and the putamen of Fischer 344 rats (200 g to 250 g) after anesthesia. The cells were injected in a volume of 5 μl and the needle was left in place for 2 rain after injection in order to limit leakage.

In the context of a subcutaneous implantation, anaesthetized Fischer rats receive 100 μl containing 10⁶ RBEZ cells. To show that such cells implanted preferentially in hypervascularized regions such as tumors, a test was carried out performing the same implantations (intracranial and subcutaneous) with a mixture of F98, C6, or 9L glioma cells (10⁵ cells) and RBEZ cells, under the same conditions as above.

After implantation, the tissues were prepared so as to perform an immunohistochemical and histological analysis.

(b) Preparation of tissues. Rats were anaesthetized with ether and, after thoracotomy, the right atrium was incised and a cannula was inserted into the left ventricle, which was then perfused sequentially with a buffer containing 120 mM NaCl, 2.7 mM KCi in phosphate buffer pH 7.4 (1 ml/g of weight) and then 3.7% paraformaldehyde (fixative). The brains were placed in the same fixative for 30 min, cryoprotected with 30% sucrose in PBS and frozen. The tissues were cut (thickness 12 μm) and mounted on gelatin-coated slides.

(c) Demonstration of the survival of the cells and of the expression of the transgene. (1) Histochemical and immunohistochemical protocol for the detection of

RBEZ cells expressing the nls-lacZ reporter gene.

(A) Histochemical analysis. The slide preparations were rinsed 3x in PBS buffer and then incubated at 37°C for 1 to 2 hr in PBS containing 0.5 mg/ml of 5-bromo-4-chloro-3-indolyl-β-galactopyranoside (X-gal), 20 mM potassium ferricyanide, 20 mM potassium ferrocyanide, and 2 mM MgCl₂. Sections incubated in the absence of X-gal substrate were used as a negative control. The sections were then rinsed in PBS buffer and mounted in 90% glycerol in PBS containing 0.02% of sodium azide. The reaction conditions did not give rise to any staining in control animals which had not received any RBEZ cell.

(B) Immunohistochemical analysis. A few sections displayed a positive reaction for laminin in the tumor (detection of tumor microvessels), for the nuclear proliferation antigen Ki67 and for the expression of markers of the blood-brain barrier, such as the endothelial glucose transporter type 1 (GLUT-1), after staining with X-gal. For staining laminin, the sections were digested for 15 min at 37°C with 0.2% pepsin in 0.01 N HCl before incubation with the immunological reagents. The sections were incubated sequentially with 1% normal goat serum and then with either rabbit anti-laminin antibodies, rabbit anti-GLUT-1 antibodies or rabbit anti-Ki67 antibodies, and biotinylated goat anti-rabbit immunoglobulin was then added (1:200 in PBS). The sections were then incubated with an avidin-biotin-peroxidase complex (1:50 in PBS) followed by an incubation in 50 mM Tris buffer containing 0.5 mg/ml of 3,2'-diaminobenzidine (Sigma) and 0.01% of hydrogen peroxide. Control slides were incubated with normal rabbit serum in place of the immune serum. The sections were mounted in 90% glycerol in PBS. (2) Results.

(A) Visualization of implanted RBEZ cells. After staining with the chromogenic substrate X-gal, the blue product of the reaction with β-galactosidase was identified in histological sections of the intracranial and subcutaneous tumors after grafting of tumor and RBEZ cells (TABLES 1 and 2). No blue reaction product was detected in the tumors implanted without RBEZ cells. Surprisingly, the grafted endothelial cells were distributed throughout all the intracranial tumors including the marginal infiltrations, but do not appear to migrate into the normal tissues.

(B) Integration of implanted RBEZ cells. Essentially all the β-galactosidase-positive cells, irrespective of their localization, stain with the anti-laminin antibodies, this being in agreement with their endothelial phenotype. Interestingly, a few grafted RBEZ cells were associated with tumoral microvascular profiles. FIG. 8 illustrates the integration of RBEZ cells in the tumor tissue (C6 cells) and its vascular network (arrow head) on a section of brain tissue counterstained with neutral red and in Nomarski optics. This suggests that RBEZ cells implanted in this way had the capacity to integrate in an anatomically correct manner in the tumor vascularization.

(C) Functionalities of integrated RBEZ cells. One of the most important features of these endothelial brain cells is their high expression of the glucose transporter type 1 (GLUT-1), expressed at the blood-brain barrier. This is important for the trans-endothelial transport of D-glucose, the essential energy-releasing substrate of the brain. The endothelial cells in 9L or other intracranial gliomas also express this transporter. In contrast, the expression of GLUT-1 decreases rapidly in endothelial brain cells in culture.

Many β-gal-positive cells were labeled with the anti-GLUT-1 antibodies. These different tests show that endothelial brain cells genetically modified ex vivo survive, integrate in intracranial gliomas and express the transgene. (D) Proliferation of implanted RBEZ cells. The expression of the

Ki67 proliferation antigen by RBEZ cells implanted in intracranial 9L tumors was examined. Many cells expressing both nuclear β-galactosidase and Ki67 antigen were observed. The number of RBEZ cells implanted per tumor section was quantified by computer-assisted image analysis using the imaging device (MCflD) supplied by the company Imagine Research Inc. (Brock, St Catherine's University, Ontario, Canada), a Hamamatsu high-resolution CCD camera and a Compaq DeskPro 486/33 computer. The total number of RBEZ cells per tumor was estimated from the number of RBEZ cells per tumor volume (12 μm adjacent section), and the total tumor volume was estimated from the limits of the tumor according to two orthogonal planes of sectioning. TABLE 1 and TABLE 2 illustrate the results obtained on implantation of the modified endothelial brain cells according to the invention in 9L gliomas.

(E) Identification of the nls-lacZ gene in tumors implanted with RBEZ cells, by PCR. Oligonucleotides complementary to DNA sequences localized on the nls-ZαcZ gene (5'-CGACTCCTGGAGCCCGTCAGTATC-3') and on the vector, downstream of the 3'LTR sequence (5'-GACCACTGATATCCTGTCTTTAAC-3'), were used as primers. PCR was carried out on genomic DNA isolated from control tumors (9L tumors; FIG. 9, lanes 2 and 3), from test tumors (tumors which have integrated RBEZ cells, implantation 14 days before isolation of the DNA; FIG. 9, lanes 4-6) and from the RBEZ (FIG. 9, lane 7) and RBE4 (FIG. 9, lane 8) cell lines. Thirty-five amplification cycles with Taq polymerase were carried out under the following conditions: (1) denaturation at 95°C; (2) hybridization at 60°C; and (3) elongation at 72°C. FIG. 9 shows the results obtained. The PCR product (400 bp) was present only in the samples containing RBEZ cells. (IV) RBE/NGF cells.

(a) Prelabeling with Hoechst bisbenzimide. See, this EXAMPLE, Section I, above.

(b) Implantation. Immediately before the grafting procedure, the cells were rinsed 3x in a grafting solution comprising PBS supplemented with MgCl₂ and CaCl₂ at a concentration of 1 μg/ml and with glucose at a concentration of 0.1%, so as to remove the DMEM-FCS medium.

A total of 50 adult male rats, divided into 2 groups, belonging to the Lewis stain and weighing approximately 300 g, received a graft of RBE/NGF cells prelabeled with the Hoechst stain, under deep anesthesia, under stereotactic conditions. Ten animals received a stereotactic implantation of RBE/NGF cells in the right basal nucleus. Another group (n = 40) was subjected to a procedure of injections at multiple sites of RBE/NGF cells (right-hand side) so as to produce a cell column 2 mm in height between the basal nucleus and the dorsal striatum. A total of 300,000 cells suspended in a grafting solution (3 :1) was injected per site using an Exmire® 10 :1 microsyringe with an external needle diameter of 0.5 mm.

As a control procedure, unmodified RBE4 cells labeled with the Hoechst stain were also grafted contralaterally (left-hand side) at the same time and using the same stereotactic levels. Coronal and horizontal sections of the grafted brains were collected between 1 week and 12 months after transplantation. The grafts examined, visualized by the fluorescence obtained using the Hoechst stain, showed a compact appearance with little cellular spreading. No tumorigenic effect on the grafted RBE/NGF cells was observed.

(c) Preparation of tissues for immunohistochemistry. One week, 3 weeks, 5 weeks, 8 weeks and 1 year after implantation, the animals were anaesthetized and perfused endocardially with 0.1 M PBS solution, pH 7.4 at 4°C, followed by a perfusion of 4% paraformaldehyde in the same buffer. The brains were removed and stored in the same buffer overnight at 4°C. The brains were then stored in PBS buffer comprising 30% sucrose for 2 days at 4W and frozen in isopentane at -60°C. Coronal and horizontal sections (thickness 30 μm) were cut using a cryostat and collected in wells filled with PBS at 4°C. The sections were divided into different groups in order to carry out an immunohistochemical analysis as well as toluidine blue staining. For immunohistochemical analysis, the sections were initially treated with PBS containing 0.4% H₂0 for 30 min, and rinsed in the same buffer. They were then incubated in a 10%o normal serum of the same animal as the one used to produce the secondary antibodies and 0.1% Triton X-100 in PBS for 1 hr, and thereafter with one of the following primary antibodies:

(1) Polyclonal primary antibody. Rabbit anti-GLUT-1 (glucose transporter specific to the brain) antibody, (1:5000, Biogenesis); rabbit anti — GFAP antibody (1 :6000, Dako); goat anti-ChAT antibody (1:100, Chemicon), rabbit anti-laminin antibody (1 :5000, Sigma).

(2) Monoclonal primary antibody. Mouse anti-p75 LNGFR (low affinity NGF receptor) antibody (1:150, clone 192, Boehringer); mouse anti-CDllb (rat macrophages) antibody (1 :1000, clone MRC OX-42, Serotec); mouse anti-rat T lymphocyte antibody (1 :2000, clone MRC OX-52, Serotec); mouse anti-rat major complex I antibody (1:1000, clone MRC OX- 18, Serotec); mouse anti-rat MHC class II (Ig) antibody (1:1000, clone MRC 0X-6, Serotec); mouse anti-EBA (blood-brain barrier antigen specific to the rat) antibody (1:1000, clone 5M171, Affiniti).

All the antibodies were diluted in PBS buffer containing 5% of normal serum (donkey serum for the polyclonal antibodies and sheep serum for the monoclonal antibodies) and 0.1% Triton X-100, and incubated for 36 hr with stirring at 4°C. The sections were rinsed and incubated with biotinylated donkey anti-rabbit IgG (1:2000, Amersham) or anti-goat IgG (1:1000, Jackson Laboratories) antibodies or biotinylated sheep anti-mouse IgG antibodies (1:600, Amersham) in PBS buffer containing 5% of normal serum and 0.1% of Triton X-100, overnight with stirring at 4°C. They were rinsed, then incubated with a biotinylated-peroxidase complex (Vector Laboratories) for 30 min and rinsed again with Tris-buffered saline (0.1 M TBS, pH 7.6). The sections were then incubated in a solution of diaminobenzidine tetrahydrochloride with nickel chloride and hydrogen peroxide (H₂O₂) in Tris-buffered saline (0.05 M, pH 7.3). The enzymatic reaction was stopped by washing the sections in the buffer. The sections were then counterstained and dehydrated, mounted on slides and observed under the microscope. A control was invariably carried out, by omitting the primary antibodies, and, under these conditions, the sections were always unlabeled.

(d) Preparation of tissues and detection of the βNGF transgene by in situ hybridization. The cellular detection of βNGF transcripts in vivo was demonstrated by in situ hybridization of the NGF(mRNA) in the graft 1 week after implantation in the adult rat brain, as well as at 3 and 8 weeks, under the following conditions: After deep anesthesia with ketamine (150 mg/kg, Imalgene), the rats were perfused with 2% paraformaldehyde in 0.1 M PBS buffer (pH 7.4, 4°C). The brains were removed and placed in this buffer for 60 min at 4°C. After cryoprotection overnight in 15% sucrose solution in 0.1 M PBS at 4°C, rapid freezing of the samples was carried out by immersion in isopentane at -60°C. The frozen brains were cut horizontally (10-14 μm) using a Microm® cryostat, then mounted on gelatin-coated slides and dried at room temperature. The sections were prehybridized for 1 hr at 40°C in 4xSSC, Ix Denhardt's buffer. Hybridization was carried out in a humid chamber at 37°C for 16 hr, using as hybridization buffer a 4xSSC, 50%) formamide, 10% dextran sulfate, lx Denhardt's buffer, 500 μg/ml of fragmented and denatured salmon sperm DNA, and 100 μg/ml of yeast tRNA mixture containing the above-mentioned NGF probe at a final concentration of 2 μg/ml. The slides were washed sequentially in 2xSSC for 1 hr at 20°C, then in lxSSC for 1 hr at 20°C, then in lxSSC for 2 hr at 37°C and in 0.5xSSC for 2 hr at 20°C. The digoxigenin-labeled, hybridized probe was detected using an immunoenzymatic detection kit (Boehringer-Mannheim) according to the manufacturer's instructions. Control procedures were carried out in parallel, either by digestion of the mRNAs with RNase A (20 μg/ml for 30 min at 37°C) or by competition with an excess of unlabeled probe (excess of the order of 40) in the hybridization mixture. A dilution of the probe to 0.5 μg/ml gives a weak but specific signal. The absence of signal was observed when the NGF probe was not introduced during the hybridization. In this case, a substantial presence of NGF transcripts was detected in the RBE/NGF grafts, reflecting a constitutive expression controlled by the LTR of this trans gene.

FIG. 10 illustrates the in vivo analysis of the expression of the NGF transgene in RBE/NGF cells, three weeks after transplantation into the nucleus basalis (basal nucleus), and FIG. 11 illustrates the control brain structures used as internal control of the in situ hybridization of the NGF messenger, in vivo. FIG. 10A and FIG. 10B show a graft (G) of RBE/NGF cells strongly expressing the NGF transgene detected by in situ hybridization. This expression for the transgene still remains as strong 3 weeks after the intracerebral grafting. FIG. 10C visualizes a control graft (G) of uninfected RBE4 cells, grafted in the contralateral hemisphere, which does not display any positive NGF signal (xl30 in 10A); (x270 in 10B and 10C, with transmitted interference contrast). FIG. 11 A illustrates the neuronal detection of NGF in the frontoparietal cortex, and FIG. 1 IB illustrates the detection of NGF in the hippocampus (x260 in 11 A; x65 in FIG. 1 IB). These FIG. were in agreement with the established description of the endogenous synthesis of NGF by the neurons of the cerebral cortex and of the hippocampus in the adult rat.

(e) Biological effect of the NGF produced by the graft on the cholinergic neurons of the basal nucleus (axonal budding effect). To explore the biological effect of the NGF transgene product secreted in vivo, the following functional test was carried out. The modified cell line was grafted as specified above in the basal nucleus, in which the cholinergic neurons display a very sensitive response to NGF. These cholinergic neurons, apart from their expression of the enzyme ChAT, may also be characterized by a substantial immunoreactivity for the p75LNGFR receptor. The latter enables the cholinergic fibers and also their cell bodies to be visualized, especially during studies of axonal regeneration. The reactional budding detected by p75LNGFR immunoreactivity was observed up to at least 3 weeks in the RBE/NGF grafts.

In the first grafted group, the biological effect of the NGF produced by the grafts on the cholinergic neurons of the basal nucleus (axonal budding) was localized in this region and did not extend beyond the limits of the latter.

FIG. 12 illustrates the biological effect of the NGF secreted by RBE/NGF cells, 3 weeks after grafting, in the nucleus basalis (NB) (action on the promotion and maintenance of the reactive axonal regrowth of cholinergic neurons damaged after transplantation) . In FIG. 12 A, a general view of the shape of an RBE/NGF graft placed in the NB was visualized using the Hoechst pre-labeling. In 12B, 12D, and 12F, the effect of the NGF produced by the endothelial cells on axonal regrowth was visualized by the strong immunoreactivity of these axonal processes for the NGF p75 receptor. This axonal regrowth took place over the entire length of the graft (G) and displayed a strong reactivity around certain blood vessels (arrows). In FIG. 12C and FIG. 12E, 3 weeks after grafting, control RBE4 grafts not infected with the NGF retroviral construction, placed in the NB of the contralateral hemisphere, were incapable of promoting and maintaining a reactive axonal regrowth of the cholinergic neurons of the NB (x65 in A, B, C, D, E, and F, horizontal plane).

However, in the 2^nd group, the budding due to the NGF secreted by the grafted cells was more extensive in the ventrodorsal axis, along the graft, linking the basal nucleus to the dorsal striatum, thereby showing the trophic and tropic effects of NGF. FIG. 13 illustrates the biological effect of the NGF secreted by RBE/NGF cells, 3 weeks after grafting, away from the nucleus basalis, and illustrates the directional growth of the extensions in growth of the cholinergic neurons of the NB, along the graft up to the level of the dorsal striatum. FIG. 13A and FIG. 13B illustrate a horizontal section passing through the dorsal portion of the graft in the striatum. In FIG. 13 A, the RBE NGF cells were visualized with the Hoechst nuclear stain. In FIG. 13B, the same section has been examined in transmitted light, showing a reactive axonal regrowth visualized with the anti-p75 NGF receptor antibody (xlOO in FIG. 13A and FIG. 13B)

A quantification of the biological effect induced by the expression of the NGF transgene was undertaken according to the method described by Gundersen et al, 96 APMIS 379-394 (1988), by calculating the area occupied by the p75LNGFR immunolabeling at the sites of implantation of the RBE/NGF and RBE4 cells. The ratio of this area to that occupied by the graft was calculated at 3 and 8 weeks after implantation and is presented in FIG. 14, where the area occupied by the p75LNGFR-positive structures (expressed as a percentage relative to the area of the graft) is plotted as ordinates.

Among the different clones tested in vivo, only the 2 more productive of NGF in vitro (clones RBE/NGF-2 and RBE/NGF-4, mentioned above) gave rise to a biological effect in vivo, as summarized in TABLE 3 below:

(f) Immunological tolerance. In order to be able to study the host's immunological reaction with respect to the grafted endothelial cell lines, immunohistochemical labeling was carried out using markers of macrophages, of the major histocompatibility complex and of lymphocytes. At 1 week, an infiltration of macrophages was observed at the transplantation site, with a decrease in their presence over time. This infiltration was linked to the surgical trauma due to the transplantation. However, no infiltration by lymphocytes was observed, even 1 month after transplantation.

These data suggest that such grafts were well tolerated by the host, which does not develop an acute rejection reaction with respect to the different endothelial cell lines grafted. The above data show that both RBE4 cells alone and RBEZ cells and RBE/NGF cells survive and integrate after grafting. RBEZ and RBE/NGF cells were capable of expressing and/or secreting the product of the transgene which, in the case of NGF, has the capacity to induce a biological effect in the brain.

EXAMPLE 5 THE TK2 CELL LINE AND THE THYMΓDINE KINASE PROGRAM IN BRAIN

TUMORS (a) Rationale for HSV-TK in cancer gene therapy. The suicide gene strategy, using herpes simplex virus (HSV) enzyme thymidine kinase (TK) in association with the anti-herpes drug Ganciclovir represent a novel therapeutic tool for gene therapy of cancer. The guanine analog Ganciclovir is metabolized into a phosphorylated cytotoxic product by successive enzymatic actions of the transferred HSV-TK gene product and to a lesser extent by the endogenous mammalian cell thymidine kinase. The triphosphorylated product is incorporated into elongating DNA chains of actively proliferating cells (i.e. cancer cells), acting as a chain terminator and leading to cell death (Cheng et al, 258 J. Biol. Chem. 12460-12464 (1983)).

In the HSV-TK suicide gene strategy, the use of viral vectors carrying the enzyme and infecting or transducing the dividing cancer cells is the most developed approach (Culver et al, 256 Science 1550-1552 (1992); Ram et al, 53 Cancer Res. 83-88 (1993); Chen et al, 91 Proc. Natl. Acad. Sci. 3054-3057 (1994); Barba et al, 91 Proc. Natl. Acad. Sci. 4348-4352 (1994)). A significant advantage of this strategy is that not all cells of a tumor need to express thymidine kinase in order to achieve a complete regression of the tumoral mass. This phenomenon, also called the "bystander effect", confers Ganciclovir sensitivity to uninfected cells through a metabolic cooperation (transfer of phosphorylated Ganciclovir through intercellular gap junctions/connexins) between adjacent cells and also to the transfer of vesicular apoptotic bodies (Li Bi et al, 4 Human Gene Therapy 725-731 (1993); Freeman et al, 53 Cancer Res. 5274-5283 (1993)).

All the current approaches are based on the transfer of the thymidine kinase gene in the host cancer cells. Our approach is original, because of the non-transfer of the HSV-TK gene in the cancer cells and its cytoplasmic expression in the genetically brain endothelial cells (RBE4 cell line as an example). The obtained HSV-TK modified RBE4 clone is called the TK2 cell line. In this case, the observed therapeutic efficacy is solely based on the "bystander effect" of the TK2 cells present in the environment of the cancer cells when co-grafted with a tumoral cell line in animal models. Such an approach that avoids viral vectors for the gene transfer into a patient cancer cells has the advantage to be safer and easier to use in human clinical applications.

(b) TK2 cell line. The TK2 cell line was obtained by transfection of the rat brain endothelial RBE4 cells (passage 40) with the pUT-649 plasmid (purchased from Cayla, France) using lipofectin (Gibco). The pUT-649 Plasmid (4.6 kb) drives the expression of the HSV-TK::Sh ble fusion gene with the enhancer and promoter of the immediate early gene of human cytomegalovirus (hlE-CMV promoter). The fusion protein in addition to its thymidine kinase activity (HSV-TK gene) confers resistance to the zeocin and Phleomycin antibiotics (Sh ble gene). Among several isolated positive clones during selection, the TK2 clone was the more resistant to zeocin at a concentration of 100 μg/ml and was sensitive to Ganciclovir at concentration in vitro ranging from 0.2 μg/ml to 2 μg/ml (maximal effect). On Western blot detection the fusion protein was strongly expressed in the TK2 cell line.

(c) 9L gliosarcoma tumor model. For efficacy tests in adult Fisher 344 male rats, 10⁴ 9L tumor cells (Benda et al, 34 J. Neurosurg. 310-323 (1971); Weizsaecker et al, 224 J. Neurol. 183-192 (1981); Nam et al, 731 Brain Research 161-170 (1996)) were co-implanted with 3x10⁶ RBE4 cells (control group) or TK2 cells (treated group) by a stereotaxic injection procedure in the right striatum, according to the protocol described below. Survival curves were plotted and analysed with the Kaplan-Meier survival curve analytic method and log-rang statistical test. Ganciclovir (Cymevan, purchased from Roche, France) was diluted in saline and was injected i.p. at a dose of 100 mg/kg once a day from day 0 to day 6 following brain transplantation of the rats.

(d) Cell suspension preparation. For intracerebral transplantation of rats, 9L tumoral cells, RBE4 (control parental cells) and TK2 cells (treatment) were rinsed with sterile PBS 0.01 M without calcium/magnesium and incubated with trypsin. When the cells detached, culture medium was added and cells were counted for viability. Mixtures of 9L plus RBE4 and 9L plus TK2 (10⁴ 9L+ 2xl0⁶ or 3xl0⁶ RBE4 or TK2) were centrifuged at 1000 g. The pellet was rinsed and resuspended twice in PBS 0.1M , glucose lOmM. The final volume was adjusted to reach the maximal concentration of 3x10⁵ cells/μl. The cells were maintained on ice during the transplantation procedure. Cell viability was assessed by trypan blue exclusion and counting, before and after transplantation.

(e) Intracerebral implantation. Adult rats (8-9 weeks old male Fisher 334, purchased from Iffa Credo, L'Arbresle, France) were deeply anaesthetised with intra-peritoneal injections of acepromazine 0.5% (Vetranquil) at the dose of 4 mg/kg body weight, followed 10 min later by ketamin (Imalgene 500) at the dose of 80-90 mg/kg. They were placed in a rat stereotaxic frame (Stoelting or Kopf). After disinfecting with 70% ethanol, the skin was incised, and a burr hole was drilled through the cranium at predetermined stereotaxic co-ordinates using the Paxinos & Watson, The Rat Brain in Stereotaxic Co-ordinates (Academic Press, New York., 1998) rat brain stereotaxic atlas. In all cases, the injection needle placed in the right striatum. Cell suspensions were prepared at a density of 2x10⁵ to 3x10⁵ cells/μl and the injection was performed with an Exmire syringe at a controlled rate of 1 μl/min by a home-made electric injector. A total volume of 10 μl was injected per rat brain. After injection, the needle was left in place for an additional period of 10 min, then slowly withdrawn. The skin was sutured with sterile silk suture (Ethicon), the rat was identified by an ear code, and allowed to recover on a heating blanket before its transfer to the animal facility.

(f) Results.

(1) A group of rats (n=16) was intracerebrally implanted with 10⁴ 9L cells mixed with 2x10⁶ TK2 cells. Ganciclovir was i.p. injected daily at a dose of 50 mg/kg from day 0 to day 6 for the treated group (n=10). Control group (n=6) was given i.p. an equal volume of saline during the same period. Animals were observed daily and sacrificed when neurological signs and considerable weight loss were recorded. Kaplan-Meier survival curves were compared by log rank statistical test analysis. Results are presented in FIG. 15. In this first group a small significant increase of median survival was observed in favor of the treated rats.

(2) A second group of rats (n=l 0) was implanted with the same number of cells. Ganciclovir was injected daily at a 100 mg/kg dose for 5 treated animals. The control group (n=5) was injected with saline. Survival curves are presented in FIG. 16. A substantial increase in the median survival (33 days versus 18) was obtained in comparison with the first group of animals treated with 50 mg/kg of Ganciclovir (21 days versus 18.5).

(3) To ensure that the therapeutic gain is not only due to Ganciclovir and . to demonstrate that thymidine kinase works in vivo, 2 additional test series of 10 rats each were performed, mixing 10⁴ 9L cells with 3xl0⁶ TK2 cells (n=5) versus 10⁴ 9L + 3x10⁶ RBE4 cells (n=5, control group with parental cells). In these series, all the animals were daily i.p. injected with a dose of 100 mg/kg of Ganciclovir. Cumulative survival curves are presented in FIG. 17 for these 2 tests. Median survival are significantly increased, which is due to the thymidine kinase activity in TK2 cell line.

EXAMPLE 6 NTC-121: A IL-2 DELIVERY VEHICLE FOR USE IN THE TREATMENT OF RECURRENT HIGH GRADE GLIOMA

(I) Purpose of the EXAMPLE.

(a) Summary. This EXAMPLE describes a human clinical trial protocol for brain tumors using a gene therapy product, NTC-121. NTC-121 is a rat brain endothelial cell line, immortalized with the E1A gene and genetically modified to express the human interleukin-2 (IL-2) gene. Preclinical results have established the safety and the efficacy of this cell line in both rodents and non-human primates and in a brain-tumor model in the Fischer-344 rat. This EXAMPLE summarizes the characteristics of NTC-121, as a gene therapy for use in the treatment of recurrent high grade glioma (grade III/IN astrocytoma).

(b) Immunotherapy and interleukin-2. The immunostimulatory cytokine interleukin-2 has been used clinically in different types of human cancer inclunding melanoma and renal cancer. However, severe side effects have been observed when this cytokine was administered by the systemic route.

The mature human IL-2 peptide consists of 133 amino acids but is synthesized as a precursor containing 153 amino acids of which the 20-residue hydrophobic signal sequence is cleaved to produce the mature protein during secretion. Human IL-2 contains a single N-linked glycosylation site at position 3 and differences in glycosylation cause size and charge heterogeneity in both natural and cell line-derived/recombinant protein. Glycosylation is not necessary for biological activity, non-glycosylated recombinant protein produced in prokaryots is both stable and fully biologically active. The natural molecule contains three cysteine residues at positions 58, 105, and 125, of which 58 and 105 form a disulfide bridge essential for biological activity. It is a hydrophobic molecule which is stable to moderate heat and low pH. Crystallography shows IL-2 to consist of 6 -helical domains. IL-2 is mainly produced by T lymphocytes and both CD4+ and CD8+ cells can secrete the cytokine. T cells require stimulation with antigen via the T cell receptor if they are to secrete detectable IL-2. A second signal such as that mediated by EL-1 is also required for significant production. The time course of IL-2 production by T cell preparations varies, mostly reaching optimal at 40-48 hr. IL-2 exerts its effects through the EL-2 receptor, consisting of three chains (α, β, 0 which interact with each other and IL-2 to effectively signal IL-2-mediated events to the cell.

The most significant effects of EL-2 are exerted on leukocytes. The effects of the cytokine generally enhance and potentiate immune responses. EL-2 dramatically stimulates proliferation of activated T lymphocytes, promoting progression through the Gl phase of the cell cycle, resulting in growth of cells and increase in cell numbers. It can also cause proliferation of resting T cells, but as these cells do not express significant amounts of the IL-2R (receptor) alpha-chain, this requires a much higher dose of EL-2 than for activated cells, and is probably not significant under physiological conditions. EL-2 also stimulates cytolytic activity of subsets of T lymphocytes, for example EL-2 activates large granular lymphocytes and natural killer (NK) cells to become lymphokine-activated killer (LAK) cells that a have non-MHC-restricted tumoricidal activity for many cultured and fresh solid tumors.

EL-2 enhances T cell motility, and induces secretion of other cytokines such as gamma-interferon, EL-4 and TNF, thus acting as a T cell differentiation factor. EL-2 also stimulates the proliferation of activated B lymphocytes and promotes the induction of immunoglobulin secretion and J chain synthesis.

Although the central nervous system is an "immunologically priviledged" site, because of the presence of the blood-brain barrier, it is now well recognized that activated lymphocytes and monocytes can be recruited into the cerebral tissue following local elevation of cytokines or chemokines in pathophysiological situations. Immunotherapy with repeated intracerebral injections of EL-2, alone or in combination with EFNγ, in patients with a recurrent malignant glioma, although triggering transient peritumoral vascular leak, was found to be relatively well tolerated and potentially effective (Merchant et al, in Advances in Neuro-Oncology, Kornblith & Walker eds., pp. 469-485, (Futura Publishing Company, Armonk NY, USA, 1997).

The major impediments in treating brain tumors by systemic administration of cytokines or chemotherapeutic agents are: (1) toxicity and (2) the relative isolation of the target tissue because of the blood-tumor and blood-brain barriers. As a result, tumor-targeted gene therapy strategies are now under investigation. Indeed, gene therapy has the theoretical potential to deliver high local concentrations of a therapeutic agent while minimizing systemic side effects. The clinical interest of this concept relies on the direct or indirect antitumor activity of the transgene product, as well as on the performances of the vector. In the present gene therapy strategy of this EXAMPLE, patients undergoing surgical debulking of recurrent glioblastoma will be treated by administration of NTC-121. These integrate into the remaining tumor margins in response to tumor-derived angiogenic factors and to release human IL-2 locally in order to boost the patient's antitumor immune response.

(II) Physical, Chemical, And Pharmaceutical Properties And Formulation. (a) Introduction. NTC-121, is a rat brain endothelial cell line, derived from the parental RBE4 cell line, genetically modified to express the human EL-2 gene product. The parental RBE4 cell line is described above (Roux et al, 159 J. Cell. Physiol. 101-113 (1994)). NTC-121 cells were derived from the parental cell line RBE4 by transfection with the plasmid vector pBCMG-hygro-hEL-2 (Roux et al, 159 J. Cell. Physiol. 101-113 (1994)), an episomal expression vector containing the human IL-2 cDNA sequence under the transcriptional control of a cytomegalovirus (CMV) promoter including a rabbit β-globin intron, followed by a poly(A) sequence, and a hygromycin-resistant gene for selection. Individual clones were selected on the basis of their capacity of EL-2 secretion. Following subcloning (by limiting dilution) of the highest producers, NTC-121 clone was selected and further characterized. NTC-121 cells maintain the normal, non-transformed, endothelial phenotype of parental RBE4 cells (see, e.g. Durieu-Trautmann, et al. 269 J. Biol. Chem. 12536-12540 (1994); Federici et al, 64 J. Neurochem. 1008-1015 (1995); Bourdoulous et al, 25 Eur. J. Immunol. 1176-1183 (1995); Nobles et al, 115 Br. J. Pharmacol. 1245-1252 (1995); Abbott et al, in New concepts of Blood Brain Barrier, Greenwood et al, eds, (Plenum Press, New York, 1995); Rist et al, 768 Brain Res. 10-18 (1997)).

(b) Production of NTC-121. The NTC-121 cell line was derived from the parental rat brain endothelial cell line RBE4 by transfection with the human IL-2 cDNA (hIL-2) as described before. Rat brain endothelial cells grow in vitro only on collagen I-treated plastic vessels. Growth culture conditions (medium and growth factors) were set up when the RBE4 cell line was established (Durieu-Trautmann et al, 269 J. Biol. Chem. 12536-12540 (1994); Federici et al, 64 J. Neurochem. 1008-1015 (1995)). To introduce an expression vector encoding the hIL-2 protein into the RBE4 cell line, the calcium phosphate precipitation technique was used. The following plasmid was used in RBE4 cells: pPCHEL plasmid (pBCMG-hIL-2). This plasmid contains the hEL-2 cDNA sequence followed by the Hygromycin B resistance gene for selection. Cells which have stably integrated foreign DNA into their genome are selected in presence of Hygromycin B in the medium. In these conditions, 17 independant stable clones were isolated and analysed for hEL-2 production. Overall transfection efficiency was about 10^"5.

(b) cGMP Production of a NTC-121 Master Cell Bank (MCB). NTC-121 was manufactured in cGMP conditions by BioReliance (previously Magenta) which is located in Scotland on the Stirling University Innovation Park. The Master Cell Bank was prepared by serial subculture of NTC-121 cells for a total of three passages in collagen-coated flasks. At the final passage the cells were harvested by trypsinisation. The harvested cells were pelleted by centrifugation, resuspended and pooled in cryopreservation medium. The resuspended cells were dispensed in 1.0 ml aliquots into 200 consecutively numbered vials at a concentration of 1 x 10⁷ viable cells/ml. The filled vials were frozen using a controlled rate freezer to a final temperature of-130°C and were then immediately transferred to the vapour phase of a liquid nitrogen tank for storage.

(c) Clinical Lot Production. A clinical lot was prepared by serial subculture of NTC-121 cells, initially in collagen-coated flasks and then in collagen-coated roller bottles. At the final passage, the cells were harvested by trypsinisation. The harvested cells were pelleted by centrifugation, resuspended, and pooled in ciyopreservation medium (38.7 % (v/v) MEM alpha medium, 38.7% Nutrient Mixture Ham's F10, 15% (v/v) heat inactivated gamma irradiated FBS and 7.7 % (v/v) dimethylsulphoxide). The resuspended cells were dispensed into labeled vials in 1.0 ml aliquots at 1 x 10⁸ viable cells/ml. The filled vials were frozen using a controlled rate freezer to a final temperature of-130°C. The vials were then immediately transferred to the vapor phase of a liquid nitrogen tank for storage.

(d) Clinical Lot Production and Quality Controls. The 'bulk' material is tested for contamination with bacteria or fungi, mycoplasma, or adventitious viruses. An identity test has been carried out. For the final product testing, the material is tested for bacterial or fungal contamination. Also the endotoxin content and general safety of the material is assessed.

The cells were washed prior to administration and the likely impurities that may be carried over from the clinical lot material are derived from the cryopreservation medium (38.7 % (v/v) MEM alpha medium, 38.7% (v/v) Nutrient Mixture Ham ' s F 10, 15 % (v/v) heat inactivated gamma irradiated FBS, and 7.7 % (v/v) dimethylsulphoxide). A standard washing procedure is used for the removal of impurities as published by Ram et al, 3 Nature Med. 1354-1361 (1997)). This procedure is then validated.

(e) Shipping and storage conditions. After manufacture both the Master Cell Bank and the Clinical Lot material are stored at less than -150°C in the gaseous phase of liquid nitrogen in an automatic filling cryogenic storage vessel. Materials are shipped using a liquid nitrogen vapour shipper. The product must be stored in a liquid nitogen container at less than - 150°C and tested every six months for cellular viability. After thawing the material should be kept at 4°C and used within 6 hr. (f) Biological Characterization of the MCB. Ten vials of NTC-121 Master Cell

Bank were thawed, assessed for cell number and viability and cultured to provide samples for further characterisation. These samples were characterised by the tests shown in TABLE 4:

^r ΓABLE 4

BIOLOGICAL CHARACTERIZATION OF THE MASTER CELL BANK

Test Specification Result

Viability (Trypan Blue) > 70% 90%

Sterility Negative Negative

Mycoplasma, EP Negative Negative

Isoenzyme Report Result Rat origin

Extended S+L- Report Result Negative

Transmission Electron Microscopy Report Result C-type retrovirus particles

In Vitro Assay for Adventitious Viral Report Result Negative

Contaminants

In Vivo Assay for Adventitious Viral Report Result Negative

Contaminants

Extended XC Plaque Assay Report Result Negative

Mouse Antibody Production, MAP Report Result Negative

Test

Rat Antibody Production Test Report Result Negative

Extended Mink Cell Focus Assay in Report Result Negative

SC-1 Cells

High Sensitivity Reverse Report Result Positive

Transcriptase Assay

Co-cultivation of Test Cells with Report Result Negative

Human Rhabdomyosarcoma

Detector Cells

Co-Cultivation of Test Cells with Report Result Positive for reverse

MRC-5 Detector Cells transcriptase activity but no infectious virus detected

In addition, a final test was performed, using NTC-121 supernatant inoculated onto human MRC-5 cells and NRK (Normal Rat Kidney) cells, definitely demonstrated the absence of replicative infectious retroviral particles in the NTC-121 cells.

(g) . Conclusion. This Good Laboratory Practice (GLP) safety testing performed on samples of NTC-121 manufactured in cGMP conditions validates the viral and microbiological safety aspects of the NTC-121 MCB and demonstrate the absence of infectious retroviral particles in the NTC-121 cells (not infectious for both rodent and human cells). (III) Human Interleukin-2 Production In Vitro.

(a) To determine the amount of hEL-2 secreted by the isolated stable clones (17 different clones were obtained), an ELISA based on a quantitative sandwich enzyme immonoassay technique was used (Human IL-2 Quantikine Kit (R&D)). The hEL-2 amount was expressed in ng/million cells/24 hr, using the EL-2 standard curve included in the ELISA kit. A clone was selected for its high EL-2 production and subcloned by limit dilution (0.3 cell per well). Twenty clones grew from the limit dilutions, which roughly correlates statistically with the starting dilution. EL-2 production was assessed for each clone at 3 different passages (variations in EL-2 values are mainly due to errors at the level of cell counts):

The clone producing the highest IL-2 amount with the highest stability of secretion was selected, RCHEP107. The clone RCHIP 107 received the product code name "NTC-121".

To verify that the NTC-121 clone remains stable in culture in terms of EL-2 production, including after freezing/thawing process, NTC-121 was kept in culture for several passages in medium supplemented with the selective agents. At pi 8, p28, and p36, supernatant was conditioned and tested by ELISA according to our standard protocol. The negative control was RBE4 at passage 39:

TABLE 6 cell line passage concentration of IL-2 in standard deviation n=4 ng/million cells/24 hr

RBE4 p39 0 nd

NTC-121 p36 338 27

NTC-121 p28 326 27

NTC-121 pl8 412 5 (b) Conclusion. The NTC-121 clone was maintained in culture for over 3 months (up to passage 36) in presence of the selective agents. hEL-2 production demonstrated no significant variation over the culturing period.

(c) Biological activity of secreted human IL-2. A test was run to verify that human EL-2 produced by NTC-121 displays the same biological activity as the standard recombinant human EL-2. EL-2 is mainly a growth factor for T lymphocytes and is highly conserved between species.

We used the murine T cell line CTLL-2, which is the T cell line of reference used for quantification of the mitotic capacity of IL-2. This murine cell line proliferates in presence of murine, rat or human IL-2. Its proliferation rate is proportional to the amount of EL-2 in the medium until reaching a plateau of maximal proliferation. It is routinely maintained in culture with a suboptimal dose of recombinant hIL-2.

Serial dilutions of each sample were incubated with IL-2-starved CTLL-2 cells, in order to have at least two dilutions inducing a proliferation rate within the linear part of the reference curve. The proliferation rate was measured after 48 hr by adding the WSTl reagent

(see below) when 100% of control cells (no EL-2 added) were dead. EL-2 units are then calculated from the reference curve included in each assay, taking the dilution factor into account. Results are expressed in U/million cells/24 hr. The colorimetric assay used allows for proliferation and cell viability assays. This assay is based on hydrolysis of tetrazolium salts (WSTl) to formazan by the succinate-tetrazolium reductase system which belongs to the mitochondrial respiratory chain and is active only in viable cells. Increase in the number of viable, metabolically active cells results in increase in the overall activity of mitochondrial deshydrogenase in samples. This leads to increase in the amount of formazan dye formed, which directly correlates with the number of metabolically active cells in the culture.

Colorimetric analysis is performed with a Multiscan Labsystem spectrophotometer at

450 nm. Calculation of the data with the reference curve were performed through the Genesis program:

Based on the amount of EL-2 protein secreted by NTC-121 measured by ELISA, the biological activity of the protein is >10⁷ U/mg. This activity is identical to commercially available recombinant hEL-2 from different manufacturers, and confirm the quality of the product secreted by NTC-121. On the other hand, the hIL-2 Quantikine KIT is calibrated for conversion of sample values obtained by ELISA to relative equivalent NEBSC/WHO units, using the following formula: NIBSC (86/504) equivalent value (IU/ml) = 0.033 x quantikine IL-2 value in pg/ml.

Results from previous ELISA were converted and included in the table above as the fifth column. Values obtained from the CTLL-2 assay are very similar to those obtained by calculation from the ELISA ; difference is due to inherent variability of the CTLL-2 assay.

(d) Conclusion. These observations altogether confirm that hIL-2 produced by NTC-121 displays biological activity, in terms of T cell growth factor, similar to standard hIL-2. NTC-121 cells stably secrete high levels of hEL-2 (320-340 ng/10⁶ cells/24 hr at passage 28-36), which is biologically active (approximately 10,000-12,000 IU/10⁶ cells/24 hr) in a CTLL-2 lymphocyte proliferation assay.

(IV) Biological Properties In Vitro.

(a) Growth curve of NTC-121. A test was run to compare NTC-121 at early and late passages to RBE4 parental cell line in terms of proliferation rate. RBE4 (passage 36) and NTC-121 (passage 26 and passage 49) were seeded at 10⁴ cells/cm² in 8 plates (100 mm) and every day 2 plates were processed.

The curve in FIG. 18 illustrates cell number at the indicated days in culture (n=2 per point). In absence of selective agent, NTC-121 displays a proliferation rate identical to the parental RBE4 cell line. The doubling time is 20 hr and cells reach confluency between day 3 and day 4. Proliferation rate of NTC-121 is stable between passage 26 and passage 49. (b) Phenotypic characterization of NTC-121. To verify that NTC-121 maintain the same expression of endothelial markers as the RBE4 parental cell line, the following testes were performed.

(1) Analysis by cytofluorometiy . MHC class I molecules are constitutively expressed at variable levels at the cell surface on all nucleated cells in the organism, whereas MHC class II molecules are only expressed on cells of the hematopoietic cell lineage. Expression of MHC class II molecules can be induced (or enhanced for MHC class I molecules) in other cells through the action of Interferon gamma (IFNγ). The RBE4 cell line constituvely expresses MHC class I molecules, whereas MHC class II molecules can be induced by EFNγ. After IFNγ treatment, cells were resuspended and incubated with specific antibodies (Ab). Binding of the primary Ab was revealed with a secondary fluorochrome-conjugated Ab directed against the Fc region of the primary Ab. A cytofluorometer allows assessment of the proportion of positively stained cells.

The following TABLE 9 represents the percentage of positive cells and the mean fluorescence of this population in each condition:

As expected, EFNγ induced MHC class II expression (increase in the percentage of positive cells) and enhanced MHC class I expression (increase in mean fluorescence = M). This induction was specific since PECAM-l/CD31(a constitutive endothelial marker) expression was not affected by EFNγ treatment.

(2) Analysis by immunostaining of cells grown on glass slides. Endothelial cells can be characterized by the expression of specific proteins which have a particular localization in the cell due to their function. PECAM-1/CD31 is a transmembrane protein expressed in endothelial cells and platelets. VE-cadherin is a endothelial cell specific transmembrane protein expressed at adherens junctions. Beta catenin and ZO-1 are submembrane proteins expressed at adherens and tight junctions. Proteins involved in the formation of tight and adherens junctions are very specific to endothelial cells (and epithelial cells) and are all localized at cell/cell contacts. Cells were seeded on compartimentalized collagen-coated glass slides. At confluence, cells were permeabilized with detergent to allow immunodetection of intracellular proteins. Binding of the primary Ab was revealed with a secondary fluorochrome-conjugated Ab directed against the Fc region of the primary Ab. Expression and localization of the protein of interest was observed under a fluorescent microscope. The following TABLE 9 summarizes the results:

TABLE 9

ANALYSIS BY IMMUNOSTAINING OF CELLS GROWN ON GLASS SLIDES

Marker RBE4 p27 NTC-121 p21 NTC-121 p42

ZO-1 ++ ++ ++ correct cell/cell correct cell/cell correct cell/cell localization localization localization β-catenin +++ +++ +++ correct cell/cell correct cell/cell correct cell/cell localization localization localization

VE cadherin ++ ++ -H- correct cell/cell correct cell/cell correct cell/cell localization localization localization

PECAM-1 +++ +++ +++ correct cell/cell correct cell/cell correct cell/cell localization localization localization

No phenotypic difference could be observed between the RBE4 cell line and NTC-121 even at late passage in terms of expression and localization of the different proteins tested. Immunocytochemical and/or FACS analysis confirmed expression of endothelial markers (von Willebrandt factor, PECAM-1), a brain endothelial marker (Transferrin receptor), junctional complex proteins (adherens junction-associated catenins, tight junction-associated ZO-1). (3) Sensitivity of NTC-121 to G418 and Hygromycin B. This analysis was run to demonstrate the sensitivity / resistance of RBE4 and NTC121 cell lines to the agents used for selection in the culture medium. RBE4 and NTC-121 cell lines were grown in a 96-well plate with different concentrations of G418 or Hygromycin B, alone or in combination, with a fixed concentration of G418 (250 μg/ml) and increasing concentrations of Hygromycin B. At day 3 and day 4 after adding the selective agent, cell viability was quantified in each well with the WSTl assay.

The resulting curves (FIG. 19; FIG. 20) represent the amount of cells/well (mean of triplicates) in OD units (optical density units) in function of the selective agent concentration in μg/ml. Points on the y axis correspond to wells containing no selective agent. Background values from wells containing only medium and WSTl was subtracted. At day 4, OD= 3.000 corresponds to a confluent cell monolayer. The presence of G418 in the culture medium enhances the sensitivity to hygromycin

B. RBE4 is highly sensitive to hygromycin B even when used alone whereas NTC-121 is lOx to 20x less sensitive due to the expression of the hygromomycin B resistance gene. Combination of G418 (250 μg/ml) and Hygromycin B (225 μg/ml to 15 μg/ml) decreases the growth rate of NTC-121. (4) Cytotoxic activity of human and rabbit sera against NTC-121 cells.

This analysis was run to assess the sensitivity of NTC-121 to rabbit and human complement in an in vitro assay. Cellular surface glycoproteins are differently glycosylated across species. In humans and old world monkeys, glycoproteins carry -(l,2)-fucosyl determinants as well as other sugar moieties which constitute different alleles (A and B blood groups in humans). Instead, glycoproteins in all other mammalian species carry an -galactosyl determinant which is recognized by natural antibodies in humans and old world monkeys, mostly IgMs which can represent up to 4% of total IgMs, but also IgGs. In case of xenogeneic transplantation in humans and old world monkeys, glycoproteins on the surface of donor endothelial cells are the primary targets of these natural antibodies. Binding of these antibodies to endothelial cell surface triggers the complement activation cascade, ultimately leading to cell lysis and hyperacute rejection of the graft. FIG. 21 illustrate the % of lysis of NTC-121 and RBE4 cells in presence of the indicated serum dilutions. Lysis was confirmed by visual inspection of the cells. The following calculation formula was used: % of lysis = [[(NC OD - sample OD) / NC OD]] x 100. 100% of RBE4 and NTC-121 cells are lysed by human (up to ^XA dilution) and rabbit (up to 1/16 dilution) sera in 90 min. Lysis was inhibited by heat-inactivation of the sera. These results confirm the high sensitivity of NTC-121 cells (and parental RBE4 cell line) to the complement cascade in rabbits or humans.

(5) Soft Agar tumorigenicity assay. To demonstrate that NTC-121 is not tumorigenically transformed. Tumor cells are characterized by their capacity to grow in soft agar, forming cell clumps in suspension, without any need for adhesion to substrate. The cells were tested twice and were observed for 1 month in culture. This assay is performed in standard conditions according to a published procedure. (6) Observation under microscope. 9L gliosarcoma positive control proliferated rapidly, forming clumps in suspension, whereas the RBE4 and NTC-121 cell lines did not demonstrate any proliferation or clump formation in suspension in soft agar, until the end of the 1 month observation. The NTC-121 cell line, like the parental RBE4 cell line, is not transformed and that its proliferation capacity is strictly dependent upon adhesion to substrate. In vitro, NTC-121 cells display a normal cobblestone morphology at confluence, without foci formation, they do not form colonies when seeded in soft agar.

(c) Karyotype analysis. This test was run to verify that the NTC-121 cell line do not display any major chromosomal rearrangement. Karyological analysis indicated that NTC-121 cells maintain normal diploidy in 14 over the 14 NTC-121 cell profiles that have been analyzed by an independent laboratory.

(VI) Pharmacokinetics, Pharmacology, Biodistribution, Product Metabolism, And Toxicology. This Section of the EXAMPLE is a summary of preclinical pharmacokinetic, biodistribution, metabolism, and toxicity tests conducted with the NTC-121 cells (Gene Therapy Product, GTP).

(a) In Vivo Tests — Pharmacokinetics and Biodistribution of NTC-121 cells. The goal of this analysis was (1) to describe the general distribution of the NTC-121 cells after transplantation in the cerebral parenchyma and (2) to assess the likelihood that the transplanted cells disseminate to peripheral organs.

(1) Test design. 3 million NTC-121 cells were transplanted in the right striatum of Lewis rats (3 females and 3 males, syngeneic graft). Animals were sacrificed 1 day after transplantation (2 females and 1 male) to assess the risks of dispersion induced by the surgical procedure, and 21 days after transplantation (1 female and 2 males) to assess the dispersion due to the migration of the established graft and to its elimination. Organs listed below were dissected using sterile and disposable equipment (brain being the last sample to be taken) and frozen on dry ice and kept at -80°C until further processing. The list of organs and fluids processed is: Brain: ipsilateral hemisphere (right hemisphere and right olfactory bulb), contralateral hemisphere (left hemisphere and left olfactory bulb), cerebellum, brain stem and spinal cord. Eye, lung, heart, liver, spleen, kidney, gonad, blood, feces, urine. NTC-121 cells were detected using three sets of primers: (A) EN1U/EN1L located in the neomycin resistance gene of the pElA-Neo vector used to establish the RBE4 cell line; (B) PH1U/PH1L located in the hygromycin resistance gene of the pBCMG-hIL2 vector used to establish the NTC-121 cell line; and (C) PI1U/PI1L located in the second intron of the rabbit β-globin gene and in the human EL2 gene of the pBCMG-hEL2 vector used to establish the NTC-121 cell line. Also, BP1U/BP1L, BP2U/BP2L, BP3U/BP3L, and BP4U/BP4L located in bovine papillomavirus type 1 sequence of the pBCMG-hEL2 vector used to establish the NTC-121 cell line. Primers were purchased from Eurogenetec. Primer specificity was confirmed using a BLASTN program (Altschul et al, 25(17) Nucleic Acids Res. 3389-402 (1997)). PCR analysis was performed following procedures designed to prevent carry-over contaminations. PCR set up, DNA sample manipulation, and PCR amplification and product analysis were carried out in different laboratories. Pipettes and tube racks were decontaminated daily by UV radiation. The strategy adopted was based on an amplification of the DNA samples in duplicate. One sample in each duplicate was spiked with a fixed amount of NTC-121 genomic DNA near the limit of detection to provide an internal sample-specific amplification control. PCR programs were optimized (MgCl₂ and primer concentration, annealing temperature and cycle number) to produce a single amplification product detectable by gel electrophoresis. Every PCR set up included a negative control (no DNA reaction) to control for the presence of reagent contamination and a positive control containing only the amount of NTC-121 genomic DNA used for the spike.

(2) PCR detection limit. In order to determine the limit of detection, in a 1^st set of tests, a 10-fold serial dilution of NTC-121 genomic DNA, from 100 ng to 0. )1 ng, in the presence of 100 ng of genomic DNA prepared from 9L cells (competitor DNA), was amplified with the PH1U/PH1L (Hygromycin), ENIU/ENIL (Neomycin) and the 4 BPU/BPL (Bovine Papilloma Virus) primer pairs and with the PI1U/PI1L (hIL2) primer pair. Results showed that the neomycin and hEL2 specific primer pairs allowed to detect 1 ng of NTC-121 genomic DNA in presence of 100 ng of competitor DNA whereas the hygromycin specific primer pair allowed to detect 0.1 ng of the same DNA. Considering that the rat diploid genome is 6xl0⁹ base pairs, 1 ng of genomic DNA is the equivalent of about 150 cells. Knowing the pElA-Neo vector is 7,692 bp long, and that NTC-121 cells contain one integrated copy of pElA-Neo the relative amount of pElA-Neo sequence in 1 ng of NTC-121 genomic DNA is about 1 - 1.5 fg. Similar calculation shows that the relative amount of the pBCMG-hIL2 vector (15.5 kb) in 1 ng of NTC-121 is about 2.5 to 5 fg. None of the 4 primer pairs designed to amplify the BPV region of pBCMG-hEL2 did work. This result is in agreement with a Southern blot analysis of the NTC-121 cell line demonstrating a deletion of the BPV region of pBCMG-hIL2 during an insertion event.

In a 2^nd set of tests, 1 ng of NTC-121 DNA was amplified using the same primer pairs, in the presence of 100, 200, 300, 400, and 500 ng of 9L DNA. This test showed that the optimum conditions were obtained when 300 ng of competitor DNA was used.

(3) PCR biodistribution of the NTC-121 cells. According to the results obtained above, 300 ng of DNA of each samples were amplified in duplicate with hygromycin (PHIU/PHIL) and neomycin (ENIU/ENIL) specific primer pairs. One sample in each duplicate was spiked with 2 ng of NTC-121 genomic DNA. Samples where a hygromycin and a neomycin PCR product was observed, were then amplified using the h-EL2

(PI1U/PI1L) primer set. The results obtained are sumerized in TABLE 10 below:

(4) Conclusion. Using the optimum PCR conditions, the biodistribution analysis of the NTC-121 cells showed that:

(A) The NTC-121 cells transplanted in right striatum of Lewis rat were detectable in the injected hemisphere 1 day after transplantation; (B) Except for 1 animal with a positive signal in the controlateral hemisphere, and 1 animal with a detectable signal in the blood, the transplanted cells did not disseminate to other part of the central nervous system or to peripheral organs, including gonads; and

(C) The transplanted cells were not detectable, in all the samples tested, 21 days after transplantation.

(b) In Vivo Studies — Pharmacology and Toxicology of NTC-121 Cells. Transplantations have been performed using NTC-121 cells grown in culture for 3 to 10 passages after thawing, the 2 to 4 last passages without antibiotic selection, in order to mimic the clinical procedures to be followed. The quantity of human EL-2 secreted in culture by NTC-121 cells was controlled immediately after each graft.

(1) Route. The cellular suspension was injected in stereotaxic conditions, with a 10 μl sterile glass syringe and a needle. The injection coordinates were based on a stereotaxic atlas of rat brain (Paxinos & Watson, The Rat Brain in Stereotaxic Coordinates (Academic Press, New York., 1998)), corresponding to the middle of striatum, at distance of brain ventricles (lateral ventricles).

(2) Rationale for the choice of route of administration. The intra-parenchymal stereotaxic injection was selected because it mimic the route of administration in glioblastoma patients.

(3) Dose. On the basis of previous studies performed in rat (Quinonero et al, 4 Gene Ther. 111-119 ((1997)) the concentration of the injected cellular suspension was

3xl0⁵ cells/:l. This concentration was the highest achievable, allowing to inject a great number of cells in a small volume of suspension. According to the analysis, the injected volume into cerebral parenchyma was 10 μl. The cell suspension was injected at a rate of 1 μl/min, according to brain transplantation standards (Brundin & Strecker, 7 Methods in Neurosiences (1991). (4) Rationale for the dose selection. The injection volume was kept as low as possible in order to limit traumatic lesions. At the maximal density of 3x10⁵ cells/μl, the dose selected was 3 million cells (10 μl) for the biodistribution and therapeutic efficacy studies of the NTC-121 cells. The dose was chosen for maximal efficiency, according to previous observations.

(c) Expression Level and Stability of the Transgene: Pharmacology Data Obtained On Fisher 344 Rat (Allogeneic Transplantation). The objectives of the 2 following tests were to define the duration of expression of the gene of interest (human EL-2) in Fisher 344 rats, following co-transplantation with 9L gliosarcoma cells (Test 1 : ELISA technique) and to visualize in situ expression of the product of this gene (Test 2: immunohistochemistry).

(1) Test l.

(A) Protocol transplantation of 3xl0⁶ NTC-121 cells and 10⁴ 9L gliosarcoma cells in the right striatum of adult Fisher 344 rat (males and females). Blood samples were taken from the retro-orbitary vessels of isofluorane anaesthesized animals, before and after the injection of cells. Some positive control samples were obtained by intramuscular injection of the NTC-121 cells. At each time point, blood samples were collected from 2 males and 2 females (n=4). (Total: 18 Fisher rats, 9 males, 9 females, number R995 to R1000, R1027 to R1032 and R1057 to R1062). (B) Processing. Sera were collected 1 day before and every day after transplantation, from Dl to D3, as well as at days D 7, 14, 21 and 28. Sample sera were quickly frozen and processed all at the same time. The analysis of serum concentrations of human EL-2 were performed by ELISA (Quantikine, R&D, batch 9833243), according to the manufacturer's instructions. (C) Results. Detection of serum levels of human EL-2 following intramuscular injection of 3xl0⁶ NTC-121 cells. Serum levels of EL-2 were 129.5±28.9, 101.3±20.6, and 15.8±12.1 pg/ml, respectively 5 hr, 1 day (Dl), and 2 days (D2) after transplantation. Serum levels of EL-2 were below the limit of detection at day 5 (D5; limit of detection = 7 pg/ml). All values are mean ± standard deviation of duplicate determinations in 4 rats (2 males and 2 females). Detection of serum levels of human EL-2 following intracerebral co-injection of 3xl0⁶ NTC-121 cells and 10⁴ 9L gliosarcoma cells. Human IL-2 could not be detected in the serum of rats grafted intracerebrally with NTC-121 and 9L cells, at 5 and 15 hr, and 1, 2, 3, 7, 14, 21, and 28 days after transplantation (4 samples at each time point).

(D) Conclusions. All (but 1) sera collected from rats grafted intracerebrally with NTC-121 cells were detected negative for human EL-2, as opposed with sera collected from rats receiving an intramuscular injection of the same number of NTC-121 cells. This observation likely reflects the presence of the blood-brain barrier which isolates the CNS from the periphery and limits the exchanges of NTC-121 derived EL-2 between the two compartments.

(2) Test 2. (A) Protocol Transplantation of 3xl0⁶ NTC-121 cells and 10⁴ 9L gliosarcoma cells in the right striatum of adult Fisher 344 rats (males and females). Following post-graft periods of 1, 7, 14, and 21 days, animals were sacrificed and brains were processed for cryostat sections. Indirect immunostaining was performed on alternate sections with specific antibodies. At each time point, 2 males and 2 females were sacrificed (n=4). (Total: 20 Fisher rats, 10 males, 10 females, number R953 to R972).

(B) Processing. Histological and immunohistochemical staining of the cerebral parenchyma at the injection site. This site was the area where the material itself was delivered at the tip of the needle (at a deepness of 5 mm).

(C) Results. Following transplantation of NTC-121 cells in the striatum of adult rats, human IL-2 protein was detected by immunohistochemistry in brain sections of all 4 animals at 1 day post-graft (n=4), of 3 animals at 7 days post-graft, and could not be detected in any of the 4 animals at 14 and 21 days post-graft.

TABLE 11 POST-ENGRAFTMENT TIME, POSITIVE CASE/TREATED CASE

Day post-graft 1 day 2 days 7 days 14 days 21 days human EL-2 4/4 2/2 3/4 0/4 0/7 Detection of grafted cells 4/4 2/2 4/4 3/4 7/7

(D) Conclusion. Human EL-2 protein was detected at the transplantation site in grafted rats at 1, 2 and 7 days post-graft, then declined over time, between 7 days and 2 weeks. This decline of human EL-2 levels was not due to cell death, since grafted cells were still detected after 21 days post-graft, but rather reflected transgene inactivation.

(d) Immune Reaction Against NTC-121 Cells: Toxicology Data In New-Zealand Rabbits (Xenogeneic Brain Grafting). The clinical application of NTC-121 being a xenogenic transplantation (rat into human), the objective of this Section was to describe the immunogenicity (Test 1) and potential toxicity (Test 2) of NTC-121 cells in a xenogeneic (rat to rabbit) brain transplantation model. (1) Test 1.

(A) Protocol. Intramuscular injection into rabbits (male and female) of 3xl0⁷ NTC-121 cells. Sera were collected before and after injection, in order to monitor a potential antibody response against rat NTC-121 cells. (Total: 2 rabbitsj 1 female NZl and 1 male NZ2).

(B) Processing. Sera were collected 1 day before (pre-immune) and 10 days after (immune) intra-muscular injection. The reactivity of these sera was tested towards RBE4 and NTC-121 cell proteins and membranes coated on the bottom of 96-well plates, by ELISA technique.

(C) Results. The comparison, at the same dilution, between pre-immune and immune sera of rabbits NZl and NZ2 revealed that a modest immune response was elicited in NZl whereas essentially no response was detected in NZ2. The sensitivity of this qualitative immunoassay (minimal detectable dose) is obtained at the 1/2700 dilution of the immune serum.

(D) Conclusions. These rabbits produce less antibodies 10 days post-graph (dpg) than rabbits 3 and 4 grafted with the same quantity of RBE4 cells. This difference could be due to individuals or due to the possibility that EL-2 can interfere with the timing of establishment of the immune response. (2) Test 2:

(A) Protocol. Intracerebral stereotaxic transplantation of 15x10⁶ NTC-121 cells (prelabeled with Hoechst dye, 5 μg/ml) into the striatum of New-Zealand rabbits (4 males and 4 females). Following post-graft periods of 7 and 21 days, animals were sacrificed and brains were processed for cryostat sections. Indirect immunostaining was performed on alternate sections with specific antibodies. At each time point, 2 males and 2 females were sacrificed (n=4). Sera were collected 1 day before (pre-immune) and 6, 10 and 20 days after graft, in order to follow the potential antibody response against xenogenic cells. (Total: 8 rabbits, 4 males and 4 females, NZ9 to NZ12, NZ17 to NZ20). (B) Processing. Titration of antibodies against NTC-121 cells before and after intracerebral injection.

(C) Results. A solid phase assay of antibody binding was first performed on a plate coated with NTC-121 proteins (as described in the protocol) with pre-immune serum and sera collected at 6 days post-graft (dpg) from rabbits NZ9, 12, 17, 18 (the day before sacrifice). The comparison of the optical densities at all dilutions (ranging from 1/100 to 1/2700) between pre-immune serum and 6 dpg serum of each animal showed no difference, indicating that intracerebral injection of 15x10⁶ NTC-121 cells did not detectable amounts of circulating antibodies within 6 days. These results are similar to those obtain with RBE4 cells at the same time. The same assay was performed using the pre-immune serum, and the 10 dpg and 20 dpg serum of rabbits NZ10, NZl 1, NZl 9 and NZ20, which were sacrificed 21 days post-graft. The serum of rabbit NZ4 (10 days following the third intramuscular injection of RBE4 cells) was used as a positive control. Results indicated that intracerebral stereotaxic transplantation of 15xl0⁶ NTC-121 cells elicited a modest humoral immune response detectable 20 days post-transplantation (20 dpg).

TABLE 13

IMMUNE REACTION AGAINST NTC-121 CELL IN RABBITS; TEST 2

Rabbit Sacrifice (day post Sera 6 dpg Sera 10 dpg Sera 20 dpg graft)

9 7 - nc nc

12 7 - nc nc

17 7 - nc nc

18 7 - nc nc

10 21 nc - +/-

11 21 nc - +/-

19 21 nc - +/-

20 21 nc - +/-

Response: - negative; +/-: very low; nc: serum not collected.

(C) Conclusions. The rabbit humoral immune response elicited by intracerebral injection of 15x10⁶ cells is similar for RBE4 and NTC-121 cells in terms of timing and amplitude. Responses remain much lower than those following intramuscular injection. This low production of circulating antibodies could reflect the immuno-priviledged status of the brain due to the presence of the blood-brain barrier.

(e) Rabbit Response to NTC-121 Cells Was Analyzed Using Antibodies Against The Following Antigens:

(1) Antigens: (A) GFAP, an astrocytic cytoskeletal antigen overexpressed in case of reactive gliosis. (B) Major Histocompatibihty Complex type II (MHC II), an antigen expressed on activated antigen presenting cells during rejection processes and not found in normal brain. (C) CD5, a pan-lymphocytic marker (T cells and B cell subsets), of immune effector cell types normally not present in the brain.

(D) Serum albumin, a blood protein, detectable in the brain parenchyma only after extravasation when the blood-brain barrier is disrupted or leaky. (E) Human EL-2 to detect the presence of the NTC-121 cells. (2) Results. GFAP expression was robust on astrocytes located in the vicinity of the graft with associated morphological changes (thick processes and swollen cell bodies) characteristic of reactive gliosis. This response appeared less intense than in rabbits grafted with RBE4 cells at 7 dpg and 21 dpg. Response initiated at the graft margin and extended radially around the cannula tract. Extension was also noticed in the ipsilateral corpus callosum. No astrocytes were found within the graft core but arranged in circle around the graft. The severity of the gliosis was variable among animals.

MHC class II expression was observed in all animals studied, as an indicator of ongoing immune rejection. The distribution of the immunostaining clearly surrounded the graft core with some infiltration of the graft. At 7 dpg, positive cells exhibited essentially small cell bodies with a large number of cell processes and a dense arborization resembling reactive microglia. Another cell phenotype was also observed with a bigger cell body and very short processes (likely macrophages). In some cases, this reaction was very close to the graft site and wide spread for other animals. Round cells were also detected as perivascular cuffs and lining the pial membranes, where Hoechst-positive cells were detected. At 21 dpg, MHC class II staining of cells of microglial/macrophagic phenotype was still observable. Also cells with a crescent-like shape, lining the blood vessels in the vicinity of the graft site, were positively stained. MHC Class II positive cells were always found in meninges.

CD5 reactivity was found in all animals at 7 and 21 dpg on round cells infiltrated within the graft, sometimes in large number. These lymphocytes in the brain parenchyma strictly co-localized with the grafted cells and around blood vessels. CD5-positive cells were also seen below the pia mater, where Hoechst-labeled grafted cells were often detected. These immune effector cells were detected at least during 3 weeks post graft.

Immunohistochemistry against hEL-2 revealed a very faint staining on few cell clusters within the graft at 7 dpg, in all but 1 animals. At 21 dpg, all animals were negative.

Staining against serum albumin observed at low magnification, delineated the area of blood-brain barrier disruption. The leakage of serum albumin was variable, but more intense at 7 dpg than at 21 dpg. TABLE 14 RABBIT IMMUNE RESPONSE TO NTC-121 CELLS

Rabbi Sex Sacrifice Hoechst Graft necrosis GFAP MHC CD5 Serum Huma t (Day) positive auto- (gliosis) II albumin n cells fluorescence leakage IL-2

9 F 7 +/- ++ + + + ++ -

12 M 7 + ++ + ++ +/- 4+ +/-

17 F 7 + -H- + + + ++ +/-

18 M 7 +/- ++ + ++ ++ +++ +/-

10 M 21 - ++ + + + + -

11 F 21 +/- ++ +++ + + + -

19 F 21 - ++ + + +/- + -

20 M 21 +/- ++ + + +/- +/- -

-: neg ative; +/-: very low; +: low; ++: strong; +^■ ++: very strong.

(3) Histochemistry. Hematoxylin/Eosin and cresyl violet/luxol fast blue stainings demonstrated hypercellularity at the border of the graft/parenchyma and around blood vessels (astrocytes, microglia/macrophages, immune cells) and a cell depletion within the cannula tract often filled with fibrinous material. Area of cell depletion was sometimes observed around the graft core, corresponding to oedema. The observation of luxol fast blue staining showed a local degradation of the myelin sheaths without any extension of demyelination in white matter structures (i.e. the corpus callosum), despite the presence of hEL-2 for several days. (3) Conclusions. Rabbits intracerebrally injected with 15xl0⁶ NTC-121 cells displayed no behavioral or neurological impairments. Disruption of the blood-brain barrier by surgical injury exposed NTC-121 cells to circulating immune effectors. MHC class II molecules were induced in parallel with infiltration of immunocompetent effector cells within the graft core. These observations likely reflect immune rejection of NTC-121 cells. Despite hEL-2 production and immune cell infiltration, NTC-121 cells did not cause neuropathological effects after a high single dose injection in the brain parenchyma of adult rabbits.

(f) Phenotypic Data On the Gene Therapy Product: Pharmacology Data Obtained In the New-Zealand Rabbit (Xenogeneic Graft) To describe the phenotype of NTC-121 cells when grafted into the rabbit brain parenchyma (xenogenic transplantation) and to assess their migration capacity and distribution in the parenchyma and/or in the blood vessels.

( 1 ) Protocol. Intracerebral stereotaxic transplantation of 15x 10⁶ NTC- 121 cells pre-labeled with Hoechst dye (5 μg/ml) in the striatum of New-Zealand rabbits (4 males and 4 females). Following post-graft periods of 7 and 21 days, animals were sacrificed and brains were processed for cryostat sections. Indirect immunostaining was performed on alternate sections with specific antibodies. At each time point, 2 males and 2 females were sacrificed (n=4). (Total: 8 rabbits grafted, 4 males and 4 females, number NZ9 to NZ12 and NZ17 to NZ20).

(2) Processing. Histological and immunohistochemical staining of the cerebral parenchyma at the injection site, at a deepness of 7 mm. Titration of antibodies against NTC-121 cells before and after intracerebral injection.

(3) Results. Cortical surface often appeared yellow/green at the level of the cannula tract at both time points, due to autofiuorescence of numerous necrotic grafted cells which could be seen by microscopy under UV light. Among these autofluorescent cells located within the graft column, clusters of Hoechst-positive cells were detected, essentially when isolated from the parenchyma (encircled by necrotic cells) or when located at the bottom of the graft. These remnant surviving cells were more numerous at 7 dpg than at 21 dpg: at this latter time point, only a few grafted cells were detected, as expected in this xenogeneic model. Outside the graft core, Hoechst-positive cells were found lining the meninges and in the ventricles (Rabbit 13) where they seemed to survive.

No Hoechst-positive cells were found in the brain parenchyma at distance from the cannula tract, as observed previously in allogeneic or syngeneic rat model. The analysis of hEL-2 expression revealed that the few expressing cells at 7 dpg exhibited a round phenotype.

(4) Conclusion. The survival of NTC-121 cells in rabbit brain is limited by the xenogenic immune rejection. (g) The Non Tumorigenicity Of The Gene Therapy Product: Data Obtained In the Non-Immunocompetent Nude Mice. To demonstrate the non tumorigenicity of NTC-121 cells in a non-immunocompetent recipient.

(1) Protocol Subcutaneous injection of 3x10⁶ NTC-121 cells into the flank of athymic Nude mice and follow-up for the detection of potential tumor formation.

(Total: 10 Nude mice, males).

(2) Processing. 10 Nude mice (Ufa Credo, L'Arbresle, France), deeply anesthetised with isoflurane, have been subcutaneously injected into the right flank, using an exmire microsyringe with 3xl0⁶ NTC-121 parental clone (n=2), or with 3xl0⁶ NTC-121 cells (n=5), or with 3xl0⁶ 9L gliosarcoma cells (positive control, n=2), or with vehicle only (PBS buffer with calcium and magnesium, 10 mM glucose) (negative control, n=l), in a volume of 10 μl. The follow-up of the animals was performed twice a week by palpation. Animals were maintained in a sterile environment, including sterile diet and water.

(3) Results. The 2 positive controls (mice injected with 9L-gliosarcoma cells) were sacrificed 15 days post-graft for ethical reason, because of the size of the tumors:

0.74 g and 0.87 g. The 2 parental clone injected mice and the negative control mouse remained normal at the examination, without any sign of tumor formation and were sacrificed after 4 months. Out of 5 mice injected with NTC 121 cells, 1 died after 5 months of unrelated causes and 4 are still alive after 9 months without any sign of tumor formation. (4) Conclusion. This demonstrates that NTC-121 cells are not tumorigenic, like the parental RBE4 cell line, and also confirm the non-tumorigenic behavior of the NTC-121 as demonstrated in vitro using soft agar.

(h) Pharmacology Data On the Therapeutic Efficacy of the Gene Therapy Product: In Vivo Survival and Imaging Studies In a Rat Brain Tumor Model (9L Model). To assess, in a model system (9L gliosarcoma in male Fisher rats), the therapeutic efficacy of the gene therapy product.

(1) Protocol. Transplantation of 3xl0⁶ human EL-2 -secreting cells (treated) or RBE4 cells (control) together with 10⁴ 9L gliosarcoma cells in the right striatum of adult male Fisher 344 rats (allogeneic graft). Animals were weighted daily and inspected to detect any clinical signs. Survival of each animal was recorded (Total: 80 male rats, 7 months, once a month). Several animals were processed for magnetic resonance imaging (MRI). (2) Processing. Animals were observed daily after tumor graft. When animals developed symptoms of ataxia, severe paresis, seizures, peri-ophtalmic encrustations, posturing, and/or 20% weight loss, they were euthanized. Day of euthanasia was considered as the eve of the day of death in plotting the survival curves. (3) Results. Several human IL-2-secreting cell lines were tested for their efficacy in extending the survival of rats intracerebrally grafted with 9L gliosarcoma cells.

One of the cell lines demonstrated the best therapeutic efficiency. The medians of survival after cumulative survival analysis of 5 different tests were 30.0 days for treated rats versus

18.0 days for controls (PO.0001). This cell line was subcloned by limited dilution in order to obtain a pure clonal population. Among the subclones, the clone RCHIP-107 exhibited the best therapeutic efficiency. At that point, the conditions of use of this clone were highly standardized and this clone was identified as the clinical candidate and was renamed "NTC-121". Transplantations have been performed using NTC-121 cells grown in culture for 3 to 10 passages after thawing, the 2 to 4 last passages without antibiotic selection, in order to mimic the clinical procedures to be followed. The quantity of human EL-2 secreted in culture by NTC-121 cells was controlled immediately after each graft. The average production of human IL-2 was

272±93 ng /million of cells/day.

Animals co-implanted intracranially with 9L gliosarcoma cells and NTC-121 cells survived significantly longer than control animals co-implanted with 9L cells and RBE4 cells.

The medians of survival after cumulative survival analysis of eight different tests were 39.0 for treated rats versus 18.0 days for controls (PO.0001).

Ten out of 39 treated rats are still alive (25%), with 1 year survival for the oldest rat and 5 months for the youngest. Twenty-one animals were processed for brain MRI. T2 -weighted MR images were performed in the coronal plane with slices of 1 mm width without enhancing contrast agent.

The 9L gliosarcoma tumors were characterized by hyper-signals. Three representative cases are presented below:

(A) MRI of an intracerebral tumor at 17 days after co-implantation of 9L and RBE4 cells (control animal, number R828). A hyper-signal in the right hemisphere was associated with hemispheric swelling, mass effect and collapse of the right ventricle.

Arrows point to the tumor and oedema. This animal died 1 day later (day 18). (B) MRI of an intracerebral tumor at 40 days after co-implantation of 9L and NTC-121 cells (treated animal, number R814). At this time point, all control animals were dead. An abnormal signal was observed in the right hemisphere, made of an hyper-signal and an hypo-signal. The hyper-signal (arrows) likely reflects growing tumor and oedema, while hypo-signal (large arrow) likely reflects haemosiderin deposit resulting from haemorrhage. Tumor and oedema were localized along the needle tract, between the entry point and the site of injection. Note also a slight dilatation of the right lateral ventricle, but no hemispheric swelling or mass effect. This animal died 17 days later (day 57).

(C) MRI of an intracerebral tumor at 120 days after co-implantation of 9L and NTC-121 cells (treated animal, number R746). A hypo-signal was observed along the needle tract (arrow). No hyper-signal was detected. Note dilatation of the right lateral ventricle. This animal was still alive 300 days after transplantation.

(4) Conclusion. Survival of rats bearing intracranial 9L gliosarcoma was greatly enhanced by co-implantation with human EL-2 secreting-NTC-121 cells. This cell line was obtained after screening and subcloning of several cell lines secreting various amounts of human IL-2. Long term survival was achieved in 25% of cases. There is a good correlation between long term survival and MRI data.

(i) Limitation of Tumor Progression In Fisher 344 Rats Co-Implanted Intracranially With 9L Gliosarcoma Cells and NTC-121 Cells to analyze the effect of the NTC-121 cells on intracerebral tumor progression.

(1) Protocol. Transplantation of 3xl0⁶ NTC-121 cells (treated) or RBE4 cells (control) together with 10⁴ 9L gliosarcoma cells in the right striatum of adult male Fisher 344 rats (allogeneic graft). At 13 days post-transplantation, MRI was performed on 10 animals. At 18 days post-transplantation, these 10 rats were sacrificed and brains were submitted to histological examination (Total: 10 male Fisher 344 rats, number R865 to R874).

(2) Processing. Tumor area was quantified by computer-assisted image analysis (BIOCOM 2000). Tumor volume was estimated by the formula: volume=(square root of maximum tumor cross-sectional area)³. (3) Results. Mean tumor volume (mean ±± SD) was 0.74 ± 1.3 mm3 in the treated group (n=5), versus 65.14 ± 45.56 mm3 in the control group. Intracranial 9L tumors established in the presence of NTC-121 cells (n=5) were significantly smaller than 9L tumors established in the presence of control RBE4 cells (n=5) (P<0.0001; Student t-test). All ten animals were processed for brain MRI (5 controls, 5 treated). MR T2 -weighted images were recorded in the coronal plane without enhancing contrast agent. Two representative cases are presented:

(A) MRI of intracerebral tumor at 13 days after co-implantation of 9L and RBE4 cells (control animal, number R872). A hyper-signal (arrows) extended over the entire right striatum, without hemispheric swelling. Five days later, the rat was sacrificed and the tumor volume calculated. On histological brain section the tumor volume was 54.4 ± 3.9 mm³, in agreement with the hyper-signal observed on MRI.

(B) MRI of intracerebral tumor at 13 days after co-implantation of 9L and NTC-121 cells (treated animal, number R867). A hypo-signal was observed along the needle tract. Note a very small hyper-signal (arrow), sign of a small tumor mass or of oedema. Five days later, no tumor mass could be detected on histological brain sections of this rat, in good agreement with the MRI data.

(4) Conclusion. Co-implantation of NTC-121 cells limited the progression of gliosarcoma 9L tumor to a very large extent. There is a good correlation between histological analysis of brain sections and MRI data.

(j) In Vivo Demonstration of the Stability of the Gene Therapy Product After Thawing: Efficacy Data Obtained In the Fisher 344 Rat/9L Brain Tumor Model. To define, in the animal model (gliosarcoma 9L and male Fisher rats), the therapeutic efficacy of the gene therapy product obtained directly from thawing preclinical vials to mimic the clinical procedures.

( 1 ) Protocol. Transplantation of 3x 10⁶ NTC- 121 cells (treated), directly after thawing, together with 10⁴ 9L gliosarcoma cells in the right striatum of adult male Fisher 344 rats (allogeneic graft). Cells conditioned with 2 different freezing media were tested: (1) 20%) foetal bovine serum (FBS) or (2) without serum (HBSS buffer). Animals were weighed daily and inspected to detect any clinical signs. Survival of each animal was recorded. (Total: 10 male Fisher rats, number R895 to R904). (2) Processing. Animals were observed daily after tumor graft. When animals developed symptoms of ataxia, severe paresis, seizures, peri-ophtalmic encrustations, posturing, and/or 20% weight loss, they were euthanized . The day of euthanasia was considered as the eve of the day of death in plotting survival curves.

(3) Results. The medians of survival after analysis were 30.0 and 40.0 days for group 1 (20% FBS) and group 2 (HBSS buffer) respectively, versus 18.0 days for controls (P<0.0001). The difference between the two treated groups was not statistically significant.

(4) Conclusion. NTC-121 cells obtained directly after thawing retained their therapeutic efficiency. This efficiency was similar to the results obtained with NTC-121 cells collected from cell culture flasks.

(k) Preclinical Toxicology In Two Species, Rodents And Non-Human Primates (GLP Test). The objective of the test was to determine the toxicity of the test article "NTC-121 " following a single intra-cerebral administration to the Cynomolgus monkey followed by a 4 week observation period. This GLP test was adapted from OECD 409 and directive 91/507 /EEC and ICH guideline S6 (July 16, 1997), (CPMP/SWP/112/98 draft) and Guidance for human somatic cell therapy and gene therapy from CBER/US FDA (1998) (draft version).

(1) Design of the monkey test. The test was conducted according to the following design:

Group 1 animals (control) received the graft media as a control article (PBS-glucose). (2) Test article preparation. Preparation: the test article was diluted in the graft media (PBS-glucose) to provide a final concentration of 300 000 cells/μl. The preparation was performed by the cell biology laboratory. Storage: refrigerated (about +4°C). Frequency of preparation: once only for each animal (on the day of treatment). Stability of the diluted test article: 6 hr. The method and procedures are kept in the raw data of the test. According to the theoretical values given by BioReliance (the manufacturer of the cells), the number of live cells injected was as follows:

(3) Rationale for the dose selection. A low estimation of the human injected dose will be approximately 2,000,000 cells/kg (based on published studies). The maximum volume which can be administered in a healthy Cynomolgus monkey by the intracranial route is 100 μl/animal. This volume represents approximately 35,000,000 cells, i.e. approximately 14,000,000 cells/kg for a monkey of 2.5 kg. This dose level is approximately 5x the intended starting human therapeutic dose per body weight.

(4) Preclinical toxicology GLP test of NTC-121 in non-human primates. A single dose toxicity test by the intra-cranial route (intra-cerebral administration) in order to mimic the clinical conditions was performed in the monkey. The objective of this analysis was to determine the toxicity of the test article NTC-121 following a single intracerebral administration to the Cynomolgus monkey followed by a 28 days observation period. The maximum volume which can be administered in a healthy Cynomolgus monkey by the intracranial route is 100 μl/animal at a dose level of 300 000 cells per μl. This volume represents approximately 35xl0⁶ cells, i.e. approximately 14xl0⁶ cells/kg for a monkey of 2.5 kg. This dose level is approximately 5x the intended starting human therapeutic dose. A total of 10 animals were used in this analysis, 6 treated (3 males and 3 females) and 4 controls (2 males and 2 females). The NTC-121 test article was administered as a single dose with a dose volume of lOOμl/animal. Control animals received the graft media as control article (PBS-glucose) under a volume of 100 μl/animal.

Morbidity/mortality checks were performed at least twice daily. Clinical examinations were performed daily. Clinical signs indicative of potential neurological effects were observed once a week. A full clinical examination was performed before the initiation of treatment and at termination. Body weight was recorded weekly for each animal. Food consumption was measured daily for each animal. Rectal temperature was measured daily for each animal from day 2, 3 or 4 to day 8. Clinical laboratory determinations were performed pretest, on day 10 and at the end of the analysis. Blood sampling for toxicokinetic evaluations of serum human EL-2 was performed pretest, 10, 24, and 48 hr after treatment then once on days 8 and 28. All animals were killed on day 28. Selected organs were weighed. Tissue samples were fixed and preserved at necropsy for all animals. In addition, selected samples were taken and frozen for possible PCR analysis. Selected tissues from all animals were examined histopathologically. (5) Results. (A) No mortality occurred during the test.

(B) Only a few clinical signs were observed during the test (in both groups), such as lacrymation (possibly due to stereotaxic frame), liquid or soft faeces (considered to be incidental), restless. None of these signs were considered to be related to the test article administration. In addition, a slight higher mean rectal temperature in treated animals (probably due to the expression of EL2) was observed compared to controls. This could be either due to a transient systemic effect of JL2 in post surgery or to a central effect of locally secreted EL2.

(C) No treatment-related variation was observed in body weight or food consumption. (D) In both sexes, all observed variation in hematological parameters (essentially decreases in hemoglobin, red blood cells and white blood cells) were of the same magnitude in control and treated animals.

(E) The variations seen in clinical chemistry parameters were not considered to be related to the test article administration, since they were observed in both groups.

(F) No significant increase in serum human IL-2 level was observed after the treatment.

(G) There were no variations in the organ weights which were considered to be due to the administration of the test article. (H) No systemic changes were seen at necropsy.

(I) The slight inflammatory response observed at the surface of the brain in most animals was considered to be due to the surgery. (J) An inflammatory reaction with neovascularisation was present at the injection site in two treated animals, and along the needle tract in 1 of these 2. This could indicate a repair process or some action of the test article.

(6) Conclusion. Based on the results of this test, the intracerebral administration of NTC-121 to a dose level of 30xl0⁶ live cells to the Cynomolgus monkey followed by 28 days for observation, was not associated with death or general clinical signs, except a very slight increase of the rectal temperature in the treated animals. Treatment was associated with an inflammatory reaction with neovascularisation at the injection site in 2 treated animals, and along the needle track in 1 of these 2. (1) Preclinical Toxicology GLP Test of NTC-121 in Rats.

(1) A single dose toxicity test (limit test/maximal exposure to the product) by the intravenous route was performed in the rat. The objective was to determine the toxicity of the test article NTC-121 following a single intravenous bolus administration to the Fischer 344 rat. In the main test, a dose level of 60x the intended starting human therapeutic dose was used. A total number of 50xl0⁶ NTC-121 cells under a volume of 2.0 ml were injected in 5 males and 5 females. The control article (PBS-glucose) was injected in 5 males and 5 females under a volume of 2.0 ml. Morbidity/mortality checks and clinical examinations were performed 15, 30, and 45 min after administration of the test article, then at 1, 2, 3, 4, and 5 hr and daily for the 14 day observation period. Individual body weights were recorded during the main test, immediately before treatment (dayl), on days 2, 4, 7, 11, and 15. All animals were killed after the end of the 14 day test period, after the final observation (day 15). A sample of spleen from 1 treated male with a macroscopic abnormality and a corresponding sample from an additional male were fixed and preserved at necropsy and examined histopathologically. (2) Results:

(A) No mortality occured during this test.

(B) No clinical signs were observed.

(C) The slight difference in body weight between control and treated animals in both sexes was of no biological significance. (D) The only macroscopic abnormality was an hypertrophic spleen in 1 treated male with moderate extramedullary haematopoiesis and moderate focal capsular fibrosis at histopathological examination. These abnormalities were not related to administration of the test article.

(3) Conclusion. Based on the results of this test, the administration of the test article NTC-121 up to the dose level of 50x10⁶ cells per rat by the intravenous route, was not associated with any toxicological effect. The dose level of 50x10° cells per rat could therefore be considered as a NOAEL (No Observable Adverse Effect Level).

(VH) Safety Of The Intracranial Injection Of NTC-121.

(a) Safety of the cell line. NTC-121 cells (like parental RBE4 cells) did not form tumors when injected intracerebrally or subcutaneously in athymic nude mice. The same observation was made when the cells were transplanted in allogeneic (Fischer rats) hosts. Immune rejection in xenogeneic animals might be associated with deleterious inflammatory reaction. Observations in rabbit (day 7 and 21 post-implantation) and cynomolgus monkey brains (day 28; see section 6) revealed that only very few remaining cells were detectable 21-28 days after transplantation and that an inflammatory reaction, associated with limited necrosis, was present at the injection site. No evidence of toxicity was reported in cynomolgus monkeys after intra-cerebral injection of an average of 30 million NTC-121 cells.

In Phase I/II trials in humans, a survival time of about 2 weeks of mouse fibroblasts (viral vector-producing cells) transplanted in glioblastoma patients brains was reported by Ramebal et al, (1997). No toxicity was observed as a consequence of immune rejection of xenogeneic cells implanted in the human brain.

(b) Safety of IL-2. Clinical trials show that systemic IL-2 therapy can achieve significant regression in 20 to 25% of patients with advanced cancer, mostly renal cell carcinoma and melanoma. However, adverse effects have been described. The vascular leak syndrome is the most frequent and severe complication of systemic EL-2 therapy. Also, flu-like syndrome are frequent but usually mild with generalized malaise, fever and chills. A variety of skin complications have been described including life threatening skin reactions. Hypothyroidism is the major endocrine complication. Differences in toxicity seems to be related to EL-2 dose and/or schedule of administration. However, in human trials with local injection of EL-2 in the brain, no toxicity was observed as extensively discussed in NTC-121 clinical protocol.

(c) Arising from surgery. The surgical procedures carry a risk for loss of neurological functions, non-neurologic complications and death. The risk depends on the preoperative condition of the patient, size and location of the tumor, and associated disease. The risk for an individual patient can be determined prior to a surgical decision and discussed with the patient .

EXAMPLE 7

JL-10 GENE TRANSFER TO INTRACRANIAL 9L GLIOMA: TUMOR INHIBITION AND COOPERATION WITH IL-2

Summary. This EXAMPLE examines the effects of interleukin- 10 (IL-10) and combination IL-10 + EL-2 gene transfer on test brain tumor growth in vivo. 9L gliosarcoma cells were engineered to stably express murine IL-10 (9L-EL-10 cells) and implanted subcutaneously or to the caudate/putamen of syngeneic rats. The growth of tumors expressing EL-10 was substantially reduced compared to that of control tumors (p <0.05). Intracranial tumors expressing EL-10 and EL-2 were established by co-implanting 9L-EL-10 cells with endothelial cells engineered to express EL-2. At 14 days post-implantation, tumors expressing EL-10 + EL-2 were 99% smaller than control-transfected tumors (p O.0001). This extent of anti-tumor effect could not be achieved by expression of EL-10 or EL-2 alone within tumors. Neither IL-10 nor a combination of JL-10 + EL-2 gene delivery inhibited tumor growth in severe combined imnmunodeficient (SCID-Beige) mice (p > 0.05). Immunohistochemical analysis revealed that EL-10 + EL-2 gene delivery markedly increased T-cell infiltration within the striatum ipsilateral to tumor cell implantation. These findings establish that IL-10 expression, particularly in combination with EL-2 expression, can have significant immune-dependent anti-tumor actions within intracranial gliomas.

Introduction. Interleukin- 10 (EL-10), produced by the Th, subset of CD₄ cells, suppresses cytokine production by the Thi subset of CD4⁺ helper T-lymphocytes. EL-10 also inhibits the production of numerous pro-inflammatory cytokines by monocytes. EL-10 expression has been detected in human malignant gliomas and at higher levels in malignant vs. low grade tumors. This has led to the hypothesis that endogenous EL-10 functions to suppress anti-glioma immunity within brain. Despite the potentially immunosuppressive and anti-inflammatory actions of endogenous EL-10, evidence is mounting that transgenic IL-10 produced at high levels by engineered tumor cells can inhibit growth of systemic tumors by either stimulating anti-tumor immunity or inhibiting tumor-associated angiogenesis. Cell culture. The 9L cell line was originally established from a nitrosourea-induced gliosarcoma in Fisher 344 rats (Schmidek et al, 34 J. Neurosurg. 335-40 (1971); see above, EXAMPLE 4). 9L cells and endothelial cells were both grown at 37°C in 5% CO₂/95% air in Dulbecco's modified Eagle's medium (DMEM; Mediatech, Washington, DC) supplemented with 10% (v/v) fetal bovine serum (HyClone), 2 mM L-glutamine, 5 μl/ml gentamycin, and 300 μg/ml Geneticin (0418, GIBCO). Endothelial cell cultures were supplemented with 5 μg/ml basic fibroblast growth factor. Endothelial cells were originally isolated from brains of Lewis rats and immortalized by transfection with adenovirus 2 E1A gene under transcriptional control of the SV40 promoter. One clone, designated RBE4, was subsequently transfected with the replication-defective MFG-NB retroviral vector containing a modified lacZ gene (nls-lacZ) (see above, EXAMPLE 4). These cells, designated RBEZ, were cultured on fibronectin-coated substrata as described above.

Cytokine gene transfections. Murine IL-10-producing 9L cells (9L-IL-10) were created by transfection with the plasmid pBMGneo. EL-10 in the presence of lipofectamine (GIBCO) using the procedure of Kundu et al, 88 J. Natl. Cancer Inst. 536-41 (1996). 9L— neo control cells were produced by transfection under identical conditions with the plasmid pBMGneo lacking the EL-10 cDNA insert. Stable transfectants were selected in the presence of G418 (300 μg/ml; Life Technologies).

Murine EL-2-producing endothelial cells (RBEZ — IL-2) were constructed as described (see above, EXAMPLE 6).

Assays for cytokine production by transfected cells. Transfected clonal cell lines (9L-neo, 9L-EL-10, RBEZ-hygro, and RBEZ-EL-2) were grown to confluence in 24-well tissue culture plates. Cells were subsequently incubated with serum-free DMEM (0.5 ml/well) at 37°C for 24 hr. Conditioned media were removed, centrifuged, and supernatants were assayed by ELISA according to the supplier of capture and detection antibodies (PharMingen, San Diego, CA). Standard curves were established using purified recombinant murine EL-10 and EL-2 (PharMingen). Biological activity of IL-2 produced by RBEZ cells was confirmed using the EL-2 dependent CTLL cell line (see above, EXAMPLE 6).

Tumor and endothelial cell implantation. Tumor cells were harvested, counted using a Coulter counter (Coulter Electronics, Hialeah, FL), and resuspended in sterile DMEM immediately before implantation to host animals. For intracranial implantation into the caudate-putamen of anesthetized 200-250 g (adult) male Fisher 344 rats (Charles River Laboratories, Wilmington, MA), tumor cells alone (10⁵ cells; 9L-neo or 9L-EL-10) or a mixture of tumor cells (10³) and endothelial cells (2xl0⁶ cells; RBEZ — hygro or RBEZ — IL-2) in DMEM (2-5 μl) were injected stereotactically with a 26-gauge, beveled-tip Hamilton syringe ( ee above, EXAMPLE 6). Briefly, injections were made 3.0 mm to the right of Bregma, at a depth of 4.5 mm from the dural surface. Cells were injected over a 2 min period and the needle was left in place for 2 min after injection and then withdrawn slowly to limit leakage. Intracranial implantations of 9L and endothelial cells into the caudate/putamen of severe combined immune deficient SCID-Beige mice (C.B- 17/IcrCRl-SCID/Beige; Charles River) were performed in a similar manner to rat implantations. For mice, injections were made 2.0 mm to the left of Bregma, at a depth of 2.5 mm from the dural surface. For subcutaneous implantation into the anterior flanks of anesthetized Fisher 344 rats or SCID-Beige mice, tumor cells (10⁵; 9L-neo or 9L-EL-10) in sterile DMEM (100 μl) were injected bilaterally using a 22-gauge syringe as previously described (Arosarena et al, 640 Brain Res. 98-104, (1994)). Subcutaneous tumors were measured with dial calipers and vol- umes were calculated using the formula: Volume = (length X width)/2, as previously described by Tamargo et al, 9 J. Neuro-Oncol. 131-8 (1990).

Histology. At the indicated post-implantation times, rats were anesthetized and transcardially perfused with 100 ml of phosphate-buffered saline (PBS) followed by 350 ml fixative (4% paraformaldehyde in 0.1 M phosphate buffer, pH 7.4). Mice were perfused with 75 ml of phosphate-buffered saline followed by 150 ml of fixative. Brains were removed, placed in fixative for 4 h, and cryoprotected in 15%o and then 30% (w/v) sucrose in phosphate-buffered saline. Coronal sections (30 μm) through the level of the tumor were cut using a freezing microtome (Leitz). A series of sections from each brain was stained with hematoxylin and eosin (H and E), and maximal tumor cross-sectional areas were determined by computer-assisted image analysis using the Microcomputer Imaging Device (MCED) M4 software package (Imaging Research, Canada). Intracranial tumor volumes were estimated by the formula: Volume = (square root of maximal tumor cross-sectional area)³, as described (see above, EXAMPLE 6).

For the immunohistochemical detection of T-cells, brain sections were placed for I h at room temperature in a solution containing 3% normal horse serum. 0.2% Triton X-100, 0.2% porcine gelatin, and 0.02% sodium azide in phosphate-buffered saline. Sections were then incubated for 3 days at 4°C in MRC OX-52 at 1:1500, diluted in the above diluent. MRC OX-52 (Serotec) recognizes a cell surface antigen which is largely restricted to cells of the T-lymphocyte lineage (Robinson et al, 57 Immunology 527-31 (1986)). Antibody staining was visualized by the avidin-biotin-peroxidase complex method, using a Vectastain Elite ABC kit (Vector Laboratories). Following a 9 mm incubation in diaminobenzidine and three rinses, sections were mounted on slides, dehydrated through an alcohol gradient, cleared in Hemo-De, and coverslipped.

Inhibition of9L tumor growth by IL-10 gene transfer. The in vitro production of EL-10 by stably transfected clonal 9L cell lines was confirmed by an enzyme-linked immunoadsorbent (ELISA) assay. A clonal cell line that produced IL-10 at levels approximately 50-fold higher than control cells was selected for subsequent in vivo studies (TABLE 17). Production of IL-10 in vivo was determined by comparison of plasma samples obtained from animals bearing intracranial 9L-IL-10 or 9L — neo (control) tumors. The approximate 30-fold increase in plasma EL-10 levels found in animals implanted with JL-10-secretina tumor cells confirmed the secretion of transgenic EL-10 in vivo (TABLE 17).

TABLE 17 Expression of IL-10 by transfected 9L glioma cells in vitro and in vivo EL-10 (ng/ml)

Conditioned media

9L control 2.3 + 0.2

9L— IL-10 106.2 + 4.9

Plasma

9L control 0.05 ± 0.01

9L— EL-10 1.67 + 0.59

JL-10 within conditioned media of control-transfected OL cells to 4) and JL-10-transfected 9L cells (n = 3) or within plasma obtained from SCID-Beige mice bearing intracranial control-transfected 9L tumors (n=6) and IL-10-transfected 9L tumors n = 6) was determined by ELISA (limit of assay sensitivity > 31 pg/ml). Conditioned media was collected for 24 hr from confluent cell monolayers. Plasma was obtained from mice beanie intracranial 9L-control and 9L— JL-10 tumors of comparable size at the time of sacrifice (post-implantation day 14). * p < 0.001, Student's t-test. Data represents tests ± S.E.M.

To determine whether transgenic EL-10 production within subcutaneously (s.c.) implanted tumors was growth inhibitory, tumor growth was examined following the s.c. implantation of 9L-EL-I0 or 9L-neo (control) tumor cells into syngeneic Fisher 344 rats. Tumors developed in all animals, regardless of whether they received IL-10-producing cells or control-transfected cells. 9L — EL-10 tumors grew at a significantly slower rate than control tumors (p <0.05). By 34 days post-implantation, the volumes of 9L-JL-10 tumors averaged approximately 25% of control tumors.

To determine the effect of transgenic IL-10 on intracranial (i.e.) tumor growth. 10⁵ 9L — EL-10 or 9L — neo (control) cells were implanted into the caudate/putamen of Fisher 344 rats. Animals were sacrificed at post-implantation day 11 and the sizes of intracranial tumors were quantified. Similar to our results with s.c. tumors, the i.e. 9L-EL- 10 tumors were found to be significantly smaller than control tumors (p <0.05). These anti-tumor effects of IL-10 gene transfer required an intact host immune system (see below) and were, therefore, not secondary to autocrine effects of gene transfer on the inherent malignant characteristics of the tumor cell lines.

IL-10 and IL-2 cooperate in vivo to produce significantly greater tumor inhibition. We asked if endothelial cell-based EL-2 gene delivery could augment IL-10-mediated tumor inhibition. In these tests, the number of EL-2-secreting endothelial cells was reduced from that of our prior report (10-fold reduction relative to the number of 9L glioma cells) to minimize the anti-glioma effects of EL-2 and thereby increase the likelihood of identifying either additive or synergistic actions of IL-2 combined with EL-10 gene transfer.

JL-10 + JL-2 tumors were generated by implanting to the caudate putamen a mixture of glioma cells that produce IL-10 (9L-JL-10) and non-tumorigenic endothelial cells that produce JL-2 (RBEZ-IL-2). Control tumors were generated by using a mixture of control-transfected glioma cells (9L-neo) and control-transfected RBEZ cells (RBEZ-hygro). We also generated tumors expressing EL-2 alone (RBEZ-EL-2 cells + control 9L-neo cells) and tumors expressing IL-10 alone (9L-JL-10 cells + control RBEZ-hygro cells). In each case, we implanted 10⁵ glioma cells and 2xl0⁶ endothelial cells. Animals were sacrificed 6 or 14 days post-implantation and brains were evaluated for tumor growth. Histological examination of hematoxylin and eosin (H and E)-stained brain sections revealed that, at both time points, EL-10 + EL-2 tumors were substantially smaller in volume than control tumors and were considerably smaller in volume than tumors expressing EL-10 or IL-2 alone.

Quantitation of intracranial tumor volumes showed that EL-10 + EL-2 tumors were approximately 88% smaller than control tumors at 6 days post-implantation (p <0.001) and were over 99% smaller than controls (p <0.001) at 14 days post-implantation. In comparison, EL-2 tumors were approximately 20% smaller than control tumors at 6 days and 30% smaller than controls at 14 days. IL-10 tumors were about 50% smaller than control tumors at 6 days and 70% smaller than controls at 14 days. Interestingly, JL-2 tumors and IL-10 tumors grew significantly in size from 6 days to 14 days (p <0.01) while the JL-10 + IL-2 tumors appeared to decrease in volume during that period (p = 0.07).

The cooperative action of EL-10 and IL-2 was also reflected in the survival of animals bearing intracranial tumors. Fisher 344 rats were implanted intrastriatally with the same combinations of glioma and endothelial cells as described above. Although delivery to tumors of JL-10 or EL-2 alone resulted in inhibited tumor size, the survival of animals bearing tumors engineered to secrete either cytokine alone was not increased in comparison to animals bearing control tumors (p > 0.05). In contrast, simultaneous intra-tumoral expression of JL-10 and IL-2 significantly prolonged animal survival (p <0.01). Regardless of the treatment group, however, post-mortem evaluation revealed that all animals ultimately died with a substantial intracranial tumor burden.

Anti-tumor effects of IL-10 and combination IL-10 + IL-2 gene transfer are immune-dependent. To determine if the anti-tumor responses to JL-10 and combination JL-10 and IL-2 are immune-dependent, we compared the growth of 9L-JL-10 and 9L — neo (control) cells after s.c. and i.e. implantation in T-cell-, B-cell-, and NK cell-deficient SCID-Beige mice (C.B-17/IcrCrl-SCID/Beige). As observed in rat hosts, tumors developed in all animals that received inoculations of either 9L— EL-10 or control-transfected tumor cells. Subcutaneous 9L-IL-I0 and control tumors grew at essentially identical rates in these animals (p > 0.05). In accordance with the s.c. tumors, i.e. 9L — EL-10 and control tumors did not significantly differ in size at the time of sacrifice (14 days post-implantation) (p > 0.05).

To determine if the anti-tumor activity of combined EL-10 + EL-2 gene delivery is also dependent on the host immune response, we compared the i.e. growth of EL-10 + IL-2 tumors and control (9L-neo + RBEZ-hygro) tumors in SCID-Beige mice. At post-implantation day 14, animals were sacrificed and tumor sizes quantified. No significant differences in tumor volumes were observed between EL-10 + JL-2 tumors and control tumors (p > 0.05).

The host T-cell response to control and cytokine-producing intracranial tumors were also examined. Brain sections from animals sacrificed at post-implantation days 6 and 14 were immunocytochemically stained with the antibody MRC OX-52, which is directed against a pan T-cell surface antigen (Robinson et al, 57 Immunology 527-31 (1986)). This revealed a relatively low density of T-cells within control tumors and few stained cells in the peritumoral striatum. In contrast, T-cells were markedly increased in density within tumors and their surrounding striatum in animals implanted with EL-10 + IL-2 tumors. The increase in infiltrating intratumoral and peritumoral T-cells seen in brains bearing IL-10 + IL-2 tumors was also substantially more pronounced than that seen in animals bearing either IL-10 tumors or EL-2 tumors. There was no evidence for T-cell infiltration in hemispheres contralateral to tumor cell implantation in any of the control or EL-10 + EL-2 tumor-bearing animals.

Discussion. The effects of JL-10 expression on tumor growth and anti-tumor immune responses within the CNS had not been previously explored. We show in this EXAMPLE that transgenic EL-10 production by 9L gliosarcoma cells inhibits systemic and orthotopic intracranial 9L tumor growth through immune-mediated mechanisms. The anti-tumor actions of transgenic tumor cell-derived IL-10 are interesting in light of the known immunosuppressive actions of this cytokine in other settings. For example, IL-10 inhibits production of the pro-inflammatory cytokines EFNγ and TNFα in a variety of systems. Such effects would be expected to inhibit anti-tumor immune responses. JL-10 gene expression has been found within human gliomas using reverse transcriptase PCR (Huettner et al, 146 Am. J. Pathol. 317-22 (1995)) leading to a hypothesis that endogenous JL-10 functions to suppress anti-tumor immunity within brain tumor patients. However, while levels of IL-10 observed in unmanipulated wild-type tumors are typically low, the glioma cells engineered to express IL-10 in this EXAMPLE secrete transgenic EL-10 at rates 50-fold higher than control glioma cells and EL-10 immunoreactivity in plasma of animals bearing EL- 10-secreting tumor cells was elevated 30-fold in comparison to animals bearing control tumors. As is well-recognized with other cytokine-based tumor therapies, the high levels of transgenic EL-10 expression achieved within tumors following gene transfer (relative to endogenous levels) are likely to be critical to the anti-tumor effects seen in our EXAMPLES. This EXAMPLE shows that combining EL-10 and JL-2 gene transfer results in a greater than additive and possibly synergistic anti-tumor effect, by the criteria of tumor growth inhibition, animal survival, and tumor/peritumoral T-cell infiltration. The mechanism of action of JL-10 and combination EL-10 + EL-2 against intracranial 9L tumors was found to be immune-mediated, since the anti-tumor effect was lost in SCED-Beige mice that lack T, B, and NK cell functions. This conclusion is also supported by the enhanced infiltration of T-cells within tumors secreting both IL-10 and EL-2. We can also conclude from the immune-deficient animal test that tumor inhibition mediated by JL-10 is not due to potential autocrine effects on the inherent malignant characteristics of the tumor lines from the manipulations required for gene transfer. Finally, the allogenic endothelial cells used for EL-2 delivery is important in the cooperative anti-tumor actions of combination IL-10 and IL-2 in our tests. The presence of allogenic endothelial cells in all test groups, including controls, indicates that any important role for alloantigens must be EL-10 or IL-2 dependent. Conclusion. In this EXAMPLE, we showed that JL-10 gene transfer can inhibit glioma growth within brain through immune-mediated mechanisms. We examined the response of subcutaneous and intracranial rat 9L gliomas to JL-10 gene transfer. Our results establish that JL-10 gene delivery can have significant anti-glioma actions within the rat CNS. Since the Thi cytokine JL-10 synergizes with the Th₂ cytokine JL-10 in vitro, and the combined effects of these cytokines had not been evaluated previously in any tumor model, we examined the effect of IL-10 gene transfer in combination with EL-10 gene transfer on glioma growth. We showed that IL-10 and EL-10 cooperate in vivo to significantly inhibit glioma growth within the brain and to prolong animal survival. Finally, we show that an anti-tumor immune response underlies the inhibition of glioma growth following EL-10 gene transfer and combination IL-10 + IL-2 gene transfer.

EXAMPLE 8 DELIVERY OF HUMAN FIBROBLAST GROWTH FACTOR-1 GENE TO BRAIN BY MODIFIED RAT BRAIN ENDOTHELIAL CELLS

Summary. Fibroblast growth factor (FGF) is an endothelial cell mitogen that can be neuroprotective for other cell types within the central nervous system. We established brain microvascular endothelial cell lines that secrete FGF-1 with the ultimate goal of examining their usefulness as a cellular platform for FGF gene delivery to brain. A chimeric gene consisting of the secretory sequence of FGF-4 linked at the 5' end of human FGF-1 (sp-hst/KS3:FGF-l) was transfected into rat microvascular endothelial cells previously altered to express the lacZ reporter gene (RBEZ), and numerous clones were found to secrete FGF-1 (RBEZ-FGF). Immunoblotting of conditioned medium demonstrated an 18-kDa protein corresponding to FGF-1. Conditioned medium from RBEZ-FGF cells enhanced [³H]thymidine incorporation in BALB/c3T3 fibroblasts by up to sevenfold when compared with conditioned medium of control cell lines, corresponding to as much as 110 ng of active FGF-l/mg of cell protein/24 hr. RBEZ-FGF cell lines remained contact-inhibited and proliferated independent of exogenous endothelial mitogens, in contrast to control lines that are mitogen-dependent. Incubation of PC 12 cells with RBEZ-FGF cells or their conditioned medium induced neurite outgrowth by PC 12 cells. RBEZ-FGF cells survived following implantation to neonatal and adult rat caudate-putamen for at least 21 days based on 5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside (X-gal) histochemistry, and FGF-1 gene expression by these cells in vivo was demonstrated by in situ hybridization and reverse transcriptase-POR. These findings show that endothelial cells can be useful for FGF gene delivery to the CNS.

Introduction. We describe in this EXAMPLE the properties of brain microvascular endothelial cell lines genetically altered to express a chimeric human FGF-1 gene consisting of the hst/KS3 signal sequence of FGF-4 fused in-frame to FGF-1 (Forough et al, 268 J. Biol. Chem. 2960-8 (1993)). Endothelial cells expressing the sp-hst/KS3:FGF-l gene are shown to retain their contact-inhibited phenotype but no longer require exogenous endothelial mitogens for optimal proliferation in vitro. FGF-secreting endothelial cells or their conditioned medium was also found to be neurotrophic in vitro as evidenced by their ability to stimulate PC 12 cell neurite formation, a property neuronal differentiation. Finally, endothelial cells engineered to secrete human FGF-1 in vitro organize with blood vessels and express human FGF-1 in vivo following their implantation to developing rat brain. This EXAMPLE shows that the paracrine and autocrine effects of FGF may make endothelial cells that secrete FGF particularly useful for cell-based gene delivery to brain. Endothelial cells. Endothelial cells used in this EXAMPLE were originally cultured from microvessels that had been isolated from the brains of Lewis rats and immortalized using the adenovirus 2 El A gene (see above, EXAMPLE 1). A well-characterized clone of these immortalized endothelial cells (RBE4) was subsequently transfected with an nls-lacZ gene and cultured in Dulbecco's modified Eagle's medium (DMEM; Mediatech, Washington, DC, U.S.A.) containing 10% fetal bovine serum (HyClone), 0.1 M HEPES, 2 M L-glutamine, and 300 μg/mlG418 at 37°C in 5% CO₂/95% air (see above, EXAMPLE 2). RBEZ cells were transfected with the mammalian expression vector pBCMG-hygro-FGF using the polycationic transfecting reagent Lipofectamine (GibcoBRL). This plasmid contains a chimeric gene consisting of the secretory sequence of FGF-4 linked at the 5' open reading frame of human FGF-1 (sp-hst/KS3:FGF I₁-₁5₄) under the control of a cytomegalovirus (CMV) promoter and the gene for resistance to hygromycin-B (Forough et al, 268 J. Biol. Chem. 2960-8 (1993)). Control cells were generated by the same methodology using the pBCMG-hygro vector that lacked the sp-hst/KS3:FGF-l sequences. Stable cell lines following transfection with pBCMG-hygro-FGF (RBEZ-FGF) or control pBCMG-hygro (RBEZ-hygro) were selected in culture medium containing 250 μg/ml hygromycin B.

Western blot. Human FGF-1 production by endothelial cell lines was examined by Western blot analysis as described by Forough et al, 268 J. Biol. Chem. 2960-8 (1993). Serum-free conditioned medium was generated by incubating confluent monolayers of RBEZ-FGF and control RBEZ-hygro cells with 5 ml pre 10 cm diameter dish of serum-free tissue culture medium for 48 hr at 37°C. Cell lysates were prepared by incubating confluent 10-cm-diameter dishes with 1 ml of 0.01 M Tris-HCl (pH 7.5), 10 m MgCl₂, 0.5% Triton X-100, 10 μg/ml aprotinin, and 1 mM phenylmethylsulfonyl fluoride for 30 min at 4°C. Cell lysates were clarified by centrifugation at 10.000 g, and the supernatants were then incubated with heparin-Sepharose at 4°C for 24 h. The heparin-Sepharose was then rinsed with phosphate-buffered saline, heated to 100°C for 5 min with Laemmli sample buffer, and then centrifuged at 10,000 g. The supernatant fractions were then fractionated by 12.5% (wt/vol) sodium dodecyl sulfate - polyacrylamide gel electrophoresis. The electrophoresed samples were then electrophoretically transferred to nitrocellulose membranes. The membranes were then incubated in Tris-buffered saline [0.05 M Tris-HCl (pH 7.4) containing 0.15 M NaCl] containing 2% bovine serum albumin and then in Tris-buffered saline containing rabbit anti-human FGF-1. [³HJ Thymidine incorporation. The biological activity of sp-hst/KS3-FGF-l protein that was secreted by RBEZ-FGF cells was quantified by examining its ability to stimulate DNA synthesis in cultured BALB/c 3T3 fibroblasts as previously described (Forough et al, 268 J. Biol. Chem. 2960-8 (1993)). Conditioned medium consisting of DMEM supplemented with 0.5% (vol/vol) charcoal-filtered fetal bovine serum. 0.1 HEPES, 2 mM L-glutamine, and 16 U/ml heparin was collected for 48 h from confluent six- well plates (4 ml of medium per well) of RBEZ-FGF and RBEZ-hygro cell lines. The conditioned medium or nonconditioned medium supplemented with 0.1-10 ng/ml purified FGF-1 as the control was incubated with quiescent BALB/c 3T3 fibroblasts in 24-well plates (0.3 ml of medium per well) for 18 hr at 37°C. [³H] Thymidine (1.0 μCi/ml, 5 Ci/mM) was then added to each well, and the cells were incubated for another 4 hr at 37°C. The medium was removed, and the cells were fixed with 500 μl per well of 5% trichloracetic acid for 1 hr at 4°C. The fixed cells were then solubilized at room temperature for 1 hr in 300 μl of 0.1 MNaOH. The solubilized material was neutralized by addition of 300 μl of 0.1 HC1, and the radioactivity was quantified by scintillation spectroscopy.

Endothelial proliferation assays. Growth assays were conducted over seven consecutive days. RBEZ-FGF, RBEZ-hygro, and parental RBEZ cells were plated at 5,000 cells per well of 24-well plates containing 0.5 ml per well of medium consisting of DMEM supplemented with 10% fetal bovine serum, 0.1 M HEPES buffer (pH 7.4), 2 mM L-glutamine. and 16 U/ml heparin. In certain tests control RBEZ and RBEZ-hygro cell lines were also cultured in medium supplemented with either 1 ng/ml purified FGF or medium conditioned by RBEZ-FGF cells. At 24 hr intervals, cells from triplicate wells were trypsinized and counted using a Coulter Cell Counter. Fresh medium was added to each remaining well on day 4 of each 7-day test.

PC12 neurite formation. PC12 cells were grown on collagen-coated tissue culture dishes in DMEM containing 4.5 g/L glucose, 7% fetal bovine serum, 7% horse serum, 100 U/ml penicillin, 100 μg/ml streptomycin, 2 mM glutamine, and 0.1 M HEPES (pH 7.4) at 37°C. For neurite outgrowth assays, PC12 cells were seeded at 10,000-20,000 cells per well in collagen-coated 24-well tissue culture plates containing 300 μl of medium per well. After 24 hr, triplicate wells received endothelial cell conditioned medium (60 μl per well), obtained from confluent RBEZ-FGF and RBEZ-hygro cells as described above. Medium was replenished with an equivalent concentration of fresh conditioned medium at 48 h. For co-culture studies, instead of adding conditioned medium, RBEZ-FGF or control

RBEZ-hygro cells were added at 2.5-15xl0³ cells per well in endothelial cell medium to the PC12 cell structures. All cultures were maintained at 37°C and at the indicated times rinsed with phosphate-buffered saline, labeled with the endothelial cell-specific marker l,l'-dioctadecyl-3,3,3',3'-tetramethylindocarbocyanine perchlorate acetylated low-density lipoprotein (Dil-acyl-LDL; co-cultures only), and fixed with 3.7% formaldehyde. Neurites were visualized by phase-contrast microscopy, and their length was measured by computer-assisted image anlaysis with use of the Micro-computer Imaging Device (MCED) software package of Imaging Research (Brock University, St. Catharines, Ontario, Canada). Endothelial cell implantation. Sterotaxic implantation of endothelial cells to rat brain was performed as described (see above, EXAMPLE 6). In brief, cells were trypsinized and resuspended in DMEM immediately before implantation. For intracranial implantations, 10⁶ cells in 5 μl and 2.5x10⁵ cells in 2 μl were stereotactically implanted to the left hemisphere of anesthetized male Lewis rats weight 225-250 g and postnatal day 9 Lewis rat pups, respectively, using a 26-gauge, beveled-tip Hamilton syringe. With Bregma as a landmark, injection site coordinates were L 3.0 mm at a depth of 4.5 mm for adult animals and L 2.5 mm at a depth of 3.5 mm for pups. Anesthetized rats were killed 7, 14, and 21 days after implantation either by decapitation or by perfusion with 4% paraformaldehyde in phosphate-buffered saline.

Histochemistry and immunohistochemistry. RBEZ-FGF cell survival was assessed in 30 μm thick histological sections by staining with 5-bromo-4-chloro-3-indolyl β-D-galactopyranoside (X-gal), which generates a blue reaction product in the nuclei of cells that express the nls-lacZ reporter gene. Slide-mounted sections were incubated for 3 hr in 1 mg/ml X-gal (Boehringer-Mannheim) in 20 mM potassium ferrocyanide, 20 mM potassium ferricyanide, 2 mM magnesium chloride, 0.02% Nonidet P-40 (Calbiochem), and 0.01%) sodium deoxycholate. These conditions resulted in no background staining in normal age-matched control brains and in brains implanted with control endothelial cells (RBE4) that lack the nls-lacZ reporter gene as described (see above, EXAMPLE 2). To determine whether RBEZ cells had become associated with blood-brain barrier microvessels in vivo, some X-gal-stained sections were also incubated overnight at 4°C with rabbit anti-glucose transporter GLUT-1 and immunohistochemically processed using the ABC method (Vector). In situ hybridization. Sense and antisense oligonucleotide probes (45-mer) for human FGF-1 were synthesized. The 45-mer sequences were labeled with multiple residues of 5'-[α-³³P]ATP by tailing with terminal deoxynucleotidyl transferase under conditions designed to obtain tail lengths of ~15 residues.

Hybridization was carried out essentially as described by Wisden et al, in Molecular Neurobiology: A Practical Approach, Chad & Wheal, eds. 205-225 (Oxford University Press, New York, 1991). Cryostat-cut sections (10 μm thick) from flash-frozen brain were lightly fixed with paraformaldehyde, dehydrated, and stored in 95%o ethanol at 4°C. Sections were acetylated, demyelinated, and air-dried. Oligonucleotide probes were diluted to ~0.5 pg/μl in hybridization buffer containing 50% deionized formamide, 4x saline-sodium citrate (SSC), 10% dextran sulfate, 5x Denhardt's solution, 200 μg/ml cleaved salmon sperm DNA, lOOμg/ml polyadenylic acid, 120 μg/ml heparin, 25 mM sodium phosphate, and 1 M sodium pryophosphate. Probe solution (70-100 μl) was applied to each slide and covered with a glass coverslip. Hybridization was carried out at 42°C overnight. The sections were washed in lx SSC at 55°C for 30 min, with agitation. After brief rinses in lx SSC, O.lx SSC, 70% ethanol, and 95% ethanol, slides were air-dried, dipped in Kodak NTB2 emulsion, and exposed for 1-2 weeks. After the emulsion was developed, sections were lightly counterstained with cresyl violet. Reverse transcriptase PCR Brains that had received RBEZ-FGF cells or transfected control endothelial cells (RBEZ-hygro) were sectioned (1 mm thick) in the coronal plane immediately following their removal from animals, and the sections were individually flash-frozen in liquid nitrogen. Sections were homogenized, and RNA was extracted. Total RNA (1 μg) was reverse-transcribed using Moloney murine leukemia virus reverse transcriptase (2.5 μg/μl; Perkin-Elmer Cetus, Norwalk, CT) and oligodT (2.5 μM). The reaction mixture also contained 5 mM MgCl₂ and lx PCR buffer consisting of 50 m KCl, 10 mMTris-HCl (pH 8.3), 1 mMdeoxynucleotide triphosphates, and 1 U/μl Rnase inhibitor. The reagents were incubated at 42°C for 1 h, heated to 99°C for 5 min to denature Moloney murine leukemia virus reverse transcriptase, and cooled at 5°C for 5 min. The 5 μl of cDNA from the reverse transcription reaction was subjected to PCR in the presence of 5' and 3' primers at 0.15 μ each, 1.25 U of Taq polymerase (Perkin-Elmer), 2 M MgCl₂, and lx PCR buffer. PCR was performed in a DNA thennal cycler (Perkin-Elmer) for 35 cycles. Each cycle consisted of 95°C for 1 min and 60°C for 1 min. The reaction products were visualized by 1.5% agarose gel electrophoresis. The sense and antisense primers used for FGF PCR were 5'-CAAACTCCTCTACTGTAGCAACGGG-3' (SEQ ID NO:2) and

5'-TTGCTTTCTGGCCATAGTGAGTCCG-3'(SEQ JX> NO:3), respectively. The sense and antisense primers for β-actin PCR were 5'-TTGTAACCAACTGGGACGATATGG-3' (SEQ ID NO:4) and 5'-GATCTTGATCTTCATGGTGCTAGG-3'(SEQ ID NO:6), respectively (see above, EXAMPLE 6). Results. Culture medium conditioned by hygromycin-resistant endothelial cell clones transferred with either pBCMG (RBEZ-hygro)or pBCMG-FGF (RBEZFGF) was initially screened for functional FGF-like mitogenic activity based on its ability to stimulate [³H]thymidine incorporation into BALB/c 3T3 fibroblasts. Numerous RBEZ-FGF clones were found to secrete substantial amounts of FGF-like mitogenic activity, in contrast to the absence of mitogenic activity in the conditioned medium of the control RBEZ-hygro cells examined. When compared with the mitogenic activity of purified FGF-1 standards, confluent monolayers of RBEZ-FGF cell lines were found to secrete as much as 110 ng of FGF-l/mg of cell protein/ 24 hr.

Conditioned medium and cellular extracts obtained from RBEZ-FGF and control RBEZ-hygro clonal cell lines were examined for the presence of heparin-binding chimeric sp-hst/KS3FGF-l polypeptide. Immunoblot analysis of conditioned medium and cell lysates following their adsorption and subsequent elution from heparin-Sepharose demonstrated an 18-kDa immunoreactive protein produced by RBEZ-FGF cells but not RBEZ-hygro cells. This demonstrates specific expression and secretion of recombinant sp-bst/KS3:FGF-l protein in the RBEZ-FGF clonal cell lines. Autocrine effects of endothelial FGF expression. We asked if sp-hst/KS3:FGF-l expression altered the phenotypic properties of RBEZ cells. No differences in in vitro morphology or intercellular relationships were observed among parental RBEZ cells, RBEZ-FGF cells, and RBEZ-hygro cell lines using phase-contrast microscopy. RBEZ-FGF cells continued to be contact inhibited and to form well-organized cobblestoned monolayers. Likewise, there were no differences in the ability of RBEZ-FGF and RBEZ-hygro cells to incorporate rapidly the endothelial cell-specific marker Dil-acyl-LDL.

Peak proliferation in vitro of RBEZ-hygro cell lines remained dependent on the addition of endothelial mitogens to growth medium, similar to parental RBEZ cells. In contrast, RBEZ-FGF lines were no longer growth factor dependent and proliferated maximally in the absence of exogenously added mitogens. RBEZ — FGF proliferation in the absence of exogenous mitogen was equivalent to that achieved by RBEZ and RBEZ — hygro cells in the presence of 1 ng/ml exogenously added recombinant FGF. We examined the effects of RBEZ — FGF conditioned medium on the proliferation of RBEZ — hygro cells in vitro. Conditioned medium obtained from RBEZ-FGF cell lines substantially enhanced endothelial cell growth when compared with control conditioned medium. These findings suggest that RBEZ-FGF cells have a growth advantage due to the receptor-mediated autocrine stimulatory effects of secreted sp-hst/KS3:FGF-l.

Neurotrophic effects of endothelial cell-based FGF expression. We asked if RBEZ-FGF cells or their conditioned medium would stimulate neurite outgrowth in the rat pheochromocytoma cell line, a model of neuronal differentiation. PC 12 neurite formation was enhanced up to eightfold when the PC 12 cells were cultured in the presence of conditioned medium derived from RBEZ-FGF cells as opposed to that obtained from control RBEZ-hygro cells (p = 0.0001). In contrast, RBEZ-hygro cells had minimal effect on neurite formation. Neurite outgrowth was also stimulated ~ 15-fold when PC 12 cells were cultured in the presence of RBEZ-FGF cells. sp — list/KS3:FGF-l expression following endothelial cell implantation to brain. We asked if RBEZ-FGF cells could be implanted to rat brain and if sp-hst/KS3:FGF-l gene expression persisted following implantation. RBEZ — FGF cells were implanted to the caudate — putamen of postnatal day 9 neonatal Lewis rats. Animals were killed at post implantation days 7, 14, and 21, and brain sections were stained with X — gal to detect the nls-lacZ reporter gene product in implanted RBEZ cells. Numerous X-gal-positive RBEZ — FGF cells were found within the ipsilateral caudate — putamen at all post-implantation times examined, and the majority of cells were associated with microvessels in the neonatal animals, especially at the later post-implantation times examined. Host vessels that became associated with the implanted cells continued to express the blood-brain barrier — specific endothelial cell glucose transporter GLUT-1, indicating maintenance of a blood — brain barrier phenotype. Implanted RBEZ-FGF cells were also found to express GLUT-1, suggesting the brain-specific differentiation of these implanted cells. This is consistent with our findings following their implantation to gliomas (see above, EXAMPLE 4).

RBEZ-FGF cells were also implanted to the caudate putamen of adult Lewis rats, and brains examined by X — gal histochemistry as late as post-implantation day 21 revealed cell survival. In contrast to that found following implantation to neonatal rats. RBEZ cells implanted to adult brains formed multicellular aggregates, with some cells organizing into vascular forms at the borders of the larger cell aggregates. No blue X — gal reaction product was found in hemispheres contralateral to those that received RBEZ-FGF cells or in brains that received endothelial cells lacking the lacZ reporter gene. Two complementary approaches were used to examine sp-hst/KS3:FGF-l expression by RBEZ-FGF cells 7 — 21 days following their implantation to rat caudate — putamen. Brain sections were examined for human FGF-1 expression by in situ hybridization using a ³³P-oligonucleotide probe (45-mer) for human FGF-1. Specific hybridization was observed in the caudate putamen of the implanted hemispheres but not in contralateral hemispheres. Total RNA was isolated from 1 mm thick sections of individual brain hemispheres obtained from adult and neonatal animals 7 and 14 days following the implantation of RBEZ — FGF cells or control transfected endothelial cells (RBEZ — hygro) and subjected to reverse transcriptase PCR using primers specific for human FGF — 1 and rat β — actin. Human FGF-1 -specific reaction product was generated by RNA obtained from tissue sections corresponding to the region of brain that had received RBEZ — FGF cells but not by RNA obtained from sections contralateral to RBEZ — FGF cell implantation or from hemispheres implanted with control endothelial cells. Similar results were found in neonatal brains implanted with RBEZ — FGF or control endothelial cells. These findings confirm endothelial cell — based human FGF-1 gene expression in adult and neonatal brains implanted with RBEZ-FGF cells.

Discussion. This EXAMPLE describes the in vitro and in vivo properties of rat brain endothelial cell lines that have been genetically altered to secrete a chimeric human FGF-1 protein consisting of the signal peptide of the hst/KS3 gene linked to human FGF-1

(sp-hst/KS3:FGF-l). We demonstrate that brain microvascular endothelial cells can maintain their characteristic in vitro morphology, growth pattern, and endothelial phenotype following a series of sequential plasmid-based and retroviral-based genetic manipulations. We chose to examine the effects of this chimeric gene because the secretory signal sequence provides a mechanism for the transfected endothelial cells to secrete efficiently and thereby to deliver FGF-1 to neighboring cells.

Rat brain microvascular endothelial cells transfected with the sp-hst/KS3:FGF-l gene construct synthesize and secrete a protein in vitro that is immunologically and functionally comparable to human FGF-1. The mitogenic activity of endothelial cell-derived sp-hst/KS3:FGF-l was demonstrated by its ability to stimulate DNA synthesis in BALB/c 3T3 fibroblasts and to stimulate endothelial proliferation in control rat brain endothelial cells. The endothelial cell lines that express and secrete sp-hst/KS3:FCF-l displayed peak growth in vitro independent of exogenous endothelial mitogen, consistent with their autocrine stimulation by sp-hst/ KS3 :FGF- 1. It is interesting that the expression of sp-hst/KS3 :FGF- 1 by endothelial cells did not alter their morphology or contact-inhibited growth pattern in vitro. This contrasts with the ability of sp-hst/ KS3:FGF-1 gene transfer to mediate the transition of NEH 3T3 cells to a transformed phenotype (Forough et al, 268 J. Biol. Chem. 2960-8 (1993)).

Conclusion. We have shown that immortalized rat brain endothelial cells can be successfully implanted to gliomas and that the implanted endothelial cells proliferate, integrate with host-derived glioma blood vessels, and express the brain-specific endothelial cell marker GLUT-1 (see above, EXAMPLE 4). Also, derivative endothelial cell lines that have been genetically altered to secrete immune-activating cytokines can be used to stimulate anti -tumor immune responses and to inhibit glioma growth in vivo (see above, EXAMPLE 4). In this EXAMPLE, we show that rat brain endothelial cells that express sphst/KS3:FGF-l can be implanted to adult and neonatal rat striatum. Cell survival for as long as 21 days (the latest post-implantation time examined) was found using histochemical methods to detect reporter gene expression. In vivo expression of the sp-hst/KS3:FGF-l transgene by implanted cells is shown by reverse transcriptase PCR and in situ hybridization.

Interestingly, the pattern of endothelial cell implantation differed in adult versus neonatal striatum. After 1 week post-implantation, the implanted endothelial cells appeared to be more diffusely distributed and almost exclusively associated with identifiable striatal blood vessels in the neonatal animals. Within adult animals the implanted cells persisted as a small mass with a smaller percentage of cells at its margin appearing to be integrated with host blood vessels. In neither case were /ocZ-expressing cells found in the hemisphere contralateral to the site of stereotactic implantation.

EXAMPLE 9

HUMAN FGF-1 GENE DELIVERY PROTECTS AGAINST QUINOLINATEINDUCED

STRIATAL AND HIPPOCAMPAL INJURY IN NEONATAL RATS Summary. In this EXAMPLE, the effects of cell-based FGF-1 gene delivery on quinolinate-induced neurotoxicity in the developing rat brain were examined. Control endothelial cells (RBE4), and RBEZ-FGF cells were implanted into right striatum at post-natal day (PND) 7. On post-natal day 10, quinolinate (150 nmol), an endogenous N-methyl-D-aspartate (ΝMDA) receptor agonist, or vehicle alone was injected into striatum ipsilateral to cell implantation. Injury was quantified in coronal sections obtained from post-natal day 17 animals by comparing striatal and hippocampal volumes ipsilateral and contralateral to the site of quinolinate injection. Human FGF-1 specific transgene expression in vivo was shown by Northern blot and RT-PCR up to 14 days after cell implantation in control animals, and up to 4 days after quinolinate exposure. Quinolinate reduced the size of ipsilateral striatum by 37% and hippocampus by 38% in animals pre-implanted with control endothelial cells. In contrast, quinolinate reduced the size of striatum by only 14% and had no effect on hippocampal size in animals pre-implanted with RBEZ-FGF cells. Thus, FGF-1 gene delivery protected the developing striatum and hippocampus from quinolinate-induced volume loss by 62% and 100%, respectively. Intrastriatal quinolinate resulted in a significant decrease in density of NOS⁺CA3 hippocampal neurons (-38%) without affecting the density of NOS⁺ neurons in hippocampal regions CA1, dentate gyrus or striatum. This response of CA3 NOS⁺ neurons appeared to be only partially reversed by FGF-1 gene delivery. Our results show that intracerebral FGF-1 gene expression within the developing brain can protect striatum and hippocampus from quinolinate-mediated injury.

Endothelial cell implantation. Endothelial cell implantation to the brains of neonatal Lewis rats was performed under stereotaxic control as described previously (see above, EXAMPLE 8). Briefly, cells were trypsinized, resuspended as single cell suspension in DMEM and maintained on ice prior to implantation to the brain. Post-natal day 7 rats were anaesthetized with ether and the calvarium exposed by a midline scalp incision. The animals were positioned in a Stoelting Lab Standard stereotaxic frame (Stoelting. Wood Dale, IL, USA) fitted with a neonatal rat adapter. The unilateral micro injections of control (RBE4) and FGF-secreting (RBEZ-FGF) cells (5x10⁵ cells/animal) in a volume of 2 μL were directed into the right striatum using a 26-gauge, beveled-tip Hamilton syringe. With the toothbar set at 2.5 mm and Bregma as landmark, injection site coordinates were AP, 0 mm; L, 2.5 mm; and depth, 4.0 mm from the dura. The syringe was left in place for 2 min following the injection to limit leakage. The animals were maintained in a temperature-controlled environment using a Hova Bator chick incubator (BFG Corp., Savannah, GA, USA) set at 35-36 °C for 1 hr after the injection, and then returned to the dams.

Quinolinate neurotoxicity. On post-natal day 10, pups were placed in a temperature-controlled incubator at 35-36 °C for 30 min and then anaesthetized with ether. The calvarium was exposed by opening the incision used for endothelial cell implantation. Animals received up to 150 nmol quinolinate (Sigma, St Louis, MO, USA) dissolved in 0.5 μL of 0.1 M phosphate buffer (pH 7.4), or phosphate buffer only as control, by stereotaxic injection into the right striatum using coordinates identical to those used for endothelial-cell implantation as described above. The animals were placed in the temperature-controlled incubator at 35-36 °C for 2 hr and then all animals were returned to their dams. Anaesthetized pups were killed at the indicated times after implantation either by decapitation or perfusion with 4% paraformaldehyde in phosphate-buffered saline.

Tissue preparation. For RNA isolation, anaesthetized rats were killed by decapitation at the indicated times following quinolinate injection. Brains were quickly removed: the left and right striatum were dissected and immediately frozen on dry ice, and stored at -70 °C. For histological and immunohistochemical analyses, rats were killed at times indicated after quinolinate injection either by perfusion with 4% paraformaldehyde (Sigma, St Louis, MO, USA) in phosphate-buffered saline or by decapitation, after which brains were immersion-fixed in 4% paraformaldehyde in phosphate-buffered saline for 24 hr. Prior to immunohistochemical analysis, brains were cryoprotected in 30% sucrose and stored at -70 °C.

Histochemistry. Brain injury was assessed by histological analysis of striatum and dorsal hippocampus. Serial 25-μm coronal sections of brains immersion-fixed in 4% paraformaldehyde were cut with a cryostat (Microm, Heidelberg. Germany) and stained with 0.5% Cresyl violet. For each section, striatal and hippocampal cross-sectional areas both ipsilateral and contralateral to quinolinate injection were measured as previously described, using a video-based computerized image analysis system (Imaging Research, St Catherines, Ontario, Canada). Coronal sections selected at 250 μm intervals spanning the entire striatum rostral to the anterior commissure were analyzed. Similarly, coronal sections selected at 250 μm intervals encompassing the entire extent of hippocampus caudal to its rostral-most edge were analyzed. The volumes of ipsilateral and contralateral striatum and hippocampus for each animal were calculated by the formula Σ₅ (cross-sectional area₅ x250) with s = each individual section. The percentage of ipsilateral volume (I) relative to contralateral volume (C) of each structure was determined by the formula II C x 100 in each animal (n = 6) and expressed as mean ± SEM.

Nitric oxide synthase immunohistochemistry. For posterior striatum, coronal cryostat sections (30 μm) obtained at 90 μm intervals rostral to the anterior commissure were analyzed. For dorsal hippocampus, coronal sections (30 μm) obtained at 90 μm intervals caudal to its rostral-most edge were analyzed. Sections were placed for 1 hr at room temperature in a solution containing 5% normal goat serum (NGS), 0.3% Triton X-100, and 0.2%) gelatin in phosphate-buffered saline. Sections were then incubated with affinity-purified rabbit anti-rat neuronal nitric oxide synthase (nNOS) antibody (1:2000 dilution) in phosphate-buffered saline with 2% normal goat serum, 0.3% Triton X-100 and 0.1% gelatin at 4 °C overnight with shaking in a humidified chamber. The sections were washed with phosphate-buffered saline and incubated with secondary antibody in 2%> normal goat serum in phosphate-buffered saline (1:400, biotinylated goat anti-rabbit; Vector Laboratories, Burlingame, CA, USA) for 30 mm at room temperature. Sections were rinsed in phosphate-buffered saline and incubated with the avidin-biotin-peroxidase complex (Vectastain Elite ABC Kit, Vector Laboratories) for 30 min at room temperature. Sections were then rinsed sequentially in phosphate-buffered saline followed by 0.05 M Tris, pH 7.7, and finally in 0.05 M Tris containing 0.01% H₂O₂ and 0.5% mg/ml diaminobenzidine. Sections were mounted on slides, dehydrated, demyelinated and cover-slipped using DPX mountant (Fluka, Germany) for histology. The density of nNOS⁺ cells in striatum and specific hippocampal subfields was quantified using a Kontron KS 400 computerized image analysis system (Carl Zeiss, NY, USA). Six sections were analyzed per structure for each of six animals per test group. Mean values were then compared statistically (n = 6 animals). The analysis of injected (ipsilateral) and non-injected (contralateral) hemispheres within each section controlled for potential variations caused by sectioning, staining or fixation.

Northern blot analysis. RBEZ-FGF cells have previously been shown to secrete biologically active human FGF-1 in vitro (see above, EXAMPLE 8). Expression of the human FGF-1 transgene in vivo following RBEZ-FGF cell implantation was examined by Northern blot analysis using a 545 bp (nucleotides 1356-1901) EcoRI and Sail restriction fragment of human FGF-1 cDNA. Total RNA was isolated from ipsilateral and contralateral striatal tissue of animals implanted with RBEZ-FGF cells or control RBE4 cells using an RNeasy isolation kit (Qiagen, Santa Clara, CA, USA) according to manufacturer's instructions. Total RNA (10 μg) was electrophoretically separated in 1% agarose gels, transferred to nytran membranes (Schleicher and Schuell, USA) and cross-linked to the membrane using an automated UV Stratalinker (Stratagene, La Jolla, CA, USA). The cDNA probe was labeled with 50 μCi (α-³²P) dCTP (specific activity 3000 Ci/mmol, Amersham, Arlington Heights, JL, USA) by random-primed DNA labeling (Boehringer Mannheim, Indianapolis, USA). Unincorporated nucleotides were separated by G-50 Sephadex chromatography (Boehringer Mannheim). Denatured probe was hybridized to the membrane at 42 °C overnight in 10 ml hybridization buffer (50% formamide, 5x SSPE, 2.5x Denhardt's solution, 0.196 mg/ml salmon testes DNA and 0.5x SSC). The membrane was then washed three times in lx SSC buffer containing 0.1% sodium dodecyl sulfate at 52 °C for 1 hr each. Membranes were exposed to Hyperfilm-ECL films (Amersham) using intensifier screens at -70 °C for 2-3 days. To control for variations in the amount of total RNA in different samples, membranes were stripped of probe by washing in lx SSC containing 0.1% SDS at 85-90 °C for 10 min and then rehybridized with cDNA probe to glyceraldehyde-3 -phosphate dehydrogenase (GAPDH).

Reverse transcriptase PCR. Total RNA (1 μg) was reverse transcribed using Moloney murine leukemia virus (MuLV) reverse transcriptase (2.5 U/μL; Perkin-Elmer Cetus, Norwalk, CT, USA) and oligo d(T)ι₆ (2.5 μM). The reaction mixture also contained (final concentration): MgCl₂, 5 mM; lx PCR buffer (KC1, 50 mM; Tris-HCl, 10 mM; pH 8.3); deoxynucleotide triphosphate (dNTPs), 1 mM; RNase inhibitor, 1 U/μL, in a total volume of 20 μL. The reaction mixture was incubated at 42°C for 1 hr, heated to 99°C for 5 min to denature the MuLV reverse transcriptase, and cooled to 5°C for 5 min. The 10 μL of cDNA from the reverse transcription reaction was subjected to PCR in the presence of 5' and 3' primers at 0.15 μM each, 1.25 U of AmpliTaq DNA polymerase (Perkin Elmer, USA), 2.0 M MgCl₂ and lx PCR buffer. PCR was performed in a DNA thermal cycler (Perkin Elmer Model 480) for 35 cycles. Each cycle consisted of 95°C for 1 min and 60°C for 1 min. The PCR reaction product was visualized by 2% agarose gel electrophoresis. The sense and antisense primers used for FGF- 1 PCR were 5 ' -C AAACTCCTCTACTGTAGC AACGGG-3 ' (SEQ ID NO:2) and 5'-TTGCTTTCTGGCCATAGTGAGTCCG-3' (SEQ ID NO:3), respectively, and for β-actin PCR were 5'- TTGTAACCAACTGGGACGATATGG-3' (SEQ ED NO:4) and 5'-GATCTTGATCTTCATGGTGCTAGG-3' (SEQ JO NO:5), respectively, as described (see above, EXAMPLE 8). Quinolinate-induced neurotoxicity. The effect of injecting quinolinate into the right striatum of post-natal day 10 Lewis rats was examined. Injury assessed at post-natal day 17 was evidenced by a dose-dependent reduction in size of both striatum and hippocampus ipsilateral to the site of quinolinate injection. Quinolinate at 150 nmol was found to cause moderate striatal and hippocampal injury. No histological evidence of injury to the corpus striatum or hippocampus contralateral to the hemisphere of quinolinate injection was observed. We chose 150 nmol quinolinate in subsequent test designed to examine the neuroprotective action of FGF-1 gene delivery.

Analyses of human FGF-1 transgene expression in vivo. Northern hybridization using RNA isolated from striatum 7 and 14 days following RBEZ-FGF cell implantation confirmed FGF-1 transgene expression ipsilateral to the site of cell implantation. No detectable hybridization occurred to RNA isolated from striatum contralateral to the RBEZ-FGF cell implantation or from striatum implanted with control RBE4 cells lacking the human FGF-1 transgene.

FGF-1 expression was also examined in brains that had been injected with quinolinate 3 days after RBEZ-FGF cell implantation. Total striatal RNA was isolated from animals killed at 2 hr, 24 hr , 48 hr, and 4 days following quinolinate injection and subjected to reverse transcriptase PCR (RT-PCR) using primers specific for human FGF-1 and β-actin as controls. Human FGF-1 -specific reaction products resulted from all RNA samples obtained from ipsilateral striatum pre-implanted with RBEZ-FGF cells. No FGF-1 -specific amplification products were found using RNA isolated from hemispheres contralateral to RBEZ-FGF cell implantation or hemispheres implanted with control RBE4 cells. These findings confirmed endothelial cell-based human FGF-1 gene expression in the relevant hemisphere prior to and up to at least 4 days following quinolinate exposure. The reduction in FGF-1 RT-PCR reaction product seen from 24 hr postquinolinate injection suggests that acute quinolinate injury reduces FGF-1 transgene expression levels relative to uninjured controls possibly due to a reduction in either endothelial cell survival or reduced activity of the CMV promoter controlling FGF-1 transgene expression.

Attenuation of quinolinate-induced injury by FGF-1 gene delivery. To determine whether the expression of transgenic FGF-1 in vivo protects against quinolinate-induced neonatal striatal and hippocampal injury, animals pre-implanted with RBEZ-FGF cells or RBE4 cells as controls at post-natal day 7 received intrastriatal quinolinate injections at post-natal day 10. Striatal and hippocampal volumes were then quantified on post-natal day 17. In animals pre-implanted with control RBE4 cells, intrastriatal injection of quinolinate (RBE4/Quin group) produced a reduction in ipsilateral striatal volume that was significantly greater than that following the injection of buffer instead of quinolinate (RBE4 group) (~37% versus ~5%, p < 0.001). The reduction in striatal volume following quinolinate injection was substantially less in animals pre-implanted with FGF-secreting cells (FGF/Quin group) in comparison to animals pre-implanted with control cells (RBE4/Quin group) (~14% versus ~37%, p < 0.001). Thus, pre-implanting FGF-secreting endothelial cells protected striatum from quinolinate toxicity by 62%. No significant differences in striatal volume were found between animals that received either FGF-secreting cells followed by quinolinate (FGF/Quin group), FGF-secreting cells followed by buffer (FGF group) or control RBE4 cells followed by buffer (RBE4 group). We have already shown that intrastriatal injection of 150 nmol quinolinate at post-natal day 10 also produces significant injury in the ipsilateral hippocampus. Pre-implanting FGF-producing cells to striatum also protected against this hippocampal injury. In animals pre-implanted with FGF-producing cells on post-natal day 7 followed 3 days later by quinolinate injection (FGF/Quin group), ipsilateral hippocampal volume determined at post-natal day 17 was unchanged (+8%) when compared to the contralateral side. This was significantly different than the reduction in ipsilateral hippocampal volume found in animals that received control RBE4 cells and quinolinate (RBE4/Quin group) (~37%>, p < 0.01). Intrastriatal injection of buffer instead of quinolinate produced no appreciable change in ipsilateral hippocampal volume among animals pre-implanted with either RBE4 cells (RBE4 group) (+ 3%), or FGF-producing cells (FGF group) (+ 1%) when compared to their contralateral sides. Thus. FGF-1 gene transfer in vivo completely protected the neonatal hippocampus from quinolinate-induced volume loss.

Effect of FGF-1 on quinolinate-induced neurotoxicity to nNOS* -neurons. Histological sections obtained from animals pre-implanted with either RBEZ-FGF or RBE4 cells, and subsequently injected with quinolinate (RBEZ-FGF/Quin group and RBE4/Quin group, respectively), were stained immunohistochemically with anti-nNOS antibody to determine the effect of quinolinate on the density of nNOS neurons. Quinolinate had no effect on the density of nNOS⁺ striatal neurons under these conditions that substantially reduce striatal volume. In contrast to that found in striatum, quinolinate significantly decreased the density of nNOS⁺ neurons within the hippocampal CA3 subfield (-38%, p < 0.001), but not within CA1 or dentate gyrus. When compared to animals pre-implanted with control RBE4 cells, FGF-1 gene delivery resulted in a trend toward a smaller reduction in density of these CA3 nNOS⁺ neurons (-38%, P = 0.003 versus ~26%>, P = 0.03). The density of neurons expressing nNOS within the CA3 subfield remained significantly reduced in animals pre-implanted with FGF-secreting cells prior to quinolinate exposure.

Discussion. The present EXAMPLE describes the neuroprotective effects of endothelial cell-based human FGF-1 gene delivery in a neonatal rat model of quinolinate excitotoxicity. Our findings show that the intracerebral implantation of endothelial cells engineered to secrete a chimeric form of human FGF-1 substantially protects the developing rat brain from quinolinate-induced excitotoxic injury. We have confirmed that the direct injection of quinolinic acid into the striatum of post-natal day 10 rats produces marked structural injury with significant reductions in the size of ipsilateral hemisphere. Pre-implantation of FGF-secreting brain endothelial cells to striatum significantly attenuates the overall extent of quinolinate-induced striatal and hippocampal injury. Importantly, intracerebral expression of FGF-1 in this EXAMPLE resulted in neuroprotection against excitotoxic brain injury (66-100%).

Interestingly, we also found that neuroprotection by FGF-1 gene delivery as measured by volume loss was more pronounced in hippocampus than striatum, despite the fact that the FGF-secreting cells were implanted directly within striatum.

We also showed that quinolinate injures nNOS⁺ neurons within the developing hippocampus. We asked if exposing the neonatal brain to quinolinate affects nNOS-expressing neurons within the hippocampus. We found that quinolinate significantly reduces the density of nNOS⁺ neurons within hippocampal CA3 subfields. In contrast, we observed no quinolinate-induced reduction in the density of nNOS⁺ neurons within hippocampal dentate gyrus, CA1 subfield or striatum. Interestingly, if hippocampal volume is used as an endpoint, FGF-1 gene delivery completely protected the hippocampus from injury.

EXAMPLE 10 ENDOTHELIAL CELLS AS MOTILE PACKAGING CELLS

The primary goal of any gene therapy is to achieve successful gene transfer and expression in a target cell. Genetic therapeutic material is transfected and expressed in the target cell. The purpose of this EXAMPLE is to enhance the delivery of gene therapy vectors in patients, by using endothelial cells as motile packaging cells. The endothelial cell line of the invention is useful in gene therapy for diseases of the central nervous system, especially the brain, which is a difficult target enclosed entirely in a fixed space. Endothelial cells are injected into the brain or central nervous system of the patient. The endothelial cells then migrate to and engraft into an appropriate location. The brain endothelial cells of the invention are thus useful a motile gene therapy deliver vehicles for the treatment of neurological disease

Viruses useful as gene transfer vectors include retrovirus, which are the vectors most commonly used in human clinical trials. To generate a gene therapy vector, the gene of interest is cloned into a replication-defective retroviral plasmid which contains two long terminal repeats (LTR), a primer binding site, a packaging signal, and a polypurine tract essential to reverse transcription and the integration functions of retrovirus after infection. To produce viral vector, the plasmid form of a vector is transfected into a packaging cell line which produces Gag, Pol and Env of the retroviral structural proteins required for particle assembly. A producer cell line is usually generated using a selective marker, often a G418 resistant gene carried by the retroviral vector.

In particular, the endothelial cells have been transfected to produce packaging cell lines that produces lentiviral vectors containing the gene of interest (see, United States patent 5,665,577, incorporated by reference). Lentiviruses can infect both dividing and non-dividing cells and therefore have recently attracted much attention regarding their potential as vectors for gene delivery. These cell lines are then used as part of a system for gene transfer for treatment of neurological disease.

Alternatively, the endothelial cell line of the invention is constructed to produce retroviral gene transfer vectors using the methods of United States patent 5,614,404, describing recombinant viral vectors which coexpress heterologous polypeptides capable of assembling into defective nonself-propagating viral particles.

The endothelial cell line of the invention can be encapsulated, as described in PCT International patent application WO 97/44065, which describes biocompatible capsules containing living packaging cells that secrete a viral vector for infection of a target cell, and methods of delivery for an advantageous infectivity of the target cells.

Treatment of glioblastoma. Characteristics making glioblastoma a potential target for gene therapy include (1) the fact that gliobalastom cells are unique replicating cells within the centra] nervous system; (2) the level of understanding in the art of glioblastoma tumor biology; and (3) the fact that it is 100% fatal. Thymidine kinase, an enzyme found the herpes simplex virus, catalyzes the phosphorylation of ganciclovir. The resultant triphosphate inhibits the DNA polymerase leading to cell death.

The method of treatment involves the implantation of endothelial cells of the invention transfected with recombinant retroviral into the tumor. The retrovirus is replication deficient and infects only those cells replicating (tumor cells). Once cells are infected thymidine kinase is expressed the patient is treated with ganciclovir. In addition to the cells expressing thymidine kinase, there is evidence that neighboring cells may also be killed by this method ("the bystander effect"). See, Mineta et al, 54(15) Cancer Res 3963-6 (1994). Watkins et al, 18(2) Cancer Detect Prev 139-44 (1994).

Treatment of Parkinson's disease (PD). Parkinson's disease (PD) is an idiopathic degenerative disease of the central nervous system that leads to selective premature cell drop out of the pigmented dopaminergic cells within the substantia nigra. Loss of these cells eventually leads to the hallmark symptoms of Parkinson's disease, notably rest tremor, rigidity, bradykinesia and loss of postural reflexes. Generally, these symptoms do not begin to appear until about 80% of cells are lost within the substantia nigra.

Characteristics of Parkinson's disease making it a potential target for gene therapy include (1) an understanding of the pharmacology the basal ganglia; (2) clinical proof that modulation of the dopaminergic or glutamatergic pathway may provide symptomatic relief of the disease; (3) the severity of the disease; (4) information as to the functional anatomy of the basal ganglia; and (5) the presence of animal models of the disease.

The viral vectors contain genes that are useful for the treatment of Parkinson's disease, such as (1) tyrosine hydroxylase; (2) amino acid decarboxylase (both genes are known to be effective for gene therapy in the 6-hydroxydopamine lesioned rat model); (3) glutamate acid decarboylase (GAD) (which catalyzes the conversion of glutamate to the inhibitory neurotransmitter gamma amino-butyric acid (GABA) for the modulation of glutamatergic output of the subthalamic nucleus; also useful for treating conditions affecting dopaminergic cells, such as seizures, glaucoma, Huntington's disease, and traumatic brain injury); and (4) brain derived neural trophic factor (BDNF) (which exert trophic and

1 protective effects on dopaminergic neurons, the cell type known to degenerate in Parkinson's disease).

The ultimate goal of gene therapy in Parkinson's disease is to alter the natural history of disease. Such therapy cannot go forward into humans until first tested in animals. A common animal model for testing the effectiveness of gene therapy for Parkinson's disease is the 6-hydroxydopamine lesioned rat (see, Yoshimoto et al, 691(1-2) Brain Res 691(l-2):25-36 (1995)). To produce this model a stereotactic injection of 6-hydroxydopamine is made into the substantia nigra. This selectively destroys dopaminergic cells leaving intact post-synaptic neurons within the striatum to experience up-regulation of their dopamine receptors. Following a 2 week survival, these animals display a forced rotation (opposite direction as the side of their lesion) following challenge with a dopamine agonist such as apomorphine. Genetic therapy is determined to be effective in this model when there is a decrease in rotational behavior as compared with untreated animals.

EXAMPLE 11 PERIPHERAL ENDOTHELIAL CELLS

In addition to using endothelial cells derive from the central nervous system, the invention provides methods for using genetically modified endothelial cells of peripheral (non-central nervous system) origin, into which selected genetic material of interest has been incorporated and and in which expression of the genetic material is desired. Peripheral endothelial cells transduced with the genetic material are then transplanted into the central nervous system, as described above in EXAMPLES 4-9. Peripheral endothelial cells can also express genetic material encoding a selectable marker, thus providing a means by which cells expressing the incorporated genetic material are identified and selected for in vitro. The isolation and maintenance of endothelial cells from capillaries and large vessels

(e.g., arteries, veins) of many species of vertebrates has been well described in the literature (see, for example, United States patent 6,001,350, McGuire & Orkin, 5 Biotechniques 546-554 (1987); Booyse et al., 34 Thrombosis Diath. Haemorrh. 825-839 (1975)). Endothelial cell progenitors can also be isolated from circulating blood. In vitro, these cells differentiate into endothelial cells (see, United States patent 5,980,887, incorporated by reference). To enhance the growth of endothelial cells in culture, an endothelial cell mitogen can be added. Such endothelial cell mitogens include, for example, acidic and basic fibroblast growth factors (aFGF and bFGF), vascular endothelial growth factor (VEGF), epidermal growth factor (EGF), transforming growth factor α and β (TGF- and β) platelet-derived endothelial growth factor (PD-ECGF), platelet-derived growth factor (PDGF), tumor necrosis factor a (TNF-.alpha.), hepatocyte growth factor (HGF), insulin like growth factor (IGF), erythropoietin, colony stimulating factor (CSF), macrophage-CSF (M-CSF), granulocyte/macrophage CSF (GM-CSF) and nitric oxidesynthase (NOS). The nucleotide sequence of numerous endothelial cell mitogens, are readily available through a number of computer data bases, for example, GenBank, EMBL and Swiss-Prot.

As with central nervous system-derived endothelial cell described above, peripheral endothelial cells can also be genetically modified (see, United States patent 6,001,350, incorporated by reference). EXAMPLE 12 SVAREC INTRACEREBRAL EVALUATION Introduction. The purpose of this EXAMPLE was to investigate if an endothelial cell line of a peripheral origin (for example SVAREC) is able to integrate in the host rat brain parenchyma following a stereotactic cerebral injection as demonstrated above for the RBE4 cell line. During development, normal peripheral-type endothelial cells can be induced to become brain-like following transplantation to the brain. SVAREC (SV 40 T immortalized Aortic Rat Endothelial Cell line) is an endothelial cell line of peripheral type origin (Charreaii et al. 58(11) Transplantation 1222-9 (1994)), in contrast to the RBE4 cell line established from primary cultures of rat brain endothelial cells.

Cell suspension preparation. SVAREC and RBE4 cells were incubated in culture medium with the bisbenzimide dye Hoechst 33342 (Sigma) at the concentration of 5 μg/ml during 30 min at 37°C. Dishes were then rinsed with PBS 0.1 M calcium and magnesium free and trypsin was added. Once the cell detached culture medium was added and the cell number and viability were estimated on an hematimeter. After centrifugation (1000 g), the pellets were rinsed and resuspended twice in graft medium (PBS 0.1 M, glucose 10 mM). The final volume is adjusted to reach a concentration of 2xl0⁵ cells/μl. Control of Hoechst incorporation was made on a microscope under UV light were labeled nuclei appeared blue. Animals and transplantation procedure. Males Sprague-Dawley adult rats weighting 300-320 g (JEFA CREDO), n=9, were deeply anaesthetized by intraperitoneal injections of acepromazine 0.5% (vetranquil) 4 mg/kg body weight followed 10 min later by ketamine (Imalgene 500) 80-90 mg/kg. Animals were shaved between the ears and placed in stereotaxic frames (Stoelting). The skin was disinfected and incised and a small burr hole was drilled through the cranium at predetermined stereotaxic coordinates based on the Paxinos and Watson atlas, allowing the needle to be lowered in the right striatum. The cell suspension RBE4 (n=3) or SVAREC (n=6), 1 million cells in 5μl, was injected with a speed controller at the rate of 1 μl/min. Once the volume injected, the needle was left in place for additional 5 min and then slowly withdrawn. The skin was sutured with sterile silk (Ethicon).

Perfusion/ fixation. Deeply anaesthetised rats were perfused transcardially 7 (n=3), 14 (n=3), 30 (n=3) days following intracerebral implantation. 250 ml of PBS 0.1 M followed by 500 ml of cooled fixative (Paraformaledehyd 4% in PBS 0.1M, 4°C. The cryopreservation was obtained by 36 hours immersion in PBS 0.1 M/sucrose 20%. Brains were then frozen in isopentane and stored at -20°C up to processing for immunohistochemistry.

Evans blue injection. The rats sacrificed 30 dpg were anaesthetized with isoflurane and injected via the saphen vein with a filtrated solution of 2% Evans blue in saline at the dose of 40 mg/kg body weight. Animals were allowed to recover 1 hour before the perfusion/fixation.

Immunohistochemistry: Immunofluorescence staining on cryostate sections. Monoclonal mouse Anti-Proliferating cell nuclear antigen (PCNA, Dako, clone PC10). Sections were rehydrated in PBS 0.1 M, 2x5 min, immersed in absolute ethanol 10 min at room temperature and rainsed quickly in PBS 0.1M, saturation was obtained by incubation 30 min in PBS 0.1 M, triton 0.2%, 10% normal sheep serum. After mid rinsing the incubation with the monoclonal mouse anti-PCNA at the 1/50 dilution in the solution of saturation was performed during 2 hours at room temperature. After 4 rinsing in PBS 0.1M, a biotynilated sheep anti-mouse antibody (Amersham) diluted 1/200 in the solution of saturation was applied for one hour. Sections were then rinsed 4X5 min in PBS 0.1M, Triton 0.2%. Amplification was obtained after 1 hr incubation with streptavidin-fluorescein (Amersham) diluted 1/400 in the solution of saturation. Sections were rinsed 3x5 min in PBS 0.1M, triton 0.2% plus 1x5 min in PBS 0.1 M. The sections were then mounted in Bectashield (Vector). Monoclonal mouse anti-Glial Fibrillary Acidic Protein (GFAP, Sigma, clone G-A-5).

The same protocol was used omitting the post-fixation step in ethanol. The serum used for the saturation step was normal donkey serum. The anti-GFAP antibody was diluted 1/200. The secondary antibody (Fluorescein conjugated Donkey anti-mouse, Jackson Immunoresearch) was used at the 1/200 dilution.

Hematoxylin/Eosin staining. Slides were thawed and air-dried, dehydrated in increasing concentrations of ethanol (75,95, 100, 100%) and inversely rehydrated up to water, stained in Mayer's hemalun (ready to use, RAL) for 10 min. Sections were rinsed quickly in water, differentiated in alcohol/acid (1.5 ml HCL 35.5% in 1000ml ethanol 95%) for 15 to 20 sec, and rinsed again in water for 5 min. Sections are subsequently stained 10 min in eosine orange G solution (1%> each in water), rinsed in water for a least 30 min and checked under microscope. Section were finally dehydrated through ascending ethanol concentrations, cleared in xylen and mounted in Eukitt medium

Results. The EXAMPLE was focused on several specific points: (1) evaluation of the grafted cells spreading and position in the brain parenchyma and vessel walls at 3 time points (7, 14, 30 days) after transplantation; (2) assessment of the proliferative behavior of the cells in situ, mitosis interfering with the integration and differentiation processes; (3) evaluation of the local host response, especially gliosis and edema; and (4) evaluation of the state of the BBB at a time point for which it normally recovers after the surgical injury (30 days post-grafting).

Migration. Migration/integration of the labeled grafted cells was evaluated directly on cryostat sections under UV light illumination. Intense blue nucleus is attributed to grafted cells, especially for the early times post transplantation.

The distribution of the cells was studied at three time points engraftment as specified in the material and methods, 7, 14 and 30 days.

7 days post-graft. For the control animal (RBE4) blue nuclei were essentially localized at the site of injection (striatum). The two SVAREC animals showed only a moderate dispersion of labeled nuclei probably reflecting a modest spreading of the cells in the vicinity of the striatum. No difference between the two cell types was noticed and pictures of integration are observed in blood vessels in the immediate vicinity of the graft core (FIG. 22A).

14 days post-graft. As for the 7 days time point, the RBE4 grated animal exhibited only a modest migration whereas the two SVAREC grafted rats showed a more pronounced spreading of blue nuclei. Hoechst labeling was noticed in the striatum, and in the cortex, rostrally from the graft point. Again integration of the grated RBE4 and SVAREC cells within the vessel walls was noticed. Blue nuclei were seen lining blood vessels in luminal and abluminal positions as illustrated in FIG. 22B. 30 days post-graft. The Hoeschst labeling was still visible for both cell types but the spreading of the blue nuclei in the parenchyma was again less pronounced for RBE4 in comparison with SVAREC (FIG. 23A). Cells arboring blue nucleus were always picked out in vessel walls (FIG. 23B). Globally no increase or no reduction with time in the number of integrated cells was noticed for both cell types.

Blood brain barrier (BBB) permeability. The Evans blue intravenous injection is a classical method routinely used to evaluate the status of the blood brain barrier. In this EXAMPLE, two main questions were raised: (1) are the SVAREC cells tumorigenic?; and (2) if not tumorigenic and integrated in the vessel wall, do they retain characteristics of their original peripheric phenotype, i.e. vessel permeability?

The time of 30 days post transplantation correspond to a period for which the blood brain barrier has recover its integrity after the period of tissue remodeling following the surgical wound. At this time the blood brain barrier is normally repaired, except in tumors and potentially in sties where peripheral endothelial cells wouldn't have acquired a cerebral phenotype.

For the RBE4 and SVAREC blue Evans injected animals no leakage was noticed in the brain parenchyma 30 dpg. This demonstrates that the barrier is closed despite the integration of endothelial cells of peripheral-type origin.

Proliferation. Immunohistochemistry against PCNA was perfoπned to ensure that SVAREC cells do not proliferate in situ, an abnormal proliferation being one criteria of tumoral behavior. The comparison between RBE4 and SVAREC revealed no difference and only a few cells per slice (n < 10) where positive within and around the graft core._For 9L glioma cells grafted in the same location (tumoral positive control) showed a high proportion of PCNA positive cells. No tumoral formation was noticed in all animals studied. Gliosis. The anti-GFAP immunostaining was performed to evaluate the host brain response to the exogenous grafted cells. A strong and sustained gliosis can be observed in situation such as tumor or BBB disruption.

At 7 dpg the gliosis was moderate, consisting of reactive astrocytes surrounding and sometimes infiltrating the graft mass. This gliosis decrease at 14 dpg, and almost disappear for 30 dpg except for the RBE4 animal where numerous astrocytes where detected at the interface brain parenchyma/graft. Coloration. Hematoxylin/eosin staining was performed to evaluate the viability of the grated cells and the aspect of the graft mass. A significant proportion of cells in the graft mass showed apoptotic/necrotic aspect with swollen nucleus and condensation of apoptotic bodies. Autofiuorescence within the graft mass was noticed under epifluorescence on non treated fresh cryostat sections for all animals (RBE4 and SVAREC) at all times studied confirming that not all cells survive.

Hypercellularity was noticed at the interface graft/parenchyma, corresponding to grafted cells plus probably astrocytes and macrophages. No sign of edema or inflammation was noticed. Discussion. This EXAMPLE using a large T immortalized rat endothelial cell line of peripheral origin (SVAREC) was performed in a syngenic context (Sprague-Dawley rats) to avoid interferences of the immune system with the functional differentiation of the grafted cells. However, the control cell type, RBE4, previously evaluated in the Lewis rat brain, was here in an allogenic situation. Any of the evaluated parameters, except the migratory behavior, have shown evident differences between the two cell types. In this study the spreading of the RBE4 cells seems less intense than that of the SVAREC cells. This could be due to the allogenic versus syngenic context. In the previous study in syngenic situation, RBE4 in Lewis, long distance migration of the cells was reported.

We have controlled other aspects concerning the SVAREC cell line, particularly the capacity to divide and the related aspects of the graft mass (normal or tumoral), the hosts response or tolerance to the graft (gliosis), and the survival of the transplanted cells. At all the times studied, the SVAREC cells were in a quiescent state (PCNA negative) and no tumoral formation was noticed. Gliosis was very moderate and no edema isolating the graft mass noticed. Cell death was observed mostly in the graft core, SVAREC cell line seems less resistant than RBE4 in a syngenic environment. So no abnormal phenomenon, such as uncontrolled proliferation, excessive inflammation, massive and sustained gliosis, all parameters which could interfere with the cell differentiation thus distorting the study have been observed.

Concerning the integration capacity of the grafted cells similar results were obtained for both cell lines. Labeled nuclei with a crescent like moφhology were identified in luminal position in blood vessel located in the vicinity of the graft, demonstrating that an immortalized endothelial cell line of peripheral origin is able to integrate into adult host brain blood vessels, thus exhibiting at least one criteria or differentiation. This was observed for all the animals 7, 14, and 30 days post-implantation and probably occurs during the phase of reactive angiogenesis which usually began 4-5 days after the surgical wound. This wound healing angiogenesis is a time-restricted process and can explain why the number of integrated cells doesn't increase with time.

So mosaic blood vessels composed of cells of brain endothelial and aortic endothelial origins are generated. We have investigated a physical parameter, the in vivo permeability to Evans blue which can discriminate between intact or disrupted BBB. If endothelial cells of peripheral origin integrated into brain blood vessels have conserved their original phenotype (pre-determined differentiation) a blue Evans leakage in the brain parenchyma is expected. No leakage of Evans blue was detected in the sites where SVAREC cells integrate (i.e. blood vessels in the vicinity of the graft) indicating that the BBB is preserved. So these endothelial cells have differentiated at least in part toward a brain phenotype.

The results of this EXAMPLE are interesting because they extend the gene therapy applications of brain pathologies to peripheral-type immortalized endothelial cell lines.

As is apparent from the foregoing, the invention is in no way limited to those of its embodiments and modes of implementation and application which have just been described more explicitly; it embraces, on the contrary, all the variants which may occur to the practitioner in the field, without departing from the scope or compass of the present invention.

Claims

CLAIMSWE CLAIM:

1. An immortalized, injectable, non-tumorigenic mammalian brain endothelial cell, wherein (1) the cell displays in a stable manner at least one characteristic of differentiated mammalian brain endothelial cells selected from the group consisting of:

(a) expression of endothelial markers,

(b) secretion of vasoactive substances,

(c) expression of molecules of the maj or histocompatibility complex (MHC),

(d) expression of hormone receptors, and

(e) existence of tight junctions,

(2) the cell comprises a polynucleotide encoding an immortalizing gene, wherein expression of the immortalizing gene causes immortalization of the cell; (3) the cell comprises a promoter operably linked to a polynucleotide coding for a factor selected from the group consisting of cytokines, neurotrophins, neurotransmitters, growth factors, and neurohormones, (4) the cell is clonal, wherein the clone has been selected for (i) production of the encoded factor; (ii) non-tumorigenicity; and

(iii) the normal, non-transformed, endothelial mammalian brain cell phenotype; and (4) the cell is capable in vivo of migrating and integrating in the brain vessels or brain parenchyma of a recipient host mammal and expressing the encoded factor, wherein the cell is non-tumorigenic when integrated in the brain vessels or brain parenchyma of the recipient host mammal.

2. The endothelial cell of claim 1, wherein the immortalizing gene is selected from the group consisting of a viral oncogene, a cellular oncogene, a telomerase reverse transcriptase gene; a transforming virus, a gene that reduces tumor suppressor activity, a gene that increases telomerase activity, and a gene that inactivates the genes that restrict cell cycle progression.

3. The endothelial cell of claim 1, wherein the encoded factor is a growth factor selected from the group consisting of nerve growth factor (NGF) and fibroblast growth factor- 1 (FGF-1).

4. The endothelial cell of claim 1, wherein the encoded factor is a cytokine selected from the group consisting of interleukin-2 (IL-2) and interleukin- 10 (JL-10).

5. The endothelial cell of claim 4, wherein the cell is NTC- 121.

6. The endothelial cell of claim 1, wherein the encoded factor is thymidine kinase (TK).

7. The endothelial cell of claim 1, further comprising a promoter operably linked to a polynucleotide coding for a reporter gene.

8. A non-tumorigenic mammalian endothelial cell, wherein

(1) the cell comprises a polynucleotide encoding an immortalizing gene, wherein expression of the immortalizing gene causes immortalization of the cells;

(2) the cell comprises a promoter operably linked to a polynucleotide coding for a factor selected from the group consisting of cytokines, neurotrophins, neurotransmitters, growth factors, anti-inflammatory agents, and neurohormones,

(3) the cell is capable in vivo of migrating and integrating in the brain vessels or brain parenchyma of a host mammal and expressing the encoded factor; and (4) the cell is non-tumorigenic when integrated in the brain vessels or brain parenchyma of the host mammal.

. The endothelial cell of claim 8, wherin the immortalizing gene is selected from the group consisting of a viral oncogene, a cellular oncogene, a telomerase reverse transcriptase gene; transforming viruses, genes that reduce tumor suppressor activity, genes that increase telomerase activity, and genes that inactivate the genes that restrict cell cycle progression.

10. The endothelial cell of claim 8, wherein the oncogene is selected from the group consisting of myc, ras, rafl T antigens of papovaviruses, early proteins of papillomaviruses, and oncogenes of the Epstein-Barr virus.

11. The endothelial cell of claim 8, wherein the oncogene is selected from the group consisting of a SV40 T oncogene and the El A early region of the adenovirus 2 genome.

12. The endothelial cell of claim 8, wherein the gene that restricts cell cycle progression is selected from the group consisting of p53, Rb, the CDK-4 inhibitor (CDKN2), and prohibitin.

13. The endothelial cell of claim 1, wherein the cell is a rat cell.

14. A pharmaceutical composition, comprising:

(a) at least one immortalized, injectable, non-tumorigenic mammalian endothelial cell, wherein the endothelial cell secretes a therapeutic factor, and

(b) a pharmaceutically acceptable vehicle.

15. The pharmaceutical composition of claim 14, wherein the composition comprises between 10⁴ and 3 x 10⁵ mammalian endothelial cells/μl.

16. The pharmaceutical composition of claim 14, wherein the therapeutic factor is selected from the group consisting of nerve growth factor (NGF); fibroblast growth factor-1 (FGF-1); interleukin-2 (JL-2); and interleukin- 10 (EL-10).

17. A method for producing an immortalized, injectable, non-tumorigenic mammalian brain endothelial cell, comprising:

(a) culturing mammalian brain endothelial cells in a serum-containing culture medium, (b) transfecting the mammalian brain endothelial cells with polynucleotide comprising an immortalizing gene, (c) transfecting the mammalian brain endothelial cells of step (b) with polynucleotide comprising a promoter operably linked to a polynucleotide coding for a factor; (i) wherein the promoter can express the encoded factor in mammalian brain endothelial cells; and (ii) wherein the encoded factor is selected from the group consisting of cytokines, neurotrophins, neurotransmitters, growth factors, and neurohormones; (d) selecting the transfected mammalian brain endothelial cells of step (c) for

(i) production of the encoded factor; (ii) non-tumorigenicity; and

(iii) the normal, non-transformed, endothelial mammalian brain cell phenotype.

18. The method of claim 17, wherein the mammalian brain endothelial cells of step (a) are derived from brain microvessels.

19. The method of claim 17, wherein the serum-containing culture medium is supplemented with growth factors comprising bFGF or glutamine or both growth factors.

20. The method of claim 17, wherein the mammalian brain endothelial cells are transfected with the polynucleotide comprising an immortalizing gene between the 2^nd and the 6^th passage.

21. The method of claim 17, wherein the normal, non-transformed phenotype comprises a characteristic selected from the group conisting of:

( 1 ) expression of endothelial markers,

(2) secretion of vasoactive substances, (3) expression of molecules of the major histocompatibility complex (MHC),

(4) expression of hormone receptors, and

(5) existence of tight junctions.

22. A method for producing an immortalized, injectable, non-tumorigenic mammalian brain endothelial cell, comprising:

(a) culturing mammalian brain endothelial cells in a serum-containing culture medium, wherein the brain endothelial cells are from the cell line deposited under accession number I-l 142 at the National Collection of Microorganism Cultures (Institut Pasteur, France); (b) transfecting the mammalian brain endothelial cells with polynucleotide comprising a promoter operably linked to a polynucleotide coding for a factor; (i) wherein the promoter can express the encoded factor in mammalian brain endothelial cells; and (ii) wherein the encoded factor is selected from the group consisting of cytokines, neurotrophins, neurotransmitters, growth factors, and neurohormones; (c) selecting the transfected mammalian brain endothelial cells for (i) production of the encoded factor; (ii) non-tumorigenicity; and (iii) the normal, non-transformed, endothelial mammalian brain cell phenotype.