WO1992009688A1

WO1992009688A1 - MICE HAVING β2 MICROGLOBULIN GENE DISRUPTION

Info

Publication number: WO1992009688A1
Application number: PCT/US1991/008680
Authority: WO
Inventors: Maarten Zijlstra; Mark Bix; Rudolf Jaenisch; David H. Raulet
Original assignee: Massachusetts Institute Of Technology; Whitehead Institute For Biomedical Research
Priority date: 1990-11-21
Filing date: 1991-11-21
Publication date: 1992-06-11
Also published as: AU9076191A

Abstract

Mice heterozygous (+/-) or homozygous (-/-) for a β2-microglobulin gene disruption; a method of producing tissues which are heterozygous (+/-) or homozygous (-/-) for a β2-microglobulin disruption; tissues produced by the method; and uses therefor.

Description

MICE HAVING β2 MICROGLOBULIN GENE DISRUPTION

DESCRIPTION Background

β2-microglobulin (β2-m) is a 12 Kd polypeptide which is activated in mouse embryos already at the 2-cell stage (Sawicki, J.S. et al., Nature, 294:450-451 (1981)) and associates with the heavy chain of the polymorphic MHC class I proteins encoded by the H2-K, H2-L/D and Qa/T1a loci (Klein, J. et al., Nature, 291:455-460 (1981)).

While the role of major histocompatibility complex (MHC) class I proteins in the presentation of antigens to the immune system and in the development of the T cell repertoire is well documented (Klein, J. et al., Nature,

291:455-460 (1981); (Zinkernagel, R.M. & P.C. Doherty, Adv. Immun., 27:51-117 (1979); Bevan, M.J. and P. Fink, Immunol. , Rev. 42:3-19 (1978); Von Boehmer, H., A. Rev. Immun., 6:309-326 (1988); Marrack, P. & J. Kappler,

Immun. Today, 9: 308-315 (1988)), it has been inferred that the MHC molecules have many other non-immunological functions in mammals. For example, it has been proposed that class I molecules serve as differentiation antigens in cell-cell interactions during embryonic development (Curtis, A.S.G. & P. Rooney, Nature, 281:222-223 (1979); Bartlett, P.F. & M. Edidin, J. Cell Biol., 77:377-388 (1978)), may have a role in the function of certain hormone receptors (Verland, S. et al., J. Immun.,

143:945-951 (1989); Kittur, D. et al., Proc. Natn. Acad. Sci. USA, 84: 1351-1355 (1987); (Hansen, T. et al., Proc. Natn. Acad. Sci. USA, 86:3123-3126 (1989); Schreiber, A.B. et al., J. Cell Biol., 98:725-731 (1984); Solano, A.R. et al., Proc. Natn. Acad. Sci. USA, 85:5087-5091 (1988)), and that class I molecules secreted into the urine act as olfactory cues which influence mating behavior (Singh, P.B. et al., Nature, 327:161-164 (1987); Yamazaki, K. et al., Science, 240:1331-1332 (1988)).

Furthermore, it has been demonstrated that the β2-m protein associates with the Fc receptor in neonatal gut cells (Simister, N.E. & K.E. Mostov, Nature, 337:184-187 (1989)), induces collagenase activity in fibroblasts (Brinckerhoff, C.E. et al., Science, 243: 655-657 (1989)), and may serve as a chemotactic protein in the fetal thymus (Dargemont, C. et al., Science, 246 : 803-806

(1989)). The numerous biological functions β2-m and MHC class I molecules may have in the life cycle of the vertebrate organism are still poorly understood.

Summary of the Invention

Described herein is a mutant heterozygous (+/-) non-human mammal which has a disrupted β2-microglobulin (β2 -m) gene and particularly a mutant heterozygous non-human mammal produced by homologous recombination in embryonic stem cells. Also described herein is a mutant non-human mammal which is homozygous (-/-) for a β2 - microglobulin gene disruption and deficient in β2-microglobulin protein synthesis and cell surface MHC-I

expression. Further described is a method of producing a mutant non-human mammal which is heterozygous or homozygous for a B2-microglobulin gene disruption. A non- human mammal heterozygous for a β2-microglobulin gene disruption can be produced by any means by which the β2-microglobulin gene can be altered at an embryonic stage and, as described herein, has been carried out by homologous recombination in embryonic cells into which a replacement-type vector has been introduced, resulting in disruption of the β2-microglobulin gene. A method of producing a mutant non-human mammal, homozygous for a β2-microglobulin gene disruption, in which β2-microglobulin protein synthesis and cell surface MHC-I expression are deficient, has been developed and is described. The mutant homozygous mouse is produced by breeding two heterozygous mice, two homozygous mice or a heterozygous mouse and a homozygous mouse. Also described herein is a method of producing cells, cell lines and tissues which are heterozygous (+/-) for the disrupted β2-microglobulin gene or are homozygous (-/-) for the disruption. In (-/-) tissue, β2-microglobulin protein synthesis and cell surface MHC-I expression are deficient.

As a result of the work described herein, mammalian cells and tissues which may be useful for transplantation because of the deficient β2-microglobulin protein synthesis and cell surface MHC-I expression are available. Of particular interest are purified cell types which do not express MHC class II proteins and which have been rendered deficient in MHC class I protein expression;

purified cell types which have been rendered deficient in MHC Class I protein expression and express MHC Class II protein and tissues comprising a large percentage of either of the two cell types. Such cell types include fibroblasts, keratinocytes, myoblasts and endothelial cells. In one embodiment of the present invention, in which mammalian cells altered as described herein are used, either to provide cells needed by a recipient or to provide gene therapy, cells are removed from a donor, which can be the individual to whom they will also be transplanted (the recipient) or another individual (human or non-human). The cells are cultured, using known methods, and the β2-microglobulin genes are inactivated using the methods described herein. The resulting cells, which are deficient in MHC Class I protein, are cultured and transplanted to the recipient. Alternatively, the MHC Class I deficient cells can be further manipulated, using known methods (e.g., retroviral or other vectors) to introduce a gene encoding a product to be provided to the recipient (e.g., a drug, hormone or enzyme, such as Factor VIII, Factor IX, insulin, growth hormone). In the first instance, the MHC Class I deficient cells are themselves the therapeutic or clinical product (e.g., to provide keratinocytes for wounds, retinal cells for macular degeneration). In the second instance, the MHC Class I cells are a vehicle by which the product(s) encoded by the introduced gene(s) can be delivered.

Brief Description of the Drawings

Figure 1 is a schematic diagram showing the

structure of the β2-microglobulin locus of parental D3 cells.

Figure 2 is a schematic diagram of the targeting vector used.

Figure 3 is a schematic diagram of the predicted structure of a targeted β2-microglobulin locus. Figure 4 is a schematic diagram of the wild-type gene and the mutant β2-microglobulin gene and the predicted sizes of the correctly spliced β2-microglobulin and/or neo specific mRNA transcripts.

Figure 5 shows the results of immunoprecipitations of MHC class I molecules from metabolically-labeled embryonic fibroblasts.

Figure 6 shows cell surface expression of MHC class I molecules of purified T cells obtained from lymph nodes of 4 week old wild type (+/+) and homozygous mutant (-/-) F2 animals. Negative controls are T cells from 5 week old H-2^k (BIO.BR) mice.

Figure 7 shows results of functional studies of β2-microglobulin mice. Figure 7A shows that β2-microglobulin mutant (-/-) F2 spleen cells (0,●) from two animals stimulate little or no Balb/c (H-2^d) CTL reactive with H-2^b. Shown for comparison are cultures stimulatedwith wild type littermate (+/+), (□,■) and syngeneic spleen cells (◊). Figure 7b shows β2-microglobulin cells mutant (-/-) Con A-induced blast cells (0) serve as targets for conventional anti-H-2^b CTL from secondary MLC of BIO.BR spleen cells stimulated with wild-type (+/+) F2 spleen cells. Shown for comparison is the lysis of wild type littermate target cells (■) and BIO.BR target cells (♦).

Figure 8 shows that bone marrow cells from -/- mice fail to proliferate in irradiated, MHC-matched mice.

Figure 9 shows survival of irradiated mice inoculated with bone marrow or fetal liver. Figure 9a: Irradiated groups of B6 mice were inoculated with 5 × 10⁶ +/+ or -/- bone marrow cells, depleted of T cells, or a mixture of 5 × 10⁶ of each.

Figure 9b: Irradiated B6 mice were inoculated with 7 × 10⁶ fetal liver cells from -/-, +/+ or +/- donors of embryonic age 17.5 days (E17.5).

Figure 9c: B6 mice were pretreated with NK1.1- specific mAb, or not, prior to irradiation and inoculation with 4 × 10⁶ -/- fetal (E17) liver cells. A control group received neither fetal liver cells nor antibody.

Figure 9d: Irradiated +/- or -/- recipients were inoculated with 5 × 10⁶ -/- fetal (E16) liver cells.

Figure 9e: Irradiated B10/BR recipients were inoculated with 5 × 10⁶ -/- or +/- fetal (E16) liver cells.

Detailed Description of the Invention

A mutant non-human mammal in which one or both alleles of the β2-microglobulin gene is disrupted has been produced and characterized, as described herein. In particular, mutant mice heterozygous (+/-) for a disruption in the β2-microglobulin gene and mutant mice homozygous (-/-) for a disruption in the β2-microglobulin gene have been produced.

Disruption in the β2-microglobulin gene can be produced by any means suitable for disruption of the gene at an embryonic stage and, as described herein, has been produced by genetically manipulating mouse embryonic stem cells to disrupt the β2-microglobulin gene. The

resulting embryonic stem cells containing a β2-micro globulin gene disruption were introduced into host blastocysts and chimaeric offspring were produced. When bred with wild-type females, chimaeric males transmitted the embryonic stem cell genome carrying the β2-microglobulin disruption to their offspring, which are heterozygous for the disruption. Mice heterozygous for the β2-microglobulin gene disruption were intercrossed to produce mice homozygous (-/-) for the mutation. Assessment of the homozygous mice has shown that expression of β2-microglobulin protein and functional MHC class I antigen on the cell surface is deficient and that they lack CD4 8 cytolytic T cells and are apparently healthy and fertile.

Homozygous mammals, such as mice, cows, pigs, goats, rats, rabbits and dogs, may be useful for producing tissues which have particular advantages for transplantation because they lack functional MHC class I antigen on the cell surface. In particular, purified or substantially pure mammalian cells which do not naturally express MHC Class II protein and have been rendered deficient in MHC Class I protein (antigen) on their surfaces and purified or substantially pure cells which naturally express MHC class II protein and have been rendered deficient in MHC class I protein on their surfaces can be produced and used for transplantation. In the case of cell types which do not naturally express MHC class II protein and, as a result of modification as described herein, are deficient in MHC class I protein, matching of donor cells with a recipient (as to these two key characteristics) is not needed. In the case of cells which do express MHC class II protein and have been made deficient in MHC class I protein on their surfaces, matching Is simplified because only one of these antigens (MHC class II) must be considered. Purified or substantially pure MHC I protein deficient cells can be obtained from tissues obtained from animals produced as described herein. Alternatively, they can be obtained from a normal (non-mutant) donor mammal and altered using the method described herein.

The cells can be cultured, using known methods to produce a quantity of cells useful for transplantation. In addition, cell lines, such as human cell lines, in which the β2-microglobulln gene is disrupted, preferably on both alleles, are useful as a source of tissue and cells for transplantation.

As described herein, non-human mammals, such as mice, have been derived from blastocysts injected with mutant embryonic stem cells containing a β2-microglobulin gene disruption, produced by homologous recombination. As also described, non-human mammals, homozygous for a β2-microglobulin gene disruption, which do not express detectable β2-microglobulin protein (i.e., are characterized by essentially no expression of β2-microglobulin protein) and have little, if any, functional MHC class I antigen on the cell surface have been produced. Further described are methods of producing both types of non-human mammals; methods of producing tissue from heterozygous non-human mammals or homozygous non-human mammals of the present invention; and tissue obtained from and cell lines derived from the mutant non-human mammals, which can be used for transplantation. The present invention relates to cell lines, such as human cell lines, in which the β2-microglobulin gene is disrupted on one or both alleles and use of such cell lines as a source of tissue and cells for transplantation.

As described in Example 1, mice heterozygous for the β2-microglobulin gene have been produced. Initially, the β 2-microglobulin gene was mutated by means of a replacement-type vector (Figures 1-3). The components of the vector are described in detail in Example 1. The

structure of the β2-microglobulin locus of parental cells is represented in Figure 1, the structure of the

targeting vector is shown in Figure 2, and the predicted structure of the mutated β2-microglobulin gene is shown in Figure 3. The targeting vector was introduced into embryonic stem (ES) cells and resulted in β2-microglobulin gene disruption with high frequency. Targeted clones were identified, as described in Example 1, and injected into C57BL/6J host blastocysts. Resulting chimaeric male offspring were bred with C57BL/6J females, resulting in transmission of the ES cell genome to offspring.

Mice homozygous for the β2-microglobulin gene disruption were produced by intercrossing mice heterozygous for the disrupted gene. Characteristics of resulting homozygous mice were assessed, as described in Example 2. The β2-microglobulin gene mutation was shown to have no apparent detrimental effect on the well being or breeding performance of the animals. Homozygous animals were shown to express no β2-microglobulin

protein; wild-type (+/+) animals and heterozygous (+/-) animals were shown to express the β2-microglobulin protein (Figure 5). As is also described in Example 2, homozygous mice were assessed for cell surface expression of MHC class I molecules. As shown in Figure 6, cells from homozygous mice showed no staining with β2-micro globulin, H-2k^b or Qa-2 specific antibodies. Cells from homozygous mice showed staining with several anti-D monoclonal antibodies; there was at least a 20-fold reduction in staining in homozygous cells, in comparison with staining of wild-type cells.

As described in Example 2, intestinal cells from homozygous mice failed to show significant binding of IgG, indicating that the Fc receptor heavy chain in mice must associate with β2-microglobulin protein for

functional expression on the cell surface. As is also described in Example 2, lymphoid organs were characterized for the presence of different subsets of T cells. Results showed a 100-150-fold reduction in TCR αβ⁺ CD4-8⁺ T cells in both the adult thymus and spleen of

11 day old and adult homozygous mice. In contrast, little or no difference was observed between heterozygote and wild-type cells. The populations of TCR αβ ^dim CD4⁺8⁺

T cells and αβ⁺ CD4-8- T cells were unaltered in thymus of homozygous mutant mice. Normal numbers of αβ- CD4-8⁺ thymocytes were present in thymus of day 11 mice. The presence of the MHC class II restricted CD4⁺8- T cells and surface Ig⁺B cells are unchanged in the homozygous mutant mice.

MHC-dependent control of bone marrow graft rejection is complex and poorly understood. Recent results suggest that both hybrid resistance and the rejection of allogeneic marrow grafts by irradiated mice ("allogeneic resis tance") , are me di ated by a smal l sub s e t o f p er i pheral l euko cyte s that express the NK1.1 marker, and are, therefore, distinct from conventional T-cell-mediated rejection mechanisms (Murphy, W.J. et al. , J. Exp. Med. 166:1499-1509 (1987); Yankelevich, B. et al ., J. Immunol. 142:3423-3430 (1989); Ohlen, C. et al., Science

246:666-668 (1989)). Several hypotheses to explain the specificity of bone marrow resistance have been proposed. A recent hypothesis, based on transplantations involving mice transgenic for an MHC class I (MHC-I) gene, is that self-MHC-I specific effector cells within the NK1.1+ subset reject marrow cells which they fail to engage with their MHC-I-specific receptors, while sparing those cells they appropriately recognize (Ohlen, C. et al . , Science 246:666-668 (1989)). A prediction of this model has now been tested by engrafting normal mice with marrow and fetal liver cells from mutant mice in which the β2-microglobulin gene has been disrupted by homologous recombination. Results described herein show that the mutant hemopoietic stem cells are rejected by normal, MHC-matched, as well as MHC-mismatched irradiated mice, leading to death of the recipients. Rejection of -/- cells is abolished by depleting NK1.1+ cells from the host. These findings demonstrate that the failure to express MHC-I molecules renders marrow cells susceptible to rejection.

As described herein, transgenic mice which are either heterozygous or homozygous for a disruption in the β2-microglobulin gene are available, as are methods of producing the transgenic mice, vectors useful in the methods, methods of producing tissues which are heterozygous (+/-) or homozygous (-/-) for the disruption, tissues and cell lines which contain the β2-microglobulin gene disruption and uses for the tissues and cell lines.

That is, a transgenic mammal has been produced which is heterozygous for a β2-microglobulin gene disruption which, when present on both genes, results In a homozygous mammal in which β2-microglobulin protein synthesis and cell surface MHC-I expression are deficient (i.e., there is substantial functional inactivation of MHC-1). A transgenic non-human mammal homozygous for the β2-microglobulin gene disruption has also been produced. In particular, a transgenic mouse in which there is a β2-microglobulin gene disruption has been produced using homologous recombination. A transgenic mouse homozygous for the disruption has also been produced. Other

non-human mammals which are heterozygous or homozygous for the β2-microglobulin gene disruption can be produced by the method, described herein, used to produce

heterozygous and homozygous mice.

In the present method of producing a mammal heterozygous for the β2-microglobulin gene, embryonic stem cells are grown under appropriate conditions, such as on irradiated primary embryonic fibroblasts (see Example 1). Linearized targeting vector DNA Is introduced into the embryonic stem cells; electroporation has been used to Introduce linearized DNA of the targeting vector represented in Figure 2 Into mouse embryonic stem cells. The resulting cells are maintained under conditions appropriate for cell growth and the gene-targeted clones are identified. The gene-targeted clones are introduced into host blastocytes, from which chimaeric offspring are produced.

As described in Example 1, embryonic stem cells electroporated with linearized DNA of the targeting vector represented in Figure 2 were seeded in flasks on a feeder layer of embryonic fibroblasts and maintained under conditions suitable for embryonic stem cell growth. Cells were subsequently trypsinized and divided in half. One half was seeded into wells on irradiated embryonic fibroblasts derived from transgenic mice expressing a tk-neo construct and selected with G418 for 12 days. The other half was seeded in bulk on irradiated embryonic fibroblasts expressing the neo gene and selected with G418 for 7 days and on normal medium for 3 days. DNA from the bulk-seeded half was tested by PCR for the presence of gene-targeted clones (see Example 1). Pools scored (+) as a result of the PCR assessment were further assessed using the sibling clones selected from the well-seeded half on the basis of G418 resistance. PCR clones were isolated in this manner and two clones, designated A and B, were injected into host blastocytes. Clone A resulted in a few animals with a lower degree of chimaerism than Clone B, which resulted in extensive healthy chimaera. Verification that chimaeric offspring were produced was carried out by known methods (See

Example 1).

Chimaeric males are bred with wild-type (+/+) females, resulting in transmission of the β2-microglobulin gene disruption to offspring. Homozygous (-/-) non-human mammals (mice) have been produced by intercrossing heterozygous animals. They can also be produced by crossing two homozygous animals.

Although the present method has been described with specific reference to mice, other transgenic animals can also be produced by the method. Non-human mammals heterozygous for the β2-microglobulin gene disruption can be crossed with a homozygous animal to produce homozygous offspring. Alternatively, two homozygous non-human mammals can be bred to produce homozygous offspring.

A targeting vector or plasmid useful in producing the β2-microglobulin gene disruption is also described herein. Such a targeting vector includes a nucleic acid sequence homologous to at least a portion of the

β2-microglobulin gene and additional sequences which enable the homologous sequence to recombine with the β2-microglobulin gene and cause the desired disruption. Optionally, the targeting vector also includes a gene encoding a selectable marker, such as G418 resistance (as encoded by the neo gene), by which cells containing the β2-microglobulin disruption are identified. In one embodiment, the targeting vector of the present invention has 10kb of homology to the β2-microglobulin gene (e.g., a BamHI fragment), a neo cassette in the second exon of the gene fragment and additional sequences which make it possible for homologous recombination and expression of the neo gene to occur. In a particular embodiment, represented schematically in Figure 2, the targeting vector includes 10kb of homology to the β2-microglobulin gene (a BamHI fragment), a neo cassette in the second exon of the β2-microglobulin gene, lacking the polyadenylation site, and a tk gene at the 5' end. The neo cassette used contains the 1.1 Kb blunt ended XhoI-SalI fragment of plasmid pMClneo inserted in the blunt-ended EcoRI site of exon 2 of the 10 Kb BamHI fragment. The neo^r gene is driven by the tk promoter and upstream of the tk sequences there are 3.0 Kb of plasmid sequences. Other targeting vectors can be used for this purpose and can include plasmid sequences other than those from plasmid pMClneo, and other selectable marker genes, such as those encoding the his or the hygromycin selectable marker.

Tissues and purified or substantially pure cells obtained from mutant or transgenic heterozygous or homozygous mammals, such as mice, of the present invention or derived from such cells can be used for transplantation into other animals, such as an animal in which tissue from an organ is needed. Alternatively, a cell type can be obtained from a donor, who can be the recipient or another individual (human or non-human). Such cells include, particularly, those cell types which do not naturally express MHC class II protein (e.g., fibroblasts, myoblasts, endothelial cells, lung epithelial cells, retinal cells, keratinocytes, hepatocytes, neural cells and pancreatic cells, such as islet cells). They also include cells which express MHC class II protein. The cells are cultured, the β2-microglobulin gene(s) are inactivated (using the method described herein) and the resulting cells are then transplanted onto/into the recipient, generally after culturing. The MHC class I deficient cells themselves are, in this embodiment, the treatment or therapeutic/clinical product. For example, keratinocytes rendered MHC class I deficient can be used in treating wounds, retinal cells rendered MHC class I deficient can be used for macular degeneration and pancreatic cells rendered MHC class I deficient can be used to replace or restore pancreatic products and functions to a recipient. In another embodiment, MHC class I deficient cells produced by the present method are further manipulated, using known methods, to

introduce a gene or genes of interest, which encode(s) a product(s), such as a therapeutic product, to be provided to a recipient. In this embodiment, the MHC class I deficient cells serve as a delivery vehicle for the encoded product(s). For example, MHC class I deficient cells, such as fibroblasts or endothelial cells, can be transfected with a gene encoding a therapeutic product, such as Factor VIII, Factor IX, erythropoietin, insulin or growth hormone, and introduced into an individual in need of the encoded product.

As described herein, fetal liver cells from heterozygous transgenic (+/-H-2^b) mice protected BIO.BR

recipient mice, which are H-s^k, from lethal irradiation. Irradiated BIO.BR mice rejected allogeneic H-2^b bone marrow grafts and it was not possible to overcome the ability of either MHC-matched or MHC-mismatched mice to reject MHC-deficient (-/-) fetal cells. Results described herein suggest that it Is possible to overcome the ability to reject allogeneic hemopoietic cells. This suggestion is supported by the fact that, as described in Example 3, although irradiated normal mice given marrow or fetal live from (-/-) mice died fairly soon after receipt of the cells (e.g., 10-16 days), a 1/1 mix of (-/-) and (+/+) cells resulted in longer term survival (at least 42 days). In addition, pretreatment of normal mice, prior to irradiation and inoculation with (-/-) fetal liver cells, with an antibody reactive with NK1.1+ spleen cells, which include all natural killer cells, has been shown to abrogate rejection of the cells. Control recipients, not pretreated with antibody, died 7-12 days after transplantation.

Cells other than hemopoietic cells (e.g., cells other than bone marrow cells and liver cells) can also be obtained from mice heterozygous or homozygous for the β2-microglobulin gene disruption. Rejection of other (non-hemopoietic) tissue has been assessed. Skin transplants have been carried out, using known procedures. (See Billingham and Medawar, J. Exp. Biol., 28:385

(1951)). Briefly, in this procedure, skin is removed from donors after sacrifice and placed on anesthesized recipients on which a graft bed was previously prepared by removing a 1 cm² piece of skin. The graft is

protected by a plaster cast, animals are checked daily and the cast is removed under anesthesia. Results of this assessment are shown in Tables 1 and 2.

TABLE 1

MUTANT MICE REJECT SKIN OF ALLOGENEIC DONORS

DONOR RECIPIENT DIFFERENCE REJECTION

BALB/c (H-2^d) (129 × C57BL/6)F2 allo H-2 +

+/- (control)

(H-2^b)

(129 × C57BL/6)F2 allo H-2 + -/- (mutant)

C57B1/6 (H-2^b) control +/- minor H +

mutant -/- minor H +

TABLE 2

MUTANT SKIN IS REJECTED OVER MIKOR H/ALLO_H-2 DIFFERENCES

DONOR RECIPIENT DIFFERENCE REJECTION

129 +/+ 129 +/+ (H-2^b) - - 129 +/- 129 +/- - -

(129 × C57BL/6)F1 +/- - - C57B1/6 (H-2^b) minor H + BALB/c (H-2^d) allo H-2 +

129 -/- 129 +/- - -

(mutant)

(129 × C57B1/6)F1 - -

C57B1/6 minor H +

BALB/c allo H-2 +

As shown, mutant recipients (mice with β2-microglobulin gene disrutpion) rejected skin from allogeneic donors (Table 1). This suggests that rejection is due to MCH Class II molecules. As shown in Table 2, when mutant skin was grafted onto wild-type or syngeneic animals, rejection did not occur, suggesting that NK cells play no role. Acceptance of transformed 129 mutant fibroblasts was also assessed. This was carried out by obtaining 13 day embryonic fibroblasts, growing them in vitro for a few days, infecting them with a transforming virus with a ras or myc oncogene and selecting the transformed cells. In 129 mice, 100,000 transformed cells were shown to be tumorogenic when injected into the animals using known techniques.

TABLE 3

INTRODUCTION OF RAS-MYC TRANSFORMED 129 MUTANT

FIBROBLASTS INTO MICE

RECIPIENT NUMBER OF CELLS

129 10⁶ 10⁵ 10⁴

# of mice # of mice # of mice +/+ 8 2

+/- 9 6

-/-

Flu/nu 3

(129 × C57B1 /6)F1 3

C57B1/6 2

BALB/C 3

As shown in Table 4, tumor l atency p e r i o d was 2 - 3 weeks , at whi ch t ime animal s were sacrificed. Results showed that the mutant fibrosarcoma cells were not rejected, possibly because such cells are MHC Class II negative and, as a result of the mutant β2-microglobulin gene, are deficient in MHC-1 expression on their cell surface. TABLE 4

RAS-MYC TRANSFORMED 129 MUTANT_FIBROBLASTS

GROWTH IN ALLOGENEIC HOSTS

# OF MICE RECIPIENT TUMOR LATENCY

10 129 (H-2)^b 2 weeks

10 C57B1/6 (H-2)^b 2 weeks 3 Balb/c (H-2)^d 3 weeks

Other tissues, such as kidney, brain, pancreas and heart, can also be obtained from animals homozygous for the disruption and Introduced into a recipient, preferably of the same species. Alternatively, cells with a particular function or activity can be separated from surrounding supporting or connective cells in a tissue obtained from a homozygous mouse and introduced into a recipient. For example, islets of Langerhans can be separated from other, non-insulin producing cells in the pancreas and introduced into an individual in whom insulin production or utilization is compromised.

Isolated islet cells can be introduced, for example, into an Individual contained within an appropriate device or material and serve as a replacement or supplemental source of insulin. At the present time, an important consideration in making an artificial pancreas is the need to separate islet cells from the recipient's immune system, in order to prevent or minimize immune response and rejection. With cells produced by homozygous animals of the present invention, however, this may not be a consideration. Such cells express little, if any, functional MHC class I antigen on their surface and, therefore, may not trigger the immune response which is normally seen. Liver cells obtained from mutant heterozygous or homozygous animals can also be used therapeutically. For example, liver cells from a mutant animal can be introduced into the spleen of a recipient, from which they will migrate to the recipient's liver, where they can replace of supplement liver cells whose function has been altered (e.g., through a disease or inherited condition.

Cell lines in which the β2-microglobulin gene is disrupted, either in one or in both alleles can also be established, using known methods, and provide cells for transplantation or engraftment into a recipient.

Thus, as described herein, non-human mutant mammals in which one or both alleles of the β2-microglobulin gene is disrupted have been produced. Cells and tissues obtained from or cell lines derived from such mutant animals can be used for transplantation or engraftment and may be particularly valuable for this purpose because they lack functional MHC Class I antigen on their

surface. In some instances in which cells or tissues from mutant animals or from cell lines containing the β2-microglobulin gene disruption are introduced into non-mutant animals, additional treatment or manipulation will be needed in order to facilitate use of such cells and minimize or eliminate rejection by the recipient. For example, if hemopoietic cells homozygous for the β2-microglobulin gene are transplanted into an Individual, their use can be coupled with anti-NK cell therapy in order to prevent cell rejection or reduce the extent to which it occurs. As described herein, rejection of -/- cells is abolished by depleting NK cells from the host. This has been accomplished by introduction (e.g., through Injection) of monoclonal antibodies which recognize and remove NK cells. A limited number (e.g., one or two) of injections of such monoclonal antibodies has been shown to remove NK cells for the life of the host. In those instances in which MHC Class II molecules are responsible for rejection (e.g., as Is apparently the case with skin grafts in which the β2-mIcroglobulin gene is disrupted and, thus, MHC Class I is not present), it may also be necessary to alter Class II expression, such as by knocking out Class II genes and/or their transcription.

Tissue obtained from (+/-) or (-/-) mammals of the present invention are also useful to study mechanisms of tissue rejection, such as bone marrow rejection, and subsequently to design pretreatment methods or methods carried out after transplantation which reduce or eliminate tissue rejection.

The present invention is illustrated by the

following examples, which are not intended to be limiting in any way. EXAMPLE 1 Production of Mice Heterozygous for the β2- microglobulin Gene

Mice heterozygous for the β2-microglobulin gene were produced as follows: To mutate the β2-microglobulin gene, a replacement-type vector was designed which combined several previously used characteristics which facilitate the detection of targeting events (Thomas, R.K. and M.R. Capecchi, Cell, 51:503-512 (1987); Mansour, S.L. et al., Nature, 336:348-352 (1988)). The vector is represented in Figure 1. It contained 10 kilobases (kb) of homology to the β2-microglobulin gene, a neo cassette in the second exon lacking a polyadenylation site and a tk gene at the 5' end. Not shown in the targeting vector are the 3.0-kb plasmid sequences upstream from the tk sequences. The neo cassette used contains the 1.1-kb blunt-ended XhoI-SalI fragment of plasmid pMClneo

(Thomas, K.R. & M.R. Capecchi, Cell, 51:503-512 (1988)) inserted in the blunt-ended EcoRI site of exon 2 of the 10-kb BamHI fragment of the β2-microglobulin gene. The neo gene in this construct is driven by the tk promoter with an upstream tandem repeat of the polyoma mutant enhancer region. In Figure 3, small bars indicate the position of the primers used for PCR. A box indicates the position of the 500-base pair BamHI-BglI (β2-microglobulin) fragment used as hybridization probes in the

PCR and/or Southern Blot analyses. The 900-bp EcoRI-SalI fragment used as a neo probe was derived from plasmid pMClneo. Roman numerals in Figure 1 denote the exons of the β2-microglobulin gene. Arrows indicate the sizes of BglI, HindIII and EcoRI fragments hybridizing with the β2-microglobulin probe in DNA of parental D3 cells

(Figure 1) and the predicted sizes in a targeted β2- microglobulin gene (Figure 3). In Figures 1-3, Bg^* is a Bgll polymorphism present in exon 2 of the targeting vector (C57BL/6 DNA-derived, β2- m^b allele), but absent in the ES cell DNA (129J mouse-derived, β2-m ^a allele). B, BamHI-site, Bg, BglI site, E, EcoRI site, H, HindIII site.

The following screening procedure was devised to allow rapid identification of targeted clones:

Electroporation of ES Cells

D3 ES cells (Gossler, A. et al., Prop. Natn. Acad.

Sci. USA, 83: 9065-9069 (1986)) were routinely grown on γ-irradiated (2,000 rad) primary embryonic fibroblasts (EF) in DMEM medium supplemented with 15% FCS, 1 × non-essential amino acids (Gibco) and 10^-4M β-mercaptoethanol. Electroporation of 1 × 10⁸ D3 cells using a BTX

300 transfector (50 μF, 275V) with 25 μg ml^-1 linearized vector DNA was performed as previously described by

Thomas and Capecchi (Thomas, K.R. and M.R. Capecchi, Cell

51:503-512 (1987)). The electroporation conditions result in an 80% plating efficiency in comparison with non-electroporated D3 cells. Electroporated cells were divided into 30 Independent pools and seeded into

individual 25-cm² tissue culture flasks on a feeder layer of EF.

Selection Procedure

After three one-half day's growth in normal medium, cells of individual pools were trypsinized and divided into two halves. One half was seeded into a 48-well plate on irradiated EF derived from transgenic mice expressing a tk-neo construct (Gossler, A. et al., Proc. Natl. Acad. Sci. USA, 83: 9065-9069 (1986)) and selected for 12 days in medium containing 150 μg ml (active substance) G418. From day 6-14, the medium was supplemented with 15% D3 cell-conditioned medium (0.22-μm filtered) and 1,000 U ml recombinant mouse leukaemia inhibitory factor (LIF) (William, R.L. et al., Nature, 336:684-687 (1988)). The other half was seeded as a bulk culture in a 25-cm² flask on the irradiated neo-expressing EF. The bulk cultures were selected for 7 days in medium containing 150 μg ml G418 and subsequently for 3 days in normal medium. Thereafter, DNA derived from the bulk cultures were tested by PCR for the presence of gene-targeted clones. Subsequently, the sibling cultures in the 48-well dishes were used to isolate clones from those pools scored positive by PCR. Clones were expanded and screened by PCR. Thereafter, DNA of PCR clones was characterized by Southern-blot analyses. For PCR, DNA from bulk cultures or individual clones was isolated, treated with proteinase K, deproteinized with phenolchloroform, precipitated and washed with ethanol and dissolved in 10 mM Tris buffer pH 8.0, 1 mM EDTA by standard procedures. DNA (250 ng) was tested by PCR under conditions recommended by the supplier

(Perkin Elmer Cetus) with the addition of 2 mM MgCl₁, in 50 μ l containing 200 pmol of each primer (neo,

ATATTGCTGAAGAGCTTGGCGGCGAATGGG; β2- m ,

AAGGGAGGGAGAGAAGGAGAAGGTTAGCC). PCR was run for 40 cycles using a thermal cycler (Perkin Elmer Cetus). Denaturation was performed for 1.5 min at 94ºC, annealing for 2 min at 63ºC and extension for 4 min at 72ºC. A 25-μl reaction sample was run on an 0.8% agarose gel and blot-hybridized by standard procedures.

The results of two independent experiments are summarized in Table 5.

TABLE 5

FREQUENCY OF GENE TARGETING IN D3 CELLS

PCR pools

confirmed Targeted Estimated

PCR⁺ by cell G418^r clones/total targeting

Expt pools^* cloning^* colonies⁺ clones frequency^± 1 7/32 3/3 15 5/31 1 in 30

2 19/28 4/4 36 5/40 1 in 16- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - * Expressed per total number tested.

+ Average number per 48-well plate.

± The frequency of targeting was calculated using the Poisson distribution formula, In which the fraction of plates with no recombinants, P(O) is given by e^-m, where m is the average number of recombinants per pool (m = -In P(0)). The targeting efficiency is the ratio of m to the average number of

independent G418 clones per pool, which is half the observed number based on the assumption that each pool has on average two sibling clones. PCR⁺ , scored positive by PCR.

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

In the first experiment, 7/32 cell pools, and In the second experiment, 19/28 cell pools, were identified to contain targeted cells by PCR. A large fraction of clones present in each of a total of seven positive pools were isolated, and targeted clones were identified by the PGR reaction In each pool. To confirm that the β2-m gene in the targeted clones was disrupted, DNA from seven independent clones was characterized by Southern-blot analysis using probes specific for the β2-m and neo^r genes. DNA of six clones showed the additional diagnostic Bgll, EcoRI and Hindlll fragments predicted from disruption of the β2-m gene by the targeting vector

(Figure 2). DNA of one clone contained the diagnostic EcoRI and HindIII fragments, but lacked the BglI site in exon II, most probably caused by a small mutation or deletion which can occur during a homologous recombination event (Doetschman, T. et al., Proc. Natl. Acad.

Sci., 85:8583-8587 (1986)).

The targeting frequency in the experiments was approximately 1 in 2.5×10⁶ electroporated input cells or 1/25 of the G418^r clones. This is one to three orders of magnitude higher than the targeting frequencies reported for other genes. There are at least two factors that could contribute to the high efficiency of β2-m- gene targeting. The gene could contain a hotspot of recombination, as suggested by the clustering of rearrangements in the first intron in the DNA of independent β2-m gene variants in lymphoma cell lines. In addition, the lack of a polyadenylation site in the vector reduced the number of drug-resistant clones from non-homologous integration events by about fourfold. Northern-blot analysis indicated the presence of little, if any, β2-m-specific messenger RNA in undiffer- entiated D3 cells. Therefore, high-frequency homologous recombination in ES cells is clearly not dependent on a high steady-state mRNA level, as exemplified in the study described here for the β2- m gene, and observed previously for other genes. It has been shown, for example, that the β2- m gene in F9 cells, which express only trace amounts of β2- m mRNA, is much more susceptible to

digestion by DNAase I than it is in differentiated F9 cells, which express moderate levels of β2- m mRNA. This suggests that the gene, even if it is expressed a little or not at all is in an 'opened' chromatin conformation.

To derive animals carrying a disrupted β2- m gene, two different targeted clones were injected into C57BL/6J host blastocytes. Whereas clone A resulted in a few animals with a low degree of chimaerism, extensive and healthy chimaeras were derived from clone B (Table 6).

Chimaeric males were bred with C57BL/6J females. Results showed that three males transmitted the ES cell genome to all of their offspring and three other males transmitted it to part of their offspring (Table 6). A 100% transmission is expected for 'pseudomales', which are derived from a female host blastocyst injected with the male ES cell line. In such animals, the XX germ cells of the female host are not capable of forming functional spermatozoa. Southern-blot analysis was performed with tail DNA of 17 offspring from pseudomales B2 and B3 (Table 6), and revealed the presence of the additional EcoRI fragment predicted by inheritance of the disrupted β2-m gene in eight animals. These results, therefore, indicate that the disrupted β2-m gene is genetically transmitted according to mendelian expectations.

TABLE 6

GERM-LINE TRANSMISSION OF ES CELL GENOME

Animals from injected blastocysts

Progeny Chimaeric mice

Injected No. blastocysts - - - - - - - - - - - - - - - - - - - - - - - - - - - -+ - - - - - - - - - - - - - -

ES cells Injected* born survived total tested female male

Parental D3 18 4 4 4 2 2

Clone A 49 7 3 2 0 2

Clone B 40 17 8 8 0 8

Offspring of male chimaeras

No. of germ-line Extent of offspring with ES Chimaera coat-color No. of genome/per total⁺⁺ derived from: Mouse no. chimaerism⁺ litters (% transmission)

Parental D3 1 90% 5 21/33 ( 64%)

2 70% 3 17/28 ( 61%)

Clone A A1 5% 0/40 ( 0%) Clone B B1 80% 55 1/53 ( 2%)

B2 95% 33 16/16 (100%) B3 90% 11 7/7 (100%)

B4 50% Sterile

B5 95% 55 22/22 (100%)

B6 90% 44 5/23 ( 22%)

B7 50% 33 4/19 ( 21%)- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - * ES cells (10-20) were injected into C57BL/6J blastocysts and transferred to uteri of pseudopregnant recipients.

+ Assessed by the contribution of the agouti coat colour contributed by D3 ⁺⁺ cells (129J, agouti) on the black background of C57BL/6J mice.

Assessed by the presence of the dominant agouti coat colour in offspring after breeding with C57BL/6J females.

Initial assessment of the animals heterozygous for the β2-m-gene disruption showed no detectable phenotype. Intercrossing was carried out to obtain animals homozygous for the mutation, as described in Example 2.

Considering the various other functions the β2-microglobulin protein could have, failure to express the gene could lead to disturbance of embryonic development. The β2-microglobulin gene is expressed as soon as the two- cell stage (Sawicki, J.A. et al., Nature, 294: 450 -451 (1981)), and, thus, is one of the earliest genes to be activated during mammalian development that Is known. The phenotype of homozygous mutant mice should, therefore, help in understanding the various functions that the β2-microglobulin protein, as well as its ligands, has in the life cycle of mammals.

EXAMPLE 2 Generation and Characterization_of Mice

Homozygous for the_β2-Microglobulin Disruption

Mice heterozygous for the disrupted β2- m gene were intercrossed to derive animals homozygous for the

mutation. DNA was isolated from embryos between day 14 and 18 of gestation or from the tails of weaning mice and the genotype of each animal was determined by Southern analysis as described (Zijlstra, M. et_al., Nature,

342:435-438 (1989)). Table 7 shows that 9 of 33 embryos and 18 of 66 adults were homozygous for the mutated β2-m gene. TABLE 7 TRANlMISSION OF MUTANT β2 -m GENE Genotype of Progeny (No.)

No. of

Parents Litters (Age) -/- +/- +/+

+/ - x +/- 4 (embryonic) 9 17 7

15 (postnatal) 18 28 20

- / - X -/- 1 (postnatal) 8 0 0

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - F1 mice (129 × C57/BI/6, both haplotype H-2^b) were intercrossed and offspring were genotyped as described (Zijlstra, M. et al., Nature, 342:435-438 (1989)). -/- homozygous +/- heterozygous, +/+ wild-type.

Homozygotes were indistinguishable from heterozygous or wild-type littermates, had normal body weight and, on autopsy, showed no noticeable alterations in any organ. When bred, a normal sized litter was born and raised by homozygous parents. These results indicate that the mutation has no obviously detrimental effect on the well being or breeding performance of the animals. Animals were assessed, using the methods described at the end of this example, for transcription of β2-microglobulin mRNA, expression of β2-microglobulin protein, MHC class I cell surface expression, the presence of functional Fc

receptor, and the presence of different subsets of T cells. Assessment of transcription of β2 -microglobulin mRNA was carried out and showed that homozygous mutant mice produced a truncated mRNA. Transcription of the wild- type (+/+) β2-m results In two mRNA species of 0.8 and 1.0 Kb due to the use of alternative polyadenylation signals in exon 4 (Figure 4 and Parnes, J.R. et al.,

Nature, 302:449-452 (1983)). The mutant β2-m gene contains a 1.1 Kb fragment of plasmid pMClneo inserted into exon 2 which is transcribed from the tk promoter. The inserted neo gene has the same transcriptional orientation as the disrupted β2-m gene and lacks a polyadenylation signal.

To characterize class I specific mRNA in mutant mice, total as well as polyA+ selected RNA was isolated from a variety of tissues of +/+ , +/- (heterozygous) and -/- (homozygous) mice or from embryonic fibroblasts and examined by Northern blot analysis. The blots were hybridized to a β2-m probe, the neo probe and a MHC class I heavy chain probe. Two strong bands at 0.8 and 1.0 Kb, expected for β2-m mRNA, were observed in IFN-γ treated +/+ and +/- fibroblasts; untreated cells showed a much weaker signal. In contrast, homozygous mutant cells did not synthesize the normal β2-m mRNA species, but showed instead a band at approximately 2.0 Kb, which was also seen in +/- cells and is expected for an RNA species initiated at the β2-m promoter, transcribed through the neo cassette and terminated in exon 4. The intensity of the signal suggested that this RNA was much less abundant than the wild-type β2-m RNA. Similar hybridization signals were detected in liver, kidney, spleen, brain and lung RNA of adult mutant or wild-type mice. In addition, a faint signal of approximately 1.5 Kb was seen in +/- and -/- animals which may correspond to RNA initiated at the tk promoter. It is not known why transcription from the β2- m promoter is impaired in cells of homozygous mice. It is possible that the neo insert exerts an inhibitory effect on β2-m promoter usage or that the read-through RNA is less stable. Alternatively,

additional mutations not detectable by Southern analysis may have occurred in the promoter region and interfere with its function. Hybridization to the MHC class I heavy chain probe showed the expected signal of 1.6 Kb with the same intensity in animals of all three genotypes. The results indicate that the β2-m gene disruption prevents synthesis of normal β2-m RNA, but does not interfere with the transcription or stability of MHC class I heavy chain RNA.

Expression of the β2 -microglobulin protein was assessed and results showed that homozygous mutants produced no β2-microglobulin protein.

Embryonic fibroblasts were treated with lFN-γ, labeled with [³⁵S]-methionine and protein extracts were subjected to immune precipitation using β2- m and several

MHC class I specific antibodies. Figure 5 shows that the expected 12 Kd protein was detected by precipitation with the β2- m specific antiserum in +/+ and +/- cells, but was absent in -/- cells. Recognition of the H-2K^b heavy chain by monoclonal antibody 5F1.1.24 (anti H-2K^b, al domain), however, was dependent on association of the heavy chain with β2- m because the expected band of 46 Kd was seen in wild-type and heterozygous cells but was not detected in mutant cells. In contrast, immunoprecipitation with monoclonal antibody 28-14-8S (anti H-2D^b, α3 domain) precipitated 44 and 46 Kd heavy chain bands and the 12 Kd β2-m band in wild- type and heterozygous cells, as well as a weaker 44 Kd band in homozygous cells. The 44 Kd molecule most likely represents an immature precursor molecule still containing the terminal high mannose residues (Hansen, T.H. et al., J. Immun.,

140:3522-3527 (1988)). These data suggest incomplete or altered processing of the 44 Kd H-2D polypeptide in the absence of β2-m²². These results also confirm earlier studies that monoclonal 28-14-8S can recognize H-2D molecules even when not associated with β2-m .

Surface expression of MHC class I molecules was examined by incubating purified CD4⁺8-T cells with a panel of MHC class I specific antibodies and evaluated by FACS analyses. The data shown in Figure 6 and Table 8 failed to reveal any staining of cells from homozygousmice with any of the β2- m , H-2K^b and Qa-2 specific antibodies. However, incubation with several different anti-D^b monoclonal antibodies resulted in detectable staining which was reduced 20-fold or more when compared to wild-type cells. This observation corroborates thenotion that the H-2D^b molecules can reach the cell surface even in the absence of endogenous β2-m (Allen, H. et al., Proc. Natn. Acad. Sci. USA, 83:7447-7451 (1987); Williams, D.B., et al., J. Immun., 142:2796-2806 (1989); Potter, T.A. et al., J. Exp . Med., 160:317-322 (1984)).The native configuration of D^b as detected by these antibodies may, however, depend on exogenous β2-m derived from the culture medium. TABLE 8

Cell surface expression of MHC class I molecules mean fluroescence intensity

1st Reagent Specificity +/+ F2 -/- F2 BIO.BR

none - - 1.7 1.5 1.6

NEI-026 β2-m^b 168.2 1.5 73.4

M1/42.3.9.8 K^b, H-2^k 347.7 2.3* 154.7

H141-11 K^b, K^k 219.4 2.1 128.3

K10-56.1 K^b(α1/α2) 8.2 1.5 1.3

K7-309 K^b(α1/α2) 6.7 1.6 1.3

B8-24-3 K^b(αl) 113.6 1.9 1.3

28-13-3S K^b(α2) 139.5 2.0 1.3

5F1.1.24 K^b(α2) 61.2 1.7 3.1

28-14-8S D^b(α3) 119.3 6.3 1.3

H141-31-10 D^b(α2) 36.0 1.7 1.3

B22-249R1 D^b(α1) priv. 146.9 8.5 1.3

27-11-13A D^b(α1) 112.0 3.5 1.3

D2-262 Qa-2(α3) 38.6 1.6 1.3

1-5-9 Qa-2(α1/α2) 12.1 1.6 1.3

1-4-4 Qa-2(α1/α2) 29.6 1.6 1.3

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

CD4⁺ T cells were purified from lymph node cells, as described below, reacted with the indicated mAb followed by FITC-goat anti-mouse IgG+M (or FITC-goat anti-rat IgG for Ml/42.3.9.8), and 1 × 10⁴ stained cells analyzed on an Epics C flow cytometer. The domain specificity of the monoclonal antibody, where known, is indicated. T cell purification, staining and analysis were in the presence of 5% FCS . The numbers refer to mean linear fluorescence intensity.

* Control mean fluorescence intensity with FITC-goat anti-rat IgGis 2.0. Assessment of β2-m Role in Fc Receptor Function

Recently, the β2 -m protein has been identified as the smaller component of the Fc receptor, that mediates the uptake of IgG from milk in intestinal cells of neonatal rats (Simister, N.E. & K.E. Mostov, Nature,

337:184-187 (1989)). To examine the role of β2-m for the biological function of the mouse intestinal Fc receptor, brush borders were isolated from the small intestine of

11 day old littermates and tested for the binding of

[¹²⁵I]-labeled IgG (Simister, N.E. & A.R. Rees, Eur. J.

Immun., 15:733-738 (1985)). The results demonstrate that intestinal cells from homozygous mutant mice fail to show significant binding of IgG (0.6% ± 1.8 (mean ± S.D.)) of total cpm bound. The amount of IgG bound to brush borders from heterozygotes (6.2% ± 1.9; n=4) and wild- type mice (6.9%, 4.3%, n-2) was indistinguishable. This indicates that the Fc receptor heavy chain in mouse must associate with β2-m protein for functional expression on the cell surface. Assessment of Presence of T-Cell Subsets in Lymphoid Organs

Lymphoid organs derived from F2 animals of the three genotypes were characterized by two and three color FACS analyses for the presence of different subsets of T cells. In both the adult thymus and spleen of 11 day old and adult homozygous mutant mice a dramatic 100-150 fold reduction in TCR αβ⁺ CD4-8⁺ T cells was observed. Little or no difference between heterozygote and wild-type cells was seen. In the thymus of homozygous mutant mice, the presence of the populations of TCR αβ ^dim CD4⁺8⁺ and αβ ⁺ CD4⁺8- T cells was unaltered. In addition, αβ - CD4-8⁺ thymocytes were present in normal numbers in the thymus of young (day 11 mice). These cells are thought to represent an intermediate between αβ CD4 8 and αβ

CD4+8+ cells (Nikolic-Zugic, J. et al., Eur. J. Immun., 19:649-653 (1989); Guidos, C.J. et al., Proc. Natn. Acad. Sci. USA, 86:7542-7546 (1989)). Therefore, the results imply that MHC class I cell surface expression is only essential for the development of the TCR αβ ⁺ CD4-8⁺ T cells. This strongly suggests that differentiation of αβ CD4-8⁺ T cells from αβ^{d im} CD4⁺8⁺ thymocytes requires interaction with class I MHC molecules. Finally,. it should be noted that the presence of the MHC class II restricted CD4⁺8- T cells and surface Ig⁺ B cells are unchanged in the homozygous mutant mice.

Results showed that the presence of γ/δ ⁺ T cells was unaffected in homozygous mutant mice. Some evidence suggests that at least a portion of TCR αβ - /γδ⁺ T cells may be restricted by MHC class I-like molecules encoded by the Qa/TL region of the H-2 complex. Therefore, thymi obtained from 17.5 day-old embryos and thymus and spleen from adult mice were examined for the presence of 75 T cells. The results clearly indicated that the number of γδ⁺ cells was not significantly affected by the β2-m mutation. Furthermore, the abundance in the fetal

(E17.5) thymus of a subset of γδ cells expressing the V_γ3 gene s e gment was no rmal in the β2 - m mutant mi c e .

Results further showed that no cytotoxic T cells were produced in homozygous mutant mice. Murine alloreactive cytotoxic T lymphocytes (CTL) are almost excluslvely TCR γδ⁺ CD4-8⁺ (Klein, J. et al., Nature, 291:455-460 (1981); ZInkernagle, R.M. & P.C. Doherty, Adv. Immun., 27:51-117 (1979); Bevan, M.J. & P. Fink, Immunol., rev 42:3-19 (1978); Von Boehmer, H.A., Rev.

Immun., 6:309-326 (1988); Marrack, P. & J. Kappler, J. Immun. Today, 9: 308-315 (1988)). Therefore, spleen cells derived from F2 animals were tested for the presence of CTL-precursors (CTL-p) in a bulk-mixed lymphocyte culture (MLC) against completely allogeneic BALB/c (H-2^d) cells. The results Indicate that spleen cells derived from homozygous mutants were virtually devoid of any CTL-p. The CTL responses of heterozygous animals were indistinguishable from those obtained with wild-type animals. In addition, the ability of the mutant cells to stimulate CTL-p or serve as target cells for established CTL was examined. The results shown in Figure 7A demonstrate that β2-m negative spleen cells fail to elicit a significant CTL response by BALB/c responders. Similar results were obtained with BIO.BR responders. Therefore, the low residual H-2D^b cell surface expression in homozygous mutant cells is clearly not sufficient to trigger a vigorous CTL response.

Finally, it was shown that mutant cells can serve as target cells for anti-H-2^b CTL generated in conventional MLC, although approximately 9 -fold more CTL are required to lyse mutant compared to wild-type targets (Figure 7B). This residual killing can be accounted for by two hypotheses. First, it is possible that D^b molecules assume a functional conformation even in the absence of β2- m , albeit at a dramatically reduced level. Alternatively, bovine β2- m from the serum β2- m from the serum containing medium may associate with cell surface D and facilitate refolding of the molecule. The latter hypothesis is favored, as it is consistent with published data demonstrating the binding of serum β2- m to class I molecules on cultured cells (Bernabeu, C. et al., Nature,

308: 642-645 (1985)). Serum-free conditions will address this issue.

When mutant cells were used as either responders or as stimulator cells in a mixed lymphocyte reaction (MLR), the proliferative responses were similar to those with wild-type cells. This is consistent with the fact that the proliferation measured in a MLR is predominantly determined by the recognition of foreign MHC class II molecules by CD4⁺8- T cells.

To obtain the results described above in this example, the following methods and materials were used.

To characterize class I specific mRNA in mutant mice, a

350 bp Pstl-EcoRI fragment of β2- ml cDNA was used as a β2-m-specific hybridization probe (Daniel, D. et al., EMBO J., 2:1061-1065 (1983)). This fragment spans the cDNA sequences from the 5' start site in exon 1 until the EcoRI site in exon 2. In addition, the following fragments were used as hybridization probes: a 910 bp

EcoRI-BamHI fragment derived from pMClneo (Thomas, K.R. & M.R. Capecchi, Cell, 51:503-512 (1987)) (neo probe), a 1.4 kb Pstl fragment from pH2-d-37 containing a complete cDNA of the H-2K gene (LaLanne, R. et al., Nucleic AcidsRes., 11:1567-1577 (1983)) (MHC class I heavy chain probe) and a 1.6 kbp Pstl fragment containing the entire rat α-tubulin cDNA (tublin probe). Northern blot analysis of 15 μg total cellular RNA from primary fibroblasts derived from day 14 old F2 embryos of indicated genotype was carried out using cells grown in DMEM supplemented with 10% FCS, 1x non-essential amino-acids (Gibco). Interferon treatment was performed for 24 hours by supplementing the medium with 500 U/ml recombinant mouse IFN-7 (Amgen biologicals).

Northern blot analysis of 15 μg total cellular RNA or 0.5 μg oligo-dT selected polyA mRNA (one cycle) derived from adult liver and kidney of F2 animals of indicated genotypes was carried out using total cellular RNA Isolated by the LICL/urea method and separated on a 1.5% formaldehyde-treated agarose gel and blot-hybridized by standard procedures. Filters were stripped of hybridizing probes by treatment for 20' in 10mM Tris 7.5, 1% SDS at 80°C before reprobing.

Immunoprecipitations of MHC class I molecules from metabolically-labeled embryonic fibroblasts were carried out using embryonic fibroblasts derived from day 14 old embryos, routinely cultured as outlined below. 24 hours before labeling the normal medium was supplemented with 500 U/ml IFN-γ. Subsequently, the nearly confluent cells were washed once with medium containing 9 volumes

methionine-free DMEM/1 vol. normal DMEM, supplemented with 5% dialyzed FCS (Label medium). Thereafter, cells were incubated for 2 1/2 hrs at 37ºC In label medium with the addition of 0.2 mCi/ml [³⁵S]-L-methionine (Amersham).

Subsequently, cultures were washed once with normal medium and chased for 1 hr at 37°C in normal medium.

Thereafter, cultures were washed once with ice-cold PBS and subsequently lysed in ice-cold 100 mM tris, pH 7.4, 50 mM NaCl, 1 mM MgCl₂, 0.5% Nonidet P-40 and 1% aprotinin. Lysates were centrifuged at 12,000 × g for 15 min and 8 × 10⁶ TCA precipitable cpm of the supernatant was used for subsequent immunoisolation of class I molecules. Immunoisolation was performed as described by Williams et al. (Williams, D.B. et al., J. Immun., 142:2796-2806 (1989)) with the use of protein A-Sepharose (Pharmacia). Eluted antigen was subjected to SDS-PAGE analysis using a 12% polyacrylamide gel according to Laemmli (Laemmli, U.K., Nature, 227:680-685 (1970)). Gels were fixed in acetic acid, incubated with 22% PPO (w/v) in acetic acid, dried and exposed to preflashed Kodak X-OAR5 film at -70ºC. In Figure 5, bars indicate the migration of prestained protein markers (BioRad). NRS , normal rabbit serum; anti β2-m (Serotec); NMS, normal mouse serum;

anti-D^b, monoclonal antibody 28-14-8S (α3 domain); anti K^b, monoclonal antibody 5F1.1.24 ( α1 domain).

Cell surface expression of MHC class I molecules was assessed in purified T cells obtained from lymph nodes of 4 week old wild-type (+/+) and homozygous mutant (-/-) F2 animals. T cells from 5 week old H-2^k (BIO.BR) mice served as negative controls for staining. Purified CD4 T cells were prepared by passage of lymph node cells over nylon wool columns and panning the nonadherant cells on plates coated with anti-CD4 (GK1.5) antibodies as described (Holsti, M.A. & D.H. Raulet, J. Immun., 143:2514-2519 (1989)). Cells were reacted in the first stage with reagents specific for the al domain of K^b (28-13-3), theα3 domain of D^b (28-14-8S), or no antibody, followed in the second stage with FITC-conjugated goat anti-mouse IgG + M (KIrkegaard and Perry, Gaithersburg, MD), as described. Ten thousand cells were analyzed on an EPICS C flow cytometer equipped with a quartz flow cell.

To eliminate CD4⁺ cells, thymocytes were incubated with anti-CD4 (GK1.5) mAb and complement for 40 minutes. Viable cells were purified on Ficoll-Isopaque gradients and subjected to a second round of killing with mouse anti-rat K light chain (MAR18.5) mAb plus complement to eliminate residual CD4 cells with bound GK1.5. The viable cells were again purified on Ficoll-Isopaque gradients. To stain αβ TCR versus CD8⁺, the enriched cells were reacted with H57-597-biotin followed in a second step with allophycocyaninstreptavidin (Becton Dickinson) and anti-CD8-FITC. To stain γδ TCR versus CD8, enriched cells were first reacted with UC7-13D5 culture supernatant, followed in a second step by goat-anti-hamster IgG-phycoerythrin (reagent adsorbed with rat and mouse IgG from Caltag, South San Francisco, CA).

After washing, the cells were incubated with rat- IgG to ensure there were no free rat Ig-binding sites, and subsequently reacted with anti-CD8-FITC. In all cases the negative controls using all reagents except the

TCR-specific first reagents, gave insignificant numbers of positive cells. Cursors were set based on the

negative control samples. Dead cells were excluded based on forward and 90 degree light scatter characteristics.

One hundred thousand cells were analyzed on a FACSTAR or 3 × 10⁴ cells on an EPICS C flow cytometer.

The most striking phenotype of the β2-m mutant mice is the virtually complete absence of the mature _β ⁺ CD4-8⁺ T cell subset, in both the thymus and peripheral lymphoid organs. These data argue that encounters with class I MHC molecules are essential for the differentiation of the CD4-8⁺ T cell subset, consistent with the conclusions drawn from studies of transgenic mice expressing a defined αβ TCR in all their T cells, as well as mice treated from birth with anti-class 1 MHC antibodies. Presumably, interaction with class I molecules induces CD4⁺8⁺ T cells to differentiate into CD4-8⁺ αβ T cells, or rescues newly differentiated CD4 8 T cells from programmed cell death. Conversely, the results indicate that differentiation of the other defined thymocyte subsets does not require interactions with class I molecules. Included in the latter category are CD4⁺8⁺ αβ^dim thymocytes, CD4⁺8- mature T cells, and

CD4-8⁺ αβ -thymocytes, the latter thought to represent an intermediate between CD4-8- thymocytes and CD4⁺8⁺ thymocytes.

As expected from the absence of mature CD4-8+ T cells, the CTL responses in homozygous mutant mice were abolished. The animals, when kept under pathogen-free conditions, appear to be healthy. It will, however, by interesting to study the role of MHC class I antigens in the response to viral or other infections or in tumorigenesis by exposing the animals to infectious agents or to carcinogens. Finally, because of the deficiency in class I dependent functions, homozygous mutant mice may serve as donors or recipients in transplantation experiments across histocompatibility barriers without the need for immunosuppressive regimens. EXAMPLE 3 Engraftment of Normal Mice with

Tissue from Mutant Mice

Assessment of Ability of MHC-I-Deficient Marrow to

Reconstitute Normal Mice

To assess whether hemopoietic cells from MHC-I- deficient mice can proliferate in irradiated normal mice, bone marrow cells from homozygous β2- m mutant (-/-) mice of the 129/Sv (129) strain were transplanted to groups of irradiated normal mice. Both C57B1/6 (B6) and 129 strains are of the H-2^b haplotype, although they differ in their Qa/T1 region gene composition. Resistance to marrow grafts is usually assessed by decreased incorporation of ¹²⁵IUdR into donor marrow-derived cells in the spleens of irradiated recipients 5 days after transfer, compared to transfers of syngeneic marrow (Cudkowicz, G. and J.H. Stimpfling, Nature 204:450-453 (1964); (Murphy,

W.J. et al., J. Exp. Med. 166:1499-1509 (1987)).

Strain 129 (+/-) and B6 (+/+) were irradiated (940 rads from a ¹³⁷Cesium source, 100 rads/min.) and then received intravenous inoculation of 5 × 10⁶ bone marrow cells from +/- or -/- 129 strain mice. A control group received no marrow cells. Five days later, the mice were inoculated intraperitoneally with 3 μCi of ¹²⁵IUdR. The following day, mice were sacrificed and incorporated isotope in the recipient spleens was determined,

following extensive rinsing of the spleens with PBS.

Results are shown in Figure 8a; data are presented as geometric means with the standard error of the mean. The number of recipients in each group is indicated (n). The results of such an experiment showed that -/- 129 marrow proliferated very poorly in either fully matched +/- 129 hosts or in H-2-matched B6 hosts. In contrast, marrow cells from +/- 129 mice proliferated at least 30-fold better in either host. Since +/- 129 marrow proliferated as well in B6 hosts as in syngeneic hosts (panel a), there are no 129 strain genes in the F2 mice that prevent marrow engraftment in B6 mice. (B6 x 129)F2 and F3 mutant mice were used for all subsequent experiments.

Irradiated (980 rads) B6 mice were inoculated iv with 2 × 10⁶ bone marrow cells from +/+ or -/- (B6 × 129)F2 mice. The bone marrow cells were depleted of T cells by treatment with anti-Thy-1 plus complement (Liao, N.-S. et al., J. Exp. Med., 170:135-143 (1989)). A control group received no marrow cells. 125 IUdR incorporation was determined as described above. Results are shown in

Figure 8b and were similar to those represented in Figure 8a. Thus, β2-microglobulin mutant marrow cells fail to proliferate significantly in irradiated MHC-matched hosts. The F2 and F3 mice were used in subsequent procedures.

Assessment of Survivalof Irradiated Mice Inoculated withBone Marrow or Fetal Liver

Recipient B6 mice were between 8 and 38 weeks old and of both sexes (no effect of sex was observed). They received 980 rads within hours of intravenous inoculation with bone marrow cells. They were maintained on antibiotic water for 1-2 days prior and 14 days after

irradiation and reconstitution, except for the animals for which results are represented in Figure 9d, where antibiotic water was first provided on the day of irradiation and reconstitution. Donors were (B6 ×

129) F2, F3, F4 and F2 × F3 animals which were genotyped by Southern blot analysis of tail DNA as described

(Zijlstra, M. et al., Nature, 342:435-438 (1989)). B6 recipients In Figure 9c were pre-depleted of NK1.1⁺ cells by intraperitoneal injection of 200 μg purified PK136 mAb, specific for NK1.1, two days before and again one day before irradiation and inoculation with fetal liver cells (Koo, G.C. et al., J. Immunol., 137:3742-3747

(1986)). The following is a brief description of methods and materials for each group for which results are presented in Figure 9:

Figure 9a. Irradiated groups of B6 mice (n = 6) were Inoculated with 5 × 10⁶ +/+ or -/- bone marrow cells, depleted of T cells, as described above, or with a mixture of 5 × 10⁶ of each.

Figure 9b: Irradiated B6 mice were inoculated with 7 × 10⁶ fetal liver cells from -/- (n = 9), +/+ (n = 3) or +/- (n = 6) donors of embryonic age 17.5 days (E17.5). Fetal liver cells from individual donors were inoculated into three recipients.

Figure 9c: B6 mice were pretreated with NK1.1-specific mAb (n = 7), or not (n = 6), prior to irradiation with 4 × 10⁶ -/- fetal (E17) liver cells. A control group received neither fetal liver cells nor antibody (n = 6). In separate experiments, B6 mice (n = 6) pretreated with irrelevant mAb rejected -/- marrow normally. Figure 9d: Irradiated +/- or -/- recipients (n = 6 each) were inoculated with 5 × 10⁶ -/- fetal (E16) liver cells.

Figure 9e: Irradiated BIO.BR recipients were inoculated with 5 × 10⁶ -/- (n - 5) or +/- (n - 6) fetal (E16) liver cells

According to published reports, a graft of approximately 10⁵ marrow cells protects 50% of syngeneic

recipients from lethal irradiation, resulting in survival and long term recons titution (Muller-Sieburg, C.E. et al., J. Exp. Med., 167:1825-1840 (1988)).

Mutant hemopoietic cells not only fail ed to p ro liferate in irradiated normal mice , they failed to protect these mice from the effects of lethal

irradiation. Irradiated B6 mice inoculated with 5 × 10⁶ mutant bone marrow cells (depleted of T cells) (Figure 9a) or 7 × 10⁶ mutant fetal liver cells (Figure 9b) died

8 to 16 days later. The same doses of hemopoietic cells from wild-type +/+ or heterozygous +/- littermates resulted in long-term survival of irradiated B6

recipients (Figures 9a and 9b). A titration experiment showed that all lethally irradiated B6 mice were

protected by 1.5 × a0⁶ +/- fetal liver cells, but 80% died with even 3 × 10⁷ -/- fetal liver cells. Also, 5 × 10⁶ -/- H-2^b fetal liver failed to protect MHC -mismatched BIO.BR mice from lethal irradiation, while these animals were protected by the same dose of +/- fetal liver cells (Figure 9e).

It is unlikely that T-cell-depleted bone marrow cells or fetal liver cells from the mutant mice exert a lethal graft versus host reaction against the recipients, because irradiated recipients that received a 1/1 mixture of -/- and +/+ marrow cells survived for at least 42 days (Figure 9a). Rather, these data suggest a failure in repopulatlon of the hemopoietic system of irradiated mice by mutant marrow or fetal liver cells. This interpretation is further supported by the finding that greater than 99% of the lymph node cells from irradiated mice that received a mixture of wild-type and mutant bone marrow cells 12 weeks earlier expressed cell surface K MHC-I molecules, indicating their wild-type origin.

Assessment of source of lymph node cells in these animals was carried out as follows: Lymph node cells were stained with Ml/42 monoclonal antibody as previously described (Zijlstra, M. et al., Nature 344:742-746

(1990)). Ml/42 reacts with K^b of the H-2^b MHC-I

molecules (Stallcup, K.C., et al., J. Immunol.

127:923-930 (1981)); positive staining requires β2-microglobulin expression (Zijlstra, M. et al., Nature

344:742-746 (1990)). Southern blot analysis of genomic DNA with a β2-m specific probe demonstrated that spleen and bone marrow cells, in addition to lymph node cells, were largely, if not entirely, of wild-type origin. Not only the lymphoid (LN) compartment, but also the spleen and bone marrow compartments are virtually entirely of +/+ origin.

Previous studies have demonstrated that the pretreatment of mice with a monoclonal antibody reactive with the NK1.1 marker present on approximately 3% of normal spleen cells eliminates NK1.1⁺ cells from the mice and prevents rejection of allogeneic bone marrow transplants (Murphy, W.J. et al., J. Exp. Med., 166 :1499-1509 (1987); Ohlen, C. et al., Science, 246:666-668 (1989)); (Koo, G.C. et al., J. Immunol. 131 :3742-3747 (1986)).

As shown in Figure 9c, pretreatment of B6 mice with anti-NK1.1 monoclonal antibody before irradiation and inoculation with -/- fetal liver cells prevented subsequent mortality, while control recipients not pretreated with antibody died 7-12 days after transplantation. Most cells repopulating the lymph nodes (>80%), Figure 9b, spleen and bone marrow of anti-NK1.1-pretreated animals were of mutant origin. These data suggest that there is no intrinsic defect in the capacity of -/- stem cells to repopulate the hemopoietic system, but rather that normal MHC-matched mice reject the mutant cells. The simplest interpretation is that host NK1.1⁺ effector cells "recognize" and destroy MHC-I deficient donor cells. These effector cells are either absent or inactive in -/- mice, since irradiated -/- mice survived following transplantation of -/- fetal liver cells (Figure 9d) . Irradiated -/- mice also survived following transplantation of +/+ fetal liver cells.

While the rejection of wild-thpe allogeneic hemopoietic cells can be overcome by 5 × 10⁶ donor cells, most recipients die with even 3 × 10⁷ -/- cells. This suggests that rejection of MHC- deficient cells is more rigorous than rejection of allogeneic cells.

Two conclusions can be drawn from this experiment. First, there is no intrinsic defect in the capacity of -/- stem cells to repopulate the hemopoietic system.

Secondly, normal MHC-matched mice reject -/- hemopoietic cells, and the rejection required the participation of an NKl.l cell. The simplest interpretation is that host NKl.l effector cells "recognize" and destroy MHC-deficient donor cells. These effector cells are either absent or inactive in -/- mice, since irradiated -/- mice survived following transplantation of -/- fetal liver cells (Figure 9d).

Irradiated BIO.BR (H-2^k) mice have been shown to reject allogeneic H-2^b marrow grafts, as determined with the ¹²⁵IUdR incorporation assay (Cudkowicz, G. and M.

Bennett, J. Exp. Med. 134: 83-102 (1971)). But as shown in Figure 9e, a graft of 5 × 10⁶ +/- H-2^b fetal liver cells protected BIO.BR recipients from lethal Irradiation, suggesting that the capacity of mice to reject allogeneic hemopoietic cells can be overcome. In contrast, the capacity of mice to reject MHC-deficient fetal liver cells cannot be overcome with similar or even larger doses of fetal liver cells, since MHC-matched (Figure 9a, 9b) as well as MHC-mismatched (Figure 9e) recipients died after transplantation of -/- fetal liver cells. These results suggest that rejection of MHC-deficient hemopoietic cells is more vigorous than that of allogeneic, MHC⁺ hemopoietic cells. If a common effector cell type is responsible for rejection of both types of hemopoietic cells, these considerations may indicate that there is a higher frequency of precursor cells reactive with MHC-defIcient cells, or that individual effector cells react better with MHC-deficient cells. Alternatively, it is possible that the two types of rejection are mediated by distinct effector cell types within the NK1.1⁺ subset. The complexity of the data on marrow rejection has led to controversy on several issues, including the precise nature of the effector cell. Effector cells mediating allogeneic and hybrid resistance show many similarities with natural killer (NK) cells, including their ontogeny, radiosensitivity and shared NK1.1⁺ phenotype (Murphy, W.J. et al., J. Exp. Med. 166:1499-1509 (1987); Yankelevich, B. et al., J. Immunol.

142:3423-3430 (1989); Ohlen, C. et al., Science 246: 666-668 (1989); Kiessling, R. et al ., Eur. J. Immunol. 7:655-663 (1977); Murphy, W. et al., J. Exp. Med. 165:1212 - 1217 (1987)). However, there is no evidence that the

specificity of NK cells from a particular strain corresponds to the specificity of allogeneic or hybrid resist- ance. Rather, NK cells lyse particular tumor target cells, with no evidence of MHC-allele specificity

(Herberman, R.B. and J.R. Ortaldo, Science 214:24-30

(1981)). Furthermore, recent studies (Yankelevich, B. e t al., J. Immunol. 142:3423-3430 (1989); Dennert, G. et al., Immunogenetics 31 :161-168 (1990)) suggest that allogeneic resistance is mediated by a fraction of NK1.1⁺ cells that express αβ T cell receptors (about 25% of splenic NK1.1⁺ cells in B6 mice). Yet these latter results fail to account for the finding that T- and B cell-deficient SCID mutant mice reject allogeneic marrow grafts (Murphy, W. et al., J. Exp. Med. 165:1212-1217

(1987)). Thus, the relation of NK cells and the effector cells mediating marrow rejection requires further study.

These results show that NK1.1⁺ cells are also required for rejection of MHC-1-deficient marrow. Furthermore, the rejection of MHC-I-deficient marrow, like that of allogeneic or parental marrow, is radio- resistant. These considerations raise the possibility that rejection of MHC- I-deficient marrow is mediated by the same effector mechanism that mediates rejection of allogeneic and parental marrow. In this light, it is worth considering models to account for bone marrow rejection by irradiated mice.

Cudkowicz, Bennett and their co-workers have proposed that hemopoietic cells express target antigens encoded by allelically variable recessive genes (Hh-1 genes) within the MHC, but distinct from conventional MHC genes (Cudkowicz, G. and E. Lotzova, Transplant. Proc. 4:1399 (1973); Rembecki, R.M. et al., J. Immunol.

138:2734-2738 (1987)). On the other hand, Karre and co-workers have proposed a "missing self" model of marrow rejection, in which marrow cells that fail to be recognized by putative self-MHC-I-specifIc receptors on NK1.1⁺ effector cells are rejected (Ohlen, C. et al., Science 246:666-668 (1989)). The effector cells in this model would presumably bind to hemopoietic cells via receptors other than the MHC-I-specIfic receptors, yet reject only those cells which fail to engage their MHC-I-specific receptors. This model was proposed to account for a variety of observations, including hybrid resistance and the striking finding that B6 mice transgenic for the D MHC-I gene reject B6 marrow grafts. A third model is that MHC-I proteins on the cell surface mask non-MHC-I antigens that serve as targets for marrow draft rejection. The "masking" model has been proposed in a different context ( S to rkus , W . J . e t al . , J. Immunol .

138:1657-1659 (1987); Shimuzu, Y. and R. DeMars, Eur._J. Immunol 19:447-451 (1989); Sturmhofel, K. and G.J.

Hammerling, J. Immunol. 20 : 171-177 (1990)), to account for the findings that the loss of MHC-I-expression by tumor cell lines augments their sensitivity to NK cells (Storkus, W.J. e t al., J. Immunol. 138:1657-1659 (1987); Shimuzu, Y. and R. DeMars, Eur. J. Immunol. 19:447-451 (1989); Sturmhofel, K. and G.J. Hammerling, J. Immunol. 20:171-177 (1990); Harel-Bellan, A. et al., Proc. Natl. Acad. Sci. USA 83:5688 (1986); Karre, K. et al., Nature, 319:675-678 (1986)); note, however, that the "missing-self" model can also account for these findings.

The data on rejection of MHC-I-deficient marrow presented herein is most consistent with either the

"missing-self" model or the "masking" model. Either of these mechanisms may have evolved to eliminate variant cells that lose MHC-I expression by mutation or as a result of viral infection (Schrier, P.I. e t al., Nature 305:771-775 (1983)), and thereby evade conventional CTL immunity (Ohlen, C. e t al., Science 246: 666-668 (1989); Karre, K. et al., Nature, 319:675-678 (1986); Hoglund, P. et al., J. Exp. Med. 168:1469-1474 (1988)).

It is of interest that -/- mice fail to reject their own marrow, as evidenced by their lack of hemopoietic deficiencies as well as the experiment presented in

Figure 9d. One interpretation of this observation is that the definition of "self" with respect to marrow transplantation is determined by the environment in which the effector cells mature, by a tolerance-inducing process and/or positive selection process. It is also possible that the differentiation of the effector cells required for marrow transplantation requires developmental Interactions with MHC-I molecules, similar to the earlier finding that -/- mice fail to develop mature CD8+ T cells (Zijlstra, M. et al., Nature 344:742-746 (1990)). It will be of interest to evaluate the role of MHC-I expression in the development NK cells and the capacity to reject allogeneic marrow transplants. A complete understanding of the phenomenon reported here will likely aid In the understanding of NK cell function as well as in therapeutic bone marrow transplantation in humans, where a similar mechanism of bone marrow rejection may be operative.

Claims

1. Purified or substantially pure MHC class I deficient cells.

2. Purified or substantially pure MHC class I deficient cells of Claim 1 selected from the group consisting of:

a) MHC class I deficient fibroblasts;

b) MHC class I deficient myoblasts;

c) MHC class I deficient endothelial cells;

d) MHC class I deficient lung epithelial cells; e) MHC class I deficient retinal cells;

f) MHC class I deficient keratinocytes;

g) MHC class I deficient hepatocytes;

h) MHC class I deficient neural cells; and

i) MHC class I deficient pancreatic cells.

3. Purified or substantially pure MHC class I deficient cells of Claim 1 or Claim 2 transfected with at least one gene encoding a therapeutic product.

4. Purified or substantially pure MHC class I deficient cells of Claim 3 wherein the gene encoding a

therapeutic product is selected from the group consisting of:

a) an insulin gene;

b) a Factor VIII gene;

c) a Factor IX gene;

d) a growth hormone gene; and

e) an erythropoietin gene.

5. Use of the MHC class I deficient cells of Claim 4 to produce a delivery system for the gene encoding a therapeutic product.

6. Purified or substantially pure MHC class I deficient cells which do not express MHC class II protein.

7. Purified or substantially pure MHC class I deficient cells, according to Claim 1 or Claim 2, for use in therapy.

8. Purified or substantially pure MHC class I deficient cells, according to Claim 3 or Claim 4, for use in therapy.

9. Purified or substantially pure MHC class I deficient cells, according to Claim 3 or Claim 4, for use in transplanting to an individual to provide a delivery system for the therapeutic product.

10. Use of purified or substantially pure MHC class I deficient cells for the manufacture of a sheet of said cells for transplanting to the body of an individual.

11. Use of purified or substantially pure MHC class I deficient cells transfected with a gene encoding a drug, an enzyme or a hormone for the manufacture of a sheet of said cells for transplanting to the body of an individual.