SCID Mouse Models of Human Stem Cell Engraftment

Abstract

The discovery of the severe combined immunodeficiency (scid) mouse mutation has provided a tool for establishment of small animal models as hosts for the in vivo analysis of normal and malignant human pluripotent hemopoietic stem cells. Intravenous injection of irradiated scid mice with human bone marrow, cord blood, or G-CSF cytokine-mobilized peripheral blood mononuclear cells, all rich in human hemopoietic stem cell activity, results in the engraftment of a human hemopoietic system in the murine recipient. This model has been used to identify a pluripotent stem cell, termed “scid-repopulating cell” (SRC) that is more primitive than any of the hemopoietic stem cell populations identified using the currently available in vitro methodology. In this review, we describe the development and use of this model system, termed Hu-SRC-SCID, and summarize the discoveries that have resulted from the investigation of human stem cells in this model. Finally, we detail the recent extension of the original Hu-SRC-SCID model system based on the C.B-17-scid mouse as the murine host to the Hu-SRC-NOD-SCID model based on the NOD-scid mouse as the host. The engraftment of human stem cells in the Hu-SRC-NOD-SCID model is enhanced over that observed in the Hu-SRC-SCID model and results in exceptionally high levels of human hemopoietic cells in the murine recipient. Future directions to further improve the Hu-SRC-NOD-SCID model system and the potential utility of this model in the preclinical and diagnostic arenas of hematology and oncology are discussed.

scid mouse, Hemopoiesis, Stem cell, Differentiation, Animal model, scid repopulating cell (SRC), Hu-SRC-SCID

Introduction

The true functional measure of a long-term renewable stem cell is the capacity to engraft myeloablated recipients, repopulate their hemopoietic systems, and sustain long-term multi-lineage hemopoiesis in vivo [1,2]. This in vivo function is the “gold standard” definition of a true pluripotent hemopoietic stem cell (PHSC) and is the definition we will use throughout this review. The best characterized in vivo assays developed to quantify PHSC have utilized murine model systems. The definitive assay system in murine models is a quantitative competitive repopulation assay between phenotypically distinguishable stem cell populations [1-4]. In this system, mixtures of bone marrow cells from different donors are injected into irradiated recipients, and the relative proportion of each donor-origin population is determined as a ratio. This ratio represents the relative PHSC content of each inoculum.

Quantitative analyses of human PHSC have historically been limited to in vitro assays where the proliferative potential of stem cells is evaluated in the presence of various combinations of cytokines. The closest surrogate in vitro assays for the primitive in vivo repopulating human stem cells are the colony-forming blast (CFU-blast), the high-proliferative potential colony-forming cell (HPP-CFC) assays, and several stromal-based assays, including the long-term culture-initiating cell assay (LTC-IC) and cobblestone-area-forming assays [3, 5-8]. Even though the “extended long-term culture-initiating cell system” (ELTC-IC) recently described [9] approaches detection of a primitive progenitor population that may be functionally closer to discriminating human long-term repopulating stem cells, the available in vitro assays do not appear to reflect a true measure of human PHSC.

Over the last 30 years, a number of investigators have attempted to use animals as hosts for quantitative study of the development and differentiation of human pluripotent hemopoietic stem cells. These efforts have included transplantation of bone marrow into heavily irradiated mice [10], into mice homozygous for the nude (Hfh11^nu, hereafter referred to as nu) gene [11-13], and into fetal sheep [14, 15]. In all cases, the levels of human stem cell engraftment were low. The utility of the murine models has been limited by available methodology, while time constraints and high cost limit the utility of the sheep model.

Need for Better Small Animal Models of Human Hemopoiesis

In order to enhance the study of human PHSC, investigators have directed their attention to the development of small animal models of human hemopoiesis. The advantages of animal models, particularly small animal models, are obvious. The development, differentiation, and long-term repopulating capacity of human cells, which can only be determined in vivo, can be ascertained in a small animal model without the need for clinical studies. The animals can be biopsied and autopsied at predetermined points for analyses, and treatments of hemopoietic stem cells in vivo can be performed without concern that the manipulation might directly affect the patient's well-being. Small animals can also be manipulated to express human growth hemopoietic factors. Treatments for hemopoietic-based disorders can be screened and evaluated for in vivo efficacy, and the establishment of optimal conditions for in vitro manipulation and expansion of PHSC can be easily evaluated. The development and validation of a small animal model of human hemopoiesis would also permit rapid evaluation and application of advances in gene therapy for the treatment of hematologic disorders, AIDS, and other diseases.

Recent advances in mammalian genetics have provided a number of immunodeficient murine models for engraftment and quantitation of human stem cells. One model is mice which are triply homozygous for the beige (Lyst^bg, hereafter referred to as bg), nu, and X-linked immunodeficiency (Btk^xid, hereafter referred to as xid) [16-19] loci. Additional immunodeficient models include mice deficient in the recombination activating gene-1 (Rag1) [20] and Rag2 [21] genes and mice homozygous for the Prkdc^scid (hereafter referred to as scid) locus [22]. Although attempts have been made to utilize the bg-nu-xid mouse [16-19] and Rag1 and Rag2-deficient mice [23] as hosts for normal and malignant human hemopoietic cells, the optimal host has been determined to be mice homozygous for the scid mutation.

In 1988, the first reports of the engraftment of human cells into homozygous scid mice appeared [24, 25]. These reports included engraftment of human peripheral blood mononuclear cells following intraperitoneal injection into unirradiated recipients [24] and transplantation of fetal bone marrow and thymus fragments under the renal capsule [25]. These initial model systems were soon followed by the description of human bone marrow engraftment when transferred intravenously into irradiated scid recipients [26]. A number of investigators have now reported on the utilization of immunodeficient scid mice to investigate human hemopoiesis, and in particular, in vivo function of human PHSC. In this review, we will focus on the use of immunodeficient scid mice as hosts for in vivo evaluation of human PHSC activity.

Discovery and Characteristics of scid Mice

The scid mutation was first described in 1983 in C.B-17 strain mice [22]. C.B-17 mice are BALB/c mice congenic at an immunoglobulin heavy chain locus derived from C57BL/Ka mice [27]. Recently, the catalytic subunit of a DNA-dependent protein kinase encoded by a gene termed “protein kinase, DNA activated catalytic polypeptide” (Prkdc) has been identified as the gene disrupted by the scid allele [28-32]. The Prkdc^scid gene encodes a nonsense mutation which causes the insertion of a termination codon [33].

Mice homozygous for the scid mutation lack both humoral and cell-mediated immunity due to the absence of mature T or B lymphocytes [34]. The lack of an adaptive immune system in scid mice results from their inability to express rearranged antigen receptors [35, 36]. This defect is due to a failure to activate a DNA recombinase enzyme that requires a functional Prkdc gene [37]. Despite this defect, some long-lived antigen-receptor rearrangements do occur and result in a phenomenon described as “leakiness” [38-40]. As C.B-17-scid mice age, the proportion of mice exhibiting functional rearrangements in lymphocytes may reach 90% [40].

Three well-characterized small animal models of human stem cell engraftment based on the use of scid mice have been described [25, 26, 41]. In all cases, these basic model systems have utilized C.B-17-scid mice as hosts for human stem cells.

SCID-hu Mice

In the first described model of human stem cell engraftment in scid mice, human fetal liver or fetal bone marrow with or without human fetal thymus tissue was transplanted under the renal capsule of unirradiated C.B-17-scid mice (termed SCID-hu mice) [25]. The human cells were observed to survive and maintain many of the characteristics of normal human marrow in the scid host. Low levels of human cells were detectable in the marrow or peripheral tissues of the scid mouse hosts [25]. Investigators have used SCID-hu mice as model systems for the study of human stem cell phenotype and differentiation [42-52], human gene therapy protocols [53-55], human T-cell differentiation [56], and homing of human myeloma cells to the bone marrow [57]. SCID-hu mice, in combination with cotransplanted thymic fragments, have been used to investigate the primitive hemopoietic stem cell that populates the T-cell lineage [42, 58], for studies of human T-cell function [59, 60], and for infection of human T cells and thymus with HIV [61, 62]. Although an important experimental system, human fetal tissues are not readily available and impose a number of ethical concerns and constraints that limit their widespread use among laboratories. In addition, it takes several months for chimera formation to be detectable. Numerous reviews on this model system are available to the reader for further information [49, 61, 63].

Triple-Chimeric (Trimera) Mice

Soon after the discovery of scid mice, it was shown that lethally irradiated normal (immunocompetent) mice transplanted with C.B-17-scid marrow, followed later by a large dose of human bone marrow, resulted in low levels of human cell engraftment [41]. This model has been used to study human stem cell phenotype and differentiation [41], as well as the growth of human leukemias and lymphomas in vivo [64]. However, in these irradiated, scid marrow-engrafted, human marrow-injected mice, the role of natural antibodies derived from the irradiated immunocompetent host remains a potential obstacle [65]. In addition, the severe conditioning regimen and high doses of human marrow required for engraftment have precluded many studies of human stem cells in this model.

Intravenous Injection of Human Stem Cells into Irradiated scid Mice

The third model that has been well characterized is intravenous injection of human stem cells into irradiated scid mice [66-68]. This model system has been used to study human stem cells in vivo in a number of laboratories and is the focus of the remainder of this review.

Irradiated scid Mice Injected Intravenously with Human Stem Cells

Engraftment of Human Hemopoietic Stem Cells in C.B-17-scid Mice

The ability of intravenously injected human bone marrow cell suspensions to engraft in sublethally irradiated C.B-17-scid mice was first documented in 1992 [26]. Human cell engraftment levels were low and represented only a small percentage of the total bone marrow compartment of the host. Differentiation of the stem cells into mature progeny required administration of exogenous human cytokines, including stem cell factor (SCF), a fusion protein of interleukin 3 (IL-3) and GM-CSF termed PIXY321, and erythropoietin (EPO) [26]. The term “scid-repopulating cell” (SRC) has recently been coined to describe the human stem cell that engrafts in irradiated scid mice following intravenous injection [67, 69], and hu-SRC-SCID is the term we will use throughout the remainder of this review to describe this model system.

The hu-SRC-SCID system has been used to study many aspects of the human hemopoietic system [26, 70], to investigate growth of human leukemias in vivo [71-75] and to analyze heritable hemopoietic disorders [76]. The hu-SRC-SCID system has also been extended to investigate human stem cells present in tissues other than bone marrow. Human umbilical cord blood [66] has also been analyzed in this system. In these studies, cord blood engrafts in C.B-17-scid mice without the need for exogenous cytokines. This enhanced engraftment of cord blood stem cells was postulated to be due to the presence of immature T cells in the inoculum that provided a source of human cytokines [66].

Limitations of the hu-SRC-SCID Model

Two major limitations have precluded the widespread use of the hu-SRC-SCID model system. First, the engraftment levels of human cells are low, representing only 0.5%-5% of the total scid recipient marrow population [26]. Second, human stem cells cannot utilize many of the murine hemopoietic growth factors produced by their hosts due to the lack of species cross-reactivity of these cytokines. This results in a cytokine-deficient microenvironment that fails to support human stem cell proliferation and differentiation [26]. To attempt to overcome these obstacles, two approaches have been implemented.

First, it was reasoned that replacement of the critical human-origin cytokines required by the human stem cells would result in increased human cell engraftment in the hu-SRC-SCID system. To replace these human cytokines in the murine hosts, recent advances in deciphering the gene sequence of a number of human cytokines were combined with transgenic technology. BALB/c mice transgenic for human IL-3, GM-CSF, and SCF were generated [77]. These mice were irradiated and injected with human cord blood cells. However, even in these human cytokine transgenic mice, the engraftment levels of human cells remained low, and detectable numbers of human cells were observed in only ∼50% of the irradiated transgenic BALB/c mice injected with human cord blood cells [77].

The second approach has been based on the hypothesis that the poor engraftment of human lymphohemopoietic cells in scid mice is due to host innate immune resistance. Thus, innate immune function in scid mice as well as the absence of critical human cytokines required for human stem cell expansion and differentiation appear to be constraining the growth of human stem cells. Since an adaptive immune system is completely lacking in scid mice, the innate immune system may have increased importance in rejection of hemopoietic xenografts. This possibility has gained support following analysis of the innate immune system of C.B-17-scid mice. In contrast to the absence of functional T or B lymphocytes, C.B-17-scid mice have been documented to have an enhanced innate immune system [39]. The scid mutation in C.B-17 strain mice does not impede myeloid, erythroid, or natural killer (NK) cell development. These mice have elevated levels of hemolytic complement, granulocyte and macrophage function appear normal, and increased levels of NK cell activity are observed.

To study the possible role of innate immunity as the likely source of host-mediated resistance to xenogeneic hemopoietic cell engraftment [78-80], some investigators have focused on the NK cell population. To eliminate NK cells, scid mice have been treated with anti-asialo-GM1 antibody to deplete host NK cells prior to administration of human cells [81-84]. This approach has failed to increase dramatically the engraftment levels of human peripheral blood lymphocytes in C.B-17-scid mice, and if coupled with irradiation, shortens the survival of the murine scid recipient to only a few weeks after treatment [85].

An alternative approach has been implemented to study the role of the host background, and the host innate immune system, in resistance of scid mice to human cell engraftment. We have utilized the vast resource of known innate immune defects in various strains of inbred mice available at The Jackson Laboratory, Bar Harbor, ME. Many of these inbred strains of mice have been backcrossed with C.B-17-scid mice to generate new inbred strains of mice congenic at the scid locus. A number of these inbred strains of scid mice have been characterized and tested for their ability to support engraftment with human lymphohemopoietic cells.

Effect of Host Strain Background on Human Lymphohemopoietic Cell Engraftment in scid Mice

As a rapid screen to evaluate murine background gene effects on human cell engraftment, we first injected experimental mice with human peripheral blood mononuclear cells (PBMC). C.B-17-scid, C3H/HeJ-scid, C57BL/6J-scid, NK1.1-depleted C57BL/6-scid, DBA/2-scid, and nonobese diabetic (NOD)-scid mice were evaluated [86]. For these screening experiments, unirradiated hosts were injected intraperitoneally with 20 × 10⁶ human PBMC obtained from an individual leukapheresis donor. Each inbred strain of scid mice was deficient in one or more innate immune components. C3H/HeJ-scid mice were evaluated based on their known defects in macrophage responsiveness to lipopolysaccharide (LPS) [87]. C57BL/6-scid mice were evaluated based on their widespread use as a standard, prototypic mouse strain, and served as a normal control with which to compare the C.B-17-scid strain. C57BL/6-scid mice also express the antigen NK1.1 [88], and anti-NK1.1 antibody-injected C57BL/6-scid mice were tested to determine the potential role of NK cells in scid resistance to human cell engraftment. DBA/2-scid mice were tested based on their lack of the complement component C5, rendering these mice deficient in hemolytic complement [89]. NOD-scid mice appear to express multiple defects in innate immunity (see below).

We observed that levels of human PBMC engraftment in NOD-scid mice were always 5- to 10-fold higher than in any of the other genetic stocks of scid mice examined [86]. Similar results were obtained using human spleen cells as the donor population. Engraftment levels in NOD-scid mice were 5- to 10-fold higher than those observed in C.B-17-scid mice [90].

These results demonstrated that individual ablation of NK cell activity (NK-depleted C57BL/6J-scid) mice, loss of hemolytic complement (DBA/2J-scid mice), or defects in macrophage activation (C3H/HeJ-scid mice) failed to increase engraftment levels of human cells to those observed in NOD-scid mice [86]. These results focused our research efforts on the NOD-scid strain of mice to determine the basis for the increased support of human cell engraftment and to extend the hu-SCR-SCID model system to a hu-SCR-NOD-SCID system.

hu-Scr-Nod-Scid Model For Human Stem Cell Engraftment

Innate Immune System of NOD-scid Mice

The NOD mouse is an animal model of spontaneous autoimmune T-cell-mediated insulin-dependent diabetes mellitus (IDDM) [91]. Because of the interest in the immune basis for this disorder, the adaptive and innate immune systems of NOD strain mice have been studied intensely. It has been reported that inbred NOD strain mice have multiple defects in innate immunity. They are deficient in NK cell activity [91, 92], display defects in myeloid development and function [92-95], and are one of the ∼30% of inbred strains that lack C5 and cannot generate either the classical or alternative pathways of hemolytic complement activation [27, 96].

The NOD/LtSz-scid strain was generated by crossing the scid mutation from C.B-17-scid mice onto the NOD background. NOD/LtSz-scid mice lack an adaptive immune system; due to the absence of T cells, they do not develop autoimmune IDDM and remain insulitis- and diabetes-free throughout life [39]. However, many of the innate immune defects present in the parental NOD/Lt stock of mice are also expressed in the NOD-scid strain of mice [39].

NK Cell Activity

C.B-17-scid mice have higher NK cell activity than C.B-17 wild-type mice, presumably as compensation for the absence of a protective adaptive immune system [39]. This compensatory effect is also observed in the C57BL/6-scid strain of mice; these mice have higher levels of NK cell activity than do parental C57BL/6 wild-type strain mice [88]. In contrast to these normal strains of mice, NOD/LtSz-scid strain mice have low levels of NK cell activity, and the presence of the scid mutation does not elevate NK activity to the high levels observed in other strains of mice [39]. The NK cell activity is extremely low in NOD/LtSz-scid mice, and pre-treatment with the interferon inducer, polyinosinic acid-polycytidylic acid (poly I:C), is required to detect NK cell activity. The NK cell activity in NOD/LtSz-scid mice induced by poly I:C is unaffected by treatment with anti-NK1.1 monoclonal antibody, demonstrating that these mice are NK1.1-negative [39].

Myeloid Development and Function

In the parental stock of NOD strain mice, myeloid development and function are abnormal and can be identified by a reduced ability of their macrophages to secrete IL-1 in response to LPS [92-95]. To determine if this innate immune defect is also present in NOD/LtSz-scid mice, IL-1 secretion in response to stimulation with LPS was determined [39]. As with the parental NOD/Lt stock of mice, NOD/LtSz-scid mice are deficient in their IL-1 response to stimulation to LPS [39]. In contrast, macrophages from C.B-17-scid mice readily produce IL-1 in response to LPS stimulation. This observation confirms that one of the defects in the myeloid lineage of NOD/Lt mice is also present in NOD/LtSz-scid mice and likely represents a more generalized defect in myeloid development and function in these mice [92-95].

Hemolytic Complement Deficiency

The alternative pathway of complement does not require antibody fixation for activation and hemolytic activity and is an important component of innate immune activity [97]. However, ∼30% of mouse strains are genetically deficient in C5, which is part of the membrane attack complex and is required for the hemolytic activity of both the classical and alternative complement pathways. The C5 component of complement maps to the hemolytic complement (Hc) locus on chromosome 2 [98]. The scid mutation on C.B-17-scid mice maps to chromosome 16 [99]. Placement of the Chr 16 scid locus onto the NOD strain background did not transfer the functional Hc locus to the NOD-scid strain. Consequently, NOD/LtSz-scid mice lack a detectable hemolytic complement, while C.B-17-scid mice display elevated levels of hemolytic complement activity as compared to C.B-17 wild-type mice [39].

Additional Characteristics of NOD/LtSz-scid Mice

A major concern when using scid strain mice for studies where the presence of an adaptive immune system would be detrimental is the occurrence of a phenomenon termed “leakiness.” First described in 1988 in C.B-17-scid mice [40], productive rearrangement of T- and B-cell antigen receptors occurs spontaneously in C.B-17-scid mice, and serum immunoglobulin as well as T cell receptor-positive lymphocytes can be detected in the circulation of many of these mice. This phenomenon increases with age of the scid mouse, and can be influenced by the environment in which the animals are maintained [38, 40]. To determine the propensity of NOD/LtSz-scid mice to develop leakiness, mice were aged and analyzed for the presence of serum immunoglobulin. Less than 10% of NOD/LtSz-scid mice develop detectable levels of circulating immunoglobulin by 200 days of age [39]. In contrast, ∼90% of C.B-17-scid mice develop detectable serum immunoglobulin by this age. Older NOD/LtSz-scid mice are also able to retain MHC class I-disparate allogeneic skin grafts for several months, demonstrating the lack of functional T cells in addition to the absence of B lymphocytes.

Two additional characteristics of NOD/LtSz-scid mice may relate to their enhanced ability to be engrafted with human hemopoietic stem cells. First, there is an approximate twofold reduction of bone marrow cell counts in NOD/LtSz-scid mice as compared with NOD/Lt strain mice. This may increase the available “niches” in the marrow for human stem cells. Second, NOD/LtSz-scid mice have a slight reduction in erythrocyte counts but an increase in erythrocyte mean cell volumes. This results in the expression of a mild macrocytic anemia, even though hematocrits are normal [39].

Thymic Lymphomagenesis

An additional characteristic that would be important in a small animal model of human hemopoiesis is the ability to assess the long-term repopulating capacity of the transferred human stem cell population. However, NOD/LtSz-scid mice have been found to have a mean life span of only ∼8 months [39]. Upon transfer of the scid mutation from the C.B-17 strain onto the NOD strain, an unusually high incidence of thymic lymphomas was noted [100]. These thymic lymphomas occur in over 70% of NOD-scid mice by 40 weeks of age and are the cause of their early death [39, 100]. The development of these thymic lymphomas is imparted by the presence of the scid mutation, since NOD/Lt wild-type mice rarely develop thymic lymphomas [101]. Investigations into this phenomenon have led to the postulate that thymic lymphomas in NOD/LtSz-scid mice may result from the expression of a NOD mouse-unique endogenous ecotropic murine leukemia provirus locus [100]. The provirus locus has been mapped to the proximal region of chromosome 11 and is termed “Emv30” [100, 102]. It has been suggested that the Emv-30 provirus expression may synergize with the block in thymocyte development in scid mice, leading to activation of the NOD-unique Emv30. This hypothesis was tested by replacement of the proximal end of chromosome 11 with genetic material derived from the closely related NOR/Lt strain [103]. The new stock of mice termed Emv30^nullNOD/LtSz-scid mice develops thymic lymphomas at a significantly delayed rate but retains its ability to support the engraftment of human cells [103]. Utilization of this stock of NOD-scid mice may facilitate studies where long-term engraftment of human cells in scid mice is important.

Repopulation of NOD-scid Mice with Human Stem Cells

Based on the initial studies of human PBMC engraftment in NOD-scid mice, it was hypothesized that human stem cell engraftment would also be increased in NOD-scid recipients. Human umbilical cord blood transplantation was initially used to test this hypothesis [104]. Irradiated NOD/LtSz-scid mice injected intravenously with human umbilical cord blood were able to support 5- to 10-fold higher levels of human cell engraftment in their bone marrow than were C.B-17-scid mice. Human CD34⁺ cells were detectable in the bone marrow of NOD/LtSz-scid mice, and the engrafted human cells were also present in the peripheral circulation. Peak engraftment levels were observed at four to eight weeks after transplantation, although high levels of human cell engraftment in the marrow remained detectable at four months after transplantation. The engrafted cells phenotypically consisted of immature human stem cells and progenitor cells. Some differentiated cells were detected and were predominantly of the myeloid and B-lymphocyte lineages [104]. These observations set the stage for the widespread increase in the use of the hu-SCR-NOD-SCID model for studies of human hemopoiesis in scid mice and has quickly become the host strain of choice in most laboratories.

Characteristics of Stem Cells That Repopulate NOD-scid Mice

The initial report on the increased engraftment of human stem cells in NOD-scid mice was rapidly confirmed in a number of laboratories [105-107]. The enhanced ability of NOD-scid mice to support human stem cell engraftment was further demonstrated by comparing the levels of human cell engraftment to those observed in BALB/c-scid mice expressing transgenes for SCF, IL-3, and GM-CSF [77]. These studies also extended the initial observation of increased engraftment of human hemopoietic stem cells in NOD-scid mice to further identify the phenotypes of the engrafted human cells [105]. The scid mouse marrow analyzed four to eight weeks after injection of human cord blood cells contained human CD34⁺ cells, the majority of which were CD38⁺. However, the frequency of colony-forming cells in the CD34⁺ cell fraction and the frequency of LTC-IC recovered from scid marrow was low [105].

Two reports subsequently examined the phenotype of the human cord blood stem cell population that is capable of repopulating the hemopoietic system of NOD-scid mice [69, 107]. In both reports, the SRC was present in the CD34⁺ cell fraction and was absent in the CD34⁺CD38⁺ cell fraction. An important extension of these studies was the demonstration that human stem cells engrafted in NOD-scid mice generated human cells in secondary NOD-scid recipients [107]. This serial transfer experiment confirmed that the human CD34⁺CD38^– cells in the marrow of the primary NOD-scid recipients contained a functional stem cell population. Using limiting cell doses, it was calculated that ∼1 in 617 CD34⁺CD38^– human cord blood cells is an SRC capable of generating at least 4 × 10⁵ human cells in the recipients [69]. In this study, a significant number of these cells remained CD34⁺CD38^–, suggesting renewal as well as engraftment of human stem cells in the NOD-scid marrow.

Engraftment of human hematopoietic bone marrow stem cells is also enhanced in NOD-scid mice as compared to C.B-17-scid mice [67, 69, 106, 108]. The bone marrow stem cell capable of repopulating NOD-scid mice is present in the CD34⁺CD38^– marrow cell fraction, similar to that observed for cord blood stem cells [67]. These stem cells also generated significant numbers of human CD34⁺CD38^– cell populations in the marrow of the NOD-scid recipient and were able to generate multiple lineages of human cells. The study also clearly distinguished the in vivo SRC population from the cell populations that can generate CFC and LTC-IC in vitro [67]. In addition, the authors attempted to transduce the SRC using a retrovirus vector. The efficiency of transduction of the SRC was extremely low, while the CFC and LTC-IC cell fractions were readily transduced and were able to generate retrovirus-marked cells.

Additional studies of human stem cells in NOD-scid mice have yielded valuable information on the ability to culture functional human stem cells in vitro on human bone marrow stromal cells [108] or in serum-free conditions [109, 110]. In all cases, the in vitro culture of human stem cells resulted in lower levels of human cell engraftment in the NOD-scid recipient, irrespective of whether the human stem cell source was bone marrow or cord blood.

An important advance now made possible by the availability of the hu-SRC-NOD-SCID model is the ability to quantitatively assess primitive human stem cell frequency in mixed populations using an in vivo assay system. As noted above, the “gold standard” for the assessment of stem cell activity is the ability to repopulate the hemopoietic system of an irradiated recipient. These quantitative studies have now been initiated, and initial preliminary estimates of SRC frequency in various hemopoietic tissues have been determined using a limiting cell dose into NOD-scid mice [106]. SRC frequency was calculated to be approximately 1 per 1 × 10⁶ cord blood cells, 1 per 3 × 10⁶ bone marrow cells, and 1 per 6 × 10⁶ mobilized peripheral blood cells obtained from normal donors [106]. These results were obtained using the only presently available model system for quantitative in vivo assessment of this cell population, the hu-SRC-SCID model.

Application of the hu-SRC-NOD-SCID Model for the Study of Human T-Cell Lineage Acute Leukemia

A potential application of the hu-SRC-NOD-SCID model system is to study the growth of human leukemias in vivo, perhaps as an indicator of the future clinical outcome of the patient. To test this application, primary leukemia cells from patients with acute lymphoblastic leukemia were injected into both C.B-17-scid mice and NOD-scid mice [111]. The surprising observation was that cells from patients in “low-risk” groups failed to grow in C.B.-17-scid mice but readily grew in NOD-scid mice. Cells from “high-risk” patients readily grew in both strains of scid mice. These results suggest that the improved hu-SRC-NOD-SCID model may be too permissive for T-cell leukemia growth to allow the distinction between aggressive and nonaggressive leukemias in patients [111].

Overall, these recent reports have confirmed that levels of human cell engraftment, and more specifically, human stem cell engraftment, are consistently better when NOD-scid mice are used as recipients instead of C.B-17-scid mice (or BALB/c-scid mice transgenic for human growth factors).

Continuing Limitations of the hu-scr-nod-scid Model

For complete validation of the hu-SCR-NOD-SCID model as an assay for the most primitive human hemopoietic stem cell (PHSC), additional studies remain to be performed. Most importantly, the outcome of transplantation of human stem cells into NOD-scid mice needs to be correlated with the clinical outcome following transplantation of the same pool of human stem cells into human recipients. The present data suggest, however, that the SRC is either closely related with the true PHSC in humans, or is, in fact, the PHSC population capable of providing long-term repopulation of the entire hemopoietic system in humans. The results comparing the SRC with the precursor of in vitro detected CFC and LTC-IC suggest that the SRC is more primitive and exists at a lower frequency than do the CFC or LTC-IC precursors. The phenotype of the SRC, CD34⁺CD38^–, is similar to the phenotype of the human PHSC that can repopulate the hemopoietic system of irradiated individuals. Finally, the SRC appears to provide long-term engraftment of the recipient NOD-scid marrow (in our hands, over six months), and can generate human lymphohemopoietic cells upon secondary transfer. These are all characteristics of the stem cell that can repopulate human hemopoietic systems.

At the level of the recipient, there are still numerous deficiencies of the NOD-scid mouse that impede studies on human stem cells in vivo. The development of thymic lymphomas and death of most NOD-scid mice by eight months of age impede implementation of long-term engraftment studies past six to seven months [39, 100]. The scid mutation also imparts extreme radiosensitivity to the NOD-scid host [39, 112-116]. As little as 400 rads gamma-irradiation can result in hemopoietic insufficiency and death. The microenvironment in which the human stem cells engraft is currently unknown, but may be of predominantly murine origin. If so, the lack of appropriate cell-cell interactions due to non-species cross-reactivity, including interaction with appropriate adhesion molecules [117] or the lack of appropriate human cytokines [16], may continue to impede human stem cell engraftment and development. Finally, although the NOD-scid mouse exhibits multiple defects in innate immunity [39], it still displays low NK cell activity [39], disparate major histocompatibility loci, and intact cytolytic mechanisms such as functional expression of the perforin molecule in NK cells. These characteristics continue to limit the utility of even the improved hu-SRC-NOD-SCID model for the study of human stem cells in vivo in a small animal model.

Future Directions

As detailed above, it is first critical to establish a causal relationship between the human stem cell that repopulates NOD-scid mice and the stem cell that can repopulate human hosts. Detailed cell fractionation experiments, genetic marking and tracing experiments, and analysis of the differentiated progeny in both humans and in NOD-scid mice are required for this relationship to be established. It is clear from the available data, however, that the human stem cell that repopulates NOD-scid mice is more primitive than the populations that are detected in any of the currently available in vitro assays for human stem and progenitor cells. Pending the establishment of this relationship, the utility of the hu-SRC-NOD-SCID system will still provide valuable information on a very primitive, perhaps pluripotent, stem cell population present in humans. To further our understanding of this primitive population of cells, additional modifications of the murine recipient as described below may increase the utility of this model system.

Recent Advances in Development of New Immunodeficient Mice as Hosts for Human Stem Cells: Knockouts, Transgenics, and “Humanized” Immunodeficient Mice

Based on the documented differences in engraftment observed between NOD-scid mice and C.B-17-scid mice, it is reasonable to suggest that the genetic background of the murine host, in addition to the presence of homozygosity for the scid mutation, may influence engraftment and differentiation of transplanted human stem cells.

To provide immunodeficient recipients lacking major histocompatibility loci, the β-2 microglobulin null (β-2m^null) allele for β-2 microglobulin has recently been backcrossed onto the NOD-scid stock of mice [118]. These NOD-scid-β-2m^null mice lack expression of murine MHC class I due to the lack of β-2m required for MHC class I expression. In addition, the lack of MHC class I results in essentially undetectable levels of NK cell activity, much lower than the already low NK cell activity detected in NOD-scid mice [118]. Injection of human PBMC into these mice results in ∼5- to 10-fold higher levels of human cell engraftment than observed in NOD-scid mice, which are already 5- to 10-fold higher than observed in C.B-17-scid mice [86, 118]. Based on these results, we have recently examined the NOD-scid-β-2m^null mouse as a host for human cord blood cells. As was observed for engraftment of human peripheral blood mononuclear cells, higher levels of human cell engraftment were observed in the bone marrow of NOD-scid-β-2m^null mice than in NOD-scid mice. In addition, the engraftment of human cells could be accomplished at much lower doses of irradiation in NOD-scid-β-2m^null mice than in NOD-scid mice (unpublished observations). Two limitations are still notable in NOD-scid-β-2m^null mice. First, the incidence of thymic lymphomas is increased in frequency and occurs earlier in NOD-scid-β-2m^null mice than in NOD-scid mice, perhaps due to the almost complete deficiency of NK cell activity [118]. Second, NOD-scid-β-2m^null mice appear to be even more radiosensitive than are NOD-scid mice (unpublished observations).

To overcome these latter two limitations, the Rag1^null (LDS, unpublished data) and Rag2^null [119] alleles have been backcrossed onto the NOD/Lt stock of mice. These mice should develop thymic lymphomas at a decreased rate or frequency due to the effect of the Rag1^null allele at an earlier stage of thymocyte development than that of the scid mutation. The NOD-Rag1^null mice should also be much more radioresistant than the NOD-scid mice, since the Rag1 knockout is not known to interfere with DNA repair following gamma-irradiation. A second approach may be the in utero injection of human stem cells into NOD-scid mice. It has recently been observed that in utero injection of allogeneic cells into day 13-14 gestation NOD-scid fetuses results in engraftment of the allogeneic stem cells in the absence of irradiation [120]. If the NOD-Rag1^null mice support enhanced human cell engraftment similar to that observed in NOD-scid mice, or if human stem cells can be engrafted in utero without the need for irradiation, two of the major limitations of the hu-SRC-NOD-SCID model would be overcome.

Another limitation of the currently available hu-SRC-NOD-SCID model is the lack of species cross-reactivity between many of the cytokines that are essential for stem cell proliferation and differentiation. To overcome this limitation, the transgenes for human SCF, IL-3, and GM-CSF recently described in a transgenic BALB/c mouse [77] have recently been backcrossed onto the NOD-scid stock (unpublished observations). These mice should support increased levels of human stem cell engraftment, and the transgenes should provide at least some of the human cytokines essential for human stem cell development and differentiation. Additional NOD-scid mice under development include MHC class II^null mice, perforin^null mice, and NOD-scid mice transgenic for human MHC class I and class II molecules. The utility of these new stocks of immunodeficient NOD mice for human stem cell engraftment remains to be determined, but it is likely that continuing modification of the murine host will improve the already excellent hu-SRC-NOD-SCID model for human stem cells.

Potential Use of the Hu-SRC-NOD-SCID Model

The exciting applications that the hu-SRC-NOD-SCID model system now provides the hematologist, oncologist, transplantation immunologist, and biologists interested in autoimmunity are too numerous to detail. The utility of the hu-SRC-SCID model for study of normal and abnormal human stem cells and malignancies has already been demonstrated. The extension of these studies to the hu-SRC-NOD-SCID model should provide enhanced engraftment and differentiation of the human stem cell populations and facilitate their study in an in vivo environment previously not possible. This model system should allow detailed identification and characterization of the human pluripotent stem cell and prove readily applicable for in vivo analysis of gene therapy for genetic disorders such as sickle cell anemia and β-thalassemia which have been studied previously using the Hu-SRC-C.B-17-SCID model [76]. The extension of the hu-SRC-NOD-SCID model to studies of genetic therapy for somatic-based disorders such as adenosine deaminase deficiency has recently been reported [67] and has been shown to provide in vivo information on transduction of stem cells not currently possible using only in vitro methodology. The potential of the hu-SRC-NOD-SCID model as a diagnostic tool for hemopoietic-based malignancies and as a preclinical analysis of the stem cell activity in tissues prior to transplantation is just becoming realized. Extension of this model for establishment of hemopoietic chimeras to study transplantation tolerance and for investigation of the stem cell contribution to autoimmunity will provide additional potential avenues for clinical application. Continued improvement of the hu-SRC-NOD-SCID model system and the application of the system to the current perplexing problems facing the hematologist, oncologist, and immunologist should provide an in vivo small animal model useful in the clinical arena.

Acknowledgements

This work was supported by National Institutes of Health grants DK36024 (DLG), AI30389 (LDS), and AI38757 (RMH), and by a Diabetes Interdisciplinary Research Program grant from the Juvenile Diabetes Association International (DLG).