The earliest immunodeficient mice were nude mice with a FOXN1 gene defect preventing them from developing thymus. Therefore, they lack mature T cells and their associated immune rejection but possess B and natural killer (NK) cells, consequently rejecting human-derived cells. Later, the severe combined immunodeficient (SCID) mouse model was developed. This strain of mice is defective in T and B cells (14); hence human cells can be transplanted into it. The peripheral blood lymphocytes (PBL)-SCID and SCID-Hu mouse models we will discuss are both constructed with SCID mice.

The first breakthrough was the appearance of the Prkdc^scid (DNA activated, protein kinase, catalytic polypeptide; severe combined immunodeficiency, abbreviated as scid) mutation in CB17 mice (C.B17-SCID mice) (15). The first models of humanized mouse models were reported in 1988 by three independent groups who transplanted human hematopoietic cells into immunodeficient mice (16–18). Two of these studies used C.B17-SCID mice as recipients for human peripheral blood (17) or human fetal tissues (16). C.B17-SCID mice (C.B17-Prkdc^scid ) possess a genetic autosomal recessive mutation (scid) affecting both B and T lymphocytes. Additionally, they contain normal NK cells, macrophages, and granulocytes (15). However, engraftment occurred at a low level, and the transplanted human cells failed to generate an immune response. Limitations hampering human cell engraftment in C.B17-SCID mice include T and B cell leakiness and high levels of host NK cells, limiting the transplantation of human HSCs (19). Additionally, scid mutation resulted in defective DNA repair and increased radiosensitivity (20).

Immunodeficient mouse strains have been gradually developed and optimized, as shown in Figure 1 (TIMELINE). In 1995, the advent of non-obese diabetic SCID (NOD/SCID) mice was notably a second breakthrough involving the transfer of scid mutation onto a NOD background (NOD.CB17-Prkdc^scid ) (21). Backcrossing of the scid mutation onto the NOD genetic background resulted in an immunodeficient recipient in which increased levels of human engraftment were achieved (22). Animals homozygous for the scid mutation have impaired T- and B- cell development, and NOD background results in deficient NK cell function. Therefore, the NOD/SCID mice have the following characteristics (23): (a) lack of complement C5; hence, complement activation is inhibited; (b) deficient level and decreased killing function of NK cells; (c) lack of T and B cells and antibodies in body fluids. The optimal method for constructing a humanized mouse model is to irradiate NOD/SCID mice with sublethal doses and transplant the human embryonic liver and thymus tissue blocks under the kidney capsule, followed by tail vein injection of homologous human embryonic liver-derived CD34⁺ HSCs. The transfused human HSCs produce large numbers of B and myeloid cells, whereas T cells develop in the transplanted thymus (24). However, some human cells, such as erythrocytes, do not reconstitute well due to strong rejection by the immune system of immunodeficient mice. Notably, NOD/SCID mice have additional shortcomings; for example, they are sensitive to radiation and can only receive small irradiation doses. Moreover, leakage of T and B cells occurs in older NOD/SCID mice, and their survival period is only approximately 8 months.

In a new immunodeficient mouse model, the recombination-activating gene (RAG)-deficient mouse, targeted mutations at RAG1 ( 25) and RAG2 ( 26) loci can prevent mature T and B cell development (27), allowing the mouse to receive transplanted human cells, without causing radiosensitivity or leakiness. However, their high NK cell activity limits the reconstitution of human HSCs. To optimize RAG-deficient mice and reduce their NK cell-killing function, scientists have modified mouse genes, further optimizing the strain. In addition, major histocompatibility complex-I molecules are essential for NK cell development, containing an important component, B2m. β2 -microglobulin is required for the expression of MHC class I molecules and the lack of MHC class I molecules in NOD/SCID B2m^-/- mice prevents NK-cell development (28). Therefore, the constructed novel immunodeficient NOD/SCID B2m^-/- mice do not have the functional NK cells present in NOD/SCID mice (29); hence, they can be applied for higher levels of humanized cell reconstitution.

The third breakthrough was the humanization of immunodeficient mice homozygous for targeted mutations at the interleukin (IL)-2 receptor γ-chain locus (IL2rg; also known as the common cytokine-receptorγ-chain, γc, and CD132) in the mid-2000s (30, 31). γ_c, is a component of the receptors for ILs-2, 4, 7, 9, 15, and 21; the IL-2 receptor gamma chain (IL2Rγ) is a common signaling component of these ILs (32). The absence of IL2Rγ blocks NK cell development owing to the ablation of ILs-7 and 15 signaling, which researchers have exploited to enhance the transplantation of humanized cells with the advancement of technology (33). In addition, this mutation results in poor lymph node development and organization.

Mice with mutations in IL2rg can be selectively bred with those with mutations in RAG1, RAG2, NOD, SCID, and other genes to generate new immunodeficient mice that are more suitable for transplantation of human cells (34). Such new immunodeficient mouse lines include NOD.Cg-Prkdc^scidIl2rg^tm1Wjl (NSG) mice (35), NODShi.Cg-Prkdc ^scidIl2rg ^tm1Sug (NOG) mice (30), NOD.Cg-Rag1^tm1Mom Il2rg^tm1Wjl (NRG) mice (36), Rag2^-/-γc^-/- mice (37), C.Cg-Rag2^tm1FwaIl2rg^tm1Sug (BRG) mice (31), and C.Cg-Rag1 ^tm1MomIl2rg ^tm1Wjl (BRG) mice (36).

The three major mouse strain platforms used are NSG, NOG, and BRG mice. The acronym BRG usually relates to the BALB/c-Rag1 ^−/− or BALB/c-Rag2 ^−/− strains and the IL2rg ^null mutation. Regarding the different mice backgrounds, RG mice included the NRG and BRG. The NRG and BRG backgrounds are usually NOD and BALB/c mice, respectively. Several strains of BRG and NSG mice are available from the Jackson Laboratory, and NOG mice from Taconic. The characteristics and derivation of each immunodeficient mouse model from immunodeficient IL2rg null mice are reviewed in Table 1 .

Based on the types of human-derived cells or tissues transplanted into the strains of immunodeficient mice for humanized mouse model construction, these models can be divided into four classical types (shown in Table 2 ), discussed below.

Due to the differences between mice and humans in the growth factors and cytokines required for hematopoietic development, there are several major limitations inherent in humanized mice. Scientists are currently focused on ensuring the development and maintenance of human cells in the murine microenvironment, and human growth factors and cytokines are essential to achieve this.

Technologies to deliver human species-specific factors into mice to enhance human hematopoiesis and immune system development are diverse. The simplest approach is to inject human cytokines as recombinant proteins or through the hydrodynamic injection of plasmids. Erythropoietin is an important cytokine in terminal RBCs differentiation; however, mouse erythropoietin cannot sufficiently cross-react with the respective human receptor (50). IL-3 is a growth factor that stimulates hematopoiesis; the administration of human IL-3 to patients increases the number of reticulocytes and platelets (51, 52). In a humanized mouse model, human erythroid lineage cells can develop in the mouse bone marrow; however, mature huRBCs are deficient in peripheral blood. Notably, the provision of human erythropoietin and IL-3 can improve human erythropoiesis (53). Several factors contribute to adequate T cell homeostasis in circulation, including human T cell survival in the presence of mouse macrophages, such as ILs-7 and 15 (54). IL-7 is a central cytokine involved in hematopoietic cell development; the injection of its recombinant form transiently affects thymic lymphopoiesis but does not improve peripheral T cell homeostasis (35, 55). Higher levels of human DCs can be obtained by the hydrodynamic injection of plasmids encoding human granulocyte-macrophage colony-stimulating factor (GM-CSF), IL-4, and Fms-like tyrosine kinase 3 ligand (Flt3L) (56).

With the increasing feasibility of genetic engineering, different protocols have been used to deliver a variety of human cytokines in humanized mice, especially NSG and BRG mice. Three main genetic engineering technologies have been applied: transgenic expression of bacterial artificial chromosomes, transgenic expression of cDNA constructs driven by tissue-specific or ubiquitous promoters, and knock-in technology. Furthermore, stem cell factor (SCF) is critical for HSC engraftment, proliferation, development, and survival. Scientists found that human membrane-bound SCF increases the levels of HSC engraftment in non-irradiated NSG mice. The transgenic expression of this SCF under a cytomegalovirus promoter in NSG mice resulted in a higher engraftment level after human Hematopoietic stem and progenitor cell (HSPC) transplantation, even without recipient pretreatment (57, 58). Notably, GM-CSF was initially classified as a hematopoietic growth factor (59). The transgenic expression of cytokines, known to support human myelopoiesis, can improve the major deficiencies in the development of human myeloid cells in humanized mice. Transgenic mice expressing human GM-CSF, SCF, and IL-3 under the control of a ubiquitous cytomegalovirus promoter are generated from an NSG background (60). Furthermore, the knock-in humanization of the two adjacent genes encoding GM-CSF and IL-3 has a critical effect on the lung alveolar macrophages, primarily due to the humanization of GM-CSF (61). CSF-1 and macrophage-CSF(M-CSF) are early cytokines that promote hematopoiesis, and in the hematopoietic system, CSF-1/M-CSF is believed to act specifically on myeloid progenitors.

Similarly, the knock-in replacement of the gene encoding mouse M-CSF by its human counterpart leads to improved engraftment of functional CD33⁺CD14⁺ monocytes in the tissues of recipient mice (62). Furthermore, the most noticeable effect of IL-3 transgenesis in NSG mice is in human T cell improvement, especially the regulatory T cells (60). Additionally, the transgenic expression of membrane human SCF has a positive effect on myelopoiesis, particularly on the development of mast cells (58). Studies on dengue virus showed that human T cells developed after engraftment of HLA-A2 transgenic NSG mice with HLA-A2(+) human cord blood HSCs and were able to secrete IFN-γ, IL-2, and TNF-α in response to stimulation with dengue virus (63). Furthermore, MISTRG mice carry knock-ins for human cytokines granulocyte-monocyte and M-CSF, IL-3(I), thrombopoietin, signal regulatory protein α, and the receptor for the “do not eat me” signal regulatory protein CD47 in the Rag2 ^−/− Il2rg ^−/−(RG) background (64). Based on these mice, scientists deleted the fumarylacetoacetate hydrolase (FAH) gene using CRISPR-Cas9 technology in MISTRG to generate MISTRGFah ^−/− (MISTRGFah) mice and then humanized the liver of this model. Scientists have established a new humanized mouse model based on huHepMISTRGFah^−/−mice with a better and longer huRBC reconstitution and used the model to mimic human SCD (10).

Table 3 lists the cytokines delivered exogenously into humanized mice by injection of recombinant cytokines and hydrodynamic injection of plasmid DNA. Table 4 shows the genetic expression of human growth factors and cytokines in the genomes of the immunodeficient mice.

In 2011, Yang et al. found that macrophages in immunodeficient mice are a key factor in the rejection of human erythrocytes and that clodronate liposomes application to remove macrophages made it possible to optimize the level of human erythrocyte reconstitution in humanized mice; the proportion of reconstituted human erythrocytes in this model was approximately 3% (53). Furthermore, in 2013, Chen et al. applied plasmids encoding human IL-15 and Flt3L to enhance human NK cells in mice and administered a sufficient amount of human erythrocytes daily to study erythrocyte-related diseases, such as malaria (72). However, on stopping the administration, mice did not produce human erythrocytes; hence, this method cannot be used to study the differentiation of human HSCs into erythrocytes and their development in humanized mice. Furthermore, in a 2014 study on malaria, Chen et al. used clodronate liposomes to remove mouse macrophages and antibodies to neutralize mouse neutrophils, thereby inhibiting human erythrocyte rejection. Additionally, they applied high-pressure injection of plasmids expressing human erythropoietin and IL-3, promoting human erythrocytes reconstitution in humanized mice (73). However, the percentage of reconstituted human erythrocytes in this humanized mouse model was not very high (1.6–4%). In 2014, Wathsala et al. used gene editing to construct humanized mice by transplanting HLA-adapted human HSCs into HLA-DR4. RagKO.IL2RγcKO.NOD (DRAG) mice and the percentage of reconstituted human erythrocytes in this model was low (0.2–1%) (74). In 2017, Bing’s research showed that macrophages, neutrophils, endothelial cells, and complement C3 play important roles in rejecting human erythrocytes in NOD/SCID mice. Complement C3 depletion in the serum of mice that had already been depleted of macrophages with cobra venom factor (CVF) increased the proportion of reconstituted human erythrocytes in human HSC-transplanted mice. We found that the simultaneous application of CVF to deplete mouse complement C3 and clodronate liposomes to deplete mouse macrophages resulted in a two-fold increase in the proportion of reconstituted human erythrocytes in humanized mice compared with the application of clodronate liposomes alone (11). In 2021, Song et al. reported a new humanized mouse model in which combined human liver, and cytokine humanization improved erythropoiesis and RBC reconstitution in the circulation based on MISTRG mice. They used this model to mimic human SCD (10); however, the reconstitution rate was not ideal. The human erythrocyte reconstitution rate in these humanized mouse models is insufficient; therefore, further clarification of the mechanism of human erythrocyte rejection in humanized mice is necessary to develop a better-humanized mouse model.

Platelets are produced by the fragmentation of megakaryocytes and the differentiation of megakaryocyte-erythroid precursor cells in the bone marrow (75); Thrombopoietin(TPO) plays a major role in this process. Similar to erythrocytes, human megakaryocytes are detectable in the bone marrow of humanized mice; however, human platelets are not detectable in the peripheral blood of humanized mice. Moreover, TPO-based humanization did not significantly increase the number of platelets in the peripheral blood; however, applying clodronate liposomes to remove phagocytes significantly increased the platelet level (58, 76).

The lack of human erythrocyte reconstitution in humanized mice has plagued researchers in erythrocyte-related diseases for many years. The reason why human T, B, NK, and myeloid cells can be reconstituted, but human erythrocytes are absent in humanized mice remains to be determined. The rejection of human erythrocytes by the mouse immune system is the primary factor affecting their reconstitution in humanized mice. The mouse immune system mainly includes the complement system and innate immune cells, including macrophages, neutrophils, and endothelial cells. Our previous studies showed that the rejection of human erythrocytes by mouse macrophages was a key factor in the defective reconstitution of human erythrocytes in humanized mice. In recent research, we found that another key factor is the allogeneic rejection of human erythrocytes by neutrophils and endothelial cells in immunodeficient mice (11).

Humanized mouse models with huRBC reconstitution can be applied to malaria, SCD, aplastic anemia, and many other huRBC-related diseases (67, 97). For example, malaria is mainly transmitted through the bites of mosquitoes infected with Plasmodium, which reproduces in the human liver and is released into the bloodstream, infecting human erythrocytes. Therefore, the optimal choice is a humanized mouse model that simulates the stages of Plasmodium reproduction and its release into the bloodstream with human liver tissue and erythrocyte reconstitution (68). Currently, two main approaches are combined by scientists to construct mice with huRBCs: continuous infusion of huRBCs and transplantation of human HSCs in mice. For example, one group used plasmids encoding human IL-15 and Flt3L to enhance human NK cells in mice, followed by continuous daily infusion of large amounts of huRBCs to study malaria (72). However, once human erythrocytes were withdrawn, mice no longer had human erythrocytes in their bodies. Although we improved human erythrocyte reconstitution in humanized mouse models, clodronate liposomes injection was required, which has certain toxic side effects in immunodeficient mice. Furthermore, high doses of clodronate liposomes cause the death of immunodeficient mice in a short period; hence the observation window is insufficient. Therefore, a humanized erythrocyte mouse model that can consistently carry high proportions of human erythrocytes at different stages is urgently needed, and scientists are working to develop such a model.