Establishment of a reverse genetics system for studying human immune functions in mice

Abstract

Reverse genetics approaches in mice are widely used to understand gene functions and their aberrations in diseases. However, limitations exist in translating findings from animal models to human physiology. Humanized mice provide a powerful bridge to understanding human physiology and mechanisms of disease pathogenesis while maintaining the feasibility of working with small animals. Methods for generating humanized mouse models that allow scientists to probe contributions of particular genes have been rudimentary. Here, we established an efficient method for generating genetically modified human cord blood–derived CD34⁺ cells for transplantation, resulting in humanized mice with near-complete loss of specific gene expression by the human immune system. Mice transplanted with Cas9-edited human CD34⁺ cells recapitulate functional consequences of specific gene losses in the human immune system. Our approach enables targeted gene knockouts in humanized mice, offering a valuable tool for human gene function studies in vivo.

Humanized mouse models enable the use of reverse genetics to study immune gene functions in human disease.

INTRODUCTION

Genetically modified mice have revolutionized investigations of gene function and regulation as well as the ability to probe into the mechanisms behind disease pathogenesis (1). In the early 2000s, there was an increasing appreciation for the incongruities between animal models and human physiology and disease, which launched the development of human immune system mice, also known as humanized mice. They became an invaluable tool in biomedical research, offering a unique platform for more translationally relevant work from studying underlying mechanisms of human diseases to testing potential therapeutic strategies (2, 3). There are two common ways to “humanize” a mouse: use of humanized genetically engineered model whereby a human gene replaces the mouse ortholog via targeted gene replacement and can be used for the narrow investigation of that particular gene, or by engrafting immunodeficient mice with human hematopoietic stem and progenitor cells (HSPCs; CD34⁺ cells) (4), leading to the development of a functional human immune system within the mouse (5, 6). However, the current approach for specifically knocking out genes within the human immune system in humanized mice has faced several challenges. While gene knockouts (KO) can be achieved in human and mouse HSPCs using electroporation of Cas9 and single guide RNA (sgRNA) ribonucleoprotein (RNP) complexes in vitro (7–9), immune reconstitution to generate humanized mice using these genome-edited human HSPCs remains challenging. A recent proof-of-concept study showed that deletion of the human CD33 coding gene in CD34⁺ cells resulted in development of CD33-deficient monocytes in mice (8), demonstrating the possibility of interrogating gene-specific functions within the human immune system in mice. Here, we aimed to establish an efficient method for genome editing in human CD34⁺ cells derived from umbilical cord blood and assess the impact of these modifications in a humanized mouse model. We optimized the efficiency of gene KO in human CD34⁺ cells using electroporation of Cas9/sgRNA RNP complexes. We achieved nearly 100% KO in mice that received edited human CD34⁺ cells, generating human transgenic mice on both an NSG-SGM3 (10) and MISTRG-6-15 (11, 12) backgrounds while attaining no limitations in engraftment levels compared to controls. Furthermore, we generated RAG2-KO, TCF7-KO, CCR5-KO, and IFNAR-KO humanized mice and examined the impact of these deletions on hematopoiesis and HIV-1 infection. In summary, our study offers an efficient method for genome editing in human HSPCs, highlighting its potential for studying gene function and antiviral immunity in humanized mouse models.

RESULTS

Optimizing gene KO efficiency in human cord blood–derived CD34⁺ cells by electroporation of Cas9/sgRNA RNP complexes in vitro

Electroporation of Cas9/sgRNA RNP complexes is a powerful technique for genome editing in human primary cells (13–15). The exact parameters used for electroporation determine the ability to obtain high transduction efficiencies and high gene disruption efficiencies. We optimized various parameters, such as pulse code, as well as the concentrations and ratios of the Cas9 protein and sgRNA for effective genome editing for human CD34⁺ cells. We included two sgRNAs for each gene because human genetic variations influence the CRISPR-Cas9 editing efficiency (16, 17). To determine optimal gene KO efficiency in human HSPCs, we first tested different pulse codes using the Lonza 4D electroporation system. To do this, we targeted the CD45 coding gene, due to its trackability via flow cytometry with two sgRNAs. Cord blood HSPCs were electroporated with Cas9/sgRNA RNP complexes and subsequently cultured for 3 to 5 days in the presence of stem cell cytokines. Flow cytometry was used to assess CD45 KO efficiency and cell viability (Fig. 1A). Among the tested pulse codes, program code DZ-100 resulted in the highest and most consistent editing efficiency (Fig. 1, B and C). Specifically, this approach including two sgRNAs achieved nearly 100% CD45 KO across four cord blood donor samples while maintaining ~65% cell viability (Fig. 1, C and D). Next, we optimized the doses of the Cas9 protein and sgRNAs to achieve maximal gene KO. We observed that sgRNA concentrations below 25 μM in the 20-μl reaction system reduced the KO efficiency (Fig. 1, E to G). Consequently, for our subsequent experiments, we selected the DZ-100 program and maintained Cas9 and sgRNA concentrations at 10 and 25 μM, respectively. Under these conditions, we ensured maximum gene KO efficiency while preserving acceptable levels of cellular viability.

Fig. 1. High-efficiency genome editing in human CD34⁺ HSPCs by electroporation of Cas9/sgRNA RNPs.

(A) Experimental scheme of CRISPR-Cas9 gene KO in human CD34⁺ HSPCs. h, hours. (B to D) Optimization of the electroporation (EP) program. Representative fluorescence-activated cell sorting (FACS) plots of CD45 expression (B), percentage of edited (CD45⁻) cells (C), and percentage of viable cells after electroporation (D) are shown. Human CD34⁺ cells were electroporated with Cas9/sgRNA RNP complexes targeting human CD45 by using different program codes. Electroporated cells were cultured with stem cell factor (SCF) (100 ng/ml), FMS-like tyrosine kinase 3 (FLT3) (100 ng/ml), and thrombopoietin (TPO) (100 ng/ml) for 5 days before flow cytometry analysis. (E to G) Optimization of the input sgRNA amounts. Human CD34⁺ cells were electroporated in the presence of various amounts of Cas9/sgRNA complexes, and subsequently, CD45 expression and cell viability were measured as above. CD34⁺ cells from four different cord blood were used. Data were shown as means ± SEM. P values were calculated using the one-way analysis of variance (ANOVA) with Tukey’s multiple comparisons tests. *P < 0.05; **P < 0.01; ****P < 0.0001. n.s., not significant.

Optimizing immune reconstitution by electroporated human cord blood CD34⁺ cells in immunodeficient mice

One of the most common approaches to creating humanized mice involves transferring human HSPCs into immunocompromised mice. Given the cell viability loss during electroporation, we began by investigating how the initial number of electroporated cells and the in vitro recovery post-electroporation affect the engraftment levels (Fig. 2A). Specifically, we engrafted 3000, 10,000, or 30,000 electroporated human cord blood CD34⁺ cells into newborn NSG-SGM3 mice. These cells were electroporated with the Cas9 protein and nontargeting sgRNAs. Human immune reconstitution was evaluated 12 to 14 weeks post-engraftment by analyzing the presence of human CD45⁺ (huCD45⁺) cells in peripheral blood. Our findings indicate a direct correlation between a higher initial cell count and improved engraftment rates in the blood, spleen, and bone marrow (BM) (Fig. 2B). We subsequently assessed whether in vitro culture of electroporated CD34⁺ cells enhances engraftment efficiency. Electroporated cells were cultured for durations of 0.5, 24, or 72 hours before engraftment. We observed that the percentage of human CD45⁺ cells in the spleen, BM, liver, and lung exhibited no difference, indicating that short-term in vitro culture of CD34⁺ HSPCs did not have a substantial impact on the engraftment efficiency (Fig. 2, C and D). Moreover, the proportion of human T cells, B cells, monocytes, and natural killer (NK) cells in tissues was not affected by the duration of in vitro culture (fig. S1, A and B). Consequently, we opted to transplant a minimum of 30,000 electroporated cells with a 3-day period of in vitro culture in our subsequent experiments. The 3-day in vitro culture period provided the flexibility needed to wait for the delivery of newborn mice. To evaluate whether electroporation had any long-lasting impact on the stemness and multipotency of hematopoietic stem cells (HSCs), we engrafted MISTRG-6-15 mice using either untouched CD34⁺ cells or cells electroporated with the Cas9 protein and a scramble sgRNA. The numbers of human CD45⁺ cells in each tissue was similar between the two groups (Fig. 2E). Both groups had similar reconstitution levels and composition of human immune cells in blood and tissues (fig. S1, C to G). Furthermore, the frequency and number of BM CD34⁺ cells, including HSCs and multipotent progenitors (MPPs), were comparable between the groups (Fig. 2F). Hence, these results illustrate that our methodology successfully created transgenic humanized mice with a selective human gene KO, without impairing overall engraftment.

Fig. 2. Immune reconstitution and multilineage differentiation in mice engrafted with electroporated human CD34⁺ cells.

(A) Experimental scheme of electroporation and transplantation procedures. (B to D) Human immune cell engraftment and multilineage differentiation using in vitro cultured CD34⁺ cells post-electroporation. In (B), different doses of CD34⁺ cells were used for engraftment immediately after electroporation. 3K (n = 3), 10K (n = 4), and 30K (n = 3). [(C) and (D)] After electroporation, CD34⁺ cells were cultured in vitro for 0.5, 24, or 72 hours with the presence of SCF (100 ng/ml), FLT3 (100 ng/ml), and TPO (100 ng/ml) before engraftment of newborn NSG-SGM3 mice. Each animal received 30,000 cells. Tissues were harvested 14 weeks after engraftment to determine the reconstitution of total human immune cells by FACS. (E and F) HSPCs and immune reconstitution using unelectroporated and Cas9 control CD34⁺ cells. Newborn MISTRG-6-15 mice were engrafted with either untouched CD34⁺ cells or cells electroporated with the Cas9 protein and a scramble sgRNA. Each animal received 30,000 cells. Blood and tissues were collected 10 weeks post-engraftment to determine human immune reconstitution. The number of human CD45⁺ cells in tissues (E) as well as subsets of BM progenitors (F) were determined by FACS. Data is presented as the mean ± SEM. In (B) and (D), P values were calculated using the one-way ANOVA with Tukey’s multiple comparisons tests. In (E) and (F), P values were calculated using the two-way ANOVA with the Holm-Sidak test. *P < 0.05; **P < 0.01; ***P < 0.001.

Immune reconstitution by Cas9-edited human cord blood CD34⁺ cells in immunodeficient mice

We next aimed to evaluate human immune reconstitution after optimizing the electroporation of Cas9/sgRNA RNP complexes and engraftment conditions. To this end, we targeted the HLA-A2 gene, which is not involved in hematopoiesis and can be easily detected via flow cytometry. We first validated the editing efficiency in vitro and found that nearly 100% KO efficiency of HLA-A2 in CD34⁺ cells was achieved 5 days after Cas9/sgRNA RNP electroporation (fig. S2A). Subsequently, we transplanted these HLA-A2–edited and Cas9 control (Cas9 ctrl) cells into newborn NSG-SGM3 mice and assessed the HLA-A2 KO efficiency 14 weeks post-engraftment. We observed a complete loss of human HLA-A2 expression in the blood, spleen, BM, liver, and lung of each mouse engrafted with the edited cells (Fig. 3, A and B). In addition, we observed that the levels of reconstitution of human CD45⁺ cells in the blood, spleen, BM, liver, and lung of engrafted mice were comparable between the Cas9 control and HLA-A2 targeted groups, suggesting that the KO of HLA-A2 had no impact on the stemness, multipotency, and differentiation potential of edited CD34⁺ cells (Fig. 3, C and D, and fig. S2B). Notably, this approach achieved >90% HLA-A2 KO across five humanized mouse cohorts engrafted with CD34⁺ cells from different cord blood donors (fig. S2C).

Fig. 3. Immune reconstitution by HLA-A2–KO CD34⁺ cells in immunodeficient mice.

NSG-SGM3 mice were used in this figure. Each animal received 30,000 cells. Samples were harvested 14 weeks post-engraftment. (A) Representative FACS plots of human HLA-A2 expression by huCD45⁺ cells in mice 14 weeks after engraftment in the blood, spleen, BM, liver, and lung. (B) Percentages of human HLA-A2⁺ of huCD45⁺ cells in the blood, spleen, BM, liver, and lung. (C) Representative FACS plots of human (huCD45⁺) and mouse (msCD45⁺) immune cells in the blood, spleen, BM, liver, and lung. (D) Human immune cell reconstitution in the blood, spleen, BM, liver, and lung. Data were shown as the means ± SEM. Cas9 ctrl (n = 9); HLA-A2–KO (n = 7). P values were calculated using an unpaired two-tailed t test. ****P < 0.0001.

Next, we aimed to evaluate whether CRISPR-Cas9–edited HSPCs provided a reverse genetics system, which faithfully recapitulated known immune defects when essential genes were deleted. To do this, we used the MISTRG-6-15 mouse strain, which supports the development of human innate immune system more efficiently than the NSG-SGM3 mouse strain, so that the loss of T cells and B cells caused by targeted gene deletions would not result in a complete loss of human immune reconstitution. In mice, loss of T cell factor 1 (Tcf1, encoded by Tcf7) blocks thymocyte differentiation but has no direct impact on humoral responses (18, 19). We introduced TCF7-KO and Cas9 ctrl cord blood–derived CD34⁺ cells into MISTRG-6-15 mice to evaluate immune reconstitution. In mice engrafted with TCF7-KO CD34⁺ cells, a complete loss of T cell development was observed whereas B cells, myeloid cells, and NK cells remained unaffected (Fig. 4, A to C). A slight increase in B cell or NK cell frequency in the TCF7-KO group was likely a reflection of the loss of T cells. We confirmed that targeted deletion of TCF7 had no impact on hematopoiesis, as evidenced by the comparable levels of BM progenitors between the TCF7-KO and Cas9 control groups including total CD34⁺ cells as well as different subsets of progenitor cells (Fig. 4, D and E). These results indicate a specific ablation of mature T cells by TCF7-KO. Next, we introduced RAG2-KO and Cas9 ctrl cord blood–derived CD34⁺ cells into MISTRG-6-15 mice to evaluate the reconstitution of adaptive immunity. The successful establishment of the RAG2 KO human immune system in the mice was confirmed using deep sequencing analysis. This method revealed that >90% of human cells were edited at the RAG2 locus and that the editing efficiency was comparable between the two cohorts (Fig. 4F and fig. S3A). Both B cell and T cell populations were absent in mice engrafted with RAG2-KO CD34⁺ cells (Fig. 4G), whereas innate lymphoid cells (ILCs) were not affected in the RAG2-KO group (fig. S3, B and C). These results confirmed the essential role of RAG2 in genomic rearrangement of B cell and T cell receptors (20). Notably, the RAG2-KO animals are excellent tools to study human innate immune responses to pathogens or other diseases. Our TCF7-KO and RAG2-KO transgenic humanized mice demonstrate that CRISPR-Cas9–edited human HSPCs provide a robust reverse genetics system, which will efficiently allow scientists to probe the function and contribution of a variety of immune genes in vivo in the broader context of a complete immune system.

Fig. 4. The human immune systems developed from TCF7-KO and RAG2-KO CD34⁺ cells recapitulate established immune deficiencies.

MISTRG-6-15 mice were used in this figure. Each animal received 30,000 cells. Samples were harvested 12 weeks post-engraftment. (A) TCF1 expression by huCD45⁺ cells was determined by FACS. (B to D) Reconstitution of human immune cells in the blood (B), spleen (C), and BM (D) of mice engrafted with TCF7-KO and Cas9 ctrl CD34⁺ cells. (E) Analysis of BM total huCD45⁺ cells and human HSPCs from mice engrafted with TCF7-KO and Cas9 ctrl CD34⁺ cells. (F) Editing efficiency of RAG2. PBMCs were collected for genomic DNA extraction. Editing efficiency was determined by the MiSeq Illumina sequencing of the targeted region (see also fig. S3A). (G) T cell, B cell, and myeloid cell development in mice engrafted with RAG2-KO and Cas9 ctrl CD34⁺ cells. Representative FACS plots of CD3 and CD19 expression by blood huCD45⁺ cells were shown. Frequency of CD14⁺, CD19⁺, and CD3⁺ blood cells were determined by FACS. Data were shown as the means ± SEM. P values were calculated using an unpaired two-tailed t test or the two-way ANOVA with the Holm-Sidak test. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.

Mice with an edited human immune system have altered responses to HIV-1 infection

The humanized mouse reverse genetics system may offer insights into HIV-1 and host interactions in vivo. CCR5 and CXCR4 are the two co-receptors for HIV entry (21, 22), and CCR5-tropic HIV-1 isolates are transmitted by all routes and almost always dominate throughout the course of infection prior to the onset of AIDS (23, 24). We introduced human CCR5-KO and Cas9 ctrl CD34⁺ cells into NSG-SGM3 mice. The successful establishment of the CCR5 KO human immune system in the mice was confirmed using flow cytometry and deep sequencing analysis. These methods revealed that more than 95% of human cells were edited at the CCR5 locus, whereas total human cell engraftment remained unaffected (Fig. 5, A to C, and fig. S4A). We then infected these mice with the CCR5-tropic HIV_BaL and monitored the levels of plasma HIV-1 RNA and the CD4⁺ T cell loss in the blood. Mice that were reconstituted with the CCR5-KO immune system displayed undetectable levels of plasma HIV-1 RNA (Fig. 5D), and CD4⁺ T cell loss was prevented in these animals (Fig. 5E). Moreover, both the frequency and the number of CD4⁺ T cells in tissues including the spleen, BM, liver, and lung were higher in mice with a CCR5-KO immune system in comparison to the Cas9 control mice (Fig. 5, F and G).

Fig. 5. Mice with a human immune system developed from CCR5-KO CD34⁺ cells are resistant to HIV-1 infection.

NSG-SGM3 mice were used in this figure. Each animal received 30,000 cells. Mice were infected with HIV_BaL 14 weeks post-engraftment. (A) Representative FACS plots of CCR5 expression in blood CD4⁺ T cells before infection. (B and C) Percentage of huCD45⁺ cells (B) and CCR5⁺ CD4 T cells (C) in blood before infection. (D) Plasma viral loads measured by reverse transcription qPCR (RT-qPCR). (E) Loss of blood CD4⁺ T cells determined by FACS. (F to G) Frequency and number of human CD4⁺ T cells in tissues. Data were shown as the mean ± SEM. Cas9 ctrl (n = 6); CCR5-KO (n = 5). P values were calculated using an unpaired two-tailed t test. *P < 0.05; **P < 0.01; ***P < 0.001.

Type I interferons (IFNs) play a complicated role in HIV-1 infection as they not only suppress viral replication but also drive immune activation and exhaustion (25). On one hand, IFN-α monotherapy during clinically asymptomatic HIV-1 infection without antiretroviral therapy reduced plasma viral loads, prevented CD4 decline, and delayed disease progression to AIDS (26–28). On the other hand, in vivo blockade of IFN-α receptor (IFNAR) in simian immunodeficiency virus–infected rhesus macaques dampened up-regulation of IFN-stimulated genes, which led to heightened plasma viremia and accelerated CD4 decline (29). Although these interventions demonstrate strong anti–HIV-1 activities of type 1 IFNs, the role of type 1 IFNs induced during natural HIV-1 infection in vivo remains unclear. We introduced human IFNAR-KO and Cas9 ctrl CD34⁺ cells into MISTRG-6-15 mice as this mouse strain develops myeloid cells more efficiently than NSG-SGM3. We confirmed nearly 100% editing efficiency of IFNAR (Fig. 6A and fig. S4B) and observed that the levels of reconstitution of human CD45⁺ cells as well as T cell development were comparable between the Cas9 control and IFNAR-KO groups (Fig. 6, B to D). Upon HIV-1 infection, both groups of mice produced detectable and comparable levels of IFN-α2 (Fig. 6E). The plasma HIV-1 RNA levels were 10- to 100-fold higher in the IFNAR-KO group (Fig. 6F), which suggests that type 1 IFNs contribute substantially during acute HIV-1 infection to control viral replication.

Fig. 6. Type I IFNs suppress HIV-1 replication in vivo.

MISTRG-6-15 mice were used in this figure. Each animal received 30,000 cells. Mice were infected with HIV_BaL 12 weeks post-engraftment. (A) Editing efficiency of IFNAR. PBMCs were collected for genomic DNA extraction. Editing efficiency was determined by the MiSeq Illumina sequencing of the targeted region (see also Fig. 4B). (B to D) Percentage of huCD45⁺ (B), CD3⁺ and CD14⁺ (C), and CD4:CD8 ratio (D) cells in the blood in the indicated groups. (E) Plasma IFN-α2 levels in infected and control mice measured by ELISA. (F) Plasma viral loads were measured from 1 to 3 weeks after infection by RT-qPCR. In (A), (B), (D), and (E), P values were calculated using an unpaired two-tailed t test. In (C) and (F), P values were calculated using the two-way ANOVA with the Holm-Sidak test. *P < 0.05; ****P < 0.0001.

DISCUSSION

Here, we described a reverse genetics system that allows us to interrogate specific gene functions in the development of the human immune system and its contribution to controlling human pathogens. We used well-characterized immune genes as a proof of concept to demonstrate highly efficient genome editing and immune reconstitution of transgenic HSPCs in mice, which provides a successful reverse genetics model. We chose the MISTRG-6-15 and the NSG-SGM3 mice as they support human hematopoiesis efficiently due to their production of multiple human cytokines and growth factors. Although NSG mice are the most commonly used strain for the generation of humanized mice, 50,000 to 200,000 unedited CD34⁺ cells per mouse are needed for human immune reconstitution in the NSG model, which would be prohibitive when using transgenic cells.

Genetical ablation of certain human genes in primary CD4⁺ T cells to study HIV-1 and host interaction can provide valuable insights into HIV-1 biology that help identify previously unidentified therapeutic targets (30, 31). The platform described here has notable advantages compared with the functional studies in vitro. For example, immune cells residing in different tissues often have distinct phenotypes and functions because they interact with tissue-specific parenchymal cells and rely on different cytokines for maturation and survival (32–38). Although latent HIV-1 can be found in tissue macrophages (39–42), without an animal model lacking a T cell reservoir, it remains unclear whether tissue macrophages contribute to HIV-1 reservoirs and whether there would be viral rebound after analytic treatment interruptions. The RAG2-KO and TCF7-KO models we describe can serve as T cell–deficient humanized mice to understand the role of tissue macrophages in HIV-1 persistence. In addition, disruption of certain immune functions such as T cells or type I IFN response could help understand how the human immune system controls pathogens and how pathogens evolve under immune selective pressure. In this study, we chose genes with known immune functions as examples to model human immune reconstitution. This platform is also very well suited to study genes involved in hematopoiesis, from which immune cell reconstitution and secondary transplantations can be used as readouts of HSC functions.

Notably, there are still limitations with regard to broad applications of the reverse genetics in humanized mice. First, genome editing cannot be used to study the function of genes that are essential for hematopoiesis. Second, unlike the commonly used conditional KO systems such as the Cre recombinase, the current genome editing of the human immune system applies to all immune cells without cell type or tissue specificity. Furthermore, because the number of human CD34⁺ cells from a standard cord blood sample is limited, multi-arm large cohort studies remain challenging. Nonetheless, this study represents a step forward in the development of a robust platform to functionally study the contribution of various genes to the control of HIV pathogenesis and to directly probe the host-pathogen interface using humanized mice.

MATERIALS AND METHODS

Mice strain

NSG-SGM3 mice were purchased from the Jackson Laboratory (strain #013062), and the colony was maintained at the Washington University School of Medicine. The MISTRG-6-15 human cytokine knock-in mice were described previously (12). The mouse colony was maintained at the Washington University School of Medicine. Both male and female mice were included. All animal experiments were approved by the Institutional Animal Care and Use Committee of Washington University School of Medicine.

Human samples

Anonymous human cord blood samples were collected at the Cleveland Cord Blood Center. This study is reviewed by the Washington University School of Medicine Internal Review Board and not considered human subject research. The informed consent procedure was not required at the Washington University.

Plasmids and viruses

The replication-competent HIV_BaL was produced by infecting CD8-depleted phytohemagglutinin-stimulated peripheral blood mononuclear cells (PBMCs). The culture supernatant was collected 6 to 9 days postinfection. NL4-3-BAL was generated by replacing the BAL-01 envelope into the NL4-3 consensus sequence and was used for infections of IFNAR-KO mice. The virus was generated by transfection of 293T cells with plasmids and purified and aliquoted using LentiX at −80°C and quantified using the p24 enzyme-linked immunosorbent assay (ELISA) kit, and 10 ng per mouse was used for infection.

CRISPR KO in CD34⁺ cells

The EasySep Human Cord Blood CD34 Positive Selection Kit III (StemCell Technologies, #17897) was used to purify CD34⁺ cells from cord blood. Cas9/sgRNA RNP complexes were electroporated into CD34⁺ cells using the Lonza 4D-Nucleofector system. The recombinant Cas9 protein was obtained from IDT. The modified synthetic sgRNAs were purchased from Synthego, and the sequences are listed in table S1. RNP complexes were prepared by mixing sgRNA (100 pmol) with Cas9 (40 pmol) and incubating them for 10 min at room temperature. A total of 0.2 × 10⁶ CD34⁺ cells were washed with phosphate-buffered saline (PBS) and resuspended in 20 μl of buffer P3 (Lonza, #V4XP-3032). To deliver two specific sgRNAs targeting one gene, 2 μl of each RNP complex was then mixed with the cell suspension and transferred into a 16-well reaction cuvette of the 4D-Nucleofector system. The CD34⁺ cells were electroporated using the program DZ-100, unless indicated. After electroporation, CD34⁺ cells were resuspended in 100 μl of prewarmed Iscove’s Modified Dulbecco’s Medium and transferred to a 96-well plate to recover for 30 min at 37°C. The electroporated CD34⁺ cells were then washed with PBS before in vitro culture or transplantation into the mice.

Flow cytometry analysis

The mouse tissue cell suspensions were first incubated with Zombie NIR in PBS for 30 min at 4°C, followed by incubation with Human TruStain FcX and TruStain FcX anti-mouse CD16/32 blocking antibodies (BioLegend, #422302 and #101302) for 10 min at room temperature. The cell suspensions were then incubated at 4°C with fluorescence-conjugated antibodies for 30 min to stain surface antigens. In all experiments, stained cells were acquired on a BD LSRFortessa, X20, or Accuri C6 (BD Biosciences), and data were analyzed by the FlowJo software. The following antibodies were used for analysis: anti-mouse CD45-FITC (BioLegend, #103108), anti-human CD45-Pacific Blue (BioLegend, #304029), anti-human CD3-BV785 (BioLegend, #317330), anti-human CD19-PE-Cy7 (BioLegend, #302216), anti-human CD56-PE (BioLegend, #362508), anti-human CD14-APC (BioLegend, #301808), anti-human CD4-BUV395 (BD Biosciences, #563550), anti-human CD8-BUV737 (BD Biosciences, #612754), anti-human CD34-PE (BioLegend, #343506), anti-human CCR5-BV421 (BioLegend, #359118), anti-human CD3-APC/Cy7 (BioLegend, #300318), anti-human CD45RA-FITC (BioLegend 304148), anti-human CD90-AF647 (BioLegend 328116), anti-human CD38-PE/Cy7 (BioLegend #980312), and anti-human TCF1-AF647 (BioLegend, #655204). For HSPC analysis, a cocktail of lineage markers including CD3, CD14, CD16, CD19, CD20, and CD56 were used (BioLegend, #348803). Progenitor cells were defined according to a previous study (43). HSC: Lin⁻CD34⁺CD38⁻CD45RA⁻CD90⁺. MPP: Lin⁻CD34⁺CD38⁻CD45RA⁻CD90⁻.

Generation and HIV-1 infection of humanized mice

One- to 3-day-old newborn NSG-SGM3 mice were preconditioned with sublethal irradiation (100 centigray). No preconditioning was performed for MISTRG-6-15 mice. Mice then received 3 × 10⁴ to 4 × 10⁴ unmodified or Cas9-edited CD34⁺ cells by an intrahepatic injection. Reconstitution of human CD45⁺ cells was assessed 9 to 12 weeks after engraftment. Experimental groups were assigned randomly.

HIV_BaL was obtained through the NIH AIDS Reagent Program (#510). Some engrafted mice were infected with HIV_BaL (10 ng p24 per mouse) through retro-orbital injection. Blood samples were collected by retro-orbital or submandibular vein bleeding to quantify plasma HIV-1 RNA. The plasma viral RNA was extracted using a Quick-RNA Viral Kit (Zymo Research) and reverse transcribed using the ProtoScript II Reverse Transcriptase (NEB). A 10-fold serial dilution of HIV genomic DNA served as a standard for measuring plasma viral RNA by the HIV gag-based quantitative polymerase chain reaction (qPCR) assay.

Amplicon deep sequencing

In the first step, corresponding primers were used for the gene locus. In a second round of PCR using primers containing sample-specific barcodes and adapters, amplicons were sequenced with 2 × 150 paired-end reads with MiSeq sequencing (Illumina). The CRISPResso software was used to analyze the deep sequencing data (44).

Statistical analysis

The statistical analysis was conducted using Prism 9. Phylogenetic statistics and analysis were performed as above. The methods for statistical analysis are described in the figure legends. The error bars indicate the SEM.

Supplementary Materials

This PDF file includes:

Figs. S1 to S4

Table S1