HIV-1 infection of genetically engineered iPSC-derived central nervous system-engrafted microglia in a humanized mouse model

ABSTRACT The central nervous system (CNS) is a major human immunodeficiency virus type-1 (HIV-1) reservoir. Microglia are the primary target cell of HIV-1 infection in the CNS. Current models have not allowed the precise molecular pathways of acute and chronic CNS microglial infection to be tested with in vivo genetic methods. Here, we describe a novel-humanized mouse model utilizing human induced pluripotent stem cell (iPSC)-derived microglia to xenograft into murine hosts. These mice are additionally engrafted with human peripheral blood mononuclear cells that serve as a medium to establish a peripheral infection that then spreads to the CNS microglia xenograft, modeling a trans-blood-brain barrier route of acute CNS HIV-1 infection with human target cells. The approach is compatible with iPSC genetic engineering, including inserting targeted transgenic reporter cassettes to track the xenografted human cells, enabling the testing of novel treatment and viral tracking strategies in a comparatively simple and cost-effective in vivo model for neuroHIV. IMPORTANCE Our mouse model is a powerful tool for investigating the genetic mechanisms governing central nervous system (CNS) human immunodeficiency virus type-1 (HIV-1) infection and latency in the CNS at a single-cell level. A major advantage of our model is that it uses induced pluripotent stem cell-derived microglia, which enables human genetics, including gene function and therapeutic gene manipulation, to be explored in vivo, which is more challenging to study with current hematopoietic stem cell-based models for neuroHIV. Our transgenic tracing of xenografted human cells will provide a quantitative medium to develop new molecular and epigenetic strategies for reducing the HIV-1 latent reservoir and to test the impact of therapeutic inflammation-targeting drug interventions on CNS HIV-1 latency.

HIV-1 enters the brain during the first initial weeks of acute infection (19)(20)(21)(22) via transmigration of HIV-1-infected cells across the blood-brain barrier (BBB) (23)(24)(25).In the periphery, HIV-1 targets the CD4 + T-cells and monocytes that then disseminate infection to other tissues and organs (23)(24)(25)(26)(27)(28).In the CNS, these peripherally infected cells primarily target microglia comprising about 5%-10% of adult brain cells (29).Our recent studies on the human postmortem brain corroborate that HIV-1 infection occurs predominantly in microglia, and provirus integration is linked to inflammation-associated reprogramming of microglial transcriptomes and 3D genomes (15).Understanding the molecular mechanisms underlying the establishment and maintenance of actively and latently HIV-1-infected microglia in the CNS will help investigate ways to target and reduce CNS reservoirs and develop therapeutic interventions for microglia-associated neuroinflammation.
Genetic animal models that capture the salient features of CNS HIV-1 infection in humans are greatly needed to interrogate the precise mechanisms governing the HIV-1 life cycle in the host.These can enable the study of initial infection of the peripheral immune system, viral replication, dissemination throughout all organ systems, including the CNS microglia and other myeloid compartments, and provirus integration into the host cell genome to maintain chronic HIV-1 infection.To this end, elegant non-human primate models using various strains of simian immunodeficiency virus have long been established in the field (30,31).However, these models are still inherently limited, given that HIV-1 is a human-specific retrovirus (32).
Although HIV-1 does not replicate in mouse immune cells, the repertoire for modeling neuroHIV has been expanded by employing chimeric virus models to infect mice.The EcoHIV model utilizes a chimeric murine-tropic virus that has been genetically modified to carry a murine leukemia virus envelope coding region and, hence, can infect and simulate HIV-1 disease-like phenotype in mice (33).To study native HIV-1 in small animal models, investigators have employed human immune cell xenografted immunodeficient mouse models.While initially, studies focused on blood and peripheral systems (34,35), a recent advance has enabled CNS engraftment of human cells by using human fetal hematopoietic tissue or cord blood-derived CD34 + hematopoietic stem cells (HSC) into immunocompromised NOG mice expressing a human interleukin (IL)-34 transgene.These mice show colonization of multiple lymphoid and myeloid compartments, including differentiation of microglia-like brain cells susceptible to HIV-1 infection (36,37).Since then, another mouse model, where human IL-34 transgene expressing NOG mice are xenografted with human fetal liver tissue, fetal thymic tissue, and liver-derived CD34 + hematopoietic stem and progenitor cells (HSPC), was reported to undergo human microglia reconstitution in CNS that are susceptible to HIV-1 infection (38).Other models rely on the direct engraftment of HIV-1-infected human myeloid cells into the brain of immunocompromised mice (39).
These humanized mouse models have been extremely valuable for expanding knowledge on molecular and cellular signatures of HIV-1-infected brains.However, currently existing humanized mouse models also pose some limitations.These include the requirement of human fetal tissue or umbilical cord blood that is limited in supply and not readily amplified to conduct extensive tests with the same genotype (36,38,40) and exhibit a lower rate of xenograft reconstitution following irradiation or chemotherapy (36)(37)(38) as compared to induced pluripotent stem cell (iPSC)-derived cells.Some of these models also feature a non-physiological intracranial route of HIV-1 inoculation into the brain (36,39,41).Here, we present a novel humanized mouse model for HIV-1 infection that implements genetically engineered human iPSCs as a near-unlimited resource to reconstitute neonatal mouse brains with human microglia.The unrivaled versatility of iPSCs enables genetic material to be clonally engineered and provides an invaluable toolbox for generating reporter cell lines.Furthermore, in our model, CNS-engrafted mice are peripherally engrafted with human peripheral blood mononuclear cells (huPBMC) in early adulthood to initiate infections from peripheral vasculature.Infected huPBMCs then travel to the CNS to infect the xenografted microglia (xenoMG).Mice dually engrafted with xenoMGs and huPBMCs are infected with M-tropic HIV-1.This models a physiological route of acute CNS HIV-1 infection through virus or infected cells crossing the BBB.

Differentiation of microglia from genetically modified human iPSC in vitro
Studying CNS HIV-1 infection has been challenging due to the limited accessibility of human samples and difficulty in tracing persistently infected cellular reservoirs, even with suppressive ART.This study presents a novel human iPSC-based cell lineage tracing model that irreversibly marks HIV-1-infected cells for subsequent investigation.The iPSC is an ideal model due to its potential to differentiate into multiple cell types and to generate clonal lines (42,43).Specifically, we chose the WTC11 iPSC as our parental line for the following reasons: (i) WTC11 can be robustly differentiated into different cell fates, including microglia-like cells (44,45); (ii) WTC11 possesses a well-known stable 46XY karyotype that is highly conducive to gene editing (to date, over 60 different gene-edited WTC11 lines have been generated and are available through the Allen Cell Collection [https://www.allencell.org/cell-catalog.html]; and (iii) the complete genome sequence of the WTC11 line is publicly available, greatly facilitating the design of gene editing experiments.
We genetically modified the WTC11 line by introducing a Cre-recombinase-depend ent, CAG promoter-driven dsRed-to-eGFP switch cassette into the AAVS1 locus within the PPP1R12C gene (MSE2104 iPSC line, Fig. S1A).We considered other potential safe harbor loci, including human Rosa 26 (hROSA26) (46), CCR5 (47), and CLYBL (48).We ruled out the CCR5 locus because of its association with HIV-1 biology.Additionally, we excluded the CLYBL locus because we observed significant silencing effects during differentiation into HSCs (unpublished data).Our focus shifted to AAVS1 because of its widespread utility in the iPSC field.The AAVS1 locus is also ideal because it permits robust, stable, and reproducible expression of transgenes across multiple lineages maintained long term in culture (49).Furthermore, it mitigates pleiotropic position effects, ensuring that the inserted transgene is not silenced during meso-and ectoderm differentiation.In our transgenic MSE2104 iPSC line, the genetic switch is encoded by inserting a dsRed coding sequence followed by a stop codon into a loxP Cre-recombinase target site upstream of an eGFP coding sequence.This configuration enables Cre-dependent switching from dsRed to eGFP (Fig. S1B and C).
To generate homeostatic microglia (iPSC-derived MG, or iMG) in vitro, HPCs were treated with cytokines IL-34, transforming growth factor beta 1, and macrophage colonystimulating factor (M-CSF) (Fig. 1A) (44,(50)(51)(52)(53)56). Differentiation of HPC to iMG was followed for 20-25 days, and phenotypes were validated by measuring the surface expression of common myeloid marker CD45 and microglial markers, Iba1 and P2YR12 (Fig S2B).Macrophage-specific CD206 and monocyte-specific CD14 cell surface markers were examined to distinguish our iMGs from macrophages and monocytes.More than 90% of our iMGs expressed CD45 and CD11b, and approximately 80% expressed P2YR12 In vitro HIV-1 susceptibility of microglia derived from Cre-activated fluores cent reporter iPSC Next, we tested whether in vitro differentiated iPSC-derived cells could sustain HIV-1 infection.We constructed an HIV-1 clone that carries the M-tropic JRFL envelope to target all myeloid lineage cells, including microglia (Fig. 1B) (57)(58)(59).We selected the HIV-1 JRFL clone for its known neurotropism, specifically for myeloid lineage cells, including microglia (60)(61)(62).The HIV-1 JRFL was first isolated from the frontal lobe of a patient with HIV-1-associated dementia (60,62).Moreover, studies have demonstrated myeloid lineage cell-specific fusogenic potential of the HIV-1 JRFL (59,63).This virus also expressed Cre instead of the viral Nef, an early viral gene expression indicator.Cre-recombinase-dependent cellular changes occurred only after HIV-1 viral integration (64).Nef expression was restored by inserting an internal ribosome entry site (IRES) upstream of the Nef open reading frame, as previously described (64).Inserting the IRES to restore Nef activity is an efficient strategy for expressing heterologous genes in the context of replication-competent HIV-1 (65).Analogous fluorescent protein-expressing, IRES-carrying viral clones are infectious, capable of mediating high viremia and CD4 + T-cell depletion in vivo in humanized mice (66).
The iMGs were infected with HIV-1 JRFL-Cre, as previously described (59,64,66,67).After 48 h, HIV-1 JRFL-Cre-infected cells started to express eGFP (Fig. 1C and D).On average, we saw approximately 3% of HIV-1-infected cells (Fig. 1E).Many iMG cells continued to express dsRed.Some infected cells even appeared to co-express dsRed and eGFP (Fig. 1D).This dsRed and eGFP co-expression by infected iMGs is likely because eGFP is expressed early upon provirus integration.Dsred has a long maturation time and half-life (68,69).The Cre recombinase-mediated deletion of dsRed in infected iMGs was confirmed by DNA PCR (Fig. 1F).Hence, our fluorescent reporter iPSC line can be successfully used to generate human microglia that could serve as a tool for studying HIV-1 infection at a single-cell level in vitro and, as described below, in vivo.

A novel-humanized mouse dually xenografted with human iPSC-derived microglia and huPBMC
Animal models for studying CNS HIV-1 infection have been limited in elucidating the precise mechanisms governing CNS dissemination and establishing chronic viral reservoirs in vivo.Here, we present a novel mouse model xenografted with iPSC-derived microglia that enables tracing of HIV-1-infected cells at a single-cell resolution and, hence, could shed more light on the HIV-1 life cycle related to CNS HIV-1 infection.We use immunocompromised mice harboring the human MCSF (CSF1) knock-in allele in a Rag2 and Il2rγ knockout background to facilitate successful central engraftment of human microglia.These mice express at least one allele of the human CSF1 gene that critically supports the development of human iPSC-derived microglia in the mouse brain (44,(50)(51)(52)(53)70). Mice lacking Rag2 and Il2rγ genes have no native T-cells, B-cells, or NK cells, making them ideal hosts for xenografted human cells.Genetically modified fluorescent reporter iPSC described above were differentiated into HPC and injected intracranially into newborn pups on postnatal days 0-2, as previously described (44) (Fig. 2A).
To confirm that our genetically engineered iPSC reporter cells colonize the mouse brain as xenografted microglia (xenoMG), we collected the brains of mice (n = 6) that were centrally injected with iPSC-derived HPCs at 8 weeks of age.We observed dsRed-expressing xenografted cells in the cerebral cortex of all six injected mice forebrain.Figure 2B shows a representative coronal section of a mouse brain colon ized with dsRed+ cells.To verify that the engrafted dsRed+ cells are human and in vivo differentiated microglia, we immunohistochemically labeled xenografted mouse brain sections with human-specific nuclear antigen, Ku80 (44), and microglia-specific cell marker, ionized calcium-binding adapter molecule 1 (Iba1) and purinergic receptor P2RY12, antibodies.Indeed, the dsRed+ cells colocalized with Ku80, Iba1, and P2RY12 cell markers (Fig. 2C), confirming that the centrally injected iPSC-derived HPC differentiated into xenoMG.To develop a model system whereby infection of peripheral immune cells spreads to the CNS, xenoMG mice engrafted with iMG at birth were additionally engrafted with huPBMCs at 6-10 weeks of age.In this model, a route that transits the blood-brain-bar rier may be achieved by intraperitoneal (IP) inoculation rather avoiding the need for direct (non-physiological) injection of virus intracerebroventricularly (ICV) (Fig. 2A).This mimics aspects of a physiological route of acute CNS HIV-1 infection in people, where peripheral CD4 + T-cells or monocytes are the initial HIV-1 targets that are then trafficked to the brain to infect microglial cells, which are the primary resident HIV-1 target in the CNS.Dual-engrafted mice were IP injected with in vitro activated huPBMCs.Human leukocytes that populate the mice are predominantly memory T-cells as monitored by human leukocyte-specific cell surface markers: leukocyte common antigen CD45 and T-cell markers CD3, CD4, and CD8 (Fig S3) and are highly susceptible to HIV infection (71).

In vivo HIV-1 infection of xenoMG in humanized mouse brain
Until recently, previously described humanized mouse models for CNS HIV-1 infection were largely limited to a non-physiological route of infection by directly injecting infected cells, or the HIV-1 virus, into the brain (36,38,72).This limitation is due to the inherent tropism of HIV-1 for human immune cells and with the understanding that mouse peripheral immune cells are resistant to HIV-1 infection and, thus, cannot transmit the infection to the CNS unless dually engrafted with human peripheral immune cells.We, therefore, conducted a direct side-by-side comparison of two different modes of infection for neuroHIV: ICV (73) vs IP.
Additionally, we examined whether peripheral engraftment of human immune cells is necessary for HIV-1 infection of central xenoMG when these mice are infected peripher ally.To test the requirement of peripheral human immune cell engraftment, mice pups (n = 18) that were centrally injected with xenoMG precursors were randomized into two groups at 3 months.Group 1 included mice that were dually engrafted with xenoMG in the brain and with huPBMC in the peripheral blood (n = 6) (Fig. 2; Fig. S4), and Group 2 included mice that were singly engrafted with xenoMG in the brain (n = 12) (Fig S4).In Group 1, three mice were ICV-infected, and three were IP-infected with HIV-1 JRFL-Cre.In Group 2, two mice were ICV-infected, and two were IP-infected with HIV-1 JRFL-Cre.Eight remaining mice in Group 2 were used as uninfected control mice.
The six dually engrafted mice in Group 1 reached a median of 20% (range: ~10%-30%) huPBMC engraftment at about 4 weeks post-injection when the mice reached about 3 months of age.Group 1 mice were randomized to two modes of HIV-1 infection, ICV and IP.Peripheral HIV-1 infection was monitored weekly by HIV-1 viral qRT-PCR from mouse cheek blood.The dually engrafted (huPBMC and xenoMG) Group 1 mice had measurable viremia, with HIV-1 RNA levels 2-3 orders of magnitude above the detectability threshold as compared to the singly (xenoMG) engrafted Group 2 mice, regardless of the route of infection (Fig S4B).The HIV-1 viral copy number in the Group 1 huPMBC engrafted mice ranged at 10 1 -10 3 copies of viral RNA transcript per mL plasma (Fig S4B).This would be comparable to early stages of HIV-1 infection in human subjects whose average viral load, if left untreated, could reach 30-50 × 10 3 copies/mL before initiating effective antiretroviral therapies (ART) (74)(75)(76).
We next IP-infected a third cohort of dually engrafted mice (Group 3, n = 30) with HIV-1 JRFL-Cre.Here, we found that peripheral viremia peaked around week 3 or 4 (Fig S5 ; Table S1) when the brains were collected.A subset of mouse brains (n = 8) was for histological analysis to verify central HIV-1 infection.The brain sections of HIV-1-infected dually engrafted mice revealed human nuclear antigen positive (HuNu+) cell nuclei surrounded by a halo of HIV-1 p24 antigen in the cytoplasm, confirming successful infection (Fig. 3A).Remarkably, a subset of HIV-1 p24+ cells presented as large multinucleated HuNu+ cells (Fig. 3A arrowheads), reminiscent of the well-described multinucleated microglial nodules of HIV-1-infected cells in the encephalitic human brain, particularly in cases with severe neuroinflammation and encephalitis (77,78).In addition, cellular morphologies of xenoMGs from HIV-1-infected mouse brains included many cells with amoeboid shape and enlarged somata with fewer and shorter and wider processes reminiscent of similar findings in clinical specimens (77, 79, 80) (Fig. 3A).For additional confirmation of successful xenoMG infection with HIV-1 JRFL-Cre, we performed RNAScope fluorescence in situ hybridization (FISH) using HIV-1-specific oligonucleotide probe.There was robust HIV-1 viral RNA (vRNA) expression in the xenoMGs of HIV-1-infected mouse brains.In contrast, the brains of uninfected mice did not demonstrate any signal above the background (Fig. 3B).

Infection rate and immunological response of xenoMG linked to peripheral HIV-1 viremia
We next measured the extent of HIV-1 infection in xenoMG of mouse brain sections.We dually stained the brain sections with HIV-1-specific oligonucleotide probe for fluorescent in situ hybridization (FISH) and with HuNu antibody for immunohistochem istry.We focused on five representative mice for quantification: four dually engrafted, IP-infected mice (n = 4) and one dually engrafted, uninfected control mouse (n = 1).We calculated the ratio of HIV-1-infected xenoMGs to uninfected cells in these mouse brain sections and measured the peripheral HIV-1 viral load in the mouse plasma (Fig. 4A).The HIV-1-infected xenoMGs were defined as those that were double-positive for HIV-1 vRNA and HuNu (Fig. 4B).We observed xenoMG infection rate from as low as <10% up to 40% in the dually engrafted mouse brain sections (Fig. 4C).There was a significant correlation (R 2 = 0.97) between the percentage of HIV-1-infected xenoMG in the brain and terminal plasma HIV-1 viremia level (Fig. 4C and D).The mouse with the highest plasma viral load (1.2 × 10 4 HIV-1 copies /mL plasma) reached an average of ~30% infection rate in the xenoMG.We also observed that the terminal plasma HIV-1 viremia level significantly correlates (R 2 ≈ 1) with brain HIV-1 viral load (Fig. 4E and F).
It is well-established in the literature that chronic HIV-1 infection is linked to proinflammatory responses underlying clinical comorbidities, including HIV-1-associated neurocognitive disorder (15)(16)(17)(18).Hence, we wondered whether HIV-1-infected xenoMG exhibits higher proinflammatory cytokine expression than an uninfected mouse brain cell.To test this notion, we measured the expression of two proinflammatory cytokines, tumor necrosis factor alpha (TNFα) and IL-6, and the microglial activation marker CD68 in mouse brain tissue extracts by performing qRT-PCR using human-specific primers (81-83) (Fig. 5).As expected, the HIV-1-infected mouse brains (n = 4) expressed higher levels of TNFα, IL-6, and CD68 transcript compared to uninfected mouse brains (n = 4) (Fig. 5).Moreover, the proinflammatory marker expression significantly correlated with the HIV-1 viral load of terminally collected brain tissue (P < 0.05 on 2-sample t-test using HIV-1 infected vs uninfected brain samples) (Fig. 5).Mouse brain with low HIV-1 vRNA copies (defined as <10 copies per 100 ng total RNA) had lower expression of proinflammatory markers compared to mice with higher HIV-1 vRNA copies (defined as >10 5 copies per 100 ng total RNA) (Fig. 5).Hence, the HIV-1-infected mouse exhibited higher proinflammatory response in the brain compared to what was observed in the uninfected mouse brain, and the level of response was significantly associated with the degree of HIV-1 viral transcription.

DISCUSSION
This study presents a novel-humanized mouse model in which mice are dually engrafted centrally with human iPSC-derived and genetically engineered microglia and peripher ally with huPBMC.While these mice can be infected with HIV-1 in the brain via ICV or IP injection, as demonstrated above, the peripheral route offers a significant advantage in modeling the typical route of CNS infection in human cases.A major strength of our model lies in the remarkable versatility of iPSC, including their potential for use in genetic engineering.This includes integrating a transgenic reporter system that enables us to track the transplanted cells over an extended period, up to 4 months after transplantation into the neonatal mouse brain.
The differentiation of iPSCs into microglia, followed by their xenografting into the mouse brain, provides an abundant and renewable source of microglia.This approach also allows for genetic modification before transplantation and subsequent in vivo experimentation, thus facilitating a more intricate understanding of CNS HIV-1 patho genesis.Genetically modified stem cells have proven immensely valuable in HIV-1 research.It enables the evaluation of cellular susceptibility and resistance to HIV-1 infection by altering viral docking sites on the cell surface (84), as well as exploration of cell type-specific transcriptional responses to infection and antiretroviral treatment in a tightly controlled, isogenic environment (85,86).Despite the critical importance of these molecular and mechanistic studies, they have primarily been conducted in vitro or ex vivo, which can limit their clinical relevance.
Our novel-humanized mouse model presents a promising alternative for preclinical and translational research in vivo.It serves as a valuable platform for assessing geneti cally engineered human iPSC-derived cells within the brain, thus paving the way for developing more effective therapies targeting microglia-related HIV-1-associated CNS comorbidities, such as HIV-1-associated neurocognitive disorder (15).Moreover, our model has the potential to generate other unconventional cell-based humanized mouse models for studying HIV-1, including those involving astrocytes.The role of astrocytes in HIV-1 infection and their contribution to viral reservoirs with clinical implications lack a consensus (83,87).Even in the context of end-stage HIV-1 encephalitis, astrocytes harbor only a tiny proportion of HIV vRNA+ cells, accounting for only 0.3%-1% of the total astrocyte population (15,83); consequently, while our model system likely has the potential to generate iPSC-derived astrocyte-humanized mouse model, similar to the approach presented here for microglia, the significance of this model remains to be tested.
To develop novel therapies for treating and potentially eradicating HIV-1, we propose using genetically edited iPSCs targeting promising genes.For example, the CCR5 Δ32 mutation has shown significant promise as a curative target against HIV-1, as demon strated by multiple clinical studies indicating its high resistance to the virus (88)(89)(90)(91)(92). Similarly, SAMHD1, a host restriction factor that impedes HIV-1 reverse transcription in myeloid cells, represents a potential therapeutic target for combating HIV-1 (93)(94)(95)(96)(97). Consequently, our genetically engineered iPSC-based mouse model significantly broadens the repertoire of humanized mouse models for studying CNS HIV-1 infec tion.This expansion in model diversity amplifies our capacity to explore innovative approaches to combat this debilitating disease.
Several other humanized mouse models have been developed to study CNS HIV-1 infection (36-38, 72, 98-100).The hIL34-NOG model involves intrahepatic transplan tation of umbilical cord blood-derived CD34 + HSPCs that differentiate into human microglia in mouse brains over 6-8 months before infection with HIV-1 (36,37,72).The hu-BLT-hIL34-NOG is a modification of the hIL34-NOG model that involves transplan tation of human fetal liver, thymic tissue, and fetal liver-derived CD34 + HSPCs.This model reconstitutes the human immune system in the mouse by 16 weeks of age, and the mouse is susceptible to HIV-1 infection (32,38,41).However, these models are limited by their restricted options for genetic manipulation.Our iPSC-based humanized mouse model allows clonal expansion of genetically modified iPSCs with uniform penetrance of the altered gene.This holds tremendous potential for generating knock-out, knock-in, or conditional transgenes to definitively examine the role of genes within an in vivo context.
In addition, both models require 4-5 months for the complete reconstitution of the human hematolymphoid system before effective HIV-1 infection can occur (2).In contrast, our iPSC-based humanized mouse model offers a more expedited route to achieve CNS HIV-1 seeding.Our model utilizes huPBMCs that engraft rapidly within 4 weeks and efficiently in immunocompromised mice.This shortened timeline from engraftment to infection analysis distinguishes our approach from other humanized mouse models.
On the other hand, immunocompromised mice engrafted with huPBMC generate a predominantly memory and activated T-cell model.While this model supports robust and rapid acute HIV-1 infection kinetics in T-cells, it largely lacks other immune populations including peripheral monocytes and myeloid cells and naïve T-cell populations, yielding a less diverse immune system model.Umbilical cord blood and fetal tissues, rich in hematopoietic stem cells and have a high capacity for tissue regeneration, allow for more complete reconstitution of the human immune system and create a more robust humanized mouse model for studying diseases.Our iPSC-derived xenoMG system, which relies on T-cell-dependent infiltration of HIV-1 into the CNS, provides a platform for investigating HIV-1 infection using alternative approaches, such as transplantation of cord blood-derived HSC instead of huPBMC.The dual engraftment of Cord-blood HSCs and iPSC-derived MG may be a useful strategy to study the interplay between the peripheral immune system and the CNS in a humanized mouse model of HIV-1 infection.Furthermore, our model allows for the simulation of physiological peripheral HIV-1 infection during acute human infection.During this stage, HIV-1 infects immune cells such as CD4 + T-cells or monocytes that traverse the blood-brain barrier, subse quently infecting microglia and disseminating HIV-1 infection in the CNS (101)(102)(103)(104)(105)(106).Consequently, our humanized mouse model, dually engrafted with xenoMG and huPBMC, provides a faster alternative for achieving CNS HIV-1 seeding.It enables widespread microglial infection with proinflammatory gene expression profile activation, as demonstrated here.
Whether the infected microglia produce a virus that can disseminate extracellularly remains to be established in our system or in other humanized mouse models discussed.To date, viral outgrowth with productive infection has been demonstrated for primary microglia from a single human brain specimen, after treatment of the cultured cells with multiple chromatin-modifying drugs designed to broadly activate transcription (107).
Furthermore, our humanized mouse model also has the potential to serve as a valuable tool for quantitatively testing molecular and pharmacologic interventions.This capacity becomes especially important when assessing the impact of ART on HIV-1-infec ted microglia.Post-mortem brain and translational research studies have consistently demonstrated the persistence of HIV-1-infected microglia that are not eliminated by ART (15,(108)(109)(110).This persistence primarily stems from the compartmentalization of infected cells, impeding their complete eradication by ART.Therefore, it is imperative to meticulously dissect the precise kinetics and biology underlying HIV-1 immune evasion, both in the presence and absence of ART, as this ultimately leads to the formation of viral latent reservoirs and chronic HIV-1 infection.Understanding these mechanisms is critical for advancing our efforts toward developing a cure for HIV-1.
A limitation of this study is that despite confirming HIV-1 infection of microglia in the mouse brain through various methods, we have not readily detected Cre-activated GFP reporter transgene expression in the humanized mouse brain at four weeks post-infec tion (data not shown).Possible explanations for this may include inefficient maturation of the GFP fluorophore and/or weaker transgene expression in HIV-1-infected xenoMG.In other acute infection studies in humanized mice, the Cre-activated switch effectively marks HIV-1 DNA-positive cells for several weeks in T-cells (Satija et al. in preparation).To further develop the use of the in vivo Cre-activated switch, additional sequencing studies will be conducted to examine the maintenance of Cre sequence over time.The presence of the reporter transgenes in the virus may affect viral fitness potentially leading to the deletion of the transgene.While these results indicate challenges associated with using a Cre-activated reporter to track HIV-1 infection in vivo, further studies are needed to understand what presently limits this approach's feasibility.
Nonetheless, our Cre-activated reporter-dependent humanized mouse model presents a powerful tool for investigating the pathogenesis of both active and latent HIV-1 infection (Satija et al. in preparation).The Cre-loxP system allows for the tracking of HIV-1-infected cells in vivo, enabling the characterization of viral gene expression and the identification of cellular reservoirs of latent HIV-1.In addition, transgenic reporter systems, including the one presented here, can be utilized to track human cells in the animal host.This will facilitate single-cell RNA sequencing analysis, chroma tin analysis, epigenetic and molecular analysis of HIV pathogenesis, and the testing of therapeutic drug candidates against HIV-1 infection.These tools could be highly informative for examining questions related to latency and active viral expression in brain microglia before and after treatment with current, clinically approved antiretrovi ral drugs.Understanding the precise mechanisms of HIV-1 latency is crucial for devel oping strategies to eliminate latent viral reservoirs, a major barrier to an HIV-1 cure.Thus, utilizing humanized mouse models with genetically engineered iPSCs to study HIV-1 latency could provide invaluable insights into the pathogenesis of HIV-1 and the development of more effective treatments for HIV-1 infection.

Construction and maintenance of fluorescent reporter iPSC
The iPSC lines were constructed at the Black Family Stem Cell Institute.The Cre recombinase-dependent dual fluorescent MSE2104 iPSC line was generated by CRISPR modification of the WTC11 line to insert a dsRed-to-eGFP cassette into the AAVS1 locus within the PPP1R12C gene (chromosomal location 19q13.4-qter)under the control of a CAG promoter.The original WTC11 line was created from PBMCs of a healthy 30-year-old male donor at the University of California San Francisco, and it has been made readily available for use by third parties for research, clinical, and commercial purposes.The MSE2104 iPSC line was maintained by culturing in feeder-free condition incomplete mTeSR E8 medium (StemCell Technologies) in a humidified incubator (5% CO 2 , 37°C) with medium changed every 1-2 days.Cells were passaged approximately every 7 days, dissociating the cells with 0.5 mM EDTA in DPBS and plated onto 6-well plates (Corning) coated with growth factor-reduced Matrigel (1 mg/mL; BD Biosciences) in mTeSR E8 medium supplemented with ROCK inhibitor Thiazovivin (Tocris).Media were switched to mTeSR E8 only medium the next day.

HIV-1 JRFL-Cre clone construction and cell-free HIV-1 virion production
The HIV-1 JRFL clone is a full-length molecular clone of HIV-1 based on NL4-3 (57,88) that expresses the JRFL envelope.Cre is inserted in place of the nef gene, and Nef expression is restored by a downstream internal ribosome entry site (IRES) (81).Plasmid was amplified in Stbl2 electrocompetent E. coli and isolated using a Qiagen Midi-kit.The human epithelial 293T cell line was used to produce HIV-1 virions.293T cells were maintained in Dulbecco's modified Eagle medium (DMEM; Sigma) contain ing 10% heat-inactivated fetal bovine serum (Sigma), 100 U/mL of penicillin (Gibco), 10 U/mL of streptomycin (Gibco), and 2 mM glutamine (Gibco) (complete DMEM).Cell-free virus particles were produced by transfection of 293T cells in a 10 cm dish using Polyjet (Signajen) per the manufacturer's protocol.Virus supernatant was harvested 48 h post-transfection, filtered with a 0.45-µm filter and concentrated by high-speed centrifugation (Sorvall ST 40R centrifuge; ThermoFisher Scientific) at 100,000g for 2 h at 4°C.The pelleted virus was resuspended in DPBS, aliquoted, and stored at −80°C.

HIV-1 p24 ELISA assay
Viral stocks were quantified by NCI HIV-1 p24 ELISA kit.Corning 96-well flat-bottomed plates were coated with anti-p24 capture antibody in 0.1 M NaHCO 3 overnight at 4°C.The plate was blocked with 1% nonfat dry milk (Lab Scientific) for 1 h.The plate was then loaded with p24 standard titrations and experimental virus supernatant treated with 1% Empigen.The dish was incubated for 2 h at room temperature or overnight at 4°C and then washed 6 times with 1× TBS-0.05%Tween (TBST).Alkaline phosphataseconjugated mouse anti-HIV p24 (Cliniqa) was added (1:8,000 in TBST 20% sheep serum) and incubated for 1 h, followed by 6 TBST washes with TBST.The plate was developed with Sapphire Substrate (Tropix), and luminescence was quantitated on a FluoStar Optima plate reader.HIV p24 level was calculated using Prism software (GraphPad), using nonlinear standard curve regression.

iPSC-microglia and hu-PBMC xenograft mouse model
All procedures were performed per the Institutional Animal Care and Use Committee protocol at the Icahn School of Medicine at Mount Sinai.Neonatal immunocompromised mice [C;129S4-Rag2tm1.1FlvCsf1tm1 (CSF1) Flv Il2rgtm1.1Flv/J,JAX ID# 014593] at age P0-P2 were taken out of their home cage and placed on sterile surgical drape overlying a cooling block for 2-3 min to induce hypothermic anesthesia.ICV injection of HPC was performed using a 30G needle fixed to a 10-µL Hamilton syringe.Each mouse received 400-500 K HPCs at four cranial surface coordinates at two different depths, totaling eight different sites (44).The HPCs were resuspended in 1× DPBS at 50-62.5 K cells/μL for injection.Injected mice were allowed to recover on heating pads covered with sterile surgical drapes before being returned to their home cages.Mice were weaned from their mother at P21.At 6-10 weeks, mice hosting the central xenograft were intraperitoneally injected with human PBMCs.PBMCs were obtained from deidentified HIV-1 negative healthy blood donors (New York Blood Center), purified by Ficoll (HyClone) density gradient centrifugation, and maintained in RPMI 1640 medium (Sigma) containing 10% heat-inac tivated fetal bovine serum (Sigma), 100 U/mL of penicillin (Gibco), 10 U/mL of strepto mycin (Gibco), and 2 mM glutamine (Gibco) (complete RPMI).To minimize the donor variability effect, we used the same PBMC donors to inject all nine mice used for this study.PBMCs were activated with phytohemagglutinin-L (PHA-L; 2 µg/mL, Sigma) and IL-2 (50 IU/mL, Roche) for 3 days co-cultured with irradiated feeder PBMCs.Cells were harvested counted, and 10 7 cells resuspended in 200 µL 1× PBS were intraperitoneally injected into each mouse.One week after PBMC injection, engraftment was measured weekly by quantifying human CD45 + cells in each mouse's peripheral blood through fluorescence-activated cell sorting (FACS) on an Attune flow cytometer (ThermoFisher).On average, mice have successfully engrafted with PBMC ~4 weeks after the initial injection.

PBMC engraftment FACS analysis
An Attune flow cytometer (ThermoFisher) was used to measure the level of human PBMC engraftment in our iPSC-microglia and hu-PBMC xenografted mice.The cellular layer separated from the plasma of peripheral blood was treated with ACK lysis buffer (Gibco) to remove red blood cells.Isolated white blood cells were stained with LIVE/DEAD fixable stain (Invitrogen) at a concentration of 1:1,000 in FACS buffer (2 mM EDTA, 2% FBS in DPBS) to detect live cells.Cells were incubated for 30 min at 4°C and then washed with FACS buffer.Cells were stained with 1:100 concentration of CD45 (anti-human PE-Cy7, Biolegend), CD45 (anti-mouse Pacific Blue, Biolegend), CD3 (anti-human APC eFluor780, eBiosciences), CD4 (anti-human APC, Biolegend), and CD8 (anti-human PerCP-Cy5.5,Biolegend) for 30 min at 4°C.Stained cells were washed with FACS buffer and fixed in 4% (wt/vol) PFA for FACS analysis.

HIV-1 infection of iPSC-microglia and hu-PBMC xenografted mice
For HIV-1 infection, each mouse was injected with 250 ng HIV-1 p24 antigen either intracranially or intraperitoneally.For ICV infection, a rodent stereotaxic rig mounted with a micro pump (Stoelting) and Hamilton syringe fitted with a 30G needle was used to inject HIV-1 bilaterally into the PFC (1 µL per hemisphere) (73).The coordinates for injection were as follows: +1.5 mm anterior/posterior, ±0.5 mm medial/lateral, and 1.5 mm dorsal/ventral.The virus was injected per hemisphere at a rate of 0.25 µL per min, and four additional minutes were allowed before syringe removal.Mouse peripheral blood was collected and analyzed weekly for evidence of peripheral infection.Mice were sacrificed 4 weeks post-infection, and tissue samples were harvested and analyzed.

Peripheral HIV-1 infection qPCR analysis
Peripheral blood collected from mice was centrifuged at 10,000g for 10 min to separate plasma from cells.RNA was isolated from plasma using QIAamp Viral RNA Mini kit (Qiagen) and quantified using a NanoDrop Spectrophotometer (ThermoFisher).RNA was reverse transcribed to cDNA using the High-Capacity RNA-to-cDNA kit (ThermoFisher).A Custom TaqMan Gene Expression RT-PCR assay designed for the gag-pol region (Assay ID: AP7DXHY; ThermoFisher) was then used on the cDNA to quantify the HIV-1 viral copy number.A series of 10-fold dilutions of measured HIV-1 target RNA fragments derived from the HIV-1 NL4-3 clone was included in each assay to generate a standard curve to derive the HIV-1 copy number.

Immunohistochemistry of mouse brain sections and confocal microscopy
Mice were anesthetized with isoflurane and monitored for loss of consciousness.Mice that did not respond to toe pinch were cervically dislocated, and their brains dissected.Brains were drop fixed in 4% (wt/vol) PFA for 24 h.Fixed brains were cryoprotected in 30% (wt/vol) sucrose until they sank to the bottom of the solution for at least 48 h.Brains were cut coronally or sagittally at 20-40 μm thickness using a sliding microtome cooled with dry ice.Free-floating tissue sections were collected in 1× DPBS and 0.05% sodium azide.For immunohistochemistry staining, tissues were blocked in 1× DPBS, 0.1% Triton X-100, and 1% BSA for 1 h at room temperature.Tissues were incubated in primary antibodies diluted in 1× DPBS and 1% BSA overnight on a shaker at 4°C.Tissue sections were washed with DPBS three times at room temperature and incubated in fluorophore-conjugated secondary antibodies either for 1 h at room temperature or overnight at 4°C.Tissues were washed with DPBS three times at room temperature and then stained with DAPI (Sigma Aldrich).Tissues were washed with DPBS and then mounted on charged glass slides.Immunofluorescent sections were visualized and imaged using either Zeiss LSM780 or Zeiss LSM980 with airyscan2 confocal microscopes.Brightness and contrast settings were slightly adjusted for better visualization of some images.Primary antibodies: mouse anti-human nuclei (Ku80 1:100; Abcam, ab79220), mouse anti-human nuclei (HuNu 1:50; Millipore, mab1280), rabbit anti-Iba1 (1:100; Wako, 019-19741), goat anti-Iba1 (1:100; Abcam ab5076), and rabbit anti-P2ry12 (1:500; Sigma; HPA014518).

RNAScope Fluorescence in situ Hybridization
Formalin-fixed and cryoprotected mouse brain tissue sections were cut into coronal and sagittal sections (10-20 µm thickness) using a freezing microtome.Sections were mounted on Superfrost Plus slides (Fisherbrand) and dried for 10 min at 60°C.The slides were processed per the RNAScope Multiplex Fluorescent v2 protocol (Advanced Cell Diagnostics).A hydrophobic barrier was drawn around the mounted sections before washing the slides with DPBS.A few drops of anti-sense probe targeting approximately 3 kb of the HIV gag-pol mRNA sequence (317691-C2, Advanced Cell Diagnostics) were added to each section and incubated for 2 h at 40°C.Slides were washed and incubated with polymerizing amplifier sequences conjugated to Opal 520 dye.Slides were washed in DPBS, briefly dried, and mounted with DAPI Fluoromount-G (Southern Biotech).Imaging was performed on a Zeiss LSM780 confocal microscope.

FIG 1
FIG 1 Overview of in vitro iPSC experiments.(A) (Top) Timeline of cellular differentiation of the MSE2104 iPSC-HPC-iMG line harboring the conditional fluorescent reporter transgene at the AAVS1 locus (refer to Fig S1A).(Bottom) Corresponding stage-specific morphological appearances of cell cultures at low power magnification (4×-20×, as indicated).(B) Schematic of the M-tropic JFRL-Cre HIV-1 genome, including Cre coding cassette followed by IRES to drive Nef expression.This viral clone facilitates the Cre-dependent conditional recombination of the dsRed-to-eGFP reporter.(C) Schematic of dsRed-to-eGFP color switch observed in MSE2104 iPSC-derived microglia (iMG) infected with HIV-1 JRFL-Cre.(D) (Top) Conditional eGFP transgene expression in a subset of HIV-1 JRFL-Cre-infected iMG, and (bottom) absence of eGFP expression in uninfected iMG.Cells were infected with 0-20 ng HIV-1 p24 for 48 h.Quantification of HIV-1 JRFL-Cre-infected iMG demonstrated (E) a maximum of 3% HIV-1 infection efficiency in cells infected with the highest amount (20 ng) HIV-1 p24.Error bars are SEM of n = 3.The asterisk (*) represents statistical significance (P < 0.05) on a two-sample t-test comparing HIV-1 infected vs uninfected iMG sample.(Continued on next page)

FIG 2
FIG 2 Humanized mouse model with central xenoMG and peripheral huPBMC engraftment.(A) Timeline of the dual engraftment process and HIV-1 infection of humanized mouse.MSE2104 iPSC-derived HPCs, or precursors of microglia xenograft (xenoMG), were intracranially injected into neonatal mouse brains upon birth between days 0 and 2. Mice were subsequently injected with huPMBC intraperitoneally between weeks 6 and 10.The mice were then infected with HIV-1 about 4-week post-huPBMC injection via one of two routes: ICV or IP.(B) The whole hemisphere view of DAPI (blue) counterstained coronal brain section of a 2-month-old mouse neonatally injected with HPCs (at 4× magnification).A diffuse spread of dsRed+ cells is observed at particularly high densities in the medial temporal lobe, including in the hippocampus.(C) Immunohistochemistry on coronal brain sections of the above mouse at 40× magnification.Layers I and/or II of the adult mouse cerebral cortex demonstrate colocalization of dsRed+ cell signal with human-nuclei-specific Ku80 and microglia-specific markers, Iba1 and P2RY12.All sections were counterstained with DAPI (blue).

FIG 4
FIG 4 Percentage of HIV-1-infected xenoMG is proportional to HIV-1 viremia in the plasma and brain of dually xenografted humanized mice.(A) Schematic of tissue samples collected from representative HIV-1-infected (n = 4) and uninfected (n = 1) mice dually engrafted with xenoMG centrally and huPBMC peripherally.Cortical brain sections of the left hemisphere of the mouse brain were used for staining with HuNu antibody and HIV-1 gag-pol RNAScope FISH.The right hemisphere was used for total RNA extraction to measure central HIV-1 viremia.Cheek blood was collected to isolate plasma to measure peripheral HIV-1 viremia.(B) A representative section from an HIV-1-infected mouse demonstrates HuNu+ cells (magenta) that are HIV-1 infected (arrow), as demonstrated by co-expression of HIV-1 vRNA (green), as well as HuNu+ cells that are uninfected (arrowhead) and, hence, do not express HIV-1 vRNA.(C) The percentage of HIV-1 vRNA+ xenoMG (left y-axis, green) was proportional to the plasma HIV-1 viral load (copy/mL) (right y-axis, red) in HIV-1-infected mice (#1-4 on the x-axis) and uninfected control mouse (Ctl on the x-axis).The asterisk (*) above the bar graphs represents statistical significance (P < 0.05) on a two-sample t-test comparing individual HIV-1-infected mice to the uninfected control mouse.(D) Linear correlation between the percentage of HIV-1-infected xenoMg and complementary plasma HIV-1 viral load.The dotted line represents a fitted linear graph of data points with a significant correlation constant (R 2 ) of 0.97.(E) Brain HIV-1 viral load (left y-axis, (Continued on next page)

FIG 4 (
FIG 4 (Continued)white) is plotted with corresponding plasma HIV-1 viral load (right y-axis, red) in HIV-1-infected mice (#1-4 on the x-axis) and uninfected control mouse (Ctl on the x-axis).(F) Exponential correlation (R 2 = 1) between HIV-1 viral load in the plasma and brain.All error bars in this figure represent the SEM of at least three replicates.