Genetic in vivo engineering of human T lymphocytes in mouse models

Receptor targeting of vector particles is a key technology to enable cell type–specific in vivo gene delivery. For example, T cells in humanized mouse models can be modified by lentiviral vectors (LVs) targeted to human T-cell markers to enable them to express chimeric antigen receptors (CARs). Here, we provide detailed protocols for the generation of CD4- and CD8-targeted LVs (which takes ~9 d in total). We also describe how to humanize immunodeficient mice with hematopoietic stem cells (which takes 12–16 weeks) and precondition (over 5 d) and administer the vector stocks. Conversion of the targeted cell type is monitored by PCR and flow cytometry of blood samples. A few weeks after administration, ~10% of the targeted T-cell subtype can be expected to have converted to CAR T cells. By closely following the protocol, sufficient vector stock for the genetic manipulation of 10–15 humanized mice is obtained. We also discuss how the protocol can be easily adapted to use LVs targeted to other types of receptors and/or for delivery of other genes of interest. This protocol describes production of lentiviral vectors targeted to receptors present on specific cell types, humanization of mice, administration of the lentiviral vectors and detection of the presence of transduced cell types.


Introduction
T lymphocytes are a focus of current research in basic immunology and are used in gene therapy and immunotherapy. Their genetic manipulation is a key technology with applications in both fields. Many different methods have been described, ranging from plasmid transfer by electroporation or chemical methods to the use of viral vectors 1-3 . Retroviral vectors and lentiviral vectors (LVs) have emerged as basic technologies that can be used for genetic manipulations in cell-based gene therapy medicinal products 4 . Those products consist either of hematopoietic stem cells harboring an intact copy of the defective protein for the treatment of patients suffering from genetic diseases or lymphocytes expressing recombinant proteins such as chimeric antigen receptors (CARs) to facilitate cancer immunotherapy 5 . Most methods used for genetic manipulation of T lymphocytes today require modification and expansion of the T lymphocytes ex vivo before infusion into the patient. This is an established clinical approach used in immunotherapy and gene therapy. For the latter, it requires each patient to be treated with an individually produced T-cell product. In addition, manipulating the T cells outside the home organism alters their phenotypes and properties [6][7][8] . An alternative strategy is to genetically manipulate T lymphocytes directly in vivo, in their physiological environment. This strategy has become possible as a consequence of the development of so-called receptor-targeted LVs discriminating between T lymphocytes and other cells at the level of cell attachment and entry 9 .
In this protocol, we describe how to generate vector stocks that transfer CARs to targeted T cells in vivo and administer them to appropriate mouse models. We also provide details on how to humanize and handle mice and monitor for the presence of CAR T cells. We have used this protocol to generate CD19-CAR-specific T lymphocytes in humanized mouse models 10,11 . These in vivo-generated CAR T cells eliminated CD19-positive tumor cells and B lymphocytes even when the CAR was delivered exclusively to CD4-positive T lymphocytes 12 . expression of the gene to a specific cell type. Receptor usage can be manipulated by adding targeting ligands to the vector surface that exhibit high affinity for a cell surface receptor selectively expressed on the desired target cell type. By simultaneously destroying natural receptor usage, gene delivery to non-target cells is eliminated 13 . This approach is limited to certain viral glycoproteins because it requires extensive protein engineering. Suitable glycoproteins can be derived from paramyxoviruses, particularly measles virus (MV) and Nipah virus (NiV). In both viruses, receptor attachment and membrane fusion functions are split across two glycoproteins. This is in contrast to the glycoprotein G of vesicular stomatitis virus (VSV), which is commonly used in conventional LVs and has both functions combined. By engineering paramyxoviral glycoproteins, LVs have been generated that deliver genes selectively into distinct cell types, such as cancer cells, subtypes of neurons or endothelial cells 14,15 . Making use of the distinct surface markers of T-cell subtypes, CD8 for cytotoxic T lymphocytes and CD4 for helper and regulatory T cells, enabled the generation of LVs targeted to specific T-cell subtypes [16][17][18] . CD8-LV carries the NiV G protein fused to a human CD8-specific single-chain antibody (single-chain variable fragment (scFv)). Although the engineered NiV glycoproteins are better incorporated into vector particles, they rely on membrane proximal target receptor binding for proper cell entry 19 . CD4-LV, which binds to the membrane distal domain of CD4 via a human CD4-specific designed ankyrin repeat protein (DARPin), is therefore based on the MV glycoproteins. CD8-LV and CD4-LV have enabled the in vivo delivery of CARs specific for the CD19 antigen present on B-cell malignancies. A single intravenous injection of CD8-LV into humanized mice was sufficient to induce the formation of CD8 + CD19-specific CAR T cells, whereas CD4 + lymphocytes remained unmodified 11 . The in vivo-generated CAR T cells eliminated CD19 + B lymphocytes as well as tumor cells 10 . Conversely, CD4-LV mediated the exclusive generation of CD4 + CAR T cells, which were equally active in eliminating tumor cells 12 .

Applications of the protocol
The in vivo generation of CAR T cells using CD8-LV or CD4-LV is currently the most prominent application of receptor-targeted LVs. This protocol describes the generation and use of a myc-tagged second-generation CD19-specific CAR. A prior version of this protocol has been made available via Protocol Exchange 20 . The protocol can be applied to any other type of CAR and thus is adaptable to the constant improvements in CAR T cell technology. However, the procedure for CAR T cell detection does need to be adapted if different types of immunological tag or reporter gene are used in combination with the CAR.
In addition to targeting cancer cells, another application of CAR T cell therapy is the targeting of chronically infected cells 21 . The humanized mouse model described in the current protocol is engrafted with a spectrum of human cells susceptible to HIV, which leads to chronic viremia accompanied by HIV-induced loss of human CD4 + T cells. It is thus well established for the testing of anti-HIV gene therapies. Moreover, the model recapitulates HIV latency and generates human immune responses similar to those seen in patients with HIV 22 . HIV therapies have predominantly been evaluated in NOD/shi-SCID, γc −/− (NOG) 23,24 or NOD/SCID, γc −/− (NSG) 25,26 mice. Therefore, CD4-LV or CD8-LV delivering CARs directed against HIV can also be evaluated in these humanized mice to determine whether CAR T cells are generated and whether there is subsequent elimination of HIV-infected cells 27,28 .
Beyond CARs, the provided protocol can be used to package any gene of interest into CD4-LV or CD8-LV. Recombinant T-cell-receptor genes are an obvious related example, while antiviral genes, cytokines or interfering RNA are other options, with the latter being relevant for more basic research. When used ex vivo, gene delivery rates in the same range as those reported for CAR genes can be expected. In vivo, however, it must be kept in mind that CARs (and also recombinant T-cell receptors) mediate a selective advantage for transduced T cells, potentially resulting in their preferential proliferation. Accordingly, we have observed between 5-and 25-fold higher rates of transduced cells after delivery of CAR genes compared to reporter genes 11 .
In addition, other types of receptors can be targeted if their extracellular parts are readily accessible by the vector particle. Target receptors can be expressed on other human T-cell subtypes, on completely different cell types or even on cells of other species, such as murine CD8 or CD4. In addition to the requirement for a target receptor, the coding sequence for a suitable targeting ligand, preferably an scFv or DARPin, is required. Once confirmed that the engineered glycoprotein composed of targeting ligand and NiV G protein or MV H protein is well expressed on the surface of packaging cells, the procedure provided here for the generation of CD4-LV or CD8-LV stocks can be followed by exchanging the G or H protein-encoding plasmids. When targeting receptors of other species, it has to be kept in mind that restriction factors may interfere with proper transduction 29 . For primary lymphocytes from non-human primates, switching to simian immunodeficiency virus-derived vectors is a straightforward solution. Both, CD4-LV and CD8-LV, are highly active on non-human primate lymphocytes 17,18 .

Alternative methods for receptor-targeted gene delivery
The other LV-based system for receptor-targeted vectors to date relies on engineered Sindbis virus (SINV) glycoproteins 30 . The engineered SINV-LVs have been targeted to various cell types including human T lymphocytes via CD4 31 or CD3 32 . However, data demonstrating that these vectors can successfully target primary cells is scarce, and the in vivo generation of CAR T cells has not been demonstrated with this technology. SINV-LVs have often been equipped with tissue-specific promotors to achieve sufficient selectivity. Although a side-by-side comparison has not been undertaken, it seems that they do not reach the selectivity demonstrated for engineered paramyxoviral glycoproteins as illustrated, for example, by the discrimination between CD4 + and CD8 + T lymphocytes 33 . An almost absolute selectivity in distinguishing between on-target and off-target cells can be of ultimate importance when toxic or oncogenic genes are delivered. For example, for CARs, the inadvertent delivery of a CD19-CAR into the patient's malignant cells can result in severe adverse events with a fatal outcome, as described for CAR delivery by VSV-LV 34 . Independently from selectivity, paramyxovirus glycoprotein-based LVs follow a different cell entry mode from that of other LVs. SINV-LVs and VSV-LVs require endocytosis of the target receptor for proper entry and transduction. MV and NiV pseudotyped LVs enter cells directly at the cell membrane under neutral pH 35 . Blockage of endocytosis enhances their gene delivery activity, especially when receptors with a high endocytosis rate are targeted 36,37 . Conversely, this means that receptors with absent or low endocytosis rates can be targeted.
There has been enormous progress in the development of non-viral vector systems during the past years. Nevertheless, there was some surprise when the in vivo generation of murine CAR T cells was achieved with synthetic nanoparticles, because this requires stable integration of the delivered genetic information in the T-cell genome 38 . This was achieved by a copackaged transposase and T-cell selectivity via an incorporated CD3-specific antibody. Although this is remarkable progress for the gene-delivery field, it is also obvious that such nanoparticles will require further improvement to reach the activity and selectivity of LVs. It remains to be seen if CD3-targeted nanoparticles can be engineered that deliver CARs not only to murine but also to human T cells.

Alternative mouse models for in vivo CAR T cell generation
The humanized mouse with reconstituted human hematopoietic and immune cells is a powerful tool for investigation of human biological systems and for translational research 39,40 . Humanized mice enable direct access to the dynamics of the human immune-hematopoietic system. Two strains of immunodeficient NOD/SCID mice homozygous for targeted mutations at the Il2rg locus are available: the NOD/SCID/IL2rγc null (NOG) strain 23,24 and the NOD/SCID/IL2rγ null (NSG) strain 25,26 . Compared to NOD/SCID mice, transplantation of human CD34 + cells, also called hematopoietic stem and progenitor cells (HSPCs), into NSG or NOG recipient mice robustly improved the hematolymphoid engraftment efficiency and at least partially supported the maturation of human T and B cells, as evidenced by the development of Ig-producing human B cells as well as human CD4 + and CD8 + T cells in secondary lymphoid organs 24,25,41 . These NSG or NOG mice have become the gold standard for evaluation of human hematopoiesis and immunity in the context of gene and cancer therapy, including CAR T cell therapy 42,43 .
We describe humanization of the NSG model because it is the most widely distributed model used in Europe and beyond. Therefore, the procedure for humanization described here will be widely applicable beyond the in vivo evaluation of targeted vectors. Humanized mice are also commercially available. However, to obtain a homogenous cohort of humanized mice, we highly recommend that they be generated in house. It is important to note that HSPC-humanized NSG mice are unable to develop a functional human innate immune system and thus do not support human myeloid, natural killer cell, erythroid and macrophage lineage development 44,45 .
Two different mouse strains can be used as alternatives. The first is c-Kit receptor-mutant mice (c-Kit mutant mice) on the NSG background, a strain that supports unprecedented levels of human engraftment, including myelo-erythroid differentiation [46][47][48][49] . The second strain is MI(S)TRG mice 50 , which are immunodeficient rag2 −/− IL2rG −/− mice 51 in which the human genes encoding M-CSF, IL-3, GM-CSF and TPO were knocked into their respective mouse loci. In addition, these mice are transgenic for human SIRPα, which allows mouse phagocytes to tolerate human engrafted cells. However, neither of these mouse models expresses human HLA molecules on thymic epithelial cells. Therefore, human T cells developing in CD34 + -humanized NSG mice lack the ability to recognize antigens in a human HLA-restricted manner. However, when engrafted with CD34 + cells and a functional autologous human thymus, education of T cells on human HLA was achieved (BLT-mice) 50,52,53 . The BLT mouse model, although it has major advantages, is much more cumbersome to generate, and human fetal liver and thymic tissue is not easily accessible by many research groups.

Limitations of this protocol
Although selectivity for target cells is excellent for receptor-targeted LVs including CD4-LV and CD8-LV, the amount of particles able to transfer genes in vector stocks often lags substantially behind that of VSV-LV. To compensate for that, it is of utmost importance that the complete production process is performed under the optimal conditions provided in this protocol for the production of CD8-LV or CD4-LV. Notably, the expected yield of particle numbers does not differ between targeted and non-targeted LVs. We therefore assume that the functional activity and stability of the engineered glycoproteins is reduced. Functional titers could indeed be improved by switching from MV glycoproteins to those of NiV, which are three to four times more efficiently incorporated into the lentiviral particle 19 . Likewise, DARPins instead of scFvs improved titers because of the former's higher stability 54 .
In this context, it is important to mention that functional titers provided in transducing units (t.u.) per volume very much depend on the particular experimental conditions and the cell type used. Yet, they are required to confirm the activity of a vector stock as a general quality check. They are, in our experience, rarely predictive for the in vivo performance of vector stocks, especially when comparing targeted and non-targeted vectors. Here, receptor-targeted LVs usually outperform VSV-LV, which attaches to multiple cell types in vivo via the LDL receptor, resulting in a completely different biodistribution 54 . However, also ex vivo, CD4-LV and CD8-LV can reach or even outperform VSV-LV when applied in the presence of transduction enhancers and/or on minimally activated T lymphocytes that express low levels of the VSV receptor 33 .
Often, a single surface marker is not sufficient to define a cell type of choice. Many subtypes of T lymphocytes exist, and they can be distinguished through combinations of particular surface markers. Therefore, receptor-targeted LVs that require more than one cell surface marker are desirable. However, such vectors have not yet been described. The display of two different targeting ligands on the particle surface expands rather than restricts the tropism of that vector, because each target receptor can be used separately. Notably, in vivo selectivity is influenced by many more parameters than just receptor expression. First, cells with a high target receptor density will be preferentially transduced over those with a lower density. Second, local administration routes directly into the tissue of choice (e.g., lymphoid tissues for CD4-LV and CD8-LV) can be an option to prevent encounters of the vector particles with the unwanted cell type. Third, promotor choice, miRNA target sequences and restriction factors expressed in particular cell types can prevent gene expression in some target receptor-positive cell types. An example of CD4-positive cells that are not transduced by CD4-LV are monocytes, which are resistant to LVs that lack vpx, the virion-associated protein encoded by most simian immunodeficiency virus strains and HIV-2 55,56 . These issues should be carefully considered when designing novel types of receptor-targeted LVs.

Experimental design
The process we describe here to achieve in vivo CAR T cell generation can be split into four parts (Fig. 1). The generation of CD8-LV or CD4-LV vector stocks is the crucial first stage to achieve sufficient in vivo gene delivery rates ( Fig. 1a; Steps 1-18). The second part, which runs in parallel to the vector production part, is the generation of homogenously humanized mice (Fig. 1b, phase 1; Steps 33-74). Part 3 includes preconditioning of the mice and vector administration (Fig. 1b, phase 2; Steps 75-78). In the last part, mice are monitored for 1-8 weeks for the production of CAR T cells (Fig. 1b, phase 3; Steps 79-86).

Production and quality control of vector stocks
The generation of CD8-LV and CD4-LV stocks differs from that required for conventional VSV-LVs because they carry paramyxoviral glycoproteins. An additional plasmid is required for the transfection step, and particle concentration requires a more sensitive approach, achieved by sucrose cushion centrifugation to reduce shear forces. We recommend retaining a small aliquot of packaging cell supernatant before concentration for assessment of vector particle activity to control the concentration step. In addition, depending on the stability of the displayed scFv, syncytia formation in  the packaging cells may occur, which makes the timing of particle harvest crucial 57 . Overall, it is recommended to follow the instructions for vector stock production as precisely as possible. The quality of the starting components, especially the plasmids and the packaging cells, is of ultimate importance. As a positive control, VSV-LV stocks should be generated in parallel or in a test run and result in ≥5 × 10 11 particles/ml with 10 8 -10 9 t.u./ml. Using an identical transfer vector plasmid for generation of the VSV-LV control stock is fundamental, because the encoded gene influences the delivery activity of the vector stock. Functionality of the LVs has to be confirmed by incubation of target receptor-positive cell lines with different concentrations of the LV stock to determine the transducing units (t.u./ml) by staining of transgene-positive cells via flow cytometer. For quantitative comparisons necessary for in vivo applications, however, the particle number of the LV stock appears to be more important. This can be measured either by nanoparticle tracking analysis (NTA), which provides very precise particle numbers and the size of the particles, or via p24 ELISA. When producing new types of targeted LVs, we recommend using GFP as a transgene for initial testing of proper gene transfer activity. In addition, selectivity can best be assessed with GFP by comparing gene transfer into target receptor-positive and -negative cells. The amount of GFP in the latter should be similar to that seen for background activity. Expression and incorporation of the glycoproteins should be verified by western blot analysis of vector stocks. As a final quality control stage, CAR-encoding LVs should be tested for functionality on primary cells such as peripheral blood mononuclear cells (PBMCs) to check that CAR T cells develop and kill target cells (e.g., CD19 + cells in the case of CD19-CAR T cells) in vitro.

Generation of humanized mice
Immunodeficient NSG mice lack murine B, T, natural killer and functional myeloid cells and are readily engrafted by CD34 + cells, subsequently resulting in the development of a human blood system. To obtain a robust and homogeneous engraftment, NSG mice are preconditioned with a sublethal dose of busulfan, which reduces the number of residual murine progenitors and creates space in the bone marrow that can be used for the engraftment of human progenitor cells. Human CD34 + cells are injected into the blood stream of NSG mice to humanize them. The quality of the CD34 + cells, the age of the mice, sterile housing and experimental conditions are all crucial to obtain an efficient engraftment, as explained in this procedure.
The final part of this stage of the protocol includes a quality control check of the extent of NSG humanization over time. This is required to identify the optimal time point for vector particle injection. To increase transduction levels in vivo, we describe pretreatment with IL7, which slightly activates the T cells and makes them more permissive to the vector that is subsequently injected. However, if an agonistic CD3-targeted LV is used, IL7 pretreatment can be omitted 37 .
As a complementation to full humanization with blood stem cells, NSG mice can also be engrafted with human PBMCs before LV administration (Fig. 2). This mouse model is less expensive and easier to set up. We therefore use it for an initial analysis of the in vivo performance of newly generated vector types. Moreover, tumor cells can be administered to this mouse model to provide a target for the in vivo-generated CAR T cells. Because of the xenoreactive setting, however, T lymphocytes in this model are more activated than those of fully humanized mice. PBMC-transplanted mice therefore cannot replace, but only complement, fully humanized mice. Detection of in vivo-generated CAR T cells Monitoring of the mice after vector administration, detection of CAR T cells and evaluation of target cell-specific transduction in blood can be performed at different time points, and tissues are analyzed at experimental endpoint conditions. Although protocols for CAR T cell detection and analysis in mouse models have been published, it is important to emphasize that the kinetics of CAR T cell development after in vivo gene transfer differ substantially from that of transplanted ex vivo-generated CAR T cells. Ex vivo-generated CAR T cells are expanded in culture and then transplanted into mouse models in high numbers, which results in an instant high concentration of CAR T cells in vivo. Timing and detection methods therefore have been adapted in this protocol to the kinetics of in vivo CAR T cell generation.
The most direct detection of CAR T cells is flow cytometry using fluorescently labeled antibodies recognizing the CAR. In addition, PCR can be used to detect the integrated vector sequence in the genome of T lymphocytes. The detection of the in vivo-generated CAR T cells can be challenging when the signal-to-noise ratio obtained by flow cytometry is low. It is then crucial to have sufficient numbers of control mice available. Such mice injected just with PBS or control particles are essential to set the gating for the identification of CAR-expressing T lymphocytes.

Biological materials
Plasmids See Fig. 3 for details. All plasmids described here are available either from commercial suppliers (e.g., Addgene) or from the authors upon signing material transfer agreements (MTAs) c CRITICAL All plasmids should be adjusted to a concentration of~1 µg/µl to use similar volumes in each production. We recommend plasmid production by a commercial supplier with certified quality. The absence of RNA and endotoxin ≤100 endotoxin units (EUs)/mg DNA are most important. Alternatively, plasmids can be prepared using Macherey Nagel Maxiprep kits (Nucleo Bond Xtra Midi 100, cat. no. 740410.100) or an equivalent kit. • Transgene plasmid: pS-CD19.CAR-W 11 c CRITICAL This plasmid contains a myc-tag for detection. Other tags, like NGFR, can be used as well. Alternatively, any other transgene plasmid packagable by HIV-derived LVs can be applied. When setting up vectors targeted to other receptors, we recommend using GFP to follow gene transfer activity and target cell selectivity. • Second-generation HIV packaging plasmid: pCMV-dR8.91 58 . This plasmid is commercially available (Life Science Market, cat. no. PVT2323) c CRITICAL Third-generation packaging plasmids can be used as well. In this case, plasmid ratios have to be adapted (see Step 4). • Envelope plasmids used for CD8-LV production: plasmid encoding NiV envelope glycoprotein G fused to a CD8-specific scFv (pCAGGS-NiV-GcΔ34-αCD8opt 19 ) and NiV envelope fusion protein F-encoding plasmid (pCAGGS-NiV-FcΔ22 52 ) • Envelope plasmids used for CD4-LV production: plasmid encoding MV envelope glycoprotein H fused to CD4-specific DARPin 29.2 (pCG-Hmut-CD4.DARPin29.  59 and J76S8ab (CD8-positive Jurkat cells; cells can be provided upon signing an MTA) 18 ) and primary human lymphocytes (Steps 24-32) derived from buffy coats of healthy donors regularly checked for the absence of human pathogens and having provided informed consent c CRITICAL Any cell line expressing the target receptor of the used LV (e.g., CD4 and CD8) at a sufficient level can be used for titration. This should be checked via flow cytometry before first use (see Fig. 4a as an example) ! CAUTION The cell lines used in your research should be regularly checked to ensure that they are authentic and not infected with mycoplasma. • Human CD34 + cells (alternatively called HSPCs) isolated from cord blood obtained from the 'Etablissement Français du Sang' (Besançon, France) upon informed consent. See Reagent setup for details on how to isolate these cells.
• NOD.Cg-Prkdc scid Il2rg tm1Wjl /SzJ (NSG mice; The Jackson Laboratory, stock no. 005557) • Other immunodeficient strains can also be used for immune system humanization. Busulfex conditioning has to be adapted ! CAUTION All experiments using live rodents must conform to governmental and institutional laws and guidelines and be approved by a local ethics committee. We housed our NOD/SCIDgC −/− (NSG) mice in the animal facility 'Plateau de Biologie Expérimentale de la Souris' (Ecole normale supérieure (ENS) de Lyon, Lyon, France). Experiments were performed in accordance with the European Union guidelines upon approval of the animal experimentation protocols by the local ethics committee and the French government (authorization agreement number C2EA-15: Document de saisine du Comité d'Evaluation Commun au CentreLéon Bérard, à l'Animalerie de transit de l'ENS, au PBESet au laboratoire P4 (CECCAPP), Lyon, France).

Reagents
LV production and analysis • DMEM high glucose without L-glutamine with sodium pyruvate (Biowest, cat. no. L0106-500) • RPMI 1640 without L-glutamine (Biowest, cat. no. L0501-500) • FBS, heat-inactivated (56°C, 30 min) (e.g., Sigma, cat. no. F7524) c CRITICAL We test new lots experimentally after adaptation of the cells for ≥2 weeks in medium supplemented with FBS by monitoring cell growth and morphology as well as efficiency of vector production.  Regulatory elements and open reading frames are indicated for each plasmid, respectively. The chimeric intron is derived from introns of chicken β-actin and rabbit β-globin. CAG, chicken beta-actin promoter; CMV, human cytomegalovirus immediate early enhancer and promoter; ICD, intracellular signaling domain; LTR, long terminal repeat; SFFV, spleen focus-forming virus promoter; SIN, self-inactivating; TMD, transmembrane domain; WPRE, woodchuck posttranscriptional regulatory element; βg, beta-globin.

Reagent setup
Medium for cultivation of Lenti-X 293T cells This medium is DMEM supplemented with 10% (vol/vol) FBS and 2 mM L-glutamine (store at 4°C for~1 month).

Medium for medium exchange before transfection
This medium is DMEM supplemented with 15% (vol/vol) FBS and 2 mM L-glutamine (store at 4°C for~1 month). Note that the medium contains 15% (vol/vol) FBS to compensate for dilution with the transfection mixture (containing DMEM without further additives).

Cultivation of Lenti-X 293T cells
Cultivate Lenti-X 293T cells in DMEM containing 10% (vol/vol) FBS and 2 mM L-glutamine. For passaging, detach with 0.25% (wt/vol) trypsin (PBS without Mg 2+ /Ca 2+ /1 mM EDTA). Cells should be passaged twice a week at a 1:8-1:10 ratio. To have enough cells for the production, expand your cells to~18 T175 flasks 4 d before seeding. Check the condition of the cells before seeding. c CRITICAL We strongly recommend cultivation of cells without antibiotics to avoid hidden contamination. LV production is also possible with HEK-293T/17 cells (American Type Culture Collection CRL-11268), but yields will be two-fold lower.
c CRITICAL Cells should not be passaged >20 times.

PEI solution
Prepare PEI solution in several dilutions to achieve the desired concentration. First, prepare a 25% (wt/wt) solution by weighing 5 g of PEI (molecular weight: 25,000 Da); add a triple amount of water and dissolve by mixing. As a next step, prepare a 100 mM (referring to free phosphate residues) stock solution by mixing 0.71 g of the 25% (wt/wt) PEI solution with 39.5 g of water. Finally, prepare the 18 mM (0.92 µg/µl) ready-to-use solution by mixing 9.25 g of the 100 mM PEI stock solution with 35 ml of water, adjust to pH 7 with 1 N HCl (should be~350 µl) and fill up to 50 g with water.

Sucrose solution for concentration of LV stocks
Prepare 20% (wt/vol) sucrose in PBS without Mg 2+ /Ca 2+ and sterile-filter through a Nalgene 0.2-µm filter (can be stored at 4°C for years). c CRITICAL Regularly check for contamination.

Freezing medium
The freezing medium is FBS and 10% (vol/vol) dimethyl sulfoxide (prepare fresh).

Complete stem cell growth medium
This medium is stem cell growth medium supplemented with SCF (100 ng/ml), TPO (30 ng/ml) and FLT3 (100 ng/ml). Stocks or cytokines should be stored at −20°C. c CRITICAL The complete CellGro medium should be prepared shortly before using it.
Blocking solution for FACS staining of in vivo samples (panel 2) Add 1 µl each of mouse and human FcBlock per sample. 2 µl of FcBlock + 48 µl of FACS wash buffer = 50 µl/sample (further referred to as 'Tube 1').
Antibody mixture for FACS staining of in vivo samples (panel 2) Add 1 µl of each anti-human antibody listed in Table 1 to 40 µl of FACS wash buffer per sample. 1 µl × 10 different antibodies + 40 µl of FACS wash buffer = 50 µl/sample (further referred to as 'Tube 2').
c CRITICAL The volume of each mAb may vary depending on the antibody clone, fluorochrome-conjugate and its concentration. Titration of all mAbs is recommended, especially when using other fluorochromes.
c CRITICAL Isotypes are used to determine the background signal from nonspecific binding of the particular isotype of a given antibody to the cells. They are used at the same concentration as the corresponding antibody of interest.
Viability dye for FACS staining of in vivo samples (panel 2) Use a 1:1,000 dilution of Fixable Viability dye eFluor 780 (further referred to as 'Tube 4').
c CRITICAL The standard plasmid can be diluted in advance to a concentration of 1 × 10 9 molecules/µl, divided into aliquots and stored at −20°C in DNA low-binding tubes for several years. Lower dilutions should be prepared fresh and stored at 4°C until usage. The standard plasmid must be handled separately from the actual samples to avoid cross-contamination.
Isolation of human CD34 + cells from cord blood (CD34 + cells are also commercially available (ABCellBio, Lymphobank and Merck/Sigma-Aldrich) and provided cryopreserved (−150°C) in cryotubes.) ! CAUTION Human tissues should be handled using biosafety level 2-recommended protocols and should be collected and used in accordance with all institutional and governmental ethics guidelines. c CRITICAL We use cord blood collected in citrate phosphate dextrose-coated bags obtained from the 'Etablissement Français du Sang' (Besançon, France) upon informed consent.  Resuspend the cells at a maximum concentration 1 × 10 8 cells in 500 µl of PBS buffer. 7 Proceed to magnetic separation with the AutoMACS Pro Separator using the Posseld2 program adapted for CD34 + positive separation according to the manufacturer's instructions. c CRITICAL If no AutoMACS Pro Separator is available, one can instead proceed with Steps 8-11 using manual magnetic separation. 8 Preincubate the MACS separation column with PBS/2% (vol/vol) FBS and let labeled cells pass through this first column placed on a magnetic device. 9 Wash the column once with 2 ml of PBS buffer. 10 Remove the column from the magnetic device to flush out CD34 + cells with 1 ml of PBS buffer using a 5-ml syringe. 11 Repeat Steps 9 and 10 once.
c CRITICAL The purity of CD34 + cells is usually 90-95% using either the manual procedure or the AutoMACS Pro Separator. This should achieve 75-90% cell confluency for transfection. We recommend adding 8 ml of medium to the culture plates, then diluting the cells to a concentration of 2 × 10 6 cells/ml and adding 10 ml of this cell suspension to the plates. We prefer to transfect cells in 15-cm culture plates instead of T175 flasks, although transfection of cells in T175 flasks is also possible. There is a higher risk of contamination in plates; however, they allow a quicker workflow because they are easier to handle, especially when many plates have to be processed. c CRITICAL When handling the plates, be careful not to move your hand or material above an uncovered plate, to avoid contamination. Evenly spread the cells on the plates by carefully tilting the plate back and forth and from left to right after seeding.
Transfection of Lenti-X 293T cells • Timing Day 2, late afternoon c CRITICAL Transfection can also be performed in the morning (if cell density is >75%). In this case, medium is replaced (Step 9) 6-8 h after transfection on the same day. Harvesting of LV particles (Step 10) should still be carried out 2 d after transfection, as described. Alternative transgene plasmids can be used. When setting up LVs targeted to other receptors, we recommend using GFP for validation. Plasmid ratios may have to be adapted when using transgenes of different sizes. For third-generation packaging LVs, plasmid amounts have to be adapted as follows. Instead of using 577.8 µg of pCMV-dR8.91, use 385.0 µg of the plasmid coding for Gag/Pol (pMDLg/pRRE, addgene #12251) and 192.77 µg of the plasmid coding for Rev (pRSV-Rev, addgene #12253). The other plasmid amounts are not changed. 2 Observe cells under the microscope to check for their condition and possible contamination. The cells should be equally distributed over the plate (no cluster formation) with a confluency of 75-90% and should not be growing threedimensionally. c CRITICAL The color of the medium should be orange, not red (low cell density) or turbid yellow (indicating contamination). The cells must cover ≥75% of the plate's surface for optimal transfection. Ideally, they should be above this threshold, although they should not be too confluent because they are in this culture for a further 2 d. Step 3 (medium change of ≥20 plates) should be completed before this step to avoid exceeding this incubation time. 7 After the incubation period, pipette 4.6 ml of the transfection mixture from Step 6 to each plate using a 10-ml pipette. Carefully tilt the plates to disperse the medium evenly. c CRITICAL Avoid disturbing the attached cells by slightly tilting the plate and carefully pipetting the solution slowly to its border. Distribution of the transfection mixture drop by drop is not necessary. Transfect only 10 plates at a time and leave the rest of the plates in the incubator to keep the cells in optimal culture condition. 8 Return the dishes to the tissue culture incubator with a humidified atmosphere containing 5% CO 2 at 37°C and incubate them overnight for 16-18 h. c CRITICAL Stack the dishes horizontally to make sure that cells are completely covered with medium at any time.

Medium change • Timing Day 3, early morning (~16 h after transfection) c
CRITICAL If cells were transfected in the morning, the medium is replaced 6-8 h after transfection on the same day. 9 Gently aspirate the medium from the cells and add 18 ml of DMEM (10% (vol/vol) FBS, 2 mM L-glutamine). Incubate the plates in a tissue culture incubator with a humidified atmosphere containing 5% CO 2 at 37°C for 24 h. Note that when using T175 flasks, 20 ml of DMEM is required. c CRITICAL Take only 10 plates at a time out of the incubator to keep the cells in optimal culture condition. Change the medium of only two plates at a time, to avoid cell dehydration. c CRITICAL The extra volume in the pipette is necessary to avoid bubbles in the tube, because the 5-ml mark of a 10-ml pipette is still visible when being inserted in the centrifuge tube containing the supernatant. c CRITICAL The sucrose cushion should be pipetted very gently to avoid mixing of the two phases. 13 Balance the tubes by addition of PBS without Mg 2+ /Ca 2+ until the weight difference is <0.1 g.
c CRITICAL Weighing can be done outside the cell culture hood, although the tubes have to be kept closed. However, PBS has to be added under the cell culture hood to maintain sterility. 14 Centrifuge for 24 h at 4°C (4,500g; acceleration: 6; deceleration: 6) in a benchtop centrifuge. You can also concentrate the vectors by ultracentrifugation (SW32 Ti Rotor) for 2 h at 100,000g, but due to the volume restrictions, we prefer low-speed centrifugation. If you decide on ultracentrifugation, we recommend using open-top thinwall polypropylene tubes (Beckmann Coulter, 38.5 ml, 25 × 89 mm, cat. no. 326823) for centrifugation of the supernatant. These tubes have to be filled to the maximum to prevent collapsing. If necessary, fill up the supernatant to exactly 30 ml with PBS without Mg 2+ /Ca 2+ before adding 5 ml of sucrose solution. If production is downscaled to <10 plates per vector, concentration of LVs can be performed in 50-ml conical centrifuge tubes. In this case, the supernatant should be aspirated with a 10-ml pipette and filtered with a 0.45-µm syringe filter. One filter can be sufficient for two plates by carefully detaching the pipette without clogging the filter (e.g., by keeping an~3-ml volume in the pipette). The filtered supernatant of two culture plates is transferred in a 50-ml centrifuge tube and underlayed with 4.5 ml of 20% (wt/vol) sucrose by aspirating 7.5 ml of the sucrose solution with a 5-ml pipette and stopping to expel the solution when it hits the 3-ml mark on the pipette. c CRITICAL The centrifugation time should be set for ≥24 h. Set the centrifuge on 'Hold' to be sure that it does not finish before you are ready to proceed.

Resuspension of concentrated LVs • Timing Day 5
15 Carefully remove the tubes from the centrifuge and discard the supernatant. Remove residual liquid by leaving the tube for 5 min upside down in a rack lined with paper towels. Wipe the tubes with fresh paper towels without touching the pellet. Change gloves after that step. A tiny, beige-brownish pellet should be visible at the bottom of the tube. c CRITICAL Pour the supernatant quickly but carefully and avoid bubbles. 16 Add 600-1,000 µl of PBS without Mg 2+ /Ca 2+ to each pellet (equals 60-100 µl/plate) without touching it and let the pellet resuspend for 30 min by placing the tubes on ice on a plate shaker. 17 Pipette up and down 80 times with an electronic pipette. Alternatively, it is possible to use a conventional mechanical pipette for resuspension of the pellet. c CRITICAL Avoid foam formation during resuspension of the pellets, to prevent aerosols. 18 Pool the liquids from all resuspended pellets in one of the conical tubes and distribute aliquots of the vector stock in low-binding tubes in volumes of 15-200 µl (15-25 µl for subsequent NTA or p24 ELISA and titration; 200 µl for subsequent in vivo analyses). Store the vectors at −80°C. c CRITICAL Resuspension of the pellet in too little PBS can be counterproductive for efficient gene transfer. This is a particular issue for LVs containing MV envelope proteins. We recommend the addition of 100 µl of PBS per plate when starting with a new type of targeted LV. j PAUSE POINT LVs can be stored at −80°C. Although formal test results are not available, you can store the LVs for several years in our experience. However, if stored long term, vectors should be tested for functionality in vitro before in vivo application. If you want to proceed directly with the titration, be sure to freeze/thaw the respective aliquot first to make it comparable with LVs used after storage (e.g., for in vivo applications).

Quantification of particle number and gene transfer activity of LVs
Determine particle number via NTA • Timing~1 h per stock c CRITICAL Particle numbers of LVs can also be measured via p24 ELISA according to the manufacturer's protocol. In our laboratory, we often use both methods for each vector stock. c CRITICAL Regularly check PBS solution for the absence of salt crystals before usage. Always sterilefilter PBS for ≥24 h before preparation of LV dilutions, to reduce air bubbles. 19 Determine particle number using your preferred method. We use NTA with Nanosight NS300 (Malvern Panalytical) according to the user's manual, diluting LVs 1:1,000-1:10,000 in sterile-filtered (0.2 µm) PBS without Mg 2+ /Ca 2+ (typically 1:3,000) in a total volume of 1 ml for measurement. Ensure that the particle number per frame ranges between 20 and 50 events for analysis. Perform at least three consecutive measurements for each dilution to obtain robust results. The possibility of measuring the concentration and particle size of LV stocks simultaneously while also seeing inhomogeneity and aggregates is the advantage of NTA. Representative data measured via NTA are shown in Fig. 4b. We provide two different NTA software scripts that can be used (Supplementary Data 1; also available at figshare (doi:10.6084/m9.figshare.13221602)). We prefer to use the script for continuous flow, which is only applicable in combination with a syringe pump (Malvern Panalytical), because it is time saving and simplifies the work flow. If a syringe pump is not available, you can still perform a measurement, but you will have to use the stop flow script. to use will depend on the corresponding transgene. For the example we describe here, an anti-myc-PE antibody was used for detection of the CAR (Fig. 4c). c CRITICAL When producing other targeted vectors, the cell line for titration and the antibody used for detection of the transgene must be adjusted. Vector stocks should have ≥1 × 10 7 t.u./ml (Fig. 4d).

Gene transfer activity on primary cells • Timing 8-11 d c
CRITICAL It is very important to check whether the LVs are able to successfully transduce primary cells. In this section, we describe how to use human PBMCs purified from whole blood. Alternatively, buffy coats from healthy donors, purified via a Ficoll gradient (e.g., Histopaque 1077; Sigma Aldrich, cat. no. 10771) can be used, as described in the manufacturer's protocol. 24 In preparation for the activation of 2 × 10 6 PBMCs, coat one well of a 24-well plate with 500 µl of 1 µg/ml anti-human CD3 mAb (clone: OKT3; Miltenyi Biotec) and incubate for 2 h at 37°C or overnight at 4°C. 25 Remove the medium and replace with 2% (wt/vol) BSA in PBS without Mg 2+ /Ca 2+ (sterile-filtered) for 30 min at 37°C. 26 Wash with PBS without Mg 2+ /Ca 2+ twice. 27 Seed 2 × 10 6 PBMCs in 2 ml of TCM supplemented with 3 µg/ml anti-human CD28 mAb (clone: 15E8; Miltenyi Biotec) and 50 U/ml IL-2 into the plate prepared in Steps 24-26 and incubate the plate for 72 h at 37°C, 5% CO 2 and 90% humidity. c CRITICAL After 72 h, the PBMCs should be activated. c CRITICAL PBMC activation can be adjusted according to the amount of cells needed. For example, you can activate 1 × 10 7 PBMCs in one well of a 6-well plate using 1 ml of anti-human CD3 mAb (1 µg/ml) for coating, and seeding cells in a total volume of 6 ml per well. These mice will be used in Step 51 after 30-36 h. In the meantime, if using fresh CD34 + cells in the next section, proceed with culture of CD34 + cells (as described in Reagent setup).
Cell preparation for human CD34 + cells before injection • Timing 40-50 min c CRITICAL This procedure needs to be performed 24 h before injection into preconditioned mice in Step 56. c CRITICAL CD34 + cells from three to four donors can be pooled to obtain enough cells to humanize a larger cohort of mice with homogeneous engraftment levels. c CRITICAL It is crucial to use CD34 + cells with a purity that is >90% to obtain a high level of human cell engraftment in NSG mice. A contamination of T cells (>3%) might result in low human cell engraftment. 38 Warm up 30 ml of CellGro medium in a 37°C water bath. 39 If using frozen CD34 + cells, take CD34 + cells from the liquid nitrogen tank and thaw them in the water bath at 37°C for 1-2 min. Take cells out of the water bath when they are almost thawed. If using fresh CD34 + cells instead, proceed straight to Step 41. Cell preparation for injection • Timing 1 h 47 In preparation for transplantation, collect cells in 1.5-ml tubes. 48 Count cells using a hemocytometer after trypan blue staining (as described in Step 43). 49 Spin down CD34 + cells at 300g for 10 min at room temperature and resuspend them with sterile PBS without Mg 2+ /Ca 2+ to 0.7--2 × 10 5 cells per 35-μl total volume for each mouse to be injected. The dose of CD34 + cells injected will determine the speed of NSG humanization. If one injects 2 × 10 5 CD34 + cells per mouse, 40% human cell reconstitution in blood should be achieved 12-16 weeks after engraftment. c CRITICAL Cell suspension must be prepared in 35 μl of PBS for retro-orbital injection.

Transplantation of human cells • Timing 1 h (for 20 mice)
c CRITICAL This part of the protocol must be carried out by personnel trained in animal experimentation. c CRITICAL Intravenous injection can also be performed into the tail vein. This has the advantage that there is no need for anesthesia of the mice but does require an appropriate 'contention box' and a mouse-tail illuminator for dilatation of the tail vein.
c CRITICAL It is also important to perform Steps 51-58 on control mice, but instead of CD34 + cells, inject a vehicle using a buffer without cells. 50 Take the CD34 + cells from Step 49 to the animal facility (experimental area). 51 Anesthetize mice. We anesthetize mice using a decontaminated induction box infused with 3.5% isoflurane-enriched air until the mouse is non-responsive and recumbent and demonstrates a slower, even respiratory pattern. Anesthesia is maintained using a nose cone releasing 1-2% isoflurane-enriched air. c CRITICAL Mice should be anesthetized because the needle is being placed in the retrobulbar space. c CRITICAL Be sure to perform anesthesia and retro-orbital injection under a laminar flow (PSM2, sterile conditions). Decontaminate the surface of the laminar flow and your gloves regularly with 70% (vol/vol) alcohol before touching the NSG mice. c CRITICAL It is preferable to use inhalant anesthetic, because it ensures rapid induction and quick recovery times. c CRITICAL Ensure that the mouse is completely under anesthesia before injecting the cells. This usually takes 2-5 min and can be checked by pressing the foot pad to make sure that the mouse has no reflexes. 52 Remove the mouse from the isoflurane chamber and place it on a sterile compress with its belly facing the surface of the laminar flow. c CRITICAL Injection (Steps 54-57) takes <15 s per mouse. Mice can be maintained under anesthesia using a nose cone and 1-2% (vol/vol) isoflurane, but injection can also be performed directly after withdrawing the mouse from the induction box if personnel are trained to perform it quickly. 53 Mix the cells immediately before each injection by pipetting up and down. Load the syringe (29-gauge needles, 0.3-to 0.5-ml insulin syringes) with the cell suspension at a volume of 30 µl per mouse. The same syringe can be used for several mice. c CRITICAL Air bubbles in the syringe must absolutely be avoided before injection into the bloodstream, because this will lead to cardiac arrest. 54 Protrude the mouse's right eyeball from the eye socket by applying gentle pressure to the skin at the dorsal and ventral part of the eye. 55 Introduce the needle bevel down at an angle of~30°into the medial canthus. 56 Inject the cell suspension slowly and smoothly and remove the needle slowly and smoothly once the injection is complete. 57 Make sure that there is little or no bleeding. 58 Place mice back into the cage for recovery. A warming device is not required, because the injection procedure takes only a very short time (<15 s). Thus, the mouse is usually ambulatory within 30-45 s. c CRITICAL The NSG mice are immunodeficient and should be housed in a sterile environment according to national and institutional requirements. Mice must be strictly handled under a laminar flow, get sterile food and water and be kept in sterile cages. This is essential to ensure an efficient high-level humanization of these mice. If this is not respected, opportunistic infections of NSG mice can strongly impair CD34 + cell engraftment. The complete humanization process can take 12-16 weeks depending on the CD34 + cell donor and the number of injected cells. We recommend proceeding to the next step after 8 weeks.
Determination of the humanization level in the peripheral blood (PB) • Timing 2-3 h c CRITICAL We recommend taking blood from the mice every 3 weeks from 8 weeks after injection to monitor humanization in the PB (i.e., at 8, 11 and 14 weeks after injection). 59 Apply a drop of tetracaine 5 min before sampling of blood from the eye.
c CRITICAL Mice can be briefly anesthetized with isoflurane (as described in Step 51) to facilitate blood sampling. 60 Perform retro-orbital blood sampling by penetrating the retro-orbital sinus in mice with a sterile hematocrit capillary tube or Pasteur pipette. c CRITICAL Sterile tubes or pipettes are required to avoid periorbital infections and potential longterm damage to the eye. The eyelid is pulled back to protrude the eye and facilitate blood harvesting. 61 Take~100 µl of blood per mouse in 1.5-ml sterile microcentrifuge tubes containing 20 µl of CPD (for plasma collection and genomic DNA (gDNA) analysis) and immediately pipette 50 µl in a new tube containing 20 µl of CPD and place on ice for FACS staining. j PAUSE POINT Samples can be kept on ice for ≤12 h before proceeding to the next step.
62 Add 50-100 µl of PBS without Mg 2+ /Ca 2+ to the blood (1:1 ratio). 63 Transfer 50 µl of 2×-diluted blood into 5-ml FACS tubes containing 50 µl of FACS wash buffer supplemented with the antibody cocktail described in panel 1 (Table 1). Alternatively, staining and RBC lysis (Step 68) can be performed directly in a 96-well plate (conical bottom), especially if many samples (>10 mice) are processed. 64 Incubate for 30 min at 4°C in the dark. 65 Add 500 µl of FACS wash buffer. 66 Centrifuge at 300g for 10 min at 4°C. 67 Discard the supernatant by inverting the tube and gently tapping it on a paper towel to remove the remaining drop of supernatant. 68 Add 700 µl of 1× RBC lysis buffer for 10 min at room temperature (dark). j PAUSE POINT Cells can be retained in the FACS tubes on ice for 1-2 h before MACS Quant measurement. 73 First, gate the white blood cells by granularity and size (forward scatter versus side scatter). Then, evaluate by gating for hCD45 (negative for mCD45) combined with hCD3 (total T cells), hCD19 (total B cells) or hCD14 (monocytes) using flow cytometry analysis (Fig. 5). 74 Determine the human immune reconstitution by applying the following formula: CRITICAL Use this calculation to identify the time point when the human cell engraftment has become a relevant fraction in blood. The percentage of human cells is more relevant than absolute cell numbers. Animals with ≥40% engraftment are suitable for use in further experiments.

? TROUBLESHOOTING
In vivo CAR T cell generation: IL7 conditioning • Timing 45 min (for 20 mice) c CRITICAL Human IL7 needs to be administered to the humanized NSG mice 4 d and again 1 d before vector application. 75 Resuspend human IL7 according to the manufacturer's protocol. The IL7 stock solution is usually at 100 µg/ml. 76 Prepare a working solution at 2 µg/ml in sterile PBS without Mg 2+ /Ca 2+ . 77 Inject 100 µl of the working solution of huIL7 (200 ng) via the subcutaneous route. To do this, restrain the animal by grasping the skin along its back and insert the needle at the base of the skin fold between your thumb and finger. Administer IL7 in a steady, fluent motion with a 29-gauge 0.5-ml syringe. c CRITICAL Inject the control group with PBS using the same route of administration.
In vivo CAR T cell generation: administration of the vector • Timing 1 h 78 Inject a single dose of 2 × 10 11 LV particles intravenously using the same procedure as for human CD34 + cell injection (Steps 51-58). c CRITICAL A maximal volume of 100 µl can be injected into the eye. Alternatively, up to 200 µl can be injected into the tail vein.
CAR-T longitudinal analysis • Timing 1-8 weeks 79 Monitor mice every 3-4 d. For longitudinal analysis, sample blood from the mice as described in Steps 59-61 every week after the first IL7 injection. Proceed to the next steps to euthanize mice at the chosen time point after LV injection or when clinical endpoints are reached. c CRITICAL Appropriate experimental endpoints (physical appearance, behavioral changes and weight loss) should have been agreed upon and outlined in the protocol that was submitted to the local ethics committee before commencing the protocol.

In vivo sampling and analyses
Euthanization and organ collection • Timing 2 h (for 20 mice) c CRITICAL At least two trained people are necessary for mouse euthanasia and organ collection to enable quick processing of organs before the cells start to die. 80 Anesthetize mice by intraperitoneal administration of 100 mg/kg ketamine/20 mg/kg xylasin. 81 Collect >500 µl of blood in a 1.5-ml tube containing 100 µl of CPD via retro-orbital blood sampling or intracardiac puncture and keep it on ice until further use. c CRITICAL Put 20 µl of blood into a 1.5-ml tube and store at −80°C for gDNA extraction and qPCR (Step 86B). 82 Perform cervical dislocation. 83 Collect mouse tissues (spleen, lymph nodes, liver and others) in FACS wash buffer and immediately place on ice. j PAUSE POINT Organs can be stored at 4°C overnight before cell isolation for the analysis by flow cytometry (Step 86A). The following step (Step 84) needs to be performed immediately. 84 For histology and immunohistochemistry, fix the tissue in 4% (vol/vol) formalin (for 24 h) and wash in 70% (vol/vol) ethanol. j PAUSE POINT Keep the organs in alcohol until paraffin-embedded slices are prepared from the fixed tissues. (xi) Spin at 300g for 10 min at 4°C. (xii) Resuspend the pellet with 5-10 ml of PBS without Mg 2+ /Ca 2+ to reach~1-5 × 10 6 cells/ml. (xiii) Count cells with trypan blue solution (as described in Step 43). Note that the cell suspension should be diluted 20-fold with PBS for cell counting in a 96-well plate. (xiv) Place 1 × 10 6 cells in a FACS tube and 2 × 10 6 cells in a 1.5-ml tube for qPCR for further analysis (Step 86A and B). j PAUSE POINT Cells can be stored on ice for 1-2 h before proceeding with Step 86. Freeze the remaining cells at −150°C in a freezing medium. (C) Mononuclear cell isolation from bone marrow (i) Put the femur into a 10-cm dish. Note that the tibia can also be processed using this procedure.  (Table 2) for the following steps. Note that fluorophores can be exchanged dependent on the laser configuration of the flow cytometer used. The panel can be extended by, for example, adding anti-human CD20 (LT20), anti-human CD69 (FN50), anti-human CD71 (AC102) and anti-human TIM3 (F38-2E2). Naive and stem cell subsets may be additionally identified by including CD45RO. In this case, naïve T cells are identified as CD45RA + CD62L + CD45RO − , whereas stem cells are CD45RA + CD62L + CD45RO + . (i) Wash the purified cells from Step 85A, B and C (there should be 1 × 10 6 cells from each organ in separate micronics) twice with 500 µl of FACS wash buffer and pellet cells at 400g for 5 min at 4°C. (ii) Resuspend the cell pellet with blocking solution (tube 1) and incubate for 10 min at 4°C. (iii) Add 50 µl of prepared antibody mixture (tube 2) or isotype control mixture (tube 3) into the respective well and incubate for 30 min in the dark at 4°C. c CRITICAL Fluorescent minus one controls should be included in case of a spread/ spillover of one channel into the other. Fluorescent minus one controls enable all ambiguity to be removed from the compensated plots and help to distinguish false positive from actual positive signals. (iv) For compensation, use Ultracomp eBeads. This is necessary to compensate spillover signals from channels with overlapping fluorescence spectra. Add 1 µl of each mAb to a drop of the compensation beads in micronics. c CRITICAL Vortex the beads before use. (xi) Proceed to sample acquisition by flow cytometry and analysis by FlowJo or FCS express software (Fig. 6).
? TROUBLESHOOTING (B) Determine the number of vector copies associated with gDNA extracted from huNSG mouse tissues by qPCR • Timing 2 h for enrichment of cells, 3-4 h for isolation of gDNA and 3 h for qPCR (i) Enrich 2 × 10 6 cells from bone marrow or spleen for a CD4 + or CD8 + cell population with the respective MicroBeads (Miltenyi) according to the manufacturer's protocol. This allows the selectivity of the LV for the CD8 + or the CD4 + population by qPCR to be confirmed later in the procedure.    Fig. 6 | Detection of in vivo-generated CAR T cells by flow cytometry. Human cells are identified within tissue obtained from humanized mice as a living single lymphocyte population (left diagrams) that expresses CD45 (top right diagrams). Cells within the CD45 gate are further gated for CD3 and CD19 populations (center right diagrams). Cells within the CD3 gate are gated for human CD4 and CAR expression via its myc tag (bottom right diagrams). Further from the CAR gate, the phenotype or the exhaustion of the cells is determined on the basis of the expression of CD45RA and CD62L or Lag-3 and Tim-3, respectively. The red frame highlights CD19 elimination, the green frame highlights the CAR + signal in the CD4 + fraction and the yellow frame highlights the absence of a CAR signal in the CD4 − fraction in mice administered the CD4-LV vector compared to control. The axis scales for forward scatter and side scatter are linear, and other axes identifying specific populations are in logarithmic scale. These data were published previously in ref. 12 . Animal experiments were performed in accordance with the regulations of the German Animal Protection Law and the respective European Union guidelines (permit number F107/1011, Regierungspräsidium Darmstadt). K, thousand.
j PAUSE POINT Cells can be frozen at −80°C as a pellet before further analysis. Perform the enrichment before freezing the cells. c CRITICAL Start with 2 × 10 6 cells from each bone marrow or spleen sample as a starting material before enrichment to obtain~1 × 10 5 cells for DNA isolation. c CRITICAL Dilute with DNase-and RNase-free water at all steps. c CRITICAL We use WPRE to quantify CAR gene transfer and human albumin as a housekeeping gene. Alternatively, other primers and target sequences can be used. (iv) Transfer 15 µl of the reaction mixture into the wells of a 96-well PCR plate and then add 10 µl of sample containing 100 ng of DNA to the plate without mixing to achieve a total volume of 25 µl. As a control, transfer 10 µl of serial dilutions (from 1 × 10 6 to 1 × 10 0 molecules/µl) to the plate containing the reaction mixture. As a negative control, add 10 µl of DNase and RNase-free water to the plate. c CRITICAL All samples and controls should be measured in duplicates or, preferably, triplicates.
(v) Seal the plate and centrifuge at 20g for 1 min at 4°C to mix samples with the reaction mixture. c CRITICAL 1 × 10 7 molecules in total have been added to your sample when using a volume of 10 µl of the standard plasmid with a concentration of 1 × 10 6 molecules/µl per well. This has to be taken into account during analysis with the LightCycler 480 software, because you have to provide the total amount of standard.

Troubleshooting
Troubleshooting advice can be found in Table 3.  Fig. 4a as an example for sufficient receptor density. Use an alternative cell line for quantification of gene transfer activity The vector stock contains a sufficient number of vector particles (>5 × 10 11 /ml), but they are inactive in gene transfer (<2 × 10 6 t.u./ml) Check concentration step: determine gene transfer activity in unconcentrated harvest (taken in Step 10). Loss of gene transfer activity during concentration should not be more than twofold Check the identity and functionality of all plasmids used, especially surface expression and receptor binding activity of the MV-H-or NiV-G-encoding plasmids 60  Inadequate qPCR due to poor DNA quality DNA might have been lost in preparation (Step 86B (ii)). Determine the DNA concentration and be sure to use an accurate amount of DNA for qPCR. Make sure that DNA is of sufficient quality by gel electrophoresis Inadequate qPCR due to poor design of primers and probe Redesign primers and probe using appropriate software When a positive signal is observed from FACS, this could be a false-positive signal Compensate FACS settings using beads as well as single stainings to avoid spillovers (Step 86A(iv)) When a negative signal is observed from FACS, this could be the result of an inactive vector stock Improve production conditions as described above vector stock. Any therapeutic or gene of interest can be packaged into CD4-LV or CD8-LV using the provided protocol. However, yields of vector particles and gene transfer activities might differ. This holds true also for LVs targeted to surface receptors other than CD4 or CD8, for which the same protocol for vector stock generation can be followed.

Reporting Summary
Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability
Source data for Figs. 4 and 5 are provided. Data shown in Fig. 6 were previously published in ref. 12 . Last updated by author(s): Jan 5, 2021 Reporting Summary Nature Research wishes to improve the reproducibility of the work that we publish. This form provides structure for consistency and transparency in reporting. For further information on Nature Research policies, see our Editorial Policies and the Editorial Policy Checklist.

Statistics
For all statistical analyses, confirm that the following items are present in the figure legend, table legend, main text, or Methods section.

n/a Confirmed
The exact sample size (n) for each experimental group/condition, given as a discrete number and unit of measurement A statement on whether measurements were taken from distinct samples or whether the same sample was measured repeatedly The statistical test(s) used AND whether they are one-or two-sided Only common tests should be described solely by name; describe more complex techniques in the Methods section.
A description of all covariates tested A description of any assumptions or corrections, such as tests of normality and adjustment for multiple comparisons A full description of the statistical parameters including central tendency (e.g. means) or other basic estimates (e.g. regression coefficient) AND variation (e.g. standard deviation) or associated estimates of uncertainty (e.g. confidence intervals) For null hypothesis testing, the test statistic (e.g. F, t, r) with confidence intervals, effect sizes, degrees of freedom and P value noted For manuscripts utilizing custom algorithms or software that are central to the research but not yet described in published literature, software must be made available to editors and reviewers. We strongly encourage code deposition in a community repository (e.g. GitHub). See the Nature Research guidelines for submitting code & software for further information.

Data
Policy information about availability of data All manuscripts must include a data availability statement. This statement should provide the following information, where applicable: -Accession codes, unique identifiers, or web links for publicly available datasets -A list of figures that have associated raw data -A description of any restrictions on data availability We have included a full data availability statement in our manuscript, including a doi to the data repository (figshare) Figures 4 and 5 have associated raw data, which are added as "Source data" and have been deposited in figshare nature research | reporting summary April 2020 Field-specific reporting Please select the one below that is the best fit for your research. If you are not sure, read the appropriate sections before making your selection.

Life sciences Behavioural & social sciences Ecological, evolutionary & environmental sciences
For a reference copy of the document with all sections, see nature.com/documents/nr-reporting-summary-flat.pdf

Life sciences study design
All studies must disclose on these points even when the disclosure is negative.

Sample size
For animal experiments we used sufficient numbers (usually n>5) to achieve statistical relevance.
Data exclusions Mice were excluded from further experiments, if humanization levels are <40%

Replication
Representative data sets based on many biological replicas are shown.
Randomization Not relevant, since all available mice were used for humanization.

Blinding
Blinding not relevant here, since the read-out for CAR T cell presence is quantified by FACS and qPCR which are independent from experimentor's influence.

Reporting for specific materials, systems and methods
We require information from authors about some types of materials, experimental systems and methods used in many studies. Here, indicate whether each material, system or method listed is relevant to your study. If you are not sure if a list item applies to your research, read the appropriate section before selecting a response.

Wild animals
The study did not involve wild animals.
Field-collected samples The study did not involve samples collected from the field.
Ethics oversight NOD/SCIDgC-/-(NSG) mice were housed in the animal facility "Plateau de Biologie Expérimentale de la Souris (PBES)" (ENS de Lyon, Lyon, France). Experiments shown in Fig. 5 were performed in accordance with the European Union guidelines upon approval of the animal experimentation protocols by the local ethical and the French government (Authorization agreement number C2EA -15: CECCAPP, Lyon, France). Experiments shown in Fig. 6 were performed in accordance with the European Union guidelines upon approval of the animal experimentation protocols by the local ethical and the French government (Authorization agreement number C2EA-15: CECCAPP, Lyon, France).
Note that full information on the approval of the study protocol must also be provided in the manuscript.

Flow Cytometry
Plots Confirm that: The axis labels state the marker and fluorochrome used (e.g. CD4-FITC).
The axis scales are clearly visible. Include numbers along axes only for bottom left plot of group (a 'group' is an analysis of identical markers).
All plots are contour plots with outliers or pseudocolor plots.
A numerical value for number of cells or percentage (with statistics) is provided.