Ex vivo long-term expansion of human hematopoietic stem and progenitor cells as a tool for modeling vector integration sites and clonality

TL;DR

This study shows how to expand human blood stem cells in the lab to track how gene therapy vectors integrate into DNA, helping predict potential risks.

Contribution

A novel in vitro model using APU compounds enables long-term expansion of HSPCs to study vector integration and clonal dynamics in human cells.

Findings

APU compounds support long-term expansion of CD34+ HSPCs and maintain stemness for up to 5 weeks post-transduction.

Long-term culture in APU reveals high-risk integrations of mutagenic vectors in genes like MEIS1 and SUSD6.

Xenotransplantation showed reduced clonality compared to in vitro models, suggesting APU enhances clonal diversity.

Abstract

Gene therapy (GT) using retroviral vectors (RVs) is efficacious in treating monogenic diseases. However, there is an inherent risk for severe adverse effects due to insertional mutagenesis. Preclinical safety assessment and patient monitoring are inevitable in GT. To assess the genotoxic risk of novel RV vectors, mainly murine hematopoietic stem and progenitor cells (HPSCs) are routinely used, because human HSPCs cannot be immortalized in vitro using mutagenic vectors. In this study, we aim to identify early signs of clonal outgrowth by performing integration site analyses (ISA).

The small molecules A83-01, pomalidomide, and UM171 (APU) were used for the ex vivo expansion, lentiviral transduction, and long-term cultivation of umbilical cord blood-derived HSPCs. We determined the influence of APU on the stemness of HSPCs and their differentiation capacity via single-cell RNA sequencing (scRNA seq) and in xenotransplantation studies. To track vector insertion site dynamics, we transduced 7-day expanded HSPCs with a mutagenic or a safer RV. ISA was conducted in human HSPCs over a 5-week cultivation in vitro and compared to the bone marrow of xenotransplanted mice to assess clonal skewings.

APU supported the expansion of CD34+CD38−CD45RA−CD90+EPCR+ HSPCs. scRNA seq confirmed the enrichment of HSC signature genes in APU-expanded HSPCs compared to the clinically used medium SFT3 (SCF, FLT3-L, TPO, IL-3). After RV transduction, APU still maintained around 30% of CD34+ cells for 5 more weeks. Without the compounds, already 2 weeks post-transduction, less than 10% of cells were CD34+. The long-term culture allowed the detection of high-risk integrations of the mutagenic SIN-LV.SF in MEIS1 or SUSD6 due to their increasing abundance over time. Bone marrow of xenotransplanted mice was less clonal but did not support the outgrowth of insertional mutants. Overall, APU increased clonal diversity.

Our findings propose that long-term cultivation of transduced HSPC in APU allows for outgrowth of clonal integration sites. The decrease of clonality has been observed in gene therapy patient’s years after treatment. Thus, the in vitro model could be used to develop novel human HSPC-based genotoxicity assays that predict insertional mutagenesis, in addition to existing preclinical biosafety assays.

The online version contains supplementary material available at 10.1186/s12967-026-07700-6.

Figures

C). Cryopreservation of expanded cells accelerated the decrease in viability after thawing. However, the expansion of CD34^+^CD38^−^CD45RA^−^CD90^+^ cells was still superior in conditions with Poma (Supplementary Fig. 1D). Hence, we titrated Poma and analyzed the level of cells in early and late apoptosis after 14 days (Annexin V^+^ DAPI^+^ staining - Supplementary Fig. 1E). The frequency of Annexin V^+^ cells correlated with the concentration of Poma. The 10-fold lower concentration (0.2 µM) of Poma resulted in good viability while still maintaining the HSC immune phenotype in APU conditions.

C-D). Over time, a subset of the cMEMP cluster highly expressed *GP1BB*, *CAVIN2*, *HBD*, and *PPBP*, linked to platelet and erythrocyte differentiation (Supplementary Fig. 2E). We could confirm the myeloid differentiation potential of APU-exp HSPCs and their mildly reduced erythroid differentiation bias via colony forming assay (Supplementary Fig. 2F-H). In contrast, Poma promoted the differentiation of cDCs (Fig. [4](#Fig4)A). After 14 days, Poma further promoted the differentiation of monocytes to an inflammatory or activated state (Supplementary Fig. 3A-C). Generally, the presence of some

Effect of APU on HSPCs after 24 h cultivation on the single-cell level. (A) UMAP of uncultivated and 24 h-cultivated HSPCs and their individual cluster abundances in the bar plot below. (B) Feature plots of the HSC signature genes *HLF*, *MECOM*, *NKAIN1*, *PROM1*, *AVP*, and *CD34*. (C) Dotplot comparing the average expression of HSC signature genes among the 24 h-cultivated HSPCs. (D) Volcano plot displaying the log2 fold changes (FC) and the -log10 p values of DEGs between HSPCs cultivated for 24 h in SFT + APU (APU_24h) compared to SFT (SFT_24h). Significant (*p* > 0.05, Wilcoxon Rank Sum

A-B, Supplementary Fig. 7B), the cells only caused a sufficient human chimerism (> 5%) in 4/18 mice: Only 1/9 mice transplanted with SFT3-exp cells and 3/9 mice with APU-exp cells showed substantial engraftment (Supplementary Fig. 8A). Since our experience in the previous cohorts was based on CD34^+^CD38^−^CD45RA^−^ cells as starting material (Supplementary Fig. 4–6), the frequency of engraftable HSC was underestimated here using CD34^+^ cells (Supplementary Fig. 8A). The mCherry expression within the hCD45^+^ cells in the BM was either abundant in virtually all cells (> 97% for mouse 63 and 6

Increased lentiviral transduction efficiency in APU-exp cells. (A) Graphical scheme of the combined transduction (td) and expansion protocol for UCB-derived CD34^+^ cells. Td was performed with 1 × 10^5^ CD34^+^ 24 h post-thawing with a self-inactivating lentiviral vector encoding for mCherry driven by the CBX3.EFS promoter (SIN-LV.CBX3.EFS.mCherry) in APU/SFT3 medium supplemented with protamine sulfate and Synperonic^®^ F-108 in a U-bottom 96-well plate. On day 1 post-td, HSPCs were seeded for expansion. Td efficiency and stem cell phenotype were determined on day 6 post-td, and vector copy n

APU_0.2 allows long-term cultivation of CD34^+^ cells up to 5 weeks after lentiviral transduction. (A) Experimental setup of transducing expanded HSPCs. Purified CD34^+^CD38^−^CD45RA^−^ cells were expanded for 7 days either in SFT3 or APU_0.2 medium before transduction (td) with the SIN-LV.SF.eGFP (SF) or the SIN-LV.EFS.eGFP (EFS) vector. 24 h post-td, HSPCs were either transplanted into immunocompromised mice or further cultivated. Long-term culture allowed downstream analyses. (B) Expansion factor after 7-day culture in SFT3 or APU_0.2 medium determined by Casy. Paired t-test revealed non-si

Ex vivo long-term cultivation enables the modeling of viral integration sites in HSPCs. (A) Diversity and clonality of vector integration sites (with VCN > 0.02) in in vitro-cultured HSPCs (24 h, 1–6 weeks post-td) and in the BM of xenotransplanted mice, 16 weeks post-transplantation, measured by the Gini and Shannon indices. Dot size indicates the inverse of the UC50 value, which stands for the number of insertions present in 50% of the cells within a sample. (B-C) Unique integration sites (UIS) during long-term in vitro culture in SFT3 or APU_0.2 medium after transduction with LV-EFS.eGFP (E

Integration site analysis (ISA) reveals higher clonality in APU_0.2-transduced HSPCs after hematopoietic reconstitution. (A) Analysis of bone marrow (BM) samples 16 weeks post-transplantation. VCN (x-axis) and eGFP expression within hCD45^+^ cells (y-axis and color-scaled) are plotted with dot sizes that correspond to the human chimerism (hCD45^+^). Correlation between factors was calculated by the Spearman r index. (B) Number of UIS in the BM of mice transplanted with transduced HSPCs (EFS or SF) either in SFT3 or APU_0.2 medium. P values determined by two-way ANOVA using Šídák’s multiple com

Full text

Background

Gene therapy (GT) using retroviral vectors (RVs) is a promising treatment option for patients suffering from monogenic diseases. In the past 25 years, over 55 clinical GT studies demonstrated the effective and relatively safe treatment of primary immunodeficiencies, bone marrow (BM) failure syndromes, hemoglobinopathies, and inborn errors of metabolism [1]. To date, four FDA-approved self-inactivating (SIN) lentiviral (LV) therapies that modify hematopoietic stem and progenitor cells (HPSCs) ex vivo for the treatment of genetic diseases are currently on the market [2]. Despite efforts to improve RV designs with SIN vectors, weaker promoters, and genetic insulators, hematopoietic stem cells (HSCs) are highly affected by the risk of insertional mutagenesis due to their high self-renewal capacity [3]. In the clinical study testing the SIN-LV “Lenti-D”, which encodes for the ATP-binding cassette subfamily D member 1 (ABCD1) gene to treat cerebral adrenoleukodystrophy, 7 of 67 patients developed hematological cancers 14 to 92 months after treatment as of 2024 [4]. Insertion site analysis (ISA) revealed vector integrations in high-risk loci, such as the myelodysplastic syndrome 1 protein (MDS1) and ecotropic virus integration site 1 (EVI1) complex locus protein (MECOM-EVI1) or PR/SET Domain 16 (PRDM16), promoting transformation. Additionally, the adverse effects of insertional mutagenesis might only be visible in certain cell types. For example, integrations into LMO2 have been reported to promote clonal T cell proliferation [5], which is also an important aspect to consider in the growing field of immunotherapies using chimeric antigen receptors to equip T cells. Hence, preclinical genotoxicity assessment of novel RVs and the monitoring of insertion sites in patients after GT are essential to prevent future severe adverse events. Animal models transplanting transduced murine HSPCs can shed light on the clonal development of vector integration sites and the risk for leukemic events in vivo [6–8]. Unfortunately, the time period to detect transformation after transplantation in wild-type or disease-model mouse strains can be longer than a year or only detectable after serial transplantations. The tumor-prone mouse model cyclin dependent kinase inhibitor 2 A (Cdkn2a^−/−^) was successfully used to increase the likelihood for insertional transformation of RV-transduced murine HSPCs in a shorter period of time. Due to the knock-out of the tumor suppressor gene, the burden on the animal strain is high, as the mice always develop cancer [9]. The risk of RV-mediated insertional mutagenesis can also be tested in vitro using cell culture-based genotoxicity assays. However, the FDA/EMA-accepted in vitro immortalization assay (IVIM) and the surrogate assay for genotoxicity assessment (SAGA) also require murine HSPCs [10–12]. Despite the high sensitivity of IVIM and SAGA, they do not fully reflect the insertional landscape and the associated side effects in human HSPCs. There is an evident lack of predictive preclinical model systems using human HSPCs. A first human-based in vitro genotoxicity assay, ^h^InGeTox, was introduced by Suleman and colleagues [13]. ^h^InGeTox aims to score the oncogenic risk of LV vectors based on integration sites, differentially expressed genes (DEGs), deregulated signaling pathways, methylation changes, and the identification of truncated genes caused by the vector. However, the assay leverages human induced pluripotent stem cells and hepatocyte derivatives as a starting material. How predictive this material will be for hematopoietic stem cell-based therapies remains to be shown.

Developing humanized genotoxicity assays with HSPCs is hampered due to the difficulty of transforming the cells in vitro. Moreover, CD34^+^ cells quickly differentiate in culture. It is hard to maintain stem and progenitor cells until insertional mutants can become dominant. Small molecules, such as the pyrimidoindole derivative UM171, enable the expansion of umbilical cord blood (UCB)-derived HSPCs [14, 15]. UM171 can reduce stress responses during ex vivo culture after transduction and increase engraftment success with an increased clonality of transduced HSPCs in immunocompromised mice [16, 17]. Zonari and colleagues also demonstrated the beneficial effects of UM171, Prostaglandin E2 [18], and StemRegenin 1 [19] on the transduction efficiency and ex vivo expansion of HSPCs [20]. In 2016, Wataru Ebina from the Rossi lab proposed the use of A83-01 and pomalidomide as a triple combination (APU) to further elevate HSC expansion [21]. A83-01 inhibits the evolutionary, highly conserved transforming growth factor beta (TGFβ) pathway and pomalidomide (Poma) targets Notch signaling [22–24], which both influence quiescence, proliferation, and differentiation of HSCs [25]. Ebina observed a ~ 200-fold expansion of CD49f^+^ HSC after 14 days in APU medium [26]. The molecular effects of APU on HSCs have not been investigated so far but could be the basis to maintain stem and progenitor cells long enough in culture to screen for early signs of clonal dominance after RV transduction.

After optimizing the APU media composition and culture conditions to our needs, we conducted an in-depth characterization of APU-expanded (-exp) HSPCs at the single-cell level to investigate the potential of the expanded HSPCs for their use in genotoxicity assessment of RVs in vitro. Since regulatory authorities also demand animal studies as part of preclinical safety assays, we also transplanted the transduced HSPCs into immunocompromised mice. The in vitro long-term culture for over 5 weeks post-transduction enabled us to model the increase of clonal integration sites, representative to dynamics seen in patients months or years after GT [27].

Methods

Human umbilical cord blood (UCB) cells

UCB was kindly provided by the prenatal medicine and obstetrics facility at Hannover Medical School (MHH) from terminal deliveries and with the consent of the families in conformity with the standards of the MHH Ethics Committee and the Helsinki Declaration of 1975 under the revision of 1983. The study has approval from the ethics commission of MHH to use the cells for the described purposes (approval #2791 − 2015). Peripheral blood mononuclear cells (PBMCs) were isolated with the Biocoll^®^ separation solution (Bio&SELL, Feucht, Germany) for density gradient centrifugation and were enriched for CD34^+^ cells with the magnetic-activated cell sorting (MACS) kit (Miltenyi Biotec, Bergisch Gladbach, Germany) according to the manufacturer’s protocol and cryopreserved until further use in 90% fetal calf serum (FCS) and 10% Dimethylsulfoxid (DMSO). Fluorescence-activated cell sorting (FACS) for CD34^+^CD38^−^CD45RA^−^ cells was performed with the FACSAria Fusion cell sorter (Becton-Dickinson, Franklin Lakes, NJ, USA). Thus, CD34^+^-enriched HSPCs were thawed and blocked with Fc receptor blocking reagent (Miltenyi Biotec) before staining with antibodies against CD34 (APC, clone HG12, catalog #345804, BD Biosciences, Franklin Lakes, NJ, Unites States), CD38 (FITC, clone HB7, catalog #11-0388-42, Thermo Fisher Scientific, Waltham, MA, USA), and CD45RA (BV605, clone HI100, catalog #304134, BioLegend, San Diego, CA, USA). For dead cell exclusion, samples were resuspended in FACS buffer with 0.2 µg/ml DAPI (Sigma-Aldrich, Saint Louis, MO, USA).

Expansion of UCB cells

Purified CD34^+^or CD34^+^CD38^−^CD45RA^−^ cells were cultivated at low cell density (5000 cells/ml in a 96-well plate). StemSpan™ serum-free expansion medium (SFEM, STEMCELL Technologies, Vancouver, Canada) was supplemented with 100 U/ml penicillin and 100 µl/ml streptomycin (PAN-Biotech, Aidenbach, Germany). Additionally, stem cell factor (SCF), FMS-like tyrosine kinase 3 ligand (FLT3-L), thrombopoietin (TPO), and interleukin-3 (IL-3) were purchased by Peprotech (part of Thermo Fisher Scientific) and used at 100 ng/ml (referred to as “SFT3”). SFT3 was supplemented with 1 µM A83-01 (“A”, Tocris, Bristol, United Kingdom), 2 µM or 0.2 µM Pomalidomide (“P” or “Poma”, Selleck Chem, Houston, TX, USA), and 35 nM UM171 (“U”, APExBio, Houston, TX, USA) and referred to as “APU” [21]. Half-medium changes were performed twice a week, and wells were expanded or split if wells reached 70–80% confluence. Expanded cells were counted either with the Casy cell counter (Roche Innovatis/Schärfe System, Reutlingen, Germany) or the CellTiter-Glo^®^ luminescent cell viability assay (Promega, Madison, WI, USA). Based on the luminescence, the ATP content was determined. The number of cells was assessed by interpolating the ATP content to a standard curve of known cell numbers.

Flow cytometry

For immunophenotyping, expanded HSPCs were stained with targeted fluorophore-coupled antibodies against CD34 (APC, clone HG12, catalog #345804, BD Biosciences), CD38 (BV605, clone HB7, catalog #562665 BD Bioscience), CD45RA (BV605, clone HI100, catalog #304134, BioLegend), CD90 (PE-Cy7, clone 5E10, catalog #328124, Biolegend), endothelial protein c receptor (EPCR) or CD201 (BV786, clone RCR-252, catalog #743556, BD Biosciences), and C-X-C motif chemokine receptor 4 (CXCR4; PE-Cy5, clone 12G5, catalog #306508, Biolegend). To characterize differentiation, expanded cells were stained for CD11b (PE-Cy7, clone ICRF44, catalog #301322, BioLegend), CD14 (AF700, clone 63D3, catalog #367114, BioLegend), CD33 (BV605, clone P67.6, catalog #366612, BioLegend), CD123 (APC, clone 7G3, catalog #560087, BD Biosciences), CD133 (PE-Dazzle594, clone 7, catalog #372812, BioLegend), and CD34 (BV785, clone 561, catalog #343626, BioLegend). Flow cytometric analysis was performed with the Cytoflex S (Beckman Coulter, Brea, CA, USA) cytometer and data were analyzed with the CytExpert (Beckman Coulter) and FlowJo (Tree Star, Ashland, OR, USA) software.

ScRNA seq

7-day expanded and uncultivated cells were multiplexed in a cell hashing-based experimental approach via oligo-tagged antibodies (Experiment ID: S01-S03). Accordingly, cells originating from independent expansion conditions and the uncultivated CD34^+^CD38^−^CD45RA^−^ cells were labeled with differently tagged TotalSeq™-B Hashtag Derived Oligo (HTO) antibodies from Biolegend (TotalSeq™-B0251 anti-human Hashtag 1 Antibody [catalog #394631, barcode: GTCAACTCTTTAGCG], TotalSeq™-B0252 anti-human Hashtag 2 Antibody [catalog #394633, barcode: TGATGGCCTATTGGG], TotalSeq™-B0253 anti-human Hashtag 3 Antibody [catalog #394635, barcode: TTCCGCCTCTCTTTG], TotalSeq™-B0254 anti-human Hashtag 4 Antibody [catalog #394637, barcode: AGTAAGTTCAGCGTA], TotalSeq™-B0255 anti-human Hashtag 5 Antibody [catalog #394639, barcode: AAGTATCGTTTCGCA]) with a concentration of 0.1 µg/1 × 10^6^ cells prior to sample pooling. Antibody-labeled cells were sorted for DAPI-negative living cells with the FACSAria Fusion cell sorter (Becton-Dickinson), combined in three separate pools, and loaded on three lanes of a Chromium chip (10x Genomics, Pleasanton, CA, USA). Library preparation for single-cell mRNA seq was performed according to the Chromium NextGEM Single Cell 3ʹ Reagent Kits v3.1 CellSurfaceProtein User Guide (Manual Part Number CG000206 Rev D, 10x Genomics). Fragment length distribution of generated libraries was monitored using the Bioanalyzer High Sensitivity DNA Assay (catalog #5067 − 4626, Agilent Technologies, Santa Clara, CA, USA). Libraries were quantified using the Qubit^®^ dsDNA HS Assay Kit (catalog #Q32854, ThermoFisher Scientific). Equal molar proportions of generated mRNA expression libraries were pooled and sequenced by Novogene on an Illumina NovaSeq6000 sequencer using one S4 flowcell with 300 cycles. Sequencing was performed according to the following settings: 150 bp as sequence read 1 and 2; 8 bp as index read 1; no index read 2. The data processing of S01-S03 and the details for the run with 14-day expanded HSPCs (Experiment ID: N3917-N3918) were described in the Supplementary Material and Methods.

Gene set enrichment analysis (GSEA)

GSEA was performed on differentially expressed genes (DEGs) determined by the Seurat function “FindMarkers”. Gene sets were downloaded from the Molecular Signature Database (MSigDB: https://www.gsea-msigdb.org/gsea/msigdb/human/genesets.jsp). The normalized enrichment score (NES) was calculated on the pre-ranked DEGs with the R package “fgsea” [28].

Xenotransplantation studies

8- to 12-week-old NOD.Cg-Kit^W− 41J^ Tyr^+^ Prkdc^scid^ Il2rg^tm1Wjl^/ThomJ (NBSGW) mice from the Jackson Laboratory (Bar Harbor, ME, USA) were transplanted with human UCB-derived HSPCs without irradiation. Either 1 × 10^4^ uncultivated CD34^+^CD38^−^CD45RA^−^ cells or the expanded yield after 7- or 14-days were thawed and resuspended in 150 µl PBS for transplantation into one mouse. Cells were slowly injected into the tail vein of the mice using a 25G syringe. Mice received broad-spectrum antibiotic prophylaxis for two weeks by adding 10 mg/ml Ciprofloxacin (Fresenius Kabi, Bad Homburg von der Höhe, Germany) to their drinking water. Blood was collected after 8, 12, and 16 weeks from the retro-orbital sinus to monitor the presence of human CD45^+^ (hCD45^+^) hematopoietic cells by flow cytometry. Thus, blood samples were depleted for erythrocytes by using the Pharm Lyse™ Lysing Buffer (BD Biosciences) before Fc receptor blocking with the human (TruStain FcX™, catalog #422302, Biolegend) and murine (anti-CD16/CD32, catalog #14-0161-82, eBioscience, San Diego, CA, USA) Fc receptor blocking solution. Hematopoietic cells were stained with human anti-CD45 (BV650, clone HI30, catalog #563717, BD Biosciences) and murine anti-CD45 (mCD45, PerCp-Cy5.5, clone Ly-5, catalog #12-0451-83, eBiosciences) antibodies. Additionally, the abundance of T cells (anti-CD3, APC, clone OKT3, catalog #48-0037-42, eBioscience), B cells (anti-CD19, APC/Cy7, clone HIB19, catalog #302218, Biolegend), myeloid cells (anti-CD33, FITC, clone HIM3-4, catalog #303304, Biolegend), and NK cells (anti-CD56, Pe_Cy7, clone NCAM16.2, catalog #335826, BD Biosciences) was determined within the mCD45^−^hCD45^+^ fraction. The final analysis was performed 16 weeks post-Tx. BM cells were isolated by flushing tibiae, femora, and iliac crests. Spleen was harvested and sieved through a 70-µm cell strainer (Sarstedt, Nümbrecht, Germany) to collect a single cell suspension. Finally, human chimerism and immune cell distribution were determined in BM, peripheral blood (PB), and the spleen using the aforementioned flow cytometry stainings.

Transduction of freshly thawed CD34+ cells

CD34^+^ cells were thawed and seeded in a 48-well plate with a cell density of 1 × 10^5^ cells/ml either in SFT3 or APU medium. The next day, 1 × 10^5^ cells were transduced at an MOI of 10 in 100 µl medium (U-bottom 96-well plate) with the lentiviral SIN.LV.CBX3.EFS.mCherry vector, which is driven by the elongation factor 1-alpha promoter [29]. The medium was supplemented with 4 µg/ml protamine sulfate (Thermo Fischer Scientific) and 1 mg/ml Synperonic^®^ F-108 (catalog #07579, Sigma-Aldrich) as transduction enhancers [30]. After 16–18 h, HSPCs were washed and expanded to ten wells of a 96-well plate (flat-bottom) with fresh SFT3 or APU medium. On day 4 post-transduction (post-td), we performed a half-medium change before the cells were cryopreserved on day 6 post-td.

Transduction of expanded HSPCs

7-day expanded HSPCs were counted to seed 1 × 10^5^ cells per well of a flat-bottom 96-well plate for transduction. Cells were transduced either with the SIN-LV.SF.eGFP [31, 32] vector or the SIN-LV.EFS.eGFP [29, 33, 34] vector at an MOI of 5 and with RetroNectin (48 µg/ml, Takara Bio, Kusatsu, Shiga, Japan) and Synperonic^®^ F-108 (1 mg/ml, Sigma-Aldrich) as transduction enhancers. After 24 h of inoculation with the lentiviral suspension, cells were harvested, washed, and re-seeded in 96-wells with a cell density of ~ 15,000–20,000 cells/well. Alternatively, transduced cells were collected for DNA isolation or cryopreserved for xenotransplantation.

Vector copy number (VCN) determination by qPCR or droplet digital PCR (ddPCR)

VCN per diploid cell was determined by detecting the amount of vector-derived woodchuck hepatitis virus post-transcriptional regulatory element (WPRE) compared to the genomic reference gene polypyrimidine tract binding protein 2 (PTBP2) by Taqman qPCR on an Applied Biosystems StepOnePlus System (Foster City, CA, USA). 100 ng of genomic DNA from transduced samples were applied per PCR reaction in technical triplicate. The ABI Taqman Fast Advanced Master Mix (Thermo Fisher Scientific) was set with 660 nM WPRE and PTBP2 primer and 150 nM of the respective probes. The PCR was run in 40 cycles. Alternatively, VCN was determined with the QX200 ddPCR system (Bio-Rad, Feldkirchen, Germany), with primer pairs and probes for the WPRE element and for the PTBP2 gene in combination with the ddPCR supermix for probes—no dUTP (catalog #1863023, Bio-Rad). The number of viral sequences was normalized to the genomic reference sequence of PTBP2 to calculate the VCN.

Insertion site analysis using INSPIIRED

INSPIIRED (Integration Site Pipeline for paired-end reads) was developed to determine vector integration sites [35, 36]. Thus, genomic DNA was isolated from HSPCs using the DNeasy Blood & Tissue Kit (Qiagen). DNA was sheared by sonication using the S220 Focused-ultrasonicator (Covaris, Woburn, MA, USA). Purification was performed with AMPure Beads (Beckman and Coulter) in a 0.7-fold ratio. Fragmented DNA was end-repaired with the NEBNext Ultra II End Repair/dA-Tailing module (New England Biolabs (NEB), Frankfurt, Germany) and ligated with specific linkers using the NEBNext Ultra II Ligation Module (NEB). After repeating the purification of the DNA with the AMPure beads, nested PCRs were run to amplify the vector-genome junctions by adding Illumina adapter sequences, using specific index primers, and sample-specific linker primers. To assess the success of the PCR, the PCR products were loaded on 2% agarose gels for visualization and measured with the 1x dsDNA high-sensitivity (HS) assay by Qubit™ (Thermo Fisher Scientific). Lastly, PCR products were pooled into DNA libraries and transferred to the Research Core Unit Genomics (RCUG) of MHH for quality control with a Bioanalyzer device and analysis by Illumina sequencing on flow cells with 1 million clusters. Linker and primer sequences used for PCR reactions are available upon request. Downstream bioinformatic processing was performed according to the protocol by Berry and colleagues [35]. The analysis files necessary to run the INSPIIRED pipeline were downloaded from GitHub (https://github.com/BushmanLab/INSPIIRED). De-multiplexed sequences in FASTQ format were generated according to the individual index primers used, quality checked, aligned, and annotated to the human genome (hg38). The plasmid vector sequences served as a reference for the long terminal repeats (LTR) regions and vector trimming. The processing and alignment statistics were exported before uploading the results to a local database. After creating a sample management database, reports with all integration site data were generated and used for customized post-processing in Excel and GraphPad Prism (GraphPad Prism Inc., San Diego, CA, USA).

Statistical analysis

Statistical analyses were performed with Prism 10 or RStudio. Data are presented as mean ± SD in bar plots or as median with the upper and lower quartiles in violin plots. Statistical significance between the two groups was determined by a paired t-test. More than two groups were analyzed by One-way ANOVA if only one variable is considered (e.g., expansion factor) and by Two-way ANOVA if two categorical factors were included (e.g., expansion factor and starting population). P values < 0.05 were considered significant and indicated by an asterisk (p < 0.05 = *, p < 0.01 = **, p < 0.001 = ***, p < 0.001 = ****) or stated above the brackets.

Results

The small molecules A83-01, pomalidomide, and UM171 expand immune-phenotypic CD34+CD38-CD45RA-CD90+EPCR+ HSCs

To establish a biosafety assay with human HSPCs for modeling insertional mutagenesis, we need to define culture conditions to expand HSPCs while preventing vast differentiation. In many GT protocols, HSCs are cultured and transduced in “SFT” media [37]. However, the HSPC proliferation rate in SFT is low, which can have a negative impact on transduction efficiency. Hence, IL-3 is often added to SFT (SFT3) to promote the cycling of HSPCs [38, 39]. IL-3 can also promote the differentiation of HSPCs. To maintain the stem cell phenotype while supporting proliferation, we added the small molecules A83-01, pomalidomide (Poma), and UM171 (APU) to the basal SFT3 medium. As previous studies reported a direct correlation between stem cell purity and the expansion potential of murine HSCs, we tested this hypothesis for human cells [40, 41]. Therefore, we seeded equal numbers of CD34^+^ and CD34^+^CD38^−^CD45RA^−^ cells in SFT3, SFT3 + UM171, or SFT3 + APU conditions to assess the immune phenotype, expansion rate and viability after 7 days in culture (Fig. 1A). Specifically, we aimed to determine the number of stem and progenitor cells remaining in culture after prolonged expansion. We and others defined immune-phenotypical long-term (LT)-HSCs by the marker combination CD34^+^CD38^−^CD45RA^−^CD90^+^EPCR^+^ to distinguish them from CD34^+^CD38^−^CD45RA^−^CD90^+^ short-term (ST)-HSCs, or more differentiated CD34^+^CD38^+^CD45RA^+^CD90^−^ progenitors (Gating strategy in Fig. 1B) [42, 43]. In SFT3, the expanded CD34^+^CD38^−^CD45RA^−^ cells remained ~ 70% CD34^+^, predominantly expressed CD45RA^+^CD38^+^ but lacked EPCR and CD90 expression (Fig. 1B-C). The addition of APU or UM171 to SFT3 preserved CD34 expression at levels > 94%. CD90 and EPCR expression were significantly higher in APU conditions (p < 0.0001, Supplementary Table 1). APU-expanded (-exp) HSPCs harbored more than ~ 60% CD34^+^CD38^−^CD45RA^−^CD90^+^ immune-phenotypic ST-HSCs and ~ 30% CD34^+^CD38^−^CD45RA^−^CD90^+^EPCR^+^ immune-phenotypic LT-HSCs. Starting populations of purified CD34^+^CD38^−^CD45RA^−^ HSPCs demonstrated a clearer stem cell immune phenotype after expansion (Fig. 1B-C) compared to CD34^+^ starting populations (Supplementary Fig. 1A). Overall proliferation levels after 7 days were similar among all tested media (Fig. 1D). Also, the purity of the starting population had minor effects on total proliferation (Supplementary Fig. 1B). However, the number of ST-HSCs after expansion differed significantly between the groups. ST-HSC levels were 12.5-fold higher in APU medium compared to the SFT3-condition, and 4.4-fold higher compared to the sole addition of UM171 (Fig. 1E).

Fig. 1. Ex vivo expansion of HSPCs in A83-01, pomalidomide, and UM171 (APU). (A) Experimental scheme of the 7-day ex vivo expansion protocol. CD34^+^CD38^−^CD45RA^−^ cells from umbilical cord blood (UCB) were seeded in low cell density for HSPC expansion and analyzed via flow cytometry for their HSC immune phenotype. (B) Representative plots for the gating strategy of CD34^+^CD38^−^CD45RA^−^CD90^+^EPCR^+^ cells in SFT3 (basal medium), UM171, and APU. (C) Surface marker expression of CD34, CD38, CD45RA, CD90, and EPCR in N = 3 UCB donors after 7-day expansion of CD34^+^CD38^−^CD45RA^−^ cells in SFT3, UM171, and APU with 2 µM or 0.2 µM pomalidomide (Poma). (D) Expansion of the total cell number in one 96-well counted with Casy cell counter. (E) Fold change of expanded ST-HSCs after 7-days. The number of expanded CD34^+^CD38^−^CD45RA^−^CD90^+^ cells was divided by 1,000 (seeded cell number) and then normalized to SFT3. (F) Percentage of CD34^+^CXCR4^+^ HSPCs after expansion. The mean is displayed with the SD by the error bars. Statistical analysis in D-F was performed with one-way ANOVA and Tukey’s test for multiple comparisons in N = 4 UCB donors (* = p < 0.05, ** = p < 0.01)

In contrast to the positive effects of Poma in preserving HSCs, it also decreased cell viability. This effect was even stronger when cells were cultivated long-term. After 14-day cultivation, 2 µM Poma alone and all combinations containing Poma (APU, PU, AP) had a lower viability compared to the conditions without Poma (Supplementary Fig. 1C). Cryopreservation of expanded cells accelerated the decrease in viability after thawing. However, the expansion of CD34^+^CD38^−^CD45RA^−^CD90^+^ cells was still superior in conditions with Poma (Supplementary Fig. 1D). Hence, we titrated Poma and analyzed the level of cells in early and late apoptosis after 14 days (Annexin V^+^ DAPI^+^ staining - Supplementary Fig. 1E). The frequency of Annexin V^+^ cells correlated with the concentration of Poma. The 10-fold lower concentration (0.2 µM) of Poma resulted in good viability while still maintaining the HSC immune phenotype in APU conditions. This optimized medium served as the basis for long-term culture assays to assess clonal kinetics after transduction.

As APU-exp cells might also be used in xenotransplantation experiments, we analyzed the effects of APU on CXCR4 expression. This chemokine receptor is required for HSC homing and migration to the BM niche [44, 45]. Interestingly, the percentage of CD34^+^CXCR4^+^ cells was decreased by Poma and could be partially rescued by UM171 (Fig. 1F). Again, reducing Poma concentration to 0.2 µM increased the frequency of CD34^+^CXCR4^+^ cells to levels observed for UM171-exp HSPCs, while preserving the HSC immune phenotype (Fig. 1C).

To determine which compounds contributed to ST- and LT-HSC expansion, we performed single and minus-one compound conditions in the same experimental setting as in Fig. 1A. Surprisingly, PU supported the EPCR^+^ LT-HSC phenotype even without A83-01 (Fig. 2A). The TGFβ inhibitor A83-01 only showed a noticeable effect in combination with Poma. Although Poma alone resulted in a 3.5-fold higher percentage of CD90^+^EPCR^−^ ST-HSCs, the percentage of CD34^+^ cells was equal to SFT3.

Fig. 2. Single-cell RNA sequencing (scRNA seq) analysis reveals expression of HSC signature genes in APU-exp HSPCs. (A) Surface marker expression of CD34, CD38, CD45RA, CD90, and EPCR after 7-day expansion in SFT3 or SFT3 supplemented with single (A, P, U), minus-one (AP, AU, PU), and complete APU (APU with 2 µM or 0.2 µM Poma) compound conditions. (B) Experimental scheme for the scRNA seq experiment using the 10X Genomics Chromium technology and hashtag oligo (HTO)-labeled antibodies for sample multiplexing. UCB-derived CD34^+^CD38^−^CD45RA^−^ HSPCs were analyzed uncultivated, 24 h cultivated in SFT, SFT + APU, or SFT + APU_0.2, and 7-day expanded in the conditions described in (A). (C) Uniform Manifold Approximation and Projection for Dimension Reduction (UMAP) displaying the integrated data of all samples with cluster annotation: Hematopoietic stem cells/multipotent progenitors (HSCs/MPPs), early and late cycling megakaryocytic-erythroid-mast cell progenitors (cMEMPs), granulocytes (neutrophils, eosinophils, basophils/mast cells), mast cells, monocytes and macrophages (Mono’s/Neutro’s), cycling MPPs (cMPPs), and cycling dendritic cells (cDCs). (D) Dot plot showing the average expression of the respective marker genes used in cluster annotation

The effects of APU compounds on the single-cell level

To apply the APU-exp HSPCs to genotoxicity assays, we wanted to verify their promising HSC immune phenotype on the transcriptional level. The transformation of early progenitors by a mutagenic vector integration relies on the presence of stem and progenitor cells at the time of transduction. To assess the frequency of HSC in our expanded culture at the single-cell level, we performed scRNA seq with the 10x Genomics Chromium platform. The setup comprised HSPCs expanded for 7 days in single (A_7d, P_7d, U_7d), minus-one (AP_7d, AU_7d, PU_7d), and complete APU medium (APU_7d) in comparison to the basal medium (SFT3_7d) and uncultivated CD34^+^CD38^−^CD45RA^−^ cells (uncultivated) (Fig. 2B). Additionally, we aimed to verify the HSC expansion capability of 0.2 µM Poma in the APU combination (APU_0.2_7d) compared to the standard concentration of 2 µM Poma described by Ebina. We also compared the short-term APU cultivation without IL-3 to detect initial intracellular signaling cascades after 24 h (APU_24h, SFT_24h). Later, we included HSPCs expanded for 14 days to see long-term cultivation effects.

TotalSeq™ Hashtag antibodies were used to label cells of each condition. After sample demultiplexing and their import into a Seurat-based workflow for UMAP visualization of all integrated samples, we performed clustering and cell type annotation (Fig. 2C, detailed description of data processing in Supplementary Material and Methods). Marker genes were enriched in the specific cell types (Fig. 2D). All marker genes with a log2 fold change > 3 and expressed in > 20% of the cluster are shown in Supplementary Table 2. We also submitted the marker genes to OpenAI’s ChatGPT (GPT‑4, last accessed 16 September 2025; Supplementary Table 3), further supporting the choice of the selected cell types.

Early effects of APU on the gene expression of HSPCs

In many gene therapy protocols, transduction is performed as early as possible to avoid loss of stem cells by differentiation. Hence, we investigated whether APU could further reduce initial transcriptional changes, leading to loss of stemness. In the early conditions (uncultivated and 24 h cultivated, Fig. 3A), most of the cells were hematopoietic stem cells/multipotent progenitors (HSCs/MPPs) or cycling megakaryocyte erythrocyte progenitors, and mast cell progenitors (cMEMPs) (Fig. 3A). MPPs began to cycle (cMPPs) after 24 h, and their abundance increased over time (Supplementary Fig. 2A-B). The HSC_MPP_1 cluster harbored the highest expression of HSC signature genes, such as HLF, MECOM, NKAIN2, PROM1,* AVP*, and CD34 (Figs. 2D and 3B), was mostly in G1 phase (Supplementary Fig. 2A-B), and had the earliest rank in pseudo time analysis (Supplementary Fig. 2C-D). In the HSC_MPP_2 and _3 clusters, more cells switched to the S phase (Supplementary Fig. 2A-B). Initially, the general cluster distribution of 24 h-cultivated HSPCs looked identical among all media conditions (SFT, SFT + APU, and SFT + APU_0.2) (Fig. 3A, bar graph). However, comparing the average expression of 13 HSC signature genes [46–50] within the HSC/MPP clusters revealed that the APU compounds increased RUNX1, MLLT3, CRHBP, SPINK2, ANGPT1, NKAIN2, MECOM, and HLF expression (Fig. 3C).

Fig. 3. Effect of APU on HSPCs after 24 h cultivation on the single-cell level. (A) UMAP of uncultivated and 24 h-cultivated HSPCs and their individual cluster abundances in the bar plot below. (B) Feature plots of the HSC signature genes HLF, MECOM, NKAIN1, PROM1, AVP, and CD34. (C) Dotplot comparing the average expression of HSC signature genes among the 24 h-cultivated HSPCs. (D) Volcano plot displaying the log2 fold changes (FC) and the -log10 p values of DEGs between HSPCs cultivated for 24 h in SFT + APU (APU_24h) compared to SFT (SFT_24h). Significant (p > 0.05, Wilcoxon Rank Sum test) genes with an FC > 1 are colored in green (up in APU_24h) or yellow (up in SFT_24h). (E) Gene set enrichment analysis (GSEA) on the DEGs of APU_24h vs. SFT_24h using HSC-related gene sets from the Molecular Signatures Database (MSigDB), showing the normalized enrichment score (NES). (F) GSEA on the DEGs of APU_24h vs. SFT_24h and of SFT3 + APU_7d vs. SFT3_7d (HSC/MPP cluster) using TGFβ-related gene sets from the MSigDB

Next, we analyzed the effects of APU addition within the first 24 h without IL-3 by determining the differentially expressed genes (DEGs; Fig. 3D, Supplementary Table 4). We used the DEGs to assess the enrichment for HSC gene sets from the Molecular Signatures Database (MSigDB; Fig. 3E). APU_24h scored with a positive normalized enrichment score (NES) for all gene sets linked to HSC homeostasis, proliferation, migration, and HSC gene upregulation. In contrast, gene sets describing HSPC differentiation and TGFβ signaling were downregulated in APU_24h (Fig. 3E-F). Especially SKIL and SMAD7, two crucial regulators of TGFβ signaling, were strongly downregulated in APU_24h *(*average log2 fold change ~ -3 compared to SFT_24h, Supplementary Table 4). We hypothesize that the ALK 4/5/7 inhibitor A83-01 in the APU medium inhibited the TGFβ signaling cascade within 24 h. Since low levels of TGFβ signaling maintain HSCs in healthy hematopoiesis, we concluded that early TGFβ inhibition benefits the preservation of stemness already within 24 h of ex vivo expansion, whereas in SFT-only medium, higher TGFβ signaling most likely stimulates differentiation.

The long-term culture effects of APU

The use of 7-day expanded HSPCs could increase the starting material to test multiple vectors and replicates in the assessment of the preclinical biosafety of RVs. However, we had to ensure that the expansion maintained HSCs as the target cell type. The cell type annotation revealed differentiation processes after 7 days of expansion in all conditions, most likely driven by IL-3. Terminally differentiated cells, such as granulocytes (neutrophils, eosinophils, basophils), mast cells, and cycling dendritic cells (cDCs) were present (Fig. 4A). In line with their cell fate, the differentiated cells were detected later in the pseudo time analysis (Supplementary Fig. 2C-D). Over time, a subset of the cMEMP cluster highly expressed GP1BB, CAVIN2, HBD, and PPBP, linked to platelet and erythrocyte differentiation (Supplementary Fig. 2E). We could confirm the myeloid differentiation potential of APU-exp HSPCs and their mildly reduced erythroid differentiation bias via colony forming assay (Supplementary Fig. 2F-H). In contrast, Poma promoted the differentiation of cDCs (Fig. 4A). After 14 days, Poma further promoted the differentiation of monocytes to an inflammatory or activated state (Supplementary Fig. 3A-C). Generally, the presence of some cell types varied among the conditions in the scRNA seq dataset. Mast cells were more abundant in 7-day expanded cultures without Poma, and the highest number was observed with UM171 (Fig. 4A). But did the addition of APU preserve stem cells after 7 days? In flow cytometry, A83-01 alone did not promote a clear HSC immune phenotype (Fig. 2A). However, in scRNA seq, the combination of A and AU harbored the highest percentage of HSCs/MPPs after 7 (Fig. 4A) and even 14 days of culture (Supplementary Fig. 4A-C). Hence, we analyzed the HSC gene expression in the HSC/MPP clusters of all single, minus-one, and complete APU conditions of the 7-day expanded HSPCs (Fig. 4B). APU_7d and APU_0.2_7d showed the highest average expression for most HSC genes (HOXA9, PRDM16, CD34, AVP, ANGPT1, NKAIN2, MECOM, HLF). Although APU_7d/APU_0.2_7d outperformed AU_7d in 9 of 13 HSC genes, AU_7d had a higher average expression of MLLT3, CRHBP, and PROM1 and had the highest percentage of the HSC_MPP_1 cluster, which could be linked to a downregulation of HSC genes by Poma. Some single compounds upregulated the HSC gene expression individually. For example, ANGPT1 and MECOM were upregulated by Poma or MLLT3 by A83-01 alone. Nevertheless, the expression of the HSC signature genes in uncultivated HSPCs differed from all expanded cultures (Supplementary Fig. 3D). Compared to 7-day expanded or 24 h cultivated HSPCs, the uncultivated HSPCs harbored a higher expression of NKAIN2, PROM1, SPINK2, CRHBP, AVP, and MLLT3. Interestingly, the genes RUNX1 and SPINK2 were expressed in uncultivated HSPCs, then downregulated after 24 h, but again upregulated after 7 days, especially under the conditions harboring AP.

Fig. 4. Effect of APU on HSPCs after 7-day culture on the single-cell level. (A) UMAP of 7-day expanded HSPCs and their individual cluster abundances in the bar plot below. (B) Dotplot comparing the average expression of HSC signature genes among the 7-day expanded HSPCs only within the progenitor clusters (HSC_MPP_1–3, cMEMP_1–2, cMPP). (C) Gene set enrichment analysis (GSEA) of the marker genes within each 7-day expanded HSPCs condition within the HSC/MPP clusters using the JAANTINEN set. Normalized enrichment score (NES) indicates enrichment in upregulated HSC genes

Additionally, we determined DEGs of each 7-day expanded compound condition within the HSC/MPP cluster using the Seurat function “FindMarkers” to perform gene set enrichment (GSEA). AU_7d HSPCs scored the highest NES for the “Jaatinen_Hematopoietic_Stem_Cell_Up” gene set (NES_AU_ = 1.97), followed by APU_0.2 (NES_APU_0.2_ = 1.17) (Fig. 4C). All APU complete and minus-one conditions scored a positive NES, while the single compounds and SFT3 were negatively enriched for the HSC genes. We concluded that the APU expansion generated immune phenotypical HSCs with a transcriptional signature related to stemness, but myeloid lineage commitment was inevitable. Thus, we hypothesized that the 7-day APU-exp HSPCs could be used for RV transduction and potentially also for long-term cultivation afterward.

7-day expansion in APU results in increased hematopoietic repopulation in vivo

To test whether APU-exp HSPCs could also be used for in vivo studies, the expanded HSPCs were xenotransplanted into the immunocompromised NBSGW mice. Evaluating the engraftment potential would be the ultimate measure of stemness and self-renewal capacity. The mouse strain does not require irradiation to reach a similar human chimerism seen in irradiated NSG (NOD.Cg-Prkdc^scid^Il2rg^tm1Wjl^/SzJ) mice due to their additional Kit^W−41 J^ mutation [51]. We found that the number of transplanted cells, the duration of expansion, the donor, and the treatment of the HSPCs all had massive impacts on the human chimerism in the NBSGW model.

Transplantation cohort 1 (Tx1) received 7-day bulk culture expanded HSPCs (Supplementary Fig. 4A-E). We transplanted 1 × 10^6^ expanded HSPCs per mouse (SFT3-exp = 48,13% and APU-exp = 92,36% CD34^+^, Supplementary Fig. 4E). Compared to the basal SFT3-condition, APU increased the human chimerism in peripheral blood (PB), BM, and spleen (Supplementary Fig. 5A-F). Only APU-exp and uncultivated HSPCs developed CD3^+^ cells in the PB of the mice. However, in most cohorts, the human chimerism of APU-exp HSPCs was still lower compared to uncultivated cells. In cohort 2 (Tx2), we transplanted 14-day expanded HSPCs from the same experiment and donor set as the 7-day experiment. The number of ST-HSC increased by a factor of ten during the additional cultivation week in APU medium (Supplementary Fig. 4B + F). However, we assumed that prolonged cultivation would decrease the number of engraftment-capable cells and transplanted ~ 1 × 10^7^ cells per mouse. Unexpectedly, the 14-day expanded HSPCs saturated the BM niche across all groups (Supplementary Fig. 6A-C), although SFT3-exp HSPCs harbored only 15% CD34^+^ (APU-exp = 54% CD34^+^, Supplementary Fig. 4F). Despite the presence of fewer ST-HSCs, and significantly lower EPCR expression in SFT3-exp HSPCs than in APU-exp HSPCs, all six recipient mice reached a human chimerism of over > 93% (hCD45^+^) in the BM and a hCD34^+^ frequency like the uncultivated HSPCs. The strongest differences among the groups were detected in the PB and in the spleen (Supplementary Fig. 6D-H). We hypothesized that the high cell number of expanded cells without HSC phenotype but myeloid progenitors within the grafts (~ 1 × 10^7^ cells/mouse transplant) most likely contributed to the occupation of the BM niche. The high human chimerism in all groups complicated the assessment of differences in the expansion media properties, although we appreciated the repopulation of the BM despite the 2-week expansion.

To prevent an oversaturated BM niche for a reliable readout, we seeded cell numbers ranging from 2,500 to 10,000 CD34^+^CD38^-^CD45RA^-^ cells after APU expansion (Supplementary Fig. 7A-B). The cells were expanded in low-cell-density, which resulted in the expected HSC immune phenotype as observed in previous experiments (Supplementary Fig. 7C). Then, the expanded cells were cryopreserved and thawed 16–18 h before transplantation to have a similar setup to gene therapeutic protocols. After 18 weeks post-Tx, we observed the clearest difference in the BM at a level of 5,000 transplanted HSPCs (Supplementary Fig. 7D). Generally, APU showed a superior repopulation of the niche for each tested cell number. Hence, we adjusted the cell number to 5,000, included the lower and less toxic Poma concentration (APU_0.2), and UM171-exp HSPCs as a comparison in the xenotransplantation (7E). However, non-expanded, APU_0.2-expanded, and UM171-expanded again equally repopulated the BM, spleen, and PB, although APU and APU_0.2 elicited a superior HSC immune phenotype (Supplementary Fig. 7F).

In conclusion, expanded and non-expanded HSPCs show substantial engraftment but without significant differences between media or expansion in the given experimental setup. Since the cells did not lose their repopulation potential, we aimed to explore the benefits of APU addition in LV transduction protocols.

APU expansion medium supports lentiviral transduction of CD34+ cells

Most GT protocols describe that patient-derived CD34^+^ cells are transduced around 24 h after thawing. Hence, we initially explored whether the APU medium would support lentiviral transduction itself and whether those transduced HSPCs could be expanded for transplantation.

CD34^+^ cells were thawed on day − 1, transduced on day 0, and cultivated in SFT3 or APU medium until day 6 (Fig. 5A). We thawed cryopreserved UCB units of three donors and counted around 1 × 10^6^ CD34^+^ cells per unit, but observed a 10-fold loss of cells 24 h later (Fig. 5B). Lentiviral transduction was performed with a self-inactivating lentiviral vector encoding for mCherry (SIN-LV.CBX3.EFS.mCherry) to verify transgene expression by flow cytometry. On day 6 post-transduction (post-td), the cells expanded equally in SFT3 and APU medium. Still, only APU could maintain the expression of stem cell markers (Fig. 5C). The mean percentage of CD34^+^ cells was around 9% higher in APU-exp cells compared to SFT3-exp (SFT3-exp = 67.1%, APU-exp = 76.4% CD34^+^). Cells expressing ST-HSC markers CD34^+^CD38^-^CD45^-^CD90^+^ and the LT-HSC marker EPCR^+^ were significantly higher if HSPCs were expanded in APU medium (p < 0.001 for CD90^+^, p < 0.0001 for EPCR^+^).

Fig. 5. Increased lentiviral transduction efficiency in APU-exp cells. (A) Graphical scheme of the combined transduction (td) and expansion protocol for UCB-derived CD34^+^ cells. Td was performed with 1 × 10^5^ CD34^+^ 24 h post-thawing with a self-inactivating lentiviral vector encoding for mCherry driven by the CBX3.EFS promoter (SIN-LV.CBX3.EFS.mCherry) in APU/SFT3 medium supplemented with protamine sulfate and Synperonic^®^ F-108 in a U-bottom 96-well plate. On day 1 post-td, HSPCs were seeded for expansion. Td efficiency and stem cell phenotype were determined on day 6 post-td, and vector copy number (VCN) was determined on day 8 post-td. Each donor replicate of expanded cells was split for transplantation into three mice. (B) Number of cells after thawing CD34^+^ cells from N = 3 UCB donors (day − 1), the next day (day 0), and after expansion (day 6 post-td) counted with Casy. (C) HSPC immune phenotype of expanded cells 6 days post-td. Expression of CD34^+^, CD34^+^CD38^−^CD45RA^−^CD90^+^ (labeled CD90^+^), and CD34^+^EPCR^+^ in expanded HSPCs. (D) Td efficiency in the bulk cell population, in CD34^+^ HSPCs, and CD34^+^CD38^−^CD45RA^−^CD90^+^EPCR^+^ (labeled as LT-HSC). (E) VCN per diploid cell determined by qPCR. Bars indicate the mean of n = 3 UCB donors with the SD displayed by the error bars; Statistical significances were determined by paired t-test; p < 0.05= *, p < 0.0001= ****

The transduction efficiency was higher with APU medium compared to SFT3 (Fig. 5D). This was true for the bulk culture and HSPC subpopulations (Fig. 5D). In line with the higher transduction efficiencies, APU-exp cells harbored a significantly higher mean vector integration load than SFT3-exp cells (APU_VCN_ = 1.46, SFT3_VCN_ = 1.03, p < 0.05, Fig. 5E).

For a pilot study of transplanting the transduced CD34^+^, we decided to split the expanded HSPCs due to the experience of oversaturation of the niche (Supplementary Fig. 5–7). The expanded sample was divided into three aliquots, each of which was transplanted into a separate mouse. However, although being in a range of cells similar to the previously described cohorts (~ 5.3 × 10^5^-1.4 × 10^6^ per aliquot, Supplementary Fig. 4A-B, Supplementary Fig. 7B), the cells only caused a sufficient human chimerism (> 5%) in 4/18 mice: Only 1/9 mice transplanted with SFT3-exp cells and 3/9 mice with APU-exp cells showed substantial engraftment (Supplementary Fig. 8A). Since our experience in the previous cohorts was based on CD34^+^CD38^−^CD45RA^−^ cells as starting material (Supplementary Fig. 4–6), the frequency of engraftable HSC was underestimated here using CD34^+^ cells (Supplementary Fig. 8A). The mCherry expression within the hCD45^+^ cells in the BM was either abundant in virtually all cells (> 97% for mouse 63 and 69) or almost not detectable at all (< 0.4% for mouse 62 and 66) (Supplementary Fig. 8B). Furthermore, we determined the vector copy number (VCN) of hCD45-enriched BM of these mice. The VCN of BM samples showing high mCherry^+^ expression had VCNs of 3.75 (mouse 63) and 3.42 (mouse 69), compared to BM samples with a low VCN (62_VCN_ = 0.103, 66_VCN_ = 0.307), which correlates with a low mCherry^+^ expression (Supplementary Fig. 8C). Although APU enhanced the transduction efficiency, the low cell number 24 h after thawing and the few engrafted samples demanded an improved protocol.

APU_0.2 supports the long-term culture of CD34+CD38-CD45RA-CD90+ HSPCs after lentiviral transduction by inhibiting monocytic differentiation

Expanding HSPCs before transduction could boost their viability during the prestimulation phase and increase the number of cells to conduct preclinical safety studies. In the murine genotoxicity assay “IVIM”, at least nine independent transductions are needed for the test vector, and three replicates for the positive and negative controls to elucidate statistical differences [10]. Without expansion, several UCB units would be required to compensate for the loss of cells after 24 h. Therefore, we tested the use of 7-day APU_0.2-exp HSPCs compared to SFT3-exp (Fig. 6A). We adjusted the protocol for the following experiments to the less cytotoxic, 10-fold lower Poma concentration in the APU medium (APU_0.2). The HSPCs expanded by 40 to 100-fold after 7 days, with no significant difference in the choice of medium, although the expansion factor tended to be higher in SFT3 (Fig. 6B). However, the HSC immune phenotype (CD34^+^CD38^-^CD45RA^-^CD90^+^) was again significantly more preserved in APU_0.2 compared to SFT3 (p < 0.05, Supplementary Fig. 9A).

Fig. 6APU_0.2 allows long-term cultivation of CD34^+^ cells up to 5 weeks after lentiviral transduction. (A) Experimental setup of transducing expanded HSPCs. Purified CD34^+^CD38^−^CD45RA^−^ cells were expanded for 7 days either in SFT3 or APU_0.2 medium before transduction (td) with the SIN-LV.SF.eGFP (SF) or the SIN-LV.EFS.eGFP (EFS) vector. 24 h post-td, HSPCs were either transplanted into immunocompromised mice or further cultivated. Long-term culture allowed downstream analyses. (B) Expansion factor after 7-day culture in SFT3 or APU_0.2 medium determined by Casy. Paired t-test revealed non-significant (ns = p > 0.05) differences. (C) Flow cytometric analysis of eGFP expression on day 7 post-td to determine the td efficiency. (D) Vector copy number (VCN) per diploid genome was determined via ddPCR. (E) Expansion factor over 5 weeks post-td either in SFT3 or APU_0.2 medium. Cell numbers were determined by Casy and divided by the seeded cell number from the week before. Paired t-test over the whole period of time between APU_0.2 and SFT3 to assess the expected mean difference (diff.) between the groups, the 95% confidence interval (CI), and the p value. (F) Microscopic images 3 weeks post-td. Imaging was performed with CellCyteX using 10x magnification. Scale bar length indicates 100 μm. (G-H) Monocytic markers (CD11b, CD14, CD33) were determined via flow cytometry 3, 4, and 6 weeks post-td of HSPCs cultivated in SFT3 (G) or APU_0.2 (H). (I) Dim and bright CD11b expression in HSPCs cultivated for 3 weeks post-td. Exemplary gating strategy of the CD11b^+ bright^CD14^+^ population for SFT3 cultivated cells. (J) Percentage of CD34^+^ cells during the long-term culture in SFT3 or APU_0.2 up to 5 weeks post-td. Two-way ANOVA with Tukey’s multiple comparisons test. Individual values are biological replicates with N = 3 UCB donors. Error bars indicate the mean ± SD

We investigated expanded HSPCs, transduced with the mutagenic SIN-LV.SF.eGFP (SF) vector or the safer SIN-LV.EFS.eGFP (EFS) vector, as well as non-transduced controls (mock) [33, 52]. In a single transduction using the transduction enhancers Synperonic^®^ F-108 [53] and RetroNectin, we achieved transduction efficiencies ranging from ~ 38% to 86%, with no significant differences between vector type or medium (Fig. 6C). In line with the level of eGFP expression, the range for the VCN per diploid cell varied between ~ 0.7–5.7 (Fig. 6D).

To use the expanded cells in safety assays for insertional mutagenesis, cells need to proliferate to allow for the development of clonal dominance without differentiating too rapidly. Hence, we compared the cell numbers of the transduced and untransduced HSPCs for 5 weeks post-td. HSPCs cultivated in APU_0.2 medium had a significantly higher expansion capacity compared to SFT3 in the mock and SF-transduced conditions (p < 0.05, Fig. 6E). Especially at the last two time points, APU_0.2 cultured cells expanded twice as much as in SFT3 medium. In EFS-transduced cells, APU_0.2 also increased expansion, although without a statistically significant difference compared to SFT3 medium. Cells in SFT3 medium began to adhere and differentiate earlier, which might have caused the decline in the expansion (Fig. 6F).

The increased level of differentiation in SFT3 was also confirmed by the higher expression of monocytic markers (Fig. 6G-H). While APU_0.2 medium inhibited CD14 expression during the first week of culture, the SFT3 medium promoted CD14 expression, visible at 2 weeks post-td. Both media promoted CD11b^+^ cells already visible 2 weeks post-td. However, the SFT3 medium promoted predominantly CD11b^+ bright^ cells (CD11b^+ bright^mean > 60%), of which > 95% were CD14^+^ (Fig. 6I, Supplementary Fig. 9B). In contrast, APU_0.2 contained mostly CD11b^+ dim^ cells, but also maintained significantly higher CD34^+^ expression over 5 weeks post-td (Fig. 6J). Moreover, 5 weeks post-td, significantly more CD34^+^CD38^-^CD45RA^-^CD90^+^ HSPCs were still present in APU_0.2 compared to SFT3 (Supplementary Fig. 9C, p < 0.0001). There were no significant differences in the surface marker expression among the types of vector transduction, irrespective of the medium. However, we observed a tendency for transduced HSPCs to differentiate more rapidly than mock HSPCs.

Long-term culture in APU_0.2 enables tracking of insertion sites and clonal development of transduced HSPCs

Since there was no obvious sign of transformation by the mutagenic vector in the flow cytometry analyses or by proliferation, we decided to investigate the integration sites. The appearance of clonal integrations could be a hint for transformation before immune phenotypical changes occur in the time-frame of a cell culture experiment. As described in the experimental scheme in Fig. 6A, we harvested cells weekly for DNA isolation to conduct integration site analysis (ISA). We observed a highly polyclonal integration pattern in the samples 24 h and 1–3 weeks post-td, measured by the Gini and Shannon diversity indices (Fig. 7A). However, the diversity decreased each week (Fig. 7A, Supplementary Fig. 10A-B). In line with the diversity indices, the number of UIS was the highest 7 days post-td and declined over time in culture (Fig. 7B-C). The mean of the UIS in APU_0.2 cultured HSPCs was higher than in SFT3-exp HSPCs at each time point for SF and from day 14 post-td for EFS. After fitting an exponential decay curve to the data points, the trend of APU_0.2 supporting a higher number of UIS over a longer time was confirmed. The VCN showed a similar development to the UIS but remained relatively stable after 7 days post-td (Supplementary Fig. 10C). Additionally, we determined the exact chromosomal position of the vector integration to assess whether insertional mutagenesis played a role during culture. ISA revealed a gradual increase in the contribution of specific integrations over time in all media and vectors (Fig. 7D-G, Supplementary Fig. 10D-E). We observed integrations within genes () and within 50 kb of cancer genes (~). In a representative sample of EFS-transduced HSPCs cultured in SFT3, the integration with the highest contribution was within LRRC19 (3.1%), which encodes for the leucine-rich repeat-containing 19 protein involved in NF-κB activation and induction of proinflammatory cytokines [54]. In contrast, the mutagenic SF vector integrated into MEIS1* (3.7%), a high-risk locus in GT due to its involvement in the proliferation and survival of leukemia cells [55, 56]. Moreover, SF integrated into the tumor suppressor gene BRCA2 [57]. Thus, our in vitro culture was able to model the emergence of integrations potentially involved in insertional mutagenesis. Since APU_0.2 media supported a prolonged culture of immune phenotypical HSPCs, we observed even higher contributions. APU_0.2-exp HSPCs transduced with SF harbored integrations with a contribution of almost 9% in SUSD6, which is abundantly expressed in AML and solid tumors [58]. Modeling the appearance of integrations nearby or within relevant proto-oncogenes involved in leukemia in vitro could be a powerful tool for the prediction of clonal development, which is observed in patients suffering from suspected unexpected serious adverse events (SUSARs) years after GT.

Fig. 7. Ex vivo long-term cultivation enables the modeling of viral integration sites in HSPCs. (A) Diversity and clonality of vector integration sites (with VCN > 0.02) in in vitro-cultured HSPCs (24 h, 1–6 weeks post-td) and in the BM of xenotransplanted mice, 16 weeks post-transplantation, measured by the Gini and Shannon indices. Dot size indicates the inverse of the UC50 value, which stands for the number of insertions present in 50% of the cells within a sample. (B-C) Unique integration sites (UIS) during long-term in vitro culture in SFT3 or APU_0.2 medium after transduction with LV-EFS.eGFP (EFS, B) or LV-SF.eGFP (SF, C) in N = 3 UCB donors. Error bars indicate the mean (filled points) ± SD of the donor replicates (transparent points). Fitted exponential decay curves. (D-G) Representative sandart plots displaying the abundance of the top 10 integration sites and the remaining UIS in grey up to 6 weeks (42 days) post-td: EFS-transduced in SFT3 (D), SF-transduced in SFT3 (E), EFS-transduced in APU_0.2 (F), SF-transduced in APU_0.2 (G)

ISA reveals lower clonality in the BM of mice transplanted with HSPCs expanded, transduced, and cultivated in APU_0.2 than in SFT3

In vivo studies are still part of preclinical safety and efficacy studies. Therefore, we evaluated whether the presented in vitro model displayed similar dynamics to a mouse model. We transplanted the same starting population of the in vitro culture into the NBSGW mice and determined the human chimerism (% hCD45^+^) in the BM, PB, and spleen 16 weeks post-Tx (Supplementary Fig. 10F). Unfortunately, we observed human chimerism above 10% in the BM in only 5 of 20 mice and no significant differences among the conditions (Fig. 8A). The frequency of eGFP^+^ cells fluctuated but was present in all hCD45^+^ cells within the BM and partially in PB and SP (Fig. 8A, Supplementary Fig. 10G). We determined the VCN of the samples (Fig. 8A), which positively correlated with the eGFP expression within the hCD45^+^ cells (VCN vs. hCD45^+^/eGFP^+^: Spearman r = 0.6782). Similarly, the eGFP expression and the VCN had correlated with the overall engraftment but with a lower likelihood (hCD45^+^ vs. hCD45^+^/eGFP^+^: Spearman r = 0.5530, hCD45^+^ vs. VCN: Spearman r = 0.6584). Thus, we hypothesize that the vector integration itself did not massively influence the repopulation of the BM. Nevertheless, the choice of medium determined the clonality at the endpoint of the in vivo experiments, as indicated by the Shannon index, which considers the decreasing number of UIS (Supplementary Fig. 11A). We also included the diversity indices of Pielou and Simpson because they have been described as more robust for comparing the evenness of samples, especially among patients and in clinical trials [59], although they do not account for the general decrease in richness within the samples. Gini, Pielou, and Simpson were similar between APU_0.2 and SFT3, as the overall reduction in integrations is normalized in these calculations (Supplementary Fig. 11B-D). The number of UIS was significantly higher in the APU_0.2 conditions for EFS compared to SFT3 (p = 0.0211, Fig. 8B). Also, the number of UIS for SF was lower in SFT3 than in APU_0.2 (p = 0.0696). Hence, although the human chimerism gave similar results between APU_0.2 and SFT3, the number of integrations in the BM of APU_0.2-transplanted mice was higher. This higher number of clones in the BM potentially indicates that APU_0.2-transduced HSCs repopulated the niche more effectively than SFT3-transduced HSPCs. Assuming that higher clonality indicates a higher number of engraftable HSC clones, we provided an additional readout for measuring the engraftment potential of expanded and transduced HSPCs using ISA.

Fig. 8. Integration site analysis (ISA) reveals higher clonality in APU_0.2-transduced HSPCs after hematopoietic reconstitution. (A) Analysis of bone marrow (BM) samples 16 weeks post-transplantation. VCN (x-axis) and eGFP expression within hCD45^+^ cells (y-axis and color-scaled) are plotted with dot sizes that correspond to the human chimerism (hCD45^+^). Correlation between factors was calculated by the Spearman r index. (B) Number of UIS in the BM of mice transplanted with transduced HSPCs (EFS or SF) either in SFT3 or APU_0.2 medium. P values determined by two-way ANOVA using Šídák’s multiple comparisons test. Error bars indicate the mean ± SD. (C-D) Top 10 integration sites within the BM of mice harboring EFS- or SF-transduced HSPCs (with VCN > 0.02) cultured in SFT3 (C) or APU_0.2 (D). Abundance and gene symbol of the UIS with the indication whether integration is within a transcription unit (*), within 50 kb of a cancer-related gene (~), or associated with lymphoma in humans (!) according to the INSPIIRED pipeline

Overall, the integration site pattern in the BM samples was less clonal than the latest time points of the in vitro culture (Fig. 7A, Supplementary 11E-F), which is reasonable since the endpoint in the mice was 16 weeks post-Tx compared to 6 weeks in vitro. Additionally, the vector integrations had higher contributions in vivo than the in vitro samples. For example, one of the lowest clonality was seen in mouse 116 (LV-EFS), which harbored an integration in COL4A4 with a contribution of > 98% (Fig. 8C). COL4A4 encodes for the collagen type IV alpha 4 chain, known to be present in the basement membrane, but has not been described to be involved in hematological malignancies. Still, we also observed integrations associated with lymphoma in humans (marked in the graph with “!”), such as mouse 108, which harbored integrations by LV-EFS into TCF3, which has been associated with B-cell acute lymphoblastic leukemia if the gene expression is perturbed (Fig. 8D) [60, 61]. The SF vector also showed integrations in genes linked to leukemia, such as ELAVL1 [62] or GRB10 [63, 64] (mouse 107). In mouse 102, we observed the highest VCN (VCN = 2.15) and eGFP expression (hCD45^+^/eGFP^+^ = 90.2%) combined with a moderate human chimerism (hCD45^+^ = 45.1%). One could hypothesize that the integration into the epigenetic regulator ZNF217 could have influenced the repopulation of the niche or led to a lower clonality due to its described involvement in leukemia, specifically B-cell acute lymphoblastic leukemia [65, 66]. However, we did not observe any symptoms of leukemia in the timeframe of the experiment.

Discussion

Modeling integration sites of retroviral vectors in vitro could be an additional tool to evaluate the safety profile of novel RVs. Our aim was to establish culture conditions enabling long-term expansion of human HSPCs to allow for the detection of clonal retroviral integration sites. Therefore, we leveraged the small molecules APU before, during and after RV transduction. APU decreased differentiation for > 5 weeks after efficient LV transduction and allowed for monitoring integration sites over time. We observed increasing abundances of certain integration sites after 5–6 weeks cultivation in vitro suggesting clonal development. Due to the absence of an immune phenotype or growth advantage after transduction with a mutagenic vector, we propose that ISA technology could be the starting point for generating human HSPC-based preclinical safety prediction tools.

To allow the use of human HSPCs in preclinical biosafety assays, we needed to reduce differentiation processes during ex vivo cultivation. Therefore, we used the small molecules APU, described by Ebina [21]. We could confirm the strong expansion of immune phenotypical LT-HSCs (CD34^+^CD38^−^CD45RA^−^CD90^+^EPCR^+^) in APU medium, but also determined adverse effects by pomalidomide. Fortunately, by decreasing the pomalidomide concentration, the viability and expression of CXCR4 could be increased without loss of the HSC surface markers. Nevertheless, the use of pomalidomide, especially for clinical purposes, must be assessed carefully. Even lower concentrations or only temporary supplementation might further reduce negative impacts of pomalidomide, despite its ability to enhance CD34^+^CD38^−^CD45RA^−^CD90^+^EPCR^+^ cell expansion.

On the transcriptome level, APU already maintained an HSC phenotype and low TGFβ signaling in the early phase of expansion. We hypothesize that the ALK4/5/7 inhibitor A83-01 could have played a major role in the first 24 h. Surprisingly, after 7 days of expansion, A83-01 alone did not influence HSC surface marker expression or gene expression. A83-01 only showed an effect in combination with UM171 and/or pomalidomide. In contrast, we previously demonstrated that treating murine HSPCs with A83-01 for 14 days strongly inhibited mast cell differentiation and maintained lineage-negative cells [40]. However, in the current study, the scRNA seq data indicated that pomalidomide, and not A83-01, decreased mast cell differentiation, while UM171 supported the presence of mast cells. In line with our data, an upregulation of mast cell-specific genes by UM171 has also been described in scRNA seq analyses [67]. Recently, Roy and colleagues mentioned the advantage of UM171-expanded CB units harboring mastocytes and DCs, which are described to reduce graft-versus-host disease and enhance anti-leukemia effects after HSCT [68–70]. Besides the presence of HSCs in APU-exp cultures, the differentiation to cDCs and mast cells by APU could have a beneficial effect during engraftment, making the APU-exp HSPCs also worthy for UCB transplantation settings. However, we want to emphasize that AU was ranked closely after APU regarding the expression of HSC genes and could already be highly applicable for expansion settings if pomalidomide is of concern.

To allow the use of APU-exp HSPCs also for in vivo assays, we assessed their engraftment potential in NBSGW mice. APU demonstrated clear benefits compared to the basal medium SFT3, but the expanded cells were still inferior to uncultivated HSPCs in xenotransplantation studies using 7-day expanded HSPCs. Nevertheless, we saw a CD3^+^ population after transplantation of APU-exp HSPCs, comparable to uncultivated cells, which was absent in the SFT3 condition. The scRNA seq data confirmed the strong difference between uncultivated and cultivated HSPCs, even for as short as 24 h. Johnson and colleagues already described that within 24 h ex vivo, HSCs experienced a severe loss of their function [71], which we could not rescue by APU. We and others hypothesize that IL-3 strongly promotes myeloid differentiation, further decelerating the engraftment success [20]. Controversially, a 14-day expansion in SFT3 medium led to the highest repopulation of the BM niche of all expanded conditions in our studies. Most likely, this was linked to the engraftment of short-term repopulating progenitors, indicated by the absence of CD3^+^ cells and the high number of myeloid-biased progenitor cells potentially caused by IL-3 signaling. Ultimately, the xenograft model cannot entirely reflect the engraftment potential in humans. In our study, UM171 performed similarly well to APU_0.2-exp and non-exp HSPCs in the hematopoietic reconstitution. However, the transplantation of UM171-exp UCB cells into patients has been proven to be superior to non-exp cells in clinical studies [70, 72–74]. Hence, we believe that the observed engraftment potential of the expanded HSPCs makes them suitable for the development of preclinical assays in gene therapy.

Our institute regularly conducts the preclinical in vitro assays IVIM and SAGA using murine bone marrow-derived HSPCs for RV transduction [11]. To humanize the assay, we initially decided to transduce UCB-derived CD34^+^ cells 24 h after thawing to have a similar protocol to that in gene therapy. APU increased the transduction efficiency and maintained the HSC immune phenotype 6 days post-transduction compared to the SFT3 basal medium. However, the number of cells for transduction after thawing was too low to apply this experimental setup for a genotoxicity assay, which requires multiple replicates for the test vector and negative and positive controls. During the short time of 24 h, APU could not counteract the loss after transduction.

Fortunately, the successful expansion of HSPCs in APU allowed us to massively increase the number of viable HSPCs for LV transduction. Moreover, APU maintained immune phenotypical CD34^+^CD90^+^ HSPCs for > 5 weeks compared to the basal medium. In the mouse-based IVIM assay, mutagenic vectors cause an outgrowth of transformed clones after 15 days of bulk culture, followed by another 14 days under limiting dilution conditions. We did not observe excessive proliferation or a differentiation block in the transduced human HSPCs, which are characteristics of IVIM clones. Generally, the longevity of the human lifespan compared to that of a mouse makes human HSCs less sensitive to transformation [75]. We would have expected excessive proliferation or expression of surface markers known to be expressed on leukemic cells, e.g., CD11b, CD14, CD33, CD34, CD123, and CD38. Since none of these markers were influenced by the transduction with mutagenic vectors, we hypothesized that either other markers needed to be identified or more mutagenic vectors, potentially even overexpressing oncogenes such as HOXB8, needed to be tested. To identify additional markers, similarly to SAGA [76], the transcriptome of human HSPCs after transduction with known mutagenic vectors could be analyzed. We aim to explore a gene signature linked to transformation. Ultimately, combining transcriptomic analyses, such as scRNA seq, with ISA could determine the direct influence of high-risk integrations on the transcriptome. Specifically, early signs of transformation triggered by insertional mutagenesis would be valuable for longitudinal monitoring of patient studies. Alternatively, we aim to sensitize HSPCs to LV integrations near hotspots by using epigenetic regulators, such as quisinostat and CPI230, which have already been shown to increase the RV transduction while preserving the stemness and engraftment potential of HSPCs during ex vivo culture [77].

The prolonged in vitro culture enabled the detection of clonal outgrowth of integration sites. For example, we observed a clear increase in the abundance of integrations near the high-risk loci MEIS1 or SUSD6. Thereby, we were able to model integration site kinetics with human HSPCs after a few weeks in vitro, likewise to the ISA observations in the murine IVIM assay, which led to transformation. Similar patterns are also seen in patients with severe adverse events years after gene therapy [78, 79]. However, the model requires further verification by conducting transductions also with clinical vectors before we can predict clonal dynamics with certainty. Would known mutagenic vectors from clinical trials give rise to the same integration site profiles? Thus, we envision that the in vitro integration sites shall be compared to databases of GT patients to assess whether we could develop models for the prediction of insertional mutagenesis. Thus far, the culture model clearly supports clonal outgrowth. But are these integrations becoming dominant because of a proliferation advantage, or are they the result of the progeny of the fittest HSC in the culture? For the in vivo ISA, we observed many highly clonal integrations without association with leukemogenesis, which makes the in vivo model less representative of supporting the outgrowth of insertional mutants. Nevertheless, the number of integration sites in the BM of mice receiving APU-exp transduced HSPCs was higher than SFT3-exp. We hypothesize that the diversity could be indicative of the number of engraftable HSCs and not related to insertional mutagenesis. According to Shannon, we observed differences between SFT3 and APU, which could be a more sensitive measure of the repopulation potential of the HSPCs than assessing the human chimerism (hCD45^+^), which statistically did not differ. However, Gini, Pielou, and Simpson did not reveal differences. Generally, the four indices assess different aspects of diversity. While Gini and Simpson might judge the dominance of clones the best, Shannon can define how complex the integrome is in terms of its richness and evenness [80]. Pielou focuses on evenness and allows comparisons among species or clinical studies, especially if DNA content and sequencing depth vary [59]. However, we routinely conduct ISA with the INSPIIRED pipeline and adhere to strict guidelines, as the analyses are part of patient monitoring. Together with the UC50, Shannon, and Gini are used to report clonality and diversity in clinical trials [81].

Since the in vivo model had no advantages over the in vitro culture model, we aim to further investigate the potential of simulating insertional mutagenesis by long-term cultivation. More repetitions with different vectors are needed to build a statistical model, which could assess the fitness of the clonal integrations and their risk of causing transformation. Recently, the “MELISSA” tool was developed to estimate the frequency of gene-specific integrations and how these integrations affect the clonal fitness [82]. We envision implementing this tool for our study to classify the mutagenicity of the vectors based on the integration site development over time. Especially, the in vitro long-term culture would allow us to track dynamic changes over time, while we were only able to measure the endpoint in the mice due to low human chimerism and cell number in the peripheral blood samples. Nevertheless, the in vitro culture thus far lacks the ability to measure functional clonal fitness. While we aim to build prediction tools independent from animal experiments, transplantation of human HSPCs, which harbor clonal high-risk integrations after long-term culture, could shed light on the oncogenic risk of the vector.

Besides the assessment of LV-mediated genotoxicity, the long-term culture can also be used to evaluate other factors involved during transduction, such as new medium components and transduction enhancers. Novel small molecules that enhance expansion or transduction efficiency can also alter the epigenetic landscape in HSCs, with yet uncertain effects on insertional mutagenesis. Moreover, the long-term culture can also be used in disease modeling. After introducing disease-causing mutations, the long-term culture could reveal whether gene therapy can rescue the phenotype in efficacy studies that extend beyond the usual time frame of 1–3 weeks. Our protocol would allow measuring the transgene expression, evaluating promoter strengths, and potential silencing effects of the vector. Moreover, the toxicity and efficacy of other non-viral delivery systems, like lipid nanoparticles, could be assessed.

Conclusions

Our study confirmed the benefits of the small molecules APU during the short-term (24 h-14 days) and long-term (> 5 weeks) ex vivo expansion of UCB-derived HSPCs based on their immune phenotype, engraftment potential, and transcriptome (scRNA seq). The extended cultivation enabled the reconstruction of a vector integration site spectrum in human HSPCs in vitro, similar to the existing murine HSPC model for immortalization (IVIM assay). Thereby, ISA potentially allowed sensing early signs of clonal dominance before transformation potentially could have occurred on the immune phenotypical level. Ultimately, APU also increased the clonality of APU-expanded and transduced HSPCs compared to SFT3 in the bone marrow of xenotransplanted mice.

Supplementary Information

Below is the link to the electronic supplementary material.

Supplementary Material 1

Supplementary Material 2

Supplementary Material 3

Supplementary Material 4

Supplementary Material 5

Supplementary Material 6