Optimizing Lentiviral Vector Transduction of Hematopoietic Stem Cells for Gene Therapy

Abstract

Autologous gene therapy using lentiviral vectors (LVs) holds promise for treating monogenetic blood diseases. However, clinical applications can be limited by suboptimal hematopoietic stem cell (HSC) transduction and insufficient quantities of available vector. We recently reported gene therapy for X-linked severe combined immunodeficiency using a protocol in which patient CD34⁺ cells were incubated with two successive transductions. Here we describe an improved protocol for LV delivery to CD34⁺ cells that simplifies product manipulation, reduces vector consumption and achieves greater vector copy number (VCN) of repopulating HSCs in mouse xenotransplantation assays. Notable findings include: 1) the VCN of CD34⁺ cells measured shortly after transduction did not always correlate with the VCN of repopulating HSCs after xenotransplantation; 2) single-step transduction at higher CD34⁺ cell concentrations (2-4x10⁶/ml) conserved LV without compromising HSC VCN; 3) poloxamer F108 (LentiBOOST) increased HSC VCN by 2- to 3-fold (average from 3 donors); 4) while LentiBOOST+prostaglandin E2 combination further increased VCN in vitro, the VCN observed in vivo were similar to LentiBOOST alone. 5) cyclosporine H increased the HSC VCN to similar or greater extent with LentiBOOST in vivo. Our findings delineate an improved protocol to increase the VCN of HSCs after CD34⁺ cell transduction with clinically relevant LVs.

Introduction

Progress in the development of lentiviral vectors (LVs) for transduction of hematopoietic stem cells (HSCs) has advanced therapies for monogenic blood diseases including primary immunodeficiencies and β-hemoglobinopathies [1, 2]. LVs can deliver complex tissue-specific expression cassettes to non-dividing cells such as HSCs and have not been associated with leukemogenesis, in contrast to γ-retroviral vectors used in earlier clinical studies [3]. The clinical utility of any LV gene therapy depends on efficient high-level transduction of patient HSCs capable of long-term (LT) hematopoietic repopulation [4].

A number of methods can be used to optimize LV transduction of CD34⁺ hematopoietic stem and progenitor cells (HSPCs) [5, 6]. Strategies include two successive rounds of CD34⁺ cell incubation with LV at the high multiplicities of infection (MOI, 100-150) [7-9], and use of transduction enhancers such as polybrene, RetroNectin, rapamycin, and prostaglandin E2 (PGE2) [10-13]. Despite these methods, achieving high HSC vector copy number (VCN) is limiting for most clinical gene therapy protocols, particularly if genetically corrected progeny have no selective growth advantage in vivo [4, 14]. Moreover, the same HSC transduction protocol can produce variable results between different donors [8, 15, 16]. Hence, there remains a need for improved HSC transduction protocols to increase the efficacy of gene therapies and to reduce the manufacturing costs of clinical products.

We recently reported the initial findings of a clinical trial on LV gene therapy for infants with X-linked severe combined immunodeficiency (X-SCID) [15]. In this study, autologous CD34⁺ cells are incubated at relatively low density (1x10⁶ cells/ml) with 2-2.7x10⁸ transducing units (TU)/ml LV in two successive rounds. Although 7 of 8 individuals achieved T cell and natural killer cell recovery by 3-4 months post-treatment, the pre-infusion VCN of cell products varied between different individuals, from 0.16 to 1.13, and the two-step transduction protocol consumed relatively large amounts of LV particles, a problem that is exacerbated further in older subjects [7].

Here we examine several strategies to improve our CD34⁺ cell transduction protocol. Specifically, we evaluated the effects of CD34⁺ cell concentration during transduction, LV dosage, and transduction enhancers (alone or combined) on the VCN of CD34⁺ HSPCs after in vitro culture, and in repopulating HSCs using two mouse xenotransplantation models. As observed in previous studies, the VCN of pre-infusion HPSCs was consistently higher than in repopulating HSCs, emphasizing that mouse xenotransplantation assays are important to assess HSC transduction. By manipulating the concentrations of input CD34⁺ cells and vector particles, we developed a simplified, single-step protocol that achieves greater VCN of CD34⁺ cells and utilizes less vector than our two-step protocol employed in an ongoing clinical trial. In addition, we confirm recent findings that LentiBOOST [17], a non-ionic amphiphilic poloxamer that increases fluidity and permeability of cell membranes [18], can enhance HSC transduction by approximately 2- to 3-fold.

Materials and Methods

Human CD34⁺ cells

Human granulocyte-colony stimulating factor (G-CSF)-mobilized peripheral blood leukopaks from healthy donors were purchased from either Key Biologics, LLC (Memphis, TN, USA) or StemExpress, LLC (Folsom, CA, USA). The CD34⁺ cells were isolated using a CliniMACS instrument in the Human Applications Laboratory at St Jude Children’s Research Hospital (St Jude; Memphis, TN, USA) [19]. Cells were cryopreserved in medium containing 5% dimethyl sulfoxide, 6% Pentastarch, Plasma-Lyte A and 4% human serum albumin, and stored in liquid nitrogen until needed.

Lentiviral vectors and CD34⁺ cells transduction

Lentiviral vectors used in this study are described in Table 1 and were produced at the St. Jude Vector Core as previously described [20]. Lentiviral TU were determined after transduction of the human osteosarcoma cell line HOS by droplet digital polymerase chain reaction (ddPCR) [21] and MOI was calculated as TU/CD34⁺ cell number.

Table 1.

Lentiviral vectors used in this study are referred to by the short name.

Lentiviral vector (short name)	Production	Promoter	Gene
MND-eGFP	Transient transfection	MND	eGFP
CL20i4-EF1a-hgcOPT [27] (X-SCID 1st gen)	Producer cell line	EF1a	hIL-2Rγc
CL20i4-EF1a-hgcOPT [28] (X-SCID 2nd gen)	Producer cell line	EF1a	hIL-2Rγc
CL45-650-MND-WASp [34] (WASp)	Transient transfection	MND	hWASp

Transduction protocols are described in Table 2. Cryopreserved CD34⁺ cells were thawed and prestimulated for 24 to 48 h in X-VIVO 10 (Lonza, Basel, Switzerland) supplemented with 100 ng/ml recombinant human Flt3-Ligand, 100 ng/ml stem cell factor, 100 ng/ml thrombopoietin (Pepro Tech, Rocky Hill, NJ, USA), 1x penicillin/streptomycin (Corning, Manassas, VA, USA), and 2 mM L-alanyl-L-glutamine (Glutagro™, Corning). RetroNectin (Takara, Shiga, Japan) coated plates were used as specified. After prestimulation, CD34⁺ cells at specified density (1, 2, or 4 x 10⁶ cells/ml) were transduced with LV at the indicated TU/ml or MOI. Transductions were performed in cytokines-supplemented X-VIVO 10 media as above with 8 μg/ml protamine sulfate (Fresenius Kabi, Lake Zurich, IL, USA). Transductions for xenotransplantation experiments were supplemented with 1% human albumin (HA, Grifols Biologicals Inc., Los Angeles, CA, USA) where indicated (Table 2). Transduction enhancers tested in this study were 10 μM PGE2 (Cayman chemical, Ann Arbor, MI, USA), 1mg/ml LentiBOOST (Sirion Biotech GmbH, Martinsried, Germany), and 8 μM Cyclosporine H (CsH, Sigma-Aldrich, St. Louis, MO, USA) either alone or some combination. Transduced CD34⁺ cells were washed, counted, and either cultured in vitro for VCN, or transplanted into immunodeficient mice.

Table 2. Featured lentiviral transduction protocols.

The LV X-SCID protocol is currently employed in clinical trials for gene therapy to treat newborns with X-linked severe combined immunodeficiency (X-SCID). It is only referenced here, but the other protocols were explored experimentally. The 2-hit protocol mimics the LV X-SCID protocol. TU, transducing units; MOI, multiplicity of infection; HA, human albumin.

Protocol	CD34+ cells/ml	Transduction steps	TU/ml	MOI	RetroNectin	1%HA	Xenotransplant
Matrix (Fig. 1)	1-4 x 106	1	4-16 x 107	10-160	−	−	−
LV X-SCID [15]	1 x 106	2	2-2.7 x 108	200-270	+	+	NSG [30]
2-hit (Fig. 2)	1.3 x 106	2	2.6 x 108	200	+	+	NSG
2E6 (Fig. 2)	2 x 106	1	2 x 108	100	−	−	NSG
4E6 (Fig. 2)	4 x 106	1	2 x 108	50	−	−	NSG
2E6 (Fig. 3)	2 x 106	1	5.2 x 107	26	−	+	NBSGW [22]
2E6 (Fig. 4)	2 x 106	1	5 x 107	25	−	+	NBSGW

Xenotransplantation studies

The St. Jude Institutional Animal Care and Use Committee approved the use of mice in transplantation experiments, and the animal studies were performed according to relevant ethical regulations. Although no statistical methods were used for sample size estimate, the experiments were conducted with 4 or more mice per group for sufficient statistics. NOD.SCID.Il2rg^−/− (NSG) mice were initially obtained from The Jackson Laboratory (Bar Harbor, ME, USA), and bred at the St Jude Animal Research Center (ARC). One day prior to transplantation of transduced human CD34⁺ cells, 8- to 10-week-old female NSG mice underwent conditioning with 35 mg/kg busulfan (Busulfex, PDL BioPharma, Redwood City, CA, USA), and were injected with 1x10⁶ CD34⁺ cells per mouse via tail vein injection. Another immunodeficient mouse strain, NOD,B6.SCID.Il2rg^−/− Kit^W41/W41 (NBSGW) was recently described [22] and all recipient mice were purchased from The Jackson Laboratory. Six to 8-week-old female NBSGW mice were transplanted with 1-1.5x10⁶ cells per mouse, without conditioning. All mice were provided food and Baytril-water and Sulfatrim-water alternately ad libitum. Mice were randomly allocated to experimental groups before injection, and each one was identified by a unique ear tag. Blinding was done during animal allocation, and no blinding was done during experiments or outcome assessment. Blood was collected via retro-orbital puncture at 12 weeks post-transplantation as noted. After 16-18 weeks, mice were euthanized to collect spleen and bone marrow (BM). The BM cells were harvested from femurs and tibiae by either flushing or spinning [23], and analyzed for human cell chimerism and VCN. Human CD34⁺ cells in BM were purified using a magnetic sorting system using anti-human CD34 MicroBead kit (Miltenyi Biotec GmbH, Bergisch Gladbach, Germany) according to the manufacturer’s protocol.

Flow cytometry analysis

For immunophenotyping, BM cells were treated with RBC Lysis Buffer (BioLegend, San Diego, CA, USA) and stained with the following monoclonal antibodies: anti-human CD45-Allophycocyanin (Clone: HI30), CD34-phycoerythrin (PE; 581), CD19-PEcy7 (SJ25C1; BD Biosciences, San Jose, CA, USA), hCD33-PerCPcy5.5 (P67.6), and anti-mouse CD45-Brilliant Violet 605 (30-F11; BioLegend) [24]. 4′,6-diamidino-2-phenylindole (DAPI) was used to identify dead cell. All flow cytometric analyses were performed with BD LSR Fortessa (BD Biosciences) and FlowJo software (Tree Star, Ashland, OR, USA). In singlet population of live cells, the percentage of human CD45⁺ cells was calculated as [% hCD45⁺/(% hCD45⁺ + % mCD45⁺) x 100] and gating for percentages of human CD34⁺, CD19⁺ and CD33⁺ were performed on the human CD45⁺ population.

Evaluation of vector copy number (VCN)

To determine VCN pre-transplantation, 10,000 or 600 transduced CD34+ cells were cultured in 1 ml methylcellulose (MethoCult™) H4435 (Stem cell Technologies, Vancouver, Canada) for 6-7 or 13 days, respectively. For VCN analyses of single colony forming unit-granulocyte/macrophage (CFU-GM), 500 cells were cultured in 1 ml methylcellulose H4535 (3ml/group) for 13 days. Pooled cells were collected from the dishes and genomic DNA was extracted by Quick-DNA™ Mini-Prep Kit (Zymo Research, Irvine, CA, USA). To determine the VCN of human donor cells after engraftment, genomic DNA was extracted from unfractionated BM, purified hCD34⁺ cells, or spleen using the Quick-DNA™ Mini- or Micro-Prep Kit (Zymo Research). The VCN for most samples was measured by ddPCR using the QX200™ Droplet Reader and QuantaSoft software (Bio-Rad, Hercules, CA, USA) [21]. Briefly, genomic DNA was digested with MspI (New England Biolabs, Ipswich, MA, USA) and used as a template in PCR. The following primer-probe sets were used to amplify lentiviral psi (Ψ) and the human ribonuclease P/MRP subunit p30 (RPP30) for normalization (Ψ: 5’-ACTTGAAAGCGAAAGGGAAAC-3’, 5’-CACCCATCTCTCTCCTTCTAGCC-3’ and probe 5’FAM-AGCTCTCTCGACGCAGGACTCGGC-3’, RPP30: 5’-GCGGCTGTCTCCACAAGT-3’, 5’-GATTTGGACCTGCGAGCG-3’ and probe 5’HEX-CTGACCTGAAGGCTCT-3’). VCNs were determined by calculating the number of copies of Ψ to every 2 copies of RPP30. For experiments using LV X-SCID 2^nd gen, the VCNs of pre-infusion CD34⁺ cells and peripheral blood mononuclear cells (PBMCs) were measured by using real time quantitative PCR as previously described [15].

Identification of vector insertion sites

Vector insertion sites (VIS) were analyzed with quantitative shearing linear amplification PCR (qsLAM-PCR) as described with modification of the adaptor sequence [25]. Libraries were sequenced by Illumina MiSeq with a 150-bp Read length in paired run. Sequences in FASTQ files were trimmed using Cutadapt (v 1.14) with ATCCCTCAGACCCTTTTAGTCAGTGTGGAAAATCTC trimmed from forward reads and GACTGCGTATCAGT trimmed from reverse. An error rate (−e) of 0.1 was allowed and a minimum trimmed length (−m) of 20 was required. Trimmed reads were aligned to the hg19 reference using BWA (0.7.12-r1039). Aligned BAM files were converted to bed files using samtools and bedtools bamtobed. Reads were assigned to genes and gene features using R/Bioconductor libraries including ChIPpeakAnno, and TxDb.Hsapiens.UCSC.hg19.knownGene.

Statistical analysis

Statistical significance for Fig. 1 was performed by the slopes with a linear model and then comparison of the resulting slopes using one-way ANOVA with Tukey's post-hoc test using the GraphPad Prism 8 software (San Diego, CA, USA). Otherwise, analyses were performed in R=3.5.3 (R-Core Team, Vienna, Austria). Kruskal-Wallis test or analysis of variance (ANOVA) were applied to assess the overall difference among treatment groups depending on if normality assumption met. Pairwise two sample t test or exact Wilcoxon rank sum test was applied to examine whether there was significant difference between each pair of two treatment groups depending on the normality of the data. The data normality was checked using Shapiro-Wilk test. False Discovery Rate (FDR) [26] method was employed to adjust p-values for multiple testing problem. FDR-adjusted p-value <0.05 was considered significant for analyses as noted. In all figures, *, **, or *** refer to p<0.05, < 0.01, or <0.001, respectively.

Fig. 1. Lentiviral transduction is more efficient at higher CD34⁺ cell concentrations.

(A) CD34⁺ cells were transduced in 100 μl with lentiviral vector (LV) MND-eGFP, LV X-SCID 1^st gen, or LV X-SCID 2^nd gen at cell concentrations of 1, 2, or 4x10⁶ cells/ml, using 4, 8, or 16x10⁷ transducing units (TU) / ml. Multiplicity of infection [MOI] was calculated as virus TU / cell. (B) %GFP⁺ cells of CD34⁺ cells transduced with LV MND-eGFP and cultured in X-VIVO 10 media for 6 days. (C) Vector copy number (VCN) of CD34⁺ cells transduced with LV X-SCID 1^st gen and cultured in X-VIVO10 media for 9 days. (D) VCNs of pooled colony-forming unit cells. CD34⁺ cells were transduced with LV X-SCID 2^nd gen, seeded into methylcellulose cultures (500 cells/ml) supplemented with cytokines, and cultured for 13 days. Each experiment was performed in duplicate from a healthy donor CD34⁺ cells, and data are expressed as mean ± standard error of the mean (SEM). Statistical significance was determined by the slopes with a Linear model and then comparison of the resulting slopes using one-way ANOVA with Tukey's post-hoc test (*p<0.05 vs 1x10⁶ cells/ml).

Results

Lentiviral vector transduction is more efficient at higher CD34⁺ cell concentrations.

We performed a matrix analysis to test the effects of CD34⁺ cell concentration and LV dose on transduction. Normal donor CD34⁺ cells (1, 2 or 4 x 10⁶ cells/ml) were incubated with LV MND-eGFP (Table 1) at vector doses of 4, 8, or 16 x 10⁷ TU/ml, corresponding to a calculated MOI of 10 to 160 (Fig. 1A). After transduction, cells were cultured for 6 days in X-VIVO 10 with cytokines. To approximate transduction efficiency, eGFP expression was analyzed by flow cytometry (Fig. 1B). The highest vector dose (16 x 10⁷ TU/ml), achieved maximal %GFP⁺ cells, irrespective of cell concentration. At the lowest cell concentration (1 x 10⁶ cells/ml), increasing the LV dose had minimal effects on transduction efficiency.

We performed a similar matrix analysis of LV X-SCID 1^st gen (Table 1), an LV produced from a first-generation producer cell line and under clinical study for the treatment of X-SCID [7, 27], The VCN was determined by ddPCR after cell culture for 9 days. In contrast to the LV MND-eGFP, the LV X-SCID 1^st gen produced similar VCNs (~0.75) regardless of cell concentration or vector dose, suggesting that LV transduction was plateaued at lower VCN for the 1^st gen vector (Fig. 1C). We next tested the LV X-SCID 2^nd gen described in [28]. Following transduction, CD34⁺ cells were cultured in methylcellulose supplemented with cytokines for 13 days, then VCNs were measured for each condition. Similar to the results obtained with LV MND-eGFP (Fig. 1A), transduction level increased in vector dose (MOI)-dependent manner (Fig. 1D) suggesting that the 2^nd gen vector could plateau to the higher VCN than the 1^st gen LV X-SCID. Transduction with LV X-SCID 2^nd gen at the highest dose (16 x 10⁷ TU/ml) showed optimal VCN (about 0.9), irrespective of CD34⁺ cell concentration.

Overall, these findings indicate that optimal dosing for each LV preparation must be determined empirically. For some preparations (i.e, MND-eGFP and X-SCID 2^nd gen), it may be possible to reduce the number of LV particles required to achieve optimal HSC VCN by transducing CD34⁺ cells at higher density.

A “single hit” LV transduction protocol at high cell concentration modifies HSCs efficiently.

We next performed xenotransplantation studies to investigate conditions that optimize the VCN of repopulating HSCs after CD34⁺ cell transduction with LV X-SCID 2^nd gen (Table 2, Supplementary fig. 1) according to three different conditions: 1) “2E6”: a single transduction of 2 x 10⁶ cells/ml with 2 x 10⁸ TU/ml LV (MOI 100); 2) “4E6”: a single transduction of 4 x 10⁶ cells/ml with 2 x 10⁸ TU/ml LV (MOI 50); and 3) “2-hit”: similar to our standard clinical protocol [15, 19] that includes two successive transductions of 1.3 x 10⁶ cells/ml with 1.3 x 10⁸ TU/ml (MOI 100) on RetroNectin-coated plates (Table 2). One million transduced cells per mouse were transplanted into busulfan-treated NSG mice via intravenous injection. Eighteen weeks post-transplantation, mice were euthanized, and BM cells were analyzed for human donor cell engraftment, lineage composition, and VCN. Donor chimerism, measured as % human (h) CD45⁺ cells, was similar to that reported in other NSG xenotransplantation studies (Fig. 2A) [13, 29, 30, 31]. Chimerism of individual mice in the 4E6 and 2-hit groups was more variable compared to the non-transduced and 2E6 groups, although no significant difference was detected in % hCD45⁺ for each group. The human B cell (hCD19⁺) and myeloid cell (hCD33⁺) compositions were similar between groups (Fig. 2B). Pre-infusion VCNs were 1.89, 1.65, and 3.3, respectively for the 2E6, 4E6 and 2-hit groups (Fig. 2C). The 2E6 group showed the highest VCN (mean 1.02) in repopulating BM cells after xenotransplantation, although this was not significantly different from that of the 2-hit group (mean 0.8) (Fig. 2D). Thus, the VCN of pre-infusion CD34⁺ cells may not correlate with that of repopulating HSCs. Moreover, “single-hit” LV transduction at higher CD34⁺ cell concentrations (2 to 4 x 10⁶ cells/ml) may be equivalent or superior to a 2-hit transduction protocol performed at lower CD34⁺ cell concentration. Note that the 2E6 protocol consumed 2-fold less LV than the 2-hit protocol. Therefore, we employed this single transduction at a higher cell concentration 2 x 10⁶ cells/ml without RetroNectin-coating in the following experiments.

Fig. 2. Efficient HSC transduction using a “single hit” protocol at high cell density.

CD34⁺ cells (1x10⁶ cells/ml) were prestimulated for 2 days and transduced with LV X-SCID 2^nd gen. Untd: non-transduced; 2E6: 2x10⁶ cells/ml, 1 hit, MOI 100; 4E6: 4x10⁶ cells/ml, 1 hit, MOI 50; 2-hit: 1.3x10⁶ cells/ml, 1 hit per day for 2 days, MOI 200 in total. Ten thousand cells were cultured in 1ml methylcellulose supplemented with cytokines for 6 days to measure pre-infusion VCN, and the rest cells were injected into busulfan-conditioned NSG mice. Mice were sacrificed 18 weeks post-transplantation to analyze human cell engraftment, lineage composition, and VCN. (A) Percentage of human (h) CD45⁺ cells in bone marrow (BM), where each dot represents a single mouse and bars represent the mean for each group. (B) Percentage of hCD19⁺ (B cells), and hCD33⁺ (myeloid cells) present in engrafted hCD45⁺ cells (mean ± SEM). (C) Pre-infusion VCN from pooled methylcellulose cultures. (D) BM VCN for each transplanted mouse is shown as a dot and the mean for each group is expressed as a bar. HSC, hematopoietic stem cell.

LentiBOOST improves transduction of repopulating HSCs.

The clinical efficacy of LV gene therapy depends on achieving adequate post-infusion VCN, which can be limited by innate immune responses and other unknown causes [29, 32, 33]. Recently, PGE2 [13] and poloxamer F108 (LentiBOOST [17]) have been shown to improve LV transduction of CD34⁺ cells. We tested the effects of these molecules on transduction of CD34⁺ cells with LV X-SCID 2^nd gen (Table 1).

CD34⁺ cells from two different healthy donors were transduced with LV X-SCID 2^nd gen in a single hit protocol at 2 x 10⁶ cells/ml (MOI 26), with or without PGE2 and/or LentiBOOST (Table 2). A portion of transduced cells were seeded into methylcellulose cultures and the resultant hematopoietic colonies were analyzed to determine the pre-infusion VCN. The remainder of transduced cells were transplanted into NBSGW mice to determine the VCN of repopulating HSCs. This immunodeficient strain harbors a Kit mutation that facilitates human CD34⁺ cell engraftment and obviates the need for myelotoxic conditioning [22].

In vitro pre-infusion VCNs from four different donors, two of which were analyzed subsequently by xenotransplantation, ranged 1.06 to 2.30 for the untreated LV-control, and 1.91 to 6.68 for the transduction enhancer-treated groups (Supplementary Fig. 2). PGE2, LentiBOOST, and LentiBOOST+PGE2 significantly increased VCNs by 2.3-, 2.0-, and 3.2-fold, respectively, compared to control. Noticeably, the combination of LentiBOOST+PGE2 consistently yielded the highest VCN compared to either compound alone. For two donors used in xenotransplantation, individual CFU-GM colonies were also picked to estimate gene marking (Supplementary table 1). PGE2, LentiBOOST, and LentiBOOST+PGE2 increased the percentage of vector-positive colonies (LV=8.0%±0.1; PGE2= 37.2%±3.5; LentiBOOST= 63.0%±5.3; LentiBOOST+PGE2= 86.0%±2.8).

Transduced CD34⁺ cells from two different donors were transplanted into NBSGW mice. At twelve weeks post-transplantation, peripheral blood mononuclear cells (PBMCs) were sampled to assess VCNs. Without transduction enhancers, the VCN was approximately 0.5 in PBMCs derived from CD34⁺ cells of either donor. By contrast, the VCN ranged 0.8-1.5 for the PGE2 group, 1.8-2.3 for the LentiBOOST group, and 1.8 to 2.5 for the LentiBOOST + PGE2 group (Supplementary Figs. 3A,B).

At sixteen weeks post-transplantation, mice were euthanized to assess human cell engraftment, lineage composition, and most importantly VCNs of hematopoietic cells of different tissues. The human cell chimerism in BM of NBSGW mice averaged approximately 90% (Figs. 3A,B), representing substantial improvement over what was achieved after xenotransplantation into NSG mice (compare to Fig. 2A). Engraftment levels in all groups were comparable without statistical significance (Figs. 3A,B). Human CD19⁺ (B cell) and CD33⁺ (myeloid cell) fractions were similar among recipient mice that received cells from the same donor, but varied between donors (Figs. 3C,D). The fraction of human CD34⁺ cells in recipient BM was similar for both donors throughout all groups, suggesting that transduction enhancers did not negatively impact HSC engraftment (Figs. 3C,D).

Fig. 3. Transduction of repopulating HSCs is improved with LentiBOOST.

CD34⁺ cells from two donors were independently prestimulated for 1 day and transduced with LV X-SCID 2^nd gen at 2x10⁶ cells/ml using a MOI of 26, with or without transduction enhancers (PGE2, LB, LB+PGE2). Following each transduction, 10,000 cells were cultured in 1 ml methylcellulose supplemented with cytokines for 7 days to measure pre-infusion VCN and the rest were injected into NBSGW mice (5 mice per group; 1-1.5x10⁶ cells/mouse). Sixteen weeks post-transplant, mice were sacrificed to measure VCN in BM and spleen. Human cell engraftment and lineage composition were also measured in the BM. Experimental results for Donor 1 (A, C, E, G) and Donor 2 (B, D, F, H) are shown. (A, B) Percentage of hCD45⁺ cells in BM, where each dot represents a single mouse and bars represent the mean for each group. (C, D) Percentage of hCD19⁺, and hCD33⁺, and hCD34⁺ cells present in engrafted hCD45⁺ cells (mean ± SEM, * FDR-adjusted p<0.05 vs. LV, #p<0.05 vs. Untd, LV, and PGE2). (E, F) Pre-infusion VCN for each condition from pooled methylcellulose cultures. (G, H) BM VCN for each transplanted mouse is shown as a dot and the mean for each group is expressed as a bar (* FDR-adjusted p<0.05, **p<0.01, ***p<0.001 vs. LV). In H, * PGE2 vs. LB, and LB+PGE2. Untd, non-transduced; LV, lentiviral vector; PGE2, prostaglandin E2; LB, LentiBOOST; BM, bone marrow.

The VCNs of all human cells from BM or spleen were lower than pre-infusion levels (Table 3, Figs. 3G,H, Supplementary Figs. 3C-F), but similar to that of 12 week PBMCs (Supplementary Figs. 3A,B). The decrease in VCN relative to pre-infusion levels was lowest for the LentiBOOST group, and particularly high for the PGE2 group (Table 3). Despite variation between donors, all transduction enhancer-treated groups showed improved VCN over the untreated LV control. The mean VCN of human cells in both BM and spleen ranged from 0.7 to 2.2 for the PGE2 group, 1.6 to 3.1 for the LentiBOOST group, and 2.0 to 3.0 for the LentiBOOST+PGE2 group, representing 2.2-, 4.0-, and 4.4-fold enhancement of VCN by PGE2, LentiBOOST, and LentiBOOST+PGE2, respectively (Table 3, Figs. 3G,H, Supplementary Fig. 3). The average VCN of donor CD34⁺ cells purified from recipient BM was similar to that of whole BM (Supplementary Figs. 3E,F, Table 3). Together, these results indicate that LentiBOOST improves HSC transduction to a greater extent than PGE2 and that the addition of PGE2 to LentiBOOST did not further improve transduction of HSCs.

Table 3. Pre and post infusion VCN of cells transduced with LV X-SCID 2^nd gen in the presence or absence of transduction enhancers.

CD34⁺ cells from two normal donors were transduced with LV X-SCID 2^nd gen and transplanted into NBSGW mice (n=5 per group). The mean VCN ± standard deviation (SD) for pre-infused cells, and blood (12 weeks), whole BM, BM hCD34⁺ cells, and spleen (16 weeks) post transplantation are shown. The fold decrease in VCN of indicated tissues [preinfusion VCN/mean VCN of each group] is shown below. VCN, vector copy number, LV, Lentiviral vector; PGE2, prostaglandin E2; BM, bone marrow.

VCN	LV		PGE2		LentiBOOST		LentiBOOST+PGE2
	Donor 1	Donor 2	Donor 1	Donor 2	Donor 1	Donor 2	Donor 1	Donor 2
Preinfusion	2.30	1.06	5.60	2.44	4.63	2.77	6.68	4.42
Blood	0.64 ± 0.03	0.43 ± 0.07	1.53 ± 0.13	0.80 ± 0.09	2.31 ± 0.35	1.77 ± 0.23	2.54 ± 0.6	1.81 ± 0.31
BM	0.86 ± 0.15	0.43 ± 0.04	2.14 ± 0.29	0.91 ± 0.10	3.05 ± 0.79	1.89 ± 0.12	2.83 ± 0.87	2.35 ± 0.91
hCD34+	0.89 ± 0.18	0.45 ± 0.06	2.15 ± 0.42	0.88 ± 0.13	3.12 ± 0.78	1.97 ± 0.39	2.95 ± 1.33	2.43 ± 0.92
Spleen	0.64 ± 0.06	0.37 ± 0.05	1.58 ± 0.18	0.70 ± 0.10	2.36 ± 0.34	1.62 ± 0.17	2.58 ± 0.34	1.96 ± 0.31
Fold decrease versus preinfusion VCN
Blood	3.62	2.49	3.66	3.04	1.79	1.57	2.63	2.44
BM	2.66	2.49	2.62	2.69	1.35	1.46	2.36	1.88
hCD34+	2.59	2.37	2.61	2.76	1.32	1.40	2.27	1.82
Spleen	3.62	2.86	3.55	3.51	1.75	1.71	2.59	2.26

To determine the effects of PGE2 and LentiBOOST on clonal complexity, we analyzed vector insertion sites (VIS) of donor-engrafted cells in NBSGW recipient BM by quantitative shearing linear amplification-PCR (qsLAM-PCR) [25]. The total VIS number in all groups ranged from 391 to 2349 with no clone exceeding 6% in any group (Supplementary Figs. 4A,C, and 5). The LentiBOOST group exhibited the highest number of unique integration sites (up to 2349) commensurate with the associated increase in VCN relative to other groups (Supplementary Fig. 4B,D).

Cyclosporine H also improves transduction of repopulating HSCs.

We next tested the effect of LentiBOOST on transduction of CD34⁺ cells from an independent donor with a different clinically relevant LV. During the course of this study, cyclosporine H (CsH) was shown to increase LV transduction of HSCs [29]. Therefore, we compared the effects of CsH, LentiBOOST, or LentiBOOST+PGE2 on transduction of CD34⁺ cells with LV WASp [34] (Table 1), an LV now in preclinical studies of gene therapy for Wiskott-Aldrich syndrome (WAS). 2 x 10⁶ cells/ml were transduced with LV WASp at 5x10⁷ TU/ml (MOI 25) with PGE2, LentiBOOST, LentiBOOST+PGE2, or CsH (Table 2, Supplementary Fig. 1). A fraction of cells were seeded into methylcellulose cultures to measure pre-infusion VCN and the remainder were transplanted into NBSGW mice. At 16 weeks post-transplantation, unlike two previous transplantations, the LentiBOOST+PGE2 and CsH groups trended toward slightly lower percentages of engrafted hCD45⁺ and hCD34⁺ cells compared to LV control or LentiBOOST groups (Figs. 4A,B). This result suggests that LentiBOOST+PGE2 and CsH may have some effects on engraftment, but the current study was not configured to address this parameter. Specifically input cell dose was too high and it will be necessary to perform dose-limiting or competitive repopulation assay to better address the effect of transduction enhancers on engraftment. The human cell lineage composition was similar for each group except that PGE2 treated cells exhibited slightly increased fraction of CD19⁺ and commensurately reduced % hCD33⁺ cells (Fig. 4B).

Fig. 4. HSC VCN is improved for another clinically relevant vector using LentiBOOST or cyclosporine H.

CD34⁺ cells (1x10⁶ cells/ml) were prestimulated for 1 day and transduced with LV WASp at 2x10⁶ cells/ml using a MOI of 25, with or without transduction enhancers (PGE2, LB, LB+PGE2, or CsH). Ten thousand cells were cultured in 1 ml methylcellulose supplemented with cytokines for 7 days to measure pre-infusion VCN and the rest were injected into NBSGW mice. Mice were sacrificed 16 weeks post-transplant to measure VCN in BM and Spleen. Human cell engraftment and lineage composition were also measured in the BM. (A) Percentage of hCD45⁺ cells in BM, where each dot represents a single mouse and bars represent the mean for each group (* FDR-adjusted p<0.05 vs. LB). (B) Percentage of hCD19⁺, and hCD33⁺, and hCD34⁺ cells present in engrafted hCD45⁺ cells (mean ± SEM, *p<0.05 vs. LV and LB). (C) Pre-infusion VCN was measured from pooled methylcellulose cultures. (D) BM VCN for each transplanted mouse is shown as a dot and the mean for each group is expressed as a bar (* FDR-adjusted p<0.05 vs. LV). * CsH vs. LB, and LB+PGE2, * PGE2 vs. LB, LB+PGE2, and CsH. Untd, non-transduced; LV, Lentiviral vector; PGE2, prostaglandin E2; LB, LentiBOOST; CsH, cyclosporine H; BM, bone marrow.

Compared to LV control without transduction enhancer, addition of PGE2, LentiBOOST, and LentiBOOST+PGE2 increased the pre-infusion VCNs, from 0.46 to 1.4, 0.77 and 2.16, respectively (Fig. 4C, Table 4), similar to what we observed for transduction with LV X-SCID 2^nd gen (see Fig 3D, above). Cyclosporine H increased pre-infusion VCN to 1.04. The VCN in PBMCs at 12 weeks post-transplantation was increased by LentiBOOST, LentiBOOST+PGE2 and CsH but not by PGE2 (Supplementary Fig. 6A, Table 4). Similarly, at 16 weeks post-bone marrow transplantation, the LentiBOOST and CsH groups in recipient BM exhibited significantly higher VCNs compared to the control group (Table 4, Fig. 4D, Supplementary Figs. 6B,C). For example, the VCN of unfractionated donor-derived cells was enhanced 2.5- and 3.4-fold by LentiBOOST or CsH, respectively. Effect of LentiBOOST is similar to those obtained using the LV X-SCID 2^nd gen (Fig. 3, above). These results also independently verify a previous study showing that CsH enhances HSC transduction. Of notes, CsH increased the repopulating HSC VCN to a greater extent compared to LentiBOOST alone in this experiment (Fig. 4D and Supplementary Figs. 6B,C).

Table 4. Pre and post infusion VCN of cells transduced with LV WASp in the presence or absence of transduction enhancers.

Normal donor CD34⁺ cells were transduced with LV WASp and transplanted into NBSGW mice (n=5 per group). The mean VCN ± SD for pre-infused cells, and blood (12 weeks), whole BM, BM hCD34⁺ cells, and spleen (16 weeks) post transplantation are shown. The fold decrease in VCN of indicated tissues [preinfusion VCN/mean VCN of each group] is shown below. VCN, vector copy number; LV, Lentiviral vector; PGE2, prostaglandin E2; CsH, cyclosporine H; BM, bone marrow.

VCN	LV	PGE2	LentiBOOST	LentiBOOST+PGE2	CsH
Preinfusion	0.46	1.40	0.77	2.16	1.04
Blood	0.15 ± 0.03	0.20 ± 0.05	0.48 ± 0.09	0.42 ± 0.08	0.68 ± 0.2
BM	0.22 ± 0.07	0.22 ± 0.05	0.54 ± 0.10	0.47 ± 0.04	0.75 ± 0.11
hCD34+	0.21 ± 0.12	0.23 ± 0.08	0.53 ± 0.15	0.44 ± 0.09	0.73 ± 0.17
Spleen	0.15 ± 0.03	0.28 ± 0.06	0.48 ± 0.08	0.50 ± 0.04	0.66 ± 0.14
Fold decrease versus preinfusion VCN
Blood	3.11	6.93	1.6	5.19	1.53
BM	2.09	6.42	1.43	4.64	1.39
hCD34+	2.16	6.04	1.46	4.92	1.43
Spleen	3.14	5.07	1.61	4.29	1.58

Discussion

Applications of LV-mediated gene replacement therapy for hematological disorders continue to expand and improve in parallel with technological advances. An essential criterion for success in autologous gene therapy is to achieve high-level transduction of long-term HSCs to support therapeutic transgene expression in a high proportion of patient blood cells [4]. The magnitude of transgene expression required for successful gene therapy varies between different disorders and cell types. In particular, higher VCNs (≥1) are required when gene rescue does not confer a survival or selection advantage to the affected lineage. In addition, HSC transduction can vary between donors and some complex LVs are inherently less efficient for HSC transduction [35].

Our group is currently engaged in clinical trials examining the safety and efficacy of LV gene therapy for infants with X-SCID. We demonstrated that clinically effective VCNs are attained using a protocol in which 1 x 10⁶ autologous CD34⁺ cells/ml are transduced with LV at total MOI of 200-270 on two consecutive days [15]. In a related study of older children and adults with X-SCID, clinically effective HSC VCNs were achieved via two-step transduction at high MOI (100-150) [7].

Aiming to increase HSC VCN in a simpler transduction procedure, we compared several transduction protocols and small molecule enhancers. Importantly, we employed xenotransplantation assays in order to assess the VCN of long-term repopulating HSCs after transduction with clinical or clinically relevant LVs. Our results confirm several prior studies and illustrate optimized approaches to improve HSC transduction.

Compared to our current clinical CD34⁺ cell transduction protocol, similar VCNs can be achieved after a single-step transduction of 2-4 x 10⁶ CD34⁺ cells/ml with lower MOI (25-100), thereby requiring up to 8-fold less LV and shortening the duration of ex vivo culture (Table 2). These findings are in agreement with those of Uchida et.al., who found that higher cell densities (2-4 x 10⁶ cells/ml) improved LV transduction of CD34⁺ cells by increasing cell-cell contact during culture [36].

Recently, several groups have shown that various small molecules, collectively referred as transduction enhancers, could increase the transduction efficiency of CD34⁺ cells by lentiviral vector particles through a number of distinct mechanisms [13, 17, 29, 37] (reviewed in [3]). For example, Heffner et. al. reported that PGE2 increased CD34⁺ cell transduction, likely by acting on post-entry of LV particles [13]. Ozog et. al. showed that caraphenol A enhances LV delivery into HSCs by altering innate antiviral factors [37]. Schott et. al. tested the effects of eight transduction enhancers including PGE2 and LentiBOOST on LV transduction of CD34⁺ cells in vitro and concluded that a combination of LentiBOOST and protamine sulfate produced optimal enhancement of VCN [38]. Petrillo et. al. showed that CsH improves the transduction of repopulating human HSCs by inhibiting the interferon-induced transmembrane protein 3, which blocks LV entry [29].

We initially hypothesized that LentiBOOST and PGE2 may enhance transduction synergistically through different mechanisms: LentiBOOST to promote vector particle entry and PGE2 to modify post-entry factors. Indeed, the measurement of VCN in vitro shortly after transduction supports this hypothesis in multiple donors and LV preparations (Supplementary Fig. 2, Figs. 3E,F, and Fig. 4C). Moreover, two independent groups reported recently that LentiBOOST-like poloxamer compounds combined with PGE2 increased the VCN of repopulating HSCs in a xenotransplantation model, although neither compound was examined alone [36, 39]. Hence, we were surprised to discover that LentiBOOST alone enhanced LV transduction of repopulating HSCs to a similar degree as that LentiBOOST + PGE2 (Figs. 3G,H, and Fig. 4D). In addition, we confirmed that CsH improves the VCN of repopulating HSCs to a greater extent to LentiBOOST (Fig. 4D). More transplantation studies are necessary to confirm the effect of CsH in comparison to LentiBOOST. Further experiments with CsH and LentiBOOST will be performed in our laboratory.

These results were obtained in xenotransplantation experiments using CD34⁺ cells from three normal donors and using two different clinical LVs. Overall, our findings confirm previous work showing that LentiBOOST consistently enhances LV transduction of human HSCs. Surprisingly, in our dataset addition of PGE2 to LentiBOOST conferred no additive or synergistic increase in VCN in vivo. More generally, our data further highlight the important point that analysis of CD34⁺ cells shortly after LV transduction does not consistently predict the VCN of bone marrow repopulating cells, thereby emphasizing the importance of xenotransplantation assays to assess the effects of transduction protocols on HSCs. Loss of gene marking from in vitro to in vivo was more dramatic with PGE2 alone or LentiBOOST+PGE2 combination, whereas LentiBOOST alone or CsH represented the least VCN decrease pre- and post-graft (summarized in Tables 3 and 4). Although VCN of pre-graft is a representative parameter to validate lentiviral gene therapy drug products, this parameter may not be necessarily translated into clinically relevant readout in vivo. Additional studies using CD34⁺ cells from patients with hematopoietic diseases, and varying viral concentrations and injecting cell dose will be required to determine whether the combinations of transduction enhancers will lead to additional improvements of VCN in engrafted HSC without toxicity.

There are some caveats to the interpretation of our results. First, the additive enhancement of transduction with LentiBOOST and PGE2 noted in cultured CD34⁺ cells may indicate potential clinical benefits since transduced progenitor cells could expedite the production of differentiated functional cells, such as T lymphocytes in X-SCID patients. Along these lines, functional T cell progenitors can exist for many years in vivo [40]. Second, reconstitution of lentiviral vector-transduced human HSCs may differ after autologous transplantation compared to xenotransplantation into mice. Hence, further studies are necessary to make firm conclusion on combinatorial use of transduction enhancers.

As the VCN of repopulating HSCs is increased by transduction enhancers, it is important to examine their effects on clonal diversity. We used VIS analyses to show that transduction enhancers increase the clonal diversity of repopulating HSCs (Supplementary Figs. 4 and 5). Moreover, the high number of unique VIS (up to 2,349) from a single mouse is remarkable and indicates the utility of NBSGW mice for examining highly polyclonal human HSC reconstitution. Importantly, use of transduction enhancers was not associated with the emergence of dominant repopulating HSC clones. In this regard, human donor HSC chimerism was consistently higher in NBSGW mice recipients compared to the NSG strain. Thus, use of the NBSGW xenotransplantation model combined with transduction enhancers allowed us to evaluate the repertoire of LV integration sites more comprehensively.

In conclusion, our results demonstrate that a simplified lentiviral transduction procedure can be used to improve the VCN of repopulating HSCs for LV gene therapy. Use of a single vector dose at relatively high cell concentration (2 to 4 x 10⁶ cells/ml), and a transduction enhancer (LentiBOOST or CsH) increased the long-term HSC transduction by approximately 2- to 3-fold over a clinical gene therapy transduction protocol and required much fewer LV particles. Based on some of this work, we have amended our current clinical trial for infants with X-SCID (LVXSCID-ND, NCT01512888) to include a 1-hit transduction with 2^nd gen LV at higher CD34⁺ cell concentration without any transduction enhancer. Thus, our findings should help to simplify manufacturing protocols and reduce costs associated with LV gene therapy for blood disorders in general. Future studies should be directed toward validating the findings reported here across multiple HSC donors, LV vectors and blood diseases, particularly with respect to the efficacy and safety of transduction enhancers individually and in combination.