Endothelial Cells Promote Expansion of Long‐Term Engrafting Marrow Hematopoietic Stem and Progenitor Cells in Primates

Abstract Successful expansion of bone marrow (BM) hematopoietic stem and progenitor cells (HSPCs) would benefit many HSPC transplantation and gene therapy/editing applications. However, current expansion technologies have been limited by a loss of multipotency and self‐renewal properties ex vivo. We hypothesized that an ex vivo vascular niche would provide prohematopoietic signals to expand HSPCs while maintaining multipotency and self‐renewal. To test this hypothesis, BM autologous CD34+ cells were expanded in endothelial cell (EC) coculture and transplanted in nonhuman primates. CD34+C38− HSPCs cocultured with ECs expanded up to 17‐fold, with a significant increase in hematopoietic colony‐forming activity compared with cells cultured with cytokines alone (colony‐forming unit‐granulocyte‐erythroid‐macrophage‐monocyte; p < .005). BM CD34+ cells that were transduced with green fluorescent protein lentivirus vector and expanded on ECs engrafted long term with multilineage polyclonal reconstitution. Gene marking was observed in granulocytes, lymphocytes, platelets, and erythrocytes. Whole transcriptome analysis indicated that EC coculture altered the expression profile of 75 genes in the BM CD34+ cells without impeding the long‐term engraftment potential. These findings show that an ex vivo vascular niche is an effective platform for expansion of adult BM HSPCs. Stem Cells Translational Medicine 2017;6:864–876


INTRODUCTION
Genetic diseases affecting the bone marrow (BM) and hematopoiesis can be treated by transplantation of gene-corrected autologous or matched allogeneic hematopoietic stem and progenitor cells (HSPCs) or allogeneic HSPCs from umbilical cord blood (CB) donors. For standard of care, HSPCs are mobilized from BM and CD34 + cells collected from apheresis product [1]. Efficacy of gene-modified HSPC transplantation (HSPCT) requires a threshold engraftment level, which is proportional to the CD34 + cell dose [1]. Many patients have predictive factors associated with poor mobilization [2,3] and would thus be ineligible for HSPC gene therapy. Allogeneic HSPCT is an alternative to autologous HSPC gene therapy; however, identification of a human leukocyte antigen-matched donor is often challenging. The side effects from allogeneic HSPCT can also be substantial and life-threatening [4]. Therefore, development of a safe and effective BM HSPC expansion platform would increase the number of patients eligible for autologous HSPC gene therapy.
We developed a vascular niche from human endothelial cells (ECs) that expands CB HSPCs with retained repopulation potential [15,16]. The focus of the present study was to evaluate the vascular niche for BM HSPC expansion. We found that EC-expanded gene-modified BM HSPCs reconstituted long-term hematopoiesis in the nonhuman primate. Our results indicate unprecedented marking in T lymphocytes, erythrocytes, and platelets, without selection. These findings reveal that EC-mediated expansion of BM HSPCs is both safe and effective and, thus, has the potential to improve gene-modified HSPCT for the treatment of lymphoid (HIV, severe combined immunodeficiency [SCID]) and erythroid (hemoglobinopathies) hematologic diseases.

CD34 + Cell Isolation and Transduction
Cell isolation, transduction, and transplantation for nonhuman primate controls were conducted as described previously [17,18]. Marrow aspiration was performed, and CD34 + cells were isolated [19]. For ex vivo studies, CD34 + cells were cryopreserved and thawed directly onto ECs, as described. For in vivo studies with EC-expanded gene-modified cells and for unexpanded controls, CD34 + cells were collected from three naïve animals and transduced as described.
For transduction, CD34 + cells were cultured overnight and transduced. The lentivirus vector SMPGW used in the present study has been previously described [18]. In brief, the concentrated lentivirus vector (4.5 3 10 8 transduction units per milliliter) expresses human P140K-MGMT, regulated by spleen focus-forming virus promoter, and enhanced green fluorescent protein (GFP), regulated by human phosphoglycerate kinase promoter. Lentivirus vector (multiplicity of infection [MOI], 10) was added and cultured overnight. The second dose of lentivirus (MOI, 10) was added the next morning and transduced for an additional 8 hours (total transduction: MOI, 10 3 2). The animals underwent myeloablative irradiation (linear accelerator source, hyperfractionated over 2 days). Transduced cells were collected, washed, and infused into the irradiated animals. For ex vivo EC expansion with primed CD34 + cells, the cells were collected from 3 naïve donors as described previously [17,18], cryopreserved, thawed, and plated onto ECs as described.
On the day of transplantation, the cells were resuspended in StemSpan SFEM containing 10 mM PGE 2 (Cayman Chemical, Ann Arbor, MI, http://www.caymanchem.com), placed on ice, and vortexed every 30 minutes for 2 hours. The infusion product (autologous gene-modified HSPC and human EC mixture) was filtered and suspended in 2% autoserum in saline. The animals underwent total body irradiation, 1,020 cGy, administered from a linear accelerator at 7 cGy/minute, as four equally divided doses 12 hours apart. Within 24 hours of radiation, cocultured HSPCs/ ECs were intravenously infused into the donor animal. Hematopoietic recovery based on the complete blood count (platelet count .20,000 per microliter and absolute neutrophil count [ANC] .500) was defined as described previously [21].

Cytospin Analysis
A total of 100,000 cells were placed in cytocentrifuge, distributed onto slides, Wright-stained, mounted, and analyzed.

Retroviral Integration Site Analysis
Mapping of integration sites (IS) was conducted on WBCs and sorted subsets, as described previously [22,23]. Genomic DNA was extracted from leukocytes using the Qiagen blood DNA mini kit (Qiagen, Hilden, Germany, http://www.qiagen.com) in accordance with the manufacturer's instructions. Lentivirus long terminal repeat (LTR)-genome junctions were amplified by modified genomic sequencing-PCR. Libraries were subjected to ion torrent semiconductor sequencing, and sequence reads were analyzed using custom Perl scripts (Dr. Grant Trobridge, Washington State University, Pullman, WA). Genomic sequences were mapped to the rhesus macaque genome (rheMac3) using a standalone version of Blat available from the UCSC Genome Browser (UCSC Genome Informatics Group, Center for Biomolecular Science & Engineering, Santa Cruz, CA, http://www.genome.ucsc.edu). Sequences corresponding to the same integration locus were grouped together to determine the total number of unique integration sites (clones) identified. The relative contributions of each clone were determined by the number of IS-associated sequence reads corresponding to that clone. A quality control check was performed to reveal clones overrepresented by PCR bias by comparing the number of IS-associated sequence reads with the number of different fragment lengths observed for each genomic locus.

RNA-Sequence Library Construction, Sequencing, and Analysis
RNA isolation, library construction, sequencing, and analysis were performed as described previously [24]. CD34 + cells from donors in the transplantation studies were subdivided into two subgroups. In group 1, unexpanded CD34 + cells went into TRIzol (Thermo Fisher Scientific Life Sciences, Waltham, MA, http:// www.thermofisher.com) for RNA extraction. In group 2, CD34 + cells plated into EC coculture for 1 week, a fraction of which were FACS sorted to deplete the human ECs (by exclusion of human CD147 + cells by staining with Tra-1-85 antibody that does not cross-react with monkey cells) and sorting on the CD34 + fraction (the cells were ∼98% CD34 + after 1 week of coculture). The second fractions were injected into nonhuman primates. All original RNA-sequence data were uploaded to the NCBI Gene Expression Omnibus database (National Center for Biotechnology Information, NIH, Bethesda, MD, http://www.ncbi.nlm.nih.gov/geo/) with a public release date of September 1, 2015.

Study Approval
Juvenile macaques (Macaca nemestrina) were housed at the University of Washington Regional Primate Research Center under conditions approved by the American Association for the Accreditation of Laboratory Animal Care. Healthy macaques were randomly assigned to the present study. The experiments were conducted under protocols approved by the institutional review board and the animal care and use committees of the University of Washington. Transplantation, priming (mobilization), and cell collection procedures and protocols have been previously described [17,18].

ECs Expand Macaque Primed and Steady-State BM CD34 + Cells
We previously showed that human ECs support the expansion of mouse HSPCs [20] and human CB CD34 + cells [16] capable of hematopoietic reconstitution. We also showed that ECs support specification of hematopoietic progenitor cells from induced pluripotent stem cells through a Notch pathway mechanism [22]. In the present study, we hypothesized that a vascular niche would support expansion of BM HSPCs for reconstitution of nonhuman primates.
To evaluate whether the vascular niche would support expansion of adult HSPCs, we compared the expansion of steady state BM CD34 + cells obtained from multiple naïve donors and primed/ mobilized CD34 + cells collected from animals after treatment with G-CSF/SCF as described previously [17,18]. BM CD34 + CD38 2 cell expansion peaked on day 7, and expansion of primed CD34 + CD38 2 cells peaked on day 10 ( Fig. 1A). Endothelial coculture supported 15-fold expansion of steady-state BM CD34 + CD38 2 cells, which was significantly higher compared with cells cultured in cytokines (10-fold expansion) or primed CD34 + cells cultured with ECs (7-fold expansion). Endothelial-expanded BM CD34 + cells maintained higher colony-forming potential compared with cytokine-expanded BM CD34 + cells or EC-expanded primed CD34 + cells (p , .005; unpaired two-tailed t test; Fig. 1B).
Analysis of hematologic recovery revealed that the ANC was .500 cell per microliter by day 17 for two animals [28]. The third animal achieved neutrophil engraftment on day 21; this delayed reconstitution, blood transfusion requirements, and failure to recover platelet counts were attributed to latent malarial infection. For animals A11224 and Z13018, platelet recovery (.20,000 cells per microliter) without transfusion occurred after neutrophil recovery, consistent with the kinetics of recovery observed after clinical mobilized peripheral blood HSPCT (range, 16-20 days) [29]. These data show that ECs do not cause infusional toxicity and do not persist and indicate that EC-expanded CD34 + cells exhibit similar engraftment kinetics to patients undergoing HSPCT.

Long-Term Multilineage Engraftment of EC-Expanded Gene-Modified CD34 + Cells
To confirm that EC-expanded gene-modified CD34 + cells contribute to hematopoiesis, transgene marking was analyzed in blood subsets after HSPCT. At the time of neutrophil engraftment, the percentage of GFP + (gene-modified) blood granulocytes was 50% in all animals (Fig. 3, left). By day +50, 25% of GFP + lymphocytes were detected. GFP + lymphocytes stabilized (A11224, Z13018) or doubled between days +50 and +100 (A11208). These data are consistent with HSPCT in historic monkeys and patients, in that the initial marking was higher in granulocytes than in lymphocytes, as myeloid reconstitution occurs before lymphocyte recovery [17]. The number of GFP + granulocytes and lymphocytes correlated positively with the CD34 + cell dose and was 1-log higher for the recipient that received the highest cell dose (Fig. 3, middle; Z13018, 60 3 10 6 CD34 + cells per kilogram). Gene marking in lymphocytes .300 days after transplantation ranged from 11% to 34%, with up to 30% marking in CD3 + T lymphocytes (supplemental online Table 1). Marking in erythrocytes and platelets from A11224 and Z13018 were ∼5% and ∼25% GFP + at days ∼700 and ∼450 after transplantation, respectively (Fig. 4A).
Analysis of marrow biopsies at day +120 revealed GFP marking in granulocytes (∼35%), monocytes (∼30%), and lymphocytes/blasts (∼20%), but no human ECs were detected (supplemental online Fig. 1). Six percent of LT-HSPCs (CD34 + CD38 2 CD90 + CD49f + CD45RA 2 ) in the marrow were GFP + . Gene marking in LT-HSPCs at day +120 after transplantation was the same as the level of gene marking in the blood +800 days after transplantation (A11224). This finding is consistent with the canonical view of HSPC ontogeny, in which committed progenitors from HSPCs undergo proliferation on route to repopulation of the hematopoietic system with lineage-committed mature blood cells [30].
We tracked engraftment by quantitative PCR analysis for detection of GFP and the lentivirus LTR, the latter to rule out contribution of lentiviral transduced (LTR + GFP 2 ) ECs. No difference was found between the GFP and LTR viral copy number per circulating leukocyte (VCN), which stabilized at 2 and 5 VCN, respectively, for A11224, Z13018, and A11208 (Fig. 3, right).

EC-Expanded CD34 + Cells Reconstitute Polyclonal Hematopoiesis
To determine polyclonal repopulation, we performed retroviral integration site analysis on blood leukocytes. For the longest engrafted animal (A11224), reconstitution was polyclonal in bulk leukocytes. Two hundred different clones were detected at each time point (Fig. 5). We examined clonal diversity in sorted blood cell subsets. Between days +182 and +322, we observed no restriction of clonal diversity in CD3 + cells, and tracked multiple clones over time. At day 312, we detected shared clones in the CD3 + , CD20 + , and granulocyte lineages, demonstrating that EC-expanded gene-modified LT-HSPCs were present at the time of infusion and were capable of sustaining hematopoiesis 1 year after transplantation. In Z13018, we also observed highly polyclonal reconstitution over time up to 107 days after transplantation, with nearly 2,000 individual clones identified at each time point (Fig. 5). Early reconstitution was so clonally diverse that not a single clone identified was found to contribute .1% of the genomes analyzed.
For A11208, we analyzed one time point (+95 days) and found polyclonal reconstitution. Based on the level of lentivirus genemodified blood leukocytes (44.6%), the overall contribution of the most abundant clone to leukocytes was ∼8.5%. This proviral integration mapped within the sequence encoding mRNA JV727540 in the rhesus genome (BGI CR_1.0/rheMac3 assembly). The function of this mRNA is unknown, and integration within this mRNA sequence has not previously been reported in conjunction with insertional mutagenesis; however, we cannot rule out the disease state in this animal (fatal malaria infection) as a possible cause for the clonal skewing observed at this time point. These data indicate that ECs promote polyclonal CD34 + cell expansion ex vivo.

Whole Transcriptome Analysis Reveals Differential Gene Expression in Expanded and Unexpanded CD34 + Cells
Given that EC-expanded cells retain engraftment potential, we next compared the gene expression profiles of expanded and unexpanded CD34 + cells. CD34 + cells from three donors cultured on ECs for 7 days were FACS depleted of human ECs with Tra-1-85 antibody, which binds to the human CD147 surface antigen but does not bind to macaque cells [22], combined with immunomagnetic selection for CD34 + cells (Fig. 6). Transcriptome analysis of populations of CD34 + cells revealed that coculture with cytokines plus ECs or cytokines alone alters the HSPC transcriptional profile (Fig. 6). Of the 75 genes that were differentially expressed in EC-expanded cells compared with the other populations, the most differentially expressed gene was FOS, which was identified Gori, Butler, Kunar et al.    [32]. Notably, CD86 is also upregulated in EC-expanded CD34 + cells. CD86 + HSCs have prolonged selfrenewal activity and lymphopoietic potential, and loss of these cells resulted in impaired lymphopoiesis in vivo [33]. These findings are consistent with previous studies indicating that the vascular niche balances HSPC self-renewal and differentiation, in part, through Notch signaling [20]. We have shown that ECs expand repopulating CD34 + cells and alter their transcriptional profile without compromising engraftment, as indicated by the long-term engraftment and robust lymphoid reconstitution achieved for animals A11224 and Z13018.

DISCUSSION
Methods that support ex vivo maintenance and/or expansion of long-term repopulating bone marrow-derived CD34 + cells would improve engraftment and therapeutic benefit for gene-modified HSPC gene therapy applications. Validation of long-term engraftment of expanded cells in a clinically predictive animal model would enable clinical trials and translation. We [15,16], and other groups [34][35][36], have shown that human CD34 + CD38 2 cells expand severalfold after coculture with endothelial cells and reconstitute hematopoiesis in vivo, with increased SCID mouse repopulation potential compared with either unexpanded or cytokine-expanded CD34 + cells. In previously published studies describing EC-mediated human BM CD34 + cell expansion from Chute et al., CD34 + CD38 2 cells cocultured with primary ECs plus cytokines expanded 5-fold [34] and 13-fold [36]. In our studies, the EC and cytokine coculture system supported up to 17-fold expansion of CD34 + CD38 2 cells (before PGE 2 exposure).
In previously published studies of the nonhuman primate, expanded autologous BM CD34 + cells reconstituted hematopoiesis in and rescued lethally irradiated baboons after coculture with porcine microvascular endothelial cells (PMVECs) [37] or human brain microvascular endothelial cells [38]. These studies provided the first evidence that endothelial cells produce prohematopoietic factors capable of simultaneously increasing the CD34 + cell dose by severalfold while maintaining long-term repopulating HSPCs. However, use of either xenogenic (porcine) or human brain ECs as an "off the shelf" HSPC expansion product is not clinically feasible, because sourcing ECs from xenogenic (pig) or adult human organs (brain tissue) for clinical use and expansion of primary unmodified ECs would present challenges.
Toward the clinical development of an HSPC expansion platform that could be an "off-the-shelf" product for use in clinical applications, we previously developed an ex vivo vascular niche from E4ORF1-transduced, Akt-activated endothelial cells from human cord blood, a source that is readily available and contains cells that are easily harvested [15,16]. These Akt-activated ECs produce additional prohematopoietic and angiocrine factors compared with unmodified primary ECs. In the present study, we demonstrated for the first time the ability of this ex vivo vascular niche to support CD34 + cell expansion, leading to long-term engraftment of EC-expanded BM CD34 + cells in a clinically relevant model. Transplant recipients showed long-term (500-800 days after transplantation) multilineage polyclonal HSPC engraftment, robust gene-modified lymphoid reconstitution, and gene marking in platelets and erythrocytes at levels not achievable previously after transplantation of mobilized/primed transduced CD34 + cells without in vivo selection. EC-expanded CD34 + cells have a unique transcriptional profile, with upregulation of genes that are important for self-renewal.
Although our data clearly demonstrate long-term engraftment of EC-expanded CD34 + cells, further studies focused on identifying the precise factors produced by ECs that maintain and expand LT-HSPCs and/or enhance homing to and lodging in the marrow would facilitate clinical translation. Chute et al. [39], and others, demonstrated that transplantation of ECs alone rescues in vivo hematopoiesis of irradiated mice, highlighting the critical role of the vascular niche in hematopoietic regeneration. Regenerative hematopoiesis through contact with prohematopoietic factors is endothelial-context dependent, as human HSPCs exposed to irradiation and then cultured with ECs recovered reconstitution potential while irradiated HSPCs cultured with cytokines alone did not rescue HSPC reconstitution [40]. Toward understanding the mechanism through which ECs support hemogenic regeneration after injury, subsequent studies in transgenic mice showed that conditional deletion of prohematopoietic factors in ECs, including vascular endothelial growth factor [41], SCF [42], chemokine C-X-C motif ligand 12 [43], Notch ligand Jagged-1 [44], and pleiotrophin [45], impaired hematopoietic reconstitution. However, supplementing expansion cultures with prohematopoietic factors alone is not sufficient to maintain the HSPC reconstitution potential, suggesting that there are likely additional unidentified acellular paracrine or autocrine factors produced by the vascular niche that are required for LT-HSPC maintenance. Current CB HSPC expansion strategies have used small molecules [14] or BM mesenchymal cells [46] or have exploited the modulation of key signaling pathways [19] that balance HSPC homeostasis, self-renewal, and differentiation. Csaszar et al. [47] demonstrated that Notch ligand delta-1 (DL1)-mediated HSPC expansion in the context of a fed-batch system reduced IL-6 cis-and trans-signaling in CB HSPCs, increasing HSPC engraftment. Although such methods have been successful for expanding CB HSPCs, they have been almost exclusively evaluated with CB . For determination of gene marking in lymphoid subsets, cells in the lymphoid subgate within the forward scatter by side scatter (FSC 3 SSC) gate were costained with fluorophore-conjugated CD3, CD4, CD8, and CD20 antibodies. The percentages of GFP + cells were then evaluated within the CD3, CD4, CD8, and CD20 subset gates. Abbreviations: GFP, green fluorescent protein; RBCs, red blood cells.  [7][8][9]. Although PGE 2 increased homing, no difference was seen in the engraftment of PGE 2 -treated mobilized CD34 + cells and untreated cells. In our study, gene-modified BM CD34 + cells were expanded on ECs and then briefly pulsed with PGE 2 , which might have affected HSPC homing. However, the increase in CD34 + CD38 2 cell numbers that occurred was attributed to EC plus cytokine cocultures and not to PGE 2 , because the cells were not transduced, cultured, or expanded in the presence of PGE 2 . We hypothesized that the combination of EC expansion, EC coinfusion, and PGE 2 pulse may synergize to improve CD34 + cell homing to, and lodging in, the bone marrow, thereby enhancing long-term hematopoietic reconstitution.
Additional transplantation studies of gene-modified CD34 + cells in nonhuman primates that evaluate the effect of each independent variable on HSPC engraftment would clarify the mechanism through which EC expansion and EC infusion independently  alone is likely not sufficient to rescue/recover the engraftment potential of an EC-expanded CD34 + product that had lost all multipotent repopulating cells during ex vivo culture. We have also tested other reagents that, like PGE 2 , showed promising results for CB CD34 + cell expansion and hematopoietic regeneration and reconstitution in mouse models. These methods included coculture of the Notch ligand DL1, mesenchymal stromal cells (MSCs), and forced expression of HOXB4. Watts et al. [48] expanded HOXB4-transduced autologous nonhuman Figure 6. Differential gene expression in CD34 + cells unexpanded or expanded with cytokines or expanded with cytokines and ECs. Hierarchical clustering and transcriptional analysis of EC-expanded CD34 + cells and cytokine-expanded CD34 + cells from three donors (A11224, R11145, Z13018) and from unexpanded CD34 + cells (two donors). Heat maps display the relative abundance of genes that were differentially expressed across the three cell populations (unexpanded CD34 + cells, cytokine-expanded CD34 + cells, and EC-expanded CD34 + cells). EC-expanded CD34 + cells were EC depleted and enriched for CD34 + cells before analysis. All other subsets were sorted according to expression of CD34 before analysis. Scaling is relative to the maximum FPKM for each gene across all samples. primate CB CD34 + cells, which had superior engraftment compared with unexpanded CB CD34 + cells. A follow-up study to investigate whether activation of Notch signaling (through coculture with MSCs expressing Notch ligand DL) synergized with HOXB4 overexpression to further improve CB expansion and engraftment was tested in the autologous nonhuman primate setting [49]. That study showed that this combined expansion strategy accelerated neutrophil and platelet recovery compared with unexpanded and HOXB4-expanded HSPC recipient historic controls. Although these methods, and others (PGE 2 , UM171, SR1), expanded CB CD34 + cells, no studies have been reported on the expansion of steady-state BM CD34 + cells, until now.
Our findings showed that ECs support expansion of BM-HSPCs (supplemental online Table 1), thus marking an important advance for HSPC gene therapy for patients who cannot tolerate mobilization but who would otherwise be candidates for gene therapy (e.g., sickle cell disease, myelofibrosis). Without mobilization, the yield from BM harvest results in a lower CD34 + cell dose. Because the success of CD34 + cell reconstitution is proportional to the cell dose, expansion of BM CD34 + cells would likely improve gene-corrected HSPC reconstitution. ECs could thus be used to expand steady-state BM for such patients.
In contrast to these expansion platforms, we provide the first evidence of a CB expansion strategy [16] that effectively translates to expansion of adult HSPCs to support hematopoietic reconstitution and long-term engraftment of steady-state adult BM CD34 + cells. We have shown for the first time that our ex vivo vascular niche expands steady-state BM CD34 + cells while maintaining their stem cell properties (i.e., self-renewal, multipotency, engraftment potential), leading to sustained long-term engraftment (800 days), and enhances lymphoid and erythroid gene-modified cell reconstitution. We have also shown that EC-mediated expansion of HSPCs is safe, as the animals did not exhibit signs of infusional toxicity or any other health problems, such as unexpected thromboembolic or hemorrhagic complications that could be attributed to cotransplantation of ECs. As such, our studies set the stage for translation of endothelial cell-mediated expansion of steadystate BM CD34 + cells.
For the recipients of EC-expanded CD34 + cells, hematopoietic recovery and engraftment occurred within the time frame observed in clinical BM HSPC transplantation (ANC .500 by days [14][15][16][17][18][19][20]. Hematopoietic recovery and engraftment was achieved earlier compared with a previous study in which baboons were transplanted with similar BM CD34 + cell doses after expansion on PMVECs (ANC .500 by day 25 6 1) [37]. The CD34 + cell dose per kilogram and the frequency of mixed hematopoietic colonyforming units (CFU-GM, CFU-GEMM) have been shown to correlate with improved long-term engraftment in patients [1,[25][26][27]. In our studies, the EC coculture supported up to 17-fold expansion of CD34 + CD38 2 cells that substantially increased the CD34 + cell dose per kilogram with a concomitant increase in CFU-GEMM. The expansion of CD34 + cells and CFCs achieved in the present study was substantially higher than the levels achieved in earlier reports evaluating EC-plus cytokine-expanded CD34 + cells in baboons [37,38]. However, similar to those previous reports, expansion of macaque CD34 + cells was lower compared with human CD34 + cell expansion with human cytokines and ECs [16], which can be attributed to the use of human ECs and human cytokines, as nonhuman primate reagents were not available. In addition to the increased cell dose and positive safety profile associated with vascular niche-expanded marrow CD34 + cells, we also achieved unprecedented levels of gene marking in CD3 + CD4 + T lymphocytes, erythrocytes, and platelets without the requirement for in vivo selection.
Our previous studies in the nonhuman primate indicated that myeloid cells typically have the highest efficiency of gene marking in vivo after HSPC transplantation. In those studies, achieving high levels of transgene expression in lymphoid, erythroid, and megakaryocyte progeny was more challenging despite polyclonal repopulation [17,18]. In addition, gene marking in the lymphocyte pool was predominantly in B cells, with ,20% in GFP + CD20 + B lymphocytes [17]. In contrast, marking in CD3 + T lymphocytes was ,5% [50] without in vivo selection. In the study by Trobridge et al. [50], in vivo selection with O 6 -benzylguanine combined with alkylating chemotherapy (temozolomide) increased marking in CD3 + cells to 7%. In contrast, here we show that reconstitution of nonhuman primates with endothelial cell-expanded BM CD34 + cells led to an unprecedented 30% gene marking in CD3 + CD4 + lymphocytes (range, 10%-30%) without in vivo selection. This high level marking in gene-modified T cells would likely improve the efficacy of HSPC and gene therapies for the treatment of HIV and genetic immunodeficiencies.
The level of marking achieved in erythrocytes and platelets after transplantation of endothelial cell-expanded HSPCs is 2-to 10-fold greater than the level previously reported by our group and others for erythroid-specific marking after transplantation with lentivirus transduced mobilized HSPC transplantation [51,52]. Therefore, the gene marking observed in the erythroid and megakaryocyte progeny of endothelial cell-expanded CD34 + cell progeny in vivo is unprecedented, as neither mobilization before transplantation nor in vivo selection with chemotherapy after transplantation were required to achieve these milestones in our study.
In summary, vascular niche-mediated expansion of steadystate BM CD34 + cells has the potential to increase the accessibility and improve the outcome of success of HSPC gene therapy for patient populations with varied hematopoietic disease indications, from frequent immunodeficiencies (infectious or genetic), hemoglobinopathies (sickle cell disease) and anemias (Fanconi anemia) to rare orphan diseases.

CONCLUSION
An ex vivo vascular niche sustains and/or expands bone marrowderived LT-HSPCs that retain multilineage blood reconstitution of the clinically relevant nonhuman primate. These findings demonstrate safety and effectiveness of a bone marrow HSPC expansion platform that could be clinically translated to HSPC gene therapy applications.