Research ArticleStem Cells

A distinct hematopoietic stem cell population for rapid multilineage engraftment in nonhuman primates

See allHide authors and affiliations

Science Translational Medicine  01 Nov 2017:
Vol. 9, Issue 414, eaan1145
DOI: 10.1126/scitranslmed.aan1145

Refining the gold standard

CD34-positive hematopoietic cells are the gold standard for stem cell therapy and transplantation of stem cell–enriched grafts. However, most of the cells within this population will not contribute to engraftment. Using a robust nonhuman primate transplantation model, Radtke et al. identified a stem cell–enriched subpopulation of CD34-positive cells that was exclusively responsible for multilineage engraftment. The cell dose of this subpopulation correlated with neutrophil and platelet recovery and reliably predicted overall transplant success. The authors observed phenotypic and transcriptomic similarities between these cells and human hematopoietic cells with high engraftment and repopulating potential. These data suggest a refined subpopulation of CD34-positive cells for use in transplantation and gene therapy/editing approaches.


Hematopoietic reconstitution after bone marrow transplantation is thought to be driven by committed multipotent progenitor cells followed by long-term engrafting hematopoietic stem cells (HSCs). We observed a population of early-engrafting cells displaying HSC-like behavior, which persisted long-term in vivo in an autologous myeloablative transplant model in nonhuman primates. To identify this population, we characterized the phenotype and function of defined nonhuman primate hematopoietic stem and progenitor cell (HSPC) subsets and compared these to human HSPCs. We demonstrated that the CD34+CD45RACD90+ cell phenotype is highly enriched for HSCs. This population fully supported rapid short-term recovery and robust multilineage hematopoiesis in the nonhuman primate transplant model and quantitatively predicted transplant success and time to neutrophil and platelet recovery. Application of this cell population has potential in the setting of HSC transplantation and gene therapy/editing approaches.


Stringent confirmation of a hematopoietic stem cell (HSC) phenotype can only be achieved after robust long-term multilineage reconstitution of an irradiated recipient. Of the various models developed, retrovirus transduction and subsequent transplantation of CD34+ cells into human or animal recipients has permitted tracking of tens of thousands of individual cells in vivo over time (1).

Most of these studies apply provirus insertion site analysis, which assigns the genomic locus of retrovirus integration as the clone signature. These reports generally identify successive waves of hematopoietic reconstitution after transplant wherein committed hematopoietic progenitor cells provide short-term recovery and HSCs provide long-term blood cell production, with waves of clones emerging and succeeding one another over the lifetime of the recipient (27). However, differences in experimental design, species studied, and the HSC definition applied have resulted in conflicting models.

Large animals such as the pig-tailed macaque (Macaca nemestrina) and the rhesus macaque (Macaca mulatta) share a close evolutionary relationship with humans and have been used as preclinical model systems to study basic HSC biology (8) or to develop specific HSC gene therapy approaches (9). A critical advantage of these models is the ability to support both short- and long-term hematopoiesis, as well as multilineage reconstitution in an autologous, myeloablative transplant setting.

We previously used the pig-tailed macaque as a stem cell transplant model (8, 9) to study long-term hematopoietic reconstitution after transplantation of autologous lentivirus gene–modified CD34+ cells in the myeloablative setting (1016). Here, we wanted to examine HSC engraftment kinetics by applying a stringent definition for HSC behavior in vivo in the pig-tailed macaque transplantation model. We retrospectively followed hundreds of thousands of hematopoietic clones in five pig-tailed macaques by insertion site analysis for up to 7 years after transplantation to identify HSC behavior. We then coupled these studies with prospective, competitive transplantation and additional retrospective analyses to identify a CD34+ subpopulation capable of immediate, long-term, and stable multilineage hematopoietic reconstitution in our nonhuman primate model.


Persistence of early-engrafting clones up to 7.5 years after transplant

We first established the time to full, multilineage hematopoietic reconstitution in the pig-tailed macaque model by discrete analysis of hematopoietic subsets in five retrospectively transplanted animals (fig. S1). We defined the day of engraftment for each lineage as the first day ≥1% fluorophore+ cells were observed with consecutive increases over the next three measurements. On the basis of these data, we established the following time frames: “initial engraftment,” 0 to 3 months; “stabilization,” 3 months to 1 year; “homeostasis,” >1 year; and “long-term” engraftment, >2 years after transplant.

To study the kinetics of individual blood cell clones in autologous CD34+ hematopoietic stem and progenitor cell (HSPC) grafts, we retrospectively analyzed the engraftment of individual lentivirus gene–modified cell clones by high-throughput retrovirus insertion site analysis (Fig. 1A). We followed 134,528 clones (range, 6830 to 43,558 unique clones per animal, n = 5) for 2 to 7.5 years after transplant (table S1). In all five animals, clones engrafting in the first 0 to 3 months contributed more than half of the detected, gene-modified blood cells stably for >2 years after transplant (Fig. 1A).

Fig. 1. Persistence of early-engrafting stem cell–like clones after transplant in nonhuman primates.

(A) In vivo hematopoietic cell clone tracking data for five retrospectively analyzed pig-tailed macaques followed for more than 2 years after myeloablative transplant of lentivirus gene–modified autologous CD34+ cells. Bars represent the contribution of clones identified in peripheral blood <3 months after transplant (first bar in each graph) over time. The number above each bar is the total number of clones identified in the time period; n, number of individual blood samples analyzed during that time period. NA, no samples available within this time period. (B) Longitudinal contributions of abundant HSC clones (maximum clone contribution > mean maximum contribution for all HSC clones identified in the animal). Each colored band represents one HSC clone. Bandwidth corresponds to the clone frequency (y axis). To be called an HSC clone, the clone signature (lentivirus insertion locus) had to be present in one short-lived and one long-lived mature blood cell lineage and present at more than a single time period of analysis, one of which had to be later than 1 year after transplant.

Early engraftment of persistent HSC clones

To deduce whether any of these early-engrafting and persistent clones displayed HSC biology in vivo, the specific clone signature (that is, genomic locus of lentivirus integration) had to be detected in multiple mature cell lineages ≥1 year after transplant and detected at additional time points. We analyzed clone signatures shared between short-lived granulocytes or monocytes and longer-lived B cells or T cells at time points ≥6 months after transplant in three of five animals (table S1). We observed 175, 291, and 740 shared HSC clone signatures, representing 0.4, 0.9, and 2.1% of all clones detected in these animals, respectively (table S1). We then tracked the most abundant of these HSC clone signatures over time (Fig. 1B). Clones were characterized as “high-abundance” HSC clones if their maximum contribution over time was greater than the mean maximum contribution for all HSC clones identified in the animal. We observed abundant HSC clones within the first 6 months after transplant, which were stably maintained throughout follow-up. These data demonstrated a subpopulation of CD34+ cells with multilineage capacity that engrafted very early after autologous transplantation and was stably maintained for years in the recipient nonhuman primates. We then sought to identify such a population for targeted cellular therapy applications.

Identification of an HSC-enriched subpopulation in the nonhuman primate CD34+ cell population

Self-renewal and multilineage potential are classic characteristics of HSCs. To determine which subpopulation of CD34+ cells could be HSCs, we prospectively tested the cross-reactivity of human HSPC markers with pig-tailed macaque and rhesus macaque stem cells (table S2). Focusing on the cross-reactive markers CD34 (17, 18), CD45RA (19), CD90 (20), CD117 (21), and CD123 (22), we identified nine (I to IX) phenotypically distinct HSPC subpopulations in unmanipulated, steady-state bone marrow (ssBM), growth factor–primed bone marrow, and umbilical cord blood (Fig. 2, A and B, and fig. S2, A to D). The average frequency of CD34+ cells in ssBM was 5.8 ± 2.9% in pig-tailed macaques and 2.2 ± 0.7% in rhesus macaques, with equal proportions of the identified subpopulations (fig. S2, E to G). Very low numbers of CD34+ cells were found in pig-tailed macaque umbilical cord blood (0.13 ± 0.04%), but increased proportions of populations I, IV, and VII were observed (fig. S2H).

Fig. 2. Identification of phenotypically defined ssBM-derived CD34+ subpopulations in the pig-tailed macaque.

(A) Cell surface expression of CD45 versus CD34, CD45RA versus CD117, and CD90 versus CD123 on pig-tailed macaque ssBM–derived white blood cells. CD45RA/CD117 expression is gated on CD34highCD45int (I) populations; CD90/CD123 analysis is shown for CD34highCD45intCD45RACD117+ (IV) populations. FSC, forward scatter; SSC, side scatter. (B) Phenotypically distinct subpopulations were labeled I to IX (int, intermediate staining). Corresponding data can be found in figs. S2 and S3. (C) Summary of the in vitro functional properties observed in each subpopulation. (n.d., not determined). Corresponding data can be found in figs. S3 to S9. CAFC, cobblestone area–forming cells; CFC, colony-forming cells.

To evaluate self-renewal capacity, we examined the proliferation potential of defined HSPC subpopulations after flow-based sorting. All subpopulations lost CD34, CD90, and CD117 marker expression and gained CD45RA marker expression when cultured (fig. S3A). Population V demonstrated the greatest proliferation potential (fold expansion ≫1), followed by populations I and IV (fold expansion >1). In contrast, populations VII and VIII demonstrated no change in cell number (fold expansion ±1), and populations II, III, VI, and IX demonstrated cell loss after culture (fold expansion <1) (Fig. 2C and fig. S3B).

We next wanted to identify which subpopulations contained HSC and multipotent progenitor cell activity and applied functional in vitro assays for differentiation of T cells (fig. S4), natural killer (NK) cells (fig. S5), granulocytes (figs. S5, S6, and S9), monocytes/macrophages (figs. S5, S6, and S9), erythrocytes (figs. S6 and S9), and megakaryocytes (fig. S7). Populations I, IV, VII, and VIII were capable of generating all of the tested lineages (Fig. 2C). Given that populations VII and VIII were more discrete subdivisions of populations I and IV, these cells were considered to be the more refined HSC/multipotent progenitor cell–containing subpopulations.

The formation of colonies containing cobblestone area–forming cells and the ability to form secondary colonies were also suggestive of primitive, multipotent cell types (23, 24). In the MS-5 stromal cell line coculture assay, we observed the greatest number of cells with cobblestone area–forming cell potential within population VII (fig. S8 and table S3). In standard colony-forming cell assays, autonomous cobblestone area–forming cells from populations VII and VIII demonstrated secondary myeloid and erythroid colony-forming cell potential (fig. S9, A to C). Overall, secondary colony-forming cell potential was most highly enriched in population VII across all species and stem cell populations from different source tissues (fig. S9, D to H).

Together, these data indicate that the most primitive nonhuman primate HSPCs were enriched in CD34+CD117+CD45RACD90+ cells (population VII), which gave rise to multipotent CD34+CD117+CD45RACD90 HSPCs (population VIII), whereas lympho-myeloid–restricted progenitor cells were enriched in population V. Given these in vitro observations, we hypothesized that these cell fractions could contribute to the early and sustained engraftment in vivo observed in animals analyzed retrospectively.

Complete multilineage engraftment from CD34+CD45RACD90+ nonhuman primate cells

To study the in vivo biology of identified HSPC subsets, we performed prospective competitive reconstitution experiments in the myeloablative, autologous nonhuman primate stem cell transplant model (Fig. 3A). Enriched CD34+ cells from an additional four animals were sorted into three distinct fractions based on CD45RA and CD90 expression: fraction i (CD34+CD45RACD90+), fraction ii (CD34+CD45RACD90), and fraction iii (CD34+CD45RA+CD90) (Fig. 3B and table S4). Together, these three fractions represented all CD34highCD45int cell phenotypes (populations IV to IX). Purified fractions were then transduced with lentivirus vectors encoding green fluorescent protein (GFP), mCherry or mCerulean fluorescent proteins to independently track in vivo performance of each cell population after transplant (Fig. 3C and fig. S10A). To ensure that expression of a specific fluorophore did not influence in vivo observations, fluorophores assigned to each transplanted cell population were varied across animals (Fig. 3C and table S4).

Fig. 3. Multilineage engraftment is exclusively driven by CD34+CD45RACD90+ cell fractions.

(A) Outline for autologous nonhuman primate stem cell transplantation experiments including time points for peripheral blood and bone marrow analysis. PB, peripheral blood; BM, bone marrow; G-CSF, granulocyte colony-stimulating factor; SCF, stem cell factor; MOI, multiplicity of infection; TBI, total body irradiation; QC, quality control. (B) Bone marrow–derived nonhuman primate CD34+ HSPCs were separated into fractions i, ii, and iii by fluorescence-activated cell sorting (FACS) based on the expression of CD45RA and CD90. These fractions contained all CD34highCD45int subpopulations as follows: fraction i (population VII), fraction ii (populations IV, VIII, and IX), and fraction iii (populations V and VI). The dashed box highlights presumed multipotent cell phenotypes expected to contribute to engraftment in vivo. (C) A total of four animals were used for this study. FACS-purified HSPC fractions were transduced in different combinations with lentivirus encoding GFP (green), mCherry (red), or mCerulean (blue), as indicated. (D) Long-term follow-up of gene marking in nonhuman primate white blood cells. CMV, cytomegalovirus. (E) Frequency of gene-marked granulocytes (CD11b+CD14+), monocytes (CD11b+CD14+), B cells (CD20+), T cells (CD3+), and NK cells (CD16+) in the peripheral blood of nonhuman primates.

Purified, transduced cell fractions from all four nonhuman primates maintained fraction-specific marker expression (fig. S10A) and colony-forming cell potential (fig. S10B) after culture. Total cell doses ranged from 0.8 × 106 cells/kg to 1.7 × 106 cells/kg at transplant (Fig. 3C). Despite these low cell doses, peripheral blood absolute neutrophil counts and platelet counts rapidly rebounded in all animals within 9 to 10 and 10 to 20 days, respectively (figs. S11 and S12). Surprisingly, peripheral blood white blood cells in all animals exclusively expressed fluorophore proteins assigned to fraction i (CD34+CD45RACD90+ cells) (Fig. 3D). Detailed analysis of peripheral blood (fig. S10C) showed stable gene marking in all cell lineages (Fig. 3E). The frequency of fluorophore-expressing T cells remained low for the first 3 months and reached stable engraftment 6 months after transplant (Fig. 3E).

All animals demonstrated robust engraftment of both unmodified (fluorophore) and gene-modified (fluorophore+) white blood cells and HSPCs in the bone marrow at 5 to 8 weeks after transplant (fig. S13A). The bone marrow reconstitution was also exclusively driven by fraction i cells, which were initially skewed toward lympho-myeloid progenitor cells, but then established a more balanced composition over time (fig. S13B). Bone marrow was harvested, and both gene-modified and non–gene-modified HSPC subpopulations were sorted to determine whether gene-modified cells displayed similar colony-forming cell potential relative to nontransduced repopulating cells. No significant differences were observed between fluorochrome+ and fluorochrome HSPCs in colony-forming cell assays (fig. S13C). These observations suggested that CD34+CD45RACD90+ (fraction i) cells were exclusively responsible for multilineage cell engraftment and initial hematopoietic recovery in peripheral blood and bone marrow after autologous, myeloablative transplantation.

Multilineage reconstitution demonstrated by clonal analysis

To test whether multipotent HSCs were represented by this early engraftment of fraction i cells, we purified fluorophore+ peripheral blood cell lineages by cell sorting from two evaluable animals at 4 months (Z13264, mCh+) and 6.5 months (Z14004, GFP+) after transplant for insertion site analysis (Fig. 4). Cell populations sorted included short-lived granulocytes (CD11b+CD14) and monocytes (CD11b+CD14+), as well as long-lived B cells (CD20+) and T cells (CD3+).

Fig. 4. Clone tracking by insertion site analysis confirms early, multilineage hematopoietic engraftment of CD34+CD45RACD90+ cells.

Venn diagrams illustrate the number of shared clone signatures between fluorophore and cell surface marker–sorted peripheral blood cell lineages in two animals at 4 months (Z13264) and 6.5 months (Z14004) after transplant. Fluorophore+ fraction i cells were mCherry+ (Z13264) or GFP+ (Z14004). Subsets include B cells (CD20+), T cells (CD3+), granulocytes (CD11b+CD14), and monocytes (CD11b+CD14+).

Clonal analysis demonstrated highly polyclonal engraftment with 6152 and 18,034 unique clone signatures detected for animals Z13264 and Z14004, respectively (Fig. 4). Of these, 694 (Z13264) and 546 (Z14004) were shared clone signatures between granulocytes and B cells, and 288 (Z13264) and 662 (Z14004) were shared clone signatures between monocytes and T cells, similar to what was analyzed retrospectively in the long-term follow-up animals. In total, 1577 clone signatures (25.6% of total) were shared by two or more lineages in animal Z13264, and 1502 clone signatures (8.3% of total) were shared by two or more lineages in animal Z14004. These data confirmed that fraction i cells conferred multilineage reconstitution in vivo after transplant. Given this observation, we next wanted to determine the range of fraction i cell doses required for hematopoietic recovery after myeloablative transplant in this animal model.

Hematopoietic recovery predicted by CD34+CD45RACD90+ cell dose

To evaluate CD34+CD45RACD90+ cell doses required for successful multilineage engraftment, we retrospectively analyzed flow cytometric data from the infusion products of 11 additional animals (table S5). Initial neutrophil engraftment (absolute neutrophil count, >500/μl) and platelet engraftment (platelets, >20,000/μl) were achieved in all animals (figs. S11 and S12). However, two animals were not able to sustain platelet and neutrophil engraftment without additional supportive care during the first 3 months after transplant (fig. S14). We therefore defined these two animals as “engraftment failures” and the remaining nine animals with long-term engraftment as “engraftment successes.”

Animals displaying engraftment success received about sixfold the mean dose of CD34+CD45RACD90+ cells/kg body weight compared to animals displaying engraftment failure (P = 0.019) (Fig. 5A). No significant differences were observed among animals displaying engraftment success and failure for CD34+, CD34+CD45RACD90, or CD34+CD45RA+CD90 transplanted cells/kg body weight (Fig. 5A and fig. S15A). Notably, we observed a strong correlation between the CD34+CD45RACD90+ cell dose transplanted and the onset of neutrophil and platelet recovery in animals displaying engraftment success (Fig. 5B), whereas total nucleated cells and number of CD34+, CD34+CD45RACD90, CD34+CD45RA+CD90, or colony-forming cells did not correlate with recovery of neutrophils or platelets (fig. S15B). These data suggest that a minimum dose of 122,000 CD34+CD45RACD90+ cells/kg was sufficient to elicit engraftment in this transplant model.

Fig. 5. Correlation between CD34+CD45RACD90+ cell dose, engraftment success, and onset of neutrophil/platelet recovery in nonhuman primates.

(A) Statistical comparison of transplanted CD34+ and CD34+CD45RACD90+ cells/kg body weight in animals with engraftment failure or long-term engraftment (unpaired, two-sided t test). (B) Logarithmic correlation of transplanted CD34+CD45RACD90+ cells/kg body weight with the day of neutrophil and platelet recovery, as calculated using Spearman’s rank correlation coefficient. The linear regression and the 95% confidence interval for each correlation are indicated with solid and dotted lines, respectively.

Enrichment of CD34+CD38low/−cells in human CD34+CD45RACD90+ cell fractions

To evaluate whether the markers CD45RA and CD90 used in the nonhuman primate (Fig. 6A) enriched a similar cell fraction in human CD34+ HSPCs, we tested our strategy on G-CSF–mobilized human peripheral blood stem cells (PBSCs) (Fig. 6B) and human umbilical cord blood (fig. S16). Classical gating strategies including CD38 (20, 25) resulted in a low to moderate enrichment for CD34+CD45RACD90+ cells (Fig. 6B). CD34+CD45RACD90+ fractions were enriched for CD38 and CD38low cells in PBSCs (Fig. 6B). This enrichment of CD38 cells was even greater in CD34+CD45RACD90+ fractions from umbilical cord blood (fig. S16).

Fig. 6. Phenotypic and transcriptomic similarities observed between human and nonhuman primate HSC-enriched cell fractions.

(A) Gating of the HSC-enriched CD34+CD45RACD90+ cell fraction in nonhuman primate ssBM and primed bone marrow. (B) Gating of an HSC-enriched cell fraction in human G-CSF–mobilized PBSCs using classical gating (left) and our new gating strategy (right) for a CD34+CD45RACD90+ subfraction. (C) Comparison of transcript expression in human and nonhuman primate HSC-enriched cell fractions to that of bulk CD34+ cells. The log2 fold change in gene expression was calculated for each species and gene individually. The number of genes included in each quadrant for all genes (gray), differentially expressed genes in human or nonhuman primate (black), and differentially expressed genes in both species (red) with a P value <0.05 is given next to the scatterplot, with genes showing the same pattern (lower left and upper right quadrants), highlighted in bold. Key genes of HSCs and hematopoietic differentiation are tagged and labeled.

Transcript expression of key molecules has been described for human HSCs (26, 27). To determine whether nonhuman primate and human HSC-enriched cells displayed similar transcript expression, we performed RNA sequencing (RNA-seq) analysis. We sorted human (CD34+CD133+CD45RACD90+) and nonhuman primate (CD34+CD117+CD45RACD90+) HSC-enriched cell fractions, isolated and sequenced polyadenylated RNA molecules, and compared these to gold standard bulk CD34+ cells. Key genes reported to be up-regulated in multipotent human HSCs such as ABCA5, HOXB3, MEIS1, BMPR2, MLLT3, SOCS2, and FZD6 (2832) were also found to be significantly (P < 0.05) up-regulated in nonhuman primate CD34+CD117+CD45RACD90+ cells, whereas low expression was observed for the differentiation markers TLR1, IL7R, IRF8, HGF, CEBPA, or MPO (26, 27) in this cell population from both species (Fig. 6C and table S6). These data suggest phenotypic and transcriptomic similarities between human and nonhuman primate HSC-enriched cell populations, which could be biologically discriminated by expression of CD34, CD45RA, and CD90.


Here, we describe a CD45RACD90+ subset of enriched CD34+ hematopoietic cells that is responsible for rapid multilineage hematopoiesis in a clinically relevant, autologous nonhuman primate HSC transplantation model. We demonstrated evidence for early-engrafting HSC-like clones, which were stably maintained long-term. These observations were supported by tracking hundreds of thousands of unique clone signatures in animals for up to 7.5 years of retrospective follow-up. We demonstrated that the phenotype and transcriptome of these cells were highly similar between nonhuman primates and humans. Biologically robust competitive repopulation experiments in four animals demonstrated the capacity of CD34+CD45RACD90+ cells to completely reconstitute hematopoiesis in the autologous, myeloablative transplant setting. We also demonstrated that limited cell doses were needed for successful hematopoietic reconstitution in this setting. The number of nonhuman primates included in the present study, validation in vivo both retrospectively and prospectively, and the tracking of individual clones to confirm these observations suggested early engraftment of HSC clones with stable contributions to hematopoiesis over time after autologous transplantation. For the fields of HSC biology and transplantation, gene therapy, and gene editing, this study identified a highly refined, small cell population capable of complete functional hematopoietic reconstitution.

These findings markedly contrast with a recent report using insertion site analysis alone to describe hematopoietic reconstitution in rhesus macaques over a decade after autologous transplantation of gene-modified CD34+ cells (33). This study identified between 846 and 5758 total clones in each of four rhesus macaques transplanted with either lentivirus- or simian immunodeficiency virus–based vector–transduced cells, which followed a biphasic reconstitution over the first year, and then sequential expansion of HSC-like clones. This study also reported that the number of clones identified in each animal was directly proportional to the CD34+ cell dose at the time of transplantation and that HSC-like clonal signatures were not observed in vivo until 7 to 9 months after transplant, gradually increasing in contribution >1 year after transplant, the time period in our study defined as homeostasis. There are notable differences between these two studies. First, different nonhuman primate species were used for in vivo studies (rhesus macaques versus pig-tailed macaques), as well as different sources of CD34+ cells (G-CSF–mobilized leukapheresis-derived CD34+ cells in rhesus macaques versus G-CSF– and stem cell factor–primed bone marrow–derived CD34+ cells in pig-tailed macaques). The fraction of repopulating HSCs declines in mobilized leukapheresis products compared to bone marrow products, with the highest proportion of HSCs found in umbilical cord blood (34). Thus, the lower number of clones transplanted and tracked in the rhesus macaque study could reflect lower HSC content in mobilized leukapheresis products.

Additionally, there are technical differences between insertion site analysis methods used to track the clones: The rhesus macaque study used a restriction enzyme–based insertion site analysis method different from the acoustic shearing–based method that we used. Our method allowed for quality control of clone abundance by capturing different fragment sizes of the same flanking genomic sequence by single-end sequencing, whereas the rhesus macaque method relied on paired-end sequencing of resulting genomic fragments. The paired-end approach can reduce bias but also displays lower sequencing efficiency, resulting in lower clone numbers. Additionally, the rhesus macaque study collected samples from each animal a total of 7 to 12 times after transplant, over as many as 140 months, with very few samples (≤3 time points) collected within the first year. Thus, the biphasic reconstitution and clonal succession observed in the rhesus macaque study may have resulted from undersampling.

A more comprehensive sampling of early time points within the first 9.5 months after transplant was reported by Wu and colleagues (35) in a side-by-side report to the rhesus macaque study. Here, the same rhesus macaque model was used (n = 3), with unique clones tracked by retrovirus DNA barcode sequencing. Monthly sampling was performed over the early reconstitution and stabilization periods we describe. Similar to our observations, cumulative clone numbers detected did not correlate with the CD34+ cell dose at time of transplant over the brief period of follow-up included in this report. This report identified multilineage clones over the first 9.5 months after transplant, beginning 2 to 3 months after infusion. Longer follow-up and additional animals will be required to demonstrate long-term persistence of these clones in the rhesus macaque model. Additionally, the DNA barcode sequencing study reported a unique ontogeny of NK cells. This analysis was not performed in our retrospective study because of lack of sorted NK cell samples. However, we observed CD34+CD45RACD90+ cells contributing to the NK compartment in competitive repopulation experiments. Additional clone tracking studies will be required to confirm clonal origins of the NK lineage in our model.

Our most striking observation was that the highly HSC-enriched CD34+CD45RACD90+ fraction accounted for all gene-modified cells observed in peripheral blood and bone marrow after transplantation in this model. This suggested that everything required for short- and long-term, multilineage hematopoietic reconstitution was contained within this cell phenotype. We observed clone signatures, which were both shared and exclusive to sorted blood cell lineages in two of the four animals studied. We also observed up to 25% of these clones shared by two or more blood cell lineages, which is a marked increase relative to animals in the retrospectively analyzed cohort. However, these studies sampled very disparate time points of follow-up in these two cohorts. Additionally, analysis in competitively repopulated animals consists of a single time point, compared to multiple time points of analysis in the retrospective cohort.

In vivo observations in competitively transplanted animals suggest that the CD34+CD45RACD90+ population could be a significant refinement in the target cell population for transplantation in the autologous setting, as well as in approaches such as gene therapy and gene editing. Currently, the gold standard for gene therapy is lentivirus transduction of enriched CD34+ cells. Focusing on CD34+CD45RACD90+ cells would result in a 20-fold reduction in overall cell numbers and thus markedly reduce materials required for vector-based or nuclease-mediated gene therapy and editing protocols. Ongoing studies are currently also evaluating whether this enrichment strategy can result in more efficient gene modification. When comparing human and nonhuman primate CD34+CD45RACD90+ HSPCs, we found significant overlap in gene expression of key stem cell factors and low expression of differentiation factors. Given the cross-species conservation of the HSC-enriched CD34+CD45RACD90+ phenotype, we would expect important implications for human transplantation and gene therapy studies.

We observed a strong correlation between CD34+CD45RACD90+ HSPCs/kg body weight and the success of multilineage engraftment, as well as the time to neutrophil and platelet recovery, further supporting the biological importance of our phenotypically defined, HSC-enriched population. Our data demonstrate a direct correlation of a phenotypically defined HSPC subpopulation with engraftment properties after autologous transplant. We also found that the correlation between CD34+CD45RACD90+ cell dose and engraftment was not affected by different stem cell sources, gene modifications, expansion protocols, or cryopreservation, making this phenotype a valuable tool for the in vitro analysis of HSC modification, expansion strategies, and quality control of stem cell infusion products.

The follow-up of animals prospectively transplanted with sort-purified and fluorophore-coded subpopulations in this study is currently limited to 1 year and remains ongoing. Whereas this length of follow-up has been predictive of long-term outcome (>7 years) in our retrospective nonhuman primate studies, longer follow-up will be required to fully assess the safety and feasibility of this approach for clinical stem cell transplantation and gene therapy. Also, our evaluation of this new phenotype in human HSC sources is currently limited to phenotypic and transcriptomic analyses. Further comprehensive in vivo studies of these human HSPC subpopulations should be performed in either prospective mouse xenograft assays or retrospectively analyzed in patients if the required marker expression data and posttransplant follow-up data are available. Finally, this study did not address whether this refined HSC-enriched phenotype requires the cotransplantation of other supporting cell phenotypes.

In conclusion, retrospective detection of persistent and HSC-like clonal signatures early after transplant, before all lineages have reconstituted, suggests a subpopulation of transplanted CD34+ hematopoietic cells that immediately contributes to long-term, multilineage hematopoietic reconstitution lasting up to 7.5 years in this clinically relevant model. Most notably, using prospective transplantation studies, we have identified a subset of CD34+ cells expressing CD90 and lacking CD45RA markers, which rapidly and exclusively reconstitutes multilineage hematopoiesis at low cell doses in the autologous, myeloablative transplant setting in nonhuman primates. This cell population is similar between humans and nonhuman primates on the basis of phenotypic and transcriptomic analyses. Moreover, enumeration in the autologous setting is rapid and predictive of engraftment success and extent of supportive care required after transplant. These findings may affect HSC transplantation and the development of gene therapy and gene editing approaches.


Study design

Retrospective insertion site analysis after autologous, myeloablative transplantation of gene-modified, CD34+ HSPCs in the nonhuman primate model suggested early, multilineage contribution of long-term engrafting blood cells. Based on these findings, the main goal of this study was to identify this subpopulation and evaluate its engraftment and differentiation potential. Combining HSPC cell surface marker expression with functional in vitro readouts, distinct HSPC subpopulations were identified in pig-tailed macaque and rhesus macaque stem cell populations from different tissue sources. Contributions of these candidate populations to short- and long-term hematopoietic recovery were tested in prospective competitive reconstitution experiments in four additional pig-tailed macaques. We longitudinally tracked the engraftment of gene-modified cells in the peripheral blood and bone marrow of pig-tailed macaques by flow cytometry, functional in vitro assays, and insertion site analysis for 1 year after transplant. Correlation of HSC-enriched target cell populations with engraftment success and hematologic recovery kinetics was established with an additional 11 retrospectively transplanted animals. Finally, relevance to human hematopoiesis was studied by flow cytometry and RNA-seq, demonstrating similar cell surface marker phenotypes and transcriptomic profiles between these two species.

Cell sources, CD34+ enrichment, and in vitro culture

Pig-tailed macaque ssBM, primed bone marrow, umbilical cord blood, and rhesus macaque ssBM, as well as human CD34+ cells were harvested, enriched, and cultured as previously described (11, 36). For more details on the CD34+ enrichment, see Supplementary Methods. Enriched CD34+ cells were cultured in StemSpan (STEMCELL Technologies) supplemented with penicillin-streptomycin (100 U/ml) (Gibco by Life Technologies) and either stem cell factor (PeproTech), TPO (thrombopoietin; PeproTech), and FLT3-L (Fms-related tyrosine kinase 3 ligand; Miltenyi Biotec), or stem cell factor and IL-3 (interleukin-3; PeproTech) (100 ng/ml each).

FACS analysis and cell sorting

Antibodies used for FACS analysis and sorting of pig-tailed macaque, rhesus macaque, and human cells are listed in table S1. Dead cells and debris were excluded via forward scatter/side scatter gating. 123count eBeads (eBioscience) were used for quantification purposes. Flow cytometric analysis was performed on an LSR II (BD) and FACSAria II (BD). Cells for in vitro assays as well as autologous nonhuman primate stem cell transplants were sorted using a FACSAria II cell sorter (BD), and sort purity was assessed by recovery of sorted cells.

Functional in vitro readouts

Analysis of sort-purified pig-tailed macaque and rhesus macaque cells in functional in vitro readouts for the evaluation and quantification of erythroid, myeloid (granulocyte, monocyte/macrophage), lymphoid (T, B, and NK cells), and megakaryocytic differentiation potentials was performed as previously described (see Supplementary Methods) (3739).

Autologous nonhuman primate transplants and ex vivo engineering of HSPCs

Autologous nonhuman primate transplants, priming (mobilization), collection of cells, and genetic engineering were conducted consistent with our previously published protocols (11). All experimental procedures performed were reviewed and approved by the Institutional Animal Care and Use Committee of the Fred Hutchinson Cancer Research Center and University of Washington (protocol #3235-01). This study was carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health (“The Guide”), and monkeys were randomly assigned to the study. Detailed protocols for the performed nonhuman primate transplants including (i) ex vivo engineering of HSPCs, (ii) animal housing and care, and (iii) preconditioning and supportive care can be found in Supplementary Methods.

RNA isolation, RNA quality control, RNA-seq, RNA-seq expression, and data analysis

Isolation of total RNA from human and nonhuman primate bulk CD34+ cells and sort-purified HSC-enriched cell fractions was performed according to the manufacturer’s instructions (see Supplementary Methods). Total RNA integrity was analyzed using an Agilent 2200 TapeStation (Agilent Technologies Inc.) and quantified using a Trinean DropSense96 spectrophotometer (Caliper Life Sciences). RNA-seq expression analysis was performed in Shared Resources at the Fred Hutchinson Cancer Research Center. Detailed protocols for data analyses can be found in Supplementary Methods. All custom R and Python codes are available on request. All other codes are publicly available and cited in the appropriate method description (see Supplementary Methods).

Insertion site analysis

Detailed protocols for the insertion site analysis, persistence analysis, and determination of HSC signatures can be found in Supplementary Methods. All custom R and Python codes are available on request. All other codes are publicly available and cited in the appropriate method description (see Supplementary Methods).


Statistical analysis of data generated from all experiments except RNA-seq analysis was performed using GraphPad Prism version 5. These statistical data are presented as means ± SEM. Significance analyses were performed with the paired/unpaired, two-sided Student’s t test (*P < 0.05; **P < 0.01; ***P < 0.001). Correlations were calculated using Spearman’s rank correlation coefficient (correlation coefficient R2 values of 0.0 to 0.19 = very weak, 0.20 to 0.39 = weak, 0.4 to 0.59 = moderate, 0.6 to 0.79 = strong, and 0.8 to 1.0 = very strong). Statistical analysis of RNA-seq data was performed using DESeq2.



Fig. S1. Engraftment kinetics in transplanted nonhuman primates.

Fig. S2. HSPC composition in nonhuman primate stem cell populations from different sources.

Fig. S3. Ex vivo culture of nonhuman primate HSPCs.

Fig. S4. T cell potential of nonhuman primate HSPCs.

Fig. S5. Lympho-myeloid potential of nonhuman primate HSPCs.

Fig. S6. Colony-forming cell potential of nonhuman primate HSPCs.

Fig. S7. Megakaryocyte potential of nonhuman primate HSPCs.

Fig. S8. Cobblestone area–forming cell potential of nonhuman primate HSPCs.

Fig. S9. Secondary colony-forming cell potential of nonhuman primate HSPCs.

Fig. S10. Quality control for competitive repopulation experiments.

Fig. S11. Neutrophil recovery in transplanted nonhuman primates.

Fig. S12. Platelet recovery in transplanted nonhuman primates.

Fig. S13. Bone marrow engraftment of gene-modified HSPCs.

Fig. S14. Long-term engraftment versus failure.

Fig. S15. Correlation of recovery with infused HSPCs and colony-forming cells.

Fig. S16. Gating of HSC-enriched cell fractions in human umbilical cord blood.

Table S1. Identification of clones displaying HSC biology in vivo.

Table S2. Cross-species reactive antibodies in nonhuman primates.

Table S3. MS-5 assay colonies.

Table S4. Characteristics of nonhuman primate transplants.

Table S5. Characteristics of nonhuman primate transplants II.

Table S6. Differentially expressed genes in human and nonhuman primate HSCs.

References (4054)


Acknowledgments: Human PBSCs were provided by S. Heimfeld. We thank H. Crawford for help in preparing the manuscript. We also thank V. Nelson, E. Curry, and K. Sze for excellent support in our pig-tailed macaque studies. Funding: This work was supported, in part, by grants to H.-P.K. from the NIH (HL115128, HL098489, and AI096111) and by funds to J.E.A. from the Fred Hutchinson Cancer Research Center and the Cuyamaca Foundation. H.-P.K. is a Markey Molecular Medicine Investigator and received support as the inaugural recipient of the José Carreras/E. Donnall Thomas Endowed Chair for Cancer Research and the Fred Hutchinson Cancer Research Center Endowed Chair for Cell and Gene Therapy. Author contributions: S.R., J.E.A., and H.-P.K. designed the study. S.R., M.A.G., and Y.-Y.C. performed in vitro experiments, competitive transplants, and data analysis. S.R. and Y.-Y.C. generated RNA-seq data. S.R., J.E.A., Z.K.N., M.E., and L.E.S. performed RNA-seq sample analyses. J.E.A., K.G.H., Z.K.N., and L.E.S. generated and analyzed insertion site analysis data. Z.K.N. generated custom R and Python scripts. S.R., J.E.A., Z.K.N., M.E., and L.E.S. generated the figures. H.-P.K. and J.E.A. funded the study. S.R., J.E.A., and H.-P.K. wrote the manuscript. All authors reviewed and edited the final manuscript. Competing interests: S.R., J.E.A., and H.-P.K. are co-inventors on U.S. Provisional Patent Applications #62/351,761 and #62/428,994 “Selection and use of cell populations for research and therapeutic purposes” and the International Patent Application #PCT/US2017/037967 “Strategies to assess and/or produce cell populations with predictive engraftment potential.” H.-P.K. and J.E.A. are consultants for Rocket Pharmaceuticals. The other authors declare that they have no competing interests. Data and materials availability: All original RNA-seq data were uploaded to the National Center for Biotechnology Information database [BioProject accession codes PRJNA320857 (nonhuman primate) and PRJNA320858 (human)].

Stay Connected to Science Translational Medicine

Navigate This Article