Attenuation of miR-126 Activity Expands HSC In Vivo without Exhaustion

miR-126 Levels Peak in Human HSC

Within the human hematopoietic hierarchy, we previously demonstrated peak miR-126 expression and bioactivity within the CD34⁺CD38⁻CD90⁺ primitive compartment (Gentner et al., 2010). However, the human progenitor hierarchy model was recently fractionated into more precise subpopulations (Figure 1A) (Doulatov et al., 2010; Notta et al., 2011). We sorted seven functionally characterized populations from human lineage-depleted cord blood hematopoietic stem and progenitor cells (lin− CB HSPC) to obtain a detailed expression profile of miR-126. The highest expression levels of miR-126 were restricted to the HSC-enriched CD34⁺CD38⁻CD45RA⁻CD49f⁺ fraction (Figure 1B). miR-126 levels remained high in CD34⁺CD38⁻CD45RA⁻CD90⁻CD49f⁻ multipotent progenitors (MPPs) and became significantly downregulated in multilymphoid progenitors (MLPs) and committed CD34⁺CD38⁺ fractions. These data demonstrated a strong correlation between the HSPC hierarchy and miR-126 expression level, suggesting a role for miR-126 in regulating stem cell function.

miR-126 Attenuates Proliferation of Early Progenitors

To interrogate the function of miR-126 in primitive human blood cells, we performed miR-126 overexpression (OE) and knockdown (KD) in primary CB HSPC by using lentiviral vectors (Figure 1C). The design and validation of these vector tools is described in Supplemental Results and Figure S1 (available online). The degree of miR-126 up- or downmodulation that we obtained was functionally relevant as the expression levels of the previously validated miR-126 target PIK3R2 were specifically and significantly decreased upon OE and increased upon KD, respectively (Figure 1D). We then assessed whether modulating miR-126 could influence the proliferation and differentiation of HSPC in vitro. Cytokine-driven serum-free long-term culture of CD34⁺CD38⁻ cells revealed a trend toward decreased total cell output upon 126/OE and a significantly increased expansion upon 126/KD without disturbance of myeloid differentiation into CD15⁺ granulocytes and CD14⁺ monocytes (Figure 1E and data not shown). To examine whether this response was associated with a specific CB subpopulation, we generated growth curves from CD34⁺ subpopulations that were transduced with the 126/KD or a scrambled control vector (Figure 2A). Interestingly, the growth of primitive CD34⁺CD38⁻CD133⁺ cells was significantly enhanced upon 126/KD, in contrast to the more committed CD34⁺CD38⁺CD133⁺ subset. Moreover, the growth advantage was most pronounced in the first 5 days of culture when the cells are the least differentiated, suggesting that the 126/KD phenotype may specifically affect stem and early progenitor cells, reflecting the physiological expression pattern of miR-126. To assess whether the cycling status of HSPC was altered upon 126/KD, cell-cycle analysis was performed 3 to 4 days after transduction. 126/KD induced a 1.7-fold increase in the proportion of cells in the S/G₂/M phases of the cell cycle as compared to nontransduced or CTRL vector-transduced cells (Figure 2B). Taken together, these data indicate that 126/KD increases the proliferation of primitive HSPC.

We then investigated whether 126/KD affected the proliferation and/or differentiation of functionally defined myeloid and erythroid progenitors in clonogenic assays. When CD34⁺ CB cells were plated 1 day after transduction, no quantitative differences in colony number or type were observed between 126/KD or control groups (Figure S2A). However, when the cells were cultured for 7 days before performing the clonogenic assay, 126/KD CD34⁺ CB yielded significantly more myeloid colonies than control vector-transduced cells (Figure 2C). When 126/KD CB were further enriched for primitive HSPC and expanded for 1 to 2 weeks, there was a significant increase in erythroid colonies upon 126/KD and a trend toward increased myeloid colonies, whereas the less primitive CD38^lo/+ fractions only yielded more myeloid colonies as seen with total CD34⁺ cells (Figures 2D and S2B). Overall, these results indicate that in the CD38⁻ compartment, 126/KD expands erythroid and myeloid clonogenic progenitors, whereas in the less primitive CD38^+/lo compartment, only myeloid progenitor output is increased. The role of miR-126 in modulating primitive HSPC is reinforced by the finding that 126/OE in CD34⁺CD38⁻ HSPC yielded a 1.23-fold decrease in total colony numbers (p = 0.017) involving both myeloid and erythroid colonies (Figure 2E and data not shown). Moreover, megakaryocyte/erythroid specification was altered by modulating miR-126 (Figures S2C–S2H and Supplemental Results). The regulatory function of miR-126 was conserved across species with very similar effects seen in murine HSPC subpopulations (Figures S2I–S2K).

Together, our data suggest that miR-126 constrains proliferation of primitive hematopoietic progenitors, consequently reducing the overall output of mature erythroid and myeloid cells upon in vitro culture.

miR-126 Knockdown Expands HSC In Vivo without Exhaustion

Although our in vitro results suggest a functional role for miR-126 in primitive HSPC, in vivo repopulation studies, where HSC reside in a physiological niche, remain the gold standard to prove that HSC self-renewal and differentiation are affected. To determine how reduced miR-126 expression would affect human HSC function, we xenotransplanted CD34⁺CD38⁻ lin− CB HSPC transduced with 126/KD and CTRL vector to similar levels as assayed in vitro. Nine weeks after transplantation, 126/KD resulted in an increase in hematopoietic output as measured by human CD45⁺GFP⁺ engraftment of immune-deficient mice (Figure 3A). Both CD33⁺ myeloid and CD19⁺ lymphoid lineages were similarly affected, suggesting that an upstream progenitor cell is expanded by 126/KD. Phenotypic analysis of the most primitive human HSPC compartment was undertaken 9 weeks after transplantation. The frequency of CD34⁺CD38⁻CD90⁺CD45RA⁻ cells within the lineage-depleted CD45⁺GFP⁺ graft was increased upon 126/KD (Figures 3B and 3C). Assessment of cycling characteristics of the CD34⁺CD38⁻CD90⁺ HSC-enriched fraction showed a 2.8-fold increase in the fraction of cells in S/G₂/M upon 126/KD (Figure 3B), indicating that the observed expansion coincides with increased cell-cycle progression. Although increased cycling typically induces progressive exhaustion of HSCs, 126/KD HSPC still showed increased contribution to hematopoiesis at 20 weeks after transplantation (data not shown). To determine whether the observed increase in phenotypic HSC correlated with an increase in HSC function, human CD45⁺GFP⁺ CTRL and 126/KD cells from primary recipient mice were sorted and transplanted into secondary mouse recipients in a limiting dilution analysis (LDA). After 10 weeks, encompassing an overall total 30 weeks repopulation, there was a 3-fold increase in functional HSC within 126/KD mice (p = 0.01; Figure 3D)—results that strongly correlate with the phenotypic analysis.

It is possible that the expansion of phenotypic human HSC within xenografts following 126/KD could be influenced artificially due to the species mismatched physiological niche. Therefore, competitive transplantation assays were performed in the Ptprc congenic mouse model, transducing Ptprc^b (CD45.2) or Ptprc^a (CD45.1) lin⁻ BM HSPC, respectively, with either the 126/KD or CTRL vector. Complete blood counts (CBC) of repopulated mice were normal, even though 126/KD vector-transduced cells tended to contribute less to active hematopoiesis, particularly within the B and T lymphoid lineages (Figures 3E and S3). Steady-state bone marrow (BM) chimerism analyzed 5–8 months after transplantation showed much higher chimerism within lin⁻Sca⁺Kit⁺ (LSK) Cd150⁺Cd48⁻ phenotypic HSC-enriched compartments of the 126/KD group (Figures 3F and S3). In contrast, 126/KD cell chimerism was reduced in Lin⁻Sca⁺Kit⁻ lymphoid precursors (Harman et al., 2008) and lymphocyte compartments. These data suggest that 126/KD expanded the murine HSC compartment and reduced lymphopoiesis in vivo. To substantiate this hypothesis, we performed serial BM transplantation into secondary recipients. The contribution of 126/KD cells increased in myeloid and lymphoid lineages and surpassed the CTRL in the myeloid compartment (Figures 3E and S3), confirming that the relative increase in phenotypically defined 126/KD HSC detected in the primary BM translates into increased multilineage contribution to hematopoiesis in the secondary hosts. BM analysis of the secondary hosts mirrored the differences in chimerism observed in corresponding populations within the primary BM but shifted toward higher contribution from 126/KD cells (Figure 3F). Switching congenic strains between 126/KD and CTRL groups yielded similar results, confirming the role of 126/KD on HSC function (Figure S3B). We also measured the cell cycle of freshly isolated LSK cells from the BM of transplanted mice and found a significantly increased proportion of 126/KD LSK cells in S/G₂/M as compared to CTRL vector-transduced cells, whereas there was no significant difference in cell-cycle distribution of the committed Lin⁻Sca⁻Kit⁺ progenitors (Figure 3G). Furthermore, we confirmed the increased S/G₂/M fraction in 126/KD cells highly enriched for HSC (LSK CD150⁺; Figure 3H). These data indicate that 126/KD increased HSC proliferation within an in vivo physiological niche.

The gold standard proof of HSC expansion requires frequency analysis using transplantation at limiting doses (LDA). Ptprc^a mice were reconstituted with BM cells from Ptprc^b mice transduced with either the 126/KD or CTRL vectors. Sixteen weeks after transplantation, mice selected on the basis of similar levels of peripheral blood chimerism (Figure S3D) showed expansion of phenotypically defined HSC in 126/KD mice (Figure S3E). We then set up an LDA from 126/KD and CTRL BM cells. Recipient mice were scored as engrafted if they displayed >0.1% Ptcrp^b cells in multiple lineages. HSC frequency was 3.9-fold higher in 126/KD versus CTRL BM (Figure 3I).

Collectively, our results establish that stable knockdown of miR-126 expanded mouse and human long-term repopulating HSC, assessed by using the most stringent phenotypic and functional assays available. Although increased cycling of HSCs is frequently associated with their exhaustion, despite 1.5 years of follow up, no HSC exhaustion or malignancy was observed in our studies, pointing to the unique regulatory properties of miR-126 in HSC.

Enforced miR-126 Expression Progressively Reduces Hematopoietic Output In Vivo

Because 126/KD induced an expansion of the phenotypic HSC compartment through increased cycling, we predicted that forced expression of miR-126 may provoke the opposite response in vivo. Following xenotransplantation of lin− CD34⁺CD38⁻ CB HSPC, the progeny of cells that were transduced with the 126/OE vector transiently engrafted mice but were progressively lost between 12 and 16 weeks (Figure 4A). No significant alteration of the myeloid/lymphoid ratio was detected; however, at 16 weeks, the mice injected with 126/OE-transduced cells displayed an 11-fold lower percentage of GlyA⁺ erythroid cells compared to the control group, which was consistent with our in vitro observations (Figure S2H). We focused on the remaining HSC compartment and were able to carry out detailed analysis by pooling BM from four to five recipient mice per time point. Evaluation of primitive cell marker expression in the human lin⁻ compartment revealed that the loss of human 126/OE cells in transplanted mice was not due to a decrease in the CD34⁺CD38⁻CD90⁺ stem cell compartment size (Figure 4B). On the contrary, we observed, on average, a 2.5-fold increase in the phenotypic CD34⁺CD38⁻CD90⁺ stem cell compartment for 12 weeks posttransplant, after which marked primitive cells were no longer recoverable (Figure 4C). In addition, cell-cycle analysis of HSC-enriched Lin⁻CD34⁺CD38⁻CD90⁺ cells with 126/OE in the xenografted BM over time showed a consistently higher percentage of cells in G₀ (Ki67⁻; Figures 4B and 4D). These data indicate that miR-126 negatively regulates the cell-cycle progression of HSC, which is consistent with our in vitro experiments and is contrary to the 126/KD results. Surprisingly, no change in the numbers of quiescent cells in the more heterogeneous CD34⁺CD38⁺ progenitor-enriched compartment was observed; instead, a significant increase in the fraction of cells in S/G₂/M phase was noted (Figures 4B and 4D). Thus, the effects of 126/OE appeared to be context dependent and distinct between stem cell and progenitor-enriched fractions.

To examine the effect of 126/OE on the fate of murine HSC in vivo, we performed competitive transplantation experiments. Lin⁻ BM cells from congenic mice (Ptprc^a and Ptprc^b, respectively) were transduced with a 126/OE or CTRL vector and injected in a 1:1 ratio into myeloablated recipients (Figure 4E). In the first month after transplantation, myeloid (Cd11b+) and B cell (Cd19+) output from 126/OE cells was increased relative to the output of control vector-transduced cells. Following this initial burst of differentiated cells, 126/OE cells declined and became progressively lost from the peripheral blood upon hematopoietic stress induced by 5-Fluorouracil administration or serial transplantation into secondary recipient mice (Figure 4E). A similar phenotype was observed when ectopically expressing miR-126 from a weaker promoter (see Supplemental Results and Figure S4).

Next, we assessed the impact of 126/OE on the early engraftment phase in a noncompetitive transplantation setting. A significant advantage of early blood reconstitution by 126/OE cells was seen for platelets and leukocytes, whereas neutropenia and thrombocytopenia were shortened (see Supplemental Results). BM analysis conducted at 3 weeks posttransplantation showed that, whereas the Lin⁺ compartment was reconstituted to similar levels in all groups, absolute numbers of Lin⁻ cells were significantly increased in 126/OE BM, primarily in the Sca⁺Kit⁻ and Sca⁻Kit⁻ compartments (Figure 3F, top). In vivo nucleoside incorporation assays revealed increased labeling in LSK, Lin⁻Sca⁺Kit⁻, and Lin⁻Sca⁻Kit⁻ cells, but not in Lin⁻Sca⁻Kit⁺ or Lin⁺ cells (Figure 3F, bottom). Analysis of the BM of engrafted mice 3 months after transplantation revealed a substantially decreased 126/OE chimerism in the LSK CD150⁺CD48⁻ HSC compartment with respect to all other BM populations (Figures 4G and S4F). Moreover, although more 126/OE LSK cells were in the S/G₂/M phase of the cell cycle, HSC-enriched fractions (LSK CD150⁺) contained significantly more 126/OE cells in G₀ (Figure 4H). This is strikingly reminiscent of the situation in human cells, in which CD34⁺CD38⁺ progenitors were cycling more frequently upon 126/OE, whereas the HSC-enriched CD34⁺CD38⁻CD90⁺ cells displayed increased quiescence.

Overall, these results illustrate that miR-126 regulation of cell fate differs depending on cellular context. During the first weeks after transplantation of 126/OE cells, specific progenitor subsets are expanded, which is reflected in an increased multilineage output. Over time, a competitive disadvantage of 126/OE at the HSC level results in a progressive reduction of the progenitor supply, leading to the collapse of hematopoietic output. A constant reduction in cell-cycle entry of HSC may be the underlying explanation for the finding that 126/OE HSC are eventually outcompeted by nontransduced HSC with normal cycling characteristics.

PI3K/AKT Signaling Is Targeted by miR-126 in Early Hematopoietic Progenitors

To gain insight into the mechanism whereby miR-126 was exerting its biological effects on HSPC, gene expression profiling was performed on sorted CD34⁺CD38⁻ lin− CB HSPC transduced with 126/KD, 126/OE, and their respective CTRL vectors. Gene set enrichment analysis (GSEA) was applied to HSPC gene expression data to identify miR-126 modulated cellular pathways. Opposing effects between 126/KD and 126/OE groups were observed with a majority of enriched pathways (Figure 5A). Furthermore, pathways most highly correlated with miR-126 expression levels centered on cell-cell contact (cell junction and adhesion), motility, migration (all upregulated in 126/KD and downregulated in 126/OE), interferon-gamma signaling, and protein acetylation (both downregulated in 126/KD). Using four published algorithms, a list of predicted miR-126 target genes was generated and compared with the gene expression data set. Predicted targets were found in six pathways that are significantly influenced by miR-126 (Figure 5A, purple lines, p < 0.05). The focal adhesion pathway showed the most significant overlap with the list of miR-126 predicted targets (Figure 5A, list 1, p = 0.0003), indicating that this may be the primary pathway under the influence of miR-126 in primitive lin− CB cells. In this pathway, PIK3R2, ITGA6, and ILK, all predicted miR-126 targets, are upregulated upon 126/KD and downregulated upon 126/OE. As the effect of miRNA modulation of gene expression may be stronger at the protein level, and to provide another unbiased approach, we performed a quantitative analysis on the proteome of K562 cells stably transduced with 126/OE or CTRL vectors by using the stable isotope labeling of amino acids in culture (SILAC) technique (Figures S5A–S5C). Interrogation of this data set by using unsupervised Ingenuity Pathway Analysis (IPA) found that, upon 126/OE, the PI3K/AKT signaling pathway was among the most significantly affected canonical pathways (Figures S5B and S5C), validating the bioinformatic analysis of the mRNA arrays. These data suggest that PI3K/AKT signaling is a major pathway through which miR-126 exerts its effects in primary human hematopoietic cells.

Because the PI3K/AKT pathway transduces crucial signals for HSC proliferation and differentiation (Buitenhuis and Coffer, 2009; Buitenhuis, 2011), we assessed the expression levels of individual components of this pathway in both K562 cells (Figure S5D) and CD34⁺ CB cells (Figure 5B) upon 126/KD. Among the investigated genes, PIK3R2/PI3Kp85β and AKT2 were significantly upregulated upon 126/KD at both the transcript and protein level, whereas their respective isoforms, PIK3R1/PI3Kp85a and AKT1, were unchanged. CRKII protein was also upregulated by 126/KD as predicted by bioinformatic analysis. Whereas PIK3R2 is well established as a direct miR-126 target (Fish et al., 2008; Guo et al., 2008), AKT2 may be regulated through indirect mechanisms because its transcripts are not predicted to contain binding sequences for the miR-126 seed sequence.

miR-126 Attenuates the PI3K/AKT/GSK3β Signaling Axis upon Extrinsic Stimuli

We next assessed whether miR-126-dependent regulation of the PI3K/AKT pathway would have an effect on signal transduction. AKT interacts with several downstream effectors implicated in stem cell regulation such as mTOR complexes and GSK3β and FOXO proteins (Buitenhuis and Coffer, 2009). AKT is activated upon phosphorylation of Ser473. We found increased phosphorylation of AKT at Ser473 (p-AKT S473) in 126/KD-transduced CD34⁺ CB HSPC under steady-state culture conditions (Figures 5C and 5D; see also Figure S5E for K562 cells). To investigate the acute response to cytokine stimulation at the single cell level in primary human cells, we assayed pAKT (S473) by flow cytometry in CD34⁺ HSPC starved in serum and cytokine-free medium prior to stimulation. Stem cell factor (SCF) containing cytokine cocktails rapidly and strongly induced pAKT (S473) in CD34+ cells (Figure S6). This response was significantly enhanced in 126/KD HSPC, and this enhancement was most evident when the analysis was focused on highly transduced cells (Figure 6A). By using the same starvation/stimulation protocol, we also found significantly increased phosphorylation of GSK3β at Ser9 in 126/KD CD34⁺ CB cells (Figure 6A). An opposite effect on AKT and GSK3β phosphorylation was observed upon 126/OE (Figure 6B). Overall, these data indicate enhanced transduction of the signal induced by SCF through the PI3K/AKT pathway in human HSPC upon 126/KD and decreased signal transduction upon 126/OE. Finally, we were able to abrogate the effect of 126/KD on HSPC proliferation by a single administration of the PI3K inhibitor LY294002 at the time of cytokine stimulation, reinforcing the notion that activation of PI3K signaling is required for the observed functional effects of 126/KD in HSPC (Figure 6C).

Collectively, these data indicate that miR-126 regulates the expression of several components of the PI3K/AKT pathway and, consequently, regulates its activation by extrinsic stimuli such as SCF.

Vector Transduction of Cord Blood and In Vitro Expansion

The following transduction protocols were used with similar results in Toronto (1) and Milan (2). (1) Lin− CB cells were thawed and plated in X-VIVO (BioWhittaker) supplemented with 20% BIT 9500 (StemCell Technologies), 2 mM L-glutamine, and 100 U/ml penicillin/streptomycin (P/S). Viral particles were resuspended in the same medium and added to cells at a multiplicity of infection (MOI) of 30–50. The mixture was then supplemented with stem cell factor (SCF) 100 ng/ml, Flt3 ligand (Flt3-L) 100 ng/ml, thrombopoietin (TPO) 20 ng/ml, and interleukin 6 (IL-6) 20 ng/ml for a minimum of 16 hr. Subsequent expansion was performed in the same medium with half the concentration of these cytokines. (2) When indicated, the following conditions, referred to as TSF6, were used: CD34⁺ cells were cultured in StemSpan medium (Stem Cell Technologies) with the same supplements and cytokines as mentioned above except BIT. After 12–18 hr of prestimulation, cells were exposed to lentiviral vectors for 12–24 hr at an MOI of 50–100. Optionally, a second hit transduction was performed after a 12 hr resting phase in fresh medium. Subsequent expansion was performed in StemSpan medium containing half the concentration of the above-mentioned cytokines (TSF6^half medium).

Mouse Xenotransplantation and Human Lin− Cell Isolation

Mouse xenografts were performed as described previously (McDermott et al., 2010). Briefly, NOD/Lt-scid/IL2Rɣ^null (NSG) mice were sublethally irradiated (225 cGy) 1 day prior to injection. Transduced cells were injected with 25 μl PBS into the right femur of each recipient mouse. After euthanizing the mice, bone marrow cells were flushed with 2 ml PBS 2% FCS, and 50 μl was stained for surface markers. Remaining cells were lineage depleted and/or used for additional surface or intracellular antigen staining. For lineage depletion, BM cells flushed from the femurs and tibias of three to five mice (>3% human cells) were combined and processed with the StemSep Mouse/Human Chimera Enrichment Kit (Stem Cell Technologies) according to the manufacturer’s instructions. However, during the antibiotin incubation step, an additional 50 μL/ml human hematopoietic progenitor enrichment antibody cocktail from the StemSep Human Progenitor Cell Enrichment Kit was added to deplete human-lineage-positive cells.

Isolation and Transduction of Murine HSPC and Bone Marrow Transplants

Donor mice between 6 and 8 weeks of age were euthanized by CO₂, and BM was retrieved by femurs, tibias, and humeri. Murine HSPCs were purified by lin− selection, using the Biotin Mouse Lineage Panel (BD PharMingen) and the StemSep Mouse Progenitor Enrichment Kit (Stem Cell Technologies). LVs transduction was performed as previously described (Gentner et al., 2010). For noncompetitive transplants, ≥10⁶ transduced cells were transplanted via tail injection into 6-week-old lethally irradiated mice (10 Gy total, fractionated into two doses with an interval of 4 hr). For competitive transplants, congenic HSPC from either B6.SJL-Ptprc^a(CD45.1) or C57BL/6NTac-Ptprc^b(CD45.2) mice transduced with either 126/KD or 126/OE and with CTRL LVs were mixed in a 1:1 ratio and transplanted (≥10⁶ cells/mouse) into 6-week-old lethally irradiated mice. Please see Supplemental Information for details of the analysis of chimerism.

Limiting Dilution Analysis

For murine experiments, Lin⁻ BM cells (CD45.2) were transduced with 126/KD or CTRL LV and transplanted into recipient mice (CD45.1). Mice with similar chimerism and transduction levels in both the 126/KD and CTRL group were identified 16 weeks after transplant, and a 126/KD BM pool (n = 3 mice) and CTRL BM pool (n = 3 mice) were generated. Limiting doses of total nucleated BM cells (BM TNCs) from each pool were mixed with 3 × 10⁵ CD45.1 supporter BM TNCs and transplanted into n = 13 lethally irradiated CD45.1 recipient mice. A mouse was scored as positive if it had >0.1% engraftment in multiple lineages by serial bleeding at 8 and 14 weeks after transplantation. Limiting dilution analysis of human CB cells was performed by sorting human CD45⁺GFP⁺ cells from pooled BM of primary mice 20 weeks after transplantation and injection of different doses into secondary recipients. A secondary mouse was scored as positive if it had >0.5% BM engraftment 10 weeks after transplantation. HSC frequency was estimated by linear regression analysis and Poisson statistics using publicly available ELDA (Extreme Limiting Dilution Analysis, http://bioinf.wehi.edu.au/software/elda/) software (Hu and Smyth, 2009).

5-FU Treatment and Secondary Transplants

5-Fluorouracil (Sigma) was administered intraperitoneally at a dose of 2.5 mg. For secondary transplants of murine experiments, stably engrafted mice were euthanized (≥14 weeks after primary transplant), and 10⁷ BM-nucleated cells from each mouse were injected into one or two secondary recipients.

Statistical Analysis

Unless otherwise indicated, mean ±SEM values are reported in the graphs. For pairwise comparisons, a Student’s t test (paired if appropriate) was used unless otherwise indicated. For three or more matched groups, a one-way analysis of variance (ANOVA) for repeated measures using a Bonferroni posttest correction was used. Chimerism (chim) values were transformed into a log-odds scale (log (%chim/(100-%chim)), given that percentages are by definition constrained between 0 and 100.