Research ArticleGloboid Cell Leukodystrophy

Identification of Hematopoietic Stem Cell–Specific miRNAs Enables Gene Therapy of Globoid Cell Leukodystrophy

See allHide authors and affiliations

Science Translational Medicine  17 Nov 2010:
Vol. 2, Issue 58, pp. 58ra84
DOI: 10.1126/scitranslmed.3001522


Globoid cell leukodystrophy (GLD; also known as Krabbe disease) is an invariably fatal lysosomal storage disorder caused by mutations in the galactocerebrosidase (GALC) gene. Hematopoietic stem cell (HSC)–based gene therapy is being explored for GLD; however, we found that forced GALC expression was toxic to HSCs and early progenitors, highlighting the need for improved regulation of vector expression. We used a genetic reporter strategy based on lentiviral vectors to detect microRNA activity in hematopoietic cells at single-cell resolution. We report that miR-126 and miR-130a were expressed in HSCs and early progenitors from both mice and humans, but not in differentiated progeny. Moreover, repopulating HSCs could be purified solely on the basis of miRNA expression, providing a new method relevant for human HSC isolation. By incorporating miR-126 target sequences into a GALC-expressing vector, we suppressed GALC expression in HSCs while maintaining robust expression in mature hematopoietic cells. This approach protected HSCs from GALC toxicity and allowed successful treatment of a mouse GLD model, providing a rationale to explore HSC-based gene therapy for GLD.


Allogeneic hematopoietic stem cell (HSC) transplantation is an established therapy for hematologic and systemic diseases (1). Although still in experimental stages, genetic manipulation of autologous HSCs represents a promising treatment approach for many monogenic diseases because it bypasses the need for a compatible donor and eliminates the threats of immunosuppression and graft-versus-host disease. HSC gene therapy has been reported to be as effective as allogeneic transplantation in the treatment of inherited immunodeficiencies such as X-linked severe combined immunodeficiency (XSCID) and adenosine deaminase–deficient SCID, as well as in the peroxisomal disorder adrenoleukodystrophy (25). Moreover, in diseases such as lysosomal storage disorders, HSC gene therapy may become even more effective than allogeneic transplantation because the therapeutic gene dosage can be increased in the macrophage and microglia cells that repopulate the affected tissues and reconstitute lysosomal enzyme activity after hematopoietic stem and progenitor cell (HSPC) transplantation (6, 7).

However, if forced ectopic transgene expression in HSPCs is toxic, counterselection of gene-modified cells could occur, resulting in loss of therapeutic efficacy or in altered cellular behavior, including the deregulation of cell growth control (8, 9). Therefore, an important goal of HSC gene therapy is minimal perturbation of HSPC function by avoiding ectopic or nonregulated transgene expression, instead using HSPCs as a silent, long-term reservoir for gene-modified hematopoietic progeny. The generation of such silent HSPCs requires a deep understanding of the regulatory machinery that operates specifically in HSPCs and not in more mature cell types.

HSC “stemness” is maintained by a complex network of signaling pathways and transcription factors that balance self-renewal and differentiation (10). MicroRNAs (miRNAs), a class of noncoding RNAs that control gene expression at the posttranscriptional level by binding to regulatory sites embedded within the 3′ untranslated region of multiple target messenger RNAs (mRNAs) (11), coordinate regulation of multiple gene targets interconnected in common biological pathways, leading to substantial changes in cell fate and behavior. Recent studies demonstrate the importance of miRNAs in governing lineage commitment of hematopoietic progenitors (12), but much less is known about miRNA function in HSCs because of the technical difficulties in obtaining homogenous cell populations for miRNA profiling. Indeed, there are no expression data that link a specific miRNA to a functionally defined HSC. Thus, sensitive miRNA detection methods are needed that combine functional repopulation assays with miRNA expression in single cells.

Here, we applied an miRNA reporter vector technology (13, 14) to identify several miRNAs associated with the primitive HSPC compartment in both mice and humans and showed that their expression can be exploited to prospectively identify and isolate human HSPCs. Moreover, we demonstrated that these miRNAs effectively suppressed transgene expression in HSPCs and protected these cells from transgene toxicity in a gene therapy mouse model of globoid cell leukodystrophy (GLD; also known as Krabbe disease), an invariably fatal lysosomal storage disorder caused by mutations in the galactocerebrosidase (GALC) gene (15), resulting in successful treatment of GLD.


An miRNA activity footprint was identified in murine HSPCs and their progeny

To attain a comprehensive picture of miRNA activity within each of the major classes of stem and progenitor stages of hematopoietic lineage differentiation, we selected a panel of candidate miRNAs and constructed bidirectional miRNA reporter vectors (BdLV) for each of them (fig. S1A). These vectors coexpress the short half-life reporter, destabilized green fluorescent protein (GFP), which contains miRNA target (miRT) sequences, and an internal normalizer [truncated low-affinity nerve growth factor receptor (ΔNGFR)] to control for vector dose, transduction level, and promoter activity. Flow cytometry enables quantification of miRNA expression in single, viable cells by calculating the fold repression (FR) of the miRNA reporter compared to a control vector containing an irrelevant or no miRT sequences (13, 14). For each miRNA, mouse HSPCs were transduced with the specific reporter BdLV and transplanted into lethally irradiated mice (Fig. 1A). We then quantified specific miRNAs in distinct hematopoietic subpopulations isolated from the reconstituted mice by using multiple surface markers (Fig. 1).

Fig. 1

Footprinting miRNA activity in prospectively identified mouse HSPCs and differentiated progeny. (A) Lin bone marrow (BM) hematopoietic stem and progenitor cells (HSPCs) were transduced with bidirectional miRNA reporter lentiviral vectors (BdLVs) and transplanted into lethally irradiated mice. Mice were euthanized 8 to 12 weeks after transplant, and multiple BM HSPC subpopulations were prospectively identified by immunophenotyping. HSC, hematopoietic stem cells; MPP1/MPP2, multipotent progenitors; GMLP, granulocyte-monocyte-lymphocyte progenitors; GMP, granulocyte-monocyte progenitors; eMEP, early megakaryocyte-erythrocyte progenitors; EP, erythrocyte precursors. (B) Reporter expression of the control (CTR)-BdLV and a miR-223 reporter BdLV (BdLV-223T) in GMP and EP isolated from the BM of transplanted mice. [x], mean fluorescence intensity (MFI) of low-affinity NGFR within the NGFR+ cells; [y], GFP MFI within NGFR+ cells. [y] was corrected for autofluorescence in the GFP channel [af]y measured in cells transduced with a BdLV lacking GFP (ΔNGFR/TK). TGR, transgene ratio. miRNA activity is expressed as the fold repression (FR). (C) Representative FACS (fluorescence-activated cell sorting) plots show expression of the CTR-BdLV (no miRT or 133aT, a muscle-specific miRT; n = 9 mice) and reporter BdLVs for miR-126 (126T; n = 10), miR-130a (130aT; n = 4), and miR-196b (196bT; n = 4) in HSPC subpopulations freshly isolated from the BM of transplanted mice as described in (A). Bar graphs on the bottom show activity of the three miRNAs in each subpopulation (mean ± SEM; ***P < 0.001; **0.01 > P > 0.001 compared to control). (D) Hematopoietic activity of eight miRNAs as measured by the mean FR of indicated reporter BdLVs in multiple cell populations isolated from transplanted mice. Lineage+ BM subsets: Pro B, CD19+CD43+; CD43 B, CD19+CD43; T Ly, CD3+; Mono, CD11b+CD48+; Granu, CD11b+CD48lo; peripheral blood: Granu, CD11b+ side scatterhi; Mono, CD11b+ side scatterlo; B Ly, CD19+; T Ly, CD3+. (E) BM from young adult Tg.126T mice was immunophenotyped as described in (A), and GFP MFI of Tg.126T mice and control GFP transgenic mice was measured to calculate the FR of the miR-126 reporter in the indicated HSPC subpopulations (mean ± SEM, n = 10 Tg.126T mice; ***P < 0.001; **0.01 > P > 0.001 compared to EP). Fluorescences on FACS plots are shown in log10 scale. SSC, side scatter.

This reporter method revealed expression of miR-223 in granulocyte-monocyte progenitors (GMPs; FR, 7.1) but not in erythrocyte precursors (EPs; FR, 1.1) (Fig. 1B). miR-126, miR-130a, and miR-196b showed the highest activity in fractions enriched for the most primitive HSC, and this activity was lost during early stages of differentiation (Fig. 1, C and D, and figs. S2 to S4). miR-126, miR-130a, and miR-196b were largely inactive in differentiated cells of the lymphoid and myeloid lineages, with the exception of terminally differentiated granulocytes, which seem to reestablish some degree of miR-126 activity. miR-223 was expressed in most Kit+Sca1+Lineage marker (Lin) (KSL) cells, and in common myeloid progenitors (CMPs) and GMPs, but was sharply down-regulated in EPs and in megakaryocyte-erythrocyte progenitors (Fig. 1B and figs. S2 and S3). As expected, miR-223 was progressively up-regulated during myeloid differentiation, whereas B and T lymphocytes were devoid of it (Fig. 1D). Also, members of the miR-17 to miR-92 cluster (miR-19, miR-93a, and miR-17-5p) were highly expressed in HSPCs (Fig. 1D). However, expression was maintained during further differentiation and decreased only in terminally differentiated B cells and granulocytes (Fig. 1D). Finally, let-7a also was expressed at high amounts in all hematopoietic cell types, including bona fide HSCs (Fig. 1D).

To rule out that the bone marrow (BM) transplantation procedure affected the isolation of defined cellular subsets and/or miRNA expression, we derived a transgenic mouse line (Tg.126T mice) that harbored germline integrations of miR-126T (Fig. 1E and fig. S5). Flow cytometric analysis of Lin BM cells of young adult Tg.126T mice showed a similar pattern of miR-126 activity, as we observed in the transplanted mice (Fig. 1E). Because miR-126 is expressed in endothelial cells, and loss of function during development results in fetal mortality due to defective angiogenesis, it was important to rule out any potential toxicity from the miR-126T reporter (16, 17). No gross phenotypic abnormalities were present in Tg.126T mice, and upon intercross, a normal number of pups were obtained that also maintained similar numbers of vector integrants as their parents (fig. S5B). Thus, miR-126T sequences expressed from the phosphoglycerate kinase 1 (PGK) promoter do not interfere with the regulation of natural miR-126 targets in endothelial cells during development. Competitive repopulation experiments maintained for at least 1 year ruled out any toxic effects in hematopoietic cells, because the peripheral blood chimerism for all major blood lineages was similar for HSPCs derived from wild-type or Tg.126T mice (fig. S5C). Collectively, these data confirm that miR-126 is physiologically expressed in HSCs and that the genetic reporter does not affect normal hematopoiesis.

HSCs are protected from conditional suicide by miR-126

To conclusively establish the activity of selected miRNAs in functionally defined cell subsets, we devised a conditional suicide system based on lentiviral vectors that expressed the herpes simplex virus thymidine kinase (TK) gene regulated by different miRT sequences (Fig. 2A and fig. S1B). For a more sensitive assay, we used a destabilized TK protein (called dTK) with a reduced half-life (fig. S6). HSPCs transduced with a suicide vector (Fig. 2A) or a GFP control were plated in semisolid medium in the presence or absence of ganciclovir (GCV) (Fig. 2B). HSPCs transduced with the control dTK vector did not give rise to colonies in the presence of GCV. Adding miR-142T sequences to the dTK transcript completely rescued colony formation, in line with expression of this miRNA in all hematopoietic lineage cells (13). miR-223T partially restored the growth of myeloid colonies, whereas erythroid colony number was significantly reduced (P < 0.001) and similar to the control (Fig. 2B). For dTK-126 and dTK-130aT, myeloid and erythroid colonies were significantly reduced (P < 0.01) or absent in the presence of GCV, respectively. These functional data are concordant with down-regulation of miR-126 and miR-130a during differentiation (Fig. 2B).

Fig. 2

Regulating the expression of a suicide gene by miRNA target sequences. (A) miRT sequences perfectly complementary to miR-142, miR-223, miR-126, or miR-130a were added to a destabilized thymidine kinase (dTK) transcript. Monodirectional lentiviral suicide vectors (right half of each vector drawing) were used for (B), whereas bidirectional suicide vectors coupling a GFP marker to the miRNA-regulated dTK or an NGFR marker to CTR-dTK were used in (C). (B) Lin HSPCs were transduced with the indicated lentiviral vectors and plated ± ganciclovir (GCV) in semisolid medium (LV-GFP, n = 2; CTRL-TK, n = 8; TK-142T, n = 4; TK-223T, n = 6; TK-126T, n = 4; TK-130aT, n = 2). Box and whiskers plots show the colony number in the cultures containing GCV divided by the colony count in respective control cultures without GCV (10th to 90th percentile). Left graph, myeloid colonies (CFU-GM/-G/-M); right graph, erythroid colonies (BFU-E/CFU-E). The “no GCV” data point shows plating variability (colony count of each non–GCV-treated culture divided by the mean colony count of all non–GCV-treated cultures; n = 26). Statistical comparisons were made against the GCV-treated, LV-GFP group. ns, not significantly different. **0.001 < P < 0.01; ***P < 0.001. (C) Lin HSPCs from CD45.1+ mice were transduced with either the CTR-dTK vector (NGFR-marked) or an miRNA-regulated dTK vector (GFP-marked; experiment 1, TK.126T; experiment 2, TK.142T). LV.NGFR/CTR-dTK– and LV.GFP/dTK-miRT–transduced cells were mixed in a 1:1 ratio and transplanted into CD45.2+ congenic mice, which were treated with GCV (experiment 1, n = 6 mice; experiment 2, n = 5 mice) for 7 to 14 days starting at day 3 after transplant or left untreated (experiment 1, n = 4; experiment 2, n = 3). Representative FACS plots (axis labels: log10 scale) show GFP/NGFR chimerism within CD45.1+ donor cells. Graphs show the fraction of GFP-expressing cells within the transduced, donor-derived cells in the blood for the GCV-treated (red) and untreated (black) mice for each blood cell type over a 7- to 8-month time period. Note that significantly more cells are GFP+ in the GCV group (P < 0.001 for all lineages, two-way ANOVA), indicating protection from suicide and a selective advantage of HSC transduced by GFP/dTK-126T or GFP/dTK-142T.

To obtain direct evidence that miRNAs are expressed in functional HSCs defined on the basis of repopulation, we developed an in vivo suicide assay. Pilot experiments involving co-transplant of TK-transduced cells with untransduced BM-supporting cells indicated that a 1-week time course of GCV given within the first 2 weeks of engraftment efficiently eliminated TK-transduced long-term repopulating HSCs (LT-HSCs) (fig. S7). We then transduced HSPCs with either an miRNA-regulated bidirectional suicide vector expressing dTK-126T or dTK-142T and GFP, or a control bidirectional suicide vector expressing dTK and ΔNGFR (Fig. 2A). Cells transduced with the control or one of the miRNA-regulated suicide vectors were then competitively transplanted into congenic mice, which either did or did not receive GCV during the engraftment phase (Fig. 2C). Long-term analysis of peripheral blood chimerism indicated that most of the NGFR+ cells were efficiently eliminated in GCV-treated mice, whereas GFP+ cells persisted in increased relative numbers (Fig. 2C). This was observed in multiple blood lineages over a >7-month time period for both dTK-126T– and dTK-142T–transduced cells (Fig. 2C). These data conclusively establish that both miR-126 and miR-142 are expressed in LT-HSC at amounts sufficient to prevent TK protein expression and cell death induced by GCV.

miR-126 marks a population of primitive human HSPCs containing all SCID repopulation activity

Because murine and human HSCs do not always share identical properties, it is essential to determine whether the HSPC-specific miRNAs we identified in the mouse were conserved and functionally important in human HSPCs. We transduced human cord blood (CB) CD34+ cells with reporter BdLVs for miR-126, miR-130a, and miR-223 (Fig. 3A). Cells were cultured in vitro under conditions that provide short-term maintenance of HSPCs, and subpopulations were identified by flow cytometry according to the expression of CD34 and CD38. Myeloid (CD13+) and erythroid (CD235+) differentiation was assessed with the methylcellulose assay. We concluded that miR-126, miR-130a, and miR-223 were present in the CD34+ HSPCs because they all suppressed their respective reporter transcripts (Fig. 3A). After the cells differentiated, miR-223 maintained activity in the myeloid lineage, but decreased during erythroid differentiation. In contrast, miR-126 activity was lost during myeloid differentiation but maintained in the erythroid progeny. These patterns of expression were functionally verified by conditional suicide assay (Fig. 3B). miR-130a was down-regulated in both the myeloid and the erythroid lineages. Quantification of miRNA activity in the respective populations (Fig. 3A) indicated that, among the miRNAs tested, miR-126 was the most potent miRNA in the CD34+CD38 CB fractions enriched for primitive HSPCs. An independent experiment (n = 3 donors) showed that the highest miR-126 activity (FR, 20 to 50) occurred in CD34+CD38CD90+ CB cells and that there was progressive loss of activity during granulocyte colony-stimulating factor (G-CSF)–induced differentiation into mature granulocytes (fig. S8). Notably, treatment of CD34+ CB cells with a miR-126 antagomir, but not with a control antagomir, resulted in increased GFP expression from BdLV-126T, confirming that the vector specifically detects miR-126 activity (fig. S9).

Fig. 3

miR-126 activity in human HSPCs. (A) CD34+ HSPCs were purified from cord blood (CB) and transduced with the control-BdLV (Control) or miRNA reporter BdLVs for miR-126 (126T), miR-130a (130aT), or miR-223 (223T). BdLV reporter expression (representative FACS plots, log10 scale) is shown in CD34+CD38 HSPCs and CD34+CD38+ progenitors 2 days after transduction (two left-hand columns), in CD34CD38+ cells after 6-day liquid culture (middle column), or in CD13+ myeloid and CD235+ erythroid cells 16 days after plating in semisolid medium (two right-hand columns). Bar graphs below show quantification of miR-126, miR-130a, and miR-223 activity in the respective subpopulations (n = 4, 3, and 3, respectively; mean FR ± SEM). (B) CB CD34+ cells were transduced with control- or miRNA-regulated suicide BdLVs (dTK-223T or dTK-126T, see also Fig. 2) and plated for colony-forming cell (CFC) assays in quadruplicate ± GCV. The number of GFP+ erythroid (BFU-E and CFU-E) or myeloid (CFU-GM, CFU-G, and CFU-M) colonies was counted 14 days after plating and normalized to the number counted in the culture without GCV. Box and whiskers plots show 10th to 90th percentile. Statistical comparisons were made against the no GCV group. *0.01 < P < 0.05; ***P < 0.001. (C) Representative FACS analysis of human HSPC fractions in nonobese diabetic–severe combined immunodeficient (NOD-SCID) mice transplanted with Lin CB transduced with BdLV-126T. Eight to 9 weeks after transplantation, recovered BM cells were pooled, depleted of both murine hematopoietic cells and human Lin+ cells by negative selection, and analyzed for NGFR and GFP expression (three independent experiments, 7 to 10 mice in each experiment). The bar graph on the bottom shows miR-126 activity in each designated cell population calculated in each of the three independent experiments. For CD19+ and CD33+ cells, the mean FR from three independent experiments (each consists of three mice) ± SD is shown. (D) miR-126-3p expression (log scale) measured by Q-PCR in primitive (CD34+CD38) or committed progenitor fractions (CD34+CD38+, CD34CD38+, and CD34CD38) of human CB. U6 and 5S ribosomal RNA (rRNA) were used as normalization controls. Shown is the mean ± SD (n = 3 technical replicates).

The most definitive assay for human HSCs is multilineage repopulation of immune-deficient mice (18). To investigate miR-126 activity in human HSCs using in vivo xenotransplantation assays, we transduced Lin CB with the miR-126 reporter BdLV and then performed intrafemoral injection into preconditioned immunodeficient mice. Eight to 9 weeks after transplantation, BM cells were analyzed by multicolor flow cytometry (Fig. 3C). miR-126 activity was highest in the LinCD34+CD38CD90+ fraction (FR, 30 to 50), which corresponds to a highly enriched HSC fraction (19), followed by CD34+CD38CD90, CD34+CD38+, and CD34CD38+ fractions, respectively. miR-126 activity was almost absent in CD19+ B cells and CD33+ myeloid cells. Thus, the highest miR-126 concentrations mark the most primitive human HSPC subcompartment. To validate the miR-126 BdLV reporter activity observed within human HSPC-enriched populations, we performed quantitative polymerase chain reaction (QPCR) for miR-126 on sorted Lin CB cells. The primitive (CD34+CD38) fraction expressed 44 times as much miR-126 as more committed progenitor fractions (Fig. 3D), confirming a miR-126 expression gradient with differentiation.

To evaluate whether miR-126 was functional in HSCs and could serve as an HSC-specific biomarker, we transduced Lin CB cells with BdLV-126T. After 2 weeks of culture, the expression of surface antigens useful for purifying HSCs was almost completely lost. miR-126high and miR-126low populations were sorted and subjected to colony assay and transplantation into immune-deficient mice (Fig. 4A). Although the miR-126low fraction gave rise only to granulocyte and granulocyte-monocyte colonies and did not engraft in the mice, the miR-126high fraction gave rise to all types of colonies and successfully engrafted the mice with multilineage differentiating cells, establishing miR-126 as a functional marker of human HSC (Fig. 4, B and C).

Fig. 4

Prospective isolation of human NOD-SCID–repopulating cells on the basis of miR-126 activity. (A) Freshly thawed Lin CB cells were transduced with BdLV-126T and kept in liquid culture for 2 weeks. NGFR+ cells were fractionated into GFPhigh and GFPlow fractions by FACS sorting. Cells were enumerated on a hemocytometer, and viability was determined by trypan blue exclusion. The NOD-SCID repopulation potential of each fraction was determined by direct transplantation into the femur of preconditioned NOD-SCID mice. The progenitor activity of each population was assessed by standard methylcellulose colony assay. (B) Methylcellulose assay for miR-126high and miR-126low cell fractions (n = 3). (C) Engraftment in NOD-SCID mice. Engraftment was defined as more than 0.1% of human CD45+ cells in BM that are either CD19+ or CD33+. Representative FACS plots are shown (axes: log10 scale).

Thus, miR-126 expression marks primitive human hematopoietic cells with SCID-repopulating potential, even after 2 weeks of in vitro culture, and allows their prospective isolation.

Forced GALC expression is toxic to HSPCs but not to differentiated hematopoietic cells

To develop a model of gene therapy, we transduced HSPCs from wild-type and GLD mice that carry a point mutation in the GALC gene, resulting in <5% residual enzyme activity (Trs) with a GALC- or GFP-expressing lentiviral vector (Fig. 5A). Transduction with the GALC vector reconstituted GALC activity in the cultured progeny of GLD cells, resulting in twofold higher enzyme activity as in GFP-transduced wild-type cells (Fig. 5, B and C). Similar expression levels were observed after transduction of wild-type murine HSPCs or human CD34+ HSPCs from normal CB or BM (Fig. 5, B and C). Unexpectedly, forced GALC expression impaired clonogenic activity of both murine and human HSPCs when compared to GFP-transduced cells (Fig. 5D). Terminal deoxynucleotidyl transferase–mediated deoxyuridine triphosphate nick end labeling (TUNEL) assay performed 2 days after transduction showed that most GALC- but not GFP-transduced HSPCs were TUNEL-positive and exhibited enlarged nuclei with condensed chromatin (Fig. 5E), suggesting that the clonogenic impairment of GALC-transduced HSPCs was a result of apoptosis. Functional impairment of HSPCs was directly caused by forced, de novo GALC expression and not by toxicity related to the vector preparation, because HSPCs transduced with a miR-142–regulated, GALC-encoding lentiviral vector showed normal clonogenic activity and absence of apoptosis. Indeed, the 142T sequence suppressed GALC enzyme expression in hematopoietic cells but not in lentiviral vector producer cells (13) (see Fig. 5B). Forced, de novo GALC expression was also toxic to LT-HSCs, because GALC-transduced murine HSPCs failed to rescue Trs mice from lethal irradiation (Fig. 5F) and were functionally impaired in vivo. To test whether toxicity by forced or de novo GALC expression also affected differentiated cells, we transduced human primary monocytes, T and B lymphocytes, as well as mouse microglia cells with GALC or control vectors (Fig. 5G). Although efficient transduction and GALC overexpression were achieved in all cell types, TUNEL assay showed low or no apoptosis in the cultures (Fig. 5G). Thus, sensitivity to GALC expression is unique to HSPCs and not observed in mature hematopoietic cells.

Fig. 5

Selective toxicity of de novo GALC expression in HSPCs and rescue by miR-126 regulation. (A) The indicated LVs were used to transduce murine and human HSPCs obtained from Galc−/− (−/−) and wild-type (+/+) mice, as well as CB and BM from normal donors, respectively. (B and C) GALC activity (B) and vector copy number (VCN) (C) were measured in the in vitro culture progeny of the transduced murine (top panels) and human (bottom panels) HSPCs [pooled data from −/− and +/+ HSPCs are shown in (C), top panel]. (D) The number of colonies retrieved from clonogenic assays (CFC) performed at the end of transduction with the indicated LV on murine −/− (top panel) and human (bottom panel) HSPCs is reported. (E) TUNEL assay was performed 2 days after transduction with the indicated LV on murine −/− HSPCs (top panel) and CD34+ cells from normal donors’ CB and BM (bottom panel). The frequency of apoptosis among transduced cells was assessed (% TUNEL+ cells). Each dot represents an individual sample (B to E). In (E), eight or more fields and ≥100 cells were counted for each sample. *P < 0.05; **P < 0.01; ***P < 0.001. (F) Galc−/− or Galc+/+ murine HSPCs were transduced with the indicated vectors and intravenously transplanted into Trs mice according to the experimental scheme. Average survival ± SD and average engraftment of the transduced cells, measured as percent of GFP+ cells or VCN detected in the BM of transplanted mice (± SD) at 120 days or at death, are shown (n = 4 to 26 per group). (§) Similar results were obtained with +/+ murine HSPCs. The survival of untreated GLD mice is shown in the first row (*, not irradiated; n.a., not applicable). (G) Human primary monocytes, B and T lymphocytes, and murine microglia were transduced with the indicated vectors. GALC activity [expressed as fold to untransduced cells (UT); UT cell activity being equal to 1] was measured on the cultured cells ≥14 days after transduction (middle panel) and TUNEL assay was performed 2 days after transduction (right panel). Data from GALC-transduced murine (m) and human (h) HSPCs [from (B)] and % TUNEL+ cells in GFP-transduced microglia are reported for comparison. Each dot represents an individual sample, in which six or more fields and ≥250 cells were counted.

miR-126 regulation rescues HSPCs from GALC expression toxicity and enables successful gene therapy of GLD

The selective toxicity of de novo GALC expression in HSPCs highlights the need to tightly regulate transgene expression in HSPCs for successful gene therapy of GLD. We thus tested the efficacy of our miR-126–based regulatory system and compared it to a transcriptional strategy based on use of the myeloid-specific CD11b promoter to target GALC expression to the differentiated HSPC progeny. Both strategies rescued the transduced HSPCs from GALC-induced toxicity (see Fig. 5, B to E). However, the reconstituted GALC activity was substantially higher (up to twice as much as the wild-type level) in the progeny of the cells transduced with GALC-126T lentivector (in which GALC expression is driven by the PGK promoter) than in CD11b-GALC–transduced cells, even when cultures transduced to similar vector copy number were compared (see Fig. 5, B and C). We also verified that the GALC-126T lentiviral vector effectively protected human HSPCs from GALC toxicity (see Fig. 5, C and D). Given the likely benefit from supraphysiological enzyme activity in mature hematopoietic cells, we selected the miRNA-regulated vector for in vivo studies of GLD therapy.

HSPCs from Trs mice were transduced with the GALC-126T lentiviral vector and transplanted into newborn Trs mice. The transplanted mice were successfully engrafted (see Fig. 5F) and showed a statistically significant longer survival with respect not only to the untreated Trs mice (P < 0.0001) but also to the mice transplanted with wild-type GFP-transduced HSPCs (P = 0.0004) (Fig. 6A). When we stratified gene therapy–treated mice according to vector copy number measured in the BM, animals with the highest vector content showed a significantly longer survival (Fig. 6B). Indeed, partial reconstitution of GALC activity was observed not only in the liver, but also in the brain, of gene therapy–treated Trs mice (Fig. 6C and fig. S10). Notably, vector copy number measured in the BM significantly correlated with GALC activity measured in the brain (Fig. 6D). When gene therapy–treated animals were stratified according to GALC-specific activity measured in the brain, mice with the highest enzymatic activity showed a significantly longer survival (Fig. 6E). These observations indicate that supraphysiological expression of GALC in hematopoietic cells allows efficient enzyme delivery to the affected nervous system and augments the therapeutic efficacy of HSC transplantation.

Fig. 6

Therapeutic efficacy of HSC gene therapy in GLD mice. (A) Kaplan-Meier survival curves of Trs mice transplanted with either Galc−/− HSPCs transduced by GALC-126T LV (n = 39) or Galc+/+ HSPCs transduced by GFP LV (+/+GFP, n = 10) and of untreated affected controls (UT) (n = 15). Untreated Trs mice survived on average 50 ± 6 days (maximal survival, 59 days); GALC-126 mice survived on average 88 ± 40 days (maximal survival, 226 days); +/+GFP mice survived on average 63 ± 14 days (maximal survival, 90 days). Log-rank test for pairwise comparison of median survival: GALC-126T versus +/+GFP, P = 0.0004; GALC-126 versus UT, P < 0.0001; +/+GFP versus UT, P > 0.05. Comparison of overall survival by one-way ANOVA with Bonferroni post test: GALC-126T versus UT, P < 0.001; UT versus +/+GFP, P > 0.05; GALC-126T versus +/+GFP, P < 0.05. (B) GALC-126T–transplanted mice were divided into two groups according to the VCN measured on total BM cells at the time of death (only animals from which adequate samples were collected, allowing for molecular analysis, are shown). Survival is shown for animals with fewer (mean ± SD = 67 ± 13 days) or greater (mean ± SD = 117 ± 43 days) than five LV copies in BM cells, five being the average VCN measured in the BM of the entire population of treated mice (**P < 0.01, Student’s t test). (C) GALC activity measured on chromatographic peak fractions (see fig. S10) isolated from the liver and brain of gene therapy–treated and control Trs and wild-type (WT) mice. Partial reconstitution of GALC activity was obtained in GALC-126T mice. (D) Positive significant Pearson correlation between VCN measured on BM and GALC activity measured in the brain of gene therapy–treated mice (P = 0.001). (E) Survival of mice having GALC activity in the brain below (mean ± SD = 71 ± 17 days) or above (mean ± SD = 121 ± 37 days) 3 nmol/hour mg, 3 nmol/hour mg being the average GALC activity measured in the brain of the entire cohort, was significantly different (**P < 0.01, Student’s t test) (only mice from which adequate samples were collected, allowing for biochemical analysis, are shown). (F) GALC activity (light blue) was qualitatively evaluated on brain sections from GALC-126T–transplanted and control mice. Representative sections from the hippocampus (top row) and pons (lower row) of GALC-126T–transplanted mice, compared to WT and untreated Trs mice, are shown. GALC assay was coupled to the myeloid Iba1 marker (upper panels) and to the hematopoietic marker CD45 (lower panels). A marked infiltration of activated hematopoietic cells is present in untreated and GALC-126T–transplanted Trs mice, the latter showing GALC activity both within the transduced HSPC hematopoietic progeny [arrowheads showing co-staining of GALC (in blue) and Iba1 (upper panel) and CD45 (lower panel) in brown] and in Iba1, CD45 nonhematopoietic cells. Magnifications, ×20 and ×40 (insets). (G) Lectin staining on representative brain sections [pons and cerebellum (Cb)] from WT, untreated Trs, and gene therapy–treated mice (sections from two different mice are shown). A substantial reduction of the lectin+ storage is evident in the treated mice. Cell nuclei are stained with crystal violet. Magnification, ×10. (H) Quantification of the lectin+ area on tissue sections from the Cb and pons of representative untreated Trs and gene therapy–treated mice (n = 3 to 5 sections from each of two representative mice per group). **P < 0.01, Student’s t test. (I) Immunofluorescence staining for the astrocyte marker glial fibrillary acidic protein (GFAP) on representative brain sections (pons and Cb) from WT, untreated Trs, and gene therapy–treated mice (sections from two different representative mice are shown). Reduction of reactive astrocytosis is observed in GALC-126T–transplanted mice compared to untreated Trs mice. Magnification, ×20. (J) Quantification of the GFAP+ area on tissue sections from the Cb and pons of WT, untreated Trs, and gene therapy–treated mice (n = 3 to 5 sections from each of two representative mice per group). **P < 0.01, one-way ANOVA with Bonferroni post test. (K) Representative toluidine blue–stained semithin sections of the sciatic nerve from untreated Trs, WT (in the inset), and gene therapy–treated mice (sections from two different mice are shown in each case). In untreated Trs, numerous demyelinated fibers (arrowheads) and degenerated axons (arrows) are visible. No degenerated axons and a substantial number of myelinated axons were noticed in gene therapy–treated mice. Magnification, ×100. All the data presented in (F) to (K) were obtained on end-stage, moribund Trs mice (40 to 50 days of age) on long-lived (>150 days) gene therapy–treated mice at sacrifice/death and on wild-type, age-matched controls (n ≥ 3 mice per group, with five or more sections per mouse analyzed at each staining).

We previously demonstrated, after the transplantation of gene-modified HSPCs, a progressive infiltration of myeloid lineage cells into the brain of mice affected by leukodystrophies or other lysosomal storage disorders (7, 20, 21). These HSPC-derived myeloid lineage cells can serve as a local source of the functional enzyme for tissue metabolic correction (20). In the central nervous system (CNS) of the gene therapy–treated mice examined here, we detected GALC activity both within Iba1+, CD45+ infiltrating hematopoietic myeloid cells and within Iba1, CD45 nonhematopoietic cells (Fig. 6F), demonstrating cross-correction of resident cells, likely due to enzyme secretion by the progeny of the transplanted and transduced HSPCs.

To determine whether the reconstitution of GALC activity in the brain of treated mice would result in prevention or reversal of lysosomal storage, we visualized globoid cells containing undegraded sphingolipids by lectin histochemistry. Although abundant storage material was present in the brain of untreated Trs mice, little storage material was detected in the corresponding regions of the brain of long-lived gene therapy–treated mice (Fig. 6, G and H). Moreover, visualization of reactive astrocytosis in the brain of treated and control animals showed a substantial reduction of the astrocyte-associated glial fibrillary acidic protein (GFAP) signal in Trs mice transplanted with GALC-126T–transduced HSPCs when compared to untreated controls (Fig. 6, I and J).

Although Trs mice do not exhibit overt demyelination in the CNS, they manifest profound myelin damage in the peripheral nervous system (Fig. 6K). The sciatic nerves from long-lived, gene therapy–treated mice demonstrated a reduction of demyelination, prevention of axonal degeneration, and better preserved architecture of the nerve when compared to untreated affected controls (fiber density: gene therapy–treated mice = 2136 ± 521 fibers/mm2; untreated Trs mice = 1285 ± 221 fibers/mm2; demyelinated fiber density: gene therapy–treated mice = 142 ± 25 fibers/mm2; untreated Trs mice = 269 ± 32 fibers/mm2; calculated on four representative sections from three representative mice per group) (Fig. 6K).

Reconstitution of enzymatic activity and increased survival were associated with a markedly improved phenotype of the treated mice when they were compared to untreated affected controls, with preserved walking abilities and reduced twitching (GLD-associated intentional tremors) (videos S1 and S2).


We have identified a subset of miRNAs that are strongly expressed in HSPCs and down-regulated in the differentiated progeny cells that arise from HSPCs. Various studies have used expression profiling to associate several miRNAs with HSPCs (2225), although most have been performed on functionally heterogeneous cell populations, making it impossible to directly establish expression in defined HSPC subsets. In addition, the detection of an miRNA within a cell population does not prove its functionality, because expression above a threshold concentration might be required to render an miRNA proficient at controlling target expression (14). The reporter BdLVs that we used in this work directly measure miRNA activity in individual cells, providing a reproducible, biologically relevant, quantitative readout of miRNA function that permits longitudinal tracking of the expression of a selected miRNA across hierarchical cell populations at single-cell resolution within live cells. Our measure of miRNA activity yielded assay values consistent with miRNA expression levels measured by Q-PCR (26, 27).

We identified miR-126, miR-130a, and miR-196b as miRNAs with the strongest activity in cell fractions enriched for HSCs by using stringent immunophenotyping. This expression gradient along early hematopoietic differentiation stages of several unrelated miRNAs suggests that these miRNAs may be involved in the maintenance of the undifferentiated state. Homeobox (HOX) genes have been shown to regulate HSC function (28). The genes for miR-196b and miR-10a are embedded within HOX clusters (29), and miR-10a, miR-130a, and miR-126 repress several HOX genes (27, 30, 31). Our finding that these specific miRNAs are expressed in functional HSCs warrants further investigation of how the HOX-miRNA network regulates HSC biology.

The differential expression of miRNAs in various HSPCs and mature cell types, as revealed by our data, provides a way to identify and isolate specific cell types without requiring knowledge of cell surface marker expression. The high miR-126 expression in the most primitive human HSPC populations was key to being able to prospectively isolate HSCs out of long-term culture of CB cells without the use of cell surface markers. This is important because markers such as CD38 are not reliable for isolating human HSCs after culture (32, 33). Our approach serves as an important new tool for human HSC research and also for future HSC-based clinical therapies that require ex vivo culture, including HSC expansion and gene therapy.

The HSPC-specific expression pattern of the miRNAs we describe reveals significant potential for improved vector design in HSC gene therapy. Lentiviral vectors could now be designed that robustly suppress transgene expression in human HSCs and early progenitors while permitting strong expression in the differentiated progeny. Our miR-126T sequence is a good candidate for this purpose, because it mediates the strongest repression of the miRNAs examined in human HSPCs and provides a broader coverage of multipotent progenitors than other miRNAs. Combinations of different miRT sequences, such as 130aT and 126T, might further improve the ratio between repression in HSPCs and expression in differentiated cells (figs. S2 to S4). As a proof of principle, the potential of 126T to protect HSCs from detrimental transgene expression was provided by the rescue of TK-transduced LT-HSC from GCV-induced suicide in vivo.

Saturation of miRNA function might occur upon miRT overexpression, as a function of dose, sequence design, and other factors (34). However, we had no indication that our PGK-driven 126T vector interfered with endogenous miR-126 function, because we did not observe a functional deficit of HSPCs in competitive transplantation, nor did we phenocopy the miR-126 knockout (16) upon transgenic delivery of the vector. These findings indicate that the lentiviral vector described in this study operates within a safe therapeutic window for miR-126 regulation.

The development of a successful preclinical gene therapy model for GLD demonstrates the importance of precisely targeting expression only to the most appropriate cell types. The aim of HSC transplantation in lysosomal disorders is to repopulate affected tissues with macrophages and microglia that express and secrete a functional hydrolase (6). The generation of enzyme-overexpressing macrophage and microglia cells by means of HSC gene transfer may allow a more rapid and improved level of enzymatic reconstitution in the affected tissues, as shown here for GLD and elsewhere for metachromatic leukodystrophy and mucopolysaccharidosis type I (7, 20, 21). Toward this goal, efficient transduction of HSPCs is essential to ensure long-term repopulation of gene-corrected cells, and transgene overexpression in the progeny of transduced HSC is also required. The GALC enzyme contributes to the maintenance of a functional hematopoietic stem cell niche (35). Therefore, de novo GALC expression is not tolerated in HSPC. We were able to exploit miR-126 to protect HSPCs from the toxicity of de novo GALC expression and allow its robust expression in their progeny.

Transplantation of gene-corrected HSPCs into newborn GLD mice resulted in the reconstitution of enzyme activity in their livers and brains, formally demonstrating that down-regulation of GALC expression by miR-126 does not occur in differentiated hematopoietic cells. Reconstitution of GALC activity in the affected nervous system was associated with an amelioration of the overall disease phenotype, with reduction of storage and neuroinflammation in the brain, prevention of severe peripheral demyelination, and improvement of survival in gene therapy–treated mice as compared to untreated affected controls. Survival of gene therapy–treated mice was increased, even in comparison with mice transplanted with HSPCs from wild-type donors. Thus, our data, particularly the correlation between vector dose and survival of the reconstituted mice, indicate that enforced enzyme expression in mature hematopoietic cells augments the efficacy of HSC transplantation in lysosomal storage disorders.

In summary, our findings provide a detailed profile of miRNA expression in murine and human HSPCs at a single-cell level. In addition, these data provide the basis for the prospective isolation of human HSCs with an miRNA as functional biomarker. Finally, we demonstrated a gene therapy approach that not only may provide GLD patients with a new treatment option but also has the potential to augment the safety and efficacy of HSC-based gene therapy for other diseases.

Materials and Methods

Lentiviral vector production and titration

Lentiviral transfer constructs are shown in fig. S1. miRT sequences were designed as previously described (13, 14).

Mouse strains

C57BL/6 and C57BL/6 Ly45.1 mice were purchased from Charles Rivers Laboratories. Tg.126T mice were generated in a C57BL/6 background with lentiviral technology as described in the Supplementary Material. FVB-Twitcher (FVB/Twi) mice were generated in our animal research facility by breeding heterozygous Twi (+/−) C57BL6 mice (Jackson Laboratory) with wild-type (+/+) FVB mice. Trs mice were a gift of D. Wenger and P. Luzi. NOD/LtSz-scid/scid [nonobese diabetic (NOD)–SCID] mice were bred from breeding pairs originally obtained from the Jackson Laboratory and maintained in the defined flora animal facility located at the Ontario Cancer Institute (OCI). All animals were handled under sterile conditions and maintained under microisolators. Procedures were performed according to the protocols approved by the Animal Care and Use Committee of the Fondazione San Raffaele del Monte Tabor (Institutional Animal Care and Use Committee 324, 335, 325) or the Animal Care Committee of OCI.

Isolation and transduction of murine HSPC and HSC transplantation

Young adult mice (5 to 8 weeks) were killed with CO2, and BM was harvested by flushing the femurs and tibias. Murine HSPCs were purified and transduced with LV and transplanted by tail vein injection as described (7, 34). See Supplementary Material for additional details on culture, transduction, transplantation, colony-forming cell assay, and TK suicide experiments.

Isolation, transduction, and transplantation of human cells

Human HSPCs were isolated from CB or BM, obtained according to procedures approved by the ethical committee of the San Raffaele Institute (TIGET01). For xenotransplantation, human CB samples were obtained from umbilical tissues according to procedures approved by the institutional review board of the University Health Network. See Supplementary Material for additional technical details on HSPC enrichment, culture, transduction, and transplantation, as well as isolation and transduction of lymphocytes and monocytes.

Flow cytometric analysis (FACS)

A complete list of antibodies and staining procedures is available in the Supplementary Material. Samples were analyzed on a flow cytometer equipped with three lasers (Canto II or LSRII, BD Biosciences). miRNA activity was calculated as described in the Supplementary Material. miR-126high and miR-126low CB cells were sorted on a MoFlo cytometer (DAKO).

Determination of GALC activity, immunohistochemistry, immunofluorescence, and histopathology

A detailed description of determination of GALC activity, immunohistochemistry, immunofluorescence, and histopathology is available in the Supplementary Material.

QPCR for the determination of vector copy number and miRNA expression in human CB is described in the Supplementary Material.

Statistical analysis

Statistical analyses were made by one-way analysis of variance (ANOVA) for repeated measurements with a Bonferroni post test correction. In the in vivo TK experiments, chimerism (chim) values were transformed into a log odds scale {lochim = log [% chim/(100 − % chim)]}, given that percentages are by definition constrained between 0 and 100. Statistical analysis was performed by two-way ANOVA (variable 1: GCV treatment; variable 2: time after transplant). For pairwise comparisons, an unpaired Student’s t test or Mann-Whitney test (95% confidence interval) was used.


  • * These authors contributed equally to this work.

  • These authors share senior authorship.

  • Citation: B. Gentner, I. Visigalli, H. Hiramatsu, E. Lechman, S. Ungari, A. Giustacchini, G. Schira, M. Amendola, A. Quattrini, S. Martino, A. Orlacchio, J. E. Dick, A. Biffi, L. Naldini, Identification of Hematopoietic Stem Cell–Specific miRNAs Enables Gene Therapy of Globoid Cell Leukodystrophy. Sci. Transl. Med. 2, 58ra84 (2010).

Supplementary Material

Materials and Methods

Fig. S1. Lentiviral vector maps.

Fig. S2. Activity of all tested miRNAs in murine HSPCs.

Fig. S3. Activity of all tested miRNAs in murine myeloid precursors.

Fig. S4. Activity of HSPC-specific miRNAs in murine differentiated cells.

Fig. S5. Characterization of the miR-126–regulated vector in transgenic mice.

Fig. S6. Destabilizing thymidine kinase.

Fig. S7. TK/GCV efficiently ablates long-term engrafting cells.

Fig. S8. Loss of miR-126 activity in human cord blood cells during granulocyte differentiation.

Fig. S9. Specificity of BdLV-126T for miR-126.

Fig. S10. GALC enzyme in tissues from control and treated mice.

Video 1. Walking abilities of a representative untreated Trs mouse.

Video 2. Walking abilities of a representative gene therapy–treated Trs mouse.


References and Notes

  1. Acknowledgments: We thank S. Rivest for providing reagents, A. Wenger and P. Luzi for Trs mice, R. Tiribuzi for help in testing GALC activity, F. Benedicenti and L. Sergi Sergi for technical help, and M. Neri and F. Cerri for help with histopathology. We thank B. Brown for helpful discussion. Funding: This work was supported by Telethon (TIGET grant to L.N. and A.B.); EU (FP5LSHB-CT-2004-005242 CONSERT, FP7 GA 222878 PERSIST, and ERC Advanced Grant 249845 TARGETINGGENETHERAPY); Fondazione CARIPLO (NOBEL); the Italian Ministry of Health to L.N.; and the National Tay-Sachs and Allied Disease Association, EU (FP7 LEUKOTREAT 241622), and the ELA Foundation (2009-00515) to A.B. Additional support to J.E.D. came from The Stem Cell Network of Canada National Centres of Excellence, the Canadian Cancer Society and the Terry Fox Foundation, Genome Canada through the Ontario Genomics Institute, Ontario Institute for Cancer Research with funds from the province of Ontario, the Leukemia and Lymphoma Society, the Canadian Institutes for Health Research, and a Canada Research Chair. Author contributions: B.G. designed and performed the research (vector design and mouse and some of the human miRNA studies), analyzed the data, and wrote the paper. I.V. performed the research and analyzed the data related to the GLD model. H.H. and E.L. designed and performed the research (human CB cells), analyzed the data, and revised the paper. S.U., A.G., and G.S. performed the research. M.A. provided crucial reagents. A.Q. performed the research (histology on the GLD mouse model). S.M. and A.O. performed and analyzed the research (biochemistry on the GLD model). J.E.D. coordinated the project, interpreted the data, and revised the paper. A.B. coordinated the project related to the GLD model, analyzed and interpreted the data, and wrote the paper. L.N. coordinated the project, interpreted the data, and wrote the paper. Competing interests: A.B., J.E.D., B.G., H.H., E.L., and L.N. have filed a patent for the use of miR-126–regulated vectors for HSPC gene therapy based on work reported in this article. The other authors declare no conflicts of interest.
View Abstract

Navigate This Article