Rescue of ATPa3-deficient murine malignant osteopetrosis by hematopoietic stem cell transplantation in utero

Abstract

Autosomal recessive osteopetrosis (ARO) is a paradigm for genetic diseases that cause severe, often irreversible, defects before birth. In ARO, osteoclasts cannot remove mineralized cartilage, bone marrow is severely reduced, and bone cannot be remodeled for growth. More than 50% of the patients show defects in the osteoclastic vacuolar-proton-pump subunit, ATP6a3. We treated ATP6a3-deficient mice by in utero heterologous hematopoietic stem cell (HSC) transplant from outbred GFP transgenic mice. Dramatic phenotype rescue by GFP osteoclasts was obtained with engraftment, which was observed in most cases. Engraftment survived for variable periods. Recipients were not immunosuppressed, and graft-versus-host disease was not observed in all pups born after in utero treatment. Thus, differentiation of unmatched HSC transplanted in utero is sufficient to prevent fatal defects in ARO and may prevent complications of ARO unresponsive to conventional bone marrow transplantation. The presence of defective cells is not a barrier to the rescue of the phenotype by donor HSC.

The potential of cell therapy with hematopoietic stem cells (HSC) is clear, although outcomes have often been unfavorable. Despite the risk, transplant is often the only practical approach to fatal genetic defects. Osteopetrosis was one of the first to which bone marrow transplantation (BMT) was applied (1). Nevertheless, this therapeutic approach cannot rescue irreversible damage (impaired vision and growth retardation) already present at birth. Approaches to improve outcomes include gene therapy or in utero transplantation. Gene therapy involves complex technology that is not yet mature. In contrast, in utero transplant (IUT) of HSC produced good results in a few humans with primary immunodeficiencies (2–6), as recently reviewed in refs. 7–9.

Several factors have limited progress of in utero transplantation. Prenatal diagnosis of rare diseases is limited to families with affected members, although screening may be feasible in the future (9). Positive pregnancies are often terminated, although effective therapy may change this practice. In utero BMT has been considered more dangerous than postnatal BMT, although recent studies in humans indicate that complications, including graft-versus-host disease (GVHD), rarely occur in utero (8). In addition, ablation of host cells cannot be used during pregnancy, so it is unclear whether donor cells can engraft in the presence of abundant host cells. Indeed, trials of in utero BMT for hemoglobinopathies have not shown good results in humans, ascribed to the abundant host red cell progenitors (8, 9). On the other hand, in utero BMT has many potential advantages. Studies in mice suggest that tolerance is easily achieved, important in humans, where matched donors are not always available (10). Tolerance to in utero HSC also would allow BMT to be performed postnatally, and microchimerism should allow a second graft without immune suppression or marrow ablation (11, 12). Most importantly, in many severe diseases, organ damage is present at birth, and full reversal is thus impossible after birth (13).

Autosomal recessive osteopetrosis (ARO) is a genetic disease that has severe effects potentially treatable in utero. Mineralized cartilage and bone cannot be degraded. Complete defects block the formation of bone marrow, leading to hepatosplenomegaly and severe anemia. Blindness and deafness usually result due to no remodeling of cranial foramina, and inability to resorb bone leads to macrocephaly and severely impaired growth in the axial and appendicular skeleton. In about half of cases, the defect is in the TCIRG1 gene (previously called ATP6i), coding for the ATP6a3 subunit of the osteoclast vacuolar-ATPase, which is essential for dissolving bone mineral (14–17). The only available treatment is BMT, which has had only limited success, because neither the growth impairment nor the cranial nerve defects are reversed. In many cases, failed engraftment or GVHD lead to fatal outcomes (18–22).

Additional considerations make ARO an excellent candidate to test in utero BMT. First, mature osteoclasts are multinucleated, possibly allowing a few fusing cells to restore function, as recently shown in other systems (23–25). Thus, bone resorption may be improved by even a low number of engrafted donor cells whose progeny fuse with defective osteoclasts. Second, osteoclasts originate from HSC, which are the only stem cells that have successfully been tested in humans. We therefore concluded that ARO is a favorable test situation for the in utero BMT approach in diseases with severe symptoms already present at birth. In addition, unlike syndromes such as severe combined immunodeficiency (SCID), where entire cell types are absent, TCIRG1-dependent ARO typically has normal or increased osteoclasts, but these osteoclasts are not functional, precluding, according to one theory, successful HSC engraftment (the “niche” is full). ARO could be useful to test to what extent this assumption is valid. We studied this phenomenon using the oc mouse, whose underlying Tcirg1 defect and phenotype are essentially identical to those in TCIRG1-dependent ARO. Outbred mice transgenic for GFP were BM donors, modeling non-HLA-matched donors and allowing engrafted cells to be identified. Our results suggest that in utero BMT in ARO patients may greatly improve and even normalize their severe clinical picture.

Materials and Methods

Mice and Transplantation in Utero. Two pairs of (C57BL/6JxB6C3Fe-a/aF1) oc^+/- mice were from The Jackson Laboratory. CMV/GFP CD-1 transgenic mice, with the GFP gene controlled by the CMV promoter/enhancer, were the kind gift of Dr. Masaru Okabe (Osaka University, Osaka) (26). Mice were maintained in accordance with Italian Ministry of Health guidelines. A final suspension of 10⁶ cells per μl obtained from bone marrow flushed from femora of 2-month-old CMV/GFP CD-1 male mice was used for intrauterine transplantation. Cells (5 × 10⁶ in 5 μl) were injected into each embryo visualized through the uterine wall. This number of injected cells ensures a significant percentage of HSC. Pregnant recipient oc^+/- females crossed with oc^+/- male mice were anesthetized with Avertin at 14.5 days postcoitum. Genomic DNA was obtained from tail biopsies. Heterozygotes (oc^+/-), homozygotes (oc^-/-) and wild-type (wt) animals were identified by PCR amplification and Southern blot (see Fig. 1 b and c).

Fig. 1.

PCR and Southern blot strategy for the identification of oc^-/- mice. (a) Schematic representation (not in scale) of the 5′ region of the mouse Tcirg1 gene. In the wt genome, an EcoRI site is located in intron 1 but is absent in the oc^-/- mouse because of a 1,579-bp genomic deletion that includes part of intron 1, exon 2, intron 2, and a portion of exon 3. Arrows represent the primers (OCwtF and OCwtR) flanking the deletion, whereas arrowheads indicate primers (OCmutF and OCmutR) internal to the deletion (see Materials and Methods for further details). The dashed bar indicates the DNA segment used as a probe. This probe identifies a 4.7-kb band in the wt and a 7.0-kb band in the mutant (bottom line). The numbered boxes indicate exons, and the lines indicate introns. (b) PCR amplification products obtained from wt (WT), heterozygous mutant (oc^+/-), homozygous mutant (oc^-/-), GFP-transgenic wt (WT^*). C- is a negative control. Marker (M), 1-kb ladder. (c) Southern blot analysis of the Tcirg1 locus in oc^-/-, oc^+/-, and wt animals. M1–5 are the in utero-transplanted oc^-/- mice that survived for more than 4 weeks (see text for details). c is a composite of two blots done with identical controls and matched size markers.

In Vitro Osteoclast Differentiation and Resorption. Mononuclear cells were isolated from spleens of control, affected, and treated animals. For osteoclast differentiation, cells were plated at 2 × 10⁴ in murine CSF-1 (25 ng/ml) and receptor activator of NF-κB ligand (RANKL) (30 ng/ml), with bone fragments or dentine slices for 7–21 days, as specified in Results. Phalloidin labeling and immune labeling for GFP and ATP6a3 were as described in ref. 27. Primary antibody for ATP6a3 was raised in rabbit, the kind gift of Jan Mattsson (Astra Hässle, Mölndal, Sweden), and was used at 1:1,500. Cells for in situ enzyme activity were fixed in 40% acetone in citrate at pH 5.4. Tartrate-resistant acid phosphatase was determined by using naphthol AS-BI phosphate substrate by fast garnet GBC coupling (Sigma) to produce red precipitate in 0.67 M tartrate, pH 5.6. Acid lakes produced by osteoclasts on bone were identified by using the fluorescent polyamine-pyrrole Lysotracker red DND-99 (Molecular Probes). Lysotracker was added at 3 μM 15–25 min before fluorescent analysis with a 560-nm low-pass excitation filter and a 600-nm emission barrier. GFP was analyzed by using excitation at 450–490 nm with a 510-nm dichroic filter and a 520-nm barrier filter.

Colony Assays in Vitro. The culture conditions were as described in refs. 28 and 29. Briefly, bone marrow cells were flushed from femora and tibiae in Iscove's modified Dulbecco's media (IMDM) with 2% FCS and cultured at 10⁵ cells per ml in IMDM containing α-thioglycerol, methyl cellulose, BSA, 5% FCS, iron-saturated transferrin, lecithin, oleic acid, cholesterol, 2 units/ml human recombinant erythropoietin, 10 ng/ml murine stem cell factor, and 10 ng/ml murine IL-3. Cultures were scored at 2, 7, and 12 days with an inverted fluorescent microscope by using standard criteria.

Bone Matrix Degradation. Collagen type I fragments generated during osteoclastic bone resorption were detected by ELISA (Rat-Laps, Nordic Bioscience Diagnostics, Herlev, Denmark). The assay was performed on mouse serum from wt, oc^+/-, and oc^-/-, both untreated and treated, as indicated by the manufacturer.

Results

Survival and Genotype After Shotgun Peritoneal Transplant of Heterozygote Crosses. Thirty-eight females, injected in utero at day 14.5 postcoitum, delivered 111 pups that survived >1 week that were subsequently monitored day by day. Mice homozygous for the Tcirg1 mutation could be recognized at 10 days by failure of eruption of the incisors, clubbed feet, and circling behavior. They gained weight normally for about the first 10 days of life, then lost weight and died at about 3 weeks (30). Genotypes of all of the pups were determined at 2 weeks by PCR using tail DNA (Fig. 1 a and b). From 14 pups, only the 563-bp oc-specific PCR product was amplified with primers flanking the deletion; these mice were identified as homozygous oc^-/-. Five of the 14 mice were alive after 4 weeks and were analyzed by Southern blot to confirm the genotype (Fig. 1 a and c). The deletion in the oc allele eliminates an EcoRI site, transforming a wt 4.7-kb band in a 7.0-kb band. Accordingly, the oc^-/- mice, both those transplanted and untransplanted, showed only the 7.0-kb band, whereas the normal band was easily seen in both the wt and heterozygous mice. This demonstrated that all of the five mice surviving more than 4 weeks were oc mutants. These five mice suggested transplant response of 35% (5 of 14). This low percentage of success is due, at least in part, to technical difficulties in performing IUT in mice. Because the IUT approach is technically more feasible in humans, a higher rate of success could be expected in ARO patients.

Mouse 1 (female) died at 3 months; the mouse was smaller than normal or oc^+/- littermates. Mouse 2 (male) survived 6 months and died during anesthesia for x-ray examination; it was smaller than its littermates, although it was able to explore its surrounding environment. Mouse 3, a female of normal size and weight, lived an apparently normal life and was mated with mouse 4. As expected, this produced homozygous oc^-/- pups. The mouse was killed for analysis at 6 months. Mouse 4 was a male of normal size and weight, indistinguishable from its wt or heterozygous littermates, and killed for analysis at 6 months of age. Mouse 4 was mated several times with wt and heterozygous females and once with mouse 3. All of the progeny were compatible with mouse 4's oc^-/- genotype. The mating results confirmed mouse 4's mutant status and that, as expected, there was no colonization of the germline by donor cells. Mouse 5, a female, although smaller than normal, did not show any obvious defects and was killed at 7 months of age for examination, while in good health. Tooth eruption was normal in all these five mice (Fig. 2 Inset). In summary, of the five mice surviving more than 4 weeks, three can be considered as asymptomatic (mice 3, 4, and 5), whereas the quality of life of the other two was improved.

Fig. 2.

X-ray analysis of whole skeletons in three mice treated with in utero BMT and controls. From left to right: M5 is the oc^-/- treated mutant female mouse 5; ♀WT is a female normal mouse; M4 is the oc^-/- treated mutant male mouse 4; M2 is the oc^-/- treated mutant male mouse 2; ♂WT is a male normal mouse. The images are from two different radiographs, using identical exposures and image sizes and placed in a composite for comparison. (Inset) Normal dentition in M4 is shown.

Skeletal Phenotype by X-Ray and Histology. X-ray examination of the whole skeleton was performed on mice 2, 4, and 5. The results are shown in Fig. 2. The bone structure of mouse 4 was indistinguishable from normal or heterozygous mice, because no signs of osteopetrosis were visible. The x-ray of mouse 5 showed large areas of normal bone structure, although in some segments, signs of the underlying disease were visible. On the contrary, mouse 2 clearly showed the osteopetrotic phenotype, although it survived 6 months.

Histological examination of the surviving oc^-/- animals confirmed that significant bone degradation had occurred, with formation of marrow space and significant remodeling in the vertebrae and femora (Fig. 3). In addition, viable osteoclasts and significant remodeling of the jaw (not illustrated) were observed, in keeping with the tooth-eruption result (Fig. 2 Inset). The untreated oc^-/- mice showed diffuse extramedullary hematopoiesis, including hematopoiesis adjacent to the vertebral bodies and often also had rachitic changes [Fig. 3 (oc-/-)]; in contrast, the transplanted long-term survivals had variable amounts of true marrow that correlated with the animal growth and behavior [Fig. 3 (mice 2, 3, and 4)] and was, apparently, completely normal in the animals with normal phenotype, despite oc genotype.

Fig. 3.

Histology of femora and lumbar vertebrae in wt, oc^-/-, and three transplanted animals. (Upper) Shown are photographs from the spinous process or body of L4 vertebrae. (Lower) Shown are the growth plate of the distal femur. Note that, in the wt, articular and growth plates are narrow, resorption of mineralized cartilage is rapid and complete, and there is abundant BM. In untreated oc^-/- mice, there was adaptive response, with some extramedullary haematopoiesis (EMH) adjacent to the vertebral body at the border of the spinal canal, and, in some of these animals, there was a rachitic-like expansion of the growth plate (RGP); Upper and Lower images of oc^-/- are from different animals, selected to show these features. In long-term surviving animals, there was always actual intramedullary BM, whose extent was the feature best correlated to growth and normal survival (compare with M2 and M4 in Fig. 2). In three (mouse 1, mouse 2, and mouse 3) of the five, marrow was seen mainly in the central portion of vertebral bodies or femoral canal (not illustrated), indicating that the graft functionally had been lost at an earlier stage but that there was adequate remodeling to allow survival [compare with osteopetrotic bone (OPB) (Lower) of mouse M2]. In mouse M4, the pattern is indistinguishable from normal, in keeping with the normal phenotype and survival (see text).

Bone Matrix Degradation Analysis. Type I collagen represents more than 90% of the organic matrix of bone. During skeleton renewal, the matrix is degraded, and collagen fragments are released. The concentrations of these peptides in serum or urine are a faithful index of bone remodeling by osteoclasts. Serum samples from wt, in utero-treated wt, heterozygous oc^+/-, and oc^-/-mice and from mice 4 and 5 were collected, and the amount of the C-terminal telopeptide (a fragment of type I collagen) was determined by ELISA. The levels of this peptide were low, as expected, in the osteopetrotic strain oc^-/- but were in the normal range in heterozygous or wt animals. Interestingly, the levels detected in mice 4 and 5 were comparable with those of wt and heterozygous mice (see Fig. 4).

Fig. 4.

Bone matrix degradation analysis in normal, osteopetrotic, and treated mice. The height of the histograms indicates the amount of degradation activity present in the serum from various mouse groups and from two treated mice (M4 and M5). n, number of mice analyzed; wt, untreated wt mice; wt IUT, wt mice born after in utero BMT; oc^+/- IUT, heterozygous mice treated with in utero BMT; M5, treated mouse 5; M4, treated mouse 4.

In Vitro Clonogenic Assay of Hematopoietic Progenitors. To test whether donor bone marrow cells, derived from mice constitutively expressing GFP and transplanted in utero, permanently engrafted in the recipient animals, we tested hematopoietic progenitors in clonogenic assays. Mouse 4 and mouse 5 were analyzed at 6 and 7 months of age, respectively, to assess long-term hematopoietic reconstitution. Cells isolated from bone marrow of in utero transplanted wt and oc^-/- mice were cultured in semisolid media containing a mixture of cytokines to support growth and differentiation of different types of progenitors. Results showed that the number of bone marrow cells was normal, with virtually normal frequencies of erythroid and myelomonocytic progenitors. Furthermore, in mouse 4, a significant percentage of the different types of colonies were fluorescent, an average of 40%, similar to the fluorescent fraction in the constitutively GFP-expressing donor cells (see Table 1). No fluorescent colony was detected in mouse 5.

Table 1. Hematopoietic progenitors of wt, oc^-/-, and in utero transplanted oc^-/- mice.

Mouse	BM cells (× 106)	CFU-E	BFU-E	Total myeloid	CFU-mix	Fluorescence (% of GFP+ colonies)
wt GFP+	17.3	26,988	4,844	8,131	4,152	35
oc–/–	0.03	0	1	4	1	0
M4	22	13,020	5,830	27,940	7,370	37

Frequency of early (BFU-E) and late (CFU-E) erythroid progenitors, pluripotent (CFU-mix), and myelo-monocytic (total myeloid) precursors, as determined by clonogenic assays. The numbers of bone marrow nucleated cells and CFU are indicated per femur.

Characterization of Normal and Mutant Macrophages and Osteoclasts. To demonstrate definitively that durable engraftment of HSCs had occurred in long-term surviving animals, spleen cells from mouse 4 were compared with those of GFP donor animals and oc^-/- animals in vitro. Cells of all genotypes appeared identical by phase-contrast microscopy (not illustrated) and plated with similar efficiency. Differentiation in CSF-1 and RANKL showed that oc^-/-, wt (GFP donor), and transplanted (mouse 4) animals all formed TRAP-expressing polykaryons (Fig. 5a). In oc^-/- cells on plastic, there was a vacuolar pattern (arrowheads), suggestive of a transport or structural anomaly, that was not seen in wt cells or cells from the transplanted mouse 4. In keeping with a subtle structural anomaly in the ATP6a3-deficient cells, phalloidin labeling for F-actin (Fig. 5b) showed anomalous actin aggregates in oc^-/- cells; oc^-/- cells also had less prominent actin rings than in wt or transplanted cells from mouse 4. To verify functional GFP-expressing cells in transplanted animals, lysotracker and GFP analysis was also performed on bone slices (Fig. 5c), which showed GFP-expressing cells in mouse 4 that produced acid lakes, areas of extracellular acid secretion maintained during bone degradation by osteoclasts (31). In keeping with this finding, membrane labeling for ATP6a3 was seen in osteoclasts from mouse 4, but no labeling was present in an oc^-/- animal (Fig. 5d). The membrane distribution of the ATP6a3 is typical of the distribution of V-ATPase in osteoclasts (27).

Fig. 5.

Properties of oc^-/-, wt, and transplanted monocytes and osteoclasts in vitro and in vivo.(a) Differentiation of spleen cells in the presence of CSF-1 and RANKL showed that oc^-/-, wt (GFP donor), and GFP-transplanted mouse 4 were all capable of producing TRAP-expressing osteoclasts (arrows). The only indication of abnormality in oc^-/-cells on plastic was an odd vacuolar pattern (arrowheads) not found in wt or transplanted cells from animal M4. Panels are 220 μm square. (b) Phalloidin labeling for attachment pattern in these cells showed anomalous intracellular actin aggregates in oc^-/-cells (arrowheads), which also showed less prominent actin rings than in wt and transplanted cells from animal M4 (arrows, Center and Right). Panels are 110 μm square. (c) Spleen monocytes plated on bone slices in CSF-1 and RANKL for 21 days showed a lysosomal pattern only when oc^-/- cells were used, as shown by vacuolar distribution of lysotracker labeling (red) (Left). Cells from the wt GFP transgenic donor showed both vacuolar distribution and occasional acid lakes (arrow), large areas of acid secretion formed during bone degradation by osteoclasts, with the cells also demonstrating GFP (green). In the GFP transplant, sparse GFP-labeled cells that occasionally also formed acid lakes are shown (arrow). Each field is 220 μm square. (d) Immune labeling for ATP6a3 was uniformly negative in sections of the untransplanted oc^-/-animal. In contrast, some giant cells from the transplanted animal had membrane labeling consistent with the normal distribution of the V-ATPase in osteoclasts. Each field is 40 μm square.

Discussion

This work shows that oc^-/- mice can be rescued by in utero BMT. Our aim was to use an experimental system recapitulating the human phenotype of a severe genetic disease, whose symptoms are present at birth, to show that damage can be prevented with a therapeutical protocol that may be reproduced in humans. The underlying mouse defect in the oc^-/- mouse is identical to that found in >50% of human ARO (14–16, 30, 32, 33). We used allogeneic BMT with unselected donors. This approach would apply to any human fetus affected by ARO without an HLA-matched donor, one of the most important factor in predicting the outcome of postnatal BMT (10).

The small size of mouse embryos makes the transfer of HSC to the umbilical vein or fetal liver technically difficult. The low percentage of success obtained (35%) could be improved in humans, where umbilical vein injection is possible, and the technical feasibility is higher. Nevertheless, our results demonstrate that the quality of life and the survival of in utero transplanted oc^-/- mice are greatly improved or even normalized by this strategy. In one mouse (mouse 4), rescue of the phenotype was complete, as assessed by growth, ability to breed, and the presence of donor cells for the normal lifespan of the animal; two other animals (mice 3 and 5) were clinically asymptomatic and had normal social behavior, although both x-ray and histological examinations showed osteopetrotic signs; however, these animals were 6 months old. These results demonstrate that transient donor engraftment may be sufficient for clinical improvement. It is noteworthy that this improvement is achieved without any additional therapy. In humans, severe vision and growth impairment remain after postnatal BMT, although treated with many supportive measures, suggesting that the prenatal period is critical for therapy. This result is in agreement with the finding that, even in postnatally treated, HLA-matched ARO children, the earlier the transplant, the better the outcome (22). Although we did not explore this possibility in this study, administration of cells from the same donor could be performed postnatally in humans, and this technique might be successful without ablation (11, 12) because of microchimerism and tolerance from the in utero transplant (34, 35).

The current paradigm suggests that the good results usually obtained in SCID mice and humans by IUT are due to a proliferative or survival advantage of donor cells that could engraft and differentiate toward the lymphocyte lineage because of the complete lack of the corresponding endogenous cells. On the other hand, the exact nature of this “advantage” and the concept of empty niche is not completely clear. Although this paradigm is widely accepted, it is not clear why, in a SCID mouse, exogenously provided hematological stem cells should engraft and replace the endogenous ones, because there is no reason why, in SCID, endogenous HSCs should not have already occupied all of the available niche sites, possibly suggesting that the presence of a “lineage-specific empty space” results in high levels of growth and differentiation factors specific for the lymphoid lineage and that these factors could be responsible for driving the differentiation of donor HSCs toward the lymphoid lineage. Similarly, osteoclast-lineage-stimulating factors are likely to be present at high levels in the oc^-/- mice and ARO patients, because, in TCIRG1-dependent patients and mice, normal or elevated numbers of nonfunctional osteoclasts are usually described (14, 36). This finding raises the possibility that the dramatic rescue found in our mice could be because of the presence of proliferative/differentiating cues acting on a few stem cells, with the effect that a number of progeny cells can differentiate into osteoclasts. With this hypothesis, the concept of empty space should be considered not only in strictly anatomical but also in functional terms.

An additional component facilitating the phenotype rescue in oc^-/- mice could be that, in the last steps of their differentiation, mononuclear osteoclasts fuse to give rise to multinuclear osteoclasts (37). Fusion between donor and host cells has been shown in the mouse and is of therapeutic interest (23–25). The fact that fusion is an obligatory step in osteoclast maturation could have contributed to our results, because two or more functioning alleles could be present in multinucleated chimeric osteoclasts. Although the level of ATP6a3 molecules needed to permit resorption is not known, it is clear that a 50% activity is sufficient for a normal phenotype: heterozygous mice and humans are unaffected (14, 30, 33, 36). Lower levels could still allow for a substantial degree of resorption. However, we were unable, in this work, to determine to what extent this peculiar osteoclast behavior contributed to the results obtained.

Clinical considerations also include the advantage provided by in utero BMT in the tolerance to donor cells. This approach is of great interest in humans, because BMT outcome in inherited hematological diseases is strictly related to availability of matched donors (10, 22), both for engraftment and to reduce the risk of GVHD. In our model we did not observe any GVHD event in the analyzed mice, although we cannot exclude that this complication occurred during pregnancy, leading to in utero death.

An additional critical point is the time at which BMT is performed. So far, most in utero transplants in humans have been performed early in the second trimester. In the mouse, most protocols require HSC injection at 13.5/14.5 days of gestation, an age at which HSCs are thought to migrate from the liver to the BM. We believe that timing is critical to achieve stable engraftment and that IUT in osteopetrotic humans should be done at a corresponding time, possible because an early prenatal diagnosis can easily be performed in ARO (16).

In conclusion, we have tested an approach to the prevention of ARO defects that may be directly applicable to humans. Survival of affected embryos without matched donors has been achieved in a murine defect identical to that found in human ARO. In addition, our results challenge, in part, the concept of empty niche and raise the possibility that other diseases in which irreversible defects are present at birth could be treated by stem cell transplantation in utero. Diseases amenable to HSC injection are prime candidates. In this regard, mucopolysaccaridosis and other diseases whose defective cells originate from the bone marrow could be tested. In addition, stem cells from other adult tissues have recently been obtained, and it may eventually be possible to devise novel approaches to deliver these stem cells to the affected tissues during fetal life.