Research ArticleLysosomal Storage Diseases

Tolerance induction and microglial engraftment after fetal therapy without conditioning in mice with mucopolysaccharidosis type VII

See allHide authors and affiliations

Science Translational Medicine  26 Feb 2020:
Vol. 12, Issue 532, eaay8980
DOI: 10.1126/scitranslmed.aay8980

When treating at birth is too late

Mucopolysaccharidosis type VII (MPS7) is a rare and severe lysosomal storage disorder, which causes dysfunction of multiple organs including the brain and may be associated with undiagnosed cases of fetal death. By the time of birth, the organ damage may already be severe and the fetus may not survive at all. Thus, the prenatal period provides the most promising opportunity for intervention. Nguyen et al. assessed two prenatal approaches, in utero enzyme replacement therapy and in utero hematopoietic stem cell transplantation, and demonstrated the potential of these treatments to improve survival and functional outcomes in a mouse model of MPS7.


Mucopolysaccharidosis type VII (MPS7) is a lysosomal storage disorder (LSD) resulting from mutations in the β-glucuronidase gene, leading to multiorgan dysfunction and fetal demise. While postnatal enzyme replacement therapy (ERT) and hematopoietic stem cell transplantation have resulted in some phenotypic improvements, prenatal treatment might take advantage of a unique developmental window to penetrate the blood-brain barrier or induce tolerance to the missing protein, addressing two important shortcomings of postnatal therapy for multiple LSDs. We performed in utero ERT (IUERT) at E14.5 in MPS7 mice and improved survival of affected mice to birth. IUERT penetrated brain microglia, whereas postnatal administration did not, and neurological testing (after IUERT plus postnatal administration) showed decreased microglial inflammation and improved grip strength in treated mice. IUERT prevented antienzyme antibody development even after multiple repeated postnatal challenges. To test a more durable treatment strategy, we performed in utero hematopoietic stem cell transplantation (IUHCT) using congenic CX3C chemokine receptor 1–green fluorescent protein (CX3CR1-GFP) mice as donors, such that donor-derived microglia are identified by GFP expression. In wild-type recipients, hematopoietic chimerism resulted in microglial engraftment throughout the brain without irradiation or conditioning; the transcriptomes of donor and host microglia were similar. IUHCT in MPS7 mice enabled cross-correction of liver Kupffer cells and improved phenotype in multiple tissues. Engrafted microglia were seen in chimeric mice, with decreased inflammation near donor microglia. These results suggest that fetal therapy with IUERT and/or IUHCT could overcome the shortcomings of current treatment strategies to improve phenotype in MPS7 and other LSDs.


Mucopolysaccharidosis type VII (MPS7) is caused by a deficiency in β-glucuronidase, resulting in the accumulation of glycosaminoglycans (GAGs) in many tissues (1). The clinical sequelae of this accumulation are variable, from hydrops fetalis [ultrasound findings signifying fetal distress and impending demise; (2)] to multiorgan dysfunction including cardiomyopathy, pulmonary dysfunction, skeletal dysplasias, hepatosplenomegaly, and neurocognitive impairment (3). Although MPS7 is considered an ultrarare disease, its true incidence in the fetus is likely higher, because it is the most common genetic defect found in fetuses with idiopathic hydrops, most of whom do not survive to birth (4).

Current treatment options for patients with MPS7 and other similar lysosomal storage disorders (LSDs) include enzyme replacement therapy (ERT) (when available) and hematopoietic stem cell (HSC) transplantation (5). The recent approval of recombinant human β-glucuronidase (rhGUS) is an important clinical development for patients with MPS7. A recent clinical trial of rhGUS in pediatric patients with MPS7 was promising, with improvement in urinary GAGs and other clinical measures (6). However, because many patients with MPS7 do not survive to birth, an in utero treatment strategy is needed.

Although ERT can improve outcomes for LSDs in general, several of its limitations could be addressed with fetal therapy. First, postnatal ERT can be limited by the development of antibodies against the exogenous enzyme, which the host immune system can see as “foreign” (7). Some patients may even require immunosuppression to tolerate the therapy (8). However, introduction of an exogenous protein during fetal development can induce tolerance in multiple settings (913). Second, because the recombinant enzyme does not cross the blood–brain barrier, some therapies have to be given intrathecally (14) and fail to adequately correct the neurologic disorder; in utero administration may penetrate brain microglia before the blood-brain barrier forms. Last, because the organ dysfunction in many LSDs is cumulative, with symptoms such as cardiomyopathy starting in utero, there is a good rationale to develop in utero therapies for many LSDs even if they do not lead to fetal demise.

HSC transplantation (HSCT) could be curative for many LSDs because donor-derived leukocytes can be a permanent source of enzyme (15), but carries a high risk of peritransplant morbidity, graft failure, and graft-versus-host disease (16). Moreover, HSCT is often not effective in correcting neurological disease, because the extent of microglial engraftment is low (17) and HSCT needs to be performed early in life to ameliorate neurological outcomes (18). In utero HSCT (IUHCT) can address several of these limitations because it circumvents the need to give immunosuppression, and mixed chimerism can be achieved with high-dose transplantation without any host conditioning (19). In the mouse model of in utero transplantation, IUHCT of high-dose allogeneic HSCs results in clinically relevant amounts of engraftment (11) and leads to donor-specific tolerance via deletion of donor-reactive T cells (10). Clinically relevant amounts of chimerism and donor-specific tolerance can be achieved after transplantation of high-dose maternal HSCs in fetal dogs without conditioning (20). Furthermore, because even low-engrafted chimeric animals are tolerant to donor antigens, boosting after birth with minimal irradiation can achieve higher amounts of engraftment (13). Although there is extensive experience with IUHCT with multiple animal models, a detailed analysis of microglial engraftment after IUHCT is lacking, likely because the field of in utero transplantation has focused on correcting hematologic disorders.

Microglia are the most important cell type to correct in addressing the neurological component of MPS7, because they are the natural storehouse of GUSb in the brain (21). As such, they are the critical cell type for reconstitution after stem cell transplantation for metabolic disorders (22). Emerging evidence suggests that microglial progenitors may have a distinct ontogeny from other myeloid cells and that they populate the brain during embryonic development (23), so that achieving chimerism in this compartment before birth may be critical for achieving adequate microglial engraftment. This strategy may also take advantage of a recently described second wave of microglial engraftment derived from precursors in the fetal liver (24). Because the blood-brain barrier in mice is permeable midgestation at embryonic day 13 (E13) to E14 (25) and the developmental cues that enable microglial migration to the brain are present before birth (23, 24), we hypothesized that IUHCT without conditioning can engraft murine brains with microglia without any irradiation or conditioning.

Here, we tested two complementary strategies, ERT and HSCT, for in utero therapy of mice with MPS7. Our results suggest that starting ERT or HSCT in the fetal period may offer an important advantage for improving clinical outcomes for patients with LSDs.


In utero ERT improves survival of MPS7−/− mice and results in systemic distribution of rhGUS

The mouse model of the disease (GUSmps/mps, hereafter referred to as MPS7−/−) (26, 27) phenocopies many aspects of the clinical syndrome. We bred heterozygous MPS7+/− mice [because MPS7−/− mice are sterile; (26)] and administered rhGUS (20 mg/kg) to all fetuses in the litter on E14.5 using our established intrahepatic injection technique (Fig. 1A) (28); control mice received vehicle or phosphate-buffered saline (PBS) injections. We chose this higher dose for the in utero injection after pilot experiments comparing the usual postnatal dose of 4 to 20 mg/kg (fig. S1). Because the half-life of the enzyme in most organs is about 100 hours (29), repeated booster injections were given postnatally starting at 3 weeks of life and every 2 weeks thereafter (Fig. 1A), similar to the clinical protocol for patients. Analysis of the overall survival of the litter showed increased survival to birth and survival to weaning in treated mice compared to controls (Fig. 1, B and C). We next analyzed the genotype composition of surviving mice, expecting 25% MPS7−/− pups in each litter with this breeding scheme. In uninjected litters, only 11.5% of MPS7−/− mice survived to weaning, suggesting either fetal loss or early neonatal demise at baseline with this strain. However, IUERT (but not PBS injection) resulted in a significant increase in the survival of MPS7−/− pups, up to 27% (P = 0.002) (Fig. 1D). Thus, IUERT specifically restores the survival of affected MPS−/− mice in the litter.

Fig. 1 In utero enzyme replacement improves overall survival to weaning in MPS7 mice, and activity is enriched in monocytes.

(A) Heterozygous MPS+/− mice were mated, and their fetuses underwent intrahepatic injection at E14.5 of either PBS control or recombinant human β-glucuronidase (rhGUS). Surviving mice underwent intravenous booster injections every other week starting at 3 weeks until harvest. Mice underwent behavioral testing at 6 weeks and at harvest, at which point they also underwent biochemical enzyme activity testing, histology, and computed tomography (CT) imaging. (B) Overall survival to birth (percentage of injected) in MPS7 or B6 litters (B6, n = 141; IUPBS, n = 121; IUERT, n = 240; ****P < 0.001, χ2 test). (C) Survival to weaning (percentage of injected) in MPS7 litters after injection of PBS (n = 121) or rhGUS (IUERT, n = 240; **P < 0.01, χ2 test). (D) Survival to weaning graphed as deviation from expected 25% survival in individual MPS7−/− pups within the litter (total mice at wean: uninjected, n = 236; PBS, n = 27; IUERT, n = 86; *P < 0.05 and ****P < 0.0001, Fisher’s exact test). (E and F) Peripheral blood was taken at multiple time points after systemic injection of rhGUS in adult mice, and GUS activity in individual leukocyte populations (representative flow cytometry histogram at day 3 after injection) (E) was measured over time (F). n ≥ 3 per group and time point. FITC, fluorescein isothiocyanate.

We next tested the tissue enzyme activity in multiple organs 4 to 7 days after in utero injection of rhGUS. We found that untreated MPS7−/− mice had undetectable rhGUS activity in homogenates harvested from each organ, whereas MPS+/− mice had decreased enzyme activity compared to MPS7+/+ littermates in each tissue (fig. S2). After IUERT, there was widespread distribution in many tissues, and GUS activity was restored up to or above wild-type rates in heart, lung, and liver, whereas increases were smaller in spleen, kidney, and bone (fig. S2).

To examine the cellular distribution and pharmacokinetics of rhGUS in blood leukocytes after systemic injection, we used a flow cytometry assay that detects the fluorescent product of a cleaved intracellular β-glucuronidase substrate (Fig. 1E) (30). Enzyme activity rates (defined as percentage of cells with detectable enzyme activity) in peripheral blood peaked at days 1 to 3 and then declined to low rates 14 days after injection (Fig. 1F), validating our every-other-week enzyme administration protocol. The highest activity of the GUS enzyme was within monocytes and B cells, whereas GUS activity was lower in neutrophils and T cells (Fig. 1F).

IUERT penetrates brain microglia, and continued treatment improves brain inflammation and grip strength

To determine whether IUERT can deliver enzyme to the brain, we analyzed brain enzyme activity at early time points after injection. GUS activity was detectable in homogenates of brain tissue at 4 and 7 days after IUERT (albeit at lower rates than in wild-type and heterozygous mice) but was not detectable in the adult MPS7−/− brains harvested 4 to 7 days after a postnatal boost (Fig. 2A).

Fig. 2 In utero ERT results in detectable enzyme within the microglia of treated mice, but postnatal therapy does not.

(A) Biochemical activity of enzyme in the brain 4 to 7 days after in utero injection or 4 to 7 days after adult injection. n ≥ 5 per group. Data are means ± SEM. ****P < 0.0001 (ANOVA with Tukey’s multiple comparisons test). (B) Representative flow cytometry plots of GUS staining in microglia and nonmicroglia 4 days after in utero injection. Graphs showing (C) percentages of cells that are positive for GUS and (D) MFI of GUS in individual cells. n ≥ 3 per group. Data are means ± SEM. *P < 0.05 and **P < 0.01 (Kruskal-Wallis with Dunn’s multiple comparisons test). MG, microglia. (E) Representative images of CD68 immunohistochemical staining (scale bars, 200 μm). (F) Integrated density graph of CD68 staining in mice harvested at 8 to 10 weeks. n = 3 per group. Data are means ± SEM. *P < 0.05 and ***P < 0.001 (ANOVA with Tukey’s multiple comparisons test). (G) Grip strength in unaffected and MPS7−/− mice with and without treatment. n ≥ 5 per group. Data are means ± SEM. *P < 0.05 (ANOVA with Tukey’s multiple comparisons test).

Microglia, the main storehouse of GUS in the brain, have critical roles in both central nervous system homeostasis and disease physiology, making the restoration of GUS in these cells critical to alleviating the neurological deficits in MPS7 (22, 31). To determine whether IUERT penetrates brain microglia, we harvested mice 4 to 7 days after IUERT and used flow cytometry to quantify GUS activity in individual microglia (defined as CD11b+CD45lo) (Fig. 2B). We detected GUS activity in microglia of treated MPS7−/− animals but not in untreated mutants (Fig. 2C). GUS activity rates in individual microglia [measured as the mean fluorescence intensity (MFI)] of GUS (Fig. 2D) were very similar in treated MPS7−/− animals compared to healthy controls. GUS uptake was seen preferentially in microglia compared to nonmicroglia (Fig. 2, C and D), consistent with the known expression of mannose 6-phosphate receptors on these cells (21).

To determine whether delivery of rhGUS to microglia has a functional effect on ameliorating neural inflammation typically seen over the life of the mice (32), we next stained brains of adult MPS7−/− animals for CD68, a marker of microglial activation (33). CD68 staining was diffuse in the brains of untreated MPS7−/− mice, but staining intensity was significantly reduced in MPS7−/− mice that received ERT in utero plus postnatally (P = 0.02) (Fig. 2, E and F). These findings suggest that ERT decreases microglia activation.

Last, to assess possible functional improvement in the mice secondary to enzyme replacement, we performed a detailed neurologic evaluation that determined grip strength (using a force-sensing bar) and basic movement (using the open-field assay) (34, 35) at age 2 months; the investigators were blinded to the experimental group. We found that ERT significantly improved grip strength of MPS−/− mice compared to untreated mice (P = 0.04) (Fig. 2G), in some cases up to wild-type results. Enzyme treatment did not significantly affect basic movement (fig. S3A) or rearing (fig. S3B).

ERT improves phenotype in multiple organs

We next allowed cohorts of MPS7−/− mice (untreated or treated with IUERT plus postnatal boosts) to mature to age 8 weeks, at which time we harvested multiple organs to examine the sequelae of GAG accumulation within lysosomes, which are readily visible with periodic acid Schiff (PAS) staining (7, 29, 3638). Although untreated MPS7−/− mice had readily detectable areas of lysosomal accumulation by histology, enzyme replacement decreased cellular accumulations to the amounts found in wild-type mice in all tissues examined (Fig. 3A).

Fig. 3 Combination of in utero and postnatal ERT improves pathologic lysosomal accumulations of GAGs as well as bone length.

Mice were harvested at 8 weeks of age, and their livers, spleens, and kidneys were examined with a PAS stain. (A) Representative images from the liver, spleen, and kidney (scale bars, 20 μm) demonstrating intracellular accumulations (vacuolated cells, solid arrows) of GAGs. Quantification is shown to the right of the histology images; n ≥ 5 per group. Data are means ± SEM. ****P < 0.0001 (ANOVA with Tukey’s multiple comparisons test). (B) Representative CT images of femurs in unaffected, MPS7−/− untreated, and MPS7−/−-treated mice harvested at 8 to 10 weeks of age. Compiled data for (C) femurs and (D) tibias. n ≥ 5 per group. *P < 0.05, ***P < 0.001, and ****P < 0.0001 (ANOVA with Tukey’s multiple comparisons test).

Correction of bone defects is especially important clinically for many patients with LSDs. We used micro-CT (micro–computed tomography) to measure the effect of ERT on the skeletal dysplasia seen in this mouse strain (26) and observed that untreated MPS7−/− mice have shorter femurs and tibias than unaffected animals, whereas in treated MPS7−/− mice, femur and tibia lengths increased significantly at 8 weeks of age (femur, P = 0.0007; tibia, P = 0.01), although they were not corrected up to wild-type lengths (Fig. 3, B to D).

In utero enzyme replacement induces tolerance to rhGUS enzyme

In utero exposure to a foreign antigen can potentially induce tolerance in numerous settings (913), whereas repeated exposure to recombinant enzymes can result in sensitization (8). Given the relative lack of homology between the human and mouse isoforms of the enzyme (39), we reasoned that even unaffected pups may mount an immune response to the human isoform and thus tested MPS+/− and wild-type mice for the presence of antibodies against rhGUS. This design was necessary given the poor survival of MPS7−/− pups without IUERT. After in utero injection of PBS or rhGUS, each mouse received a postnatal boost with rhGUS (4 mg/kg) every 2 weeks starting at 3 weeks of age, followed by immunization with 50 μg of rhGUS plus complete Freund’s adjuvant (CFA) at 6 weeks of age (Fig. 4A). We then measured amounts of immunoglobulin G (IgG) antibodies against the rhGUS enzyme, focusing on IgG1, which comprises the majority of mouse circulating antibody, and IgG3, which is also a robust trigger of an effector response (40).

Fig. 4 In utero ERT results in tolerance to rhGUS.

(A) Mice underwent in utero injection with either rhGUS enzyme or PBS followed by postnatal boosting starting at 3 weeks (arrows) and continuing every other week. At 6 weeks of age, mice underwent intraperitoneal (IP) injection of rhGUS with complete Freund’s adjuvant (CFA). Plasma concentrations of antibodies against rhGUS were measured by ELISA at 8 weeks. (B) Amounts of IgG1 (left graph) and IgG3 (right graph) antibodies against rhGUS. n ≥ 10 per group. Data are means ± SEM. *P < 0.05, **P < 0.01, and ****P < 0.0001 (Kruskal-Wallis with Dunn’s multiple comparisons test).

The IgG1 response after CFA immunization was robust in naïve animals but was slightly attenuated in mice exposed to rhGUS starting at age 3 weeks, demonstrating some ability to develop tolerance in the relatively immature immune environment of the neonatal mouse (Fig. 4B). However, there was a notable lack of IgG1 response in animals exposed to rhGUS in utero compared to both other groups. Results for IgG3 were similar to those of IgG1 (Fig. 4B). IgG2b and IgG2c testing revealed that when compared to untreated mice, tolerance was induced in both postnatally treated and in utero treated mice (fig. S4, A and B).

IUHCT results in multilineage blood chimerism and engraftment of donor HSCs in bone marrow

Although IUERT has multiple benefits in this model, the short half-life of the enzyme necessitates repeat postnatal injections, which do not penetrate the brain. We reasoned that IUHCT could be more durable and potentially take advantage of developmental events to seed donor-derived microglia in the brain. Given the poor survival of fetal MPS7 mice, we transplanted both wild-type and MPS7 mice using CX3CR1–green fluorescent protein (GFP) mice as donors so that microglia could be identified on the basis of GFP expression (41). MPS7 litters received a 20 mg/kg dose of rhGUS per fetus along with the fetal HSC transplant to improve survival to birth. We transplanted 2.5 × 106 to 5 × 106 fetal liver mononuclear cells from CD45.1.CX3CR1-GFP donor mice into B6.CD45.2 congenic recipient fetal mice on E13.5 to E14.5, injecting directly into the fetal liver per our standard protocol (Fig. 5A) (28). We examined rates of blood chimerism starting at 3 weeks and harvested some mice after 4 to 5 weeks to examine HSC chimerism (Ckit+Sca1+Lin “KLS” cells) in the bone marrow (Fig. 5B). As expected, we detected multilineage chimerism in blood (with T cells, B cells, and myeloid cells) (Fig. 5C), as we have previously published with this model of IUHCT (11). We also found engraftment of donor-derived KLS cells in bone marrow, with excellent correlation (r2 = 0.96, linear regression) between KLS chimerism in the marrow and CD45 chimerism in the blood, consistent with our previous experience of stem cell engraftment with this approach (Fig. 5D). Engraftment rates did not differ significantly between wild-type mice and MPS7 mice (fig. S5A), but survival was lower in the MPS7 litters compared to wild-type litters (fig. S5B). However, the survival of MPS7−/− pups within the litter was comparable between IUERT and IUHCT (fig. S5C).

Fig. 5 In utero hematopoietic cell transplantation resulted in multilineage blood chimerism and bone marrow HSC engraftment.

(A) Experimental design. (B) Flow cytometry gating strategy for blood and bone marrow. (C) Peripheral blood flow cytometry showing composition of CD45+ cells in donor and host compartments (n ≥ 27 per group). Data are means ± SEM. *P < 0.05 and **P < 0.01 (Mann-Whitney test). (D) Bone marrow flow cytometry showing donor cell correlation between bone marrow KLS and blood CD45 cells (n = 15).

IUHCT in MPS7 mice results in cellular cross-correction and histologic improvement

One benefit of HSCT for MPS7 (and related diseases) is that transplanted cells can cross-correct other cells that can take up the enzyme, a process mediated by mannose 6-phosphate receptors (7). We used flow cytometry to quantify cross-correction in chimeric and nonchimeric MPS7−/− mice harvested at 4 weeks of age, focusing on liver Kupffer cells and blood monocytes, which should have the highest amounts of mannose 6-phosphate receptors (42). In Kupffer cells, the percentage of cells positive for GUS activity, as well as the MFI of GUS in individual host cells, was significantly improved (percentage, P = 0.01; MFI, P = 0.04). Rates of hematopoietic chimerism correlated positively with the percentage of GUS+ cells (r2 = 0.91) and the MFI of GUS in individual cells (r2 = 0.62) (Fig. 6, A and B). There was also modest cross-correction in peripheral blood monocytes, with the percentage of host cells positive for GUS activity increasing in a chimerism-dependent manner in MPS7−/− chimeras compared to untreated controls (fig. S6). Consistent with the cross-correction, the accumulation of GAG decreased in liver and spleen but not in the kidney (Fig. 6, C and D). The reduction of storage correlated with the extent of donor cell chimerism (Fig. 6E). The amounts of chimerism needed to achieve correction were low: Even 2% chimeras had decreased lysosomal storage in their livers and spleens (Fig. 6E).

Fig. 6 IUHCT in MPS7 mice resulted in multilineage chimerism in bone and blood and some evidence of long-term tissue cross-correction.

(A) Gating strategy to identify host liver Kupffer cells (left) and quantify GUS expression by Kupffer cells (right); GUS expression by host cells is secondary cross-correction by donor cells. (B) Compiled data for the percentage of host Kupffer cells expressing GUS and its relationship to chimerism (top) and the mean fluorescence intensity (MFI) of GUS in Kupffer cells and its relationship to chimerism (bottom) (n ≥ 4 per group). Data are means ± SEM. *P < 0.05 and **P < 0.01 (Mann-Whitney test). (C) Tissue histology representative images from liver, kidney, and spleen and (D) area of lysosomal GAG accumulation measured in wild-type (WT) controls, MPS7−/− chimeras, and untreated MPS7−/− controls (scale bars, 20 μm). n ≥ 4 per group. Data are means ± SEM. ****P < 0.0001 (Mann-Whitney test). (E) Correlation between the area of lysosomal GAG accumulation and peripheral blood chimerism rate in liver, kidney, and spleen (n ≥ 11 per graph, Spearman correlation test P values shown).

Donor-derived microglia engraft throughout the brains of hematopoietic chimeras after IUHCT

We next used immunofluorescence to detect donor-derived GFP+ cells in the brains of 4-week-old wild-type mice that had undergone IUHCT after perfusing the mice to limit contamination by circulating monocytes (which can also express CX3CR1). Mouse brains were sectioned coronally (Fig. 7A). We found that donor cells had engrafted throughout the brain, as detected by multiple foci of GFP+ cells with ramified microglial morphology; these cells costained with the microglial marker Iba1 (Fig. 7B). These engrafted cells were generally found along cortical regions throughout the brain, but donor cells were also seen in deeper structures such as the hippocampus (Fig. 7B).

Fig. 7 Donor-derived microglia were found in the brain after IUHCT.

(A) Representative brain slices examined for microglial engraftment. (B) Confocal imaging revealed clusters of microglial engraftment in chimeric animals. (C) Correlation between brain chimerism and peripheral blood chimerism (n = 14). PCA, principal components analysis. Bulk RNA sequencing of donor versus host-derived microglia and donor-derived monocytes isolated from brains of chimeric animals 5 weeks after IUHCT, shown as (D) principal components analysis, (E) heatmap of top 1000 genes by variance, and (F) heatmap of “signature” microglial genes (n ≥ 7 per group).

We then compared the transcriptomes of engrafted donor microglia against endogenous host microglia and against circulating donor monocytes using bulk RNA sequencing after sorting each population using congenic markers and GFP (fig. S7A). The numbers of donor-derived microglia correlated positively with hematopoietic chimerism in the blood (Fig. 7C). Principal components analysis (Fig. 7D) and T-distributed stochastic neighbor embedding (fig. S7B) plots showed that engrafted microglia overlapped with endogenous microglia and that donor circulating monocytes were separate from both populations (Fig. 7D). Among the 23,688 unique transcripts detected, 22,793 (96%) had a less than twofold difference between donor-derived and host microglia.

Previous experiments comparing donor-derived microglia after postnatal transplantation defined several key microglial “signature” genes, some of which were differentially expressed by donor-derived microglia after postnatal HSCT (43). However, we noted that donor-derived microglia after IUHCT had similar rates of expression (less than twofold difference) in microglia signature genes such as Hexb, Tgfbr1, Tgfbr2, Mef2c, Tardbp, Kcnk13, Fcrls, Tmem119, Sall1, SiglecH, CD34, P2ry12, and Slc2a5. Only Sall3 had reduced expression compared to expression in endogenous microglia (Fig. 7, E and F). Thus, donor-derived cells engrafting in the brain after IUHCT are highly similar to host-derived microglia.

Evidence of decreased inflammation near engrafted donor cells in the brains of chimeric MPS7−/− mice

To determine whether IUHCT can rescue brain inflammation in engrafted MPS7−/− mice, we quantified CD68 staining in host cells, measuring regions of the brain with varying numbers of engrafted cells and regions of the brain without noticeable engraftment. Donor microglia were defined as GFP+ cells; host microglia were defined as Iba1+ cells without GFP signal. We noted decreased inflammation (as measured by CD68 staining) in regions with high numbers of engrafted donor cells (Fig. 8A). Using Iba1 staining, we also found that compared to areas without engraftment, areas with many donor cells present had host microglia with decreased cell density and diameter (Fig. 8B), which are other indicators of inflammation (44). We found that the extent of donor cell engraftment had a negative correlation with markers of inflammation such as host cell CD68 intensity (Fig. 8C), host cell diameter (Fig. 8D), and host cell density (Fig. 8E). However, the threshold of chimerism needed to reduce brain inflammation was higher than that needed for the liver, because the most robust results were seen with a mouse with 36% hematopoietic chimerism, whereas CD68 expression was not attenuated in mice with lower percentages of engraftment (Fig. 8F). These data provide proof of concept that microglial engraftment enabled by a single in utero transplant could improve both systemic and brain phenotypes in mice with MPS7.

Fig. 8 Immunofluorescence staining of MPS−/− brain sections revealed evidence of decreased inflammation near engrafted donor cells.

(A) Representative image showing CD68 staining (microglial inflammation) in relation to areas of engraftment with donor cells (GFP+). (B) Representative images showing GFP+ donor cells, Iba1 staining (microglia), and CD68 staining (microglial activation) (scale bars, 50 μm). Subsequent graphs represent (C) CD68 intensity per host cell, (D) host cell diameter, and (E) host cell number compared to number of donor cells per confocal image (n ≥ 7 images per graph). (F) CD68 expression (raw integrated density) per host cell graphed against number of donor cells, combining all chimeric MPS7 brains.


Here, we describe two complementary strategies for prenatal therapy of MPS7 that may rapidly be translated into the clinic. The potential benefits of fetal therapy, including increased survival, tolerance induction, and amelioration of disease phenotype, particularly for neurologic manifestations, are applicable to patients with MPS7 and multiple related LSDs.

The notion of fetal therapy for LSDs is feasible in terms of both prenatal diagnosis and fetal access. Because most LSDs are autosomal recessive, at-risk families can be identified by carrier screening, and prenatal diagnosis can readily be performed either genetically or with functional testing of amniocytes. Because MPS7 can be fatal in utero, offering prenatal therapy is especially compelling for this disease and similar LSDs that result in fetal demise (4, 45). Both IUERT and IUHCT can be given systemically by accessing the umbilical vein with ultrasound guidance, a routine procedure used for fetal blood transfusions. Once the proof of principle is established for a fatal disease such as MPS7, patients with other LSDs such as MPS1, X-ALD, or Krabbe disease (46) (diseases with neurological manifestations that could be rescued with stem cell transplantation) could also be treated. Although there is currently no routine fetal screening for these diseases (because prenatal diagnosis would not change management), having a viable in utero treatment option may change the equation for when testing may be an ethical option.

Nearly half of fetuses with MPS7 present with hydrops (47) and the disease can therefore be fatal before birth. The mouse model recapitulates this aspect of the disease, in that the survival of MPS7−/− littermates is lower than the expected Mendelian distribution. We found that IUERT improves the survival of MPS7−/− mice within the litter. The reason for the in utero mortality in this mouse strain is not known. It has been suggested that endogenous GAG fragments may activate TLR4 (Toll-like receptor 4) (48), which could trigger fetal resorption, as reported in other settings (49). In humans, the mechanisms that result in fetal hydrops in patients with MPS7 are also not known, because many affected fetuses do not survive. Nonetheless, the improved survival with IUERT up to the expected Mendelian ratio in mice suggests that in utero therapy may have a similar salutary effect in patients with hydrops secondary to MPS7. In cases of fetal hydrops secondary to anemia, for example, treatment with in utero transfusion improves survival (50).

Arguably the most appealing aspect of IUERT is the possibility of inducing tolerance to the therapeutic enzyme, because neutralizing antibodies against the recombinant enzymes have been shown to affect treatment efficacy or lead to anaphylactic reactions in some settings (51). Antidrug antibodies have been found in human MPS7 ERT trials as well (52). In this study, we demonstrate that mice receiving IUERT develop fewer antibodies (IgG1 and IgG3) against the recombinant human enzyme than mice that receive the enzyme only postnatally. These findings are consistent with other studies of in utero therapy leading to tolerance induction (913) and represent a potential avenue to improve the current available therapies for numerous mucopolysaccharidoses.

Two previous attempts to use IUHCT in mice with MPS7 did not achieve enough engraftment to treat neurologic disease. In one study, Barker et al. (53) bred MPS7 mice with a C-kit deficiency to promote better engraftment of donor cells and demonstrated engraftment of hematopoietic cells but little engraftment of microglia. The authors concluded that “toxic myeloablation appears necessary for donor cells and enzyme to cross the blood-brain barrier.” In the second study, Casal et al. (54) transplanted MPS7 mice with allogeneic cells that overexpress MPS7 by injecting into the placenta. The donor cell chimerism in their experiments was 0.1%, but there were signs that even this low amount of chimerism could “delay the onset of overt signs of disease” (54). We think that we were able to overcome the difficulties faced by these investigators for three reasons: First, we transplant a higher dose of cells directly into the fetal liver (not placenta) using custom-made micropipettes. Second, we also inject rhGUS at the same time to improve survival in this fragile strain. Third, we use CX3CR1-GFP reporter mice to enable a more sensitive and specific enumeration of donor-derived microglia. Our results indicate that it is possible to engraft microglia in multiple animals with amounts of chimerism that could result in correction of neural inflammation and cross-correction of systemic disease.

We note that the proper term for engrafted microglia in the brain may be “microglia-like” cells, rather than definitive microglia. Fate-mapping studies indicate that microglia originate from cells of yolk sac primitive HSCs alone (23). Still, other studies in zebrafish found that definitive HSCs can contribute to the permanent adult microglial pool (55). These definitive HSCs, including those of liver origin, seed the brain in a second wave of microglial engraftment at around E12.5 (24), occurring near the time of our in utero transplants. We transplanted fetal liver mononuclear cells, which contain both definitive HSC and yolk sac–derived monocytes (23). It is possible that some of the microglial engraftment that we see is secondary to monocytes in the fetal liver that are of yolk sac origin (56). Whereas previous transcriptomic and epigenetic studies reported that “engrafted brain macrophages” differed considerably from endogenous microglia (43), the cells in our study were more closely related to donor-derived microglia. For example, when we focused on signature microglial genes, there were no significant differences in expression between host and donor cells in 13 of the 14 target genes. These included Tmem119, Sall1, Kcnk13, and SiglecH, which have been cited as key genes in defining microglial identity (5760), particularly because they were expressed at low amounts by engrafted brain macrophages after postnatal HSCT. Thus, IUHCT during a more natural window in ontogenic microglial development may be able to seed the brain with donor cells that are more microglia-like than with a postnatal approach.

There were some limitations in our study—in particular, the poor survival of the MPS7−/− strain limited the number of animals available for analysis. We used fetal enzyme supplementation along with IUHCT in an attempt to maximize survival of MPS7−/− mice. We also found that although the liver histology was corrected in all MPS7 chimeras, brain engraftment appears to require a higher amount of peripheral blood engraftment, limiting the number of informative chimeric animals. Peripheral blood engraftment can potentially be improved in future studies by using methods such as CD26 blockade (61).

We are currently performing a phase 1 clinical trial of IUHCT in fetuses with alpha thalassemia major (NCT02986698). This trial involves transplantation of maternal HSCs, to take advantage of the existing tolerance between the mother and fetus in normal pregnancy (62) and to avoid a possible maternal immune response to transplanted cells (11, 63). Application of this strategy for LSDs would require some modification, because mothers of patients with LSD are carriers for the disease and would produce lower amounts of enzyme. It may be necessary to transplant stem cells that are matched to the mother (but not obtained from carriers) or to further engineer maternal HSCs to overexpress the missing protein, similar to the current strategy for autotransplantation of gene-corrected HSCs (16) in LSDs. However, these experiments are an encouraging proof of concept to further develop the idea of in utero therapy for patients with LSDs. The ideal clinical strategy would likely involve infusion of the recombinant enzyme along with HSCs to allow correction of storage until there is appreciable engraftment of HSCs.

Together, our results indicate multiple advantages to a fetal approach to treating MPS7 and potentially also a broad range of other LSDs. With existing clinical application of postnatal ERT, as well as clinical trials in place for IUHCT, these modalities have immediate clinical implications in the treatment of these life-threatening disorders.


Study design

The aim of the study was to investigate whether fetal therapy (IUERT or IUHCT) could ameliorate disease phenotype in MPS7, with additional focus on the tolerogenic and neurologic effects of both therapies. IUERT experiments were performed using the breeding scheme MPS7+/− × MPS7+/−, and littermates were analyzed separately after genotyping. IUHCT experiments were performed in the pups of both wild-type C57B6 (B6) × B6 breedings and in MPS7+/− × MPS7+/− breedings, and chimerism rates and host genotypes were determined at the time of weaning. Animals with >1% donor cells were deemed to be chimeric. In each graph, the n value reflects the total number of animals. We did not exclude outlier data points. All measures requiring manual quantification were completed in a blinded fashion—for example, with measurements of histologic accumulations, neurologic testing, bone length, and brain cell diameter. Data file S1 contains raw data from experiments with n < 20 per group.


Wild-type C57BL/6J (C57; CD45.2), B6.SJL-PtrcaPep3b/BoyJ (BoyJ; CD45.1), B6.Cg-Ptprca Cx3cr1tm1Litt/LittJ (CX3CR1-GFP, CD45.1), and B6.Cg-Gusbmps/BrkJ (MPS7+/−, CD45.2) mice were obtained from the Jackson Laboratory. All mice were bred and maintained in a pathogen-free facility at the University of California, San Francisco (UCSF). All procedures were performed according to the UCSF Institutional Animal Care and Use Committee–approved protocol. Genotype was determined by polymerase chain reaction of genomic DNA followed by Sanger sequencing as previously described (64). Homozygous mutants were confirmed with the presence of a frameshift mutation within exon 10 of the β-glucuronidase gene. We also developed a genotyping method using restriction enzyme digestion of exon 10 on chromosome 5, because this is the site of frameshift mutation in MPS7. Polymerase chain reaction was used to obtain a 448–base pair (bp) fragment containing the region of interest. Samples were then incubated with an NLAIV endonuclease (New England Biolabs), which bound directly to the wild type sequence (5′-GGCCCC-3′) and yielded fragments of 285 and 163 bp in wild-type mice. In homozygous MPS7−/− mice (5′-GGCCCG-3′), the recognition site is disrupted due to the base pair deletion at the 3 prime end, resulting in the lack of digestion and a single fragment of 448 bp. All three bands are seen in heterozygous MPS7+/− mice.

In utero ERT

Pregnant dams were anesthetized at E14.5 using isoflurane. A midline laparotomy was made, the uterus was exteriorized, and 4 μg (20 mg/kg) of vestronidase alfa (Ultragenyx) was injected into the fetal livers of recipients using pulled glass micropipettes as previously described (11, 28). The uterus was returned to the abdominal cavity, and the wound was closed in layers. Surviving mice were then given retro-orbital (intravenous) postnatal boosting of rhGUS (4 mg/kg) every 2 weeks starting at 3 weeks of age.

IUHSCT and chimerism assessment

Fetal liver mononuclear donor cells were harvested at E14.5 from CX3CR1-GFP-CD45.1 × CD45.1 pregnancies (28, 43). Two separate groups of recipients received HSCs at E13.5-E14.5: wild-type fetuses (B6.CD45.2 × B6.CD45.2 parents) or MPS7 fetuses (MPS7+/− × MPS7+/− parents). Along with HSCs, fetuses in MPS7 litters received a 20 mg/kg dose of IUERT in the same solution. Peripheral blood chimerism was assessed in all transplanted progeny at 3 weeks of age by calculating donor/(donor + host) cells. Mice with >1% circulating donor cells were considered chimeric. Lineage analysis was performed by staining peripheral blood [using antibodies against CD45.1 (A20), CD45.2 (104) (BioLegend), F4/80 (BM8), CD11b (M1/70) (Thermo Fisher Scientific), Gr-1 (RB6-8C5) (BioLegend), CD19 (1D3) (BD Biosciences), and CD3 (17A2) (BioLegend)] and bone marrow [using antibodies against CD45.1 (A20), CD45.2 (104) (BioLegend), CD117 (2B8) (Thermo Fisher Scientific), Sca-1 (D7) (BioLegend), CD217 (PAJ-17R), CD34 (RAM34) (Thermo Fisher Scientific), CD135 (A2F10), and CD16/32 (93) (BioLegend)].

Analysis of GUS tissue activity

At the time of harvest, organs were removed, flash-frozen in liquid nitrogen, and stored at −80°C until biochemical analysis. Organs harvested 4 to 7 days after IUERT were rinsed thoroughly to minimize blood contamination, because perfusion could not be performed at this age. Homogenates of the tissues, including the brain, heart, lungs, liver, kidneys, spleen, and bones, were assayed for GUS activity as previously described (65).

Analysis of GUS expression by flow cytometry

Mononuclear cell preparations from peripheral blood or harvested organs were incubated at 37°C in Iscove’s modified Dulbecco’s medium (Life Technologies) containing 2% bovine serum albumin and 100 μM ImaGene Green C12FDGlcU (Setareh Biotech) substrate for 2.5 hours as previously described (30). After incubation for 2.5 hours, d-glucaric acid 1,4-lactone (Setareh Biotech) was added to the medium to stop the reaction. Fetal brains underwent careful meningeal dissection to minimize blood contamination, because perfusion could not be performed at this age.

Further staining for flow cytometry was performed to determine which types of cells were responsible for rhGUS expression in the blood and tissues, after incubation with ImaGene Green C12FDGlcU (which is visualized in the fluorescein isothiocyanate channel on flow cytometry). Mononuclear cell preparations were incubated in fluorescence-activated cell sorting staining buffer (PBS with 2% fetal bovine serum and 2 mM EDTA) with fluorochrome-conjugated anti-mouse surface antibodies. The cells were then stained with antibodies against cell markers including CD45.1 (A20), CD45.2 (104), Ly-6G/Ly6C (RB6-8C5), CD3 (17A2) (BioLegend), CD19 (1D3) (BD Biosciences), F4/80 (BM8), GR-1 (RB6-8C5) (BioLegend), and CD11b (M1/70) (Thermo Fisher Scientific) for the peripheral blood. An additional CD11c antibody (N418) (Thermo Fisher Scientific) was added for liver samples. Cell viability was determined using LIVE/DEAD fixable yellow stain kit (Thermo Fisher Scientific). For microglial detection, cells were stained with antibodies to CD45 (A20; eBioscience) and CD11b (M1/70; BD Biosciences). Flow cytometry was performed on an LSR II (BD Biosciences) or Accuri (Accuri Cytometers Inc.) flow cytometer and analyzed using FACSDiva (BD Biosciences) or FlowJo (FlowJo LLC) software.


Brain, lung, heart, liver, spleen, kidneys, and bone were harvested from mice (at 8 to 10 weeks of age for IUERT experiments and at 4 weeks of age for IUHCT experiments) and underwent PAS staining. Images were acquired using the Leica upright microscope (Leica DM1000). Images were evaluated by a laboratory technician blinded to the treatment groups, and the total area of lysosomal cellular accumulations was counted per high-powered field in four high-powered representative fields per tissue type.

Neurologic testing

Wild-type and MPS7−/− mice underwent forelimb grip testing and basic movement testing at age 8 weeks by investigators blinded to the groups. Mice were allowed to habituate to the testing environment an hour before testing began. Grip testing was done using a Chatillon Digital Force Gauge (DFIS-2) (Ametek). Mice were held near the measuring bar until the bar was gripped by both forelimbs, and then the mice were slowly pulled away from the bar until they released the bar. Peak force was measured, and grip strength was calculated as grams of force generated divided by grams of weight.

Basic movement was assessed as previously described using the Smart-Homecage (AfaSci Inc.) (66). Infrared matrices assessed mouse activity, position, and locomotion. Mice were placed into the cage and allowed 5 min to explore it. Exploratory behaviors were assessed, including travel distance, which was then calculated using the CageScore software (AfaSci Inc.).

Micro–computed tomography

To assess bone length by micro-CT, after chest and abdominal organs were harvested, mice were fixed in 10% phosphate-buffered formaldehyde for 24 hours and then transitioned to 70% ethanol solution. They were then imaged using a Scanco Medical MicroCT 50 scanner (Scanco Medical AG) with 30-μm voxel size and 55-kV x-ray energy. Femurs and tibias were measured by an investigator blinded to the treatment groups.

Analysis of antibodies against rhGUS

Mice were immunized with rhGUS at 6 weeks of age, with each mouse receiving 50 μg of rhGUS in 0.2 ml of CFA (Sigma-Aldrich) intraperitoneally as previously described (9). Blood was collected by facial-vein puncture to measure antibodies to rhGUS enzyme by enzyme-linked immunosorbent assay (ELISA) before and 14 days after the CFA injection.

Immunized mouse serum was analyzed as previously described (67). A 96-well microtiter plate was coated at 4°C overnight with 0.5 μg of rhGUS enzyme. The wells were washed with TBST [10 mM tris, 150 mM NaCl, and 0.05% Tween 20 (pH 7.6)]. Samples were then blocked with 3% casein in PBS, washed, and incubated with 100 μl of 10-fold serial dilutions of mouse serum at 37°C for 2.5 hours. Samples were then washed with TBST and incubated with peroxidase-conjugated goat anti-mouse IgG1 (Abcam), IgG2b (Invitrogen), IgG2c (Life Technologies), and IgG3 (Invitrogen) for 1 hour at room temperature. The wells were then washed with TBST and TBS [10 mM tris (pH 7.6) and 150 mM NaCl] and incubated with 100 μl of peroxidase substrate (Bethyl Laboratories) for 10 min at room temperature. The reaction was stopped with 2.5 μl of 20% SDS. Plates were then read on an ELISA plate reader at optical density at 405 nm.

Brain immunofluorescence

Chimeric mice were perfused and harvested at 4 weeks of age and underwent brain processing and free-floating stains (68). Sections were stained with anti-Iba1 antibody (VWR), and donor cells were detected through GFP expression on confocal microscopy. For brain inflammation studies, slices were stained with anti-CD68 antibody (Bio-Rad), and integrated density per microglial cell was calculated using ImageJ.

Cell sorting and RNA sequencing

Brains from chimeric animals were perfused, harvested, and processed through papain-based enzymatic digestion, followed by magnetic-activated cell sorting removal of myelin (Miltenyi Biotec). We analyzed both wild-type B6 hosts and MPS7+/− hosts since the latter mice are phenotypically normal (69). To detect microglia, cells were stained with antibodies against CD45.1 (A20; BioLegend), CD45.2 (104) (Thermo Fisher Scientific), and CD11b (M1/70) (BD Biosciences). Donor microglia, host microglia, and donor circulating monocytes were sorted separately. RNA sequencing was performed by first isolating mRNA using DynaBeads mRNA DIRECT Kit (Thermo Fisher Scientific), followed by complementary DNA preparation using SMART-Seq v4 Ultra Low Input RNA Kit (Takara Bio), library preparation using Nextera XT DNA Library Prep Kit (Illumina), and sequencing using HiSeq 4000 (Illumina). STAR (version 2.5.2b) (70) was used to align reads to the mouse genome (version GRCm38.78). Gene counts were accumulated using STAR’s option—quantMode GeneCounts. Uniquely mapped reads were used to identify differentially expressed genes using DESeq2 (71). Principal components analysis plots were made using prcomp [R 3.6.0 (stats)], t-distributed stochastic neighbor embedding plots were made using the Rtsne package (version 0.15), and heatmaps were made using heatmap.2 (gplots v3.0.1.1). Pathway analysis was performed using DAVID (Database for Annotation, Visualization, and Integrated Discovery) software (72).


Comparisons between two groups were analyzed using the Mann-Whitney or the χ2 test. Comparisons for more than two groups were evaluated using analysis of variance (ANOVA) with Tukey’s post hoc test or Kruskal-Wallis with Dunn’s multiple comparisons test. Data are represented as means ± SEM. P < 0.05 was deemed statistically significant. Statistical analysis was performed using GraphPad Prism (GraphPad Software Inc.).


Fig. S1. Dose comparison with in utero enzyme replacement reveals higher tissue enzyme activity after 20 mg/kg compared to the 4 mg/kg dose.

Fig. S2. In utero enzyme replacement results in detectable enzyme in all tissues.

Fig. S3. Behavioral testing for open-field basic movement and rearing.

Fig. S4. Additional ELISA experiments.

Fig. S5. Peripheral blood chimerism and survival rates after IUHCT.

Fig. S6. Peripheral blood monocyte cross-correction.

Fig. S7. RNA sequencing studies and further data.

Data file S1. Original data.


Acknowledgments: We thank members of the MacKenzie, Villeda, and Shiow laboratories; members of the Center for Maternal-Fetal Precision Medicine (especially M. Norton, S. Sanders, and R. Gallagher); A. M. Montaño; and members of Ultragenyx (including V. Koppaka) for helpful discussions. Funding: This study was supported by a grant from Ultragenyx to T.C.M., funds from the UCSF Center for Maternal-Fetal Precision Medicine, CIRM New Faculty grant to T.C.M. (RN3-06532), NIH grant support 5T32HD007263-35 (Integrated Training in Reproductive Sciences) to Q.-H.N., and NIH grant support F32AG055292 (Kirschstein NRSA Postdoctoral Fellowship) to J.S. Author contributions: Q.-H.N., R.G.W., and T.C.M. designed the experiments. Q.-H.N. and R.G.W. performed the in utero injections. Q.-H.N., R.G.W., C.E., and B.W. performed tissue harvesting. C.E. performed quantification of the GAG accumulations. B.W., Q.-H.N., and R.G.W. performed flow cytometry and data analysis. J.S. and S.V. performed behavioral testing. Q.-H.N., L.K.S., J.S., and S.V. performed histological analysis of the brain. R.S. and C.E. designed genotyping methods for the MPS7 mice. G.B. and J.C. performed tissue enzyme activity quantification. J.D.M. performed micro-CT imaging analysis. Q.-H.N., R.G.W., and T.C.M. wrote the manuscript with input from all authors. Competing interests: G.B. and J.C. are employees of Ultragenyx. The other authors declare that they have no competing interests. Data and materials availability: All data associated with this study are present in the paper or the Supplementary Materials. Source data files for RNA-seq analysis have been deposited to the NCBI Sequence Read Archive under accession number PRJNA590839.

Stay Connected to Science Translational Medicine

Navigate This Article