miR-143 Regulates Lysosomal Enzyme Transport across the Blood-Brain Barrier and Transforms CNS Treatment for Mucopolysaccharidosis Type I

Abstract

During brain maturation, cation-independent mannose-6-phosphate receptor (CI-MPR), a key transporter for lysosomal hydrolases, decreases significantly on the blood-brain barrier (BBB). Such a phenomenon leads to poor brain penetration of therapeutic enzymes and subsequent failure in reversing neurological complications in patients with neuropathic lysosomal storage diseases (nLSDs), such as Hurler syndrome (severe form of mucopolysaccharidosis type I [MPS I]). In this study, we discover that upregulation of microRNA-143 (miR-143) contributes to the decline of CI-MPR on the BBB during development. Gain- and loss-of-function studies showed that miR-143 inhibits CI-MPR expression and its transport function in human endothelial cells in vitro. Genetic removal of miR-143 in MPS I mice enhances CI-MPR expression and improves enzyme transport across the BBB, leading to brain metabolic correction, pathology normalization, and correction of neurological functional deficits 5 months after peripheral protein delivery at clinically relevant levels that derived from erythroid/megakaryocytic cells via hematopoietic stem cell-mediated gene therapy, when otherwise no improvement was observed in MPS I mice at a parallel setting. These studies not only uncover a novel role of miR-143 as an important modulator for the developmental decline of CI-MPR on the BBB, but they also demonstrate the functional significance of depleting miR-143 for “rescuing” BBB-anchored CI-MPR on advancing CNS treatment for nLSDs.

Graphical Abstract

Loss and gain of transporter functions on the blood-brain barrier (BBB) during development are largely unknown. Pan and colleagues uncover a new mechanism to which microRNA-143 contributes to the decline of cation-independent mannose-6-phosphate receptor during BBB maturation. The functional significance of depleting microRNA-143 on improving brain delivery of lysosomal enzyme is demonstrated in a disease model with long-term CNS correction.

Introduction

The blood-brain barrier (BBB) comprises a network of microvessels formed by highly specialized brain endothelial cells (BrECs) that control substance influx and efflux between the circulation and the central nervous system (CNS).¹ The BBB maintains CNS homeostasis with unique characteristics, including selective expression of particular transmembrane transport receptors.² The primitive BBB is formed at the embryonic stage,³ fully developed in human at birth, but it is not completely mature and functional in mice until the early postnatal stage.^4,5 However, the molecular mechanisms that control BBB maturation, especially the spatiotemporal establishment with the loss and gain of many transport functions, are not fully understood.⁵ Moreover, a mature BBB restricts the delivery of approximately 98% of small molecule drugs and nearly all macromolecules from the blood into the brain,^6,7 thus becoming a major obstacle in treating most neurological disorders, including neurologic lysosomal storage diseases (LSDs).

LSDs are a group of inherited metabolic disorders caused by the deficiency of lysosomal enzymes or components affecting lysosomal functions.⁸ More than two-thirds of over 70 LSDs involve the CNS with a broad spectrum of severity, which makes LSDs the most common cause of pediatric neuronopathic diseases.^9,10 If untreated, the progressive CNS pathology can cause neuronal dysfunctions and premature death in patients with neurologic LSDs (nLSDs), including Hurler syndrome, the severe form of mucopolysaccharidosis type I (MPS I).^11,12 MPS I is caused by the deficiency of α-l-iduronidase (IDUA) and subsequent accumulation of glycosaminoglycans (GAGs) in multiple organs.¹³ The cation-independent mannose-6-phosphate receptor (CI-MPR), also known as insulin-like growth factor 2 receptor (IGF2R), plays a key role in intra-cellular and inter-cellular transport of most lysosomal enzymes and in mediating metabolic cross-correction in affected cells and organs. Such phenomena provide the foundation for two main options in treating some LSDs, i.e., enzyme replacement therapy (ERT) and hematopoietic stem cell transplantation (HSCT).^14,15 However, they are largely unsuccessful in reversing any neurological complications, with poor penetration of therapeutic enzymes from the circulation to the CNS most likely due to the lack of CI-MPR on the BBB in the adult brain.¹⁶ Interestingly, systemic delivery of lysosomal enzymes in postnatal mice (≤2 weeks old), but not in adult mice, resulted in CI-MPR-dependent, widespread enzyme distribution in the CNS, suggesting a developmental decline of CI-MPR on the BrECs during BBB maturation.^4,17 However, the underlying mechanisms regulating CI-MPR levels in the BBB during this early postnatal stage remain largely unknown.

MicroRNAs (miRNAs) are a class of small noncoding RNAs that posttranscriptionally regulate protein expression.¹⁸ They have emerged as major modulators of pathways involved in multiple developmental stages, including vascular functions.19, 20, 21 In the present study, we identified a CI-MPR-targeting miRNA (miR-143) in mouse brain microvasculature (BrMV) and verified its direct interaction with CI-MPR mRNA. Functional studies showed that miR-143 inhibits CI-MPR expression and its transport ability in human ECs in vitro. Moreover, genetic removal of miR-143 resulted in significant increase of CI-MPR expression in the BBB with subsequent improvement of biological function in isolated BrMV. Deletion of miR-143 in a mouse model of MPS I resulted in elevated IDUA delivery and widespread distribution of enzymes in the brain, with normalization of CNS deficits 5 months after systemic delivery of moderate levels of IDUA derived from erythroid/megakaryocytic (EMK) cells. These studies uncover an original mechanism (via miR-143) attributable to the loss of effective CI-MPR-mediated transport on the BBB during development and demonstrate the functional significance of depleting miR-143 for improving enzyme delivery across the mature BBB in the treatment of nLSDs.

Results

Developmental Decline of BBB-Anchored CI-MPR Is Associated with Elevation of miR-143 in Mature BrMV from Adult Mice

To investigate the expression pattern of CI-MPR in the BBB, we isolated BrMV from either pups (∼10 days postnatal) or adult (4–6 months) C57BL/6 mice as described.²² To evaluate relative cellular contribution of BrECs (i.e., relative purity) in each BrMV sample, immunofluorescence (IF) analysis was conducted by staining with lectin for ECs and phalloidin for all cells (Figure 1A). Using BrMV samples containing various percentages of BrECs (as determined by IF), correlation curves were established by their association with relative fold changes over capillary-depleted brain (CDB) samples by quantitative real-time PCR for mRNA levels of EC markers, Pecam1 and Cldn5 (Figure S1), which have been validated previously.²² Both sets of BrMV samples were confirmed for comparable and high percentages of BrECs using correlation curves (Figure 1B). By immunoblot analysis, we found that CI-MPR protein levels decreased (by 62%) in adult BrMV as compared to neonatal mice (Figure 1C). However, marginal reduction (∼20%) was observed at mRNA levels as determined by TaqMan qRT-PCR (Figure 1D). These findings directly confirmed the decline of CI-MPR on the maturing BBB in mice.

Figure 1.

Diminishing CI-MPR Protein Expression in Adult BrMV Is Associated with an Increase of miR-143 Compared to BrMV from Pups

(A) Representative images of immunofluorescence analysis for BrMV samples isolated from adults or pups. Slices from cytospin were stained with lectin (green) for endothelial cells, Alexa Fluor 647-conjugated phalloidin (purple) highlighting cytoskeletal F-actin filaments for all cells, and DAPI (blue) for nuclei. Yellow arrowheads indicate non-endothelial cells. Scale bar, 50 μm. (B) Relative percentages of endothelial cells in BrMV samples (n = 4). Quantitative expression levels of endothelial-specific markers (Pecam1 and Cldn5) were calculated based on correlation curves (Figure S1) derived from IF staining (percent of lectin⁺/DAPI⁺) and qPCR of various BrMV samples. (C) Immunoblotting analysis of CI-MPR protein in BrMV samples; each was isolated from brains of either 12–15 pups (∼10 days old) or 10–12 adult mice (∼4–6 month of age). Relative CI-MPR levels (intensity of signals) are indicated, with β-tubulin serving as an internal loading control. (D) Copy numbers of CI-MPR mRNA in BrMV samples by TaqMan qRT-PCR (n = 4–5). Data are presented as mean ± SEM, each with two RT reactions and quantified in triplicates. (E) Heatmap representation of murine miRNA microarray showing miRNA clustering analysis. Data are derived from two pairs of BrMV samples, and miRNAs with expression values >1,000 and p < 0.05 are listed. (F) Venn diagram showing miR-143 as the only overlapped miRNA among two sets (using prediction algorithms TargetScan and miRanda) of predicted miRNAs targeting to the 3′ UTR of CI-MPR and significantly changed miRNAs in microarray. (G) Relative expression of miR-143 and miR-145a by qRT-PCR analysis, validating the increasing abundance of the miR-143/145a cluster in adult BrMV. Data are presented as mean ± SD; n = 4 for all groups, including small RNA samples used in (E). ∗p < 0.05, ∗∗p < 0.01, by two-tailed Student’s t test.

To identify differentially expressed miRNAs that may be involved in the reduction of BBB-anchoring CI-MPR, we compared miRNA signatures in the developing or mature BrMV by miRNA microarray analysis (Figure 1E). We identified 23 miRNAs to be relatively abundant (>1,000 rpm) and significantly changed (>2-fold, p < 0.05), either reduced (7 miRNAs) or increased (16 miRNAs), in adult BrMV. Using two miRNA target-prediction algorisms, TargetScan (http://www.targetscan.org/)²³ and miRanda (http://www.microrna.org/microrna/home.do)²⁴, we analyzed the 3′ untranslated region (UTR) of murine Ci-mpr mRNA and found miR-143-3p (miR-143) (among 20 co-predicted candidates) to be the only miRNA significantly elevated in adult BrMV (Figure 1F). The increase of miR-143 was further verified by TaqMan qRT-PCR (5.4-fold higher in adult BrMV), which was concurrent with higher levels of miR-145a, a Ci-mpr non-targeting miRNA that sharing the same polycistronic precursor transcript with miR-143 (Figure 1G). Notably, in normal C57BL/6 mice, we observed a strong negative correlation between miR-143 expression and the levels of CI-MPR protein in skeletal muscle and the liver as determined by qRT-PCR and western blot analysis (Figure S2). These data demonstrated the association of diminishing levels of CI-MPR protein and the increased levels of miR-143 in adult BrMV during early postnatal development, suggesting miR-143 as a potential CI-MPR posttranscriptional regulator that may contribute to the loss of CI-MPR-mediated transport on the mature BBB.

miR-143 Directly Regulates CI-MPR by Interacting with Its 3′ UTR

Three potential miR-143-binding sites were predicted using TargetScan mouse 7.1²³ in the 3′ UTR of Ci-mpr, with one highly conserved site (binding site 3 [B3]) and two less conserved sites (B1 and B2) across vertebrates (Figure 2A), whereas miR-145a has no putative target on the Ci-mpr transcript. To dissect specific interaction between miR-143 and the Ci-mpr 3′ UTR, a reporter system was constructed by fusing the firefly luciferase (Fluc) coding sequence with either the full length of wild-type Ci-mpr 3′ UTR (WT) or UTRs with mutations at single (Mut1, Mut2, Mut3) or triple binding sites (Mut123) (Figures 2A and 2B). Upon analyses with dual-luciferase assays, we found that overexpression of miR-143, not miR-145a or mock control, suppressed Fluc activities (Figure 2C). The specificity of such downregulation was further supported by a dose-dependent inhibition with increasing amounts of miR-143 (Figure 2D). Site-directed mutagenesis study further demonstrated that mutations at sites 2 and 3 (Mut2, Mut3, and Mut123), but not site 1, completely abolished the inhibitory effect of miR-143 on Fluc expression, when comparing with non-targeting miR145a controls (Figure 2E). These data support the notion that miR-143 downregulates CI-MPR expression by strong and specific binding to its 3′ UTR.

Figure 2.

miR-143 Directly Regulates CI-MPR by Interacting with Its 3′ UTR

(A) Schematic diagram of predicted miR-143 binding sites on the 3′ UTR of Ci-mpr mRNA and luciferase-based reporter system. The wild-type (WT) UTR fragment containing three predicted binding sites (B1, B2, B3; ∇) or its mutated (X) variants were constructed downstream of firefly luciferase (Fluc). (B) Potential binding pattern of miR-143 to the WT or mutated 3′ UTRs. The minimum free energy for miR-143 binding is −19.3 kcal/mol at the B1 site, −21.8 kcal/mol at the B2 site, and −26.8 kcal/mol at the B3 site, respectively (http://rna.tbi.univie.ac.at/cgi-in/RNAWebSuite/RNAfold.cgi). The binding region in the seed sequence of each site was mutated and is highlighted in lowercase letters on the right panels. (C) miR-143, not miR-145a, specifically represses Fluc expression. Human HEK293T cells were either transfected alone with WT plasmid (mock) that expressed Fluc-WT and Renilla luciferase (Rluc, as transfection control), or co-transfected with pCMV6-miR-145a or pCMV6-miR-143 plasmids. (D) Fluc activity in 293T cells when co-transfected with increasing amounts of pCMV6-miR-143 or pCMV6-miR-145a. (E) Relative luciferase activities in 293T cells that were transfected with WT or mutated UTR-containing plasmids, together with increasing amounts of either miR-143- or miR-145a-expressing constructs. Data were derived from two independent experiments and are shown as the mean ± SD. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, by two-tailed Student’s t test. ns, not significant.

miR-143 Negatively Controls CI-MPR-Mediated Transport in Human Vascular ECs

Mature miR-143 is highly conserved across vertebrates and identical between mice and humans, with predicted target sites on CI-MPR transcripts in both species. To further understand the potential translational perspective and functional correlation between miR-143 and CI-MPR, we utilized two lines of human vascular ECs, human umbilical vein ECs (HUVECs) and a human BrEC line (hCMEC/D3),²⁵ which have different, negatively related endogenous miR-143 and CI-MPR levels (Figure S3). To manipulate miR-143 expression, two lentiviral vectors (LVs) were constructed, with one vector containing a sponge cluster of eight miR-143 binding sites (LV-143Spg) to trap endogenous miR-143 for its degradation, and the other overexpressing miR-143 (LV-miR-143) (Figure 3A). Stably transduced cell lines were generated by fluorescence-activated cell sorting (FACS) for GFP⁺ cells after transduction with LV-143Spg, LV-miR-143, or the parental GFP-expressing LVs to serve as controls. No alteration of endogenous IDUA activity was observed in human ECs after manipulation of miR-143 as compared to their control cells transduced with sham vectors, suggesting that miR-143 may have no effect on the generation and stability of IDUA (Figure S4). In HUVECs stably transduced with LV-143Spg (with 97% GFP⁺ after sorting for GFP⁺ cells), reduction of endogenous miR-143 (by 80%) was observed that was associated with an increase of CI-MPR mRNA and protein (∼1.5-fold) compared to cells transduced with a parental GFP-expressing LV as determined by qRT-PCR and western blot analysis, respectively (Figure 3B). Moreover, FACS analysis with alive cells showed significantly higher CI-MPR signal intensity in HUVEC-143Spg cells compared to control cells, suggesting increased CI-MPR distribution on the cell surface upon overexpression of 143Spg (Figure S5A). Such an increase of CI-MPR was also associated with a reduction of GFP expression levels in GFP⁺ HUVEC-143Spg cells, supporting a downregulation role of endogenous miR-143 on the expression of GFP-143Spg (Figure S5B). Conversely, overexpression of miR-143 in hCMEC/D3 significantly lowered CI-MPR mRNA and decreased protein levels by >60% (Figure 3C). The data demonstrate a negative regulatory effect of miR-143 on CI-MPR levels in human ECs.

Figure 3.

miR-143 Reversely Controls CI-MPR Expression and Affects CI-MPR-Mediated Endocytosis

(A) Diagrams of lentiviral vectors for downregulation of endogenous miR-143 (LV-143Spg) and overexpression of miR-143 (LV-miR-143). The sponge cluster contains eight miR-143 binding sites and presents as part of an artificial 3′ UTR of GFP, which was driven by a SFFV promoter. The pre-miR-143 (with an ∼370-bp flanking sequence) was cloned from the murine genome and expressed under a CMV promoter. (B and C) Downregulation of miR-143 in HUVECs (B) or the overexpression of miR-143 in hCMEC/D3 cells (C) reversely regulates CI-MPR transcripts and protein expression. Stably transduced cell lines were generated by FACS sorting for GFP⁺ cells, with parental GFP-expressing LVs serving as controls. Relative changes are shown for miR-143 expression (left) and CI-MPR mRNA (middle) by qRT-PCR, and for CI-MPR protein by western blot analysis (right panel). (D) Changes of CI-MPR-mediated IDUA uptake in binding-internalization assays. Data are shown as mean ± SD in all except for (D) (mean ± SEM). n = 6 for all and derived from two independent experiments using triplicate wells. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, by two-tailed Student’s t test. ns, not significant.

To explore CI-MPR-mediated transport function in these vascular cells, we identified maximal inhibitory dosage of M6P that can prevent uptake of IDUA enzymes in a competitive uptake assay (Figure S6). Of note, unlike brain hCMEC/D3 cells, which barely showed any non-CI-MPR-mediated uptake, HUVECs retained ∼15% IDUA uptake when approaching to a steady level, presumably due to non-specific pinocytosis or fenestration. To minimize the effect of non-specific uptake, we evaluated the CI-MPR-mediated uptake pathway using a two-phase binding-internalization assay (Figure 3D). After incubation with an equal amount of enzyme at 4°C, internalized IDUA in HUVEC-143Spg cells increased approximately 7-fold, which was significantly blocked upon M6P inhibition (Figure 3D, top panel). Conversely, miR-143 overexpression in hCMEC/D3 cells significantly repressed IDUA uptake (by ∼40%) after one cycle of binding and internalization processes (Figure 3D, bottom panel). Taken together, these results highlight the functional significance of manipulating miR-143 on CI-MPR-mediated lysosomal enzyme transport in human vascular ECs in vitro.

miR-143 Represses CI-MPR Function in BrMV

To investigate whether miR-143 is involved in downregulation of CI-MPR on the BBB in vivo, we examined the CI-MPR expression and function in adult BrMV isolated from a miR-143 knockout (miR-143KO) mouse model.²⁶ Expression of the miR-143/145a cluster is highly selective, predominantly in mesenchymal cells, particularly in vascular smooth muscle cells (VSMCs) and fibroblasts,^26,27 with substantially lower levels in endothelial or epithelial cells (<3% of SMCs).²⁸ Such a notion is consistent with our observation that minimal levels of miR-143 were found in capillary-depleted brain tissues (i.e., only ∼6% of BrMV) (Figure S7). The miR-143KO mice exhibited normal lifespan (up to 26 months) and fertility (data not shown), presumably due to functional compensation from miR-145a.²⁹ BrMV isolated from miR-143KO exhibited a marginal increase in Ci-mpr mRNA (∼1.4-fold) and an average of 2.2-fold elevation of CI-MPR protein as compared to BrMV from age-matched adult WT controls (Figures 4A and 4B). Skeletal muscles (with high endogenous expression of miR-143) also showed a significant increase in CI-MPR protein levels (∼4-fold) in miR-143KO mice than those in WT mice (Figure S8). Moreover, IF analysis demonstrated more intensified CI-MPR signals (∼3-fold) in microvascular endothelium from miR-143KO brain than those from WT mice, while comparable levels of lysosomal-associated membrane protein 2 (LAMP2, as an internal control) were detected in BrMV from both miR-143KO and WT mice (Figures 4C and 4D). These results further confirmed improved CI-MPR on BrMV upon depletion of miR-143.

Figure 4.

Loss of miR-143 Significantly Increases CI-MPR Expression and CI-MPR-Mediated Uptake in Brain Vasculatures from Adult miR143KO Mice

(A) mRNA levels of Ci-mpr in BrMV of adult miR-143 null mice as determined by qRT-PCR. Total RNAs extracted from BrMV isolations (n = 4, each from 12 mice, ∼3 months old) in each group were subjected to reverse transcription and qPCR (triplicate reactions). (B) Western blot analysis of CI-MPR expression in BrMV from WT and miR143KO mice. The number of mice contributing to each sample ranged from 9 to 12. Semi-quantification of densities in western blot by ImageJ is shown in the right panel. (C) Representative images of immunofluorescence analysis of BrMV isolates. Cytospins of BrMV samples were stained with anti-CI-MPR (red) as well as anti-LAMP2 (purple) for lysosomes, fluorescein-labeled lectin (green) for vascular endothelial cells, and DAPI for nuclei (blue). Areas within white squares were enlarged and are shown as insets. (D) Semi-quantification of fluorescence intensity shows a significant increase of CI-MPR signals, but not LAMP2, in BrMV from miR-143KO mice compared to WT mice. More than 100 BrECs from three isolations (n = 3) were quantified per genotype by ImageJ. (E) Representative images of IF analysis with antibody internalization assay using BrMV samples. Left panel: freshly isolated BrMVs were incubated with antibodies against CI-MPR and LAMP2 at 37°C for internalization, and followed by cytospinning, fixation, and staining with 2° antibodies as well as lectin and DAPI. Right panel: BrMVs were subjected to the same procedures as in the left panel except that only anti-CI-MPR was added for uptake and anti-LAMP2 antibody was included after fixation. Arrowheads indicate internalized CI-MPR antibodies. (F) Semi-quantification of fluorescence intensity of CI-MPR in antibody uptake assay (as in left panel of E). Quantification of CI-MPR signals for each image by ImageJ was normalized by DAPI signals. n = 14 images for 143KO mice and n = 9 for WT mice. Scale bars, 20 μm for (C) and (E). Data are presented as mean ± SEM. ∗p < 0.05, ∗∗∗p < 0.001, by two-tailed Student’s t test. ns, not significant.

While most CI-MPRs are localized in cytoplasmic vesicles (e.g., lysosomes, endosomes, and the trans-Golgi network) for intracellular enzyme trafficking to the lysosomes, variable amounts of receptors (less than 15%) are present on the surface of different types of cells to mediate endocytosis or recapture a variety of M6P-containing lysosome enzymes.^30,31 We evaluated the potential improvement of CI-MPR accessibility on the surface of BrMV by an ex vivo antibody internalization assay using antibodies recognizing either the ectodomain of transmembrane CI-MPR or LAMP2, a lysosome marker (as an intracellular or non-surface receptor control) (Figure 4E). While barely detectable in BrMV from WT mice, a sizeable proportion of anti-CI-MPR antibodies (but not anti-LAMP2) bound to cell-surface CI-MPR and was endocytosed into BrECs, resulting in intensive signals on BrMV isolated from miR-143KO mice (Figure 4E, left panel, white arrows). The lysosomal localization of such signals was verified by LAMP2-positive staining in the fixed samples (Figure 4E, right panel). Importantly, no LAMP2 signals were observed in any BrMV isolates (Figure 4E, left panel) except for post-fixed staining samples, validating that the signals of CI-MPR detected in BrMV from miR-143KO mice were on the cell surface. Semi-quantification analysis showed a 9.7-fold increase of CI-MPR signals in BrMV from 143KO mice compared to those from WT mice, supporting the substantial increase of surface-bond CI-MPR on the BBB. These data support the notion that genetic removal of miR-143 restores the CI-MPR levels and transporter function in BrMV of adult brain.

Deprivation of miR-143 Increases Brain Delivery of Peripheral IDUA in MPS I Mice with Widespread CNS Biodistribution after Short-Term and Long-Term Gene Transfer

To evaluate functional significance of miR-143 on CI-MPR-mediated CNS delivery of lysosomal enzymes, we generated a mouse model of MPS I with the deletion of miR-143 (MPS/143KO). The nonspecific diffusion of macromolecules across disrupted BBB may occur under certain neurological diseases or conditions.³² Thus, we examined the integrity of the BBB in MPS/143KO and MPS I mice with dextran-tracer (10 kDa) assessment (Figure S9; Video S1). Contrary to parenchymal distribution of tracer (red) in the brain of positive controls with mannitol-induced breaching of the BBB, tracer signals were only detected in association with blood vessels of brain from mice regardless of genotypes. The results indicated an intact BBB, the same as observed in normal mice, in both diseased mouse models that excluded the possibility of non-specific diffusion of IDUA (∼70 kDa) from blood.

Video S1

BBB Integrity Assessment by 10-kDa Tracer Shown in Z Stack 3D Images

To establish a practical time window for in vivo evaluation of IDUA delivery to the brain, we utilized two gene-transfer approaches for IDUA production in the circulation (Figure 5A). We first performed hydrodynamic tail vein (HTV) injection of IDUA-expressing plasmid DNA into mice so to transiently introduce high transgene expression from the liver of small rodents.^33,34 To avoid any transgene expression in the CNS, we restricted the expression of myc-tagged IDUA (IDUAmyc) in the liver by utilizing a hepatocyte-specific hybrid promoter (pLiver).¹³ Extremely high levels (up to ∼800-fold of heterozygous levels) of IDUA production were achieved 24 h and 48 h after injection in the circulation of both MPS-HTV (mean, 4,917 and 3,274 U/mL) and MPS/143KO-HTV (4,979 and 3,276 U/mL) groups (Figure 5D). Plasma IDUA generated from the liver of either MPS I or MPS/143KO mice were accessed for their capacities of CI-MPR-mediated uptake in primary fibroblasts derived from an MPS I patient (Figure S10). The comparable uptake levels suggested no alteration of M6P modification in the released form of IDUAmyc proteins by the removal of miR-143. Importantly, capillary-depleted brain tissues from well-perfused MPS/143KO-HTV mice exhibited ∼6-fold higher IDUA activity than those from MPS-HTV controls after 2-day delivery of supra-high IDUA levels in the peripheral blood (Figure 5E).

Figure 5.

Removal of miR-143 Facilitates Brain Delivery of Peripheral IDUA in MPS I Mice after Short-Term and Long-Term Gene Transfer

(A) Schematic diagram of experimental design. For short-term in vivo gene expression, adult mice (∼7 weeks old) were injected with a plasmid expressing IDUAmyc from a liver-specific promoter by liver-targeted hydrodynamic injection (HTV). For long-term HSC-mediated gene therapy, HSCs from MPS I mice were transduced with LV-P_IHK-hIDUAmyc-IRES-GFP that co-expressed IDUAmyc and GFP specifically in erythroid and megakaryocytic lineages using a hybrid promoter (P_IHK) and followed by transplantation into lethally irradiated mice (∼10 weeks old). Mice were monitored and dissected after trans-cardiac perfusion for multiple analyses as indicated. (B) Representative FACS analysis of transgene (GFP) expression in platelets, RBCs, and white blood cells (WBCs) 4 months after HSC-mediated LV-GT. Whole blood samples were stained with CD41-phycoerythrin (PE) for platelets and Ter119-allophycocyanin (APC) for RBCs. The plot of forward light scatter (FSC) versus side light scatter (SSC) indicates gates, based on size and complexity, for WBCs, platelets, and RBCs. (C) Transgene frequencies are shown by the percentages of GFP⁺ platelets detected in MPS I-GT and MPS/143KO-GT mice ∼5 months after transplantation. (D) Plasma IDUA activities in short-term HTV and long-term LV-GT mice. (E) IDUA activity in capillary-depleted brain (CDB) tissues either 2 days after HTV treatment or 5 months after LV-GT treatment. All data are presented as mean ± SEM. The number of mice used for each group (n) is indicated under each column. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. ns, not significant; ud, undetectable.

For the second approach, we conducted hematopoietic stem cell (HSC)-mediated gene therapy (GT) in MPS I or MPS/143KO mice using an LV to co-express IDUAmyc and GFP in maturing red blood cells (RBCs) and megakaryocyte/platelets with a hybrid promoter (P_IHK).^35,36 Transgene expression (GFP) was detected in platelets and RBCs (but not white blood cells) as determined by FACS analysis, suggesting that no IDUA expression would be from any migrating leukocytes (Figure 5B). Similar transgene frequencies were detected in both MPS-GT (14%) and MPS/143KO-GT (13%) groups, a level that is well feasible in GT clinical trials,³⁷ 5 months after gene transfer (Figure 5C). Such transgene expression levels resulted in moderate (∼12-fold of heterozygous levels) and sustained levels of IDUA in the circulation of both GT groups (75 and 77 U/mL) (Figure 5D). In contrast to undetectable brain IDUA found in LV-treated MPS I mice, marked levels of IDUA (∼6% of normal heterozygous levels) were observed in the brain of LV-treated MPS/143KO mice (Figure 5E). Such brain IDUA levels were significantly higher than those detected in post-HSC-transplanted MPS I mice (∼1% of heterozygous levels) that fully engrafted with HSCs from wild-type donors, as we previously reported.³⁸ Taken together, the results from delivery of either short-term with extremely high levels or long-term with moderate levels of IDUA suggest that the brain of miR-143 null mice is substantially more “permeable” to the IDUA in the peripheral blood.

To further assess the distribution of full-length IDUAmyc in the brain, we performed IF analysis with antibodies against myc-tag and various markers for different types of brain cells (Figure 6; Figure S11). In the short-term HTV study, the presence of IDUAmyc in brain capillaries was visualized as sporadic myc⁺ signals in BrECs labeled by lectin⁺ of both MPS/143KO-HTV and MPS-HTV mice 2 days after injection (Figure S11). However, co-localizations of myc⁺ with NeuN⁺ (a marker for mature neurons) or CD68⁺ (phagocytotic marker) signals were frequently detected in the cerebrum of MPS/143KO-HTV mice, but not MPS-HTV mice (Figures 6A and 6B). Semi-quantification of IF analysis showed that, unlike MPS-HTV mice with background levels of IDUA signals, evidently higher percentages of neurons (up to 30%) or brain macrophages/activated microglia (up to 54%) contained IDUAmyc delivered from the circulation in the brain stem, cerebral cortex, and midbrain of MPS/143KO-HTV mice (Figures 6C and 6D). Moreover, after long-term systemic delivery of moderate levels of IDUA, widespread distribution of full-length IDUAmyc protein was observed in NeuN⁺ neurons and CD68⁺ macrophages in the cerebrum (Figures 6E and 6G) and cerebellum (Figures 6F and 6G) of MPS/143KO-GT mice, but not in MPS-GT mice, 5 -months after HSC GT. The data demonstrated that CNS penetration of therapeutic lysosomal enzymes can be enhanced by boosting CI-MPR levels on the BBB via the depletion of miR-143.

Figure 6.

Distribution of IDUA in Brain Cerebrum and Cerebellum of Treated MPS I and MPS/143KO Animals

(A and B) IDUA distribution in the brain of HTV-injected MPS I or MPS/143KO mice that have similar plasma IDUA activities (∼3,000 U/mL at day 2). Representative images from IF analysis are shown using antibodies against myc-tag for IDUA (green) and anti-NeuN for neurons (red, in brain stem) (A) or anti-CD68 for brain macrophage/activated microglia (red, in cerebral cortex) (B), and counterstained with DAPI for nuclei (blue). Areas within yellow squares are enlarged and shown on the right. White arrows indicate IDUA⁺ neurons or macrophage/activated microglia. (C and D) Quantification of IDUA⁺ neurons (C) or brain macrophage/activated microglia (D) in MPS I and MPS/143KO brain sections 2 days after HTV injection (n = 6). Data were derived from brain sections of two pairs of MPS I and MPS/143KO mice with matching IDUA in plasma, each with three sections and >500 NeuN⁺ or CD68⁺ cells counted separately for each of the brain regions. All data are presented as mean ± SEM. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, by two-tailed Student’s t test. (E–G) IDUA distribution in the cerebrum and cerebellum of LV-GT-treated MPS/143KO and MPS I mice. Five months after HSC-mediated gene therapy with erythroid/megakaryocyte-specific LV, frozen sections were obtained from perfused animals. Representative images of cerebral cortex (E) and cerebellum (F) are shown after co-staining with anti-myc (green) for IDUA, anti-NeuN (red) for neurons (left panel), or anti-CD68 (red) for CNS macrophages and activated microglia (right panel). Areas (I–IV) within yellow boxes in (E) and (F) are enlarged and shown in (G), with white arrowheads indicating IDUA⁺ neurons (I, II) and CNS macrophages/microglia (III, IV).

Correction of Memory Impairment and Normalization of Brain Pathology Are Achieved in MPS/143KO Mice 5 Months after Systemic Delivery of Moderate Levels of IDUA Derived from EMK Cells

The advanced CNS therapeutic benefits in MPS/143KO mice were evaluated in both short-term and long-term studies with three courses. First, elevated brain GAG (substrate) accumulation, a hallmark in tissues of MPS I patients and mouse models,^13,39 was completely normalized in MPS/143KO-HTV mice, but not MPS I-HTV mice, 2 days after delivery of extremely high plasma IDUA (Figure 7A). Likewise, the abnormally elevated GAG levels detected in untreated MPS I and MPS/143KO mice were significantly reduced in the brain of MPS/143KO-GT mice, but not changed in those of MPS-GT mice. These results suggest that a complete metabolic correction can be achieved in the brain of MPS/143KO mice after a short-term delivery of peripheral IDUA at extremely high levels.

Figure 7.

Preclinical Evaluation on Brain GAG Accumulation, CNS Pathology, and Neurological Function in MPS/143KO and MPS Mice after Long-Term Peripheral IDUA Delivery

(A) Brain GAG levels after HTV injection or HSC-mediated LV-GT treatment. Two pieces of forebrain cortex were tested from each perfused animal either 2 days (HTV) or 5 months (LV-GT) after treatment, together with age-matched control groups. The number of mice used for each group (n) is indicated under each bar. (B and C) Histopathology analysis of forebrain cortex in Epon-embedded tissue sections 5 months after LV-GT. Representative images with toluidine blue staining are shown in (B), indicating brain vasculatures (white arrows) either associated or not associated with vacuolated perivascular cells (delineated by purple dash lines). The percentages of blood vessels that are associated with vacuolated perivascular cells in the forebrain of GT and control groups are shown in (C); n = 4 per group with each brain analyzed by approximately six sections and >200 vasculatures. (D) Repeated open-field test in mice 5 months after LV-GT treatment. Age-matched untreated or GT-treated MPS I and MPS/143KO mice, as well as normal heterozygote (Het) and miR-143KO controls, were evaluated for changes in horizontal activities (top panel) and grooming activities (bottom panel). n = 5–8 for GT groups and n = 8–25 for control groups. Data in all graphs are presented as mean ± SEM. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, by one-way ANOVA with Tukey’s test. ns, not significant.

Second, we examined brain pathology in mice 5 months after HSC-mediated LV-GT by scoring the percentages of brain vasculatures associated with perivascular cells that were distended with massive vacuoles (Vac⁺ vessels) in cerebral cortex of treated and control animals (Figures 7B and 7C). The percentages of Vac⁺ vessels were significantly reduced from 48% in untreated MPS I mice or 50% in MPS/143KO mice to ∼14% in the MPS/143KO-GT group, which was comparable to the heterozygote levels (p = 0.382). Importantly, we observed no amelioration of brain pathology (49%) in MPS I mice receiving comparable levels of GT treatment.

Finally, we assessed potential improvement on neurological function by a repeated open-field test, which can demonstrate non-associative memory impairment in untreated MPS I mice⁴⁰ (Figure 7D). LV-treated MPS I mice exhibited the same abnormal habituation as did the untreated controls, i.e., exhibiting minimal reduction of exploratory activity and no increase in grooming activity after being exposed to the same apparatus three times. Conversely, similarly treated MPS/143KO mice demonstrated a complete restoration of normal behavior as did heterozygous controls, suggesting correction of short-term memory deficits in MPS/143KO mice. Of note, miR-143KO mice have no discernible behavioral defects as tested. Collectively, these findings document an advanced CNS therapeutic benefits in MPS I mice with the removal of miR-143 when treated with sustained, moderate levels of IDUA in the circulation after LV-mediated HSC gene transfer at clinically relevant levels.

Discussion

In this study, we have revealed a novel role of miR-143 on the decline of a BBB-anchored CI-MPR transport system during development. Moreover, we demonstrated the functional significance of depleting miR-143 on improving CI-MPR-mediated CNS delivery of lysosomal enzyme across the mature BBB. Deletion of miR-143 leads to brain metabolic correction, as well as normalization of CNS pathology and memory impairment, 5 months after treatment with peripheral IDUA in MPS I mice when otherwise (without miR-143 alteration) no brain improvement was observed in a parallel setting.

During BBB formation/maturation, brain capillary ECs acquire unique properties (compared to ECs in peripheral organs), particularly the formation of specialized tight junctions (TJs) and the decreased/extremely low rate of transcytosis without fenestration.⁴¹ In addition to several signaling pathways suggested to be involved in TJ formation in the developing BBB,^42,43 major facilitator superfamily domain containing 2a (Mfsd2a) serves as a key regulator of BBB integrity and function by suppressing transcytosis in BrECs.³ In addition, several studies have demonstrated that brain endothelial miRNAs play critical roles in the regulation of BBB function under normal and neuroinflammatory conditions.¹⁹ In this study, we show that upregulation of miR-143 in mature BBB plays a role in the downregulation of CI-MPR, one of the “housekeeping” transporters, which contributes to the major obstacle for CNS delivery of lysosomal enzymes in treating nLSDs. Conversely, the neonatal BrMV exhibits robust CI-MPR expression at both mRNA and protein levels than those in the adult brain, which is consistent with the reported observation that efficient CI-MPR-mediated CNS delivery of lysosomal enzymes can be achieved in neonatal mice, but not in adult mice.^6,44 miRNAs are known for fine-tuning cell-specific, stage-specific gene expression to ensure correct EC specification and plasticity.⁴⁵ Our miRNA profiling study defines a set of candidates involved in promoting BBB maturation during the perinatal period. While not targeting to CI-MPR, an additional 22 miRNAs, including co-expressed miR-145, also show significant developmental change in mature BBB and may exert positive effects on the establishment of appropriate BBB functionality. Such data underscore a role of miR-143 in the selective reduction of transcytosis on the BBB and shed some light on the limited understanding of temporal and physiological changes via miRNAs in BBB maturation during development.

Under physiological conditions, the selectively expressed miR-143/145a cluster²⁸ plays concerted roles in regulating VSMC phenotypic switch,^26,46 while miR-143 also contributes to adipocyte differentiation.^47,48 Abnormal upregulation of miR-143 has been reported to inhibit angiogenesis, proliferation, and autophagy in vascular ECs or neighboring cell types, most likely by targeting to HKII, Agt2b, and Agt8.49, 50, 51 The increase of miR-143 observed in the mature BBB is consistent with the quiescent nature of the BBB (i.e., relatively low rates of angiogenesis and proliferation). The removal of miR-143 in the diseased MPS I brain apparently has no adverse effect detectable on BBB integrity, in agreement with the report that silencing miR-143 can protect the BBB from drug abuse-mediated vascular dysfunction.⁵² However, the exact involvement and function of miR-143 in fine-tuning BBB-specific, stage-specific gene expression and BBB functionality remain to be fully defined.

Interestingly, compared to the liver, CI-MPR expression is tremendously suppressed in skeletal muscles, which is correlated with much higher miR-143 levels. This observation is consistent with reports by others demonstrating that a low abundance of CI-MPR contributes to the poor response of skeletal muscle and heart to ERT treatment of Pompe⁵³ and MPS VII diseases.⁵⁴ Moreover, we observed that CI-MPR protein levels were significantly increased in skeletal muscles of miR-143KO mice. Thus, the miR-143/CI-MPR axis would exist not only on the BBB but also in various organs and cell types where endogenous expression of miR-143 is selectively high. Thereupon, modulating muscular CI-MPR expression by manipulation of miR-143 may also provide an alternative opportunity to increase muscular uptake of lysosomal enzymes in those poorly responded organs for LSD treatment.

The prospect of restoring the CI-MPR levels in mature BrMV via miR-143 depletion demonstrated in mice is likely translatable to humans with therapeutic potentials for several reasons. First, the sequences of miR-143 are identical between mice (mmu-miR-143-3p) and humans (hsa-miR-143). Second, the 3′ UTR of human CI-MPR mRNA (GenBank: NM_000876.3) contains two potential miR-143-binding sites with one highly conserved across different species, including humans, mice (B3 in Figure 2A), elephants, and opossums. The direct interaction between mature miR-143 and the Ci-mpr 3′ UTR (GenBank: NM_010515.2) demonstrated by site-directed mutagenesis and luciferase reporter assays suggests a possible modulation of CI-MPR by miR-143 in humans as well. Third, using human ECs, we showed that manipulation of miR-143 levels via LVs reversely controlled endogenous CI-MPR protein levels, leading to subsequent alteration of receptor-mediated transport of IDUA enzyme. Finally, genetic removal of miR-143 rescues CI-MPR expression and its biological function in mature BBB, with advanced CNS benefits in a murine MPS I model. Therefore, BBB-targeted manipulation of miR-143 would be translatable and beneficial for the treatment of patients with nLSDs.

The dynamic barrier function of BBB has restricted effective treatment for most neurological disorders from Alzheimer’s disease to rare nLSDs. Several strategies have been explored to either bypass the BBB with direct intracranial injection of therapeutic enzymes,55, 56, 57 or take advantages of receptors highly expressed on the BBB, such as insulin receptor,⁵⁸ transferrin receptor,⁵⁹ and low-density lipoprotein receptor-related protein 1,^13,38 by the addition of receptor-binding antibodies or peptides for adapted receptor-mediated transport. In this study, we discovered a novel mechanism contributing to the loss of effective CI-MPR-mediated transport, the natural pathway of most lysosomal enzymes,^4,17 in the BBB during development. Moreover, we document functional rescue of CI-MPR-mediated IDUA transport in human vascular ECs by downregulation of miR-143 with miR-143-sponge sequences. We also show advanced CNS therapeutic benefits with correction of brain GAG accumulation as well as CNS pathologic, phenotypic normalization in MPS I mice with miR-143 depletion (MPS/miR-143KO). Various delivery systems have been evaluated to modulate miRNAs with locked nucleic acid-modified oligonucleotides, peptide nucleic acids, or viral vectors both in basic research and clinical trials for cancer treatment.^60,61 The prospect of utilizing miRNA inhibitors may open a door for a novel CNS-targeted, CI-MPR-mediated protein delivery approach that would circumvent the implausibility of overexpressing CI-MPR via any gene delivery platform due to its large size (∼8 kb of coding sequence). Such an approach can be combined with current gene, cell, and enzyme therapies and be applicable in treating many nLSDs, such as Fabry disease, Pompe disease, and mucopolysaccharidoses.

In conclusion, our data not only provide valuable insights into the maturation of the BBB with the discovery of miR-143 as an important modulator for the decline of a transporter on the BBB during development, but they also constitute a novel proof of concept that depletion of a single miRNA is sufficient to achieve long-term CNS symptomatic correction in a murine model of nLSD treated with systemic delivery of moderate, clinically relevant levels of therapeutic protein.

Materials and Methods

Cell Lines and Animals

HUVECs and hCMEC/D3 cells were maintained in endothelial growth medium-2 MV (EGM-2 MV) (Lonza, Basel, Switzerland), which consists of endothelial basal medium-2 (EBM-2), 5% fetal bovine serum (FBS), and the supplemental growth factors vascular endothelial growth factor (VEGF), basic fibroblast growth factor (bFGF), epidermal growth factor, and insulin-like growth factor in concentrations as suggested by the manufacturer. HEK293T cells (ATCC: CRL-11268) were cultured in Dulbecco’s modified Eagle’s medium (DMEM) (Thermo Fisher Scientific, MA, USA) with 10% FBS and 1% penicillin/streptomycin. A mouse model with a homozygous deletion of the miR-143 gene was generated as previously described.²⁶ The MPS I (B6.129-Idua^tm1Clk) and WT C57BL/6 mice were purchased from The Jackson Laboratory.⁶² The MPS I/miR-143KO (MPS/143KO) mouse was established from heterozygous mating pairs. All animal procedures were performed in a pathogen-free facility at the Cincinnati Children’s Research Foundation (CCRF) in the vivarium fully accredited by the Association for the Assessment and Accreditation of Laboratory Animal Care (AAALAC) and approved by the Institutional Animal Care and Use Committee at CCRF.

Animal Perfusion

Under deep anesthesia with sodium pentobarbital (intraperitoneal [i.p.], 40 mg/kg) (Abbott Laboratories, IL, USA), each mouse underwent transcardiac perfusion with sterile, ice-cold 1× PBS for 5 min as previously described.¹³ The success of this procedure was confirmed by a loss of color in the liver and the blood vessels that flank the midline of the rib cage. The brain was then quickly removed and dissected for BrMV isolation and/or fixation.

BrMV Isolation

Murine cerebral microvessels were isolated as previously reported with modifications.²² Briefly, 8–12 brain samples from well-perfused animals were pooled and emulsified in 8–12 mL of stock buffer (25 mM HEPES, 1% dextran [molecular weight (MW) of ∼70,000 Da; Sigma-Aldrich, MO, USA] in minimum essential medium) on ice with a glass tissue grinder. After homogenization, the even mixture was sequentially filtered through 200-μm (once) and 100-μm (twice) nylon mesh (Thermo Fisher Scientific), then mixed with an equal volume of 40% dextran in stock buffer followed by centrifugation at 3,500 × g for 15 min at 4°C. The top layer was washed with ice-cold PBS and collected as capillary-depleted brain (CDB). The pellet on the bottom was collected and suspended in stock buffer and filtered through 25-μm nylon mesh (BioDesign, NY, USA). Brain microvessels retained on the mesh were washed down with stock buffer and centrifuged at 5,000 × g for 15 min at 4°C. Pellets were collected as BrMV.

miRNA Microarray Analysis

miRNA-enriched RNA fractions from BrMV samples were isolated as described⁶³ using the miRNeasy kit (QIAGEN, Hilden, Germany). RNA quantity and quality were determined using a NanoDrop ND-1000 spectrophotometer (Thermo Fisher Scientific) and an Agilent bioanalyzer (Agilent Technologies, CA, USA). Microarray was performed by sample hybridization on a μParaflo microfluidic chip (LC Sciences, Houston, TX, USA), which detects 1,265 murine mature miRNA transcripts listed in Sanger miRBase release 19.0 (http://www.mirbase.org). Data were normalized and preprocessed by a service provider (LC Sciences). Each pooled BrMV from either the same WT littermate or multiple litters was considered as a biological replicate. Two biological replicates were used, and each sample represents isolation performed on different dates.

Quantification of miRNAs/mRNAs with Quantitative Real-Time PCR

In general, total RNA or an miRNA-enriched fraction from BrMV or cell lines was extracted by an RNeasy kit (QIAGEN) or miRNeasy mini kit (QIAGEN) according to the manufacturer’s manual. For quantification of mature miR-143 and miR-145a, TaqMan miRNA assays (Invitrogen, CA, USA) were used as previously described.⁶⁴ Briefly, cDNA was synthesized using miRNA-specific (stem-looped) primers following the recommendations of the TaqMan miRNA reverse transcription kit (Invitrogen). For quantification of CI-MPR, Pecam1, and Cldn5 transcripts, cDNAs were generated from total RNA using a high-capacity cDNA reverse transcription kit (Thermo Fisher Scientific). PCR was performed by using the following primers: human CI-MPR mRNA, sense, 5′-GAGCTACGCTCATCACCTTTCTC-3′, antisense, 5′-TGGTGTACCACCGGAAGTTGT-3′; mouse Ci-mpr mRNA, sense, 5′-CACACTGATTACCTTCCTCTGTGA-3′, antisense, 5′-GGTGTACCACCGGAAGTTGTAG-3′; and a common probe (for both), 5′-FAM-ATATCAGGAAGAGGACAACT-3′. TaqMan primer/probe mixtures for Pecam1, Cldn5, and Actb were purchased from Invitrogen. Quantitative miRNA/mRNA expression data were acquired and analyzed using an ABI 7900 (Applied Biosystems, CA, USA) based on the protocol provided by the manufacturer. The PCR amplification condition included 2 min at 50°C and 10 min at 95°C, followed by 40 cycles of 15 s at 95°C and 1 min at 60°C. U6 small nuclear RNA (snRNA) (Invitrogen) was used as an endogenous control for miRNA expression analysis, while Atcb (for purity analysis of BrMV samples) or Gapdh (for CI-MPR-related quantification) were applied as internal controls for mRNA quantification. Quantification of Ci-mpr copy numbers was conducted using standard curves generated with a plasmid-containing Ci-mpr cDNA fragment. Relative expression of mRNAs or miRNAs was determined by the ΔΔC_t calculation method.

Protein Preparation and Western Blot

Total proteins were prepared from BrMV, ECs, or mouse organs using a prechilled radioimmunoprecipitation assay (RIPA) buffer with proteinase inhibitor cocktails (Santa Cruz Biotechnology, CA, USA). Proteins were separated and electrotransferred onto a polyvinylidene fluoride (PVDF) membrane (Millipore, MA, USA). After blocking with 5% skimmed milk, the membranes were incubated overnight at 4°C with primary antibodies, including anti-CI-MPR (purified rabbit anti-human and mouse CI-MPR, D3V8C, 1:100) (Cell Signaling Technology, MA, USA), anti-β-tubulin (H-235, 1:100, Santa Cruz Biotechnology), or anti-GAPDH (D16H11, 1:100, Cell Signaling Technology). Secondary antibodies (IRDye 680RD- or 800CW-conjugated immunoglobulin G [IgG] (H+L), LI-COR Biosciences, NE, USA) were used at a dilution of 1:5,000. The membrane was then exposed and visualized using a LI-COR Odyssey 9120 infrared imaging system (LI-COR Biosciences). The gray density of blot bands was analyzed by ImageJ/Fiji (National Institutes of Health, Bethesda, MD, USA).

UTR Reporter Assay

A Ci-mpr 3′ UTR fragment (1,304 bp, WT) containing predicted miR-143 binding sites was amplified from mouse genomic DNA and cloned into the pEZX-MT01 dual Fluc/Renilla luciferase reporter vector (GeneCopoeia, MD, USA). Meanwhile, the seed region in each predicted binding site was mutated to remove all complementarity to the first 1–7 (site 1) or 2–8 nt (sites 2 and 3) of miR-143 using the a QuikChange site-directed mutagenesis kit (Stratagene, CA, USA) to generate single mutants Mut1, Mut2, and Mut3 and triple mutant Mut123 (primer sequences are available upon request). Both WT and mutant 3′ UTRs were co-transfected with or without cytomegalovirus (CMV)-driven miR-143 or miR-145a overexpression constructs (pCMV6-miR-143 or pCMV6-miR-145a) into HEK293T cells in 12-well plates using Lipofectamine 2000 (Invitrogen). Cells were lysed 48 h after transfection, and the ratios between Fluc and Renilla luciferase activity were determined with a dual-luciferase assay system (Promega, WI, USA). Renilla luciferase was used as an internal control for normalization of transfection efficacy.

Manipulation of miR-143 in Human Vascular ECs

We generated a miR-143 sponge vector by taking advantage of antagomir and bulged miRNA sponge designing principles as described previously.⁶⁵ Oligonucleotides (sense, 5′-CTAGAGAGCTACAGACGTCATCTCACGATGAGCTACAGTGGATCCTTCATCTCAT-3′, antisense, 5′-CTAGATGAGATGAAGGATCCACTGTAGCTCATCGTGAGATGACGTCTGTAGCTCT-3′) were annealed, ligated, and cloned into XbaI linearized LV backbone LV-TW³⁸ downstream of the GFP, which was driven by a spleen focus-forming virus (SFFV) promoter and served as a marker for cell sorting. Sequencing data showed four unidirectional inserts, in a total of eight miR-143 binding sites in LV-143Spg construct.. For overexpression of miR-143, the CMV-miR-143 cassette was amplified from pCMV6-miR-143 vector and constructed into the same LV backbone. Lentivirus was generated by a four-plasmid packaging system and concentrated through ultracentrifugation, as previously described.⁶⁶ Virus particles were then used to transduce HUVECs and hCMEC/D3 cells in the presence of 8 μg/mL polybrene (Sigma-Aldrich). miR-143-manipulated ECs and parental vector-transduced controls were selected through FACS based on GFP expression to generate stably transduced cell populations.

Binding and Internalization of IDUA Enzyme in Vascular Cells

Binding and internalization of IDUA enzyme was performed as previously described with modifications.¹³ Briefly, stably transduced vascular ECs (5 × 10⁵ per well) were seeded in 12-well plates 24 h before assays. For specific inhibition of CI-MPR-mediated binding/internalization, mannose 6-phosphate (M6P, 1,500 μM) (Sigma-Aldrich) was added 30 min before and during the incubation. Pre-treated or untreated cells were then exposed at 4°C for 20 min to IDUA containing EBM-2 basal medium (∼20,000 U/mL), which was obtained by preculture with IDUA-overexpressing HEK293 cells. ECs were then cultured in fresh medium at 37°C for 1 h after multiple washes with ice-cold PBS. IDUA activity was assayed in cell pellets. Each experiment was performed in triplicate wells. Uptake IDUA activity was calculated by subtracting the endogenous IDUA in each cell line.

Quantification of IDUA

The catalytic activity of IDUA was determined as previously described⁶⁶ with modifications. Cells or tissue samples were homogenized and ultrasonicated in lysis buffer (150 mM NaCl and 50 mM Tris-HCl with 1% Triton X-100 [pH 7.6]). Clear lysate, plasma, or culture media were incubated with 2.5 mM fluorogenic substrate (4-methy-lumbelliferyl [4MU] α-l-idopyranosiduronic acid sodium salt; Toronto Research Chemicals) in 0.4 M sodium formate buffer (pH 3.2) at 37°C for various times, followed by the addition of 0.1 M glycine carbonate buffer (pH 11.0) to stop the reaction. The fluorescent product released in the reaction was detected using a SpectraMax M2 fluorometer plate reader (MDS Analytical Technologies, CA, USA) at an excitation wavelength of 365 nm and an emission wavelength of 450 nm, where 1 U of enzyme activity is defined as the release of 1 nmol 4MU in a 1-h reaction at 37°C. IDUA catalytic activity was presented as U/mL in plasma or medium, or U/protein amount, which was determined by a Pierce bicinchoninic acid (BCA) protein assay (Thermo Fisher Scientific). All specimens were assayed in duplicate reactions and quantified in triplicate wells in parallel with buffer controls.

CI-MPR and LAMP2 Antibody Internalization Assay

Freshly isolated BrMV samples from WT or miR-143KO mice were incubated with warm serum-free and growth factor-free EBM-2 basic medium. Internalization was performed by adding 1.8 μg/mL rabbit anti-CI-MPR (purified rabbit anti-human and mouse CI-MPR, D3V8C) alone or together with 10 μg/mL rat anti-LAMP2 (ABL-93, Abcam, Cambridge, UK) at 37°C for 1 h. Cells were then washed and cytospun onto glass slides at a speed of 1,000 rpm for 5 min. After fixation in 4% paraformaldehyde (PFA) for 30 min and permeabilization with 0.3% Triton X-100/PBS, slides were blocked and incubated with primary Ab (LAMP2, if not used in the internalization step) or secondary antibody (Ab) (Alexa Fluor 568-conjugated Ab for CI-MPR; Alexa Fluor 647-conjugated Ab for LAMP2) (Thermo Fisher Scientific) and fluorescein-labeled Lycopersicon esculentum lectin (1:100) (Vector Laboratories, CA, USA). After multiple washes, slides were mounted with Vectashield anti-fade medium containing DAPI (Vector Laboratories). Micrographs were obtained on a Nikon Ti-E inverted fluorescence microscope equipped with a 60× oil immersion objective lens, and an additional 1.5× digital zoom was applied. All BrMV from two litters of day ∼10 pups (n = 12–15) or adult mice (n = 10–12) were pooled and considered as a biological replicate. Three biological replicates were used.

Short-Term In Vivo Gene Transfer by HTV Injection

A total of 50 μg of plasmid expressing IDUAmyc driven by a liver-specific promoter was intravenously (i.v.) injected (in 0.9% NaCl solution) into the tail vein of MPS I or MPS/143KO mice (∼6 weeks old) in a volume equivalent to 10% (v/w) of body mass during a period of 5–8 s by using a 26G insulin syringe (Thermo Fisher Scientific). Mice were monitored for 5 min after the injection to ensure the recovery to normal activity. Mice were bled periodically (day 1 and day 2) from the tail vein to monitor plasma IDUA activity. Forty-eight hours after injection, mice were anesthetized and transcardially perfused with cold PBS and harvested as described above.

HSC-Mediated LV GT

Long-term IDUA delivery via LV TW-IHK-IDUAmyc-IRES-GFP, driven by a lineage-specific promoter (P_IHK), were performed as previously described.^36,38 Briefly, LVs were generated by a four-plasmid transfection system and concentrated by ultracentrifugation. The titer of viral stocks was determined in murine erythroid leukemia (MEL) cells by FACS analysis for GFP%. Lineage-depleted low-density bone marrow cells (Lin⁻) were isolated from MPS I mice via negative selection and pre-stimulated in serum-free StemSpan medium (STEMCELL Technologies, Vancouver, BC, Canada) that was supplemented with 50 ng/mL stem cell factor, 20 ng/mL thrombopoietin, 10 ng/mL interleukin (IL)-3, and 25 ng/mL Flt3 for 24 h. Lin⁻ cells were then transduced with concentrated LV twice with a total multiplicity of infection of 20 within 24 h in the presence of 8 μg/mL polybrene. LV-transduced Lin⁻ cells were transplanted into lethally irradiated (split dosages of 700 and 475 cGy) MPS I mice or MPS/143KO mice (8–10 weeks old, 10⁵ cells/mouse). Transgene frequency was periodically monitored via FACS analysis for GFP expression in different blood lineages (platelets, RBCs, and white blood cells).

Quantification of GAGs

Perfused and dissected cortical parenchyma were homogenized with 10% (v/w) ice-cold double-distilled H₂O (ddH₂O). The raw protein concentration was measured by using a Pierce BCA protein assay (Thermo Fisher Scientific). Equivalent amounts of proteins (0.5 mg) per mouse were defatted by incubating with 0.5 mL of chloroform/methanol (1:2) solution at room temperature (RT) for 3 h then washed with 100% ethanol. Pellets were re-suspended and ultrasonicated (40 s/four times on ice) in papain buffer (0.1 M sodium acetate [pH 5.5], 5 mM EDTA, 5 mM l-cysteine [pH 5.5]). The protein assay was performed again by using a Bradford protein assay (Bio-Rad, CA, USA). To release the soluble GAGs, we digested samples with 100 mM papain at 65°C for 3 h, followed by DNA digestion through incubation with DNase (1 U/μL) for 30 min at 37°C. Free GAGs in solution were quantified in triplicate by reacting with 1,9-dimethylmethylene blue chloride dye as previously described.⁶⁷ The optical density (OD) value of the color reaction was measured at 656 nm by using a SpectraMax M2 fluorometer plate reader. All GAG values were calculated by comparing with a standard curve generated with heparan sulfate (Sigma-Aldrich), and normalized to the amounts of protein determined in the protein assay.

IF Staining

Brains from well-perfused mice were dissected and postfixed by immersion in 4% PFA at 4°C overnight, cryopreserved in 30% sucrose (w/v), and frozen in Tissue-Tek CRYO-OCT (Thermo Fisher Scientific). Frozen sections (10 μm) or BrMV cytological smears were fixed in 4% PFA and permeabilized in 0.3% Triton X-100 solution and blocked with 5% goat serum before staining with the following primary Abs: mouse c-Myc (myc) Ab (1:50) (Santa Cruz Biotechnology, TX, USA), rabbit anti-NeuN (1:100, Cell Signaling Technology), rat anti-CD68 (1:100, Thermo Fisher Scientific), rabbit anti-CI-MPR (1:50), and rat anti-LAMP2 (1:100), followed by incubation with Alexa Fluor 488/568-conjugated secondary antibodies (1:500, Invitrogen) or with fluorescein-conjugated Lycopersicon esculentum lectin (Vector Laboratories), a versatile endothelial marker for the CNS.⁶⁸ Slides were mounted with Vectashield DAPI-containing medium (Vector Laboratories) and visualized under a Nikon Ti-E inverted fluorescence microscope. For IDUA distribution analysis after HTV and LV-GT, brain sections (with comparable IDUA activities in CDB) per genotype from three independent experiments were analyzed. In HTV brains, the percentages of c-myc-positive neurons (NeuN⁺) or tissue macrophages (CD68⁺) were counted and calculated from more than 500 NeuN⁺ or CD68⁺ cells derived from four to six views on each non-consecutive section. For spectrophotometric quantification of CI-MPR and LAMP2 signals, mean fluorescence density in vascular cells was measured with ImageJ/Fiji (NIH, MD, USA) by outlining vascular profiles following lectin staining. More than 100 BrECs were quantified per genotype.

Brain Pathology Analysis

Brain pathology was assessed by toluidine blue staining as previously described.³⁸ Briefly, forebrain cortex was fixed by 3% glutaraldehyde in 0.175 mol/L sodium cacodylate buffer (pH 7.4) at 4°C. The tissues were then treated with 1% osmium tetroxide, washed in 0.175 mol/L sodium cacodylate buffer, dehydrated by a graded ethanol series, and embedded in LX112 resin (Ladd Research Industries, VT, USA). Sections (0.5–1 μm) were prepared and stained with 1% toluidine blue, followed by examination for the presence of pathological storage vacuoles. For brain pathology scoring, the number and percentage of brain vasculatures that were associated with vacuolated perivascular cells were determined. A minimum of 500 vasculatures per mouse and more than three mice per group were analyzed with six sections randomly selected. The percentages of all sections were contributed to statistical analysis.

Repeated Open-Field Test

The repeated open-field test was performed as described previously.^36,40 Five months after bone marrow transplantation, the test was performed using an open-field test apparatus (60 × 60 cm) consisting of a white Plexiglas box with 25 squares (12 × 12 cm) painted on the floor (16 outer and 9 inner). Mice were placed in one of the four corners of the apparatus and allowed to explore the whole field freely for 5 min. Activity was monitored and quantified blind by two observers for horizontal activity (number of inner and outer squares crossed) and grooming activity (time). Each mouse was tested in three repeated trials with a 30-min inter-trial interval and changes between trials were calculated.

Statistical Analysis

Datasets were analyzed with GraphPad Prism 6 (GraphPad, CA, USA) either by a two-tailed unpaired Student’s t test or by one-way or two-way ANOVA as specified in each figure. For ANOVA, post hoc comparisons were performed using Tukey’s multiple comparison test. Data are presented as mean ± SD or SEM as individually specified. The exact or adjusted p values at <0.05 by either a Student’s t test or ANOVA were considered as significant.

Author Contributions

D.P. conceived the study and obtained financial support. Y.L. and D.P. participated in the study design and were responsible for project management. Y.L., X.W., K.P.R., M.D., and J.H. performed the laboratory experiments and data analysis. Y.L. and D.P. wrote the manuscript. M.X. provided miR-143KO mice and helpful discussions. Y.L. and D.P. conducted the data processing, statistical analysis, and bioinformatics investigations. The manuscript was finalized by D.P. with the assistance of all of the authors.

Conflicts of Interest

The authors declare no competing interests.