Systemic and Central Nervous System Correction of Lysosomal Storage in Mucopolysaccharidosis Type VII Mice

Abstract

Mucopolysaccharidosis (MPS) type VII patients lack functional β-glucuronidase, leading to systemic and central nervous system dysfunction. In this study we tested whether recombinant adenovirus that encodes β-glucuronidase (Adβgluc), delivered intravenously and into the brain parenchyma of MPS type VII mice, could provide long-term transgene expression and correction of lysosomal distension. We also tested whether systemic treatment with the immunosuppressive anti-CD40 ligand antibody, MR-1, affected transgene expression. We found substantial plasma β-glucuronidase activity for over 9 weeks after gene transfer in the MR-1- treated group, with subsequent decline in activity corresponding to a delayed anti-β-glucuronidase antibody response. At 16 weeks, near wild-type amounts of β-glucuronidase activity and striking reduction of lysosomal pathology were detected in livers from mice that had received either MR-1 cotreatment or control antibody. In the lung and kidney, β-glucuronidase activity was markedly higher for the MR-1-treated group. β-Glucuronidase activity in the brain persisted independently of MR-1 treatment. Activity was intense in the injected hemisphere and was also evident in the noninjected cortex and striatum, with dramatic improvements in storage deposits in areas of both hemispheres. These results indicate that prolonged enzyme expression from transgenes delivered to deficient liver and brain can mediate pervasive correction and illustrate the potential for gene therapy of MPS and other lysosomal storage diseases.

The mucopolysaccharidoses (MPS) are a group of lysosomal storage diseases, each caused by a deficiency in one of the lysosomal acid hydrolases. The result is buildup of glycosaminoglycans (GAGs) and dysfunction of multiple tissues including those of the central nervous system (CNS) (24). MPS type VII (Sly syndrome) patients lack functional β-glucuronidase. A β-glucuronidase-deficient mouse strain (5, 29) has been used to test enzyme- (25, 31, 40), cell- (3, 4, 28, 30, 34), and gene-based (13, 20, 21, 23, 26, 37, 43, 44) therapies. β-Glucuronidase secretion and uptake pathways allow for cross correction (36). In adult mice, peripherally administered enzyme (31), bone marrow transplantation (4), or implantation of β-glucuronidase-producing neo-organs (23) supplies corrective levels of enzyme systemically but the CNS remains diseased. Because the half-life of recombinant β-glucuronidase in tissues is only a few days (40), direct enzyme treatment in the CNS would require either repeated invasive bypass of the blood-brain barrier or delivery of enzyme through an implanted intrathecal catheter. Introduction of sequences encoding β-glucuronidase is an attractive alternative.

We recently reported that correction in CNS pathology occurs 3 weeks after injection of recombinant adenovirus that encodes human β-glucuronidase (Adβgluc) into the brain parenchyma of adult MPS type VII mice (13). Since MPS pathology is not confined to the CNS, alleviation of widespread disease would require injection of Adβgluc systemically as well as into the brain. However, immune responses to systemically administered adenovirus vectors (2, 9, 46, 47) or their secreted transgene products (38, 49) have been shown to limit the effectiveness of peripheral gene transfer. Moreover, investigations with mice (7) and rats (6) found that transgene expression in the CNS declined rapidly upon subsequent peripheral exposure to the same vector. We hypothesized that transient immunosuppression with the anti-CD40 ligand antibody, MR-1, might improve the therapeutic efficacy after combined systemic and brain Adβgluc injections. CD40-CD40 ligand intercellular interactions are necessary for T-cell-dependent humoral immune responses and for up-regulation of costimulatory molecules critical for T-cell activation (reviewed in reference 19). Previous studies by our group (35) and others (17, 32, 42, 48) illustrate the utility of in vivo blockade of CD40-CD40 ligand interactions at the time of vector injection for inhibiting antibody and cell-mediated responses.

To test our hypotheses, we injected Adβgluc (expressing human β-glucuronidase) both intravascularly and into the brains of MPS type VII mice, with or without cotreatment with MR-1. At 16 weeks, tissues were analyzed for transgene expression and correction of lysosomal defects.

Antibody responses are inhibited and β-glucuronidase is detected in plasma after MR-1 treatment.

MPS type VII (gus^mps/gus^mps) mice (derived from the C57BL/6 strain) (The Jackson Laboratory, Bar Harbor, Maine) at 6 to 8 weeks of age were injected with 2 × 10⁹ and 2 × 10⁷ PFU of Adβgluc into the lateral tail vein and the right striatum of the brain (13), respectively. Negative control MPS type VII mice received either no virus or recombinant adenovirus that encodes Escherichia coli β-galactosidase (Adβgal). Adβgluc (20) and Adβgal (27) are derived from human adenovirus serotype 5, with deletions in the E1 region that render them replication defective. Transcription is directed by the Rous sarcoma virus long terminal repeat. The anti-CD40 ligand monoclonal antibody, MR-1 (purified as described in reference 35), was injected into the peritoneum (500 μg per dose) on days −1, 0, 1, 2, 4, 6, 9, and 12 relative to virus injection. Mice that were not given MR-1 received analogous injections of hamster gamma globulin (control immunoglobulin) (Jackson Immuno Research Laboratories Inc., West Grove, Pa.). Enzyme-linked immunosorbent assay (35) of plasma samples showed that MR-1 effectively delayed the generation of anti-Ad immunoglobulin G (IgG). On day 41, mice given control immunoglobulin had plasma anti-Ad IgG concentration of 20,000 ng/ml, whereas negligible amounts were detected in MR-1-treated mice. However, 109 days after Adβgluc injection, the anti-Ad concentration in plasma of MR-1-treated mice was 4,000 ng/ml. For determination of anti-β-glucuronidase IgG concentrations in plasma samples, an antibody capture assay (33a) was used. Briefly, plasma samples were incubated with protein G-conjugated Sepharose beads and human β-glucuronidase (generously provided by William Sly) overnight at 4°C, washed by centrifugation, resuspended in 0.2% acetic acid, and assayed for β-glucuronidase activity by fluorometry (13, 14). Similar to its effect on the anti-Ad IgG response, MR-1 treatment blocked the generation of anti-β-glucuronidase IgG for several weeks (Fig. 1A), during which time β-glucuronidase (determined by fluorometric assay) was present at high levels in the plasma of only the MR-1-treated mice (Fig. 1B). The disappearance of β-glucuronidase from plasma coincided with the appearance of anti-β-glucuronidase IgG, suggestive of antibody-mediated clearance.

FIG. 1.

Anti-β-glucuronidase antibody and β-glucuronidase activity levels in plasma. MPS type VII mice were injected with Adβgluc on day 0 and with MR-1 or control immunoglobulin on days −1, 0, 1, 2, 4, 6, 9, and 12. Plasma samples obtained at the indicated time points were analyzed for anti-β-glucuronidase IgG (A) and for β-glucuronidase activity (B). Anti-β-glucuronidase IgG concentrations are expressed as units per milliliter of plasma, which refers to the β-glucuronidase-capturing capacity of the plasma IgG. n was 3 to 5 for Adβgluc+control immunoglobulin-injected ice; n was 5 or 6 for Adβgluc+MR-1-treated mice. Error bars represent standard errors of the means.

The MR-1 dosing regimen was based on studies that showed that the presence of MR-1 at the time of antigen exposure, encompassing the time of peak CD40 ligand expression in the spleen (3 to 4 days after antigen exposure) (39), effectively blocked antibody synthesis and memory B-cell generation (10, 11). In a previous investigation, using an MR-1 injection schedule very similar to that used in the present study, we successfully blocked anti-Ad responses at least until day 63 (the last time point tested) (35). Other investigators used fewer injections containing lower concentrations of MR-1 to prevent humoral responses to injected protein antigens (10, 11). However, with a virus vector there is the potential for continuous synthesis of foreign antigen by transduced cells, including dendritic cells (16). The delayed antibody responses detected after 9 weeks in the present study were likely a result of antigenic stimulation extending beyond the clearance time of the MR-1 (11).

β-Glucuronidase synthesis persists and corrects pathology in the liver.

Sixteen weeks after Adβgluc injection, the mice were sacrificed and β-glucuronidase activities in tissue lysates of liver, lung, and kidney were quantitated by fluorometric assay. Activities in liver lysates were similar between mice given MR-1 or control immunoglobulin (88 and 82% of wild-type activity, respectively), indicating near-complete restoration of enzyme activity in this organ. Activity in the lungs of Adβgluc-injected mice was 141% of wild-type activity for the MR-1-treated group but only 15% of wild-type activity for the control immunoglobulin-treated mice. Activity in the kidney was 18% of wild-type activity for the MR-1-treated group but less than 2% of wild-type activity for control immunoglobulin-treated animals. Wild-type (normal C57BL/6 mouse) values were 202, 310, and 287 U of activity per mg of protein in liver, lung, and kidney lysates, respectively.

An enzyme-based histochemical stain (1, 13) was used to detect β-glucuronidase in cryosections of liver and kidney 16 weeks after Adβgluc injection (Fig. 2). The intensity of staining and extent of positive cells corresponded with enzyme levels determined by the fluorometric assay. In liver sections, numerous β-glucuronidase-positive cells were detected for both control immunoglobulin- and MR-1-treated mice (Fig. 2B and C, respectively). In kidney sections, positive cells within glomeruli were detected for both Adβgluc-injected groups (Fig. 2F and G), but the staining was more intense for the MR-1-treated group (Fig. 2G). Moreover, for the control immunoglobulin-treated mice, β-glucuronidase activity in the cortex of the kidney was restricted to glomeruli, while for the MR-1-treated mice β-glucuronidase activity was also seen in cortical tubule cells. Negative control sections from livers and kidneys of MPS type VII mice (Fig. 2A and E, respectively) showed no staining, and positive control sections from normal mice (Fig. 2D and H, respectively) showed obvious positive areas. In situ hybridization for human β-glucuronidase mRNA indicated that expressing cells were distributed throughout the liver sections from Adβgluc-injected mice, while positive cells were detected rarely in the lung and not at all in the kidney (not shown).

FIG. 2.

Histochemical detection of β-glucuronidase in liver and kidney sections 16 weeks after Adβgluc injection. MPS type VII mice were injected with Adβgluc, with or without MR-1 treatment. Sixteen weeks after Adβgluc injection, 10-μm-thick cryosections from liver (A, B, C, and D) and kidney (E, F, G, and H) tissues were histochemically stained for β-glucuronidase. A red precipitation product is formed in the presence of β-glucuronidase. No staining is observed in sections from negative control (naive) MPS type VII mice (panels A and E), while sections from Adβgluc+control immunoglobulin-injected mice (panels B and F) and Adβgluc+MR-1-injected mice (panels C and G) have intensely positive cells. The positive activity in sections from normal C57BL/6 mice is shown for comparison (panels D and H). Bar, 50 μm.

Tissues from mice sacrificed 16 weeks after gene transfer were also embedded in plastic for evaluation of lysosomal distension, a characteristic of this storage disease. Processing was as previously described (13) but sections were stained with Richardson’s stain (25 mM sodium borate, 0.5% methylene blue, 0.5% azure B in water) for 24 hours. Extensive lysosomal accumulations in Kupffer cells and hepatocytes were seen in liver sections from uninjected MPS type VII mice (Fig. 3A). In striking contrast, liver sections from Adβgluc-injected (Fig. 3B) and Adβgluc+MR-1-treated (Fig. 3C) mice, similar to those from normal mice (Fig. 3D), were largely devoid of storage bodies.

FIG. 3.

Effects of Adβgluc-mediated gene transfer on storage bodies in liver and kidney of MPS type VII mice. Thin sections (0.5-μm thickness) from livers and kidneys of MPS type VII mice sacrificed 16 weeks after Adβgluc injection were analyzed for distended lysosomes, identified as unstained intracellular vacuoles. The liver from an age-matched untreated MPS type VII mouse is notably affected (A), with greatly enlarged lysosomes in Kupffer cells (closed arrow) and numerous distended lysosomes in hepatocytes (open arrow). Adβgluc injection without (B) or with (C) MR-1 treatment reduced the size and number of vacuoles to levels of normal (C57BL/6) mice (D). For an untreated MPS type VII mouse, glomerular and cortical tubule cell lysosomal accumulations in the kidney are readily apparent (E). Injection of Adβgluc without MR-1 treatment (F) is only partially effective, while Adβgluc+MR-1 injection (G) restores the phenotype to near normal (H). Bar, 10 μm.

The kidney cortex of uninjected MPS type VII mice showed severe pathology (Fig. 3E). Unlike in the liver, in the kidney Adβgluc only partially corrected the pathology in the control immunoglobulin-treated group (Fig. 3F). However, complete correction was apparent in kidney sections from Adβgluc+MR-1-treated mice (Fig. 3G), where the tissue was similar to that from normal mice (Fig. 3H).

Together, these results suggest that MR-1 was beneficial and allowed corrective levels of β-glucuronidase to circulate for uptake at sites that are poorly transduced upon intravenous delivery, such as the kidneys and lungs. Because anti-β-glucuronidase antibody ultimately appeared in the MR-1-treated mice, entrapment of immune complexes by Kupffer cells likely occurred and reduced the levels of circulating enzyme to below detectable levels. The appearance of neutralizing antibodies illustrates a potential need for immunotherapy after systemic gene transfer when the foreign transgene product is capable of eliciting a humoral response, perhaps regardless of the vector used to deliver the gene.

The persistence of β-glucuronidase-expressing hepatocytes in MR-1-untreated mice was somewhat surprising. Early reports indicate that within a few weeks following intravenous injection of E1-deleted recombinant adenoviruses, transduced hepatocytes are eliminated by a cytotoxic T-lymphocyte (CTL)-mediated immune response (45). More recent data suggest that the generation of an effective CTL response (or lack thereof) is dependent on the combination of transgene product and mouse strain (15, 41, 49). For example, after adenovirus gene transfer to the liver, C57BL/6 mice exhibit transient expression of E. coli β-galactosidase (12) and human low-density lipoprotein receptor (18), but display long-term expression of human α1-antitrypsin (2, 22) and human very-low-density lipoprotein receptor (18). Our results indicate that after combined brain and intravenous injection into MPS type VII mice, recombinant adenovirus that encodes human β-glucuronidase elicits a humoral response, but not an effective CTL response. It is not yet clear, however, whether β-glucuronidase transgene persistence is due to the C57BL/6 background, low immunogenicity of the protein, or a possible immunological impairment in lysosome-laden tissues. An additional consideration is that coinjection into the CNS may alter the immune response to intravenously injected vector. This might explain why our results differ from those of Ohashi and colleagues (26). They injected a similar dose of recombinant adenovirus that encodes human β-glucuronidase into the tail veins of mice of the same MPS strain and detected only 20% of the wild-type value for β-glucuronidase activity in liver at 16 days and substantial loss by 32 days. Differences between promoters might also be a contributing factor.

β-Glucuronidase synthesis persists and corrects pathology in the brain.

At 16 weeks after systemic and CNS administration of Adβgluc with or without MR-1 cotreatment, brains were cut along the plane of the injection tract into frontal and caudal portions. The caudal portion was cryosectioned for histochemical analysis of β-glucuronidase activity and for in situ mRNA hybridization. The frontal portion was processed for evaluation of lysosomal distension. Substantial β-glucuronidase activity was found, regardless of whether MR-1 treatment was given. Figure 4A shows that intense enzyme activity was present in the injected hemisphere, particularly in the striatum, corpus callosum, and cortex, while the noninjected hemisphere had moderate activity in the striatum and cortex. Volumetric analysis of serial sections, performed as previously described (13), indicated that a larger volume of the brain was positive for enzyme activity for the MR-1-treated group than for the MR-1-untreated group, but this difference was not statistically significant (not shown). Thus, although immunosuppression may be essential for prolonged expression in other disease models or for human CNS gene therapy, it was not necessary for long-term CNS expression of β-glucuronidase from adenovirus vectors in this study.

FIG. 4.

Adβgluc-mediated gene transfer to brain results in long-term β-glucuronidase expression and improvements in storage disease. Sixteen weeks after adenovirus vector injection, 10-μm-thick coronal cryosections of the brain were histochemically stained for β-glucuronidase activity, and 0.5-μm-thick sections of plastic-embedded brain were analyzed for distended lysosomes. Histochemistry (A) shows intense β-glucuronidase activity in the injected hemisphere and moderate activity in the noninjected cortex and striatum. Thin sections from the striatum (B) and cortex (C) of age-matched Adβgal-injected MPS type VII mice show numerous distended lysosomes within cells. In sections from Adβgluc-injected mice, cells with reduced storage are present in the injected striatum (D), noninjected striatum (E), and noninjected cortex (F). Bar, 10 μm for panels B through F.

We have previously shown that wild-type levels of β-glucuronidase in the adult murine brain are below the detection limits of this histochemistry method (13). Therefore, areas of the brain with moderate or weak staining after gene transfer should have sufficient amount of enzyme for maintenance of GAG degradation and correction of lysosomal pathology. To test this, 16 weeks after Adβgluc injection brain samples from injected and contralateral hemispheres were analyzed for widespread correction. Uninjected, age-matched control MPS type VII mice had extensive and obvious lysosome-laden cells in the striatum (Fig. 4B) and cortex (Fig. 4C). In contrast, in Adβgluc-injected mice, lysosomal storage was reduced to near normal levels not only in the injected striatum (Fig. 4D) but also in the contralateral striatum (Fig. 4E) and in the cortex of both the injected (not shown) and noninjected (Fig. 4F) hemispheres. Transgene-expressing cells, as detected by mRNA in situ hybridization, were restricted to the striatum and corpus callosum of the injected hemisphere (not shown).

The presence of enzyme and correction in areas of the brain where mRNA was not detected indicates that transduced cells produced sufficient enzyme to reach and correct distant cells, including cells in the contralateral striatum. This is in contrast to our earlier study, in which neither enzyme nor correction was detected in the contralateral striatum 3 weeks after gene transfer (13). Together, the results of these studies imply that prolonged expression following focal vector delivery exposes an increasing proportion of distant regions of the brain to corrective levels of recombinant enzyme, probably by virtue of enzyme diffusion through cerebrospinal fluid and extracellular spaces, as well as along neuronal networks. This novel finding may apply to other vector systems capable of persistent expression in the brain.

Data presented show long-term expression and consequent widespread correction in peripheral organs and the CNS and provide promise for application of gene therapy for treatment of the MPS and other storage diseases. For specific application of adenoviruses, it will be important to determine in appropriate models how prior systemic exposure to a productive adenovirus infection, at doses mimicking those in human adenovirus infections, impacts on the efficiency and efficacy of adenovirus-mediated gene transfer. Finally, adenovirus vectors devoid of sequences that encode viral genes (33), in combination with modifications in capsid proteins to improve efficiency of infection (8), may provide for maximal stability of expression.