Transgene produces massive overexpression of human β-glucuronidase in mice, lysosomal storage of enzyme, and strain-dependent tumors

Abstract

β-Glucuronidase (GUSB) is a lysosomal enzyme important in the normal step-wise degradation of glycosaminoglycans. Deficiency of GUSB causes the lysosomal storage disease mucopolysaccharidosis VII (MPS VII, Sly disease). Affected patients have widespread progressive accumulation of β-glucuronide-containing glycosaminoglycans in lysosomes. Enzyme replacement, bone marrow transplantation, and gene therapy can correct lysosomal storage in the MPS VII mouse model. Gene therapy in MPS VII patients and animals may result in massive overexpression of GUSB in individual tissues, and the toxicity of such overexpression is incompletely investigated. To gain insight into the effect of massive overexpression of GUSB, we established 19 transgenic mouse lines, two of which expressed very high levels of human GUSB in many tissues. The founder overexpressing mice had from >100- to several thousand-fold increases in tissue and serum GUSB. The enzyme expression in most tissues decreased in subsequent generations in one line, and expression in liver and marrow fell in subsequent generations of the other. Both lines had morphologically similar widespread lysosomal storage of GUSB and secondary elevations of other lysosomal enzymes, a finding characteristic of lysosomal storage disease. One line developed tumors, and one did not. These transgenic models show that massive overexpression of a lysosomal enzyme can be associated with dramatic morphological alterations, which, at least in one of the two lines, had little clinical consequence. For the other transgenic line, the high frequency of tumor development in F₂ FVB progeny suggests that the vector used to generate the transgenic lines has an integration site-dependent potential to be oncogenic, at least in this strain background.

Most patients with lysosomal storage disease (LSD) have a deficiency of a lysosomal enzyme important in the normal stepwise degradation of a complex macromolecule (1). LSD in both humans and animal models can be effectively treated if relatively low levels of the deficient enzyme are delivered to the affected tissues, as occurs in bone marrow transplantation or enzyme replacement therapy (2–4). Transgenic mice that overexpress therapeutic proteins have been used as a source of donor cells to treat models of LSD, and it has been suggested that overexpressing cells may be more effective than normal cells for correction of lysosomal storage (5, 6). Viral-mediated gene therapy has also been shown to be therapeutic for murine mucopolysaccharidosis (MPS) VII and other animal models of LSD (7). Such therapy can produce high levels of therapeutic enzyme in individual tissues (8). The toxicity of such overexpression of a lysosomal enzyme is incompletely investigated, although a recent report described tumors in multiple adult MPS VII mice receiving neonatal injection of the human β-glucuronidase (HGUSB, EC 3.2.1.31) gene in an adeno-associated virus (rAAV) vector (9).

We established 19 transgenic mouse lines with integrated copies of HGUSB transgene to determine whether massive overexpression of HGUSB had an adverse effect on the clinical phenotype or tissue morphology. Two lines, WE1 and WE18, showed much higher tissue HGUSB levels than the other 17 transgenic lines. The overexpressing transgenes, which were bred onto the C57B/6 background, led to widespread lysosomal storage of the enzyme and secondary elevation of other lysosomal enzymes but a grossly normal phenotype. However, the F₂ FVB mice of one line developed tumors. That tumors occurred in relatively high numbers in only one of the two lines suggests an oncogenic potential for the vector depending on the integration site, at least in this strain of mice.

Materials and Methods

Generation of Transgenic Mice That Overexpress HGUSB Transgene.

The HGUSB cDNA (10) was inserted into the pCAGGS vector (11), which contains the CMV enhancer, the AG derivative of the chicken β-actin promotor, rabbit β-globin intron sequence, and SV40A poly(A) addition site. The cDNA fragment containing the promoter and HGUSB cDNA was isolated from the linearized plasmid and microinjected into male pronuclei of C57B/6/FVB hybrid or FVB oocytes, which were transferred to pseudopregnant Swiss Webster females. Nineteen transgenic founder mice were identified by serum enzyme levels and PCR of tail samples. Most of the founder mice and their tail-positive offspring expressed 10- to 50-fold normal levels of HGUSB in serum.

Two transgenic lines, WE1 and WE18, both of which overexpressed HGUSB at much higher levels, were established from founders. The WE1 mice were established from injected C57B/6/FVB hybrid eggs, and the WE18 mice were established from injected FVB eggs. The higher-expressing HGUSB transgenes in founders were bred initially onto the FVB background and subsequently crossed onto the C57B/6 background for more than 12 generations. Clinical observations were made in the colony kept at the transgenic facility at Saint Louis University School of Medicine. All mice were handled with the highest standards of humane animal care, and the Saint Louis University Institutional Animal Care and Use Committee, the applicable institutional review board at Saint Louis University, approved the study. Saint Louis University is fully accredited by the American Association for the Accreditation of Laboratory Animal Care.

DNA Analysis and Enzyme Assays.

Transgenic mice that overexpressed HGUSB were identified by tail DNA by using HGUSB-specific forward primer 5′-GCTGGTGAATTACCAGATCTCTGTCAA-3′ and reverse primer 5′-GGAAATAG-AAAGGTTTCCCATTGATGAGG-3′, which amplify a 284-bp fragment from DNA from mice carrying the human transgene. Lysosomal enzyme levels were quantitated in tissue and serum, as previously described (12).

Histological, Histochemical, and Immunohistological Analysis of Transgenic Mice.

Thirteen WE1 and 10 WE18 heterozygote mice from 2 to 19 mo of age and from backcrosses F₄–F₁₃ on the C57B/6 background were necropsied. Liver, spleen, kidney, brain, heart, bone, bone marrow, and smooth and skeletal muscle were examined by light or electron microscopy (13). Tissues were embedded in Spurr's resin or paraffin and stained with toluidine blue, hematoxylin/eosin, or periodic acid Schiff (PAS) stain (14). Frozen tissues were evaluated histochemically for GUSB activity using Naphthol AS-BI (6-bromo-2-hydroxy-3-naphthoic acid 2-methoxyanilide) β-d-glucuronide (12).

To immunolocalize HGUSB, kidney, liver, and skeletal muscle from a 1-yr-old WE1 mouse was fixed in 4% paraformaldehyde in 0.1 M phosphate buffer overnight at 4°C, dehydrated through a graded series of ethanol, and embedded in LR White resin. Blocks were polymerized at 55°C for 48 h. Ultrathin sections were cut and mounted on nickel grids and were incubated for 15 min at room temperature in 0.8% bovine serum/0.1% gelatin in PBS to block nonspecific antibody binding. Sections were then incubated in goat anti-HGUSB IgG (20 μg/ml) in the same blocking buffer overnight at 4°C, washed in several changes of PBS, and incubated in 10 nM of colloidal gold-labeled protein A/G (Polysciences) in blocking buffer for 2 h at room temperature followed by washing in several changes of phosphate buffer and in a final rinse of 0.1% Triton X-100 in phosphate buffer. Sections were then fixed for 1 min in 2% glutaraldehyde, washed several times in distilled water, and counterstained with uranyl acetate. The sections were examined with a JEOL 100CX transmission electron microscope. As a negative control, normal nonimmune goat IgG was substituted for the goat anti-HGUSB.

Eight female WE18 F₂ FVB mice over 1 yr of age developed palpable s.c. tumors, and one 5-mo-old female WE18 F₂ FVB mouse had abnormal circling behavior. Because of these clinical findings, these mice were killed and necropsied, and their tumors were evaluated pathologically. A single 16-mo-old male WE18 C57B/6 F₁₀ mouse had a tumor identified incidentally at necropsy for evaluation of storage. None of the WE1 mice on the FVB or C57B/6 background had clinically apparent tumors or tumors identified at necropsy for evaluation of lysosomal storage.

Results

Massive HGUSB Overexpression in Transgenic Mice.

Of 19 independent transgenic founders generated, 17 had 10- to 50-fold normal serum GUSB levels. Two founders had tail-positive offspring with serum levels >1,000-fold normal, and these mice were selected for breeding and study. Table 1 presents tissue and serum HGUSB levels in WE1 and WE18 mice expressed as the fold increase over normal. The HGUSB levels were enormously increased in both WE1 and WE18 mice compared with normal mice. Cardiac and skeletal muscle HGUSB levels were the highest of all of the tissues evaluated in WE1 mice. In WE18 mice, heart, liver, bone marrow, and skeletal muscle HGUSB were increased >1,000-fold in the initial FVB mice, but HGUSB levels in all tissues examined fell over successive generations of backcrosses onto the C57B/6 strain, particularly in the bone marrow. In WE1 mice, only liver and bone marrow HGUSB fell over subsequent generations.

Table 1.

β-GUSB levels in WE1 and WE18 transgenic mice expressed as fold increase over normal B6 mice

Tissue	WE1 FVB	WE1* F6,7†	WE1* F10†	WE18 FVB	WE18* F6,7†	WE18* F10†
Liver	83	30	7	3,586	274	204
Kidney	149	231	150	509	201	173
Heart	8,500	8,406	8,597	6,619	737	919
Brain	314	616	594	708	127	132
Muscle	1,294	3,810	2,618	1,436	1,095	680
Marrow	88	114	11	1,720	208	18
Serum	9,250	1,225	2,085	12,650	1,962	1,978

^*Mean value based on two mice. 

^† The F_6,7 and F₁₀ mice refer to products of 6–10 backcrosses of the respective transgene onto the C57B/6 background. 

HGUSB Overexpression Causes Elevation of Other Lysosomal Enzymes.

“Secondary elevation” of lysosomal enzymes other than the deficient enzyme is common in tissues of patients with LSD and in murine MPS VII (12). Modest secondary elevation of βhexosaminidase and α-galactosidase was seen in F₁₀ mice of both WE1 and WE18 transgenic lines (Table 2). Cardiac α-mannosidosis and α-galactosidase levels were increased in WE1 mice, and there was no elevation of acid phosphatase in either line (data not shown).

Table 2.

Secondary elevation of lysosomal enzymes in F₁₀ WE1 and WE18 mice, expressed as units per milligram of protein

Tissue	Normal	WE1	WE18
β-Hexosaminidase
Liver	1,303	2,664	2,063
Kidney	2,306	2,073	2,264
Heart	305	1,926	524
Brain	2,981	3,602	3,213
Muscle	100	759	193
Marrow	781	648	835
Serum	4,675	13,425	8,605
α-Galactosidase
Liver	24	24	35
Kidney	23	30	34
Heart	3	25	10
Brain	16	30	17
Muscle	2	12	6
Marrow	4	4	5
Serum	10	22	10
α-Mannosidase
Liver	114	78	83
Kidney	99	101	110
Heart	8	26	10
Brain	22	22	20
Muscle	5	6	3
Marrow	21	17	22
Serum	3,232	4,720	3,748

Effect of HGUSB Overexpression on Phenotype and Morphology of Transgenic Mice.

Mice from both the WE1 and WE18 lines had a normal phenotype, were fertile, and showed no increase in mortality up to 18 mo of age. Both mouse lines had widespread lysosomal storage (Fig. 1). The amount of storage varied from mouse to mouse and generally tended to correlate with the elevation in tissue HGUSB levels in each line. The morphological character of the stored material was similar in both the WE1 and WE18 mice. With the toluidine blue stain, it was dark blue, and with PAS, storage was dark purple, indicating that it contained neutral hexose sugars (14). The distended lysosomes in kidney, liver, and muscle contained HGUSB identified histochemically and immunohistologically (Fig. 2). The distribution of the stored material was similar with the toluidine blue, PAS, and histochemical stains, immune label, and by electron microscopy.

Figure 1.

Lysosomal storage of GUSB in overexpressing mice. (A) The myocardial cells have abundant storage that collects in a perinuclear location (arrow). (B) By electron microscopy, the stored enzyme is electron dense. (C) Skeletal muscle has small granules of PAS-positive HGUSB in the sarcoplasm of the myocytes (arrow) and in endomysial interstitial cells (arrowhead). (D) The stored enzyme in skeletal muscle is electron dense and collects in irregular membrane-bound aggregates. (E) Hippocampal neurons contain fine cytoplasmic storage granules (arrow). (F) Neocortical neurons have electron-dense storage associated with a small amount of lipid. The storage vacuoles are small and do not distend the cells, and there is no apparent neuronal cell loss related to the storage. (G) The liver collects HGUSB primarily in the sinus-lining cells, where the material is PAS positive (arrow). (H) Fine blue granules are present in glomerular visceral epithelial cells (arrow). (I) Ultrastructurally, the storage material in the kidney is electron dense and morphologically similar to other sites, except in occasional tubules where the storage has a crystalline structure. (A, E, and H, toluidine blue, ×450; B, D, F, and I, uranyl acetate, lead citrate; B, ×2,000; D, ×5,000; F, ×3,120; I, ×36,000; C and G, PAS, ×320.)

Figure 2.

(A) With a histochemical technique that demonstrates the presence of HGUSB in tissues, a glomerulus contains abundant enzyme activity (arrow), correlating with the storage granules seen by light microscopy in the toluidine blue-stained sections. (B) Ultrastructural immunolocalization confirms that the stored material in glomerular visceral epithelial cells is HGUSB. In the skeletal muscle sarcoplasm (C) and interstitial cells (D), the distended lysosomes are decorated by colloidal gold linked to a polyclonal goat anti-HGUSB. [A, Naphthol AS-BI (6-bromo-2-hydroxy-3-naphthoic acid 2-methoxyanilide) β-d-glucuronide, ×242; B–D, colloidal gold immunolabeled for GUSB; B, ×10,000; C, ×8,000; D, ×5,000.]

Ultrastructurally, the storage material was membrane bound and complex. The electron-dense material associated in some cells with clear lipid droplets, and the storage did not distend the cell cytoplasm. Storage was most abundant in the cardiac myocytes (Fig. 1 A and B), where it tended to be perinuclear in location and in skeletal muscle (Fig. 1 C and D), where it was within both the muscle fibers and the endomysial interstitial cells. Smooth muscle cells in the gut also contained a small amount of storage. In the brain, neurons in the neocortex and hippocampus (Fig. 1 E and F), glia, perivascular cells, and leptomeningeal cells were affected. In the liver (Fig. 1G), Kupffer cells were affected primarily and often had abundant PAS-positive lysosomal storage. Bone marrow sinus-lining cells also contained storage. In the kidney, storage affected glomerular visceral epithelial cells (Fig. 1H) and renal tubular epithelial cells. Crystalline storage was seen in rare kidney tubular epithelial cells (Fig. 1I).

Tumors Affected Only One HGUSB Overexpressing Line.

Multiple tumors occurred in the original WE18 F₂ FVB mice (Table 3). Nine female WE18 FVB mice examined because of s.c. tumors or abnormal behavior all had neoplasms. The youngest transgenic mouse with a tumor, a 5-mo-old, had abnormal circling behavior and a posterior fossa meningioma. Tumors in the older animals included adenosquamous mammary carcinomas and pituitary adenomas, which were accompanied in some mice by adrenal cortical hyperplasia. Metastatic lesions were not identified, and only one tumor was found in C57B/6 mice carrying the tumor-associated transgene (Table 3). This 16-mo-old WE18 F₁₀ male had a lymphoma found incidentally at necropsy while being evaluated for lysosomal storage.

Table 3.

Neoplasms in WE18 mice

Age, mo	Sex	Pituitary adenoma	Adrenal cortical hyperplasia	Mammary adenosquamous carcinoma	Lymphoma	Meningioma
5	F					+
15	F	+	+	+
16	M				+
18	F	+		+
19	F		+	+
19	F	+*		+
19	F	+	+
19	F	NA	+	+
22	F	+	+	+
22	F	+	+	+

^*Mass described but not examined histologically; NA, pituitary not examined. M, male; F, female. 

Discussion

Gene therapy for LSD may result in expression of abnormally high levels of lysosomal enzymes. The safety of overexpression of a therapeutic enzyme is often a concern. We previously reported the introduction of the complete HGUSB gene as a transgene onto the murine MPS VII model (B6.C-H-2^bm1/ByBir-gus^mps/mps) background and demonstrated that all of the phenotypic, pathological, and biochemical abnormalities of the original MPS VII gus^mps/mps mice were corrected by a 20-fold overexpression of HGUSB expressed from the transgene (15). However, these mice had no morphological evidence of glycosaminoglycan or lysosomal enzyme storage and did not develop tumors.

The morphologic data presented here show that enormous overexpression of HGUSB in the WE1 and WE18 mice produced lysosomal storage of the enzyme. The variation in expression of HGUSB levels in different tissues of the two lines could reflect the effects of different integration sites on the regulation of expression of the transgene (16). The especially high HGUSB levels in heart and skeletal muscle are likely due to the tissue specificity of the chicken β-actin promoter. Some fraction of HGUSB activity in liver may result from hepatic fixed tissue macrophage uptake of circulating enzyme actually produced in other tissue, as suggested by the predominant localization of the PAS-positive storage material in the sinus-lining cells in the liver. The crystalline character of the storage in the renal tubular epithelial cells may be due to local cellular tonicity or pH, which could favor crystallization of HGUSB. This finding is reminiscent of the intracellular aggregation and crystallization of enzyme in lysosomes described in Chinese hamster ovary cells that overexpress the human α-galactosidase A gene (17).

The fall in HGUSB levels in subsequent generations of mice generated from the founders, particularly in the WE18 line, is likely the result of extinction of the cytomegalovirus enhancer, silencing of which has been reported to be common in vivo in gene therapy experiments (18).

The modest secondary elevation of other lysosomal enzymes seen in both transgenic overexpressing lines is less striking than observed in human and other animal models of LSD. The increase in other lysosomal enzymes in different tissues generally correlates with the highest HGUSB levels. Secondary elevation of lysosomal enzymes in LSD is partly due to increased synthesis of acid hydrolases in cells distended with storage material (in this case HGUSB) (19). In the MPS storage diseases, impaired degradation and turnover of these enzymes by accumulated glycosaminoglycans may also contribute to secondary elevation of the lysosomal enzymes. The modestly increased levels of these enzymes in serum may reflect secretion resulting from saturation of lysosomal enzyme (Man 6-P) receptors in transgolgi membranes by HGUSB (20).

Despite the enormous overexpression of HGUSB and widespread lysosomal storage of the enzyme in the WE1 and WE18 mice reported here, the mice have normal life spans and a grossly normal phenotype with the exception of tumors, which developed in one of the two lines. Except for the one lymphoma, the tumors occurred in female F₂ FVB progeny of the WE18 founders but not in WE1 mice, even though both lines have high HGUSB levels, suggesting that it was the integration site rather than the very high levels of expression that contributed to the tumors. That only one tumor was seen in backcross progeny when the WE18 transgene was placed on the C57B/6 background suggests strain and gender dependence in the development of tumors in mice carrying the WE18 transgene.

Integration site-dependent development of tumors could have at least two explanations. First, insertion of the cytomegalovirus enhancer/promoter near an oncogene might lead to overexpression of the oncogene, which in turn might contribute to tumor development. Such was the case in the recently reported child with X-linked severe combined immunodeficiency who developed a leukemia-like process of T cells after treatment with gene-corrected autologous cells. In this patient, the retrovirus had inserted into a gene in which mutations are known to be involved in childhood cancers (21, 22). It will be interesting to determine where the transgene inserted into the mouse DNA and to examine RNA from organs of both lines of overexpressing transgenic mice by microarray analysis to determine which, if any, other genes are up-regulated by each transgene. Second, insertion of the transgene might disrupt a tumor suppressor gene, which could contribute to malignancy in homozygous mice, in which both copies of the tumor suppressor would be inactive, or even in heterozygotes in which the second allele was inactivated by a mutation or a gene conversion event.

The results presented here suggest that overexpression of HGUSB is generally benign, even if overexpressed to a level producing demonstrable enzyme storage. Thus, it is unlikely that enzyme replacement therapy would ever produce toxic levels of the enzyme. However, the tumors seen in one of the two transgenic lines raise the possibility that gene therapy with vectors using this combination of promotor and enhancer could be oncogenic if stably integrated into certain sites.

The recent observation of tumorigenesis in long-term studies of MPS VII mice treated with rAAV vectors using the pCAGGS promoter (the same promotor present in these transgenes) may be relevant (9). Hepatocellular carcinomas and angiosarcomas occurred in adult MPS VII mice treated as newborns with the viral vector, which may integrate randomly into the mouse host genome. Insertional mutagenesis could explain the tumors seen in both the overexpressing transgenic mice reported here, in the MPS VII mice reported by Donsante et al. (9), and in the patient with leukemia-like process after treatment of severe combined immunodeficiency with gene therapy (21, 22). However, Donsante et al. (9) concluded that the copy number of integrated rAAV cassettes was too low (far lower than one per diploid genome) to support the hypothesis that the tumors resulted from an integration event. Whether the rAAV HGUSB-associated tumors result from the rAAV treatment or the overexpression of HGUSB in specific cell types remains to be established. Tumors have not been reported to date in experiments using the rAAV vector and the same promotor expressing other gene products.

Abbreviations

MPS

mucopolysaccharidosis

LSD

lysosomal storage disease

GUSB

βglucuronidase

HGUSB

human GUSB

rAAV

adeno-associated virus

PAS

periodic acid Schiff