Defining the Pathway for Tat-mediated Delivery of β-Glucuronidase in Cultured Cells and MPS VII Mice

Abstract

We used recombinant forms of human β-glucuronidase (GUS) purified from secretions from stably transfected CHO cells to compare the native enzyme to a GUS-Tat C-terminal fusion protein containing the 11-amino-acid HIV Tat protein transduction domain for: (1) susceptibility to endocytosis by cultured cells, (2) rate of clearance following intravenous infusion, and (3) tissue distribution and effectiveness in clearing lysosomal storage following infusion in the MPS VII mouse. We found: (1) Native GUS was more efficiently taken up by cultured human fibroblasts and its endocytosis was exclusively mediated by the M6P receptor. The GUS-Tat fusion protein showed only 30-50% as much M6P-receptor-mediated uptake, but also was taken up by adsorptive endocytosis through binding of the positively charged Tat peptide to cell surface proteoglycans. (2) GUS-Tat was less rapidly cleared from the circulation in the rat (t_1/2 = 13 min vs 7 min). (3) Delivery to most tissues of the MPS VII mouse was similar, but GUS-Tat was more efficiently delivered to kidney. Histology showed that GUS-Tat more efficiently reduced storage in renal tubules, retina, and bone. These studies demonstrate that Tat modification can extend the range of tissues corrected by infused enzyme.

Introduction

β-Glucuronidase (GUS) is an acid hydrolase that removes β-D-glucuronide residues from glycosaminoglycan chains in lysosomes. Newly synthesized GUS is glycosylated in endoplasmic reticulum and phosphorylated on terminal mannose residues of sugar chains in the Golgi apparatus. This phosphorylation of mannose is essential for enzyme delivery to lysosomes via the mannose-6-phosphate receptor (M6PR) [1,2]. Some of the phosphorylated, newly synthesized GUS is secreted rather than sorted to lysosomes and may be recaptured by the M6PR expressed on the surface of some cell types. Thus, receptor-mediated endocytosis provides an alternate route to lysosomes [2-4].

Deficiency of human GUS produces mucopolysaccharidosis type VII (MPS VII; Sly syndrome), which is characterized by accumulation of glycosaminoglycans in cells of most tissues [5,6]. Patients commonly present in infancy or childhood with coarse facial features, hepatosplenomegaly, delayed psychomotor development, short stature due to bone dysplasia (dysostosis multiplex), cardiac valve involvement, corneal clouding, and hearing loss [5,7]. The mouse model of MPS VII has most of the features of the human disease and has been used extensively to study experimental therapies [8].

Recently, therapeutic trials have been carried out for several lysosomal disorders [9]. Those trials were partially successful in alleviating symptoms and clearing storage in heart, liver, spleen, kidney, bone marrow, and lung [10-12]. Several enzyme replacement therapies have been approved and are currently part of clinical therapeutics [13-15]. Nonetheless, some tissues like brain, bone, and muscle are refractory and new strategies are needed to target these tissues.

The HIV Tat protein or Tat basic domain (amino acids 47-57) was reported to have protein transduction activity (i.e., direct and energy-independent protein delivery into cytoplasm via transduction of the denatured polypeptides across the plasma membrane by an unknown mechanism), both in cell culture systems and in vivo following injection of the Tat-tagged proteins into mice [16]. However, Davidson et al. reported that expression of a nondenatured GUS-Tat fusion protein in an adenovirus vector also led to wider distribution than native GUS following intravenous and direct brain injection of the adenovirus vector [10]. They suggested that endocytosis, rather than “protein transduction,” was involved in the mechanism of internalization of nondenatured, Tat-tagged GUS. Other recent reports also suggested that Tat-mediated internalization of peptides and nondenatured proteins is likely due to adsorptive endocytosis rather than protein transduction [17-19] and dependent on binding to cell-surface proteoglycans [20].

We sought to characterize the M6P-independent component of GUS-Tat internalization by cultured cells and to determine whether purified GUS-Tat displays in vivo distribution comparable to native GUS and has enhanced effectiveness for correction following infusion into the MPS VII mouse.

Results

Production of Peptide-Targeted Enzymes

Fig. 1 shows the constructs used for producing enzymes in stably transfected CHO cell lines. Two different C-terminal tags were added to GUS fusion proteins. The first was the original 11-amino-acid Tat protein transduction domain (PTD) described by Dowdy et al. [16]. The other was a subsequently developed 11-amino-acid sequence called PTD4, which Dowdy's group reported to be much more efficient than Tat in transferring conjugated FITC into Jurkat T cells in vitro and into mouse T cells following ip injection [16]. Both constructs produced enzyme in transfected COS cells. The specific activity of GUS-Tat was comparable to that of native GUS. Total production was reduced 30% in GUS-Tat-transfected COS cells, but the fraction secreted (80%) was higher than that for native GUS (40%). This increased fraction of GUS-Tat secreted is consistent with uptake data shown below, suggesting that GUS-Tat produced in COS cells is less well phosphorylated than native GUS produced in COS cells. Independently, metabolic double labeling with [³²P]- and [³⁵S]-methionine in HeLa cells infected with an adenoviral vector expressing GUS or GUS-Tat showed reduced phosphorylation of GUS-Tat, since the ³²P/³⁵S ratio in immunoprecipitable GUS-Tat was only 46% of that of GUS (Table 1). Production of GUS-PTD4 was only 11% reduced and the fraction secreted was slightly over 50% of the total enzyme produced, not significantly different from that of wild-type GUS. Endocytosis studies (described below) were all done on enzymes produced in stably transfected CHO cells and purified to homogeneity from medium by monoclonal affinity column.

FIG. 1.

Illustration of native and modified GUS. GUS-Tat fusion protein has 11 amino acids from the HIV Tat protein (47-57) with a 4-glycine spacer fused to the C-terminus of human β-glucuronidase. The GUS-PTD4 fusion protein has the same structure as GUS-Tat except for 6 Ala substitutions at positions 48, 50-52, 55, and 57 of the 47-57, 11-amino-acid, HIV Tat protein transduction domain (PTD).

TABLE 1.

Comparison of the phosphorylation of GUS and GUS-Tat (³²PO₄/³⁵S)

	Lysate	Sup	Total (Lysate + Sup)
Lysate refers to lysate of pellet from an 1800 g centrifugation. Sup refers to labeling media after centrifugation. In both cases, CPMs in immunoprecipitates were counted as described and all of the phosphorylation occurred on mannose residues of the oligosaccharides as described [25].
Gluc	0.11	0.18	0.13 (100%)
Gluc-Tat	0.07	0.04	0.06 (46%)
Gluc-CTB	0.06	0.04	0.05 (38%)

M6P Uptake of GUS-Tat by Fibroblasts

Results in Fig. 2A compare the uptake of purified, native GUS, GUS-Tat, and GUS-PTD4 by human fibroblasts in the presence and absence of M6P. The uptake of native GUS was completely inhibited by M6P, indicating that the uptake is mediated by the M6PR. The same is true for GUS-PTD4. GUS-Tat shows only about 50% of the M6P-dependent uptake shown by native GUS and GUS-PTD4, consistent with evidence that it is less phosphorylated than native GUS. However, GUS-Tat also shows a smaller but significant component of its uptake, which is M6PR independent (not inhibited by M6P), that neither native GUS nor GUS-PTD4 shows. We conclude that the PTD4 tag, contrary to the observations of Dowdy et al. with denatured short peptides [16], confers no advantage over native GUS in delivery of nondenatured enzyme to mammalian cells.

FIG. 2.

(A) Comparison of uptake of native GUS, GUS-Tat, and GUS-PTD4. Native GUS, GUS-Tat, and GUS-PTD4 (6000 units) were added to human MPS VII fibroblasts plated in 35-mm dishes with or without 5 mM M6P. After incubation at 37°C for 3 h, cells from triplicate plates were rinsed and lysed in 1% deoxycholate, and the enzyme activity was assayed on duplicate samples from each plate. (B) Concentration dependence of uptake of phosphorylated and dephosphorylated native GUS and GUS-Tat. Purified native GUS, GUS-Tat, dephosphorylated native GUS, and dephosphorylated GUS-Tat (2000 to 50,000 units) were added to human MPS VII fibroblasts. Cells were incubated at 37°C for 3 h, followed by treatment with 0.2% trypsin for 15 min at 37°C. Enzyme activity of the cell lysates from duplicate plates was assayed as described for (A) and averaged. (C) Concentration dependence of M6P-independent uptake of GUS-Tat at 4 and 37°C by A549 cells. GUS-Tat enzyme (2000 to 1,280,000 units) was added to duplicate plates of A549 cells in the presence of 5 mM M6P. After incubation at 37 or 4°C for 2 h, cells were rinsed, lysed, and assayed as for (A). The endogenous GUS level of 52 units/ mg was subtracted from all values. The inset shows data from cells exposed to 64,000 units/plate of M6P for 3 h at 37 and 4°C, after which one set of duplicate plates was rinsed, lysed, and assayed as for (A) and another set was treated with 0.2% trypsin for 15 min at 37°C, rinsed, lysed with 1% deoxycholate, and assayed as for (A). Endogenous GUS level of 49 units/mg was subtracted from each value.

To characterize further the dependence of uptake of GUS and GUS-Tat on M6P recognition, we dephosphorylated both enzymes using calf intestine alkaline phosphatase and compared the uptake of the dephosphorylated and phosphorylated enzymes by human MPS VII fibro-blasts. The uptake of native GUS was saturated and almost completely obliterated by enzymatic dephosphorylation (Fig. 2B). In contrast, the uptake of GUS-Tat appeared only partially saturated under these conditions. At higher concentrations (above those that saturate the M6PR), uptake exceeds the levels of uptake of the M6PR-mediated uptake system. Dephosphorylation did not abolish the nonsaturable component.

Temperature Dependence of Tat-Mediated Uptake by A549 Cells

We found that A549 cells have the same amount of Tat-mediated uptake (20 units/mg) as human fibroblasts, but a much lower level of M6PR-mediated uptake than human fibroblasts. This means that cells expressing a limited number of M6PR (like A549 cells) are still accessible to Tat-mediated delivery. Xia et al. [10] reported temperature dependence of uptake of virally expressed nondenatured GUS-Tat, suggesting an endocytic process, in contrast to Tat-mediated translocation of denatured polypeptides, which was reported to be temperature independent [16]. We measured cell association of CHO-produced GUS-Tat at 4 and 37°C (Fig. 2C). While cell-associated enzyme at 37°C exceeded that at 4°C, GUS-Tat still became cell associated at 4°C in a concentration-dependent manner. Uptake and/or cell association appears not to be saturated under these conditions at both temperatures. However, when these cells were trypsinized following incubation with enzyme (inset, Fig. 2C), nearly all the enzyme that became cell associated at 4°C was removed, indicating that although it was adsorbed to the cell surface at 4°C, it was not internalized. On the other hand, nearly all the enzyme that became cell associated at 37°C was trypsin resistant, indicating that it had been internalized. Finally, GUS-Tat, which was internalized at 37°C in the presence or absence of M6P, did not leave the cells on prolonged incubation at 37°C. There was little or no decline between 2 and 27 h after endocytosis, as was true for the native GUS internalized by A549 cells by the M6P-dependent uptake system (data not shown). The prolonged retention of GUS-Tat following uptake contrasted with the results for Tat-translocated denatured peptides, which were reported to translocate in and out of cells and to be completely lost from the cells during such a chase [16].

Inhibition of Tat-Mediated Endocytosis by Polyvalent Anions and Cations

The striking difference between the GUS-Tat and the GUS-PTD4 uptake (Fig. 2A) suggested that it was the positively charged residues of the Tat peptide that mediated the M6P-independent uptake of GUS-Tat. In PTD4, 5 of the 8 positively charged residues in the 11-amino-acid Tat peptide have been replaced by neutral alanine residues [16]. If this hypothesis were correct, heparin might be expected to bind the positively charged residues of GUS-Tat and inhibit its M6P-independent cell surface binding and uptake. Polylysine might also be expected to inhibit GUS-Tat, not by binding GUS-Tat, but rather by binding to and saturating the negatively charged cell surface elements to which GUS-Tat binds. Fig. 3 shows the experiments testing these predictions. Heparin (Fig. 3A) and heparan sulfate (Fig. 3B), two polyanions, both block the M6P-independent uptake of GUS-Tat but had no effect on the M6PR-mediated endocytosis of native GUS. Polylysine (Fig. 3C) was even more effective in inhibiting GUS-Tat uptake on a weight basis but had no effect on uptake of native GUS. These observations argue that, in the presence of M6P, GUS-Tat binds to the cell surface through its positively charged residues and is subsequently taken up by adsorptive endocytosis. Similar conclusions were reached recently by others studying uptake of whole Tat protein [19,20].

FIG. 3.

Inhibition of Tat-mediated endocytosis by polyvalent anions and cations. Native GUS (4000 units) in the absence of M6P or GUS-Tat in the presence of 2 mM M6P was added to 35-mm dishes of MPS VII fibroblasts in the presence or absence of (A) 125-500 µg/ml heparin, (B) 50-200 µg/ml heparan sulfate, or (C) 2.5-20 µg/ml polylysine in MEM + 15% fetal bovine serum. After incubation at 37°C for 4 h, the cells were washed five times with PBS and solubilized in 1% deoxycholate. Extracts were assayed for β-glucuronidase and protein. Results are expressed as percentage of uptake in the absence of inhibitors.

Clearance of Native GUS and GUS-Tat in Rats

We infused GUS and GUS-Tat intravenously into rats and analyzed the rate of clearance of the infused enzymes (Fig. 4). The native enzyme was rapidly cleared, with a half-life of approximately 7 min. GUS-Tat was cleared more slowly with a half-life of approximately 13 min. The slower rate of clearance suggests that GUS-Tat not only has fewer M6P residues than native GUS, as inferred from the reduced M6P-dependent uptake by fibroblasts (Fig. 2B), but also has fewer exposed mannose residues to mediate clearance by the highly efficient mannose receptor system [21]. The likely explanation for the latter is that nonphosphorylated high-mannose oligosaccharides are processed to complex-type oligosaccharides, which would produce longer circulating forms of the GUS-Tat enzyme.

FIG. 4.

Clearance study for native GUS and GUS-Tat in rats. Each enzyme (200,000 units) was infused into a rat tail vein and 250 µl of blood was sampled at 2-, 5-, 7-, 12-, 30-, 60-, 90-, 120-, 180-, and 240-min time points. Plasma enzyme activity was assayed. The data were averaged from two rats for each purified enzyme. Each blood sample was assayed in duplicate for enzyme activity.

Distribution of GUS-Tat and Native GUS Following Intravenous Infusion into the MPS VII Mice

When we infused GUS-Tat and native GUS into the MPS VII mouse at 1 mg/kg and sacrificed the mice 24 h after infusion, the differences in tissue distribution were small. There was slightly more enzyme delivered to kidney from the GUS-Tat infusion and slightly less delivered to liver (data not shown). This difference was still apparent in kidney 1 week following three weekly infusions of enzyme (Table 2). These enzyme levels were considerably lower than those reported in various tissues following intravenous injection of adenovirus vectors expressing GUS and GUS-Tat [10]. In addition, the adenovirus-expressed GUS-Tat levels were significantly greater than adenovirus-expressed GUS levels in heart, kidney, and muscle. The basis for these differences between distribution of injected GUS-Tat and GUS-Tat expressed in vivo from injected adenovirus vectors is not clear. It may partly reflect differences between bolus injections and continuous systemic delivery provided by enzyme-expressing hepatocytes.

TABLE 2.

Tissue β-glucuronidase levels after three times weekly infusions

Tissue	Buffer only	Native GUS	GUS-Tat
Each enzyme (5000 units/g) was infused weekly for 3 weeks. Three mice in each group were sacrificed 1 week after the last infusion and tissues frozen until enzyme assay was done. Units: nmol/h/mg.
Liver	0.3	124 ± 21.6	146 ± 7.79
Spleen	1.1	35 ± 13.7	45 ± 13.3
Brain	0.04	0.06 ± 0.01	0.15 ± 0.2
Heart	0.08	29 ± 3.5	19 ± 0.58
Kidney	0.02	1.6 ± 0.38	4.2 ± 1.09
Lung	0.001	3.3 ± 0.38	0.8 ± 0.59
Bone	0.04	6.3 ± 1.85	6.2 ± 2.03
Muscle	0.02	1.4 ± 0.36	0.7 ± 0.34
Bone marrow	0	13.9 ± 4.6	10.2 ± 1.19
Plasma	0.3	2.2 ± 1.46	0.8 ± 0.53

Comparison of Infused GUS and GUS-Tat for Clearance of Lysosomal Storage in the MPS VII Mice

We used an established protocol of three weekly infusions of 1 mg/kg delivered to 6- to 8- week-old MPS VII mice to study differences in the response (clearance of storage) seen following infusion of native GUS and GUS-Tat [22]. In the present experiments, clearance was complete in response to both GUS and GUS-Tat enzymes in liver (both sinus-lining cells and hepatocytes), in bone marrow (hematopoietic elements and sinus-lining cells of the bone marrow), and in spleen (both sinus-lining cells and trabecular fibroblasts) (data not shown). Renal tubular epithelial cells appeared to contain even more enzyme by AS-BI staining in the GUS-Tat-injected animals (Figs. 5A-5C) than the fourfold increase in GUS-Tat in extracts of whole kidney (Table 2). This increased delivery to cortical tubules correlates with more effective clearance of storage in renal tubules by GUS-Tat (Figs. 5D-5F). The retinal pigment epithelium (Figs. 5G-5I) and bone osteoblasts and osteocytes (Figs. 5J-5L) also showed greater clearance by GUS-Tat, as did cardiac valve tissue (data not shown). The central nervous system neurons, glia, leptomeninges, and cornea showed no clearance by either enzyme.

FIG. 5.

Distribution of native GUS and GUS-Tat following infusion into MPS VII mice and reduction in lysosomal storage in tissues. (A) AS-BI-stained section of MPS VII untreated mouse kidney shows no red staining, indicating no evidence of enzyme activity. (B) An MPS VII mouse treated with native GUS shows a slight blush of red staining, indicating enzyme activity within the renal tubules. (C) After treatment with GUS-Tat, there is increased enzyme activity within the renal tubules in an MPS VII mouse. (D) An untreated MPS VII mouse shows extensive lysosomal distention within renal tubular epithelial cells and visceral epithelial cells in the glomeruli. (E) After treatment with native GUS, there is a slight reduction in the amount of storage in the renal tubular epithelial cells and glomerular visceral epithelial cells. (F) After treatment with GUS-Tat, there is an even more striking reduction in storage in the renal tubular epithelial cells. (G) An untreated MPS VII mouse retina shows marked lysosomal distention within the retinal pigment epithelial cells (arrow). (H) After receiving native GUS, there is only a minimal reduction in lysosomal storage in the retina (arrow). (I) After treatment with GUS-Tat, the retinal pigment epithelial cells (arrow) in an MPS VII mouse have a marked reduction in lysosomal storage. (J) An untreated MPS VII mouse rib has marked lysosomal distention affecting the osteocytes in the bone (arrowhead) and the osteoblasts (arrow) lining the bone as well as in cells in the bone marrow. (K) After receiving native GUS, there is a moderate but incomplete reduction in the amount of lysosomal storage in the rib osteocytes (arrowhead) and osteoblasts (arrow). The bone marrow has marked reduction in the amount of storage. (L) After treatment with GUS-Tat, the osteocytes (arrowhead) and osteoblasts (arrow) both have a marked reduction in storage in the rib of an MPS VII mouse. The bone marrow also has marked reduction in storage. (A-C, AS-BI-glucuronide, 1 mm = 25 µm; D-L, toluidine blue; D-I, J-L, 1 mm = 4 µm; G and H, 1 mm = 3.7 µm).

In summary, the GUS-Tat appeared more effective at this dose in clearing storage from renal tubular cells, bone osteoblasts and osteocytes, retinal pigment epithelial cells, and heart valve tissue. In no cell type did GUS-Tat appear less effective than GUS when given at this dose.

Discussion

The reports from Dowdy's group that denatured, Tat-modified peptides and denatured proteins, even as large as Escherichia coli β-galactosidase, could translocate across membranes and even enter the brain following intraperitoneal injection raised hopes that denatured, Tat-modified lysosomal enzymes might cross the blood-brain barrier and correct central nervous system storage [16]. Unfortunately, these results were difficult to generalize to mammalian lysosomal enzymes, possibly because hydrophilic oligosaccharide side chains on the glycosylated acid hydrolases prevent proper refolding and assembly of multimeric enzymes following denaturation and impede their translocation across the membrane.

Although our attempts to extend Dowdy's findings to denatured mammalian lysosomal enzymes were discouraging, Davidson's group reported that virally expressed nondenatured GUS-Tat showed broader distribution than native GUS when both were expressed from adenoviral vectors injected intravenously or directly into brain. Furthermore, they found that Tat-modified nondenatured enzyme isolated from culture media of virally infected cells showed enhanced uptake compared to native GUS in A549 cells, and most of this enhanced uptake was M6P independent. To explore the mechanism of the M6P-independent uptake of nondenatured GUS-Tat and its potential importance to enzyme replacement therapy with recombinant enzymes, we compared the uptake properties in vitro and the in vivo distribution of purified recombinant GUS and GUS-Tat.

Like native GUS, GUS-Tat chimeric protein could be produced in large amounts in CHO-K1 cells. However, GUS-Tat is less phosphorylated than native GUS (Table 1) and the rate of M6P-dependent, receptor-mediated endocytosis is greatly reduced (Fig. 2A). Decreased phosphorylation could result from the effect of the increased positive charge or subtle changes in the three-dimensional structure of GUS, which make it less effective as a substrate for the phosphotransferase, which normally recognizes three-dimensional patches on acid hydrolases [23]. Our results suggest that the amounts of M6PR-independent uptake of GUS-Tat by different cell types are actually very similar. It is the low level of M6P-dependent uptake by A549 cells that makes the M6PR-independent uptake of GUS-Tat a large fraction of the total uptake for this cell type.

Several lines of evidence indicate that the Tat-mediated component of uptake is due to adsorptive endocytosis. This includes the evidence that M6PR-independent uptake of GUS-Tat is concentration dependent and not saturated under the conditions studied, is temperature dependent, and is inhibited by the polyanions heparin and heparan sulfate and by the polycation polylysine. Since 8 of 11 amino acid residues of Tat are positively charged, we infer that Tat behaves as a polycation like polylysine. Such polycations bind negatively charged phospholipids of cell membranes and glycosaminoglycans on the cell surface and are known to mediate adsorptive endocytosis [24].

The in vivo studies indicated that the Tat moiety does influence the rate of clearance and, while the overall tissue distribution was not much different, the GUS-Tat enzyme showed a greater concentration in kidney than native enzyme. Morphologically, the GUS-Tat enzyme also showed enhanced ability to reach and clear storage from renal tubules, retinal pigment epithelial cells, and osteocytes in bone and osteoblasts lining trabecular and cortical surfaces. An earlier study showed some advantages of a chimeric enzyme with a C-terminal, IGF-2-related peptide in clearing storage in MPS VII mice [22]. That chimeric enzyme (called GUS-GILT) was more effective in clearing visceral glomerular epithelial storage in kidney than GUS-Tat. However, GUS-Tat was more effective in clearing storage from the renal tubular epithelial cells and retinal pigment epithelial cells than GUS-GILT. Clearance from renal tubular epithelial cells is not functionally important in MPS VII, but clearance in the retinal pigment epithelia and osteocytes may have functional significance. Whether GUS-Tat will be significantly more effective in clearing storage from affected tissues like retinal pigment epithelial cells, osteoblasts, and heart valves in MPS VII mice in long-term studies, and whether differences will be functionally important, remain to be established.

We suggest that it is the ability to be endocytosed by neuronal cells that do not express much M6PR that explains why the virally expressed GUS-Tat showed greater distribution in brain following direct injection than the virally expressed native GUS [10]. While that study showed more extensive distribution of GUS-Tat than native GUS in brain following direct injection of an adenovirus vector, we found no evidence that GUS-Tat crossed the blood-brain barrier when infused intravenously. Thus, for enzyme replacement to take advantage of any added potential that adsorptive endocytosis confers on GUS-Tat in the central nervous system, we need an additional strategy to transfer GUS-Tat across the blood-brain barrier following intravenous injection.

Materials and Methods

Cell culture

We used minimal essential medium with 10% fetal bovine serum, 100 units/ml penicillin, and 100 mg/ml streptomycin for all cell cultures. Human MPS VII patient fibroblasts and A549 lung carcinoma cell lines were cultured in a humidified 37°C incubator containing 6.5% CO₂.

Construction of expression vectors

Using PCR, we amplified modified cDNAs containing nucleotide sequences for Tat or PTD4 peptides and a four-glycine spacer followed by nucleotides encoding eight amino acids from the C-terminus of human GUS, transferring the stop codon to the end of the Tat or PTD4 peptides, and cloned the PCR fragment into the XhoI site of the pGEM T-Easy vector (Promega, Madison, WI, USA). The cDNAs were then subcloned into the pCXN expression vector.

We also obtained the reported β-glucuronidase Tat construct from Dr. B. L. Davidson [10], which we subcloned into pCXN. Both GUS-Tat constructs in pCXN expressed enzyme with identical properties.

Establishment of CHO K1 stable cell lines expressing recombinant enzymes

After characterizing the expression of these constructs in COS cells, we used electroporation to establish CHO and HeLa cell lines stably expressing each of these constructs [25]. We electroporated 20 µg of each plasmid construct with 1 × 10⁷ cells and plated six 100-mm dishes to select colonies under G418. Isolated colonies were placed into 48-well plates and expanded. The highest expressing clone was expanded for large-scale production of enzyme in triple flasks.

Purification of recombinant enzymes

Conditioned medium from native GUS-, GUS-Tat- and GUS-PTD4-expressing CHO K1 cell lines was collected daily from triple flasks containing Waymouth medium with 2% fetal bovine serum and stored at −20°C for purification. Enzymes were purified on an anti-human GUS monoclonal antibody affinity column and eluted with sodium phosphate buffer containing 3.5 M MgCl₂ [22]. The MgCl₂ was removed on a P6 sizing column and purified enzyme was stored at −80°C. The specific activities for GUS-Tat and GUS-PTD4 were comparable to that for native GUS, i.e., 4-5 × 10⁶ units/mg.

Uptake experiments of recombinant enzymes with MPS VII patient fibroblasts and A549 cells

Native GUS and modified GUS enzymes, GUS-Tat and GUS-PTD4 (2000-50,000 units), were added to human MPS VII patient fibroblasts and A549 cells and incubated 3 h with or without 5 mM M6P. Incubation was for 2 or 3 h at 37°C, after which cells were rinsed, lysed in 1% deoxycholate, and assayed for β-glucuronidase activity as described [22,26]. Dephosphorylated native GUS and GUS-Tat enzymes were prepared as described [27], except that calf intestine alkaline phosphatase (Promega) was used instead of E. coli alkaline phosphatase. Three uptake inhibitors were tested on human fibroblasts: porcine intestine heparin (Elkins-Sinn Corp, Cherry Hill, NJ, USA), heparan sulfate (Sigma, St. Louis, MO, USA), and poly-L-lysine (Sigma). Native or GUS-Tat purified enzymes (4000 units) were incubated with increasing concentrations of each inhibitor for 2 h and enzyme activity of the cell lysates was assayed.

Clearance of enzyme in rats

Male Sprague-Dawley rats weighing approximately 300 g were catheterized at a carotid artery and a tail vein 1 week before experimentation and were immobilized several hours per day to adapt them to constraint. Native GUS and GUS-Tat enzymes (200,000 units) were injected in a catheter inserted in a tail vein and 250 µl of blood was sampled at 2, 5, 7, 12, 30, 60, 90, 120, 180, and 240 min after injection. Collected blood samples were centrifuged at 3000 rpm for 15 min at 4°C and enzyme activity in the plasma was assayed.

In vivo infusion of GUS and GUS-Tat in MPS VII mice

MPS VII mice were genetically manipulated to make them tolerant to human GUS enzyme [28]. Three mice in each group received three weekly injections of 1.0 mg/kg of buffer, native GUS, or GUS-Tat and were sacrificed 1 week after the third injection. Tissues were sampled for enzyme activity assays, histochemical demonstration of GUS activity using napthol-AS-BI-β-D-glucuronide (Sigma), and light microscopy. All animal procedures and use of human fibroblasts were approved by the Saint Louis University Animal Care and Use Committee and the Committee for Research using Human Subjects, respectively.

Metabolic labeling of GUS and GUS-Tat in infected HeLa cells

To assess the degree of phosphorylation of GUS and GUS-Tat, metabolic and phosphate labeling were done. Briefly, HeLa cells were infected with 20 infectious units/cell of recombinant vectors expressing GUS or GUS-Tat for 24 h. HeLa cells were starved for 1 h in Dulbecco's modified Eagle's medium lacking either phosphate or methionine, plus 5% dialyzed fetal bovine serum. For [³⁵S] methionine labeling, the same medium was used containing 100 µCi L-[³⁵S]methionine for 6-8 h. For phosphate labeling, starvation medium containing 1 mCi H₃³²PO₄ was applied for a labeling period of 6-8 h. Cell extracts were prepared from 1800g pellets in 1 ml of solubilization buffer (1% Nonidet P-40, 0.1% SDS, 0.5% deoxycholate, 10 mM Tris, pH 8.5, 1 mM MgCl₂, 14 mM NaCl) and GUS, and both the cell lysates and the supernatants (media) were immunoprecipitated. Immunoprecipitates were analyzed by SDS-PAGE and autoradiography. Blots were quantified by ImageQuant (Molecular Dynamics).