Absence of Metabolic Cross-correction in Tay-Sachs Cells

We have investigated the ability of a receptor-mediated gene transfer strategy (cross-correction) to restore ganglioside metabolism in fibroblasts from Tay-Sachs (TS) patientsin vitro. TS disease is a GM2 gangliosidosis attributed to the deficiency of the lysosomal enzyme β-hexosaminidase A (HexA) (β-N-acetylhexosaminidase, EC 3.2.1.52). The hypothesis is that transduced cells overexpressing and secreting large amounts of the enzyme would lead to a measurable activity in defective cells via a secretion-recapture mechanism. We transduced NIH3T3 murine fibroblasts with the LαHexTN retroviral vector carrying the cDNA encoding for the human Hex α-subunit. The Hex activity in the medium from transduced cells was approximately 10-fold higher (up to 75 milliunits) than observed in non-transduced cells. TS cells were cultured for 72 h in the presence of the cell medium derived from the transduced NIH3T3 cells, and they were analyzed for the presence and catalytic activity of the enzyme. Although TS cells were able to efficiently uptake a large amount of the soluble enzyme, the enzyme failed to reach the lysosomes in a sufficient quantity to hydrolyze the GM2 ganglioside to GM3 ganglioside. Thus, our results showed that delivery of the therapeutic HexA was not sufficient to correct the phenotype of TS cells.

Tay-Sachs (TS) 1 disease is a GM2 gangliosidosis attributed to the deficiency of the lysosomal enzyme ␤-hexosaminidase A (HexA) (␤-N-acetylhexosaminidase, EC 3.2.1.52). HexA is a heterodimer of an ␣and ␤-subunits encoded by two different genes, HEXA and HEXB, located on chromosomes 15 and 5, respectively. Inherited defects in the ␣-subunit gene lead to the absence of the HexA and a massive accumulation of the GM2 ganglioside and related lipids primarily in neuronal lysosomes. Consequences are a severe cellular dysfunction and a rapid progressive neurodegeneration (1)(2)(3).
The formation of the HexA is controlled by a complex mechanism that ultimately results in the association of the ␣and ␤-subunits (1-2, 12, 13). This event requires the transport of both subunits to the Golgi apparatus compartment and regulation of the ratio between them. In the Golgi apparatus, a second important event occurs, the generation of the mannose-6-phosphate recognition marker. As result of this modification, the mannose-6-phosphate receptor recognizes and targets the enzyme to the lysosomes. The enzyme is then subjected to a final proteolytic processing, and in the presence of the GM2 activator protein, the enzyme hydrolyzes the ␤-GalNAc-(1-4)-␤-galactosidase glycosidic linkage (1, 8, 11, 14 -15).
No effective treatment is currently available for TS disease. Gene therapy holds the greatest promise for genetic diseases in which single gene mutations are responsible for the metabolic alterations. Vector-mediated gene transfer (direct correction) or receptor-mediated gene transfer (cross-correction) represent two potential approaches for lysosomal storage disorders (16 -18). The first transfer is based on the introduction of the missing gene to the deficient cells. In this case, the deficient cells produce the lacking enzyme. The second transfer is based on the rationale that secreted lysosomal enzymes can be uptaken by the neighboring cells through the binding with plasma membrane mannose-6-phosphate receptor.
In this paper, we have focused on the understanding of the mechanisms leading the effectiveness of the cross-correction strategy using human TS cells as a model. Moreover, because the combination of the two approaches is requested for the diffusion of the therapeutic enzyme and for the success of the gene transfer treatment in the patients (i.e. bone marrow gene transfer, local gene transfer delivery) (19 -22), we have compared the efficacy of both approaches to restore the HexA activity in TS cells. We first transduced human fibroblasts from TS patients with a retroviral vector carrying the cDNA encoding for the human Hex ␣-subunit and defined the ability of the recombinant HexA to restore the GM2 ganglioside metabolism. Second, we have cross-corrected human TS fibroblasts by using the HexA secreted either by transduced murine fibroblasts or by humantransduced TS cells and investigated the efficiency of this mechanism.
All of our results demonstrate that although both strategies were able to give rise to adequate levels of the missing enzyme in the deficient cells, the GM2 metabolism was only restored in transduced TS cells.

EXPERIMENTAL PROCEDURES
Materials-The GM2 activator protein was kindly provided by Professor Yu-Teh Li (Tulane University, New Orleans, LA). The fluorogenic substrates such as the 4-methylumbelliferyl derivative of ␤-Nacetylglucosaminide (MUG), ␤-N-acetylglucosamine-6-sulfate (MUGS), ␤-D-galactosidase, and ␣-D-mannopyranoside and 4-methylumbelliferone, Nonidet P-40, agarose gel for electrophoresis, and mannose-6phosphate were from Sigma. Bovine serum albumin and protein assay reagent were from Bio-Rad, and DE52 DEAE-cellulose was from Whatman Biochemicals. The medium for tissue culture was from Euro-Clone (Celbio Laboratories), the fetal calf serum and Colorado serum were from Mascia Brunelli, penicillin/streptomycin was from Invitrogen, and Gentamicin (G418), Polybrene, and restriction enzymes were from Roche Molecular Biochemicals. Centriprep 30 was from Amicon. All other reagents were of analytical grade.
Cell Cultures-Human fibroblasts established from three patients with the late infantile form of TS disease were kindly provided by the Laboratorio di Diagnosi Pre/Post-natale Malattie Metaboliche, Istituto G. Gaslini Hospital (Genova, Italy). Normal human fibroblasts are routinely used in our laboratory (23). Fibroblasts were cultured in EMEM containing 10% (v/v) heat-inactivated fetal calf serum in 25-cm 2 culture flasks in a humidified atmosphere containing 5% CO 2 at 37°C. NIH3T3 mouse fibroblasts were cultured in DMEM containing 10% (v/v) heat-inactivated Colorado serum in 25-cm 2 culture flasks in the same conditions as described above. Cell growth was determined by counting cells in a hemocytometer. The viability of the cells was estimated by examining their ability to exclude trypan blue (0.1% in 0.9% NaCl).
At confluence, cells were trypsinized and harvested by centrifugation (1000 rpm for 10 min in a bench top centrifuge). After two washings with 0.9% NaCl, the pellet was resuspended in 10 mM sodium phosphate buffer, pH 6.0, containing 0.1% (v/v) Nonidet P-40 detergent and sonicated. The lysates were centrifuged at 12,000 rpm in a Eppendorf Microfuge for 20 min, and the supernatants were used as cell extracts for enzyme analysis. All procedures were carried out at 4°C.
Construction of a Suitable Retroviral Vector for Hex ␣-Subunit cDNA Transfer and Transduction-We cloned the cDNA encoding for ␣-subunit of human HexA (ATCC, Manassas, VA) into the Moloney leukemia virus retroviral vector backbone LXTN. In this vector, the cDNA is under the long terminal repeat promoter. This vector contained also the neomycin resistance gene. The tk-neo fusion gene is under the control of the tk promoter (24). The procedure for the production of the retrovirus was as follows. The E86 ecotropic packaging cell line was transfected with the vector carrying the ␣-subunit cDNA by standard calcium phosphate co-precipitation. The E86 supernatants were collected, filtered (diameter 0.45 m), and used to infect the amphotropic packaging cell lines AM12 in the presence of polybrene. After selection with neomycin (G418), the best AM12-transduced clone, was selected by titration and used as viral producer packaging cells. Viral supernatants were collected, filtered through 0.45-m pore size filters, and used to infect the designed cells. One day before the infection, TS cells were plated at 75% confluence. The following day, cells were incubated with viral supernatant in the presence of polybrene for 16 h at 37°C. After G418 selection, the transduced cells were analyzed for the ␣-subunit gene expression and for the HexA properties. The same procedure was used to transduce murine fibroblasts NIH3T3.
Administration of the HexA to TS Cells-Human TS cells were cultured in the medium (conditioned medium) containing HexA secreted by transduced NIH3T3 cells for 72 h in 25-cm 2 culture flasks at 37°C. The total enzyme activity in the conditioned medium was up to 75 milliunits. The medium was changed every 24 h for 3 days. Untreated and treated TS cells were harvested and evaluated for the HexA properties. As experimental control, TS cells were cultured in conditioned medium containing different amounts of HexA secreted either by transduced TS cells or normal fibroblasts.
In some experiments, mannose-6-phosphate was added to the conditioned medium to have a 5 mM final concentration. These cells were then analyzed for hexosaminidase, galactosidase, and mannosidase activities.
The optimum pH for the HexA activity was determined by testing the enzyme activity in citrate/sodium phosphate buffers in the pH range of 3.5-7.5 (5). To define the thermal stability, 50 l of each sample were incubated at 52°C for different times, cooled on ice for 1 h, and then assayed at 37°C for the Hex activity. The results represented the average of at least three independent experiments and are expressed as a percentage of the activity found in the controls kept on ice (5). The inhibitory effect of glucosamine toward the HexA was by incubating Hex isoenzymes at 37°C in the presence of 100 mM glucosamine. The results are the average of at least three independent experiments and are expressed as a percentage of the activity of treated samples with respect to untreated control (25).
␤-Hexosaminidase Isoenzymes Analysis-Cell lysates were analyzed by the ion-exchange chromatography on DEAE-cellulose (25). The chromatography was performed by using 1-ml column equilibrated with 10 mM sodium phosphate buffer, pH 6.0 (buffer A). The flow rate was 0.5 ml/min. Enzyme activity retained by the column was eluted by a linear gradient of NaCl (0.0 -0.5 M in 40 ml of buffer A). Finally, the column was eluted with 1.0 M NaCl in the same buffer. Fractions (1 ml) were collected and assayed for the Hex activity with the two substrates, MUG and MUGS.
Subcellular Fractionation-Cells at confluence were harvested and resuspended in 0.25 M sucrose and then homogenized in a Potter Elveheim-type homogenizer until Ͼ90% of the cells were disrupted. Differential centrifugation of the homogenate was performed at 800 ϫ g for 10 min at 4°C in the AJ20 rotor of a Beckman J2-21 centrifuge to sediment the nuclear fraction. The supernatant was then centrifuged at 13,000 ϫ g for 15 min to sediment the lysosomal fraction, and the supernatant (post-lysosomal fraction) from this step was decanted (25).

Feeding Experiments-[ 3 H]GM1 or [ 3 H]GM2
, dissolved in propan-1ol/water (7:3 by volume) were pipetted into a sterile tube and dried under a nitrogen stream. The residue was solubilized in an appropriate volume of pre-warmed (37°C) EMEM to obtain a ganglioside concentration of 5 ϫ 10 Ϫ7 M. After the removal of the original medium and rapid washing of cells with EMEM, 2 ml of the medium containing the radioactive lipid were added to each 60-mm dish, and the cells (TS fibroblasts, normal fibroblasts, transduced TS fibroblasts, cross-correct TS fibroblasts) were incubated for 3 h at 37°C. After incubation, the radioactive medium was removed, and the dishes were washed first with EMEM solution for 5 min and then with 10% fetal calf serum-EMEM for 30 min. After a chase of 15 h, cells were washed twice with phosphate-buffered saline, scraped off with a rubber policeman, and centrifuged at 1000 ϫ g for 10 min. The pelleted cells were subjected to lipid extraction (27) resulting in a delipidized pellet and a total lipid mixture. The radioactivity imaging of thin layer chromatography plates of lipid extracts was acquired with a ␤-imager 2000 instrument (Biospace, Paris, France). The radioactivity associated with individual lipids was determined with the specific ␤-vision software provided by the manufacturer (Biospace), and the radioactivity associated with total lipid extracts was determined by liquid scintillation counting (30).
Degradation of GM2 by Secreted HexA-Transduced confluent NIH3T3 cells were maintained in culture for 72 h in the presence of the 10 mM NH 4 Cl, which specifically increases all secreted lysosomal enzyme activities (31). The medium was collected and partially purified by ion-exchange chromatography on DEAE-cellulose as described above.
The assay was performed using 50 milliunits of secreted HexA toward MUG in 170 l of 10 mM, pH 4.5, citrate buffer containing 10 g of GM2 activator protein, 0.1% bovine albumin, 52 g of sodium taurodeoxycholate, and 20,000 dpm of GM2 (3 g). The enzyme reaction mixture was maintained at 36°C under continuous vortexing for 12 h. The reaction mixture was mixed with three volumes of tetrahydrofuran, the mixture was centrifuged, the clear solution was dried, and the residue was resuspended into a few microliters of chloroform/methanol 2:1 by volume. The solutions were analyzed by high performance thin layer chromatography followed by radioimaging and quantitative detection of the separated GM2 and GM3.
Other Analytical Methods-Proteins were measured by the method of Bradford (32) using the serum bovine albumin as standard. Lipids were analyzed by chromatography on silica gel on high performance thin layer chromatography plates (Merck) using the solvent system chloroform/methanol/0.2% aqueous CaCl 2 (50:42:11 by volume) in comparison with standard compounds followed by radioactivity imaging (28).

Construction of a Retroviral Vector for Hex ␣-Subunit
cDNA-The recombinant retrovirus based on a Moloney murine leukemia virus backbone LXTN vector containing the wild-type Hex ␣-subunit cDNA was produced (Fig. 1). In this vector, the ␣-subunit cDNA is expressed under the control of the promoter/enhancer sequence in the viral long terminal repeat; the tk-neo fusion gene is under the control of the tk promoter. The HSV-tk gene is contained in the vector as a safety mechanism (24). The L␣HexTN is used to produce the L␣HexTN recombinant retrovirus and transduce human fibroblasts from TS patients and NIH3T3 cells as described under "Experimental Procedures." The presence of Hex ␣-subunit cDNA in the retroviral-transduced cells was evaluated by Southern blot analysis (data not shown).
Vector-mediated Gene Transfer Strategy for TS Cells-After selection with G418, transduced TS cells display a Hex-specific activity of 4.2 Ϯ 0.8 milliunits/mg as evaluated with the fluorogenic substrate MUGS, which is hydrolyzed only by the ␣-subunit (33). This value is comparable with that measured in control fibroblasts (Fig. 2A). The secreted HexA activity in the culture medium of transduced TS cells was comparable with that of normal fibroblasts and was much higher with respect to TS cells (Fig. 2B).
The isoenzyme pattern of Hex in transduced TS cells has been determined by ion-exchange chromatography on DEAEcellulose (25). Under our experimental conditions, HexB was unretained by the column and eluted with void volume, whereas HexA or other minor forms (HexS or Hex intermediate forms) was eluted by a linear gradient of NaCl (0.0 -0.5 M in buffer A). Besides, the combination of the chromatographic analysis and the specific enzymatic assay with the two substrates, MUG, which is hydrolyzed by both ␣and ␤-subunits, and MUGS, which is hydrolyzed only by ␣-subunit, provides information about the subunit composition of the Hex isoenzymes. In L␣HexTN-transduced human TS cells, the Hex activity was composed of HexA and HexB with a Hex isoenzyme pattern similar to that of control cells (Fig. 3, b and c). No presence of HexS was detected. On the contrary, TS cells were characterized by the complete absence of HexA. They have HexB and a second enzyme form eluted by the gradient at the NaCl concentration required to elute the intermediate forms of Hex (HexI) (5). The ␤␤ structure of this form was demonstrated by its inability to hydrolyze the MUGS substrate (Fig. 3a).
Recombinant HexA from transduced TS cells and HexA from control non-transduced cells were separated from the other isoenzymes by preparative DEAE-cellulose chromatography and analyzed for their biochemical properties (5,25,34). They showed the same optimum pH of 4.5 toward both MUG and MUGS substrates (data not shown) and displayed comparable curves of heat inactivation when incubated at 52°C for different intervals of time with almost complete inactivation after 15 min (Fig. 4A). In the presence of 100 mM glucosamine (the specific inhibitor of the ␤-subunit), HexA (␣␤ dimer) from transduced and control cells lost approximately 50% of its original activity (Fig. 4B). As internal control, the effect of glucosamine was also tested on Hex B (␤␤ dimer) isolated either from control cells or from transduced TS cells and on HexS (␣␣ dimer obtained from HL60 cell line, see Ref. 9). HexB lost ϳ80% of its original activity in agreement with previous data (25), whereas HexS was activated.
GM2 Ganglioside Hydrolysis in Transduced Human TS Cells- Fig. 5 shows the radioactive lipid pattern of transduced cells fed with [ 3 H]GM1. Radioactive ganglioside was taken up by the cells and catabolyzed to GM2 3 GM3 3 neutral sphingolipids with a hydrolysis rate comparable with that of control cells, demonstrating the restoration of the metabolic defect in transduced TS cells.
Production of a Suitable Enzyme Producer Cell Lines-After transduction with the retroviral vector L␣HexTN, HexA was overexpressed in NIH3T3 cells. An increase of ϳ10-folds of the cell enzyme activity toward the substrate MUGS was observed ( Fig. 2A). Furthermore, transduced NIH3T3 cells released high levels of Hex activity in the medium of culture (Fig. 2B). No atypical intracellular Hex isoenzymes such as HexS were detected by DEAE-cellulose chromatography in transduced NIH3T3 cells, but an increase of the HexB activity was ob- served (data not shown). These data were confirmed by ionexchange chromatography analysis of secreted Hex activity of both transduced and non-transduced cells. Fig. 3, d and e, show a successful production of high levels of the secreted HexA, which before eluted ϳ10 ml of the intracellular HexA activity. The amount of the enzyme secreted into the medium of culture was in the proportion to the cell number.
As control, NIH3T3 cells were also transduced with a similar retroviral vector carrying as report gene the cDNA encoding for the enzyme ␤-galactosidase. There were no changes in the intracellular or in the secreted Hex activity in transduced cells; no increase of HexB activity was revealed by DEAE-cellulose chromatography (data not shown).
Hex isoenzymes secreted by transduced NIH3T3 cells were partially purified by preparative DEAE-cellulose chromatography and were analyzed for their biochemical properties as described above. The curves of heat inactivation, the assay in the presence of 100 mM glucosamine, and the chromatographic pattern itself confirmed that the released Hex activity corresponds to HexB and HexA with the total absence of HexS (data not shown).
The ability of the HexA secreted by transduced NIH3T3 cells to hydrolyze the natural ganglioside GM2 was determined in an in vitro assay in the presence of GM2 activator protein. The   FIG. 3.

DEAE-cellulose chromatography analysis of HexA in transduced cells. a, Tay-Sachs cells. b, transduced TS cells (t-TS)
. c, normal fibroblasts (NF). d, non-transduced NIH3T3-secreted activity, e, transduced NIH3T3 (t-NIH3T3)-secreted activity. Similar amounts of protein for each sample was loaded into a 1-ml column equilibrated with 10 mM sodium phosphate buffer, pH 6.0. After loading, the column was eluted at 15 ml with 10 mM sodium phosphate buffer, pH 6.0. HexA activity retained by the column was eluted with 40 ml of a linear gradient of NaCl (0 -0.5 M in the above buffer). Fractions (1 ml) were collected and assayed for Hex activity toward the two substrates MUG (q) and MUGS (E). retroviral-transduced NIH3T3 cells were maintained in culture for 72 h, and then the secreted HexA was separated from the other protein of the culture medium by DEAE-cellulose chromatography. The enzyme activity on the natural substrate GM2 ganglioside was tested (31). Although the enzyme activity was low (Fig. 6), the measured hydrolysis rate was comparable with that determined for the enzyme secreted from normal human fibroblasts.

FIG. 4. HexA biochemical properties. HexA from transduced human TS cells (q) and from control cells (E) displayed a comparable behavior of heat inactivation when incubated at 52°C for different intervals of time with almost complete inactivation after 15 min (A). In
Receptor-mediated Gene Transfer Strategy for TS Cells-TS cells from three patients were incubated for 72 h with the medium from L␣HexTN-transduced NHI3T3. The conditioned medium was changed every 24 h. After 72 h, cells were harvested, and the HexA activity was evaluated. The cross-correction procedure was able to restore the Hex activity toward MUGS substrate in all TS fibroblasts from patients considered in this study (Fig. 7A). The internalization of the HexA was time-dependent, the maximum uptake of the enzyme occurring after 72 h of incubation. Fig. 7B shows the Hex isoenzyme pattern in TS cross-corrected cells and demonstrates the presence of HexA uptake. This enzyme displayed optimum pH of 4.5 versus both MUG and MUGS substrates and a curve of heat inactivation when incubated at 52°C for different intervals of time comparable with that of the control enzyme (data not shown).
When cross-correction experiments were performed in the presence of 5 mM mannose-6-phosphate, no HexA activity could be recorded in TS cells (Fig. 7A). This finding suggests that the uptake process of HexA by TS cells occurs through the mannose-6-phosphate surface receptor mechanism. Cross-corrected TS cells exhibiting a MUGS-specific activity of 4.5-4.8 milliunits/mg cell protein were fed with radioactive gangliosides GM2 or GM1. Fig. 8 shows that in cross-corrected TS cells, the hydrolysis of GM2 to GM3 does not occur. GM2 ganglioside was taken up by the cells but was not catabolyzed (Fig. 8, lanes e-i), whereas GM1 ganglioside was taken up by the cells and hydrolyzed to GM2 that was not further processed but was stored (Fig. 8, lanes e-i). Identical results were obtained when TS cells were cross-corrected with HexA secreted either by L␣HexTNtransduced human TS cells or by control human fibroblasts. In these experimental conditions in TS cells, the internalized MUGS activity was 22.7 and 14.5%, respectively, with respect to the normal fibroblasts (Fig. 9). In both cases, the MUGS uptake activity was lower than that observed in cross-corrected TS cells with HexA secreted by transduced NIH3T3, but this value was in the range of the activity required to correct the phenotype (1). The lower enzyme activity is not surprising considering that the enzyme activity in the cell medium collected from normal cells and transduced TS cells was ϳ20 times lower than that present in the medium collected from NHI3T3-transduced cells (Fig. 2B). Therefore, these cross-corrected TS cells were fed with radioactive GM1 ganglioside as previously described, but again the HexA taken up by the cells was not able to hydrolyze the GM2 ganglioside to GM3 gangli-oside (Fig. 8, lane l). The same results were also obtained by using a 10ϫ conditioned medium containing HexA secreted either by L␣HexTN-transduced human TS cells or by control human fibroblasts (Fig. 9) where the percent of internalized activity was significantly increased. Thus, the absence of hydrolysis was not related to the amount of HexA taken up by the cells (data not shown). These findings suggest that the enzyme taken up by the cells could not be transported to the lysosomes where the hydrolysis of natural substrate process occurs. This aspect was further investigated. The nuclear, lysosomal, and post-lysosomal fractions were obtained by the differential centrifugation. The lysosomal fraction of cross-corrected cells had MUGS activity (Fig. 10) that was approximately one-tenth of that associated with the normal fibroblast lysosomal fraction. These data suggest that the amount of the lysosomal HexA requested to prevent the metabolic phenotype (1) may be inadequate to hydrolyze the GM2 when affected lysosomes are considered.

DISCUSSION
The restoration of the metabolic defect in human Tay-Sachs cells has been evaluated by the receptor-mediated gene transfer strategy (cross-correction). This strategy is based on the rationale that part of each lysosomal enzyme is secreted and can be uptaken by the neighboring cells via the mannose-6phosphate receptor (35)(36)(37). The hypothesis is that transduced cells overexpressing and secreting large amounts of these enzymes would lead to a measurable activity in defective cells via secretion-recapture mechanism.
We explored cross-correction by using an in vitro model. Human TS fibroblasts were incubated in HexA-containing medium and were evaluated for HexA properties, localization, and activity on radioactive natural compounds previously fed to the cells. To mimic the gene transfer procedures in the clinic, we have used as enzyme producer cells transduced with a Moloney murine leukemia virus retroviral vector carrying the cDNA encoding for the human ␣-subunit of the HexA, L␣HexTN (Fig.  1). The efficacy of the vector to restore the enzyme activity within the TS cell on natural compounds was demonstrated by the transduction of human TS fibroblasts (Fig. 5). In this context, it is known that a correct ␣and ␤-subunit association is required (2,11). A combination of the DEAE-cellulose chromatography and the enzyme assay with the two fluorogenic artificial substrates, MUGS, which is hydrolyzed only by Hex isoenzymes containing ␣-subunit, and MUG, which is hydrolyzed by Hex isoenzymes containing ␣and/or ␤-subunit, was used to verify the correct HexA formation (Fig. 3, a-c). L␣HexTN-transduced human TS fibroblasts had HexA activity and properties (optimum pH, thermal stability behavior, and similar sensibility to the glucosamine) similar to that of control cells (Fig. 4).
The HexA activity on natural compounds in the transduced TS cells was evaluated by cell-feeding experiments with radioactive gangliosides. In both TS (Fig. 5, lane c) and transduced TS cells (Fig. 5, lane d), the exogenous [H 3 ]GM1 reached the lysosomes where it was correctly converted by the galactosidase to GM2 ganglioside. Only in transduced TS cells, the GM2 ganglioside was hydrolyzed to GM3 ganglioside and neutral sphingolipids. Because the GM2 ganglioside hydrolysis requires a concerted action between both the ␣/␤-subunits of HexA and the small GM2 activator protein (1), this hydrolysis confirms the correct structure of the recombinant enzyme. Thus, the detection of the GM3 ganglioside in transduced TS cells demonstrates the restoration of the metabolic defect.
We have used the murine cell lines NIH3T3 transduced with the retroviral vector L␣-HexTN as enzyme producer cells to overexpress HexA. Many authors (38) have used these cells in similar experiments. Transduced NIH3T3 released high levels of HexA activity into the culture medium. Unexpectedly, we also observed an increase in HexB levels in transduced cells in comparison with the non-transduced cell lines. Because NIH3T3 cells constitutively express the ␣-subunit, its overexpression could negatively influence the stoichiometric process of ␣and ␤-subunit association and lead to the formation of HexS (␣␣ dimer) (9,25,39). However, in our experimental conditions, the ␣and ␤-subunit balance was naturally resolved; in fact no HexS was detected. Moreover, we are investigating whether the increase in ␤-subunit level is related to the ␣-subunit overexpression. Lacorazza et al. (40) showed similar findings in C17 neuroblastoma clone transduced with a different retroviral vector expressing the ␣-subunit gene.
To perform the cross-correction experiments, the medium containing HexA secreted by transduced NIH3T3 was used to culture TS cells. Culture media with different amounts of Hex activity were used. TS cells uptake HexA from the medium of culture (Fig. 7) by a mannose-6-phosphate receptor-dependent mechanism (Fig. 7A). The level of HexA activity restored in TS cells was comparable with that measured in control human fibroblasts ( Figs. 2A and Fig. 7A). Nevertheless, the HexA uptake could not catabolyze GM2 ganglioside (Fig. 8). Identical findings were obtained by culturing TS cells in medium containing different amounts of enzyme activity (1ϫ versus 10ϫ) secreted either by human-transduced TS cells or human normal fibroblasts. In this context, we observed that the MUGS , and postlysosomal (PL) fractions were obtained by differential centrifugation (25). Results are expressed as relative specific activity (RSA). In this figure, the distribution of the activities of other lysosomal enzymes (␣-mannosidase and ␤-galactosidase) as experimental control also is reported. RSA is the ratio of the percentage of the activity of each fractions and the percentage of the proteins in that fraction. activity internalized in TS cells was lower when compared with the activity measured in cross-corrected TS cells with the medium containing HexA secreted by transduced NIH3T3, but it was sufficient to give rise to the normal phenotype (1). We may address the observed discrepancy on the amount of the internalized MUGS activity to the higher level of the HexA released by transduced NIH3T3 cells in the medium of culture with respect to HexA released by normal fibroblasts. However, all of these data indicate that the chimerical structure of the internalized enzyme (human -the exogenous ␣-subunit -and mouse -the endogenous ␤-subunit) is not responsible for the missing GM2 ganglioside hydrolysis. Therefore, it is well known that the murine and human ␣-subunit cDNAs have very high homology of Ͼ80% (41,42) Moreover, the HexA secreted by transduced NIH3T3 hydrolyzes as does the enzyme secreted by normal human fibroblasts, GM2 to GM3 ganglioside. All together, these data suggest that the structure of the recombinant enzyme is not responsible for the missing GM2 ganglioside hydrolysis, and that most probably the enzyme does not localize into the lysosomes.
To clarify these findings, we evaluated the localization of HexA in cross-corrected TS cells. Subcellular fractionation of cross-corrected TS cells showed an atypical pattern of enzyme distribution with unexpectedly high levels of MUGS activity in all subcellular fractions. Thus, only a minor part of the cellrestored activity was associated to the lysosomes. Although this activity, if all belonging to the inner volume of lysosomes, should be sufficient to restore the GM2 ganglioside metabolism (1), it probably may be inadequate to hydrolyze the GM2 when affected lysosomes are considered. Moreover, we cannot conclude that the enzyme associates to the lysosome membranes but does not belong to the lysosomes.
In addition, the storage of lipids in the lysosomes may change membrane trafficking along the lysosomal pathway (43). In this regard, Kabayashi et al. (44) have shown that lysobisphosphatidic acid is localized to the late endosomes and may have an important role in the protein-sorting functions of endosomes. In particular, they show that the lysobisphosphatidic acid is the main component of the endosome membrane domains, and that these specialized domains are involved in sorting the multifactor receptor for insulin-like growth factor 2 and ligands bearing mannose-6-phosphate, particularly lysosomal enzymes (45). Thus, the mannose-6-phoshate mechanism may be not sufficient to target the enzyme to the lysosomes.
Alternatively, the membrane fluidity of the endosomes may be altered in the mutant cells, resulting in an abnormal lipid/ protein trafficking (38). An abnormal transport along the lysosomal pathway has been described in the mucolipidosis type IV disease by Chen et al. (46). Altered endocytic environment on the biogenesis of the lysosomes was also demonstrated in the fibroblasts of patient suffering from sialic acid storage disease (47).
The overall results reported in this study demonstrate that the gene therapy of TS disease could be achieved by a direct gene transfer strategy. Nevertheless, the therapy of the lysosomal storage disorder is based on the combination of direct and cross-correction gene transfer strategies. Indeed, the end point of both approaches is the production of a functional HexA in the lysosomes. Although direct transduction was demonstrated to be able to restore the phenotype in several lysosomal disorders (16, 18, 19, 48 -50), the efficacy of the cross-correction has been shown for very few lysosomal diseases (51,52). The clarification of the cross-correction mechanism is necessary for the therapeutic administration of the missing enzyme to pa-tients and for the distribution of the secreted enzyme by implanted engineered cells to neighboring deficient cells and across same non-permeable membranes (e.g. the blood-brain barrier, ependyma).
Here, we provide data exploring this process and showing that the internalized enzyme probably undergoes a more complex turnover via a mechanism that is not yet known. Moreover, from our data emerges the necessity to follow the enzyme topology in the cross-corrected cells by feeding natural compounds to cells before considering the enzyme replacement therapy as a potential cure for TS disease and similar genetic disorders.