Allosteric activation of acid α-glucosidase by the human papillomavirus E7 protein

Changes in the cellular carbohydrate metabolism are a hallmark of malignant transformation and represent one of the earliest discernible events in tumorigenesis. In the early stages of certain epithelial cancers, a metabolic switch is regularly observed, in which slowly growing glycogenotic cells are converted to highly proliferating basophilic cells. This step is accompanied by a rapid depletion of the intracellular glycogen stores, which in liver carcinogenesis results from the activation of the enzyme acid α-glucosidase by an as yet unknown mechanism. We show here that acid α-glucosidase is a target for the E7 protein encoded by human papillomavirus type 16, a human tumor virus that plays a key role in the genesis of cervical carcinoma. We show that expression of E7 induces the catalytic activity of acid α-glucosidase in vivo and wild type E7, but not transformation-deficient mutants bind directly to acid α-glucosidase and increase the catalytic activity of the enzyme in vitro. The data suggest that the E7 protein encoded by human papillomavirus type 16 can act as an allosteric activator of acid α-glucosidase.

It has long been known that tumor cells display characteristic alterations of the carbohydrate metabolism (for review see Ref. 1), which represent one of the earliest discernible events in tumorigenesis (for recent review see Ref. 2). In certain epithelial cancers, such as liver cancer or kidney cancer, one of the first detectable alterations is a metabolic switch, in which slowly growing glycogenotic cells, also referred to as clear cells, are converted to highly proliferating basophilic cells (for review see Ref. 3). This step is accompanied by a rapid depletion of the intracellular glycogen stores, which in liver carcinogenesis results from the activation of acid ␣-glucosidase (4), whereas the activity of glycogen phosphorylase, the other cellular glycogendegrading enzyme, is reduced throughout hepatocarcinogenesis (5). The actual content of glycogen in a cell is controlled through the balance of glycogen-synthesizing enzymes and glycogen-degrading enzymes, which themselves are under allos-teric control by various metabolites (6). It is unknown at present how the activity of these enzymes is modulated in early carcinogenesis, to first build up the clear cell phenotype and then trigger its disappearance at later stages. According to the current concept (7), the deregulation of metabolic enzymes in early tumor cells reflects changes in cellular signal transduction which lead to tumor-specific alterations of the metabolic apparatus.
Reduced glycogen storage is regularly observed in early lesions of the cervix, a finding that has been used for clinical diagnosis of cervical dysplasia for more than 60 years (8). Cervical neoplasia is tightly linked to infection by human papillomaviruses (HPV) 1 of the high risk group, e.g. HPV-16 (9), and two viral genes, E6 and E7, are required for papillomavirus-associated carcinogenesis (for review see Refs. 9 and 10). The E7 oncogene of HPV-16 cooperates with HPV-16 E6 to immortalize human keratinocytes (11). The transforming activity of E7 is sensitive to mutations in the N-terminal domain (12,13), and it was shown that the N-terminal part of the HPV-16 E7 oncoprotein mediates binding to proteins of the retinoblastoma gene family (14). Therefore E7 triggers activation of cellular genes driven by the E2F transcription factor (15), leading to the accumulation of cellular factors required for S phase entry (for review see Refs. 16 and 17).
However, the ability of E7 to transform cells is also sensitive to mutations in the C-terminal domain (18), which is of particular importance for the ability of E7 to immortalize human keratinocytes, the natural host cells of HPV-16 (19). Furthermore, whereas it was convincingly demonstrated that the interaction of E7 with nuclear proteins is essential for its transforming potential (reviewed in Ref. 20), a substantial fraction of the total E7 protein is found in the cytoplasm (21,22). In accordance with these observations, it was found that several cellular proteins interact with the C terminus of HPV-16 E7 (23)(24)(25)(26)(27)(28), and one of these additional E7 target proteins, the glycolytic control enzyme M2 pyruvate kinase (M2-PK) (28), is localized in the cytoplasm. M2-PK plays a key role in reprogramming the cellular carbohydrate metabolism in tumors (for review see Ref. 29), and it was shown that HPV-16 E7 shifts M2-PK to the tumor-specific dimeric form with decreased substrate affinity, resulting in the expansion of the intracellular pools of glycolytic phosphometabolites (28).
In an attempt to discover additional E7-binding proteins, we now identified the glycogen-degrading enzyme acid ␣-glucosi-dase as a new interaction partner for HPV-16 E7, and we found that its catalytic activity is directly controlled by the E7 protein.
In Vitro Interaction Analysis-GST fusion proteins containing various mutants of E7 were loaded (20 ng/l each) on glutathione-Sepharose 4B beads. Yeast lysates (1 mg) or purified mature and premature acid ␣-glucosidase proteins (200 pg/l) were incubated with GST-E7 fusion proteins, and bound proteins were analyzed by Western blotting (28), using anti-acid ␣-glucosidase antibodies (33). Input of the various GST-derived fusion proteins was controlled by Coomassie staining. Levels of acid ␣-glucosidase proteins bound to GST-E7 were quantitated by densitometric scanning of the autoradiogram.
Indirect Immunofluorescence-U2-OS cells were cultured in DMEM ϩ 10% FCS. For transient expression of cDNAs, cells were grown to about 80% confluence on glass coverslips coated with 0.05% gelatin. Transfection of the expression vectors pSHAG2 (37) and pJ4⍀HPV16E7 (27) was performed using the Effectene method (Qiagen, Hilden, Germany). 38 h post-transfection, cells were prepared for indirect immunofluorescence according to standard protocols, including methanol fixation. After incubation with primary antibodies (polyclonal rabbit antiacid ␣-glucosidase antibodies (33) and monoclonal mouse anti-HPV-16 E7 antibody TVG701 (38), respectively), and secondary antibodies (TRITC-conjugated anti-Mouse IgG and FITC-conjugated anti-rabbit IgG from Dianova, Hamburg, Germany), cells were washed and embed- The structure of the expression library, derived from vector pJG4 -5 by insertion of cDNA fragments from a WI-38 cDNA library, is indicated. B42-TAD refers to the B42 transactivation domain. B42 fusion proteins are expressed from the inducible GAL1 promoter, and these plasmids contain the TRP1 gene as selectable marker. B, the plasmid designated pB42-acid-␣-Gluc::TRP1 was isolated during the interaction screen; it contains the cDNA for human acid ␣-glucosidase (37). C, derivatives of yeast strain EGY48/pSH1834, expressing various LexA fusion proteins as indicated, were transformed with the plasmid pB42-acid-␣-Gluc::TRP1. pJG4 -5, expressing the unfused B42 trans-activation domain, was used as negative control. Transformants were selected for uracil, histidine, and tryptophan prototrophy and grown in glucose minimal medium (Ura Ϫ , His Ϫ , and Trp Ϫ ). Yeast cells were then streaked out onto each of three plates and incubated for 4 days at 30°C under the following nutrient conditions: control, glucose minimal medium with leucine; all strains grow; glucose, glucose minimal medium without leucine, selection for B42 fusion protein-independent activation of the LexAo6-LEU2 reporter; galactose, galactose minimal medium without leucine, selecting for B42 fusion protein dependent activation of the LexAo6-LEU2 gene. ded in Fluoromount G (Biozol, Eching, Germany). Samples were viewed by indirect immunofluorescence microscopy using the confocal scanning system MicroRadiance (Bio-Rad) in combination with a Zeiss Axiophot microscope. The following filters were used for FITC-derived and TRITC-derived fluorescence: excitation for both at 488 nm, emission for FITC at 515-530 nm, and for TRITC at Ͼ600 nm).
Glucosidase Assays-As a source of ␣-glucosidase, extracts of 14/2 cells or purified enzyme were used. 14/2 cells were lysed in buffer containing 25 mM Tris-PO 4 , pH 7.8, 4 mM EGTA, 10% glycerol, and 1% Triton X-100 either after dexamethasone withdrawal (non-induced 14/2 cells, E7 not expressed) or 4 h after re-addition of dexamethasone (induced 14/2 cells, E7 expressed). For assays with purified enzyme, the 110-kDa precursor form and the 70/76-kDa mature forms of ␣-glucosidase were isolated from the milk of transgenic rabbits or human placenta, respectively, as described previously (39,40). Cell lysates were brought to a final protein concentration of 1 g/l with 0.2 M sodium acetate, pH 4.2, and purified glucosidase preparations were brought to a final protein concentration of 2.5 ng/l. Lysates of non-induced 14/2 cells and purified enzymes were preincubated with recombinant GST, GST-E7 wild type, or GST-E7 mutants (20 ng/l) for 20 min on ice to study the direct influence of E7 on glucosidase activity. Following preincubation, the artificial substrate 4-methylumbelliferyl-␣-D-glucoside (Sigma) was added at a final concentration of 0, 0.5, 1, 2, 5, and 10 mM. All samples were incubated for 1 h at 37°C. The reaction was stopped by the addition of 0.5 M sodium carbonate, pH 10.5, and the extinction of the samples was measured in a spectrophotometer at a wavelength of 350 nm.
Quantification of Intracellular Glycogen-14/2 cells cultured in the presence or absence of dexamethasone as stated above were pelleted and weighed. Pellets were lysed in 30% NaOH. Glycogen was precipitated with ethanol, washed intensely, and pelleted by centrifugation. Glycogen pellets were dissolved in water and incubated with anthrone reagent (72% H 2 SO 4 , 0.5 mg/ml anthrone, and 10 mg/ml thiourea (Sigma)) for 15 min at 100°C. Extinction of the samples was measured at 578 nm.

RESULTS
Identification of Acid ␣-Glucosidase as E7-binding Protein-To identify additional targets for E7, a human cDNA library (31) was screened in a yeast two-hybrid experiment, using the C-terminal domain of HPV-16 E7 as bait (Fig. 1A). In this screen, the cDNA encoding acid ␣-glucosidase (␣Gluc, Fig.  1B) was repeatedly isolated. Specific interaction of acid ␣-glucosidase with wild type E7 but not with two unrelated fusion proteins, LexA-Bicoid and LexA-C-Myc, was confirmed by coexpression in yeast, followed by a selection for leucine prototrophy (Fig. 1C). A lacZ reporter gene system was then applied to map the E7 domains required for the interaction with acid ␣-glucosidase, using LexA fusion proteins containing isolated subdomains of E7 or specific E7 mutants ( Fig. 2A). As shown in Fig. 2B, this experiment revealed that full-length E7 as well as the isolated E7 C terminus strongly interact with acid ␣-glucosidase, whereas the isolated cd1-(1-17) and cd2-(18 -38) subdomains of E7 are inactive in this assay. Similarly, two mutations in the C terminus of E7, GLY58/91 (19) and ⌬79-83 (27), strongly reduced the ability of E7 to bind acid ␣-glucosidase, confirming the conclusion that sequences in the E7 C terminus mediate the binding of acid ␣-glucosidase. Similar quantities of all fusion proteins were produced in the yeast strains analyzed and cd2) and the putative zinc finger motifs (CXXC) are indicated. E7 mutants used in this report are shown. B, Saccharomyces cerevisiae strain EGY48/pSH1834, containing a LexA operator-lacZ gene (LexAo8-GAL1-LacZ::URA3), was transformed with the plasmid pB42-acid-␣-Gluc::TRP1. Plasmids encoding various LexA fusion proteins were coexpressed, as indicated, and ␤-galactosidase activity was determined. C, LexA fusion proteins from yeast strains used in B were quantitated by Western blotting, using a polyclonal antiserum to LexA (30). D, purified GST or GST-E7 fusion proteins immobilized on glutathione-Sepharose 4B beads were incubated with whole-cell extract from yeast cells, and the amount of B42-HA1-acid-␣-Gluc protein that was retained on the beads was determined by direct immunoblotting as described (28), using a polyclonal antiserum to acid ␣-glucosidase.
in Figs. 1C and 2B, and these proteins were stable ( Fig. 2C; see also Ref. 30), confirming that the inability of LexA-E7cd1, LexA-E7cd2, and LexA-E7⌬79-83 to promote reporter gene activity is due to the inability of these proteins to interact with acid ␣-glucosidase.
The yeast fusion proteins were then used to demonstrate the binding of acid ␣-glucosidase to HPV-16 E7 in vitro. When lysates of yeast cells were incubated with GST HPV-16 E7 fusion proteins, the protein containing acid ␣-glucosidase fused to the B42 trans-activation domain was specifically retained by GST-E7 but not GST protein alone (Fig. 2D), whereas the isolated B42 domain did not bind to either GST or GST-E7 (data not shown; see Ref. 28). Thus, E7 specifically interacts with the acid ␣-glucosidase part of the fusion protein.
Direct Physical Interaction between HPV-16 E7 and Acid ␣-Glucosidase-To determine if E7 can bind to acid ␣-glucosidase from mammalian cells, purified GST-E7 fusion proteins were incubated with the 110-kDa (precursor) form of acid ␣-glucosidase, which had been purified to homogeneity from the milk of acid ␣-glucosidase transgenic mice (39). As is shown in Fig. 3A, purified acid ␣-glucosidase interacts with high affinity with GST-16E7 protein but not the GST control. This finding indicates that the 110-kDa form of acid ␣-glucosidase binds directly to the C terminus of E7, and no additional cellular proteins are required. As in the two-hybrid assay (Fig.  2), deletion of amino acids 79-83 in the C terminus of E7 considerably reduced the interaction of E7 with acid ␣-glucosidase (to 68 Ϯ 5% of the value obtained with wild type E7, see also Fig. 3B), consistent with the C-terminal domain of E7 being involved in the interaction. Replacement of cysteine at position 24 in E7 by glycine nearly abolished the interaction of E7 with acid ␣-glucosidase (reduction to 14 Ϯ 3% of the value obtained with wild type E7), indicating that sequences in the N terminus of E7 are also essential for a high affinity interaction (Fig. 3A). Together, these results suggest that several domains of the E7 protein are involved in the interaction with acid ␣-glucosidase. We also analyzed binding of E7 to the mature lysosomal form of acid ␣-glucosidase, which had been purified to homogeneity from human placenta (41). Upon incubation of the purified 70/76-kDa form of acid ␣-glucosidase with GST-E7, similar amounts of the enzyme were bound to GST beads and to FIG. 3. Differential interaction of E7 with acid ␣-glucosidase subspecies. A, the purified 110-kDa form of the acid ␣-glucosidase protein was incubated with beads containing various GST-E7 fusion proteins as indicated. Upper panel, after elution from the beads, bound proteins were separated by gel electrophoresis, and acid ␣-glucosidase was detected by Western blotting; for comparison, 10% of the acid ␣-glucosidase input was also loaded on the gel. Lower panel, input of the various GST fusion proteins was controlled by Coomassie staining. B, GST-E7wt and GST-E7⌬79-83 proteins (40 ng/l each) were loaded on glutathione-Sepharose 4B beads and incubated with various amounts of purified mature acid ␣-glucosidase protein, as indicated. The levels of acid ␣-glucosidase bound to the GST-E7 proteins were analyzed by Western blotting and quantitated by densitometric scanning of the autoradiogram. Percent affinity was plotted against acid ␣-glucosidase

FIG. 4. Cytoplasmic localization of HPV-16 E7 in CaSki cells.
CaSki cells were subjected to mild lysis. Subsequently, nuclei were separated from cytoplasm by centrifugation. Nuclear and cytoplasmic fractions were separated by SDS-polyacrylamide gel electrophoresis and probed with antibodies to lamin B, M2 pyruvate kinase, calreticulin, and HPV-16 E7, as indicated. Total cellular lysates were analyzed as controls.
input. Note the logarithmic scale of the x axis. C, the purified 70/76-kDa forms of acid ␣-glucosidase was incubated with beads containing various GST-E7 fusion proteins as indicated. Upper panel, after elution from the beads, bound proteins were separated by gel electrophoresis, and acid ␣-glucosidase was detected by Western blotting; for comparison, 10% of the acid ␣-glucosidase input was also loaded on the gel. Lower panel, input of the various GST fusion proteins was controlled by Coomassie staining. beads containing GST-E7, indicating that this form of the enzyme does not specifically interact with E7 (Fig. 3C).
The 110-kDa form of acid ␣-glucosidase is found in the endoplasmic reticulum and the Golgi complex (37,42). Although a fraction of E7 is localized to the nucleus (43), cytoplasmic localization of E7 was also reported (21,22,44), suggesting that E7 and acid ␣-glucosidase may interact in the cytoplasm. To address this point, nuclear and cytoplasmic extracts of E7expressing cervical carcinoma (CaSki) cells were prepared. As expected, the nuclear protein lamin B was retrieved in the nuclear fraction, whereas both the cytosolic enzyme M2 pyruvate kinase and the ER-specific protein calreticulin were found exclusively in the cytoplasmic fraction, suggesting that a clean separation of nuclear and cytoplasmic proteins had been achieved (Fig. 4). In these experiments, a significant proportion of the E7 protein is retained in the cytoplasm, in keeping with the published literature (21).
To study further the subcellular localization of both proteins and to determine whether E7 and acid ␣-glucosidase can interact in living cells, the subcellular localization of both proteins was investigated by indirect immunofluorescence analysis. U2-OS cells, a human osteosarcoma cell line with high intrinsic transfection efficiency (35), were transfected with expression vectors for E7 and acid ␣-glucosidase. It was shown before that up to 48 h after transfection, ectopically expressed acid ␣-glucosidase is present predominantly in the 110-kDa precursor form, which localizes to the nuclear envelope, endoplasmic reticulum, and Golgi complex, similar to the localization of the endogenous 110-kDa protein (42). When U2-OS cells were tran-siently transfected with an expression vector for acid ␣-glucosidase, transfected cells could be easily identified by their bright staining, using an antibody to acid ␣-glucosidase. As was described by Wisselaar et al. (42), ectopic expression of acid ␣-glucosidase leads to the accumulation of the protein in the nuclear envelope, endoplasmic reticulum, and Golgi complex (Fig. 5A), suggesting that the signal derives from the 110-kDa form. Expression of E7 resulted in a similar cytoplasmic staining (Fig. 5A), suggesting that both proteins may be localized in similar compartments. Comparison to the Ͼ90% untransfected cells within the same coverslip establishes the specificity of both the E7 and acid ␣-glucosidase immunofluorescence signals.
To analyze further colocalization of E7 and acid ␣-glucosidase, both proteins were coexpressed by transient transfection. The cells were incubated with antibodies to both proteins and stained with anti-E7 antibodies (Fig. 5B, red fluorescence) and anti-glucosidase antibodies (Fig. 5C, green fluorescence). Both staining patterns were found to overlap, and superimposition of the pictures revealed large areas of colocalization, as revealed by yellow fluorescence (Fig. 5D). Together, these experiments clearly establish that E7 and ␣-glucosidase colocalize in mammalian cells.
E7-dependent Activation of Glucosidase Activity-To determine if E7 can modulate the function of acid ␣-glucosidase in vivo, we assayed the activity of the enzyme in extracts from 14/2 cells (34), in which expression of the E7 gene can be induced from a dexamethasone-inducible promoter (Fig. 6A). As shown in Fig. 6B, expression of E7 in 14/2 cells resulted in  (Table I). In control experiments using normal rat kidney cells, we found that the substrate affinity of acid ␣-glucosidase was not significantly affected by the addition of dexamethasone to the cells (Fig. 6C), strongly suggesting that the difference in enzymatic activity obtained in 14/2 cells depends on the expression of a functional E7 gene. To assess the ability of the C-terminal deletion mutant E7⌬79-83 to modulate the enzymatic activity of acid ␣-glucosidase in vivo, a new cell line was constructed in which the HPV-16 E7 mutant ⌬79-83 is expressed under control of the murine mammary tumor virus promoter. Expression of the E7 mutant in this cell line was verified by Western blotting. We found that both cell lines did not express E7 protein in the absence of dexamethasone, and in cells induced by addition of dexamethasone, the E7 mutant was expressed to a similar level as wild type E7 (Fig. 6A). We found that, in untreated cells, i.e. in the absence of any E7 protein, the activity of acid ␣-glucosidase was similar if not identical in extracts obtained from 14/2 cells and 14/⌬79-83 cells. In contrast to the results obtained with wild type E7 (Fig. 6B), expression of the E7 mutant did not significantly alter the enzymatic activity of acid ␣-glucosidase ( Fig. 6D and Table I), indicating that the sequence in the C terminus of E7 that mediates binding of E7 to acid ␣-glucosidase is also required for the modulation of the enzymatic activity.
To analyze if E7 may be able to modulate the activity of acid ␣-glucosidase in vitro, extracts prepared from non-induced 14/2 cells (E7 not expressed) were incubated with purified recombinant GST-E7 fusion proteins, followed by a determination of the acid ␣-glucosidase activity. In these experiments, we found FIG. 6. Modulation of acid ␣-glucosidase activity by E7 in vivo. A, expression level of the E7 protein in the cell lines 14/2 and 14/⌬79-83 after dexamethasone withdrawal (ϪDex.) and 4 h after induction of E7 by dexamethasone (5 g/ml; ϩDex.). 200 g of total cellular lysate was separated on a 12.5% SDS-polyacrylamide gel electrophoresis as indicated, transferred to a nylon membrane, and analyzed with a monoclonal antibody against HPV-16 E7 (Santa Cruz Biotechnology). B, enzymatic activity was measured in 14/2 cells after dexamethasone withdrawal (ϪDex.) and 4 h after induction of E7 by dexamethasone (5 g/ml; ϩDex.), respectively. C, activity of acid ␣-glucosidase was determined in NRK cells either without previous treatment or after incubation with dexamethasone (5 g/ml) for 4 h. D, enzymatic activity was measured in 14/⌬79-83 cells after dexamethasone withdrawal (ϪDex.) and 4 h after induction of E7 by dexamethasone (5 g/ml; ϩDex.), respectively. whereas GST alone had no effect (Table II). Addition of either the ⌬79-83 or GLY24 mutant of HPV-16 E7 did not significantly affect the substrate affinity of acid ␣-glucosidase (Table  II). Similarly, the catalytic activity of the purified recombinant 110-kDa form of acid ␣-glucosidase could be stimulated by purified E7 protein in vitro, whereas E7 mutants ⌬79-83 and GLY24 were much less active in this assay. In contrast, the activity of the 70/76-kDa enzyme, which is not bound by E7 (Fig. 3), was not affected by the viral oncoprotein. Taken together, these data suggest that the ability of E7 to bind the 110-kDa form of acid ␣-glucosidase is required for the modulation of the enzyme activity by E7, and activation of acid ␣-glucosidase by E7 is independent of any additional cellular proteins.
Modulation of Intracellular Glycogen Levels by E7-As a major function of acid ␣-glucosidase is the production of glucose from glycogen (45), we wondered whether expression of E7 in our experimental cell line would result in a change of the cellular glycogen content. To analyze this question, intracellular glycogen was determined in 14/2 cells in which E7 expression was either not induced or induced for 4 h. This experiment revealed that expression of E7 for 4 h resulted in the degradation of about 40% of the cellular glycogen content (Fig. 7). To test if increased degradation of glycogen may be related to the observed activation of acid ␣-glucosidase within the cells (see Fig. 6), conduritol B epoxide, a specific inhibitor of acid ␣-glucosidase (45), was added to the culture medium of the cells. Although we still noted a significant decrease of the glycogen content under these conditions, glycogen degradation was significantly reduced by conduritol B epoxide treatment. The results of three independent experiments demonstrate that 14.7 Ϯ 3% of the total cellular glycogen is degraded by acid ␣-glucosidase upon expression of E7 (Fig. 7). DISCUSSION We show here that HPV-16 E7 binds directly to the purified 110-kDa form of acid ␣-glucosidase and increases its normally very low substrate affinity (46) to a level that is even higher than observed for the mature form of the enzyme. For both the physical interaction and the modulation of enzymatic activity, no additional proteins are required, suggesting that E7 can act as a direct allosteric activator of acid ␣-glucosidase. Expression of wild type E7 but not a C-terminal mutant increases glucosidase activity in vivo, which leads to increased glycogen breakdown in vivo.
Although it is generally assumed that the interaction of HPV-16 E7 with members of the retinoblastoma protein family is critical for its ability to transform mammalian cells, it is now clear that the interaction of E7 with members of the pRb family is not sufficient for cell transformation. Mutations in the Cterminal part of E7, which leave pRb binding intact, render E7 unable to immortalize human keratinocytes (18,19) and to elicit the formation of warts in an animal model for papillomavirus-associated diseases (47). These findings suggest that domains in the C-terminal part of E7 target additional cellular pathways that are involved in the oncogenic activity of E7 (for recent review see Ref. 48). We have previously identified M2 pyruvate kinase as a new binding partner for HPV-16 E7, and we found that the interaction of E7 with M2-PK changes the catalytic properties of that enzyme (28), suggesting that E7 directly interferes with regulation of the cellular carbohydrate metabolism. Our present results extend these initial findings by showing that E7 directly targets another metabolic enzyme, acid ␣-glucosidase. Whereas both M2-PK and acid ␣-glucosidase bind to the C terminus of E7, the sequence requirements for both interactions are different, since E7 mutant GLY24 fails to bind acid ␣-glucosidase ( Fig. 3) but interacts with M2-PK with wild type affinity. 2 Together, these results establish that E7, via its C-terminal domains, controls additional, pRbindependent pathways that probably play a role in cell transformation.
The failure of the mature (70/76 kDa) form of acid ␣-glucosidase to bind E7 indicates that specific structural properties of the 110-kDa form determine its binding affinity. The 70/76-kDa form of acid ␣-glucosidase, which cannot bind to E7 in vitro, is found exclusively in the lysosomes, whereas the 110-kDa form is found associated with the nuclear envelope, endoplasmic reticulum, and Golgi complex (42). In biochemical fractionation experiments, a significant proportion of the E7 protein is retained in the cytoplasm (Fig. 4), in keeping with the published literature (21). The immunofluorescence analysis reported here suggests that part of the E7 protein is localized to the endoplasmic reticulum (Fig. 5), where it colocalizes with the 110-kDa form of acid ␣-glucosidase.
What could be the reason that an oncogenic virus like HPV-16 has evolved a protein, which targets acid ␣-glucosi-2 W. Zwerschke, unpublished results.

TABLE II
Modulation of acid ␣-glucosidase activity in vitro Extracts of non-induced (E7 not expressed) 14/2 cells and preparations of purified acid ␣-glucosidase were incubated with recombinant GST, GST-E7 wild type, and GST-E7 mutant proteins, as indicated. K m values were determined, using 4-MUG as substrate in varying concentration, as described. Mean values of at least three independent experiments Ϯ S.D. and the statistical significance are shown. NS, not significant.  dase? The data shown in Table I, Table II, and Fig. 3 would suggest that the ability of E7 to modulate acid ␣-glucosidase activity is genetically linked to cell transformation by HPV-16 E7. Together with our previous observation that E7 modulates the activity of type M2 pyruvate kinase (28), these data suggest that E7 has the potential to affect directly the cellular carbohydrate metabolism. This is not too surprising, given the fact that during malignant conversion drastic alterations of the carbohydrate metabolism are regularly observed (49). Whereas the precise role of glycogen turnover in tumorigenesis is still to be defined (for review see Ref. 3), the data reported here would suggest a direct link between a glycogen-degrading enzyme and a papillomavirus oncogene.