Expression of Bisecting GlcNAc in Pediatric Brain Tumors and Its Association with Tumor Cell Response to Vinblastine¹

Alteration of N-linked oligosaccharide expression during malignant transformation is a well-described phenomenon and has been observed in many tumors types (1, 2, 3). The β1–4 GlcNAc branched oligosaccharide structure, commonly referred to as bisecting GlcNAc, is characterized by the presence of a β 1,4 GlcNAc linkage to the central mannose as shown below:

\

/

Formation of the β 1,4 branching is catalyzed by GnT-III³(4), and the resulting product interacts specifically with E-PHA lectin (5). This lectin has, therefore, been used extensively as a tool to determine expression of the bisecting GlcNAc in various normal and abnormal tissues (6). Expression of the bisecting GlcNAc has been detected on several types of glycoproteins including serum proteins (7), differentiation-related molecules (8), and growth factor receptors (9, 10). This molecule has also been found to be associated with hematological and hepatological malignancies. Elevated GnT-III activity was detected in patients with chronic myelogenous leukemia (CML) in the blast crisis as compared with healthy controls or with patients with other hematological malignancies (11). In a recent study, GnT-III activity, measured in 159 serum samples from patients with hepatocellular carcinoma, has been found to be enhanced as compared with healthy controls (7). This increase in GnT-III activity seemed to be specific because patients with other hepatological diseases such as cirrhosis or chronic hepatitis did not show any changes in the activity of this enzyme. More recently, we have reported that the bisecting GlcNAc structure is expressed in glioma cell line U373 MG (9), which suggests that this structure may also be associated with nonhematological and nonhepatological malignancies. The physiological significance of its overexpression in these tumors, however, is not yet understood.

One of the major problems encountered during tumor treatment with chemotherapeutic drugs is the development of MDR. Overexpression of P-glycoprotein on continuous exposure of tumor cells to antineoplastic agents represent the best described mechanism to explain the development of drug resistance (12, 13, 14). This protein encoded by the MDR gene (mdr1) is a transmembrane energy-dependent efflux pump that, among many chemotherapeutic agents, has vinblastine as a substrate (15). P-glycoprotein has three N-glycosylation sites that contribute to its proper routing and stability (16). The nature of these glycosylation sites and the effect of their targeting on the function of P-glycoprotein and on the tumor cell response to drugs, however, have not yet been studied.

In the present study, we have investigated expression of the bisecting GlcNAc in pediatric brain tumors and its possible implication in the modulation of P-glycoprotein-mediated cell response to vinblastine. For these purposes, we have conducted a histochemical study in a series of 90 pediatric brain tumor and normal brain tissue specimens. The implication of this molecule in the modulation of glioma cell response to vinblastine was investigated in a glioma cell culture model by measuring cell sensitivity to this drug in the presence and in the absence of E-PHA lectin. The effect of this lectin on drug resistance mediated by overexpression of P-glycoprotein was also determined in a glioma cell line transfected with the mdr1 gene.

Eighty archival specimens, paraffin-embedded, were obtained from patients with pediatric brain tumors at the time of operation. Normal brain samples were obtained at autopsy from 10 patients without any brain pathology at the Children’s Memorial Hospital (Chicago, IL). Tumors included 10 juvenile pilocytic astrocytomas (grade I), 8 low-grade astrocytomas (grade II), 13 anaplastic astrocytomas (grade III), 6 glioblastoma multiforme (grade IV), 14 medulloblastomas, 25 ependymomas, 1 oligodendroglioma, and 3 cerebral neuroblastomas. The tumors were classified and graded according to the criteria of WHO (17). The patients’ ages varied between a few days to 16 years. Human glioma cell lines U373 MG and U251 were purchased from the American Type Culture Collection (Rockville, MD). The U251 MDR cells were provided as a gift by Dr. Alexander Kolchinsky (18). DMEM and FBS were purchased from BioWhittaker (Walkersville, MD). Vinblastine, rhodamine 123, the MTT, E-PHA, biotinylated E-PHA, and agarose-linked E-PHA were purchased from Sigma (St. Louis, MO). Antibody C19 against P-glycoprotein was purchased from Santa Cruz Biotechnology Inc. (Santa Cruz, CA). Streptavidin-biotinylated horseradish peroxidase complex and the enhanced chemiluminescence reagents (ECL) were purchased from Amersham (Arlington Heights, IL) Immobilon-P transfer membrane for Western blot was purchased from Millipore (Bedford, MA).

Sections were mounted on microscope slides and then deparaffinized and hydrated according to the method described previously (19). Lectin histochemical staining was performed after blocking the endogenous peroxidase, aldehyde groups, and nonspecific binding sites with 3% H₂O₂, ammonium chloride, and 3% BSA as reported previously (20). The sections were then incubated with biotinylated E-PHA [2 μg/ml in 20 mm Tris buffer (pH 7.4)] for 1 h at room temperature. After washing with PBS, streptavidin-biotinylated peroxidase complex was added and incubated at 1:500 dilution in Tris buffer for 45 min. E-PHA binding was detected by the addition of the peroxidase substrate according to the manufacturer’s instructions. The results were recorded as follows: (a) negative when there was no staining; (b) positive when staining was encountered throughout the specimen; and (c) focal positive staining when less than 10% of cells were stained.

Glioma cell response to vinblastine was carried out by using a modified MTT assay (21). The cells were seeded at 10⁴ cell/well in 96-well plates and maintained in culture for 24 h at 37°C in DMEM supplemented with 10% FBS. The medium was then changed to serum-free DMEM with or without E-PHA (25 nm). Two h later, vinblastine was added at the indicated concentrations and incubated for an additional 48 h. The effect of these treatments on cell proliferation was assayed by the MTT precipitate formation technique (21). Briefly, the wells were washed twice with PBS, and the cells were incubated with 0.5 mg/ml MTT in PBS for 4 h at 37°C. PBS was then aspirated and the MTT precipitate was solubilized in 100 μl of HCl 0.5n in isopropanol. The absorbance of this solution was measured at 450 nm, and the cell survival was estimated in per cent related to the nontreated cells.

Expression of the mdr1 gene that codes for P-glycoprotein was measured by reverse transcription PCR. The sequences GAA GGC CTA ATG CCG AAC AC and CAG TGA CTC GAT GAA GGC ATG, were used as the forward and reverse primers, respectively, to amplify a 415-bp fragment of the mdr1 gene. Total RNA was extracted from 75-cm² flasks by using the Tri-reagent technique as described by the manufacturer. cDNA was made, and amplified by using a thermocycler (Perkin-Elmer, Branchburg, NJ) and the following PCR conditions: denaturation at 94°C for 20 s, annealing at 55°C for 20 s, and extension at 72°C for 30 s. The three-step PCR reaction was performed for 35 cycles.

An analysis of P-glycoprotein glycosylation in U373 MG cells was carried out as follows: (a) the cells were seeded in 75-cm² flasks containing DMEM supplemented with 10% FBS and were incubated at 37°C; (b) when the cells reached confluency, they were washed with cold PBS, and the monolayer was solubilized by the addition of 200 μl of lysis buffer [50 mm HEPES (pH 7.4), 150 mm NaCl, 100 mm NaF, 1 mm MgCl₂, 1.5 mm EGTA, 10% glycerol, 1% Triton X-100, 1 μg/ml leupeptin, and 1 mm phenylmethyl sulfonyl fluoride]; (c) insoluble material was removed by centrifugation during 5 min at 10,000 × g; (d) the cell lysates (500 μg of proteins) were incubated at 4°C overnight with 100 μl of E-PHA linked to agarose beads in 1 ml of PBS containing 1 mm MgCl₂ and 1 mm CaCl_2; (e) the beads were then washed four times with PBS; (f) 30 μl of Laemmli buffer was added to the pellet; (g) the mixture was boiled for 5 min and proteins in the supernatant were separated by SDS-PAGE; (h) the presence of P-glycoprotein was detected by Western blot using anti-P-glycoprotein C19; and (i) reactive complexes were detected by strepavidin-peroxidase and ECL reagents as described previously (9).

Cells were seeded in DMEM containing 10% FBS onto round glass coverslips in 24-well plates at 10⁴ cell/well. After 24-h incubation at 37°C, the medium was changed to a serum-free DMEM. Rhodamine 123 (80 μm) was added and incubated for 30 min at 37°C (22). The cells were then washed three times with serum-free DMEM and incubated in the same medium in the presence or the absence of E-PHA lectin (25 nm) for 4 and 15 h, respectively. After washing three times with PBS, the glass coverslips containing the cells were mounted onto microscope slides, and photographs were taken under fluorescence microscopy.

Expression of the bisecting GlcNAc in normal brain and brain tumor specimens was examined on paraffin-embedded material. Detection of this structure was carried out by lectin histochemistry with E-PHA as described in the “Materials and Methods” section. Representative staining are shown in Fig. 1, and the total histochemistry results are displayed in Table 1. As shown in Fig. 1 (Normal brain) normal brain tissue was not stained by E-PHA except for blood vessels (in the center of the illustration). This result agrees with the finding reported earlier in the normal rat brain, in which no E-PHA staining was detected (20). Taken together, these data suggest that normal brain tissue, at least in these two species, does not express the bisecting GlcNAc structure. An analysis of pediatric brain tumor specimens showed that most of the low-grade astrocytomas (Grade I and II) tested in our study were negative. A representative sample (Fig. 1, Low Grade Astrocytoma) displayed a staining pattern similar to that of the normal brain tissue, with staining mainly associated with blood vessels. In high-grade astrocytomas (Grade III and IV), E-PHA reactivity was, however, strongly enhanced and consistently observed in the majority (73%) of samples. Similarly, most (72%) of the ependymoma specimens were E-PHA-positive. In medulloblastomas, the most frequent brain tumor in children, however, the majority of specimens tested were E-PHA-negative.

A detailed analysis of these cases (Table 1) showed that E-PHA staining was consistently negative in all of the normal brain specimens. Different levels of reactivity were, however, observed in pediatric brain tumors, particularly in astrocytomas and ependymomas. These results suggest that expression of the bisecting GlcNAc is differentially expressed in pediatric brain tumors as compared with the normal brain tissue. A more detailed analysis of E-PHA staining in astrocytomas showed that the proportion of E-PHA positive cases vary with the tumor grade. Of the 10 grade-I astrocytomas, one was focally positive and one was positive for E-PHA staining. In the grade-II astrocytomas, two of the eight cases were positive. There is, however, a strong increase in the proportion of E-PHA positive cases in high-grade astrocytomas (grade III and IV) where 14 of 19 cases are positive. In this category, all of the six glioblastomas (grade IV) tested were found to be strongly positive. This interesting finding suggests that expression of the bisecting GlcNAc in astrocytomas may have a useful application as a possible marker for tumor progression. Expression of the bisecting GlcNAc also appeared to be consistently enhanced in ependymomas (Fig. 1 and Table 1). Of 25 specimens tested, 18 were E-PHA positive. The proportion of positive cases was very similar between high grade astrocytomas (Grade III and IV) and ependymomas, although these two tumor types have different morphological phenotypes. Other brain tumor specimens tested (including 14 medulloblastomas) displayed only a few E-PHA-positive or focal positive cases.

One common problem in the management of high-grade astrocytomas and ependymomas is their ability to develop resistance to chemotherapy (23, 24). Disruption of drug transport across plasma membrane is one factor that affects tumor cell response to cytotoxic drugs. To understand the implication of the bisecting GlcNAc in the modulation of brain tumor response to drugs and its effect on P-glycoprotein function, we used a human glioma cell line U373 MG that express the bisecting GlcNAc (25). Cell response to vinblastine was studied in the presence and in the absence of E-PHA. As shown in Fig. 2, pretreatment of U373 MG cell line with E-PHA (25 nm) for 1 h at 37°C prior to the addition of vinblastine enhanced sensitivity to this drug, whereas treatment with E-PHA alone had no effect. These data indicate that the bisecting GlcNAc may be an important factor in modulating vinblastine transport in U373 MG cells.

To determine whether E-PHA affects cell sensitivity to vinblastine through interaction with P-glycoprotein, we first studied the expression of this transporter and its ability to interact directly with E-PHA in U373 MG cells. Expression of the MDR gene (mdr1) that codes for P-glycoprotein was evaluated in these cells by reverse transcription-PCR. As shown in Fig. 3,A, a DNA fragment with an expected length of 415 bp was detected in U373 MG cells, which indicated that the molecular message for P-glycoprotein was expressed in these cells. P-glycoprotein itself was also found to be expressed in U373 MG cells (Fig. 3,B). As indicated in Fig. 3 B, we have found that P-glycoprotein is among the glycoproteins precipitated by E-PHA linked to agarose beads. This suggests that at least one of the N-glycosylation sites of P-glycoprotein contains the bisecting GlcNAc structure. These data also suggest that direct interaction between E-PHA and P-glycoprotein may affect the drug efflux function of this transporter and, therefore, provide an explanation for the enhancement of glioma cell sensitivity to vinblastine in the presence of E-PHA.

Determination of rhodamine 123 efflux is a widely used technique to evaluate P-glycoprotein mediated drug efflux (22). To determine whether E-PHA affects the P-glycoprotein-mediated drug efflux function, we measured the effect of this lectin on rhodamine 123 efflux from U373 MG cells. After the cells were loaded with rhodamine, they were washed, and the culture medium was changed. Rhodamine efflux was measured at 4 and 15 h of incubation. In the absence of E-PHA, most of the dye disappeared after 4 h, and the efflux was complete after 15 h of incubation in the new medium (Fig. 4). In the presence of E-PHA, however, a large amount of rhodamine 123 was retained in the cells even after 15 h of incubation. The inhibition of rhodamine efflux by E-PHA constitutes an additional argument in favor of direct interaction between E-PHA and P-glycoprotein. P-glycoprotein-mediated drug efflux could be inhibited through the bisecting GlcNAc structure present on this transporter.

The inhibition of P-glycoprotein function by E-PHA may suggest that drug resistance mediated by overexpression of P-glycoprotein may also be reversed by this lectin. To verify this hypothesis, we used a drug-resistant glioma cell line, U251 MDR, generated by transfection with the mdr1 gene. Generation of this cell line and the characterization of its resistance to cytotoxic drugs as compared with the wild-type cell line U251, have been reported previously (18). As shown in Fig. 5, sensitivity to vinblastine in the U251 MDR cells was strongly reduced as compared with the nontransfected cells, which indicated that overexpression of P-glycoprotein in these cells results in increased resistance to this drug. Cotreatment of U251 MDR cells with E-PHA lectin and vinblastine resulted in enhancement of cell sensitivity to this drug. The synergistic action between E-PHA and vinblastine in these cells is in agreement with the fact that targeting the bisecting GlcNAc on glioma cells may inhibit P-glycoprotein drug efflux function and reverses MDR mediated by this transporter.

The purpose of the present work was to study expression of the bisecting GlcNAc in pediatric brain tumors and to study its implication in the modulation of glioma response to chemotherapy. Antibodies able to interact with the bisecting GlcNAc structure are not commercially available. We, therefore, used E-PHA lectin, which has been shown to interact specifically with this structure (5), to detect its expression in a series of 90 pediatric brain tumors and normal brain tissue specimens. Our results show that negative staining with E-PHA lectin was observed in all of the 10 normal brain tissue specimens tested (Fig. 1 and Table 1). A similar observation has been reported recently in which there was also no detectable expression of the bisecting GlcNAc in any part of the rat brain tissue (20).

The bisecting GlcNAc structure is, however, expressed in pediatric brain tumors, and it is related to only certain tumor types (Fig. 1 and Table 1). The differential expression of the bisecting GlcNAc between normal brain and these tumors suggests that this molecule may be potentially useful as a marker for tumor diagnosis. More importantly, comparison of the expression of this bisecting GlcNAc within the different tumor grades in astrocytoma (Table 1) showed that the proportion of E-PHA-positive specimens seemed to increase progressively with tumor grade. This observation suggests that expression of the bisecting GlcNAc may also be relevant in predicting tumor progression in certain subsets of brain tumors. Interestingly, the expression of E-PHA in ependymomas was correspondingly high (72% versus 73% in astrocytomas). Similarity between these two tumor types have also been reported in the expression of other molecular markers, such as EGF receptor (26) or the calcium-binding protein S100 β (27). This similarity is not surprising inasmuch as gliomas seemed to be able to differentiate into ependymomas (28, 29), and malignant ependymomas were shown to de-differentiate toward glioblastoma multiforme (30). Our data may, therefore, represent an additional argument in favor of the relationship between these tumors.

Because the bisecting GlcNAc is differentially expressed in gliomas as compared with the normal brain tissue, we hypothesized that this molecule can be used to target the tumor tissue and to affect tumor cell behavior. One of the major problems in chemotherapy treatment of gliomas is that continuous exposure of tumor cells to chemotherapeutic drugs generally leads to the development of MDR. In view of this, we have investigated the potential implication of the bisecting GlcNAc in modulating glioma cell response to chemotherapy. Treatment of U373 MG cells with E-PHA enhances their response to vinblastine (Fig. 2), which suggests that the bisecting GlcNAc structure in glioma may have a physiological role in the modulation of tumor behavior. Enhancement of cell sensitivity to vinblastine by E-PHA suggests that this lectin may interact with P-glycoprotein and inhibits its drug efflux function. We have shown that P-glycoprotein is not only expressed in U373 MG cells but that this transporter also contains the bisecting GlcNAc structure required for E-PHA binding (Fig. 3). These data provide an explanation to the observed enhancement of cell response to vinblastine on treatment with E-PHA and suggest that interaction between this lectin and P-glycoprotein may inhibit the drug efflux mediated by this transporter. Evaluation of P-glycoprotein-mediated drug efflux is usually determined by rhodamine 123 efflux (22). In our cellular model, we have found that the treatment of glioma cells with E-PHA results in the inhibition of rhodamine efflux (Fig. 4). This suggests that the inhibition of P-glycoprotein function by E-PHA results in drug sequestration into the cell and, therefore, in increased drug toxicity. Our findings also suggest that expression of the bisecting GlcNAc structure in glioma cells, particularly on P-glycoprotein, has the potential to be used as a target to reverse drug resistance that characterize these tumors. In agreement with this, E-PHA lectin has been found to reverse resistance to vinblastine because of overexpression of P-glycoprotein in the U251 cells transfected with the mdr1 gene (Fig. 5).

In conclusion, we have observed that the bisecting GlcNAc is differentially expressed in pediatric brain tumors as compared with the normal brain tissue. In astrocytomas, expression of this molecule seemed to correlate with the tumor grade and, therefore, may represent a useful tool for diagnosis and grading of these tumors. Modulation of glioma cell sensitivity to vinblastine through the bisecting GlcNAc suggests that this molecule may have the potential to be considered as a unique target for development of new drugs against pediatric brain tumors.

We thank Dr. Alexander Kolchinsky (University of Illinois, Chicago, IL) for providing us the resistant (U251) cell line and Dr. Barbara Mania-Farnell and David George for their valuable assistance in the preparation of the manuscript.