Fabio Dall Olio 1, Mariella Chiricolo 1, Erminia Mariani 2,3 and Andrea Facchini 2,3

Transcription

1 Eur. J. Biochem. 268, (2001) q FEBS 2001 Biosynthesis of the cancer-related sialyl-a2,6-lactosaminyl epitope in colon cancer cell lines expressing b-galactoside a2,6-sialyltransferase under a constitutive promoter Fabio Dall Olio 1, Mariella Chiricolo 1, Erminia Mariani 2,3 and Andrea Facchini 2,3 1 Dipartimento di Patologia Sperimentale, Università di Bologna, Italy; 2 Laboratorio di Immunologia e Genetica, Istituto di Ricerca Codivilla-Putti, I.O.R., Bologna, Italy; 3 Dipartimento di Medicina Interna e Gastroenterologia, Università di Bologna, Italy An elevation of b-galactoside a2,6-sialyltransferase (ST6Gal.I) enzyme activity and an increased a2,6-sialylation of cell membranes are among the most prominent glycosylation changes associated with colon cancer; both modifications correlate with a worse prognosis. In our previous studies, we have frequently observed a discrepancy between the ST6Gal.I level within a colon cancer sample or cell line and the respective level of reactivity with the a2,6-sialyl-specific lectin from Sambucus nigra (SNA). In this study, we have investigated quantitatively the biosynthesis of the sialyl-a2,6-lactosaminyl epitope in two colon cancer cell types expressing the ST6Gal.I cdna under the control of a constitutive promoter. By measuring the amount of ST6Gal.I mrna using competitive RT-PCR, the expression of a2,6-sialylated lactosaminic structures with SNA and anti-cdw75 Ig, and the presence of unsubstituted lactosaminic termini by Erythrina cristagalli lectin, we reached the following conclusions: (a) a high proportion of the cell surface lactosaminic termini remains unsubstituted, even in the presence of a very high ST6Gal.I activity. This proportion is strongly dependent on the cell type; (b) ST6Gal.I-transfected colon cancer cells do not express the CDw75 epitope; (c) the level of ST6Gal.I enzyme activity only partially correlates with the mrna level; (d) despite the control by a constitutive promoter, the ST6Gal.I mrna is not constantly expressed over time; and (e) a very large portion of the enzyme molecules is secreted in the extracellular milieu. These results indicate that posttranscriptional and post-translational mechanisms play a pivotal role in the control of a2,6-sialylation in colon cancer cells. Keywords: colon cancer; glycosyltransferases; glycosylation; a2,6-sialyltransferase; CDw75. The biosynthesis of the oligosaccharide moieties of glycoconjugates is a complex process and the fine regulation of this process is only partially understood. Although it is generally thought that the biosynthesis of a given oligosaccharide structure is regulated mainly by the expression of the cognate glycosyltransferase, the enzyme product relationship might in some cases correlate very poorly [1,2], suggesting that other factors play a relevant role. b-galactoside a2,6-sialyltransferase (ST6Gal.I) is the only enzyme able to catalyze the a2,6-sialylation of lactosamine (Galb1,4GlcNAc) [3], a structure commonly found in the antennae of N-linked chains but present also in O-linked chains and in glycolipids (reviewed in [4]). One of the most characteristic glycosylation changes in colon cancer is represented by an elevation of ST6Gal.I enzyme activity in cancer tissues, in comparison with the surrounding normal mucosa [5]. Clinical studies indicate Correspondence to Dr F. Dall Olio, Dipartimento di Patologia Sperimentale, Via S. Giacomo 14, Bologna Italy. Fax: , Tel: , dallolio@alma.unibo.it Abbreviations: DIG, digoxigenin; ECL, Erythrina cristagalli lectin; FITC, fluorescein isothiocyanate; L-PHA, L-phytohemagglutinin; SNA, Sambucus nigra agglutinin; ST6Gal.I, b-galactoside a2,6-sialyltransferase. (Received 17 July 2001, revised 20 September 2001, accepted 21 September 2001) that ST6Gal.I enzyme activity is further increased in metastatic tissues [6] and that a high level of ST6Gal.I in primary cancer is associated with a poor prognosis [7]. Also, the level of a2,6-sialylation of membrane glycoconjugates, as determined by the binding of NeuAca2,6Gal-specific lectins from Sambucus nigra (SNA) and Tricosanthes japonica, is increased in colon cancer [8 10] and a strong reactivity is associated with a worse prognosis [11]. Thus, it is important to clarify the relationship between these two parameters for potential clinical use. In recent studies, we have observed that the level of ST6Gal.I enzyme activity in a given colon cancer sample only partially correlates with the level of SNA reactivity, suggesting the existence of other mechanisms of regulation [12,13]. Moreover, while the increase of ST6Gal.I enzyme activity in colon cancer is a general phenomenon (the exceptions constituting, 10% of all cases), some studies reported an increase of ST6Gal.I mrna only in a subset of cases [1,14]. Together, these data suggest that multiple factors control the expression of the oligosaccharide product. Many of these factors might be cell specific, for example, the stability of the mrna, the efficiency of the translational machinery, the stability and the post-translational processing of enzyme proteins, the expression of other glycosyltransferases involved in the biosynthesis of backbone oligosaccharide structures. We have undertaken this study to investigate quantitatively the sequence of biochemical events, from mrna expression to the final oligosaccharide product, in a model system designed to minimize the contribution of these

2 q FEBS 2001 a2,6-sialyltransferase-transfected cells (Eur. J. Biochem. 268) 5877 cell-specific factors. Our model system is comprised of colon cancer cell clones derived from transfection of the human colon cancer cell lines SW48 and SW948 with the ST6Gal.I cdna driven by a constitutive promoter. These two cell lines are morphologically and phenotypically very different [15] but both are devoid of endogenous ST6Gal.I and do not contain any a2,3 sialyltransferases that may act on lactosamine [16]. Clones derived from transfection of the same cell type may be considered virtually identical, except for the level of ST6Gal.I expression. By measuring in the same cell preparations the amount of ST6Gal.I mrna, the level of enzyme activity and the expression of a2,6-sialylated sugar chains, we demonstrate that, even in a simplified model system, the levels of mrna and enzyme activity sometimes correlate poorly. In addition, we show that the level of a2,6-sialyl-substitution of lactosaminic termini may be more dependent on the cell type than on the level of ST6Gal.I enzyme activity. MATERIALS AND METHODS Cell lines and transfection experiments The cell lines SW48 and SW948 [15], both lacking a detectable ST6Gal.I activity, were routinely grown in Leibovitz s L-15 medium with 10% fetal bovine serum. The medium was changed regularly every 2 days. Preliminarily, both cell lines were cloned to minimize the cell heterogeneity and a random clone from each cell line was subjected to transfection with a full length rat liver ST6Gal.I cdna, expressed under the control of the CMV promoter in the eukaryotic expression vector pcdna/3 [17]. Control transfection was performed in parallel with an expression vector with no insert. G418-resistant clones were screened for SNA reactivity and some SNA-positive clones were expanded and further characterized. Cells were released when they were < 80% confluent by trypsin treatment and divided in three aliquots: one was stored at 280 8C for RNA preparation, one was used for FACS analysis and one for the preparation of a whole cell homogenate. Conditioned culture media were obtained as follows: an identical number of 48-6A3 cells was seeded in two flasks. Cells were allowed to recover for 2 days, then the cells of one flask were harvested, counted and used to prepare an homogenate, while the medium of the other flask was changed. After 24 h, the cells of the second flask were harvested, counted and used to prepare an homogenate, while the conditioned medium was filtered. The ST6Gal.I activity of the homogenates and of the conditioned medium was measured as detailed below. FACS analysis Approximately cells were washed with NaCl/P i containing 2 mg:ml 21 BSA (NaCl/P i /BSA) and incubated for 30 min in 50 mg:ml 21 FITC-conjugated lectins (Sigma, Milan, Italy) or 0.5 mg:ml 21 anti-cdw75 Ig (Pharmingen, San Diego, CA, USA) at 4 8C. In the last case, the incubation with anti-cdw75 Ig was followed by a 30-min incubation with a secondary, FITC-labeled goat anti-(mouse IgG) Ig. Negative control cells were incubated only with the secondary labelled antibody. After a wash with NaCl/P i / BSA, cells were resuspended in the same solution and analyzed by FACS. The analysis was performed by FACStar Plus (BD, Bedford, MA, USA) equipped with a 5-W argon ion laser operating at a wavelength of 488 nm in order to excite FITC fluorochrome, and green fluorescence was selected using a 515-nm long pass filter and a band pass 530F1 30 nm filter. The photomultiplyer was operating at 460 V. ST6Gal.I enzyme activity The protein concentration of the whole cell homogenate was determined by the Lowry method. ST6Gal.I enzyme activity was measured in the range of linearity with respect to time and enzyme concentration, using CMP-[ 14 C]sialic acid as a donor and asialotransferrin as an acceptor, as described previously [18], in the presence of 0.5% Triton X-100. The ST6Gal.I activity of culture media, conditioned for 24 h, was measured as above, but in the absence of detergents. Western blot analysis One hundred micrograms of protein from whole homogenates were electrophoresed under reducing conditions on a 7.5% polyacrylamide gel and electrotransferred to a Hybond membrane (Amersham, Little Chalfont, UK), as recommended by the manufacturer. Membranes were blocked by incubation for 1 h at room temperature with 2.5% blocking reagent (Boehringer, Mannheim) in NaCl/P i (20 mm sodium phosphate buffer, 130 mm NaCl) containing 2% Tween-20 (NaCl/P i /Tween). After three washes with NaCl/P i /Tween, blots were incubated for 1 h at room temperature with 1 mg:ml 21 digoxigenin-conjugated SNA (SNA DIG) (Boehringer) dissolved in NaCl/P i /Tween containing 10 mg:ml 21 BSA. Membranes were washed as above and incubated with horseradish-peroxidase-labeled anti-dig Ig (Boehringer) diluted 1 : in NaCl/P i / Tween containing 10 mg:ml 21 BSA for 1 h at room temperature. After a final wash, the reaction was developed with SuperSignal West Pico Chemiluminescent Substrate (Pierce, Rockford, IL, USA) and detected with an autoradiography film. Quantification of ST6Gal.I mrna Quantification of mrna was performed by competitive RT-PCR. Total RNA was extracted from frozen cell pellets by the RNaZolB method (Biotecx Laboratories, Houston, TX). cdna was prepared from 1 mg of total RNA, using the TaKaRa RT-PCR kit, version 2.1 (TaKaRa, Shouzo, Japan), according to manufacturer s instructions, with random 9-mers as primers. Firstly, the presence of the endogenous human ST6Gal.I transcript was assessed by subjecting the cdnas to PCR amplification for 30 or more cycles, using the conditions and the primer pair previously described [13]. The cdnas were then PCR amplified using either rat ST6Gal.I-specific primers FD1 (5 0 -ATTAAGCTTCTCTCC TGTCTGGACCAT-3 0 ) and FD2 (5 0 -ATTCTCGACTGGA GAAGGATAAGGGTG-3 0 ), for 28 cycles of the following program: 94 8C, 1 min, 60 8C 1 min, 72 8C 1 90 s, forming a 1284-nucleotide product, or b-actin-specific primers ACTL.2 (5 0 -CTGAACCCCAAGGCCAACCGCGAG-3 0 ) and ACTR.2 (5 0 -CCTGCTTGCTGATCCACATCTGCTG GAAG-3 0 ) with 24 cycles of the following program: 94 8C

3 5878 F. Dall Olio et al. (Eur. J. Biochem. 268) q FEBS s, 72 8C 210 s, forming a 754-nucleotide product [19]. Each reaction contained, in a final volume of 50 ml, 1 ml (for b-actin) or 2 ml (for ST6Gal.I ) of cdna, 1 Taq polymerase buffer, 1.7 mm MgCl 2, 0.2 mm dntps, 0.25 mm each primer and 0.5 U of InViTAQ DNA polymerase (Eppendorf, Milan, Italy). PCR products were analyzed on a 2% agarose gel, stained with ethidium bromide. The expression vector used for transfection was utilized as a ST6Gal.I standard. For the preparation of the competitor, the ST6Gal.I expression vector was treated with Eco47III, which cleaves a 362-nucleotide fragment from the insert, generating blunt ends. The open plasmid was gelisolated, ligated with T4 ligase and used to transform JM109 competent cells (Promega, Madison, WI, USA). The plasmids contained in four randomly chosen colonies were prepared and subjected to PCR analysis yielding a product of the expected size (922 nucleotides). The standard and competitor for b-actin were a kind gift of M. Trinchera, University of Insubria, Italy, and yielded products of 754 and 474 nucleotides, respectively. The expression of b-actin in the different cdna preparations was measured as follows and used to normalize the ST6Gal.I signal. A given amount of competitor (usually 10 pg) was co-amplified with various amounts of standard, ranging from 2.5 to 40 pg, and the intensity of the respective bands was quantified by the Kodak Digital SCIENCE 1D software. These data were used to generate a calibration curve, by plotting the amount of standard used vs. the standard/competitor ratio, essentially as described previously [1]. Sample cdnas were amplified in parallel, in the presence of the same amount of competitor. The standard/competitor ratio obtained for each sample was used to calculate the amount of target actin cdna by interpolation on the curve. The amount of ST6Gal.I mrna was quantified with the same approach, using 100 fg of competitor and variable amounts ( fg) of standard. The determined amount of ST6Gal.I mrna was normalized for the respective amount of b-actin mrna. RESULTS Relationship between ST6Gal.I expression and a2,6-sialyl-substitution of lactosaminic chains To study the effect of differential ST6Gal.I expression on the level of a2,6-sialyl-substitution of membrane glycoconjugates, the cell membrane of ST6Gal.I transfectants was probed with two lectins and subjected to FACS analysis. The lectin SNA recognizes the product of ST6Gal.I [20], while the lectin from Erythrina cristagalli (ECL) is specific for terminal unsubstituted lactosaminic chains [21]. Thus, the first lectin provides direct information on the degree of a2,6-sialylation, while the second provides information on the presence of unsubstituted lactosaminic chains. The flow cytometric profiles of three SW48-derived ST6Gal.I-transfectants (clones 48-N4, 48-4E1 and 48-6A3), one SW948-derived ST6Gal.I-transfectant (948-F4) and three mock-transfectants (48-neo1, 48-neo5, and 948-neo5) are shown in Fig. 1. The ST6Gal.I enzyme activity of the same cell preparations and their mean fluorescence with the two lectins are reported in Table 1. While the enzyme activity of neo clones was below the detection limit, the level of expression reached by clone 48-6A3 was higher than that of hepatocarcinoma HepG2 cells, a cell line known to express a Fig. 1. FACS of ST6Gal.I-transfected and mock-transfected cell lines stained with SNA-FITC or ECL-FITC. Cells were stained with the fluorescent lectins as described in Materials and methods and analyzed by FACS. Thin black line, autofluorescence; gray line, SNA- FITC; thick black line, ECL-FITC. Data are representative of three experiments.

4 q FEBS 2001 a2,6-sialyltransferase-transfected cells (Eur. J. Biochem. 268) 5879 Table 1. ST6Gal.I enzyme activity and lectin reactivity of ST6Gal.Iand mock-transfected cell lines. The same preparation of cells was used for the determination of ST6Gal.I enzyme activity and FACS analysis with lectins SNA and ECL. Enzyme activity data are the mean ^ SD of four independent determinations on the same homogenate. A value of 0 indicates an activity below the detection limit of 1.5 pmol:mg protein 21 :h 21. ND, not determined. Cell line Enzyme activity (pmol:mg protein 21 :h 21 ) Lectin reactivity (mean fluorescence channels SNA ECL 48neo neo N4 40 ^ E1 44 ^ A3 97 ^ neo F4 12 ^ HepG2 79 ^ 5 ND ND very high level of ST6Gal.I activity [18]. The SNA reactivity of neo clones was very low but the expression of ST6Gal.I activity, even at low levels, induced a shift towards SNA reactivity in all transfected clones. However, in SW48-derived clones, a progressive increase of activity (the activity of 48-6A3 is more than double than that of 48-N4) induced only a slight increase in SNA reactivity. Moreover, the decrease of ECL reactivity, which should parallel the increase of SNA reactivity, was negligible. The SNA and ECL peaks became progressively closer as the ST6Gal.I activity of the cell lines increased, but this phenomenon appears to be largely due to the increase of SNA reactivity, rather than to the decrease of ECL reactivity. This pattern was consistent with the presence, in SW48-derived clones, of a large number of lactosaminic termini, only a minority of which are accessible to ST6Gal.I. Consistent with a ST6Gal.I activity much lower than that of SW48-derived clones, the SNA reactivity displayed by 948-F4 cells was remarkably lower than that of SW48-derived clones. However, in the cell type SW948, this was sufficient to induce a remarkable decrease of ECL reactivity. It should be noted that the expression of ST6Gal.I at a very high level in 48-6A3 cells induced a shift in ECL reactivity of < 50 fluorescence channels (cf. the ECL reactivity of 48-neo1 or 48-neo5 with 48-6A3), while an activity more than eightfold lower induced a shift of ECL reactivity by < 150 channels in 948-F4 cells. This strongly suggests that in the SW948 cell type, the lactosaminic chains are more accessible to ST6Gal.I, resulting in a more pronounced masking of terminal galactose, even in the presence of low levels of enzyme activity. It has been previously shown that the presence of an antenna b1,6-linked to the trimannosyl core strongly reduces the overall ability of the N-linked chain to act as a substrate for ST6Gal.I [22]. To test whether the reduced accessibility displayed by the SW48 cell type was due to an higher degree of b1,6-branching of N-linked chains, two mock-transfected SW48 clones and two SW948 clones were FACS analyzed after staining with L-phytohemagglutinin (L-PHA), a lectin specific for the Fig. 2. FACS of mock-transfected cell lines stained with L-PHA- FITC. Cells were stained with L-PHA-FITC and analyzed by FACS, as described in Materials and methods. Thin black line, autofluorescence; thick black line, L-PHA-FITC. Representative of three experiments. b1,6-branching [23]. As shown in Fig. 2, the L-PHA reactivity displayed by the two SW48-derived clones was higher than that of the two SW948 clones, supporting the notion that an higher expression of b1,6-branched N-linked chains might be at the basis of the relative resistance to a2,6-sialylation displayed by SW48-derived cells. It is well known that the CDw75 antigen is a lymphocyte cell surface epitope, the expression of which is dependent on ST6Gal.I [24]. For this reason we asked whether ST6Gal.I expression is able to induce CDw75 reactivity in cell types that are not known to physiologically express CDw75. Figure 3 shows CDw75 expression and SNA reactivity of cell lines 48 6A3, 948-F4 and, as a positive control, of the B-lymphocytic cell line Louckes. It should be noted that all

5 5880 F. Dall Olio et al. (Eur. J. Biochem. 268) q FEBS 2001 Fig. 3. SNA and anti-cdw75 Ig flow cytometric analysis. CDw75-positive human lymphoblastoid Louckes cells, 48-6A3 cells and 948-F4 cells were probed with either anti-cdw75 Ig, followed by a secondary fluorescent Ig, or with SNA, and FACSanalyzed. The control was made by omitting the primary antibody. The mean fluorescence intensity is indicated. A representative of three experiments is shown. the three cell lines are reactive with SNA and that Louckes and 48-6A3 express similar levels of SNA reactivity, but only the former cell line reacts with anti-cdw75 Ig. This data confirms that the presence of a2,6-sialylated lactosamine is a necessary but not sufficient condition for the expression of the CDw75 epitope. An SNA Western blot analysis of mock- and ST6Gal.Itransfected clones (Fig. 4) revealed an overall weaker banding pattern in 48-neo1 and 48-neo5 in comparison with 48-N4, 48-4E1 and 48-6A3. The most remarkable Fig. 4. SNA Western blot analysis of mock- and ST6Gal.Itransfected cells. One hundred micrograms of cell homogenates were electrophoresed on a 7.5% gel in denaturing conditions and electrotransferred on a nitrocellulose membrane, which was successively stained with SNA-DIG as described in Materials and methods. Arrowheads indicate the migration position of prestained molecular mass standards. qualitative difference among these clones was the presence of SNA-reactive bands of high molecular mass (< 200 kda), which became progressively more intense with increasing activity. It is probable that the banding pattern of 948F4 cells is not distinguishable from that of 948-neo5 cells because of their low enzymatic activity. Analysis on a 3% acrylammide gel (not shown) allowed us to rule out the presence of SNA positive glycoproteins of very high molecular mass. Expression of ST6Gal.I mrna A preliminary investigation failed to reveal the presence of endogenous human ST6Gal.I transcripts after 30 cycles of PCR amplification. However, amplification for 35 or more cycles revealed the presence of very faint bands, indicative of an extremely low level of transcription of the endogenous gene (data not shown). The level of expression of rat ST6Gal.I mrna was carefully determined by competitive RT-PCR analysis in the same cell preparations used for enzyme assay and FACS analysis. The left panels of Fig. 5A show the PCR products generated by the co-amplification of a fixed amount of competitor and increasing concentrations of standard cdnas for b-actin and ST6Gal.I. These data were used to generate calibration curves by plotting the amount of standard used against the standard/competitor ratio; these curves were then used to calculate by interpolation the exact amount of actin and ST6Gal.I mrnas present in the cell preparations. In the right panels of Fig. 5A are shown the PCR products generated by the co-amplification of the cdnas from cell lines and the same amount of competitors used for the calibration curves. As expected, no ST6Gal.I mrna was present in the mocktransfected cells. The amount of ST6Gal.I mrna measured in the four ST6Gal.I-transfected clones, normalized for the relative amounts of actin cdna, is reported in Fig. 5B. The three SW48-derived clones expressed a level of mrna significantly higher than that of 948-F4 cells, in agreement with enzyme activity data. However, the relative level of expression among SW48-derived clones was not consistent with the corresponding enzyme activity data. In fact, 48-N4 cells, which showed the lowest enzyme activity, contained approximately double amounts of mrna than 48-4E1 and 48-6A3 cells. On the other hand, these two cell lines express a similar level of mrna, but the enzyme activity of the latter was about double than that of the former (see Table 1).

6 q FEBS 2001 a2,6-sialyltransferase-transfected cells (Eur. J. Biochem. 268) 5881 Fig. 5. Competitive RT-PCR analysis of mockand ST6Gal.I-transfected cell lines. (A) Using a fixed amount of competitor (10 pg for b-actin and 100 fg for ST6Gal.I) and the indicated amounts of standards, calibration curves were constructed, by plotting the standard/competitor ratio vs. the amount of standard used. cdnas from mock- or ST6Gal.I-transfected cell lines were amplified in the presence of the same amounts of competitors used for the calibration curve and their amount was calculated by interpolation on the curve. The size of standard and competitor products for ST6Gal.I is 1284 and 922 nucleotides, for b-actin is 754 and 474 nucleotides. (B) Amount of ST6Gal.I mrna, normalized for the b-actin mrna present in transfected cell lines. Data are the mean ^ SD of four independent determinations. *, significantly lower than 48-N4 using Student s t-test (P, 0.05); **, significantly lower than 48-N4 (P, 0.01), 48-4E1 and 48-6A3 (P, 0.05) using Student s t-test. (C) The ST6Gal.I mrna expression and the corresponding enzyme activity, measured in two different preparations of 48-6A3 cells. Even though it is generally assumed that transduced genes driven by a constitutive promoter are expressed at a constant level, this unexpected observation prompted us to investigate this point further. The mrna expression and the enzyme activity level were compared in two different preparations of the cell line 48-6A3 (Fig. 5C). 48-6A3 1 is the same preparation characterized in Fig. 5A, while 48-6A3 2 was harvested two months later. The two cdna preparations gave identical standard/competitor ratios for the actin gene, while the 48-6A3 2 preparation contained a double amount of ST6Gal.I mrna, but expressed a lower enzyme activity. Again, the level of enzyme activity did not parallel the amount of mrna. The same behavior was shown by different preparations of 48-N4 cells (data not shown). Together, these results indicate that: (a) the amount of mrna may undergo variations even when the gene is driven by a constitutive promoter; and (b) factors other than the amount of mrna profoundly affect the activity of ST6Gal.I. Secretion of ST6Gal.I A potentially important post-translational mechanism of ST6Gal.I regulation involves the secretion of a soluble, catalytically active portion of the molecule in the extracellular milieu. To assess the relevance of this mechanism in our experimental system, we determined the amount of soluble ST6Gal.I activity secreted in the extracellular milieu (Table 2). Unlike the cell-bound enzyme, which requires the presence of detergents for full activity, soluble ST6Gal.I was active in the absence of Triton X-100. Over a 24-h period, the total ST6Gal.I activity synthesized by the cells of the flask is the sum of the total cell-bound activity plus the total secreted activity minus the Table 2. Secretion of ST6Gal.I in the extracellular milieu by 48-6A3 cells. An identical number of 48-6A3 cells was seeded in two flasks. Cells were allowed to recover for 2 days, then the cells of one flask were harvested and counted, while the medium of the other flask was changed. After 24 h the cells and the conditioned medium of this flask were harvested and the ST6Gal.I enzyme activity present in the cells and media was measured. The data are the mean ^ SD of four independent determinations. The enzyme activity in the medium was measured in the absence of detergents. The low activity present in unconditioned medium, due to the presence of serum, was subtracted. The total cell-bound and secreted activities were calculated by multiplying the activity detected in a given amount of sample by the total amount of sample. Time (hours) Number of cells ST6Gal.I activity (pmol:mg protein 21 :h 21 ) Total cell bound ST6Gal.I activity Total secreted ST6Gal.I activity ^ ^ ^ ^ ^ 320

7 5882 F. Dall Olio et al. (Eur. J. Biochem. 268) q FEBS 2001 total cell-bound activity present at time 0, ie, [( ) 2 298] ¼ Of the total amount of activity synthesized by the cells in 24 h, < 78% (2711) is released in the extracellular milieu. These data indicate that a large portion of the ST6Gal.I molecules are continuously released. It is reasonable to expect that this mechanism plays an important role in determining the steady state level of the cell associated enzyme. DISCUSSION In this study we have investigated the sequence of molecular events involved in the biosynthesis of the sialyl-a2,6-lactosaminyl epitope in a transgenic model system. The use of transfectants generated from a single cell clone allows the cell-specific factors to be minimized, while the lack of any a2,3-sialyltransferase enzyme active on lactosamine in both cell types used for transfection [16] minimizes the possible competition between sialyltransferases. A comparison of enzyme activity and FACS analysis data of SW48-derived clones indicates that a minority of the Galb1,4GlcNAc termini are modified by ST6Gal.I, while the bulk remains unsubstituted, even in the presence of the very high level of enzyme activity displayed by 48-6A3 cells. In fact, the SNA reactivity of 48-6A3 cells is only slightly stronger than that of 48-N4 and 48-4E1, even in the presence of an approximately double level of enzyme activity, whereas the ECL reactivity is only slightly reduced by ST6Gal.I expression. In the different cell type SW948, the expression of a low level of ST6Gal.I activity results in a much more pronounced decrease of ECL reactivity, strongly suggesting that the Galb1,4GlcNAc termini are more readily accessible to ST6Gal.I in this cell type. Studies performed by van den Joziasse et al. [22,25] clearly showed that the activity of ST6Gal.I from bovine colostrum is profoundly affected by the number and types of lactosaminic branches carried by substrate N-linked chains. In fact, the presence of a branch b1,6-linked to the trimannosyl core (a structure commonly present in tetra-antennary N-linked chains) severely inhibits the activity of the enzyme towards all the lactosaminic branches of the molecule. Thus, a reasonable explanation for the relative resistance to a2,6-sialylation displayed by the SW48 cell type and for the better accessibility of glycans of the SW948 cell type might reside in the degree of branching of their N-linked oligosaccharides. This interpretation is supported by the higher reactivity for L-PHA displayed by the cell type SW48. The importance of the underlying sugar chain structure in determining the antigenic profile of the cell is highlighted by the fact that anti-cdw75 Ig failed to recognize a2,6-sialylated epitopes on the surface of the ST6Gal.I-transfected colon cancer cells, even in the presence of an SNA reactivity comparable with that of CDw75-positive Louckes cells. This happens despite the fact that the CDw75 antigen may also be expressed by nonlymphocytic cells, such as the malignant gastric epithelium [26]. As demonstrated by studies on T and B cells, the length of the polylactosaminic chains underlying the a2,6-sialyl-epitope plays a pivotal role in determining the reactivity for different a2,6-sialyl-specific antibodies [27]. The role of two classes of sugar binding proteins, namely siglecs and galectins, in mediating cell recognition phenomena is increasingly recognized. Members of the siglec family recognize sialic acid in different linkages, while galectins recognize terminal galactose, displaying a specificity somehow complementary to siglecs. Thus, the interactions of a cell with the lectins of the two classes is regulated by the balance between the terminal sialic acid and galactose residues. Current results show that this balance might be dramatically dependent on the cell type and relatively independent on the level of expression of ST6Gal.I. An accurate estimation of the ST6Gal.I mrna expression and a comparison with the corresponding level of enzyme activity leads to unexpected conclusions. Consistent with enzyme activity data, the level of mrna expression in 948-F4 cells is remarkably lower than that of SW48-derived clones. However, there is a clear discrepancy between the mrna level and the enzyme activity displayed by 48-N4 and 48-6A3 cells. This discrepancy can be explained assuming that the mrna level is not the only factor regulating enzyme activity. The observed fluctuation in the level of mrna (Fig. 5C) would not be surprising for a cell line expressing the endogenous sialyltransferase gene, but is unexpected for a gene driven by a constitutive promoter. Assuming that the transcription rate of the transduced gene remains unchanged over time, it can be hypothesized that the stability of the mrna might be influenced by the cellular metabolism. The fact that the cell preparation with the highest level of mrna display a lower level of enzyme activity points to post-transcriptional and post-translational mechanisms of regulation. Several of these mechanisms have been documented in the rat ST6Gal.I molecule. First, the activity may be dramatically affected by glycosylation [28,29]. Second, the enzyme can form disulfide-bonded dimers that are devoid of activity [30]. Third, the enzyme molecule exists in two variants differing for the presence of a Tyr or a Cys at position 123, with different catalytic efficiency and intracellular turnover [31]. Fourth, the enzyme may undergo phosphorylation [32]. Other points of regulation of ST6Gal.I enzyme activity occur at the level of sorting mechanisms. As demonstrated by studies on rat hepatoma cells, a large percentage of cell-bound ST6Gal.I molecules may be delivered to a degradation compartment soon after synthesis [33]. Moreover, the catalytic portion of ST6Gal.I may be released by rat liver [34] and in other model systems, including human colon cancer cell lines [35], through the proteolitic cleavage of the catalytic domain [34,36,37]. Recent data demonstrate that cells transfected with a cdna encoding only the soluble portion of ST6Gal.I, sialylate substrates much less efficiently than their counterparts transfected with the full length cdna [38], indicating that only the cell bound form is able to effectively sialylate substrates. Current results demonstrate that the secretion of ST6Gal.I is extremely active in our model system. The rate of secretion might conceivably be affected by many factors, the identification of which will require further investigation. However, it is reasonable to expect that secretion mechanisms play a pivotal role in determining the steady state level of the cell bound ST6Gal.I molecules. ACKNOWLEDGEMENTS We thank Dr Marco Trinchera of the University of Insubria, Italy for the gift of b-actin standard and competitor cdnas. This work was supported by grants from the University of Bologna (ex60% and selected research topics) and from Progetti Finalizzati del Ministero della Sanità.

8 q FEBS 2001 a2,6-sialyltransferase-transfected cells (Eur. J. Biochem. 268) 5883 REFERENCES 1. Kudo, T., Ikehara, Y., Togayachi, A., Morozumi, K., Watanabe, M., Nakamura, M., Nishihara, S. & Narimatsu, H. (1998) Up-regulation of a set of glycosyltransferase genes in human colorectal cancer. Lab. Invest. 78, Yang, J.M., Byrd, J.C., Siddiki, B.B., Chung, Y.S., Okuno, M., Sowa, M., Kim, Y.S., Matta, K.L. & Brockhausen, I. (1994) Alterations of O-glycan biosynthesis in human colon cancer tissues. Glycobiology 4, Weinstein, J., de Souza e Silva, U. & Paulson, J.C. (1982) Sialylation of glycoprotein oligosaccharides N-linked to asparagine. Enzymatic characterization of a Galb1,3(4)GlcNAc a2,3-sialyltransferase and a Galb1,4GlcNAc a2,6-sialyltransferase from rat liver. J. Biol. Chem. 257, Dall Olio, F. (2000) The sialyl-a2,6-lactosaminyl-structure: biosynthesis and functional role. Glycoconj. J. 17, Dall Olio, F., Malagolini, N., Di Stefano, G., Minni, F., Marrano, D. & Serafini-Cessi, F. (1989) Increased CMP-NeuAc: Galb1,4Glc- NAc-R a2,6-sialyltransferase activity in human colorectal cancer tissues. Int. J. Cancer 44, Gessner, P., Riedl, S., Quentmaier, A. & Kemmner, W. (1993) Enhanced activity of CMP-neuAc: Galb1 4GlcNAc: a 2,6-sialyltransferase in metastasizing human colorectal tumor tissue and serum of tumor patients. Cancer Lett. 75, Lise, M., Belluco, C., Perera, S.P., Patel, R., Thomas, P. & Ganguly, A. (2000) Clinical correlations of a2,6-sialyltransferase expression in colorectal cancer patients. Hybridoma 19, Dall Olio, F. & Trere, D. (1993) Expression of a2,6-sialylated sugar chains in normal and neoplastic colon tissues. Detection by digoxigenin-conjugated Sambucus nigra agglutinin. Eur. J. Histochem. 37, Sata, T., Roth, J., Zuber, C., Stamm, B. & Heitz, P.U. (1991) Expression of a 2,6-linked sialic acid residues in neoplastic but not in normal human colonic mucosa. A lectin-gold cytochemical study with Sambucus nigra and Maackia amurensis lectins. Am. J. Pathol. 139, Yamashita, K., Fukushima, K., Sakiyama, T., Murata, F., Kuroki, M. & Matsuoka, Y. (1995) Expression of Sia a2!6galb1!4glcnac residues on sugar chains of glycoproteins including carcinoembryonic antigens in human colon adenocarcinoma: applications of Trichosanthes japonica agglutinin I for early diagnosis. Cancer Res. 55, Vierbuchen, M.J., Fruechtnicht, W., Brackrock, S., Krause, K.T. & Zienkiewicz, T.J. (1995) Quantitative lectin-histochemical and immunohistochemical studies on the occurrence of a(2,3)- and a(2, 6)-linked sialic acid residues in colorectal carcinomas. Relation to clinicopathologic features. Cancer 76, Dall Olio, F., Chiricolo, M. & Lau, J.T. (1999) Differential expression of the hepatic transcript of b-galactoside a2,6-sialyltransferase in human colon cancer cell lines. Int. J. Cancer 81, Dall Olio, F., Chiricolo, M., Ceccarelli, C., Minni, F., Marrano, D. & Santini, D. (2000) b-galactoside a2,6-sialyltransferase in human colon cancer: contribution of multiple transcripts to regulation of enzyme activity and reactivity with Sambucus nigra agglutinin. Int. J. Cancer 88, Petretti, T., Kemmner, W., Schulze, B. & Schlag, P.M. (2000) Altered mrna expression of glycosyltransferases in human colorectal carcinomas and liver metastases. Gut 46, Leibovitz, A., Stinson, J.C., McCombs, W.B. III, McCoy, C.E., Mazur, K.C. & Mabry, N.D. (1976) Classification of human colorectal adenocarcinoma cell lines. Cancer Res. 36, Dall Olio, F., Malagolini, N. & Serafini-Cessi, F. (1992) Enhanced CMP-NeuAc: Galb1,4GlcNAc-R a2,6-sialyltransferase activity of human colon cancer xenografts in athymic nude mice and of xenograft derived cell lines. Int. J. Cancer 50, Dall Olio, F., Chiricolo, M., Lollini, P. & Lau, J.T. (1995) Human colon cancer cell lines permanently expressing a2,6- sialylated sugar chains by transfection with rat b-galactoside a2,6-sialyltransferase cdna. Biochem. Biophys. Res. Commun. 211, Dall Olio, F., Malagolini, N., Guerrini, S., Lau, J.T. & Serafini- Cessi, F. (1996) Differentiation -dependent expression of human b-galactoside a2,6-sialyltransferase mrna in colon carcinoma CaCo-2 cells. Glycoconj. J. 13, Bardoni, A., Valli, M. & Trinchera, M. (1999) Differential expression of b1,3galactosyltransferases in human colon cells derived from adenocarcinomas or normal mucosa. FEBS Lett. 451, Shibuya, N., Goldstein, I.J., Broekaert, W.F., Nsimba-Lubaki, M., Peeters, B. & Peumans, W.J. (1987) The elderberry (Sambucus nigra L.) bark lectin recognizes the Neu5Ac (a2-6)gal/galnac sequence. J. Biol. Chem. 262, Iglesias, J.L., Lis, H. & Sharon, N. (1982) Purification and properties of a D-galactose/N-acetyl-D-galactosamine- specific lectin from Erythrina cristagalli. Eur. J. Biochem. 123, Joziasse, D.H., Schiphorst, W.E., van den Eijnden, D.H., Van Kuik, J.A., Van Halbeek, H. & Vliegenthart, J.F. (1987) Branch specificity of bovine colostrum CMP-sialic acid: Galb1,4Glc- NAc-R a2,6-sialyltransferase. Sialylation of bi-, tri-, and tetraantennary oligosaccharides and glycopeptides of the N-acetyllactosamine type. J. Biol. Chem. 262, Hammarstrom, S., Hammarstrom, M.L., Sundblad, G., Arnarp, J. & Lonngren, J. (1982) Mitogenic leukoagglutinin from Phaseolus vulgaris binds to a pentasaccharide unit in N-acetyllactosaminetype glycoprotein glycans. Proc. Natl. Acad. Sci. USA 79, Bast, B.J., Zhou, L.J., Freeman, G.J., Colley, K.J., Ernst, T.J., Munro, J.M. & Tedder, T.F. (1992) The HB-6, CDw75, and CD76 differentiation antigens are unique cell- surface carbohydrate determinants generated by the b-galactoside a2,6-sialyltransferase. J. Cell Biol. 116, Joziasse, D.H., Schiphorst, W.E., van den Eijnden, D.H., Van Kuik, J.A., Van Halbeek, H. & Vliegenthart, J.F. (1985) Branch specificity of bovine colostrum CMP-sialic acid: N-acetyllactosaminide a2,6-sialyltransferase. Interaction with biantennary oligosaccharides and glycopeptides of N-glycosylproteins. J. Biol. Chem. 260, David, L., Nesland, J.M., Funderud, S. & Sobrinho-Simoes, M. (1993) CDw75 antigen expression in human gastric carcinoma and adjacent mucosa. Cancer 72, Kniep, B., Schakel, K., Nimtz, M., Schwartz-Albiez, R., Schmitz, M., Northoff, H., Vilella, R., Gramatzki, M. & Rieber, E.P. (1999) Differential expression of a2 6 sialylated polylactosamine structures by human B and T cells. Glycobiology 9, Chen, C. & Colley, K.J. (2000) Minimal structural and glycosylation requirements for ST6Gal I activity and trafficking. Glycobiology 10, Fast, D.G., Jamieson, J.C. & McCaffrey, G. (1993) The role of the carbohydrate chains of Galb-1,4-GlcNAc a2,6-sialyltransferase for enzyme activity. Biochim. Biophys. Acta 1202, Ma, J. & Colley, K.J. (1996) A disulfide-bonded dimer of the Golgi b-galactoside a2,6-sialyltransferase is catalytically inactive yet still retains the ability to bind galactose. J. Biol. Chem. 271, Ma, J., Qian, R., Rausa, F.M. III & Colley, K.J. (1997) Two naturally occurring a2,6-sialyltransferase forms with a single amino acid change in the catalytic domain differ in their catalytic activity and proteolytic processing. J. Biol. Chem. 272, Ma, J., Simonovic, M., Qian, R. & Colley, K.J. (1999) Sialyltransferase isoforms are phosphorylated in the cis-medial Golgi on serine and threonine residues in their luminal sequences. J. Biol. Chem. 274,

9 5884 F. Dall Olio et al. (Eur. J. Biochem. 268) q FEBS Bosshart, H. & Berger, E.G. (1992) Biosynthesis and intracellular transport of a-2,6-sialyltransferase in rat hepatoma cells. Eur. J. Biochem. 208, Kaplan, H.A., Woloski, B.M., Hellman, M. & Jamieson, J.C. (1983) Studies on the effect of inflammation on rat liver and serum sialyltransferase. Evidence that inflammation causes release of Galb1,4GlcNAc a2,6-sialyltransferase from liver. J. Biol. Chem. 258, Dall Olio, F., Malagolini, N. & Serafini-Cessi, F. (1992) The expression of soluble and cell-bound a2,6-sialyltransferase in human colonic carcinoma CaCo-2 cells correlates with the degree of enterocytic differentiation. Biochem. Biophys. Res. Commun. 184, Colley, K.J., Lee, E.U., Adler, B., Browne, J.K. & Paulson, J.C. (1989) Conversion of a Golgi apparatus sialyltransferase to a secretory protein by replacement of the NH 2 -terminal signal anchor with a signal peptide. J. Biol. Chem. 264, Weinstein, J., Lee, E.U., McEntee, K., Lai, P.H. & Paulson, J.C. (1987) Primary structure of b-galactoside a2,6-sialyltransferase. Conversion of membrane-bound enzyme to soluble forms by cleavage of the NH 2 -terminal signal anchor. J. Biol. Chem. 262, Zhu, G., Allende, M.L., Jaskiewicz, E., Qian, R., Darling, D.S., Worth, C.A., Colley, K.J. & Young, W.W. Jr (1998) Two soluble glycosyltransferases glycosylate less efficiently in vivo than their membrane bound counterparts. Glycobiology 8,