Figure 1.
Schematic of the CIRES concept.
The expression patterns of about 1000 glycan-related genes were profiled in a set of six different cell lines (A–F) by comparing the microarray binding of cellular cDNA and reference polyA(+) RNA and calculating the relative expression values (Table S1). The polygons in the left web graphs represent the relative gene expression profiles of eight glycan-related genes selected as examples. In these graphs, the difference in relative gene expression is expressed on a log scale, where the edge of the polygon corresponds to the strongest expression in each cell line (A–F). The same set of six cell lines were examined for cell-surface glycan expression using fluorescently labeled plant lectins and flow cytometry; the strength of the glycan expression is plotted as relative values among the six lines, where the edge of the polygon represents the strongest expression (web graph on top right). The glycan expression profiles were analyzed for correlations with the glycan-related gene expression profiles. Similarities and dissimilarities between the profiles were assessed using Pearson's correlation coefficient, which has values ranging from -1 (no correlation) to 1 (perfect correlation). A complete list of the genes found to be positively or negatively correlated with plant lectin staining patterns is presented in Table S2. Genes known to affect the biosynthesis of an epitope were selected from among the correlated genes (shown for each lectin in the tables on the right in Figures 2–6). A correlated gene identified by CIRES was confirmed as the gene responsible for regulating the biosynthesis of a particular glycan by transferring the gene into another cell line of the set, via gene transfer techniques such as retrovirus-mediated overexpression, and looking for a related change in epitope expression.
Figure 2.
CIRES analyses of staining profiles obtained using lectins with known epitope expression-regulating enzymes.
(A, C, D) Expected glycan structures for lectin recognition (left), web graphs of the lectin staining profiles (depicted as polygons) obtained using a set of six B-cell lines (middle), and the correlation indexes (CI, Pearson's correlation coefficient for profile matching) of the relevant genes that correlated with the plant lectin staining profiles and the P values of the correlations (right). The correlation orders of the glycan-related genes selected from the complete list of correlated genes (Table S2) are indicated as numbers in parentheses in the box for each gene, with a smaller number indicating a stronger correlation between gene expression and glycan expression profiles. Genes with a negative correlation are indicated by an N before the order number. The lectins used were (A) PHA-L4, (C) SSA, and (D) PNA. Lectin epitopes shown in the figures are taken from the literature unless otherwise specified [17], [51]. (B) Namalwa cells were infected with MSCV harboring MGAT5-IRES-EGFP. Control cells were infected with empty vector (IRES-EGFP) or the same vector encoding B3GNT2 or MGAT3. Flow cytometry results for PHA-L4 staining were compared between EGFP-positive cells (solid line) and EGFP-negative cells (dashed line).
Figure 3.
CIRES analyses of staining profiles obtained using lectins that recognize terminal glycan structures and have unknown epitope-expression-regulating enzymes.
Presentation is the same as in Fig. 2 except that the plant lectins used were (A) LCA, (B) UEA-I, and (C) RCA 120. N.D. in the gene order list indicates that no gene was determined to have a correlation with the lectin staining.
Figure 4.
CIRES analyses of staining profiles obtained using lectins that recognize internal glycan structures and have unknown epitope-expression-regulating enzymes.
Presentation is the same as in Fig. 2 except that the plant lectins used were (A, B) DSA and (C–E) PHA-E4. N.D. in the gene order list indicates that no significant correlation was detected. (B, D) Flow cytometric staining patterns for EGFP-positive Namalwa cells are shown in bold lines, and those for EGFP-negative (control) cells are shown in gray dashed lines. The overexpression of MGAT5 resulted in a subtle (60%) increase in DSA staining. The overexpression of MGAT3 resulted in a 2-fold increase in PHA-E4 staining. (E) PHA-E4 lectin blotting was performed using the membrane fraction of the same set of cell lines. A plot of the quantified signals reveals differences in the PHA-E4 staining profile among the six cell lines (C, E), as discussed in the text.
Figure 5.
CIRES analyses of staining profiles obtained using lectins that recognize multiple glycan structures and have unknown epitope expression-regulating enzymes.
Presentation is the same as in Fig. 2 except that the plant lectins used were (A–B) MAM, (C) WGA, and (D) Con-A. The epitopes of the two different lectins of MAM, MAL and MAH, are illustrated separately. WGA essentially recognizes a cluster of N-acetyl groups, as indicated, and thus required Neu5Ac as a Sia species. Con-A recognizes mannose-containing glycans with varying affinities. High-mannose-type glycans (upper diagram in (D)) bind best to this lectin. (B) Namalwa cells were infected with retroviruses encoding various GlcNAc transferases. The MAM staining patterns of the EGFP-positive populations of each infectant are shown. MGAT5 overexpression did not shift the staining pattern in comparison with the vector control (data not shown).
Figure 6.
CIRES analyses of staining profiles obtained using the lectin ABA.
(A) Presentation is the same as in Fig. 2 except that ABA was used. (B) Effect of sialidase treatment on the binding of ABA and PNA in B cells. (See text for the specificity of the sialidase and Fig. 2D for the PNA epitope.) Mean fluorescence intensity (MFI) values for the staining with each lectin (bold lines) are shown at the top. Dashed lines indicate the results from the non-staining control.