We read, with considerable dismay, a recent Research Article on cholera toxin (CT) internalization(Torgersen et al., 2001), in which the authors extensively challenged methods, results and conclusions that we had published four years ago (Orlandi and Fishman, 1998). As space limits a point-by-point rebuttal of their comments and critique of the many deficiencies in their study, we encourage readers to evaluate our response by comparing both papers. As a preface to our reply, we state that most aspects of CT intoxication are generally accepted, such as its structure and receptor, its mechanism of retrograde trafficking through the Golgi and ER, its mechanism of activating adenylyl cyclase and its pathophysiological effects on human enterocytes. However, we believe that the crux of the dispute is our differing views on the relationship between CT internalization and intoxication. Whereas most of the cell-surface-bound CT is internalized, only a small percentage is activated on release of the enzymatic A1 peptide(Kassis et al., 1982; Orlandi and Fishman, 1998). Thus, to understand the mechanism of CT action, one must determine not only the pathway(s) for CT internalization but also whether the uptake leads to intoxication of the cell.
Our paper focused on whether both CT internalization and activation are mediated by caveolae or by detergent-insoluble glycolipid-enriched complexes(DIGs) (also known as lipid rafts) in cells deficient in caveolin and caveolae. Torgersen et al. primarily were interested in showing caveolae-independent endocytosis of CT. Our approach was to compare CT uptake and action in three cells that have no, low or high levels of caveolin and caveolae, and to use the cholesterol modifiers filipin and β-cyclodextrin(βCD) to selectively inhibit caveolae/DIG-mediated endocytosis. Chlorpromazine (CPZ) and diphtheria toxin (DT) served as inhibitor and probe for clathrin-mediated uptake. One of our cell lines, human intestinal CaCo-2,played a major role in their study and appears to be the source of many of their repetitious complaints. By using anti-CT-A1 antibodies to quantify CT uptake, we found 58% inhibition by filipin. As Torgersen et al. found only 17% inhibition by using a different method, they speculated that our assay may have overestimated CT uptake if the antibodies could not reach CT clustered in the narrow necks connecting caveolae to the cell surface, and if filipin could somehow alter the necks and increase antibody binding. They ignored our second assay in which cells were labeled with rhodamine-conjugated CT-B at 15°C. When warmed at 37°C, there is an extensive redistribution of fluorescence from the plasma membrane to the perinuclear region that is blocked by filipin but not CPZ. Even when Torgersen et al. found filipin to be ineffective on two other cell lines, they failed to show that the filipin was active. Filipin is known to be unstable in solution. This led them to a circular argument: as CT uptake is only slightly inhibited by filipin, it must not be via caveolae/DIGs. Thus when they found that βCD inhibits CT internalization in CaCo-2 cells by 43% (similar to our 39%), they concluded that the uptake is clathrin-dependent based on the weak effect of filipin and cited studies showing that βCD also blocks the latter pathway. Surprisingly, two of the four references cited were not relevant. We found that both βCD- and filipin-treated, but not CPZ-treated, CaCo-2 cells remain sensitive to DT. Others have shown that these agents selectively inhibit caveolae/DIG-mediated, but not clathrin-mediated, endocytosis in a variety of cells (Puri et al.,2001; Wolf et al.,2002).
We also assayed CT activation and activity by A1 and cAMP formation, respectively. Filipin totally blocks both CT activation and activity in CaCo-2 cells, and βCD inhibits CT activity by 98%. Thus, both filipin and βCD are more effective in inhibiting CT activity than endocytosis. Although filipin- and βCD-treated cells still internalize substantial amounts of the bound CT by other pathways, CT remains inactive. In this regard, filipin blocks CT-B trafficking from plasma membranes to Golgi,but not clathrin-mediated endocytosis of CT-B and transferrin in COS-7 cells(Nichols et al., 2001). We found that filipin also inhibits CT activity in A431 and Jurkat cells that are rich in and lacking caveolin and caveolae, respectively. Thus, the disruption of CT intoxication of cells independent of the presence of caveolin and caveolae led us to conclude that CT internalization and activation are mediated through cholesterol- and glycolipid-rich microdomains rather than a specific morphological structure. Torgersen et al. challenged our thesis by asserting, “DIGs have been proposed to act as the vehicle for CT entry in Jurkat T lymphoma cells (Orlandi and Fishman, 1998), but there are no data indicating how DIGs might be internalized.” We refer them to a review(Simons and Ikonen, 1997) and an article on endocytosis of a GPI-anchored protein through DIGs in Jurkat cells (Deckert et al.,1996).
Torgersen et al. chide us for not investigating the role of dynamin in caveolae-mediated uptake and finally for suggesting that CaCo-2 cells have caveolae based on small amounts of caveolin. The papers that link dynamin with caveolae-mediated uptake were published while ours was in press(Oh et al., 1998; Henley et al., 1998). Although dynamin is now known to be involved in both clathrin- and caveolae-mediated endocytosis, we are not aware of any role in DIG-mediated uptake. Regarding the presence of caveolin and caveolae in CaCo-2 cells, others agree with us(Mayor et al., 1994; Field et al., 1998). Regardless, our major thesis is that of the role of DIGs and not caveolae per se in CT internalization and intoxication. Finally, we are not dogmatic about our conclusions as some cell types may use a different pathway for CT activation. Neurons, although lacking caveolin/caveolae, have DIGs to which CT binds, but the internalization and activation of the toxin is clathrin-mediated (Shogomori and Futerman,2001).