It's not at all uncommon for effects that occur in vitro (in test tubes, basically) to fail to match effects that occur in vivo (inside the living organism). It's always pleasing when we can figure out one of these discrepancies.
In a recent article in PLoS One titled Lectin-Based Food Poisoning: A New Mechanism of Protein Toxicity, Miyake, Tanaka, and McNeil resolve a curious discrepancy between what happens inside your intestines and what's been seen in the lab. It's been known for a long time that ingesting lectins -- proteins that bind carbohydrates -- from certain improperly cooked vegetables can cause toxicity, including nausea and diarrhea. This lectin exposure causes cell death in those cells lining the intestine, but, confusingly, cause no toxicity when directly applied to cells in culture, even at very high doses.
In this study, the researchers tested a novel hypothesis -- perhaps lectins don't affect intact membranes, but instead mess with membrane repair? Indeed, your intestinal cells are subject to a host of stresses, and are constantly repairing their membranes via exocytosis, a process lectins are known to inhibit. The researchers damaged the membranes of individual cells with a laser or with physical abrasion and then tested their ability to fix themselves in the presence or absence of lectins. They found that, fitting their proposed model, lectins stuck on the cells' surfaces dramatically inhibited membrane repair.
This is a clean, classic science piece, putting together a number of observations to make a new hypothesis, and then testing that hypothesis. It's also a good reminder that your intestines are a very dynamic place, constantly undergoing damage and reconstruction.