When cells use DNA that encodes for short, hairpin RNA, the RNA they produce following these DNA instructions is converted by the cell into a short stretch of double-stranded RNA. One of these strands is then taken up by an enzyme, present in all cells, which uses it as a “guide strand” to find and destroy any matching strands of messenger RNA. This prevents that matching messenger RNA from being used to produce whatever protein the message describes (hence, the term, “RNA interference,” which is a way to prevent the production of a particular protein.) Cells are normally doing this all the time, using hairpins of RNA they produce for their own purposes, to regulate their own protein production.
We “hijack” this mechanism to reduce the production of a specific protein of our choice (such as huntingtin) when we add DNA to cells encoding for our designed shRNA. However, if too much of our “designer” shRNA is made by the cell, the resulting guide strands can hog too much of the available enzyme in the cell, displacing the guide strands the cell is normally using to regulate itself, and disrupting the well-being of the cell. In this case, harm is done to the cell not because the shRNA reduced the production of the protein it was designed to target. It could target anything, or nothing. Rather, harm is done because the presence of too much shRNA occupied too much of the cell’s mechanisms for using guide strands — hence the term, “mechanistic toxicity.”