4.2. Recommendations for Negative Screens

A standard dropout viability screen (negative selection screen) relies on the fact that some of the shRNAs in the screen are either cytotoxic or cytostatic (presumably by interfering with an essential target gene). Cells with shRNAs that do not inhibit growth, then, grow normally populating the culture in which the cells with the lethal shRNA do not propagate. The endpoint analysis involves looking for shRNA sequences that are underrepresented or depleted in the sample population relative to the original library.

Length of the Screen

For a dropout viability screen to work, the cells need to be cultured long enough for the cells with unaffected growth to significantly increase their proportion relative to the affected cells. For this to occur, the cells need enough time in culture for multiple doublings. The length of any particular screen may need to be altered depending on the specifics (e.g. cell growth rates, types of targets of interest, if additional compounds are used). However, typically, we find allowing for ~10 population doublings after transduction to be optimal.

If the screen is not run long enough, all the shRNA counts will be in a narrow range and it will be difficult to identify significantly depleted shRNA sequences from background variability. If the screen is run too long, the range of representation of shRNA sequences will become broader due to the natural growth variance in different cells in the population. This phenomena, often referred to as genetic drift, will increase the background variance of the screen and, if the spread becomes too broad, can also make it difficult to identify significantly depleted shRNAs from background variability. For optimal results, it is important to run the screen long enough that shRNAs that have an effect on cell growth/viability will be distinct from background variance but not to the point where background variance becomes large enough to confound the ability to call hits.

For a negative selection screen, the puromycin or other antibiotic selection it is not essential, but provides a way to reduce the total number of cells and makes cell culture handling more manageable. In a typical screen about 30%-40% cells are transduced, the remaining 60%-70% of cells without virus are not needed. Unless you want to maintain a larger than needed cell population throughout the screen, it makes sense to perform antibiotic selection to get rid of unwanted cells.

MOI of Transduction, Number of Cells, Amount of DNA

As mentioned above, the number of cells stably transduced with the shRNA library at the time of transduction should exceed the complexity of the shRNA library by at least 200-fold. For a library with 27,500 shRNAs, the starting population should be at least 5.5 million infected cells, and for a library with 55,000 shRNAs, the starting population should be at least 11 million infected cells. The MOI of transduction should be kept at or below 0.5, to ensure that the majority of transduced cells carry only one integrated provirus. After transduction, the ideal is to never discard any cells at any time during the experiment (e.g. at treatment, harvesting, DNA purification, etc.). However, this is often not practical— especially for a negative screen where most of the cells propagate normally. If the number of cells becomes too large and you are forced to discard a fraction, the number of remaining cells should always exceed the complexity of the library by at least 1,000-fold (e.g. keep at least 27 million cells after every splitting step, for a 27K library). Similarly, when amplifying barcodes from isolated DNA, you should always use all the genomic DNA recovered from cell samples, up to the amount corresponding to 1,000 cells × the number of shRNA in the library.

Baseline Controls for Negative Screens

In a simple screen aimed at identifying shRNAs which are cytotoxic in a given cell line, we typically use the library itself as the baseline control, since the shRNA frequency distribution in plasmid and packaged lentiviral library is virtually identical. The plasmid library has already been sequenced as part of the QC when we made the library, so it is not necessary to re-sequence the library at this point. If you would also like to use transduced cells as a baseline control, typically we recommend harvesting and sequencing genomic DNA from them by about 18 hours post-transduction so that the cells have not had too much time to proliferate and express shRNAs to affect the library distribution. In more complex experiments, aiming at identifying differential toxicity between isogenic cell lines, or between compound-treated and non-treated cells, other baselines will be needed.