Select Page

DECIPHER Pooled shRNA Library Viability Screen

Below is an example screen and data analysis using a DECIPHER library. Please also refer to the user manual before proceeding with your experiments.

Objective

The screen aims to identify prostate-specific gene products with therapeutic potential in prostate neoplasia treatment. A viability screen is performed on a set of model prostate carcinoma cell lines. Data from DU145 cell line is presented below.

Experiment Planning

We are interested in finding examples of shRNA that are selectively toxic to prostate cancer, thus, a negative selection (viability) screen was chosen.

There are two sets of controls to include in the experiment. The first is intended to verify that the screen is able to discriminate between shRNA targeting essential and nonessential genes. DECIPHER libraries include luciferase shRNA as internal baseline reference; these are the sequences that are known to be neither depleted nor enriched. We do not expect to see positive selection in this screen and so include no enrichment control. A positive control for viability screen would be shRNA with known general cytotoxic properties. There are over 50 known cytotoxic genes that were confirmed in previous DECIPHER screens and would be expected to show up in the list of hits. Examples include the proteins from the ubiquitin-proteasome pathway such as PSMA3, PSMA7, PSMC1, PSMC6, PSMD3, and RPL30.

As we are interested in the prostate-specific gene products, a set of specificity controls was also included by running several reference non-prostate cell lines: K562 (hematopoietic), HeLa, Hep3B, and 293 (epithelial), all in triplicate.

An important decision to be made at the planning stage is knockdown time. After selection step, the cells should be grown long enough to allow shRNA expression and knockdown, yet not too long to minimize the effects of genetic drift and clonal selection which may bias the results and make them less reproducible between the triplicates. Based on our previous data, we chose to grow cells for 7 days post-transduction with a puromycin selection step 48 hours post-transduction. The cells should be maintained without splitting, as this enhances the effects of genetic drift; thus, slow-growing cells can be kept for ten days or longer, while fast-growing cells will overrun the practical size limit of the experiment rather quickly.

In this experiment, we used HT-sequenced plasmid library as a control for non-selected library representation. However, using an early time point is preferable, as it will be a control not only for the heterogeneity at the sequencing and insertion steps, but also variability at the packaging and transduction steps. For the first time performing an experiment, a time course is highly advisable to find the optimal point where the depletion is most pronounced yet the effects of genetic drift are still minimal. Possible intervals of 1, 5, 10 or 1, 7, 14 days post-transduction are suggested, depending on the cell growth rate.

Workflow

Figure 1 presents a workflow for this experiment. The initial packaging step is not depicted below as pre-packaged library was available. For information on how to order pre-packaged DECIPHER libraries, please contact Cellecta at sales@cellecta.com.

Workflow of a DECIPHER shRNA Library screen

Figure 1. Workflow of a DECIPHER shRNA library screen.

Data Analysis

High-throughput sequencing data for this experiment was returned as raw data in QSEQ format, broken up in 100 files per lane. These were merged into a single file with TXTCollector (http://bluefive.pair.com/txtcollector.htm), and then converted to TAB format using Cellecta’s Deconvoluter software. Next, TAB files were processed by Deconvoluter into .csv files containing identified and annotated sequences and counts for statistical analysis. Please see Deconvoluter manual for details.

Quality control for reproducibility in replicates using DECIPHER pooled shRNA libraries

Figure 2. Quality control for reproducibility in replicates using DECIPHER pooled shRNA libraries.

Figure 2 graphically represents the quality control for reproducibility. Each panel is one replicate plotted against the median baseline values on a log2 scale. The points are annotated in color depending on their distribution around the midline in panel A (replicate 1) and plotted in panels B and C (replicates 2 and 3) using the same colors. As evident from the color distribution, the vast majority of points assume the same positions in all three replicates. Black points represent luciferase shRNA. These cluster tightly around the midline, showing that the effects of genetic drift are small and ruling out non-functional selection.

Scatter plot of shRNA representation for a viability screen with DU145 prostate cells. Time points are day 0 and day 8.
# of redundant shRNAsTotal Genes: 408
2310
359
421
512

Figure 3. Scatter plot of shRNA representation for a viability screen with DU145 prostate cells. Time points are day 0 and day 8.

Figure 3 presents a scatter plot of shRNA representation at day 0 (sequenced plasmid library) vs. day 8, each point a median of replicates performed. The majority of shRNA exert no selective pressure and are neither enriched nor depleted. These are represented by the points clustering close to midline. Most fall within two-fold deviation from the midline, as marked off by the parallel lines. The points marked in black represent the hits, i.e. statistically significant depletion in sequence representation. Amount of depletion, reproducibility in replicates, and the number of unique depleted shRNA sequences per gene were considered in determining the hits. This is done to exclude off-target effects of individual shRNAs. Table 2 summarizes the results, with a total of 408 genes identified.

Luciferase, our negative control, is not shown in Figure 3.

Figure 4. Pie Chart of biological processes represented in DU145 prostate viability screen. Click here for a larger view.

The 408 hits can be further classified using the GO categories. Figure 4A shows the breakdown by biological processes, with over 50% being involved in metabolism, cellular processes, or communication. Figure 4B breaks down the results by cellular components. Figure 4C – by molecular processes, and Figure 4D – by protein classification.

Appendix 1 [see DECIPHER-DU145-Viability-Gene-Hits-and-Prostate-Biomarkers.xls] shows 50 genes identified in the screen that are known to have altered expression profiles in prostate cancer. This result serves as the screen’s proof of principle, demonstrating its ability to identify known hits.