There are no delays on our products or support as a result of COVID-19. If there is a way we can assist you, we are here to help - Contact us

Investigating KRAS synthetic lethal/co-dependency interactions using siRNA and CRISPR

A number of siRNA and shRNA screens have identified targets that exhibit differential dependencies between KRAS mutant and KRAS wild-type tumours, but there is poor overlap between these published studies. Next generation screens that exploit both isogenic cell lines and cancer cell panels, and use a combination of knockdown (si/shRNA) and knock-out (CRISPR-Cas9-sgRNA) methodologies might be more effective at identifying novel targets that withstand validation. However, if we are to detect co-dependence as well as synthetic lethal interactions, screens must be performed under conditions where mutant KRAS alleles are essential for growth.

At Horizon Discovery we are combining our expertise in cancer cell biology and functional genomic screening to provide a more definitive target validation cascade in KRAS mutant cell lines. Our initial results indicate that RAF1 is an important mediator of mutant KRAS biology. Below we highlight some of the work generated using DLD1 colorectal carcinoma and A549 lung carcinoma cell lines, but you can also read about our study of KRAS function in a panel of non-small cell lung carcinoma cell lines.

3D conditions reveal a dependency on oncogenic KRAS

Horizon Discovery's proprietary gene editing technology - rAAV - was used to knockout the endogenous KRASG13D mutation from DLD-1 colon carcinoma cells. This creates a pair of isogenic cell lines that differ only in their KRAS mutation status and which avoids confounding genetic factors. In a standard 2D format, proliferation rates of the cell line pair were identical. However, knockout of KRASG13D reduced proliferation rates in 3D (Figure 1).

Figure 1. Proliferation of DLD-1 cells in 3D is dependent on KRAS. DLD1 isogenic cells were cultured for 72h (2D) or 7 days (3D: soft agar assay, 0.4%) before proliferation was quantified using Alamar Blue (10% v/v).

Similarly, inhibition of KRAS expression by siRNA also had a greater impact on proliferation in 3D, confirming that the dependency of DLD-1 cells on KRAS for robust proliferation is only evident under 3D conditions (Figure 2).

Figure 2. DLD1 cells are more sensitive to KRAS knockdown in 3D. DLD1 cells were transfected with siRNA using Lipofectamine RNAimax and seeded into soft agar for 7 days before proliferation was quantified using Alamar Blue (10% v/v). Plates were scanned using the IncucyteZOOM to record representative images.

Screening in 3D sensitises cells to knockdown of KRAS synthetic lethal targets

Collectively our data indicate that the proliferation of DLD-1 cells in 3D culture is significantly reduced when expression of either WT or oncogenic KRAS is compromised. Therefore, we asked whether genes reported as synthetic lethal targets of oncogenic KRAS have a greater effect on DLD-1 cell proliferation in 3D culture.

Several siRNAs to putative KRAS synthetic lethal targets had an effect on the viability of in DLD-1 cells cultured in 2D conditions and for most targets this effect was increased in 3D culture conditions (Figure 3).

Figure 3. Putative KRAS synthetic lethal targets have a greater effect on cell proliferation in 3D. DLD-1 cells were screened in 2D (72h) and 3D (7d) using an siRNA library containing putative KRAS synthetic lethal targets. Proliferation was assessed using Alamar Blue (10%v/v). siRNAs are ranked according to their sensitivity in 2D. Symbol size is an indication of significance.

Targets such as PLK1, TBK1, BCL-XL and RAF1, where multiple publications suggest a link to KRAS mutant cancers, showed a co-dependency with KRAS (Figure 4A). These findings were extended to a panel of KRAS-mutant colon lines, where the requirement for RAF1 expression correlated highly with the requirement for KRAS expression (R=0.89; p=0.0032) (Figure 4B).

Figure 4. Hit confirmation using the DLD1 2D/3D assay and KRAS colon cell line panel. (A) A subset of siRNA hits from the screen show increased sensitivity in 3D in DLD1 cells. (B) A panel of KRAS mutant colon lines were screened in 2D (72h) using siRNA to RAF1 or KRAS.

Validating RAF1 as a potential drug target

Although we found good correlation between sensitivity to KRAS and RAF1 depletion, we were unable to unambiguously validate RAF1 as a target by si/shRNA methods (Figure 5).

Figure 5. Effective depletion of RAF1 by shRNA does not result in sustained anti-proliferative phenotype. (A) Five dox-inducible shRNAs targeting RAF1 were transduced into A549 cells, where four showed similar antiproliferative effects. (B) Western blot data showing all shRNAs tested were equally efficient at depleting RAF1 protein levels.

Using CRISPR-Cas9, we found knockout of RAF1 expression was more anti-proliferative, demonstrating the advantages of using a more penetrant sgRNA based screening approach (Figure 6). At later timepoints, we did see cell outgrowth in wells treated with RAF1 sgRNA, but we anticipate this population is made up of cells containing in-frame deletions that result in a functional RAF1 protein: this hypothesis can be confirmed by PCR fragment analysis.

Figure 6. Disruption of RAF1 by sgRNA results in an anti-proliferative phenotype in A549 cells. Five sgRNA sequences that resided upstream of the kinase domain were designed using our in-house Guidebook algorithm. These were cloned into an "all-in-one" lentiviral vector, pLentiCRISPR (v2) (Sanjana et al 2014) containing the individual sgRNA, Cas-9 enzyme and puromycin resistance gene and used to produce virus. A549 cells were spinfected with virus and selected for 7 days in puromycin before seeding cells for proliferation assays (confluency) using the IncucyteZOOM .


  • Using X-MAN DLD-1 isogenic cell lines, we have shown that 3D culture formats can more readily reveal KRASG13D dependency
  • We have validated this system using a siRNA panel of putative KRAS synthetic lethal targets which are more anti-proliferative in 3D
  • CRISPR-Cas9 approaches may provide better validation tools for targets, such as kinases, where RNA interference technologies lead to ambiguous results

Some of this work is part of the European research program COLTHERES (Modeling and predicting sensitivity to targeted therapies in colorectal cancers) financed by the European Commission within the 7th Framework Program

(Grant agreement no: 259015)