4. Results
4.2 Optimization of systematic high-content cell screening in gemcitabine-treated
While gemcitabine is the predominating chemotherapeutic drug in pancreatic adenocarcinoma, the survival after diagnosis of this type of cancer has not improved substantially over the years. Investigating Chk1 and MK2 in the gemcitabine response, we could show that the cellular fate following this treatment highly relies on DNA damage mediators. To shed light on novel factors which influence the cellular response to gemcitabine treatment, we performed a high-content screen using RNA interference (RNAi) to deplete pancreatic cancer cells of single proteins. After that, we treated the cells with gemcitabine and used the immunofluorescence-based quantification of γH2AX accumulation as global readout for DNA damage signaling. We aimed to identify so far unknown mediators of the gemcitabine response in pancreatic cancer cells. The screening conditions were optimized as follows.
4.2.1 siRNA-mediated knockdown efficiency and cell growth morphology
The efficiency of RNAi does not only depend on sequence and quality of the siRNA construct, but strongly varies between cell types and reagents. As RNAi is the method on which the high-content screen for DNA damage mediators was based, knockdown conditions needed to be optimized in a way that at least established siRNA constructs reliably led to the depletion of the respective target protein. Therefore, we first tested the siRNA-mediated knockdown with standardly used siRNAs and transfection reagents. Of all reagents, LF 2000 achieved the best cell survival and knock down efficiency after trans-fection (data not shown). To quantify knockdown efficiency, all were transfected with siRNA
to p53 and absolute p53 levels were determined using quantitative immunofluorescence analysis (Fig. 4.6, right column). Although BxPC-3 cells have the lowest absolute p53 levels, knockdown of this protein was not successful. This excluded the cell line for the screening procedure. All other cell lines displayed good knockdown efficiencies.
Another aspect in a cell line’s suitability for quantitative immunofluorescence is the accessibility of the region of interest (ROI). As we defined the nucleus as ROI, it is essential that the image processing software can separate neighboring cell nuclei. This is not possible with cells growing in multiple cell layers. We therefore took bright light microscopy images
Figure 4.6 siRNA-mediated knockdown efficiency and cell growth morphology of pancreatic cancer cells.
(a) BxPC-3, (b) MIA PaCa-2, (c) PANC-1 and (d) PaTu8902 were transfected with siRNA targeting p53 and fixation and staining for immunofluorescence was performed 48 h after transfection. p53 levels were determined using quantitative
immunofluorescence and total p53 levels were used as criteria for the RNAi potential of the cell line. Light microscopic pictures of control-transfected cells were taken to assess cell growth morphology.
of the different cell lines at different levels of confluence (Fig. 4.6, left column), showing that MIA PaCa-2 and PANC-1 cells tend to detach from the dish surface and MIA PaCa-2 and PaTu8902 cells grow in multiple cell layers. Both effects resulted in a substantial cell population not available for quantitative immunofluorescence analysis.
Thus, BxPC-3 could not be used for efficient transfection, while all other cell lines had restrictions which led to a decrease in ROIs suitable for γH2AX quantification.
4.2.2 Treatment and readouts
Gemcitabine-induced H2AX phosphorylation occurs in a dose- and time-dependent manner. Gemcitabine dose and treatment time were optimized to be high enough to distinguish γH2AX levels in treated and in untreated cells but not to induce maximum γH2AX which would overlay the effects of a knockdown. In the screen, we wanted to be able to examine the influence of knockdowns on γH2AX at a time point when other mediators of the DDR are still activated and later onset processes as DNA repair and apoptosis do not predominate yet.
As expected, γH2AX levels increased with gemcitabine concentrations (data not shown). At intermediate concentrations of gemcitabine (200 nM), increased H2AX phosphorylation was evident 24 h after the beginning of treatment and further intensified after that.
In contrast, Chk1 phosphorylation, considered earlier in the DDR than H2AX phosphorylation, started to increase already after 12 h of gemcitabine treatment and peaked after 24 h with markedly decreasing levels after that (Fig. 4.7). The activation of Chk1 went along with an increase in total Chk1 protein levels, which remained elevated when Chk1 phosphorylation already decreased. This already marks an adaption process of the cell to cope with enhanced replicative stress and could have influence on the readout. To further exclude γH2AX induction being a secondary effect of apoptosis, PARP cleavage was determined. In PANC-1 cells, PARP cleavage was not yet detectable after 24 h of treatment, but clearly present after 48 h (Fig. 4.7).
Thus, our results demonstrate that, for our purpose, treatment of cells with 300 nM gemcitabine for 20 h to 24 h is ideal. We therefore chose these conditions for screening.
As already detailed in section 1.2.2, gemcitabine mainly acts during the process of DNA replication. Mediators of the gemcitabine response that specifically act during that process are of interest for two reasons: First, if the influence of a knockdown on γH2AX is emphasized during S-phase, it is to assume that the knocked down protein is part of a DDR
specific for gemcitabine. Second, if the DDR was activated mainly during S-phase, this would target rapidly proliferating cells as found in cancer.
We thus aimed to identify those cells in our screen that have entered S-phase at the time of gemcitabine treatment. Replicating cells incorporate the nucleoside analog ethinyldeoxyuridine (EdU) when incubated with it. EdU can be coupled to a fluorescent dye by a chemical reaction and quantified using immunofluorescence. This creates additional readouts to the phosphorylation of H2AX in the screen: Overall EdU incorporation is a measure for the influence of a knockdown on DNA replication and was used as minor readout. EdU incorporation can further be used as gate to identify cells which actively replicate their DNA. In addition to the overall accumulation of γH2AX, we gated cells for EdU incorporation and quantified the accumulation of γH2AX in EdU-positive cells, too.
Our data show that incorporation of EdU into the DNA is almost completely blocked after gemcitabine treatment, probably due to a complete stalling of DNA replication caused by gemcitabine. The labeling of replicating DNA therefore needs to take place before adding gemcitabine to the cells. Optimization experiments revealed that the EdU labeling is useful as a readout if it takes place 2 h prior to gemcitabine treatment (data not shown). At that point replicating cells have incorporated EdU to a measurable extend and are clearly distinguishable from non-replicating cells. Longer EdU labeling times would lead to more
Figure 4.7 Central indicators of the DNA damage response and apoptosis are activated in a time-dependent manner.
(a) MIA PaCa-2 and (b) PANC-1 cells were treated with 300 nM gemcitabine for the indicated period of time and cell lysates were analyzed by immunoblotting.
EdU-positive cells, but would exceed the objective of the procedure which is to identify replicating cells at the time of gemcitabine administration.
4.2.3 Variability of the readout γH2AX after transfection with control siRNA
As mentioned above, gemcitabine-induced γH2AX should vary upon depletion of DNA damage transmitters. The influence on γH2AX accumulation was investigated for the following control knockdowns:
Ring finger protein 8 (RNF8) is a known mediator of the DDR. It mediates the ubiquitination at Lys63 of the histone variants H2A and H2AX, which leads to the amplification of DNA damage signaling on the one hand and facilitates the recruitment of DNA repair factors on the other hand (Mailand et al. 2007). We previously ascribed a role for MK2 in the DDR, and that its depletion leads to reduced H2AX phosphorylation. ATR and Chk1 are well-characterized mediators of the response to replicative stress.
The knockdown of RNF8 reduced γH2AX levels in gemcitabine-treated MIA PaCa-2 and PANC-1 cells, and knockdown of Chk1 leads to an extensive induction of γH2AX in both treated and untreated cells. MK2 knockdown did not reliably alter γH2AX levels in MIA PaCa-2 cells, but reduced γH2AX in gemcitabine-treated PANC1 cells. Interestingly, depletion of ATR did not induce H2AX phosphorylation in MIA PaCa-2 and PANC-1 cells,
Figure 4.8 γH2AX level in MIA PaCa-2 and PANC-1 cells are modulated by positive and negative controls within the library.
Cells were transfected with siRNA against the indicated targets and treated under screening conditions 48 h after transfection. Cells were fixed and stained for immunofluorescence and γH2AX levels were quantified and corrected for background fluorescence.
although its knockdown is a well-established positive control in DNA damage screens using U2OS cells in our lab (Fig. 4.8).
4.2.4 Conclusions
Overviewing all results of the optimization experiments, we chose PANC-1 to be most suitable for high-content screening. BxPC-3 cells are not accessible for siRNA-mediated knockdown, PaTu8902 cells are highly gemcitabine-resistant with hardly any variation in endogenous γH2AX level and MIA PaCa-2 cells tend to grow in multiple cell layers. Also, MIA PaCa-2 did not react to two of four control transfections in one plate.
The screening protocol was optimized to a time period of 2 h for the label of actively replicating cells followed by 22 h of 300 nM gemcitabine treatment in PANC-1 cells.