Vertical suppression of the EGFR pathway prevents onset of resistance in colorectal cancers

Abstract

Molecular targeted drugs are clinically effective anti-cancer therapies. However, tumours treated with single agents usually develop resistance. Here we use colorectal cancer (CRC) as a model to study how the acquisition of resistance to EGFR-targeted therapies can be restrained. Pathway-oriented genetic screens reveal that CRC cells escape from EGFR blockade by downstream activation of RAS-MEK signalling. Following treatment of CRC cells with anti-EGFR, anti-MEK or the combination of the two drugs, we find that EGFR blockade alone triggers acquired resistance in weeks, while combinatorial treatment does not induce resistance. In patient-derived xenografts, EGFR-MEK combination prevents the development of resistance. We employ mathematical modelling to provide a quantitative understanding of the dynamics of response and resistance to these single and combination therapies. Mechanistically, we find that the EGFR-MEK Combo blockade triggers Bcl-2 and Mcl-1 downregulation and initiates apoptosis. These results provide the rationale for clinical trials aimed at preventing rather than intercepting resistance.

Figure 1. The RAS-MAPK pathway drives resistance to EGFR blockade in CRC cells.

CCK81 (a) and DiFi (b) cells were infected with pathway specific cDNAs and treated with panitumumab. GI50 values are reported in logarithmic scale. Data are representative of three biological replicates and error bars represent s.d. Details on the constructs are listed in Supplementary Table 1.

Figure 2. Concomitant blockade of EGFR and MEK halts the emergence of resistance in CRC cell lines.

CCK81 (a), DiFi (b), C99 (c), NCIH508 (d) and HCA-46 (e). Cell lines were seeded at 20 million (CCK81 and C99) and 10 million (DiFi, NCIH508 and HCA-46) cell density and treated with cetuximab alone (340 nM), pimasertib alone (250 nM) and with the combination of the two drugs from day 0 or at the time of acquired resistance to cetuximab. Cells were detached and counted at least once a week as described in Methods section. A single biological replicate is represented for each cell model. Non-linear fit with exponential growth curve (Graphpad Prism) was applied to data points to show growth kinetics.

Figure 3. Vertical blockade of the EGFR pathway prevents the emergence of resistance in PDX.

A cetuximab sensitive CRC PDX (xenopatient) was expanded to create four cohorts of four mice each. (a) After randomization, mice were treated with vehicle (n=4) cetuximab (n=4), pimasertib (n=4) or the combination (n=4) for 6 weeks after which drug treatment was stopped and restarted (red arrows and black line). Treatment with vehicle served as control. Error bars represent s.e.m. (b) Individual mice from cetuximab or combo treated cohorts are represented as percentage of single tumours shrinkage on treatment at three time points after treatment start.

Figure 4. Mathematical modelling of cetuximab and combinatorial treatment in vivo.

(a) Fit of model (equation 1 described in the previous page) to average tumour volume of PDX1 cetuximab-treated replicate experiment. Fitted curve (blue) is the sum of sensitive (green) and resistant (red) populations, whose predicted behaviours are shown in inset. (b) Fit of model (equation 2 described in the previous page) to average tumour volume of PDX1 combo treated replicate experiment.

Figure 5. Immunohystochemical analyses of PDX derived samples.

Two levels of magnification are presented to better appreciate sample size. (a) Ki67 staining of one representative example of Vehicle, Pima and Cmab treated mice and the entire combo treated mice from which we were able to obtain tissue at the end of the experiment. Percentage of positivity: Vehicle 6%, Pima 24%, Cmab 8% and Combo 0%. (b) Hematoxylin/Eosin staining of one representative example of Vehicle, Pima and Cmab treated mice and the entire combo treated mice from which we were able to obtain tissue at the end of the experiment. Only in mouse #5 a small amount of cancer cells is detectable, the rest are composed only by necrotic tissue. Scale bars, 50 μm (a, right panels), 100 μm (a, lefts panel; and b, right panels) and 500 μm (b, left panels).

Figure 6. Mathematical modelling of cetuximab and combinatorial treatment in vitro.

(a), Fit of model (equation 1) to cell number data of DiFi cell line treated with cetuximab. (b) Fit of model (equation 2) to the cell number data of DiFi cetuximab-resistant population re-challenged with the combo.

Figure 7. Biochemical analysis and apoptosis evaluation of CRC cells treated with EGFR and/or MEK inhibitors.

CCK81 (a) and DiFi (b) were treated with cetuximab (Cmab, 340 nM), pimasertib (Pima, 250 nM), or with the combo of the two drugs at the indicated time points, whole-cell extracts were subjected to western blot analysis and compared with untreated cells with phospho-EGFR (Tyr 1068), total EGFR, total AKT and phospho-AKT (Ser 473), total ERK1/2 and phospho-ERK1/2 antibodies. Actin was included as a loading control. (c,d) Concomitant blockade of EGFR and MEK triggers apoptosis. CCK81 and DiFi cells were treated at different time points with cetuximab, pimasertib or both. Nuclei fragmentation (c) and caspase-3 activation (d) were measured. Rates of apoptosis due to combination are approximatively twofold higher compared with those due to cetuximab, in line with our model predictions. (e) Apoptotic pathway activation. The indicated CRC cell lines were treated with cetuximab (Cmab, 340 nM), pimasertib (Pima, 250 nM), or with the two drugs for 48 h. Whole-cell extracts were subjected to Western blot analysis and compared with untreated cells using BAK, Bax, Bid, NOXA, PUMA, Bim, Bcl-2, Mcl-1, Bcl-XL and active caspase-3 antibodies. Actin was included as a loading control.