Acquisition of chromosome instability is a mechanism to evade oncogene addiction

Abstract

Chromosome instability (CIN) has been associated with therapeutic resistance in many cancers. However, whether tumours become genomically unstable as an evolutionary mechanism to overcome the bottleneck exerted by therapy is not clear. Using a CIN model of Kras-driven breast cancer, we demonstrate that aneuploid tumours acquire genetic modifications that facilitate the development of resistance to targeted therapy faster than euploid tumours. We further show that the few initially chromosomally stable cancers that manage to persist during treatment do so concomitantly with the acquisition of CIN. Whole-genome sequencing analysis revealed that the most predominant genetic alteration in resistant tumours, originated from either euploid or aneuploid primary tumours, was an amplification on chromosome 6 containing the cMet oncogene. We further show that these tumours are dependent on cMet since its pharmacological inhibition leads to reduced growth and increased cell death. Our results highlight that irrespective of the initial CIN levels, cancer genomes are dynamic and the acquisition of a certain level of CIN, either induced or spontaneous, is a mechanism to circumvent oncogene addiction.

Keywords: Chromosome instability; breast cancer; cMet; mouse models; resistance.

Figure 1. Non‐regressed K and KM tumours show high levels of chromosomal instability

1. A

   Representative micrographs of a time‐lapse microscopy of K and KM non‐regressed tumour cells (H2B‐GFP green). Top: mitotic cell with a chromatin bridge (yellow arrow). Bottom: mitotic cell with misalignment and cytokinesis failure resulting in binucleation (yellow arrow).

2. B

   Percentage of mitotic errors in K and KM primary tumours and in K and KM non‐regressed tumours. Ns: not significant; *P < 0.01, ****P < 0.0001; one‐way ANOVA, Turkey's multiple comparison test. Mean ± SEM. Exact P values are indicated in Appendix Table S1.

3. C

   Percentage of cells in K and KM primary tumours and in K and KM non‐regressed tumours with the indicated mitotic errors. Scale bar 20 μm.

Data information: K primary (n = 4; 74 cells), KM primary (n = 5; 84 cells), K non‐Regr (n = 7; 219 cells), KM non‐Regr (n = 5; 247 cells).Source data are available online for this figure.

Figure 2. Recurrent somatic copy number alterations in K and KM non‐regressed tumours

1. A

   Somatic copy number alterations in each individual Kras and Kras/Mad2 non‐regressed tumours. 7 K non‐regressed and 16 KM non‐regressed tumours were sequenced. Focal deletion (DEL, green), focal amplification (AMP, light green), whole chromosome gain (WCG, dark grey) and loss (WCL, light grey), partial chromosome gain (PCG, dark blue) and loss (PCL, light blue) and gross chromosomal rearrangement (GCR, red).

2. B

   Overlaying of the amplifications in chromosome 6 across multiple regions showing that all amplicons contained the cMet oncogene. Colour of the block corresponds to the segment mean (the degree to which a genome segment is lost (blue) or gained (red).

3. C

   Immunostaining of phospho‐cMet in non‐regressed tumours carrying or not the amplification in chromosome 6. Yellow numbers indicate the total number of cMet‐positive cells/total number of cells counted. Scale bar 100 μm. Numbers of the K and KM non‐regressed tumours are described in Table 1.

Source data are available online for this figure.

Figure EV1. Related to Fig 2. Somatic copy number alterations in K and KM primary and non‐regressed tumours

1. A

   Average of somatic copy number alteration (SCNAs) in K and KM non‐regressed tumours. Unpaired t‐test, P = 0.34. Boxes and whiskers represent min to max values. The central line represents the median.

2. B

   Percentage of SCNAs per chromosome in K (green) and KM (blue) non‐regressed tumours.

3. C

   Heatmap of all chromosomes in the non‐regressed tumours.

4. D

   Frequency of somatic copy number alterations per chromosome in Kras and Kras/Mad2 primary tumours. Focal amplification (AMP, dark green), whole chromosome gain (WCG, dark grey) and loss (WCL, light grey), partial chromosome gain (PCG, dark blue) and loss (PCL, light blue) and gross chromosomal rearrangement (GCR, red).

Source data are available online for this figure.

Figure EV2. Related to Fig 2. cMet analysis in non‐regressed tumours

1. A

   Genome‐wide log₂ of chromosome 6 of control mammary gland and log₂‐ratio plots of K and KM non‐regressed tumour biopsies showing no amplification in the control (upper panels) and a small amplification in the non‐regressed tumours (middle and bottom panels).

2. B

   Quantitative RT–PCR analysis of cMet in 10 K and 22 KM non‐regressed tumours.

3. C

   Correlation analysis of cMet mRNA and gene amplification in K (green) and KM (blue) non‐regressed breast tumours.

Source data are available online for this figure.

Figure 3. cMet amplification is not found in primary tumours

1. A

   Tumour diameter before and after doxycycline withdrawal in 4 K and 5 KM breast tumours with cMet amplification. 0 indicates when doxycycline was removed. Each colour represents one tumour. Blue and green squares indicate the timeframe between doxycycline withdrawal and the moment in which tumours resumed growth.

2. B

   Genome‐wide log₂‐ratio plots of chromosome 6 of two primary tumour biopsies and their corresponding non‐regressed tumour showing no amplification in the primary tumour (upper panels) and a small amplification in the non‐regressed tumours (bottom panels, yellow arrow).

3. C

   Immunostaining of phospho‐cMet in 3 biopsied primary tumours (PT) and their corresponding non‐regressed tumours (KM5, KM7 and KM8, which are also shown in Fig 2C). Scale bar 100 μm.

4. D

   Representative two‐dimensional scatter plots constructed with overlaid dPCR data of the reference (VIC) and cMet (FAM) from one tumour without cMet amplification and one with cMet amplified. Dots represent results of independent PCRs in the wells of a digital PCR chip. Reactions in the bottom left corner (yellow) are negative for both targets, while the ones in the top right corner (green) are double positives. Reactions in the top left (blue) and bottom right (red) corners are positive for cMet and the reference targets, respectively.

5. E

   Representative photographs of FISH staining with a probe for Met (red signal) and a probe for a reference gene EML4 (green signals) The upper panel is a negative example for cMet amplification containing 2 red and 2 green dots (white arrows). The lower panel shows an example of a tumour with cMet amplification (several red dots) and 2 green dots (yellow arrow). Scale bar 10 μm.

Source data are available online for this figure.

Figure EV3. Related to Fig 3. cMet staining in primary tumours

Immunostaining of phospho‐cMet in one non‐regressed tumour (KM15) as a positive control (as shown in Fig 2C) and 34 primary tumours (PT). Yellow numbers indicate the total number of cMet‐positive cells/total number of cells counted. Scale bar 100 μm.Source data are available online for this figure.

Figure 4. cMet amplification is not clonally dominant in primary tumours

1. A

   Schematic of the experiment. K and KM animals were set on doxycycline food until mammary tumours developed. Primary tumours were collected and single cells injected into Rag2^−/− animals to recapitulate the tumours followed by either no treatment, drug treatment or switched to a normal diet.

2. B

   Relative volume of tumours grown in 12 Rag2^−/− animals after treatment with Tepotinib or vehicle control (3K and 3KM tumours for each condition). No statistical significance was found by one‐way ANOVA. Mean ± SEM. Exact P values are indicated in Appendix Table S2.

3. C

   Quantification of cMet copy number detected by digital PCR or FISH in primary tumours injected into Rag2^−/− and treated with vehicle control or Tepotinib.

4. D

   Quantification of cMet copy number detected by digital PCR or FISH in primary tumours injected into Rag2^−/− and fed with doxycycline or after doxycycline withdrawal.

Source data are available online for this figure.

Figure 5. cMet amplified non‐regressed tumours respond to cMet treatment

1. A

   Schematic of the experiment. K and KM animals were set on doxycycline food until mammary tumours developed. Then, switched to a normal diet and non‐regressed tumours collected. Single cells were then injected into Rag2^−/− animals to recapitulate the non‐regressed tumours followed by drug treatment or vehicle control treatment.

2. B

   Relative volume of tumours grown in Rag2^−/− animals after treatment with Tepotinib or vehicle control. One‐way ANOVA, Tukey's multiple comparison. Ns: not significant; **P < 0.024. Mean ± SEM. A total of 19 Rag2^−/− mice injected with cMet expressing tumours were treated with Tepotinib and 14 used as controls. For tumours with no cMet, 13 Rag2^−/− mice were treated with Tepotinib and 9 were used as control. Exact P values are indicated in Appendix Table S3.

3. C

   Immunostaining of phospho‐cMet in the same tumours grown in Rag2^−/− mice after treatment with Tepotinib or vehicle control. Scale bar 200 μm.

4. D

   Quantification of pH3 in tumours grown in Rag2^−/− animals after 5 days treatment with the cMet inhibitor Tepotinib or vehicle control.

5. E

   Quantification of caspase 3 in tumours grown in Rag2^−/− animals after 5 days treatment with the cMet inhibitor Tepotinib or vehicle control.

Data information: (D, E) One‐way ANOVA, Tukey's multiple comparison; **P < 0.005; ****P < 0.0001. Mean ± SEM. At least four tumours per condition were analysed, and a minimum of 35 fields of view (FOV) or 12,000 cells were counted for each condition. Exact P values are indicated in Appendix Tables S3 and S4, respectively.Source data are available online for this figure.

Figure EV4. Related to Fig 5. Proliferation and apoptosis in non‐regressed tumours

Immunostaining of pH3 and caspase 3 in tumours with and without cMet amplification after treatment with Tepotinib or vehicle control. Scale bar 200 μm.Source data are available online for this figure.