1. Profiles of Contributing genes to sensitive/resistant PhenotyPe of 11 Different
onCology theraPeutiC agents aCross 240 Cell lines.
O’Day, C., Ovechkina, Y., Marcoe, K.F., Keyser, R., Yoshino, K., Nguyen, P., Hnilo, J., Shively, R., Mulligan, J., Bernards, K., Chesnut-Speelman, J., Lin, T., and Wang, S.
Ricerca Biosciences LLC – Bothell, WA, USA
PurPose Figure 2. CVs in the nuclear channel were low. The CV of control wells was averaged over 3 independent
experiments. Overall, 91% of cell lines had a CV of less than 20% and only 11 non-adherent cells had
Figure 4: Sensitive and resistant cell lines to Erlotinib with genetic biomarkers. Figure 7: Sensitive and resistant cell lines to VX-680 with genetic biomarkers.
a CV greater than 25%. The log of the difference from the average IC50 value was plotted against the 240 cell lines. Kras mutations correlated The log of the difference from the average EC50 is plotted against the various cell lines. CTTNB1 mutations predominate
A number of targeted therapies have been shown to be effective in the treatment of cancer, such as Imatinib for
with resistance. Genetic analysis of mRNA data was not complete, but the few known cell lines that overexpress EGFR in the most sensitive cells lines and tended to be of colon/GI origin. Those cell lines with CTTNB1 mutations that were
treating chronic myelogenous leukemia, Erlotinib for non-small-cell lung cancers and Sorefinib for metastatic lung
All cells were grouped according to morphology and the CVs of the control wells were averaged and binned according were identified as sensitive. Cell lines with EGFR mutations that were not sensitive also had mutations in the Ras/Raf resistant were not of colon origin. APC mutations tended to be intermediate/resistant with no APC mutations in the 50
cancer. The sporadic sensitivity to these therapeutic agents has launched an investigation of correlation between
to the range of CVs: a) adherent cells; b) semi-adherent cells; c) non-adherent cells. pathway. most sensitive cell lines.
cancer phenotype and genotype. We have developed an in vitro cellular assay to evaluate the relationship between
tumor genotypes (Affymetrix SNP and gene expression chips and Sanger mutation data) and cancer cell sensitivity a) b) c)
for over 240 human tumor cell lines. A panel of targeted therapeutics was used to show the usefulness of this in vitro
approach for developing anticancer drugs. Data for nine of the eleven evaluated therapeutics is shown below. Data
was not shown for Staurosporine, Paclitaxel or Doxorubicin in the interest of space. Stomach
Lung Liver Endocrine
and colon
MethoDs
Growth and assay conditions were established for all 240 cell lines. Compounds were added in half-log dilutions for 10
concentrations using tipless acoustic transfer with an Echo 550. An additional “time zero” (T0) plate also was seeded
at the same density and analyzed for cell number on day one to determine the number of doublings. Seventy-two Table 2: Panel of mutations generated from Sanger Database
hours after compound addition, the cells were fixed and stained with antibodies for activated caspase-3 and phospho-
histone H3. Nuclei were stained with DAPI. Cells were imaged with a 4X objective on an IN Cell Image Analyzer and All cell lines were analyzed for:
analyzed with the Developer software tool. Data was plotted with in-house Math IQ graphing software using nonlinear 1. Mutation data (Available on most cell lines. Twenty two % of the cell lines did
regression analysis. Data was analyzed for cell count (% of control), fold induction of apoptosis (% of control) and fold not have mutation data.)
induction or decrease in G2 (% of control). All data was normalized to control wells. Reference compound data was 2. SNP analysis (Affymetrix SNP 500K array)
analyzed and pooled. Cell lines were binned to sensitive and resistant lines based on acceptable in vivo dosage levels
3. Gene expression data (Affymetrix U133 plus 2.0 array) Figure 5: Sensitive and resistant cell lines to CL-1040 with genetic biomarkers.
or a marked delineation of sensitivity. Sensitive and resistant cell lines were then correlated to mutation spectrums to
determine genes underlying the corresponding phenotype (Fig.1). Mutation data was used to analyze these cell lines.
Table 3: Colon/GI cancers with CTNNB1 mutations
are sensitive to Aurora inhibition.
ConClusions
Genetic data was generated in house, through the CABIG site and the Sanger site. All data
Analysis of expression and SNP data is currently underway. was normalized through RMA normalization Gene copy number and mRNA expression data. The log of the difference from the average EC50 value was plotted against the various cell lines. Ras/Raf mutations CTNNB1 mutations made up 4 of the 5 most sensitive cell
predominated in the sensitive cells and RB mutations conferred resistance. The one Braf mutation that was not sensitive, Summary of Erlotinib
https://cabig.nci.nih.gov/caArray_GSKdata/ lines in the 32 colon/GI cancers shown below. B-catenin
Figure 1: Assay Workflow harbored a G464V mutation rather than the activating V600E mutation. • Kras mutations confer resistance to treatment
50+ Gene Mutation data mRNA levels are being examined for over-expression.
• 3 EGFR mutations were present in the 240 cell lines. One was sensitive
http://www.sanger.ac.uk/genetics/CGP/Celllines APC mutations predominate in the resistant cells but and the two others were intermediate/resistant because of Ras/Raf
“The mutation data was obtained from the Sanger Institute Catalogue Of Somatic Mutations In
G464V
APC mutations are thought to be mutually exclusive mutations
Cancer web site, http://www.sanger.ac.uk/cosmic Bamford et al (2004) The COSMIC (Catalogue of CTNNB1 mutations. Other genes found through
of Somatic Mutations in Cancer) database and website. Br J Cancer, 91,355-358.”
expression analysis might give a more complete picture Summary of CL-1040
of this sensitivity. Braf mutations clearly predominate in the sensitive cell lines
Figure 3: Distribution plot of sensitive and resistant cell lines. • 15 of the 30 most sensitive cell lines had Braf mutations
• None of the 30 most resistant cell lines had Braf mutations
• One somewhat resistant Braf mutation was a G464V rather than the
EC50 values were plotted against IC50 values for many of the 240 cell lines. For some agents that generated typical V600E mutation
incomplete growth inhibition (GI), GI50 values or max % growth inhibition was plotted against the EC50 values (Fig. Ras mutations had a positive correlation with sensitive cells but weaker
3a. and i). Sensitive cell lines were selected by a clear demarcation from the others such as in figures 3b, c, e, f, and Sensitive than Braf
g. For Geldanamycin (Fig. 3d) all cell lines responded over a small range but a few were resistant. For Dasatinib • All of the 30 most sensitive cell lines contained Braf or Ras mutations
(Fig. 3c) CML lines were most sensitive. However, a subset of the cell lines did not demonstrate good growth • 1 out of the 30 most resistant cell lines had a Ras mutation
Rb mutations appear exclusively in the resistant cell lines
inhibition when results were adjusted for the number of cells plated. Similarly, with Everolimus (Fig 3a) two groups
• 8 of the 30 most resistant cell lines had a Rb mutation
of sensitive and resistant cells were apparent but of the sensitive lines, some showed poor growth inhibition when
• 0 of the 30 most sensitive cell lines had a Rb mutation
results were adjusted for the number of cells plated. CL-1040 did not show a clear demarcation between sensitive
• Rb protein intersects the Ras/Raf pathway at CDK-cyclin control of S
and resistant and thus cutoff was made based on in vivo dosage levels. phase transcription
Summary of VX-680
Figure 6: RB mutations and Ras/Raf activation are competing processes. CTNNB1 mutations predominate in the sensitive cell lines
Intermediate
• 4 out of the 33 colon/GI cell lines have a CTTNB1 mutation and these
Figure 1: Cells are plated in 384 wells and treated with inhibitor Cells harboring a Ras/Raf activating mutation, activate CDK cyclin and ERK, which phosphorylate RB. Phosphorylated mutations are 4 of the 5 most sensitive cell lines to VX-680
(staurosporine) for 72 hours. Cells were fixed and stained with RB dissociates from the transcription complex and E2F is able to transcribe RNA and translate proteins needed for • 5 of the 8 CTNNB1 mutations occur in the sensitive cell lines determined
anti-activated caspase 3 (green), anti-phospho-histone H3 (red) to have a GI50 of 0.005 to 0.025; One CTNNB1 mutated cell line is in
entry to S phase. However, if RB is mutated, E2F is constitutively activated and MEK inhibition is not able to inhibit the resistant category
and DAPI for cell number (blue). All data is normalized to vehicle transcription. Hence, Rb mutations confer resistance to CL-1040. • 4 of the 4 colon cancers with CTTNB1 mutations are sensitive
control wells and reported as % of control (nuclear count) or fold
APC mutations do not exist in the sensitive lines but predominate in the
induction (apoptosis and cell cycle). Data is binned into sensitive intermediate/resistant cell lines
and resistant cell lines and analyzed for genetic correlations. • 0 of the 30 most sensitive cell lines have APC mutations.
• 2 of the 30 most resistant cell lines have APC mutations. APC
mutations may be in the resistant/intermediate cell lines because they
are exclusive of CTNNB1 mutation
Table 1. Multiplexed cytotoxicity assay parameters are robust. Intra-assay variability in EC50 of Staurosporine
on HCT-116 over 10 independent experiments. Data will be further validated by correlating genes from
Resistant expression analysis and gene copy number. Analysis
Cell line HCT-116 was propagated and plated in 10 different experiments. Reference compound controls were added of this data is currently in progress.
in accord with our standard assay. Data was collected, analyzed, and averaged with standard deviations. Results are
reported above.