Bernhard Kuster (ed.), Kinase Inhibitors: Methods and Protocols, Methods in Molecular Biology, vol. 795,
DOI 10.1007/978-1-61779-337-0_16, © Springer Science+Business Media, LLC 2012
Investigation of Acquired Resistance to EGFR-Targeted
Therapies in Lung Cancer Using cDNA Microarrays
Kian Kani, Rafaella Sordella, and Parag Mallick
Clinical tools to accurately describe, evaluate, and predict an individual’s response to cancer therapy are a
field-wide priority; in many advanced cancers, only 10–20% of individuals will have a clinical benefit from
therapy, yet we treat the entire population. Furthermore, many therapies are initially effective, but lose
effectiveness over time. Here we describe methods to derive in vitro models of resistance to EGFR tyrosine
kinase inhibitors. We additionally describe approaches to characterize possible mechanisms of resistance by
genomic and transcriptomic approaches.
Key words: Gefitinib, Erlotinib, Non-small-cell lung cancer, Therapeutic response, Proteomics,
Mechanistically, changes in cancer cell growth patterns and rates
may originate with alterations in growth signaling networks. The
concept of “oncogene addiction” has been demonstrated in a variety
of mouse tumorigenesis models in which continued expression of
the transforming oncogene is required for tumor maintenance.
Numerous targeted therapeutics attempt to halt the growth of a
tumor via specific disruptions in these networks. For example, the
efficacy of drugs such as imatinib illustrate the importance of BCR-
ABL in chronic myelogenous leukemia and the dramatic responses
of epidermal growth factor receptor (EGFR)-mutant lung tumors
to EGFR tyrosine kinase inhibitors (TKIs), such as erlotinib and
gefitinib, suggest that even complex epithelial cancers may exhibit
dependency on a single activated kinase for survival. Specifically
234 K. Kani et al.
in the case of lung cancer, much excitement has recently been
generated by the finding that a group of non-small-cell lung cancer
(NSCLC) patients benefit highly from treatment with selective
EGFR inhibitors (1). The clinical success of tyrosine kinase inhibi-
tors (TKIs) has dramatically influenced research and medical
oncology over the past decade. Unfortunately, these therapies are
often only effective in a small percentage of patients. Furthermore,
even when initially effective, therapies can lose effectiveness over
time, presumably through the tumor’s molecular evolution of
resistance to the therapeutic. Despite extensive study, de novo and
acquired resistance to targeted therapeutic agents remains a major
obstacle to improving remission rates and achieving prolonged
Through the use of derived models of resistance numerous
mechanisms of response and resistance have been uncovered.
NSCLC patients with activation mutations in the EGFR gene
exhibit dramatic sensitivity to targeted TKIs (2). These mutations
remove some or all of the negative allosteric regulatory mecha-
nisms in the EGFR kinase domain (3). The most frequent activating
mutations are the in-frame deletions of leucine-747 to glutamic
acid-749 (DLRE) and the leucine to arginine substitution at codon
858 (L858R), which account for 44% and 41% of all the mutations
in EGFR cases in NSLC, respectively (4, 5). Subsequent studies
also identified a mutation of threonine to methionine at 790 that
confers resistance. In addition to mutations in the EGFR, amplifi-
cation of c-MET has recently been identified as another potential
acquired resistance mechanism. In a preliminary study, 22% (4/18)
of cases of NSCLC with acquired resistance to gefitnib/erlotinib
have been shown to harbor c-MET amplification (6). Interestingly,
some of these tumors contained both the T790M mutation and
c-MET amplification. Although the role of c-MET amplification
in EGFR TKI-acquired resistance has yet to be definitively proven,
current studies suggest that the presence of T790M and c-MET
amplification could account for approximately 50% of all cases (6).
Thus, despite considerable progress in identifying acquired resis-
tance mechanisms, new determinants of EGFR TKI have yet to be
uncovered in the majority of tumors that have relapsed.
The accumulation of genetic/epigenetic aberrations that
ultimately lead to the acquisition of resistance to a particular TKI
treatment in fact usually results in dramatic changes in the expres-
sion pattern of multiple genes. In this regard, a strong correlation
between gene expression and copy number variations has been
observed. As such, gene expression profiling or microarray analysis,
by enabling the measurement of thousands of genes in a single
RNA sample, provides a powerful research tool for elucidating
genetic interactions and response to drug treatments.
Simplistically, while DNA sequence analysis instructs us on
what a cell could possibly do as a consequence of the accumulations
23516 Investigation of Acquired Resistance to EGFR-Targeted Therapies…
of genetic aberrations, the expression profile data conveys us
what it is actually happening in a given cell in a particular cell state.
The basic idea behind all the different variety of microarray
platforms that have been developed is simple: a glass slide or
membrane is spotted or “arrayed” with DNA fragments or oligo-
nucleotides that represent specific gene coding regions. Purified
RNA is then fluorescently or radioactively labeled and hybridized
to the slide/membrane. After thorough washing, the raw data is
obtained by laser scanning or autoradiographic imaging. At this
point, the data may then be entered into a database and analyzed
by a number of statistical methods. The up-front design of a study
is critical to a research study’s ability to draw sound conclusions in
the particular case of identifying signaling pathways or genes
whose disregulation could contribute to the acquisition of resistance
to a given drug treatment. Here, we introduce several techniques
for deriving models of resistance. We also describe genomic and
transcriptomic techniques for characterizing the mechanism of
resistance of those models.
1. NSCLC cell lines: e.g., HCC827 cells.
2. Type II classification biological hood and safety cabinet.
3. Cell growth media compatible for cells under investigation:
e.g., Dulbecco’s Modified Eagle’s Medium (DMEM) contain-
ing 1% of dialyzed fetal bovine serum.
4. Gefitinib: 99% purità, prepare 10 mM stock solution in DMSO.
5. Cell counter, microscope.
6. 96-well plates: flat bottom for cell culture.
7. Cell viability assay kit: e.g., the MTS or MTT assays are com-
mercially available from many sources.
8. Ethyl methane sulfonate (EMS): 600 ?g/ml in cell culture
DNA can be prepared for sequencing by use of several commer-
cially available kits.
1. DNeasy Blood and Tissue Kit.
2. Pipettes and pipette tips.
4. Microcentrifuge tubes (1.5 ml or 2 ml).
5. Microcentrifuge with rotor for 1.5 ml and 2 ml tubes.
6. Ethanol: 96–100% purity.
2.1. In Vitro Selection
2.2. DNA Sequencing
236 K. Kani et al.
7. Phosphate-based saline (PBS).
8. Lysis buffer: Autoclaved PBS supplemented with 1 mM EDTA,
0.5% NP40, 25 mM DTT, 5 mM MgCl2, and 1× RNase
1. Trizol reagent.
3. Ethanol: 200 proof quality.
4. Dulbecco’s phosphate-buffered saline (DPBS).
5. Sodium acetate: 3 M, pH 5.2.
6. RNAse-free water: 1 ml of 0.1% (v/v) diethylpyrocarbonate
(DEPC) in 999 ml of water. Let stand overnight. Autoclave
prior to use.
7. Cell scraper.
8. Centrifuge tubes: 15 ml, round bottom, polypropylene.
9. Centrifuge tubes: 50 ml, conical, polypropylene.
10. Plastic tubes: 1.5 ml.
11. PCR tubes: 0.2 ml, thin wall.
12. Centrifugal filter: MicroCon 100.
13. High-speed centrifuge.
1. PCR tubes: 0.2 ml, thin wall (or plate).
2. Plastic tubes: 1.5 ml, RNAse free.
3. Primers for genes of interest.
4. DNA polymerase: JumpStart Taq kit.
5. Reverse transcription: ImProm-II Rt system.
1. Ethanol: 200 proof quality.
2. Dulbecco’s phosphate-buffered saline (DPBS).
3. Sodium acetate: 3 M, pH 5.2.
4. Nucleotides: dATP, dCTP, dGTP, dTTP, 100 mM each, store
frozen at −20°C.
5. Primers: 1 mg/ml pd(T)12–18 in water, store frozen at −20°C.
6. Oligo primer: anchored, 5?-TTT TTT TTT TTT TTT TTT
TTV N-3?, resuspend at 2 mg/ml, store frozen at −20°C.
7. Fluorescent dyes: 1 mM CyTM3-dUTP and 1 mM CyTM5-
dUTP, store at −20°C, protect from light.
8. Rnase inhibitor: RNasina, store at −20°C.
9. SUPERSCRIPTTM II Rnase H` Reverse Transcriptase Kit:
store at −20°C.
10. Human C0t-1 DNA: 10 mg/ml, store frozen at −20°C.
2.3. RNA Isolation
2.4. Quantitative PCR
23716 Investigation of Acquired Resistance to EGFR-Targeted Therapies…
11. Ethylenediaminetetraacetic acid (EDTA): 0.5 M, pH 8.0.
12. NaOH: 1 N.
13. Tris(hydroxymethyl)aminomethane (Tris)–HCL: 1 M, pH
14. TE buffer (1×): 10 mM Tris, 1 mM EDTA pH 7.4.
15. Diethylpyrocarbonate (DEPC)-treated water: 0.1% v/v dieth-
ylpyrocarbonate in water.
16. Tris-Acetate electrophoresis buffer (TAE, 50×): 2 M Tris ace-
tate pH 8.0, 55 mM EDTA in DEPC-treated water.
17. Centrifuge tubes: 15 ml, round bottom, polypropylene.
18. Centrifuge tubes: 50 ml, conical bottom, polypropylene.
19. Plastic tubes: 1.5 ml.
20. PCR tubes: 0.2 ml, thin wall.
21. Centrifugal filter: MicroCon 100.
22. High-speed centrifuge for 15-ml tubes.
23. Nucleotides: 100 mM dGTP, dATP, dCTP, and dTTP.
24. First-Strand Buffer: 5×, provided with Superscript II kit.
25. Poly d(A): 8 mg/ml in water.
26. Yeast tRNA: 4 mg/ml in water.
27. SSC buffer (20×): 3 M sodium chloride, 300 mM sodium
citrate, pH 7.0.
28. Denhardt’s blocking solution (50×): 1% Ficoll type 400, 1%
polyvinylpyrrolidone, and 1% bovine serum albumin.
29. For our studies, we used the Affimetrix platform including the
Affimetrix U95Av2 array that contains approximately 12,600
human genes and the Affymetrix GeneChip Scanner 3000.
Other alternative systems may also be used.
1. Maintain cells for several passages to ensure normal growth
(see Note 1).
2. Plate the cells at 3,500 cells per well in quadruplicate in 96-well
plates (flat bottom, cell culture plated).
3. 18 h after cell plating, wash the cells with fresh medium and
then add medium supplemented with different concentrations
(e.g., threefold dilution series) of gefitinib (see Note 2).
Incubate the cells for 2–3 days (see Note 3).
of Gefitinib Resistant
in Cell Culture
3.1.1. Titration Method
238 K. Kani et al.
4. Measure cell viability, for example, using the MTS assay with
UV detection at 490 nM in a plate reader. Determine the drug
concentration at which 50% of all cells are viable (IC50 value).
This drug concentration is used as a starting point for subse-
quent cell dosing.
5. Start dosing the cells at one order of magnitude below the
determined IC50 value (e.g., 100 nM).
6. Plate the cells in a 10-cm cell-culture plate.
7. Incubate the cells with 10 nM gefitinib (see Note 4) and
replenish drug containing media every 24 h.
8. Monitor cell viability and increase gefitinib concentration as
the cells regain normal growth kinetics (see Note 5).
9. Expand the cells to desired quantities for subsequent experiments
(see Notes 6 and 7).
In addition to the above described titration method, a rapid selec-
tion protocol with or without exposure to mutagen can be employed.
A mutagen such as EMS can be used to increase genetic aberra-
tions. In this case, cells are treated prior to selection with EMS at a
concentration of 600 ?g/ml and allowed to recover for 72 h.
1. Maintain the cells for several passages to ensure normal
2. Measure IC50 by performing a MTS viability assay with different
concentrations of gefitinib as described in Subheading 3.1.1.
3. Plate the cells in a 10 cm cell culture plate at a 30% confluency
(i.e., 6 × 104 cells per 10 cm2).
4. Incubate the cells with three times the IC50 in the presence of
5% serum (selection media). In the case of the HCC827 cells,
this corresponds to 1 ?M gefitinib.
5. Replenish every 24 h with new selection media (RPMI, 5%
FBS, Strep/pen Invitrogen).
6. Expand the cells to the desired quantities for subsequent
1. Grow cells to 65% confluence in 10-cm plates and wash three
times with PBS.
2. Scrape the cells off the plate with 400 ?l of lysis buffer.
3. Add 40 ?l proteinase K. Check pH. If required, adjust pH
using hydrochloric acid or glacial acetic acid.
4. Add 400 ?l Buffer AL provided in the DNeasy kit (without
added ethanol) to the sample and mix thoroughly by vortexing.
5. Add 400 ?l ethanol (96–100%) and mix again thoroughly by
vortexing. It is important that the sample and the ethanol are
mixed thoroughly to yield a homogeneous solution.
3.1.2. Rapid Selection
of Resistant Cell
Populations by DNA
23916 Investigation of Acquired Resistance to EGFR-Targeted Therapies…
6. Pipette the mixture from step 5 (including any precipitate)
into the DNeasy Mini spin column placed in a 2-ml collection
tube (provided). Centrifuge at ?6,000 × g for 1 min. Discard
the flow-through and collection tube.
7. Place the DNeasy Mini spin column in a new 2-ml collection
tube (provided), add 500 ?l Buffer AW1 (provided in the
DNeasy kit), and centrifuge for 1 min at ?6,000 × g. Discard
the flow-through and collection tube.
8. Place the DNeasy Mini spin column in a new 2-ml collection
tube (provided), add 500 ?l Buffer AW2 (provided in the
DNeasy kit), and centrifuge for 3 min at 20,000 × g to dry the
9. Place the DNeasy Mini spin column in a clean 1.5 ml or 2 ml
microcentrifuge tube (not provided) and pipette 400 ?l Buffer
AE (provided in the DNeasy kit) directly onto the DNeasy
membrane. Incubate at room temperature for 1 min and then
centrifuge for 1 min at ?6,000 × g to elute the sample.
10. Measure the DNA quality and purity by use of OD 260/280
and send out for commercial sequencing (see Note 8).
Here, we provide a reliable protocol to assess the expression of
most commonly used epithelial to mesenchymal transition (EMT)
markers by using QPCR-based techniques (see Note 9).
Many protocols and commercially available kits are currently
available for the generation of RNA. We routinely employ the
following method based on a combination of phase extraction
1. Wash subconfluent cells (i.e., approximately 5 × 106 cells) twice
with cold PBS on ice. Scrape the cells with 1 ml of cold PBS
and transfer them into sterile RNase-free 1.5-ml microcentri-
fuge tubes. Pellet by centrifugation at 13,000 × g for 10 s and
remove the supernatant by gently aspiration.
2. Lyse cells by repetitive pipetting in 1 ml of Trizol.
3. Add two-tenth volume of chloroform (i.e., 200 ?l) and shake
vigorously for 15 s.
4. Let mixture stand for 3 min at room temperature. Centrifuge
at 13,000 × g for 15 min at 4°C.
5. Take off the aqueous phase (i.e., upper phase) and transfer it to
a polypropylene tube.
of Markers of
3.3.1. RNA Isolation
240 K. Kani et al.
6. Add 0.53 volumes of ethanol (i.e., 500 ?l) slowly while vortexing
(see Note 10).
7. Let mixture stand for 10 min at room temperature and centri-
fuge at 13,000 × g for 10 min at 4°C.
8. The RNA forms a pellet on the side or bottom of the tube.
Discard the supernatant.
9. Wash the pellet two times by adding 1 ml of 75% ethanol and
centrifuging at 10,000 × g for 5 min at 4°C.
10. Dry the pellet for 5–10 min and resuspend RNA in 80 ?l of
DEPC-treated water at 65°C.
11. Resuspend RNA at approximately 1 mg/ml in DEPC-treated
12. Concentrate the sample by centrifugation (500 × g) in a
MicroCon 100 filter unit (see Note 11).
13. Determine the concentration of RNA by spectrophotometry.
Store at −80°C (see Note 12). To measure the RNA concentra-
tion, take 2–5 ?l RNA sample and dilute with RNase-free water
to a final volume of 1 ml in a 1.5-ml microcentrifuge tube (i.e.,
200–500 dilution of the RNA sample). Transfer 1 ml of RNase-
free water to a clean cuvette and read absorbance as blank.
Pipette the diluted RNA sample in to a clean cuvette and read
absorbance at 260 nm and 280 nm. To determine RNA con-
centration of the original sample, use the following formula:
[RNA ?g/?l] = A260 × 33 × dilution factor/1,000.
Many commercial kits are available for the reverse transcriptase
reaction. The following protocol has been optimized for the
ImProm-II Rt system.
1. Preparation of a reaction master mix is highly recommended to
give best reproducibility. Mix all reagents but template in a
common mix, using ~10% more than needed. A 20 ?l reaction
contains 4 ?l reaction buffer, 3 mM MgCl2, 0.5 mM of each
dNTP, 1 ?/?l ribonuclease inhibitor (we suggest recombinant
RNAasin), and 1 ?l of reverse transcriptase (we suggest the
Improm-II system, see Note 13).
2. Anneal RNA and oligos for 5 min at 25°C.
3. For each reaction, combine 15 ?l of reverse transcriptase mix
with RNA and oligos for a final volume of 20 ?l.
4. The first-strand cDNA is synthesized by incubating the reaction
for 60 min at 42°C (i.e., the extension reaction) and termi-
nated by incubating at 70°C for 15 min.
Quantitative PCR uses the linearity of DNA amplification to
determine relative amounts of a known sequence in a sample by
using a fluorescent reporter in the reaction. When the DNA is in
3.3.3. Quantitative PCR
24116 Investigation of Acquired Resistance to EGFR-Targeted Therapies…
the log linear phase of amplification, the amount of fluorescence
increases above the background. The point at which the fluores-
cence becomes measurable is called the Threshold Cycle (CT) or
1. A reaction master mix (see Note 14) of all reagents but
template is assembled as a common mix according to Table 1.
2. Aliquot the master mix into a 200 ?l PCR tube or plate.
3. Add 1 ?l of RT reaction.
4. Mix gently by vortexing and briefly centrifuge to collect all
components at the bottom of the tube.
5. Perform thermal cycling according to Table 2 (see Note 17).
The first step in gene expression profiling analysis is the RNA isola-
tion. We refer to Subheading 3.3.1 for a detailed protocol.
1. Anneal the oligo dT(12–18) primer to the RNA in the reaction
mixture defined in Table 3 using a 0.2-ml thin-wall PCR
3.4. Gene Expression
3.4.1. RNA Labeling
Pipetting scheme for QPCR
Volume*Reagent Final concentration
25 ?l 2× JumpStart Taq ReadyMix1.5 units Taq DNA polymerase, 10 mM Tris–HCl,
50 mM KCl, 1.5 mM MgCl2, 0.001% gelatin,
0.2 mM dNTP, stabilizers
– ?lForward primer 50–1,000 nM (see Notes 15 and 16)
– ?l Reverse primer50–1,000 nM
– ?l Template DNA10–100 ng
– to 50 ?l Water
50 ?lTotal volume
PCR cycle conditions
Initial denaturation94°C 2 min*
Denaturation94 °C 15 s
Annealing/extension60°C or 5°C below lowest primer TM
(Optional) Hold4°C – only if products will be run out on a gel
242 K. Kani et al.
tube so that incubations can be carried out in a PCR cycler
(see Note 18).
2. Heat to 65°C for 10 min and transfer on ice for 2 min.
3. Add 23 ?l of the reaction mixture defined in Table 4 containing
either Cy5-dUTP or Cy3-dUTP nucleotides, mix well by
pipetting, and use a brief centrifuge spin to concentrate the
liquid in the bottom of the tube.
4. Incubate at 42°C for 30 min. Then add 2 ?l Superscript II.
Make sure that the enzyme is well mixed in the reaction
volume and incubate at 42°C for 30–60 min (see Note 19).
5. Stop the reaction by adding 5 ?l of 0.5 M EDTA (see Note 20).
6. Add 10 ?l of 1 N NaOH and incubate at 65°C for 30 min
to hydrolyze residual RNA. Cool to room temperature (see
7. Neutralize the reaction by adding 25 ?l of 1 M Tris–HCl,
8. Desalt the labeled cDNA by adding the following neutralized
reaction: 400 ?l of TE pH 7.5 and 20 ?g of human C0t-1
DNA. Transfer the solution to a MicroCon 100 cartridge.
Pipette to mix and spin for 10 min at 500 × g.
Pipetting scheme for RNA annealing
Addition for Cy5
Addition for Cy3
Total RNA (>7 mg/ml) 150–200 ?g 50–80 ?g
dT(12–18) primer (1 ?g/?l) 1 ?l 1 ?l
DEPC H2Oto 17 ?lto 17 ?l
Pipetting scheme for RNA labeling
Component Volume [ml]
5× first-strand buffer8
10× low T dNTPs mix4
Cy5 or Cy3 dUTP (1 mM) 4
0.1 M DTT4
Rnasin (30 ?/?l)1
Superscript II (200 ?/?l)2
24316 Investigation of Acquired Resistance to EGFR-Targeted Therapies…
9. Wash again by adding 200 ?l TE pH 7.5 and concentrating to
about 20–30 ?l by centrifuging at 500 × g for 8–10 min.
10. Recover the sample by inverting the concentrator over a clean
collection tube and spinning for 3 min at 500 × g.
11. Take a 2–3 ?l aliquot of the Cy5-labeled cDNA for analysis,
leaving the remaining 18–28 ?l for hybridization.
12. Run this probe on a 2% agarose gel (6 cm wide × 8.5 cm long,
2 mm wide teeth) in Tris Acetate Electrophoresis Buffer (TAE,
see Note 22).
13. Scan the gel on a fluorescence scanner (setting: red fluores-
cence, 200 ?m resolution, see Note 23).
1. As an initial step, it is necessary to determine the volume of
the hybridization solution required (see Note 24). Usually,
0.033 ?l of hybridization solution is used for each mm2 of
slide surface area covered by the coverslip for each array is used
(i.e., an array covered by a 24 mm by 50 mm coverslip will
require 40 ?l of hybridization solution).
2. For a 40 ?l hybridization, pool the Cy3- and Cy5-labeled cDNAs
into a single 0.2-ml thin-wall PCR tube and adjust the volume
to 30 ?l by either adding DEPC-treated water, or by removing
water in a vacuum concentrator. If using a vacuum device to
remove water, do not use high heat or heat lamps to accelerate
evaporation as the fluorescent dyes might be degraded.
3. For a 40-?l hybridization, combine the components specified
in Table 5.
4. Mix the components well by pipetting, heat at 98°C for 2 min in
a PCR cycler, cool quickly to 25°C, and add 0.6 ?l of 10% SDS.
5. Centrifuge for 5 min at 14,000 × g. The labeled cDNAs have
a tendency to form small, very fluorescent, aggregates
Pipetting scheme for array hybridization
Component High sample blockingHigh array blocking
Cy5 + Cy3 probe30 ?l 28 ?l
Poly d(A) (8 mg/ml)1 ?l2 ?l
Yeast tRNA (4 mg/ml)1 ?l2 ?l
Human C0t-1 DNA (10 mg/ml)1 ?l0 ?l
20× SSC6 ?l6 ?l
50× Denhardt’s blocking solution1 ?l (optional)2 ?l
Total volume40 ?l40 ?l
244 K. Kani et al.
which result in bright, punctate background on the array slide.
Hard centrifugation will pellet these aggregates, avoiding their
introduction to the array.
6. Apply the labeled cDNA to a 24 mm × 50 mm glass coverslip
and then touch with the inverted microarray (see Note 25).
7. Place the slide in a microarray hybridization chamber. Add 5 ?l
of 3× SSC to the reservoir and seal the chamber. Submerge the
chamber in a 65°C water bath and allow the slide to hybridize
for 16–20 h (see Note 26).
8. Wash off unbound fluorescent cDNA by incubating in prewarmed
2× SSC, 0.2% SDS washing solution at 65°C for 5 min. Then,
wash the array three times in 2× SSC, 0.2% SDS washing
solution at RT for 2 min. Subsequently, air-dry the array.
9. Remove the hybridization chamber from the water bath, cool
and carefully dry off. Unseal the chamber and remove the slide
(see Note 27).
10. Place the slide, with the coverslip still affixed, into a Coplin jar
filled with 0.5× SSC/0.01% SDS wash buffer. Allow the coverslip
to fall from the slide and then remove the coverslip from the jar
with a forceps. Allow the slide to wash for 2–5 min.
11. Transfer the slide to a fresh Coplin jar filled with 0.06× SSC.
Allow the slide to wash for 2–5 min (see Note 28).
12. Transfer the slide to a slide rack and centrifuge at low speed for
3 min in a clinical centrifuge equipped with a horizontal rotor
for microtiter plates (see Note 29).
13. Obtain image of array. Images are typically obtained using
microarray scanners. A large variety of microarrays scanners are
available with different features such as sensitivity, resolution,
scan area, rapidity of scan, excitation and emission filters, and
autofocus capability. For our studies, we used the Affymetrix
GeneChip Scanner 3000.
1. The first step in any analysis of microarray data is the extraction
of quantitative information from the images resulting from the
readout of hybridizations. As a preliminary analysis, the data
from each experiment is assessed for overall quality and to
determine the extent of difference between the experimental
sample and the reference sample. A segmentation of the array
is initially performed to decode the array information. Since
each element of an array is printed automatically to a predefined
location and because specific orientation markers are given, the
initial positioning of the grid can be automatically aligned.
2. The background of the microarray image is usually not uniform
over the entire array. Therefore, it is necessary to extract local
background intensities. Owing to many technical reasons,
3.5. Data Analysis
3.5.1. Array Segmentation,
Extraction, and Target
24516 Investigation of Acquired Resistance to EGFR-Targeted Therapies…
the changes of fluorescent background across an array are
usually gradual and smooth. In case more abrupt changes
are observed, we recommend caution in the interpretations of
the data. Conventionally, pixels near the bonding box edge are
taken to be the background pixels, and the average gray level
of these pixels are used for the estimation of the local back-
3. A fixed thresholding method is used in image analysis to deter-
mine the signal corresponding to a specific target over the
background. A threshold value (T) corresponding to changes
of over three times the value of standard deviation (s) over the
local background mean (m) is usually used (e.g., T = m + 3s).
However, this method based on a simple fixed thresholding
fails quite often due to variability of the background and the
signal, particularly when the signal is weak (a frequent finding
in cDNA array experiments). To avoid these problems, a more
sophisticated thresholding method such as the Mann–Whitney
method may be implemented (7).
4. The local background value is subtracted from the reported
sample intensities from the red channel (R) and the green
channel (G), and then the ratio (R/G) is calculated. Clearly,
the ratio measurement is the ratio of two average intensity
measurements. We usually use Affymetrix Expression Console
QC metrics to pass the image data, although other software
packages are available and can be used.
A second general step in the analysis of gene expression profile
data is the collection of experimental data into a database that
supports both further mathematical analysis and connection to
the available knowledge about the structure and function of the
individual genes (Database Design & Development). Usually, this
step is automatically generated by current software packages.
1. Gene annotation provides functional information on a given
gene. In addition, depending on the tool utilized this analysis
can also provide other useful information such as for example
the location of each gene within a particular chromosome.
Having identified some set of regulated genes, the next step in
expression profiling analysis involves looking for patterns
within the regulated set. The way in which particular genes’
expression patterns vary across the samples can be analyzed in
a variety of ways. In expression clustering analysis, a query asks
whether the similarities of patterns between genes suggest
involvement in common processes, or whether differences in
the individual gene expression patterns can be aligned with
differences in the sample types. In Discriminative Gene List
analysis, genes having the most differential behavior between
3.5.2. Data Visualization
3.5.3. Gene Annotation
246 K. Kani et al.
samples can be used to try to identify the particular cellular
activities that differ between classes of samples. By using these
two types of analysis, one can start to make biological sense of
expression profiling data. Many analytical tools are currently
available for data mining and to determine whether, for example,
proteins made from genes with similar patterns of expression
perform related functions, whether they are chemically alike or
have similar subcellular localization.
2. As part of available tools, Gene ontology analysis provides a
standard way to define these relationships. Gene ontologies
start with very broad categories, e.g., “metabolic process,” and
break them down into smaller categories, e.g., “carbohydrate
metabolic process,” and finally into quite restrictive categories
such as “inositol and derivative phosphorylation.” The Molecular
Signatures Database, the Comparative Toxicogenomics Data-
base and the Ingenuity Gene Network Diagram are examples
of resources to further categorize genes.
1. Data analysis of microarrays has become an area of intense
research. Simply stating that a group of genes are regulated
by at least twofold lacks a solid statistical footing. The usual
typical three replicates in each group are in fact not sufficient
from a statistical perspective and can easily create a deceptive
difference greater than twofold. Rather than identifying differ-
entially expressed genes using a fold change cutoff, one can
use a variety of statistical tests such as ANOVA, all of which
estimate how often we would observe the data by chance alone.
This type of analysis also allows the identification of genes that,
despite subtle variation in expression, could have important
2. In addition to treatment (i.e., drug) or comparison of multiple
cell lines, it is also helpful to include information such as
hybridization, operator, scanner, etc. as variable control items.
3. The output of an ANOVA test includes two values: a p-value
and a mean square for each factor tested. The mean of the
mean square for each factor is used to quantify the contribu-
tion of each factor to the experimental variation. The p-values
indicate instead the significance of an observed difference in
the expression. The lower the p-value, the more confidence
one has that the difference detected are not due to random
chance. A p-value of 0.05 is typically defined to indicate
significance, since it estimates a 5% probability of observing
the data by chance. A p value alone of course is not sufficient
since for example analysis of 10,000 genes on a microarray
would result in the identification of 500 genes at p < 0.05
even if there were no difference between the experimental
groups. A variety of analysis packages from bioinformatics
3.5.4. Validation by
24716 Investigation of Acquired Resistance to EGFR-Targeted Therapies…
companies are available to account for this so-called multiple
hypothesis testing. Among these tools, we use the Significance
Analysis of Microarrays (SAM) system. In addition, we suggest
following the MAQC Project recommendations for microarray
Ultimately, changes in the expression of certain genes of interest
should be further validated by more direct methods such as quan-
titative PCR and western blot analysis. While high throughput
DNA microarrays lack the quantitative accuracy of QPCR, it takes
about the same time to measure the gene expression of a few dozen
genes via QPCR as it would to measure an entire genome using
DNA microarrays. Thus, we suggest performing semiquantitative
DNA microarray analysis experiments to identify candidate genes
and then perform QPCR on some of the most interesting candi-
date genes to validate the microarray results. Western blot analysis
of some of the protein products of differentially expressed genes
is also recommended, since the mRNA levels do not necessarily
correlate to the amount of expressed protein. We refer to the previous
section for a detailed protocol for QPCR analysis.
1. Cell growth should be carefully monitored, as some cells take
longer to obtain a resistant phenotype (see Subheading 3.1.1).
2. Drug resistance can be obtained by incremental titration of a
particular drug over a several month time course. For example,
the IC50 of the HCC827 cell line treated with gefitinib is in
the nanomolar range (10–100 nM). Once the IC50 has been
determined for a particular cell line and drug, the initial dosage
can be set. For example, if the IC50 for the HCC827 cell line
with gefitinib is ~100 nM, then one could begin dosing with
10 nM of gefitinib for 1 month. The exact concentration of
gefitinib has to be established by a case-by-case basis, but initial
dosing at concentrations above the IC50 can result in apopto-
sis and the release of cell debris that can prevent proper cell
attachment and growth (see Subheading 3.1.1).
3. The rate of the development of an acquired resistance cell
line also depends on the frequency of drug replenishment.
Gefitinib is an anilinoquinazoline and thus will undergo latent
hydrolysis in aqueous and is insoluble in alkaline pH. Therefore,
the replenishment of the gefitinib containing media will
determine the effective inhibition time of EGFR. The half-life
of the therapeutic in vitro will determine the rate and possibly
influence the mechanism of resistance. This parameter should
3.5.5. Validation by QPCR
and Western Blot Analysis