SUPPORTING DOCUMENTS (online only)
Method-S1: Cell culture
For in vitro assays using human gallbladder carcinoma cell lines, RCB1129 and RCB1130 cells were purchased from Japanese Riken Cell Bank and SNU308 cells from Korean Cell Line Bank. SNU308 and RCB1130 cells were derived from primary tumors, while RCB1129 cells were from a nodal metastasis. RCB1129 and RCB1130 cells were grown in Dulbecco’s modified Eagle’s medium supplemented with 5% heat-inactivated fetal bovine serum (FBS). SNU308 cells were maintained in RPMI1640 medium supplemented with 10% heat-inactivated FBS.
Method-S2: Confocal immunofluoresce
Cells were fixed in 4% paraformaldehyde at 4°C for 1 hour, washed once with ice-cold PBS for 5 min, and then treated with 4% bovine serum albumin (Sigma-Aldrich, St. Louis, MO) for 1 hour at room temperature. Afterwards, cells were incubated with indicated primary antibodies against EGFR (1:100, B01P, mouse polyclonal, Abnova, Taipei, Taiwan) and pEGFR (1:100, specifically against Tyrosine 1086, rabbit polyclonal, Zymed, San Francisco, CA)for 1 hour atroom temperature. Cells were then washed and further incubated for 1 hour at room temperature with 1 ml of diluted (1:200) fluorescence-labeled secondary antibodies (Invitrogen, Carlsbad, CA), including either anti-mouse IgG conjugated with Alexa fuor 488 (#A31620) or anti-rabbit IgG with Alexa fuor 565 (#A11036) to probe EGFR and p-EGFR, respectively. To better delineate the nuclear contours, Hoechst dye was used for nuclear counterstain. Immunostained cells wereexamined under a digital Eclipse C1 confocal microscope system (Nikon, Tokyo, Japan).
Method-S3: Fractional immunoblotting
NE-PER Nuclear and Cytoplasmic Extraction Reagent (Thermo Fisher Scientific, Fremont, CA, USA) was used to fractionate nuclear and cytoplasmic proteins according to manufacturer’s instructions. From three cell lines, equal amounts of protein (25 μg) extracted were separated on 4-12% gradient SDS-PAGE gel NuPAGE(Invitrogene, Carlsbad, CA) and then transferred to PVDF membranes (Amersham Biosciences, Buckinghamshire, UK). After blocking with 5% skimmed milk in TBST buffer at room temperature for 1 hour, the membranes were probed with antibodies at 4 oC overnight against EGFR (1:1000, EP38Y, rabbit monoclonal, Epitomics, Burlingame, CA) and pEGFR (1:500, specifically against Tyrosine 1086, rabbit polyclonal, Zymed, San Francisco, CA). GAPDH (6C5, 1:10000, Millipore,Beverly, MA) and Histone H2B (EP819Y, 1:1000, rabbit monoclonal, Epitomics, Burlingame, CA) were used for the loading and fraction purity controls of cytosolic and nuclear lysates, respectively. The membranes were then incubated with the secondary antibodyat room temperature for 1.5 hours to visualize the targeted proteins with enhanced chemiluminescence reagents (Amersham Biosciences, Buckinghamshire, UK).
Method-S4: ChIP assay demonstrating pEGFR binding to iNOS gene promoter
SNU308, RCB1129, and RCB1130 cells with or without EGF treatmentwere treated with 1% formaldehyde at room temperature for 15 min. The cross-linked chromatin was then prepared and sonicated to an average size of 500 bp. The DNA fragments were immunoprecipitated with antibodies to pEGFR (specifically against Tyrosine 1086,polyclonal, 1:25, rabbit polyclonal, Zymed, San Francisco, CA) or control rabbit immunoglobulin G (IgG) at 4 °C overnight. After reversal of the cross-linking, the immunoprecipitated chromatin was amplified by primers related to specific regions of the iNOS genomic locus. The primersequenceswereiNOS/-157-ChIP(S)-5’-ctggcagtcacagtcataa-3’ and iNOS/+24-ChIP(AS)-5’-gttttcgactcgctacaaag-3’. The amplified DNA products were resolved by agarose gel electrophoresis and confirmed by sequencing.
Method-S5: Transfection of piNOSpromoter-luciferase construct for reporter assay
Based on the published literature, the most critical region of iNOS promoter transcriptionally regulated by EGFR was adopted to establish an iNOS promoter-luciferase construct(Cancer Cell 2005;7: 575-89). The latter was transfected into three gallbladder cancer cell lines with Lipofectamine 2000TM (Invitrogen, Carlsbad, CA)for reporter assay. At 6hours post-transfection, cells were changed to 1.5 ml of complete medium and then treated with EGF (Peprotech, Rocky Hill, NJ) for 15 hours as indicated. The luciferase activity in the cell extracts was measured with the luciferase assay system (Promega, Madison, WI).
Supplementary method-S6: Selection and clincopathological characterization of both TMA (training) and whole section (test) sample sets
In the training set used for tissue microarray (TMA) construction, 112 consecutive cases of gallbladder carcinoma undergoing primary surgery between 1986 and 2004were retrievedfrom the archivesof Chi-Mei and Chang Gung hospitals, 2 tertiary medical centers in Southern Taiwan.For the test set using whole sections for immunohistochemistry, 58 gallbladder carcinomasconsecutively resected between January, 2000 and March, 2010 were identified from the database of ChanghuaChristianHospital, a tertiary medical center in Central Taiwan. These cases were classified for histological types and tumor grading according to the latest WHO classification [24].Other histopathological features evaluated included the presence or absence of spontaneous tumor necrosis and vascular invasion. In addition, pathological staging was determined based on the6th edition of American Joint Committee on Cancer (AJCC) system [25]. Surgical procedures were cholecystectomy in 82 and 29 patients and extended operation in the remaining 22 and 29 patients in the training andtest sets, respectively. None of the patients enrolled in both sample sets receivedpreoperative chemotherapy or radiation therapy.In the training set, follow-up data were available in 100 and 74 cases with informative data of immunohistochemical scores (median, 12 months; range, 1 to 234 months) andEGFR gene dosage(median, 13 months; range, 1 to 234 months), respectively. The follow-up duration of 58 cases in the test set for pEGFR immunostain ranged from 1 to 55 months (median, 13 months). For 39 and 29 survivors in the training and test sets, the median durations of follow-up were 22 and 18 months, respectively.In addition, 20 cases of gallbladder precancerous dysplastic leions were also retrieved for immunohistochemicalanalysis to evaluate the roles of EGFR and phospho-EGFR nuclear import in carcinogenesis. These consisted of 10 cases with low-grade dysplasia alone, 5 cases with high-grade dysplasia alone, and 5 cases with concurrent high-grade dysplasia and invasive adenocarcinoma (see Table-S4).
Method-S7: Immunohistochemistry
The TMA blocks were constructed and whole tissue sections preparedfor immunohistochemistry with different conditions of antigen retrieval. For the training set, 5 tissue cylinders (1 mm in diameter) from tumor areas were punched from a representative paraffin block for each specimen using a precision instrument (Beecher Instruments, Silver Spring, MD) and arrayed into recipient blocks. In addition, we also arrayed three each carcinomas of the colorectum, lung, and breast for the purposes of orientation and external positive controls in each TMA block. A whole tissue section was recut at 2-μm from a representative tissue block for each case in the test set of another 58 cases. Both TMA and whole sections were pretreatedwith 0.05% trypsin in PBSbuffer for15 minutes at room temperature before incubation with EGFR (RTU-EGFR-384, ready-to-use, mouse monoclonal, COOH terminus, Novocastra, Newcastle upon Tyne, UK). Those used for pEGFR(1:25, specifically against Tyrosine 1086, rabbit polyclonal, Zymed, San Francisco, CA) and iNOS (1:100, mouse monoclonal, Clone 6, BD Biosciences, Franklin Lakes, NJ) were heated by microwave treatment in a10mM citrate buffer at pH 6for 7 minutes, followed by TBS wash for 15 minutes. Primary antibodies were detected using the ChemMateDAKO EnVision kit (DAKO, K5001, Carpinteria, CA).
The percentage of tumor cells with moderate to strong immunoreactivity to cytoplasmic pEGFR (C-pEGFR), N-EGFR, N-pEGFR, and iNOS (with cytoplasmic staining only) was separately scored for multiple cores by one pathologist (C.F.L) to obtain anaverage expression index for each casecontaining 2 or more preserved tissue cores.In the TMA training set, the medians of average express index were 20% for iNOS and 50% for cytoplasmic pEGFR (C-pEGFR), N-EGFR, and N-pEGFR and adopted as the cutoffs to define overexpression. By using the same cutoff (50%), immunoexpression of C-pEGFR and N-pEGFR was evaluatedby another pathologist (H.C.T) for the whole sections of gallbladder carcinomatous and precancerous dysplasticlesions in the three most active areas at 200X power filed.
Method-S8: CISH assay to detect EGFR gene amplification
CISH assay for evaluatingEGFR gene dosage was performed on TMA slides, which were incubated at 55°C overnight, deparaffinized in xylene, and dehydrated in graded ethanols. Pre-heating at 98–100°C was carried out using the provided buffer (Zymed, San Francisco, CA) for 15 minutes. The tissue was then digested with pepsin for 10 min at room temperature. After application of Zymed SpotLight® digoxigenin-labeled EGFR probe (Zymed,San Francisco, CA), the TMA slides were cover-slipped and edges sealed with rubber cement. The slides were heated at 95°C for 5 minutes, followed by overnight incubation at 37°C using a moisturized chamber. Post-hybridization wash was performed the next day and followed by immunodetection using the CISHTM polymer detection kit (Zymed,San Francisco, CA).
Method-S9: Mutant-enriched PCR assay
DNA isolation
Genomic DNA was prepared from eight 5-μm-thick sections of paraffin-embedded tumor material by using RecoverAllTM total nucleic acid isolation kit (Ambion, Austin, TX) according to the manufacture’s protocols.Paraffin blocks were trimmed to minimize the amount of non-neoplastic tissue priorto sectioning.The concentration of DNA was quantifiedusing NanoDropspectrometer.
Mutant-enriched PCR amplification and bidirectional sequencing
Given that GBC specimens may contain only a small fraction of carcinoma cells, conventional PCR for whole tissue sections can result in a potential bias to preferentially amplify the larger amount of wild-type genes, epically in well-differentiated cases with low tumor cellularity. To minimize the false negative detection of mutated EGFR gene, we applied mutant-enriched PCR to amplify exon 19 and exon 21 of EGFR gene, following the methodology reported by Asano et al [12] with minor modifications. In brief, mutant-enriched PCR is a two-step PCR with restriction digestion to selectively eliminate intermittentPCR products of wild-type genes,thus enriching the mutated genes.
For the first round of amplification of exon 19, the sequences of primers weresense: 5'-ATCCCAGAAGGTGAGAAAGATAAAATTC-3' and antisense: 5'-CCTGAGGTTCAGAGCCATGGA-3'. In this reaction, 17 thermal cycles of 20 seconds at 94°C, 30 seconds at 60°C,and 20 seconds at 72°C were run for each specimen by using 25 ng of sample DNA, 150µmol/L dNTPs, 2 pmol of each primer,and 0.25 units of HotStarTaq DNA polymerase (Qiagen, Inc., Valencia,CA).Ten IU of MseIwere then incubated with 2 µL of the first PCR products at 37°C for 4 hours to digest the TTAA sequence (spanning codons 747 and 748) in wild-type genes. TTAA is frequently deleted in exon 19 mutants according to the previous studies on pulmonary adenocarcinomas, resulting in the enrichment of deletion-typegenes[12]. Afterwards, an aliquotwas used as a template for the second round of PCR amplificationfor 40cycles under the same conditions as the first round PCR. The product of the second amplification was analyzedon 12% PAGE. Because the length deleted in mutants of exon 19, spanningcommonly deleted codons 747 to 749, is multiples of 3 bps, difference inthe sizes of the PCR amplicons allows for distinguishing mutantfrom wild-type amplicons.
The sequences of the primersfor the first round PCR of EGFR exon 21were outer sense primer: 5'- CTAACGTTCGCCAGCCATAAGTCC -3' and outer antisense primer: 5'- GCTGCGAGCTCACCCAGAATGTCTGG-3'. We used the restriction enzyme MscI to digest TGGCCAsequence in the first round PCR amplicon of the wild-type gene under the sameconditions and reagent make-up as those for exon 19. In this approach, EGFR exon 21mutant would not be eliminated, giventhesubstitutionof (T>G)GGCCAresulting in the enrichmentof the L858R mutant.After digestion usingMscI, the second PCR was performed using an inner forward primer:CAGCCATAAGTCCTCGACGTGG and inner reverse primer: CATCCTCCCCTGCATGTGTTAAAC, the product of which was subsequently digested with Sau96I (GGNCC)for RFLP analysis using electrophoresis on 12% PAGE. Since Sau96I digests the mutant type of codon 858(CGG) but not wild type, even partialdigestion of the2ndPCR product indicates the presence of a mutation at codon 858.
To confirm the validity of mutant-enriched PCR assays, 12 selected gallbladder carcinomas with wild genotypes of exons19 and 21werebidirectionallysequencedusing ABI3100 Prism Genetic Analyzer (Applied Biosystems).We found that all 12 cases, including those with EGFR amplification detected by CISH, showed wild genotypes in EGFR exons 19 and 21, confirming the specificity of methodology applied.
Table-S1: Clinicopathological, immunohistochemical, and molecular findings of gallbladder carcinomas in the tissue microarraytraining set
Parameter / Cohort for immunostains / Cohort for EGFR gene statusGender / Male / 35 (33.7%) / 24 (31.6%)
Female / 69 (66.3%) / 52 (68.4%)
Age / =<60 / 27 (26.0%) / 21 (27.6%)
61-70 / 35 (33.7%) / 20 (26.3%)
71-80 / 29 (27.9%) / 25 (32.9%)
81-90 / 12 (11.5%) / 10 (13.2%)
>=90 / 1 (1.0%) / 0 (0%)
AJCC Stage / I / 44 (42.3%) / 32 (42.1%)
II / 55 (52.9%) / 39 (51.3%)
III / 2 (1.9%) / 2 (2.6%)
IV / 3 (2.9%) / 3 (3.9%)
Histological type / Tubular / 83 (79.8%) / 58 (76.3%)
Papillary / 11 (10.6%) / 10 (13.2%)
Other / Adenosquamous (5), Undifferentiated (4), Clear cell (1) / Adenosquamous (4), Undifferentiated (3), Clear cell (1)
Histological grade / Grade 1 / 53 (51.0%) / 40 (52.6%)
Grade 2 / 31 (29.8%) / 20 (28.9%)
Grade 3 / 16 (15.4%) / 12 (15.8%)
Grade 4 / 4 (3.8%) / 2 (2.6%)
Vascular invasion / Present / 48 (46.2%) / 34 (44.7%)
Absent / 56 (53.8%) / 42 (55.3%)
Perineurial invasion / Present / 14 (13.5%) / 11 (14.5%)
Absent / 90 (86.5%) / 65 (85.5%)
Tumor necrosis / Present / 50 (48.1%) / 38 (50%)
Absent / 54 (51.9%) / 38 (50%)
N-EGFR / Low expression / 79 (76%) / 61 (80.3%)
(<50 vs. >=50) / Overexpression / 25 (24%) / 15 (19.7%)
C-EGFR / Low expression / 82 (78.8%) / 62 (81.6%)
(0-2+ vs. 3+) / Overexpression / 22 (21.2%) / 14 (18.4%)
N-pEGFR / Low expression / 42 (40.4%) / 26 (34.2%)
(<50 vs. >=50) / Overexpression / 62 (59.6%) / 50 (65.8%)
C-pEGFR / Low expression / 32 (30.8%) / 16 (21.1%)
(<50 vs. >=50) / Overexpression / 72 (69.2%) / 60 (78.9%)
EGFR amplification / No amplification / 65 (85.5%) / 65 (85.5%)
Amplification / 11 (14.5%) / 11 (14.5%)
iNOS / Low expression / 43 (41.3%) / 25 (32.9%)
(<20 vs. >=20) / Overexpression / 61 (58.7%) / 51 (67.1%)
N; nuclear; C, cytoplasmic; p, phosphorylated. No essential difference in the clinicopathological characteristics was noted between cohorts for immunochemistry and EGFR gene testing (p>0.100 for all comparisons).
Table-S2: Clinicopathological and immunohistochemical findings of another 58 gallbladder carcinomas in the validation whole section test set
Parameter / Number (Percent)Gender / Male / 27 (46.6%)
Female / 31 (53.4%)
Age / =<60 / 11 (19.0%)
61-70 / 19 (32.8%)
71-80 / 21 (36.2%)
81-90 / 6 (10.3%)
>=90 / 1 (1.7%)
AJCC Stage / I / 18 (31.0%)
II / 11 (19.0%)
III / 21 (36.2%)
IV / 8 (13.8%)
Histological type / Tubular / 51 (87.9%)
Papillary / 5 (8.6%)
Other (Signet ring cell) / 2 (3.4%)
Histological grade / Grade 1 / 12 (20.7%)
Grade 2 / 26 (44.8%)
Grade 3 / 10 (17.3%)
Grade 4 / 10 (17.3%)
Vascular invasion / Present / 10 (17.2%)
Absent / 48 (82.8%)
Perineurial invasion / Present / 9 (15.5%)
Absent / 49 (84.5%)
Tumor necrosis / Present / 26 (44.8%)
Absent / 32 (55.2%)
N-pEGFR / Low expression / 29 (50.0%)
(<50 vs. >=50) / Overexpression / 29 (50.0%)
C-pEGFR / Low expression / 26 (44.8%)
(<50 vs. >=50) / Overexpression / 32 (55.2%)
Table-S3 Univariate log-rank and multivariate Cox regression analyses of disease-specific survival for the test set
Parameter / Univariate analysis / Multivariate analysisNo. of case / No. of event / p-value / H.R / 95% CI / p-Value
Gender / Male / 27 / 4 / 0.0345* / 1 / 0.0441*
Female / 31 / 12 / 4.270 / 1.039-17.543
Age / <70 / 28 / 5 / 0.0198* / 1 / 0.0283*
>=70 / 30 / 11 / 3.999 / 1.159-13.798
AJCC Stage / I / 18 / 2 / 0.0057* / 1 / - / 0.0251*
II-IV / 40 / 14 / 7.104 / 1.277-39.519
Histological type / Papillary / 5 / 1 / 0.0124* / 1 / 0.1084
Tubular / 51 / 13 / - / -
Other / 2 / 2 / 2.835 / 0.795-10.115
Histological grade / Grade 1 / 26 / 8 / 0.7414
Grade 2-4 / 32 / 8
Vascular invasion / Absent / 48 / 10 / 0.0005* / 1 / - / 0.0999
Present / 10 / 6 / 2.886 / 0.816-10.199
Perineurial invasion / Absent / 49 / 11 / 0.1754
Present / 9 / 5
Tumor necrosis / Absent / 32 / 7 / 0.2437
Present / 26 / 9
N-pEGFR / Low expression / 29 / 3 / 0.0208* / 1 / - / 0.0223*
(<50 vs. >=50) / Overexpression / 29 / 13 / 5.573 / 1.277-24.316
C-pEGFR / Low expression / 26 / 8 / 0.3959
(<50 vs. >=50) / Overexpression / 32 / 8
*, statistically significant; N; nuclear; C, cytoplasmic; p, phosphorylated; HR, hazard ratio; CI, confidence interval
Table-S4 Immunohistochemicalfindings of EGFR and phospho-EGFR staining in precancerous dysplastic lesions of gallbladder
Category / N-EGFR / N-pEGFRNumber / Low expression / Overexpression / Low expression / Overexpression
Low-grade dysplasia / 10 / 10 / 0 / 10 / 0
High-grade dysplasia / Noassociated carcinoma / 5 / 5 / 0 / 5 / 0
with
associated carcinoma / 5 / 4 / 1 / 3 / 2
Figure-S1phospho-EGFR reactivity was absent in low-grade dysplastic epithelial cells of gallbladder(A) but overexpressed in the nuclei of glandular epithelial cells featuring high-grade dysplasia (B). .