Professional Documents
Culture Documents
ADAM28: A Potential Oncogene Involved In: Asbestos-Related Lung Adenocarcinomas
ADAM28: A Potential Oncogene Involved In: Asbestos-Related Lung Adenocarcinomas
Asbestos-related lung cancer accounts for 4–12% of all lung cancers worldwide. Since putative mechanisms of carcinogen-
esis differ between asbestos and tobacco induced lung cancers, tumors induced by the two agents may be genetically dis-
tinct. To identify gene expression biomarkers associated with asbestos-related lung tumorigenicity we performed gene
expression array analysis on tumors of 36 patients with primary lung adenocarcinoma, comparing 12 patients with lung
asbestos body counts above levels associated with urban dwelling (ARLC-AC: asbestos-related lung cancer-adenocarci-
noma) with 24 patients with no asbestos bodies (NARLC-AC: non-asbestos related lung cancer-adenocarcinoma). Genes
differentially expressed between ARLC-AC and NARLC-AC were identified on fold change and P value, and then priori-
tized using gene ontology. Candidates included ZNRF3, ADAM28, PPP1CA, IRF6, RAB3D, and PRDX1. Expression of these
six genes was technically and biologically replicated by qRT-PCR in the training set and biologically validated in three inde-
pendent test sets. ADAM28, encoding a disintegrin and metalloproteinase domain protein that interacts with integrins,
was consistently upregulated in ARLC across all four datasets. Further studies are being designed to investigate the possi-
ble role of this gene in asbestos lung tumorigenicity, its potential utility as a marker of asbestos related lung cancer for
purposes of causal attribution, and its potential as a treatment target for lung cancers arising in asbestos exposed
persons. VC 2010 Wiley-Liss, Inc.
V
C 2010 Wiley-Liss, Inc.
ADAM28, A POSSIBLE ONCOGENE IN ARLC-AC 689
multiplicative model (Vainio and Boffetta 1994; MATERIALS AND METHODS
Liddell 2001; Lee, 2001). More recent reviews
Sample Collection and Subject Selection
tend to support a cumulative exposure model for
lung cancer as a consequence of past asbestos ex- The study population described previously
posure (Churg, 1993; Egilman and Reinert, 1996; (Larsen et al., 2007) consisted of 36 cases of pri-
Billings and Howard, 2000; Hodgson and Darn- mary lung adenocarcinoma. Lung asbestos fiber
ton, 2000). count was measured using the bleach digestion
Asbestos fibers have been shown to accumulate technique described by (Roggli, 1992) and
near mitotic spindles, and it has been postulated (Wright, et al., 2008). Given that urban dwellers
that asbestos interferes directly with chromo- without occupational asbestos exposure have
somes (Hesterberg and Barrett, 1985; Dopp et al., fewer than 20 asbestos bodies/gram wet weight
1995). In vitro studies using DNA repair deficient (AB/gww) lung tissue (Roggli et al., 1982), we
Cell lines have reported time and dose depend- classified subjects with asbestos fiber burdens >
ent increases in DNA double-strand breaks after ¼ 20 AB/gww as ‘‘moderately exposed’’ (n ¼ 12)
asbestos exposure (Okayasu et al., 1999), and and subjects with 0 AB/gww as ‘‘nonexposed’’
others have suggested that asbestos can physically (n ¼ 24). Subjects were clinically characterized
shear DNA (Libbus et al., 1989). Fiber size, for tumor phenotype (age, gender, tobacco expo-
shape, and diameter appear to be important varia- sure, stage, lobe location, and differentiation)
bles in asbestos-cell interactions, with long (Table 1). Subjects gave informed written con-
straight needle-like fibers showing a high interac- sent for tissue resection and this study was
tion rate associated with altered cellular morphol- approved by The Prince Charles Hospital Human
ogy (Cardinali et al., 2005). Several pathways Research Ethics Committee.
have been implicated in asbestos-induced lung
injury with cellular functions, including oxidative
stress response, inflammation, DNA damage RNA Extraction and Gene Expression Analysis
response, mitochondrial activity, and apoptosis Total RNA was extracted from 20–30 mg of
thought to be triggered by asbestos fibers fresh frozen lung tumor tissue using the TRI-
(Nymark et al., 2008). Free radicals that trigger ZOLTM reagent (Invitrogen, California), treated
oxidative stress responses may originate from with DNaseI (Ambion, Texas) to remove
direct chemical interactions with the fiber surface genomic contamination and reverse transcribed to
or from macrophages activated during inflamma- cDNA using a two-step reaction. RNA quality
tory processes (Dusinska et al., 2004). It has been was assessed with an Agilent 2100 Bioanalyzer
postulated that generation of excessive amounts (Agilent Technologies, California) and quantified
of reactive oxygen species (ROS) and reactive by Nanodrop (Thermo Scientific, Delaware).
nitrogen species (RNS) result in DNA damage RNA quality for the independent test set was
and chromosomal aberrations (Dusinska et al., assessed by gel electrophoresis and examination
2004; Topinka et al., 2004). of A260/A280 (<1.8 excluded) and A260/A230 (<0.7
Microarray technology has shown that gene excluded) ratios.
expression profiles differ between closely related Microarray experiments were performed
tumor types (Dyrskjot et al., 2003; Komuro et al., according to MIAME guidelines as described pre-
2005). Furthermore, gene expression based classi- viously (Larsen et al., 2007). Total RNA from
fiers predicting recurrence, prognosis, survival, each tumor sample was hybridized with Human
and treatment response have been developed for Universal Reference RNA (Stratagene, California)
a many types of cancer (Schramm et al., 2005; to 22K Human Oligo microarrays printed by the
Larsen et al., 2006; Raponi et al., 2006). The aim British Columbia Gene Array Facility using the
of this study was to investigate whether tumor Operon Human Genome Oligo set v2.0 contain-
gene expression profiles differ between individu- ing 21,329 70-mer probes, representing 14,200
als with evidence of prior asbestos exposure known transcripts. Arrays were scanned using an
(measured by pulmonary asbestos lung fiber Affymetrix GMS418 confocal scanner (Affyme-
counts) and those without. We sought to identify trix, California) and probe quality assessed using
biomarkers of asbestos carcinogenicity in lung Imagene V5.1 (BioDiscovery, California). All data
adenocarcinomas that could be technically and were subject to filtering and Lowess normalization
biologically validated test and in independent procedures using AVADIS bioinformatics software
datasets. (Version 4.3, Strand Genomics, Bangalore, India).
TABLE 1. Demographics for the TPCH Training and Independent Test Set
Of the 21,329 probes on the chip 20,597 passed fil- formed using BRB ArrayTools Version 3.5; devel-
tering criteria and were used in expression analy- oped by Dr Richard Simon and Amy Peng Lam,
sis. All expression data has been deposited in the http://linus.nci.hih.gov/brb/tool.htm. Group com-
National Center for Biotechnology Information parisons, correlations, and associations were per-
(NCBI) Gene Expression Omnibus (GEO Acces- formed using v2 tests and two-tailed t tests where
sion GSE20875). appropriate using SPSS statistical software (Ver-
sion 13, SPSS Inc Chicago, Illinois). A P value less
than 0.05 (two-tailed) was considered statistically
Bioinformatics and Statistical Analysis significant.
Candidate genes were identified by meeting
criteria of (a) statistical significance (P < 0.001)
and (b) magnitude of expression change (fold Gene Ontology and Pathway Analysis
change >1.5). Ontologies were obtained by Gene Over-represented gene ontologies were deter-
Ontology and Pathway Analysis (P < 0.05). Selec- mined using the Expression Analysis Systematic
tion of genes at P < 0.001 (by student t test) Explorer (EASE) Online Interface: Chromosome,
allowed control of the false discovery rate, limit- GO Biological Process, GO Molecular function,
ing the likelihood of selecting false positives. GO Cellular component, and Kyoto Encyclopedia
Hierarchical clustering analysis was performed on of Genes and Genomes (KEGG) pathway (Ver-
differentially expressed genes at the P < 0.001 sion 1.21, http://david.niaid.nih.gov/david/ease.htm).
level using the Pearson absolute distance measure Genes were selected from differential gene ex-
with complete linkage to identify whether dis- pression analysis at the P < 0.05 level (Fisher
tinct gene expression profiles existed between exact test) and compared with all probes showing
ARLC-AC and NARLC-AC. We also used three large variations on the microarray. Relaxation of
independent datasets to test the validity of candi- P value stringency allowed the identification of
date genes in asbestos-related tumors. Statistical significant pathways using a larger number of
analyses for microarray experiments were per- genes to allow visualization of all biologically
Gene Fold
Probe ID Genbank symbol Gene description change Chromosome Map P-value
H200000836 NM_014045 LRP10 DKFZP564C1940 protein 1.806 14 14q11.2 8.36E-04
H200002032 NM_006147 IRF6 Interferon regulatory factor 6 1.671 1 1q32.3-q41 7.35E-04
H200006838 AL133117 THOC-003 Homo sapiens mRNA; cDNA 1.929 X Xq25-q26.3 8.81E-05
DKFZp586L1121 (from
clone DKFZp586L1121)
H200007710 AL050050 DKFZp566D133 DKFZp566D133 protein 1.669 9 14q13.2 2.87E-05
H200007800 NM_004160 PYY Peptide YY 1.512 17 17q21.1 2.24E-04
H200008192 NM_014265 ADAM28 A disintegrin and metallopro- 1.943 8 8p21.2 7.43E-04
teinase domain 28
H200008312 U92980 DT1P1A10 Hypothetical protein 1.559 X Xp11.22 1.35E-04
DT1P1A10
H200008482 NM_002574 PRDX1 Peroxiredoxin 1 1.593 1 1p34.1 1.04E-04
H200011070 NM_006419 SCYB13 Small inducible cytokine B sub- 1.569 4 4q21 3.66E-04
family (Cys-X-Cys motif),
member 13 (B-cell
chemoattractant)
H200013546 NM_022842 FLJ22969 Hypothetical protein FLJ22969 1.684 3 3p21.31 1.13E-05
H200014023 NM_018354 C20orf46 Chromosome 20 open reading 1.707 20 20p13 8.94E-05
frame 46
H200014241 NM_002708 PPP1CA Protein phosphatase 1, cata- 1.705 11 11q13 2.01E-05
lytic subunit, alpha isoform
H200016011 U80113 IGHV4 Human immunoglobulin heavy 1.796 14 14q32.33 7.43E-04
chain variable region (V4–
31) gene, partial cds
H200016279 NM_004283 RAB3D RAB3D, member RAS onco- 1.787 19 19p13.2 6.10E-05
gene family
H200017809 Z00008 IGKV1D-8 Immunoglobulin kappa variable 1.583 2 2 8.36E-04
1D-8
H200018915 AB051436 KIAA1133 Novel C3HC4 type Zinc fin- 1.811 22 22q12.1 8.52E-05
ger (ring finger)
H200019430 AK022017 0 Homo sapiens cDNA 1.602 3 3q13.2 1.75E-05
FLJ11955 fis, clone
HEMBB1000890
H200019567 AK024482 ZBT20 Homo sapiens mRNA for 1.579 7 7p22.2 1.74E-04
FLJ00076 protein, partial
cds
H200019644 NM_052875 MGC10485 Hypothetical protein 1.614 11 11q25 1.05E-03
MGC10485
H200020055 BC011360 LRP16 Homo sapiens, clone 1.771 11 11q11 2.32E-05
IMAGE:4177360, mRNA
H200020854 AK055590 Pseudogene Homo sapiens cDNA 1.829 0 1q32 4.13E-04
FLJ31028 fis, clone
HLUNG2000570, weakly
similar to 40S RIBOSOMAL
PROTEIN S10
The top 20 genes were selected on the basis of high magnitude (>1.5-fold) and statistical significance (P < 0.001). From this list, six high priority
candidates were selected for technical replication and biological validation.
ontologies and networks. The top five networks lism, small molecule biochemistry, and (5) lipid
significantly over-represented (P < 0.05), metabolism, small molecule biochemistry, and
included (1) inflammatory disease, cellular move- molecular transport. Significantly over-repre-
ment, nervous system development and function, sented molecular and cellular functions, included
(2) antigen presentation, cell-mediated immune cellular growth and proliferation, cell morphology,
response, humoral immune response, (3) cellular energy production, molecular transport, and
growth and proliferation, post-translational modi- nucleic acid metabolism. Figure 3 illustrates all
fication, lipid metabolism, (4) DNA replication, functions and pathways over-represented in
recombination and repair, nucleic acid metabo- ARLC-AC.
Figure 2. Bar graph depicting gene expression changes for ZNRF3, group. For the Wikman cell line data, ratios were calculated by divid-
PPP1CA, PRDX1, RAB3D, ADAM28 and IRF6 in TPCH Microarray data, ing the asbestos-treated/control for each cell line and time point.
TPCH qRT-PCR (Training and Testing sets), Nymark Microarray Data TPCH denotes The Prince Charles Hospital. ADAM28 demonstrated
and Wikman cell line gene expression data (A549, BEAS-2B and consistent up-regulation of gene expression across all datasets.
Met5A). Fold changes are the mean of ARLC-AC/mean NARLC-AC
Figure 3. Ingenuity Pathway Analysis of differentially expressed genes identified for ARLC-AC at P <
0.05. Bar graph highlights significantly over-represented functions in asbestos-related lung adenocarcinoma.
[Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]
tumors, notably 2p16, 9q33.1, and 19p (Nymark types, and methods of prior asbestos exposure
et al., 2006, 2009; Kettunen et al., 2009). In the assessment. While we investigated differences in
current study only one of our candidates was the gene expression profiles of moderately
found in a region previously associated with copy exposed (> ¼ 20 AB/gww) and nonexposed pri-
number loss (RAB3D, 19p13.2) suggesting that mary lung ACs, the Wikman group compared
the altered expression of candidates identified in subjects with high occupational exposures and
this study are not necessarily driven by copy nonexposed subjects with lung cancer patients of
number alterations. A recent review by Nymark mixed histologies. This and most other asbestos
et al., (2008), provides an excellent overview of exposed cohorts who develop lung cancer have
the many molecular and genetic changes in also smoked tobacco. Thus comparative studies
asbestos-related lung cancer. In a comparison of are effectively comparing tumor expression pro-
matched normal and lung tumor tissue, Ruosaari files of asbestos and tobacco exposed subjects
et al., (2008) have identified several functional with tobacco only exposed subjects. Whether
pathways significantly different between asbestos- ADAM28 upregulation would also occur in
exposed and nonexposed lung cancer patients tobacco treated cells is not known. It is possible
including ion transport, NF-jB signaling, DNA that combined exposure to both asbestos and
repair, and spliceosome and nucleosome com- tobacco increases the expression of ADAM28
plexes. Furthermore, many pathways downregu- more so, than exposure to either carcinogen
lated in asbestos-exposed subjects were related to alone. Further functional experiments are
protein ubiquination, involved in multiple cell required to demonstrate whether tobacco, asbes-
processes, including cell cycling, apoptosis, and tos or tobacco plus asbestos induce the greatest
DNA repair. Consistent with previous investiga- change in ADAM28 expression. Finally, to our
tions, we also found over-representation of NF- knowledge there are no publicly available data-
jB related gene ontologies. sets investigating ADAM28 expression in lung ep-
A limitation of this study is the potential for ithelium or preneoplasia from ARLC-AC and
Type I errors. Primary selection by P value NARLC-AC making it difficult to investigate
allows control of the false discovery rate (Benja- whether alteration of ADAM28 expression is an
mini and Hochberg, 1995). By combining this early event in ARLC-AC. Further studies may be
with secondary selection of those genes whose required to determine whether ADAM28 expres-
expression differed between the groups by great- sion is altered in early preneoplastic lesions.
est magnitude, our candidate identification pro- In conclusion, we have identified ADAM28 as a
cess incorporated essential elements of statistical potential asbestos-related cancer biomarker with
and biological significance (Chen et al., 2007). To consistent upregulation of ADAM28 across in-
focus down to a manageable number of candi- house and independent datasets. ADAM28 has
dates, we prioritized on biological functions that previously been implicated as an oncogene in
seemed most relevant to differences in tumor NSCLC (Ohtsuka, et al., 2006) with a role in cell
biology between ARLC-AC and NARLC-AC. proliferation and metastases (Okada, 2007). Gene
Nonetheless, some genes excluded on ontological expression changes observed for ADAM28 could
groups could nevertheless be relevant. be due to asbestos exposure independent of
Another potential limitation is the relative size tobacco smoke exposure but additional experi-
of the training set (N ¼ 36). Increasing the sam- ments are needed to confirm this. Further func-
ple size of this study would lend greater power to tional studies, including transfection experiments
our analysis and perhaps avoid overfitting to the and immunohistochemistry (IHC) are required to
particular tumors within the sample; however, confirm that protein expression parallels mRNA
microarray technology is limited by cost and tech- expression, although the lack of a commercial
nical requirements. A larger sample size may antibody verified for use in formalin-fixed paraf-
have permitted further lowering of the false dis- fin-embedded tissue and relatively modest fold
covery rate such that the differentially expressed change differences pose a challenge for accurate
gene list contained fewer falsely declared genes. quantification of protein staining by IHC.
To our knowledge only one public dataset of
lung cancers with annotated asbestos exposure or
lung fiber burden information is available (Wik- ACKNOWLEDGMENTS
man et al., 2007,) and this is limited by its rela- The authors acknowledge Harriet Wikman
tively small sample numbers, mixed histology (from The Finnish Institute of Occupational
Okada Y. 2007. Modulation of the microenvironment and adhe- Schramm A, Schulte JH, Klein-Hitpass L, Havers W, Sieverts H,
sion of cancer cells by ADAMs (a disintegrin and metalloprotei- Berwanger B, Christiansen H, Warnat P, Brors B, Eils J, et al.
nase). Verh Dtsch Ges Pathol 91:29–38. 2005. Prediction of clinical outcome and biological characteriza-
Okayasu R, Takahashi S, Yamada S, Hei TK, Ullrich RL. 1999. tion of neuroblastoma by expression profiling. Oncogene
Asbestos and DNA double strand breaks. Cancer Res 59:298– 24:7902–7912.
300. Sehgal A, Keener C, Boynton AL, Young RF, Vermeulen SS,
Powis G, Montfort W. 2001. Properties and biological activities of Yonemura KS, Kohler EP, Aldape HC, Simrell CR, Murphy
thioredoxins. Annu Rev Pharmacol Toxicol 41:261–295. GP. 1997. Isolation and characterization of a novel gene from
Rangasamy T, Cho C, Thimmulappa R, Zhen L, Srisuma S, Kensler human glioblastoma multiforme tumor tissue. Int J Cancer
T, Yanamoto M, Petrache I, Tuder R, Biswal S. 2004. Genetic 71:565–72.
ablation of Nrf2 enhances susceptibility to cigarette smoke- Sun S, Schiller JH, Gazdar AF. 2007. Lung cancer in never smok-
induced emphysema in mice. J Clin Invest 114:1248–1259. ers–A different disease. Nat Rev Cancer 7:778–790.
Raponi M, Zhang Y, Yu J, Chen G, Lee G, Taylor JM, Macdonald Topinka J, Loli P, Georgiadis P, Dusinska M, Hurbankova M,
J, Thomas D, Moskaluk C, Wang Y, Beer DG. 2006. Gene Kovacikova Z, Volkovova K, Kazimirova A, Barancokova M,
expression signatures for predicting prognosis of squamous cell Tatrai E, et al. 2004. Mutagenesis by asbestos in the lung of
and adenocarcinomas of the lung. Cancer Res 66:7466–7472. lambda-lacI transgenic rats. Mutat Res 553:67–78.
Roemer A, Schwettmann L, Jung M, Roigas J, Kristiansen G, Schnorr Vainio H, Boffetta P. 1994. Mechanisms of the combined effect of
D, Loening SA, Jung K, Lichtinghagen R. 2004. Increased mRNA asbestos and smoking in the etiology of lung cancer. Scand J
expression of ADAMs in renal cell carcinoma and their association Work Environ Health 20:235–242.
with clinical outcome. Oncol Rep 11:529–536. Wang X, Liu J, Zhang H, Xiao M, Li J, Yang C, Lin X, Wu Z,
Ruosaari S, Hienonen-Kempas T, Puustinen A, Sarhadi VK, Holl- Hu L, Kong X. 2003. Novel mutations in the IRF6 gene for
men J, Knuutila S, Saharinen J, Wikman H, Anttila S. 2008. Van der Woude syndrome. Hum Genet 113:382–386.
Pathways affected by asbestos exposure in normal and tumour Wikman H, Ruosaari S, Nymark P, Sarhadi VK, Saharinen J, Van-
tissue of lung cancer patients. BMC Med Genom 1:55–63. hala E, Karjalainen A, Hollmen J, Knuutila S, Anttila S. 2007.
Roggli VL. 1992. Tissue Digestion Techniques in Pathology of Gene expression and copy number profiling suggests the impor-
Asbestos-Associated Diseases. Boston, Massachusetts: Little tance of allelic imbalance in 19p in asbestos-associated lung
Brown and Co. pp. 402–410. cancer. Oncogene 26:4730–4737.
Roggli VL, McGavran MH, Subach J, Sybers HD, Greenberg SD. Winter SL, Bosnoyan-Collins L, Pinnaduwage D, Andrulis IL.
1982. Pulmonary asbestos body counts and electron probe anal- 2007. The interaction of PP1 with BRCA1 and analysis of their
ysis of asbestos body cores in patients with mesothelioma: A expression in breast tumors. BMC Cancer 7:85.
study of 25 cases. Cancer 50:2423–32. Wright CM, Bowman RV, Tan ME, Martins MU, McLachlan RE,
Ruiz L, Traskine M, Ferrer I, Castro E, Leal JF, Kaufman M, Passmore LH, Windsor MN, Clarke BE, Duhig EE, Mahar
Carnero A. 2008. Characterization of the p53 response to onco- AM, et al. 2008. Lung asbestos content in lungs resected for
gene-induced senescence. PLoS One 3:e3230. primary lung cancer. J Thorac Oncol 3:569–576.