ONCOGENOMICS Deaths in the United States and Other Developed 2000)

Identi®cation of genes over-expressed in small cell lung carcinoma using suppression subtractive hybridization and cDNA microarray expression analysis

Chaitanya S Bangur*,1, Ann Switzer1, Liqun Fan1, Matthew J Marton2, Michael R Meyer2 and Tongtong Wang1

1Tumor Antigen Discovery, Corixa Corporation, 1124 Columbia Street, Seattle, Washington WA 98104, USA; 2Rosetta Inpharmatics, 12040 115th Avenue NE, Kirkland, Washington, WA 98034, USA

To identify genes that are dierentially over-expressed in rate for patients diagnosed with NSCLC is 10 ± 15%, Small Cell Lung Carcinoma (SCLC) we have used a while patients diagnosed with SCLC have an even combination of suppression subtractive hybridization and worse prognosis with a 5-year survival rate of 55% cDNA microarray to analyse the expression pro®les of (Ginsberg et al., 1997; Ihde et al., 1997). Although, 2400 cDNAs clones. Genes that are over-expressed in SCLC is initially highly responsive to radiation and SCLC were identi®ed using 32 pairs of ¯uorescence- chemotherapy (50 ± 90%), most patients tend to relapse labeled cDNA samples representing various lung tumors with highly resistant disease within a year of treatment and normal tissues. This comprehensive approach has (Ihde et al., 1997). Most patients diagnosed with SCLC resulted in the identi®cation of 209 genes that are have regional lymph-node involvement or disseminated dierentially over-expressed in SCLC. Quantitative real- disease at the time of initial diagnosis (Ihde, 1992), time PCR was used to further validate the expression of which strongly correlates with the poor prognosis 43 genes in SCLC tumors and various normal tissues. observed for these patients. Thus, the problem with Discussed in this report are nine genes, which showed the the management of lung cancer can be attributed to the most promising SCLC tumor to normal tissue dierential lack of reliable early detection methods and the expression pro®les, including seven known and two novel inadequacies of current treatment protocols. This has genes. The large number of dierentially expressed genes warranted the eorts towards identi®cation of new identi®ed from this analysis and the characterization of molecular targets for designing more reliable diagnostic these genes will provide valuable information in better methods and novel therapeutic protocols for the understanding the biology of SCLC and help us in treatment of this malignancy. developing these gene products as potential targets for SCLC tumors are distinguished by their neuroendo- diagnostic as well as therapeutic usage. crine phenotype and frequently stain for neuroendo- Oncogene (2002) 21, 3814 ± 3825. DOI: 10.1038/sj/ crine markers such as neuron-speci®c enolase, onc/1205480 chromogranin, and synsptophysin (Guinee et al., 1994). SCLC cells are also found to secrete several Keywords: small cell lung carcinoma (SCLC); micro- autocrine growth factors and hormones including array; subtractive hybridization; real-time PCR; gene gastrin-releasing peptide, neurotensine, vasopressin, expression and cholecystokinin (Kalemkerian, 2000; Williams, 1997). Some of the autocrine growth factors have been used as prognostic markers of this disease with varying Introduction levels of success and reliability (Feld et al., 2000). However, their use as diagnostic and therapeutic Lung cancer is the most common cause of cancer targets has not been very successful (Kalemkerian, ONCOGENOMICS deaths in the United States and other developed 2000). Thus the need of new reliable and eective countries. It accounts for approximately 30% of all targets for the early detection and management of this cancer deaths and in the United States alone 4150 000 malignancy is more than ever. It has been well people die of lung cancer every year (Parker et al., documented that cancer cells go through a lot of 1997). Lung tumors are divided into two major types, genetic and epigenetic changes (Lengauer et al., 1998; NSCLC and SCLC. 80% of lung cancer incidences are Williams, 1997), which is suggested to bring about attributed to the NSCLC type, while SCLC accounts dramatic changes in the gene expression pro®les of for the rest 20% of cases. The overall 5-year survival these cells. The past decade has seen the development of several technologies for pro®ling these changes in gene expression, including dierential and subtractive hybridization, dierential display, SAGE, and micro- *Correspondence: CS Bangur; E-mail: [email protected] Received 11 January 2002; revised 6 March 2002; accepted 18 arrays (Gray and Collins, 2000). These approaches March 2003 have proven to be extremely useful in providing a SCLC gene expression profiling CS Bangur et al 3815 comprehensive look at the biology of various cancers as well as in the identi®cation of new targets for the development of novel therapeutic approaches for the treatment of these cancers (Backert et al., 1999; Nacht et al., 1999; Nocito et al., 2001; Ono et al., 2000; Sgroi et al., 1999; Wang et al., 2000; Xu et al., 2000; Yang et al., 1999). Recent studies have documented a large number of genetic alterations present in SCLC (Girard Figure 1 Subtraction eciency as determined by depletion of et al., 2000; Wistuba et al., 2000a,b), which is GAPDH from the subtracted cDNA population. Equal amounts indicative of possible changes in the gene expression of subtracted and unsubtracted cDNAs were subjected to PCR pro®le. However, little has been done in exploring ampli®cation for the indicated number of cycles using GAPDH these changes in SCLC and the only study reported to gene speci®c primers. PCR products were run on a 1% agarose gel and stained with ethidium bromide. M, molecular weight date was very limited in scope and was designed to marker address the dierences between the SCLC subtypes at the molecular level (Anbazhagan et al., 1999). This study was undertaken to carryout a comprehensive analysis of the changes in gene expression in unsubtracted cDNAs, GAPDH speci®c PCR product SCLC and was speci®cally focused towards identi®ca- is visible by the 23rd round of ampli®cation. However, tion of genes that are dierentially over-expressed in in case of the subtracted cDNAs, the GAPDH speci®c SCLC. To accomplish this we used a combination of PCR product is visible only after 33 rounds of technologies including suppression subtractive hybridi- ampli®cation. This is indicative of preferential deple- zation (SSH) for cDNA subtraction (Diatchenko et al., tion of GAPDH and most ubiquitously expressed 1996), high-density cDNA microarray and quantitative genes from the subtracted cDNA mixture. Sequence real-time RT ± PCR. Several recent reports have shown analysis of 48 ± 96 randomly picked clones gave us a the successful use of similar technologies in identifying good estimation of the complexity of the subtracted novel and previously unknown genes found to be libraries. Generally the libraries generated by the dierentially expressed in breast, prostate, head and suppression subtractive hybridization method are of neck, and lung squamouscell carcinoma (Villaret et al., high complexity because of the normalization of highly 2000; Wang et al., 2000; Xu et al., 2000; Yang et al., abundant messages during hybridization. However, in 1999). Here we report the identi®cation of 209 genes case of the SCL2 subtracted cDNA library, we that are dierentially over-expressed in SCLC com- repeatedly recovered cDNAs for GRP and ASH1, pared to normal tissues and the initial characterization which are known to be over-expressed in SCLC (Ball et of nine of these genes. Further characterization of these al., 1993; Yamaguchi et al., 1983). Approximately 30% nine genes and other genes identi®ed in this study will of the clones from this library contained cDNA inserts be helpful in de®ning a panel of diagnostic and representing either of the two genes (data not shown). therapeutic targets for SCLC and better our under- The repeated recovery of these genes was an indication standing into the biology of this malignancy. of successful subtraction for enriching genes dierentially expressed in SCLC. However, it also meant that the SCL2 library had a set of highly redundant clones and was not complex enough to be screened further by Results microarray. To overcome this problem, two additional subtracted libraries, SCL3 and SCL4, were constructed Generation and characterization of SCLC specific cDNA using a driver cDNA pool for hybridization that libraries included the cDNAs for GRP and ASH1 in addition To enrich for genes preferentially expressed in SCLC, to the nine normal tissues used for the SCL2 library we generated four SCLC tumor speci®c cDNA libraries (see Materials and methods). The inclusion of the GRP using the suppression subtractive hybridization method and ASH1 cDNAs in the driver pool decreased the (Diatchenko et al., 1996). These subtracted libraries are percentage of clones that represented these two genes referred to as SCL1, SCL2, SCL3 and SCL4 (for to around 5% in the SCL3 and SCL4 libraries. Thus details, see materials and methods). The initial by spiking in the cDNAs for the two highly abundant characterization of these subtracted libraries was done genes we were able to generate libraries that were of by estimating the eciency of subtraction and sucient complexity that could be exploited further by sequencing of 48 ± 96 randomly picked clones from cDNA microarray analysis. each library. The eciency of subtraction was determined by comparing the abundance of glycer- cDNA microarray analysis aldehyde-3-phosphate dehydrogenase (GAPDH) cDNAs before and after subtraction. GAPDH is a Twenty-four thousand randomly selected cDNA clones good representative for constitutively expressed genes, from SCL1, SCL3, and SCL4 subtracted libraries were which are highly abundant in most tissues. Figure 1 PCR ampli®ed and arrayed onto 32 replicate glass shows the depletion of GAPDH from the subtracted slides (microarrays). Each microarray was then cDNA mixture for the SCL3 library. For the analysed with a pair of Cy3 and Cy5 ¯uorescent dye

Oncogene SCLC gene expression profiling CS Bangur et al 3816 labeled sample to determine the expression pro®les of the cDNA clones in the tumor and normal tissue represented by the labeled samples. In all 32 labeled sample pairs (Table 1) were used to get a comprehensive picture of the expression of the cDNAs in a variety of tumor and normal tissues. Microarray fabrication, generation of the Cy3 and Cy5 labeled cDNA samples, hybridization, imaging, quantitation and normalization of the microarray expression data was performed by Rosetta Inpharmatics (Kirkland, WA, USA) (see Material and methods). To identify genes that are dierentially over-expressed in SCLC the mean ¯uorescent intensity of a given cDNA clone in the eight SCLC samples was compared with its mean ¯uorescent intensity in all the normal tissue samples. Figure 2 Figure 2 Scatter plots of cDNA microarray analysis. Mean illustrates the expression pro®les of all the cDNA ¯uorescence intensities of the 2400 cDNA clones in the eight clones on the chip, where the mean ¯uorescent SCLC samples (SCLC sample group) were plotted against their intensity of the eight SCLC samples was compared mean ¯uorescence intensities in the 32 normal tissue samples with the mean ¯uorescent intensity of the thirty-two (normal tissue sample group). The cDNA clones within the normal tissue samples. Clones that have the highest encircled region are dierentially over-expressed in SCLC. The lower dotted line indicates that the mean ¯uorescence intensity in likelihood of being dierentially over-expressed in SCLC sample group is twofold greater than the mean ¯uorescence SCLC are indicated to be within the encircled area. intensity in the normal tissue sample group and vice versa for the In all 490 cDNA clones with a ratio of two or more upper dotted line between the SCLC and normal tissue samples were identi®ed and sequenced for further analysis. Sequence the identity and the number of time each gene was analysis showed that the 490 clones formed 209 unique recovered by the microarray analysis. cDNA contigs, possibly representing 209 unique genes. Of the 209 cDNA contigs 123 were found to be Identification of cDNAs differentially expressed in small previously known genes and the remaining 86 cell lung cancer represented novel gene sequences. Table 2 summarizes The objective of this study was to identify genes that are not only dierentially over-expressed in SCLC but Table 1 Flourescent-labeled samples used for microarray analyses also have very little to no expression in majority of sample pair Cy3 Labeled Sample Cy5 Labeled Sample normal tissues. To identify such genes we visually examined the expression of each of the 209 genes, 1 Normal Lung 1 Adenocarcinoma 1 2 Adenocarcinoma 1 Normal Lung 1 identi®ed from the initial analysis, in each tumor and 3 Adenocarcinoma 1 Normal Lung 1 normal tissue samples tested. We looked for genes that 4 Adenocarcinoma 2 Normal Heart showed reasonable expression in SCLC samples while 5 Adenocarcinoma 3 Normal Liver having very little to no signal in majority of the normal 6 Adenocarcinoma 4 Normal Lung 2 7 Adenocarcinoma 5 Normal Skeletal Muscle tissue samples. Based on these criteria, we were able to 8 Adenocarcinoma 6 Normal Spleen come up with 43 genes for further characterization. 9 Adenocarcinoma 7 Normal Stomach Here we are reporting the characterization of nine 10 Adenocarcinoma 8 PBMC (Activated T Cells) genes with the most promising SCLC tumor to normal 11 Adenocarcinoma ± LPE1 Normal Bladder tissue expression pro®le, these include seven known 12 Adenocarcinoma ± LPE2 Normal Bone Marrow 13 Squamous Cell Carcinoma 1 Normal Bronchus and two novel genes. The identity of the nine genes 14 Squamous Cell Carcinoma 2 Normal Esophagus with their corresponding candidate designation is 15 Squamous Cell Carcinoma 3 Normal Kidney summarized in Table 3. Also, listed is the number of 16 Squamous Cell Carcinoma 4 Normal Lung 3 times each of these genes was recovered from the 17 Squamous Cell Carcinoma 5 PBMC (Activated B Cells) 18 Squamous Cell Carcinoma 6 PBMC (Resting) microarray analysis and the fold over-expression of 19 Squamous Cell Carcinoma 7 Normal Skin these genes in SCLC tumors compared to normal tissue 20 Squamous Cell Carcinoma 8 Normal Thymus as determined by both microarray and quantitative 21 Squamous Cell Carcinoma 9 Normal Tonsil real-time RT ± PCR analysis. 22 Squamous Cell Carcinoma 10 Normal Trachea 23 Atypical Carcinoid (METs) Normal Lymph Node 24 SCLC ± Primary 1 Normal Lung 4 Microarray expression profile of genes over-expressed in 25 SCLC ± Primary 2 Normal Pituitary Gland SCLC 26 SCLC Cell Line (NCI H69) Normal Adrenal Gland 27 SCLC Cell Line (HTB 175) Normal Pancreas The microarray expression pro®les of the nine genes 28 SCLC Cell Line (DMS 79) Normal Spinal Cord 29 SCLC Cell Line (HTB 171) Normal Thyroid Gland described in Table 3 are shown in Figure 3, with the 30 SCLC Cell Line (NCI H128) Normal Brain Cy3 and Cy5 labeled sample pairs listed on the left of 31 Colon Carcinoma 1 Normal Colon each panel (also refer to Table 1). The expression level 32 Colon Carcinoma 2 Normal Small Intestine of each gene in a given sample is shown as pseudo

Oncogene Table 2 Genes that are over-expressed in SCLC Recovery GenBank Recovery GenBank frequency Ratio* GenBank identity Acc. No. frequency Ratio* GenBank identity Acc. No. 48 8.08 Homo sapiens achaete-scute complex homolog-like 1 L08424 1 2.14 Homo sapiens CAGF28 U80735 36 8.86 Homo sapiens gastrin-releasing peptide K02054 1 2.44 Homo sapiens casein kinase I alpha isoform L37042 31 5.30 Homo sapiens SOX4 X70683 1 3.11 Homo sapiens cDNA clone YB34C04 AF14734 18 3.58 Homo sapiens carcinoembryonic antigen M37398 1 2.57 Homo sapiens cDNA DKFZp434M0326 AL137681 10 2.34 Homo sapiens cDNA FLJ20136 AK000143 1 7.30 Homo sapiens cDNA DKFZp434N1435 AL133574 10 5.65 Homo sapiens DNA topoisomerase II J04088 1 2.76 Homo sapiens cDNA DKFZp564P046 AL049339 8 9.33 Homo sapiens Pr22 protein Z11566 1 5.17 Homo sapiens cDNA FLJ10099 AK000961 8 5.63 Homo sapiens similar to ubiquitin-conjugating enzyme E2 AF161499 1 2.88 Homo sapiens cDNA FLJ11003 AK001865 7 8.63 Homo sapiens pituitary tumor transforming gene protein 1 AF095287 1 2.14 Homo sapiens cDNA FLJ11494 AK021556 6 2.48 ESTs BE094484 1 2.18 Homo sapiens cDNA FLJ14660 AK027566 5 2.60 Homo sapiens methionine adenosyltransferase II, alpha L43509 1 4.52 Homo sapiens cDNA FLJ20259 AK000266 5 2.30 Homo sapiens monokine induced by gamma interferon X72755 1 2.12 Homo sapiens cDNA FLJ20916 AK024569 5 2.44 Homo sapiens PAK-interacting exchange factor beta NM_003899 1 2.12 Homo sapiens cDNA FLJ21772 AK025425 4 2.42 Homo sapiens apoptosis-related protein TFAR15 AF022385 1 3.27 Homo sapiens cDNA FLJ22272 AK025925 4 3.63 Homo sapiens coactosin-like protein L54057 1 2.30 Homo sapiens CGI-204 mRNA AF285120 4 2.73 Homo sapiens dihydropyrimidinase related protein-3 D78014 1 2.06 Homo sapiens chemokine receptor CXCR4 AF052572 4 2.45 Homo sapiens gene amplified in osteosarcoma (OS4) AF000152 1 3.28 Homo sapiens CHL1 potential helicase U75967 4 3.74 Homo sapiens histone H2B.1 M60751 1 3.46 Homo sapiens cisplatin resistance related protein CRR9p XM_034732 4 2.92 Homo sapiens hnRNP type A/B M65028 1 4.44 Homo sapiens claudin 1 AF101051 4 2.40 Homo sapiens hnRNP-D like AB017018 1 3.59 Homo sapiens coatomer protein complex, subunit beta 2 X70476 4 2.91 Homo sapiens lactate dehydrogenase-A U13679 1 2.41 Homo sapiens COBW-like protein AF257330 4 3.97 Homo sapiens voltage-gated sodium channel, type III alpha XM_033673 1 2.18 Homo sapiens cyclin E2 AF091433 4 3.20 Homo sapiens SOX2 Z31560 1 4.57 Homo sapiens cyclin-dependent kinase inhibitor p18 AF041248 4 2.38 Novel cDNA 1 4.91 Homo sapiens DEK oncogene X64229 3 2.09 ESTs BE548287 1 2.09 Homo sapiens DNA polymerase alpha-subunit X06745 Bangur CS profiling expression gene SCLC 3 2.41 ESTs AW850519 1 2.23 Homo sapiens DNA-binding protein B M24070 3 2.44 ESTs BG288243 1 2.03 Homo sapiens E-1 enzyme AF113125

3 3.26 ESTs AW956024 1 2.26 Homo sapiens ERK activator kinase L11285 al et 3 2.74 ESTs AA774715 1 4.30 Homo sapiens F-box only protein 9 NM_012347 3 2.92 ESTs AW295665 1 2.09 Homo sapiens growth-arrest-specific protein L13720 3 2.22 Homo sapiens CGI-89 protein AF151847 1 2.25 Homo sapiens GTT1 protein NM_020151 3 2.97 Homo sapiens guanylate binding protein isoform I M55542 1 2.27 Homo sapiens guanine nucleotide regulatory protein U01147 3 5.94 Homo sapiens mesoderm specific transcript homolog (MEST) D87367 1 2.46 Homo sapiens heart-type calpastatin AB026049 3 2.29 Homo sapiens neuronal cell adhesion molecule AJ001054 1 2.58 Homo sapiens histone macroH2A1.1 AF054174 3 3.51 Homo sapiens P311 protein U30521 1 3.53 Homo sapiens hnRNP-E2 X78136 3 2.97 Homo sapiens STAT1 NM_007315 1 3.33 Homo sapiens HSP40/DNAJ homologs AB028859 3 2.65 Homo sapiens tumor protein D52 NM_005079 1 5.35 Homo sapiens immunoglobulin kappa heavy chain Y14735 3 2.29 Novel cDNA 1 2.46 Homo sapiens JKTBP2, JKBP1 AB017018 2 2.04 ESTs BE173915 1 2.22 Homo sapiens karyopherin (importin) beta 1 L38951 2 2.57 ESTs BG031569 1 2.23 Homo sapiens KIAA0038 D26068 2 2.49 ESTs BE787315 1 4.26 Homo sapiens KIAA0101 NM_014736 2 2.78 ESTs BG431541 1 3.16 Homo sapiens KIAA0166 NM_014708 2 2.28 ESTs AI024509 1 2.17 Homo sapiens KIAA0430 NM_019081 2 3.89 ESTs AA974984 1 2.73 Homo sapiens KIAA0546 XM_049055 2 4.32 ESTs AW956024 1 2.35 Homo sapiens KIAA0632 NM_015545 2 2.79 Homo sapiens alcohol dehydrogenase 7 X76342 1 2.13 Homo sapiens KIAA0689 NM_015235 2 2.01 Homo sapiens calpastatin D16217 1 2.37 Homo sapiens KIAA1335 XM_029763 2 2.02 Homo sapiens cathepsin B L16510 1 2.95 Homo sapiens KIAA1724 XM_040280 2 3.59 Homo sapiens CDK4-inhibitor p16-INK4 L27211 1 7.73 Homo sapiens LAG-3 X51985 Continued Oncogene 3817 Oncogene 3818

Table 2 (Continued ) Recovery GenBank Recovery GenBank frequency Ratio* GenBank identity Acc. No. frequency Ratio* GenBank identity Acc. No. 2 2.65 Homo sapiens cDNA DKFZp434B0920 AL137728 1 3.31 Homo sapiens Mad2B protein AF139365 2 3.07 Homo sapiens cDNA DKFZp434M1317 AL133074 1 2.64 Homo sapiens matrin 3 AF117236 2 2.43 Homo sapiens cDNA DKFZp564D173 AL110212 1 2.42 Homo sapiens matrix metalloproteinase 12 L23808 2 2.86 Homo sapiens cDNA DKFZp564O163 AL110216 1 2.30 Homo sapiens membrane-type serine protease 1 AF133086 2 2.04 Homo sapiens cDNA DKFZp586I0521 AL137567 1 2.69 Homo sapiens microsomal signal peptidase subunit AAK14919 2 2.70 Homo sapiens cDNA FLJ12953 AK023015 1 3.51 Homo sapiens mitogen-activated protein kinase 6 NM_002748 2 2.46 Homo sapiens cDNA FLJ21813 AK025466 1 5.49 Homo sapiens MYT1 kinase U56816 2 9.67 Homo sapiens CRMP-1 D78012 1 3.26 Homo sapiens neuronal pentraxin 1 U61849 2 2.90 Homo sapiens cyclin-dependent kinase 2 M68520 1 2.57 Homo sapiens NRAS-related gene NM_007158 2 2.73 Homo sapiens deleted in polyposis locus M73547 1 2.25 Homo sapiens nuclear receptor co-repressor/HDAC3 complex XM_028239 2 2.53 Homo sapiens DNA topoisomerase II binding protein AB019397 1 2.53 Homo sapiens nucleoporin-like protein 1 X89478

2 2.45 Homo sapiens dopa decarboxylase M76180 1 2.73 Homo sapiens nucleosome assembly protein 1-like 1 AB027013 profiling expression gene SCLC 2 2.52 Homo sapiens elongation factor-1 gamma Z11531 1 6.92 Homo sapiens NUF2R AF326731 2 2.35 Homo sapiens fibronectin AF169675 1 2.47 Homo sapiens oriP binding protein L29096 2 2.27 Homo sapiens glutamine synthetase pseudogene U08626 1 2.10 Homo sapiens PAPS synthetase AF097721 2 2.43 Homo sapiens guanine nucleotide binding protein, beta 1 NM_002074 1 3.52 Homo sapiens PEG10 XM_029267 2 2.15 Homo sapiens membrane protein-like protein U21556 1 2.05 Homo sapiens polyamine modulated factor-1 AF141310

2 2.43 Homo sapiens MS4A8B NM_031457 1 2.54 Homo sapiens presenilins associated rhomboid-like protein AF197937 Bangur CS 2 2.60 Homo sapiens NADH:ubiquinone oxidoreductase B17 subunit AF035840 1 3.61 Homo sapiens proteasome subunit, alpha type, 2 D00760 2 2.12 Homo sapiens neuronal protein NP25 AF112201 1 2.95 Homo sapiens SEC61 alpha subunit isoform 1 AF084458 2 3.31 Homo sapiens Pumilio 2 NM_014676 1 2.18 Homo sapiens rab geranylgeranyl transferase, alpha-subunit Y08200 2 2.57 Homo sapiens SMC4L1 XM_003062 1 2.54 Homo sapiens radixin L02320 al et 2 4.76 Homo sapiens U5 snRNP 100 kD protein AF026402 1 2.64 Homo sapiens retinoblastoma susceptibility protein RB1 M28419 2 2.41 Homo sapiens ubiquitin specific protease 14 NM_005151 1 2.04 Homo sapiens ribonucleoprotein particle C M16342 2 2.00 Novel cDNA 1 3.33 Homo sapiens RNA-binding protein BRUNOL2 AF248648 1 3.12 ESTs R11971 1 2.26 Homo sapiens Sad1 unc-84 domain protein 1 AF202724 1 2.33 ESTs AW972155 1 2.46 Homo sapiens secretagogin Y16752 1 3.23 ESTs BF507964 1 2.47 Homo sapiens serine carboxypeptidase 1 precursor protein AF282618 1 2.43 ESTs BG403577 1 5.00 Homo sapiens solute carrier family 2, member 1 XM_046330 1 2.41 ESTs AL135742 1 2.13 Homo sapiens SOX21 AF107044 1 2.23 ESTs BE815960 1 2.48 Homo sapiens splicing factor HCC1 L10910 1 2.33 ESTs BG775873 1 2.83 Homo sapiens splicing factor, arginine/serine-rich 2 NM_003016 1 2.14 ESTs BE539395 1 2.38 Homo sapiens splicing factor, arginine/serine-rich 9 U30825 1 2.44 ESTs BE779237 1 4.27 Homo sapiens topoisomerase-related function protein TRF4-1 AF089896 1 2.15 ESTs BG180508 1 2.08 Homo sapiens TRAF4-associated factor 2 U83194 1 2.05 ESTs AW937833 1 2.29 Homo sapiens UBX domain-containing 1 NM_025241 1 2.45 ESTs BF206211 1 2.22 Mus musculus synaptotagmin-like 3 NM_031395 1 2.14 ESTs BG547204 1 2.14 Novel cDNA 1 3.31 ESTs BE748520 1 2.15 Novel cDNA 1 2.18 ESTs BG575084 1 3.31 Novel cDNA 1 2.31 ESTs AW172964 1 2.33 Novel cDNA 1 3.54 ESTs AA287338 1 2.65 Novel cDNA 1 2.24 ESTs AW978427 1 5.10 Novel cDNA 1 2.63 ESTs AW023845 1 3.83 Novel cDNA 1 2.29 ESTs AW151275 1 2.05 Novel cDNA 1 6.43 ESTs BF837415 1 2.17 Novel cDNA 1 2.06 Homo sapiens adenylate kinase 2B U54645 1 2.49 Novel cDNA 1 2.46 Homo sapiens alpha-methylacyl-CoA racemase AF158378 1 2.17 Novel cDNA 1 2.03 Homo sapiens APM-1 Y14591 1 2.00 Novel cDNA Continued SCLC gene expression profiling CS Bangur et al 3819 color image. Also, listed are the fold over-expression values for sample pairs where it was observed that the given gene was over-expressed by two-fold or more in GenBank the SCLC tumor sample (Cy3) compared to the corresponding normal tissue sample (Cy5). It is evident from the microarray images that these genes are over- expressed in SCLC tumors compared to majority of the normal tissues. In some cases expression is also seen in adenocarcinoma and/or squamous cell carcinoma. L978P, L580S and L1435P are members of the SOX family genes (Wegner, 1999). Although, they belong to the same gene family their expression levels and

the ration listed is the sum average of all distribution in lung tumors are very dierent (Figure 3). L978P is highly dierentially expressed not only in SCLC but also in lung adenocarcinoma and squamous cell carcinoma. L580S and L1435P on the other hand are expressed only in a select few SCLC and squamous cell carcinomas. L1437P, L1438P, and L1439P are highly over-expressed in majority of the SCLC samples (Figure 3). Detectable expression of these genes can also be seen in various hematopoetic tissues especially activated T-cells and bone marrow. L1437P is also expressed in normal stomach, while L1438P expression can be observed in normal tonsil (Figure 3). L986P and L1423P are not only expressed in SCLC tumors but also in the metastatic atypical carcinoid tumor, while L1424P expression is seen only in the metastatic atypical carcinoid tumor. L986P and L1423S expression can also be detected in normal neurological tissues such as brain and pituitary gland, this is not surprising as SCLC are classi®ed as neuroendocrine tumors and are known to express several neuroendocrine markers.

GenBank Recovery Quantitative real-time PCR analysis For those genes that were identi®ed as being over- expressed in SCLC by microarray expression pro®ling, further analysis was performed using quantitative real- time RT ± PCR analysis. This was done to con®rm the expression pro®les of the genes by an independent method and to get additional quantitative information regarding the expression of these genes in both tumor and a variety of normal tissues. The real-time RT ± PCR analysis was carried out using a panel of SCLC primary tumor and cell lines, a metastatic atypical carcinoid, pools of adenocarcinoma and squamouscell carcinoma, normal lung, and other normal tissues (see legend for Figure 4). As evident from the results of the real-time RT ± PCR analysis (Figure 4) the overall expression pro®les for these genes correlated well with their microarray expression pro®les (Figure 3). Similar to the microarray expression data L978P is expressed in all of the SCLC samples, atypical carcinoid, and the adenocarcinoma pool present on

) the real-time RT ± PCR panel (Figure 4A). L580S is expressed in only three of the eight SCLC samples, with signi®cant expression also seen in normal brain, pituitary gland, spinal cord, trachea, and salivary gland Continued (Figure 4A). L1435P is expressed only in one of the SCLC samples as well as normal stomach (Figure 4A), which correlate well with its microarray expression Table 2 ( Recovery frequency Ratio*11111 2.331 2.621 3.52 3.69*Ratio represents 2.47 that Homo GenBank fold sapiensthe identity over-expression beta-subunit hits of 3.11 signal genes transducing Homo in proteins sapiens SCLC Homo GS/GI 2.58 breast sample sapiens Homo group carcinoma-associated ATP sapiens compared antigen synthase Homo Bruton's to isoform alpha AF070603 sapiens tyrosine normal I subunit apolipoprotein kinase-associated tissue L1 protein-135 sample group. For genes AF227899 recovered multiple times U77948 1 Homo sapiens Homo bullous sapiens 1 pemphigoid BAF57 antigen 1 6.43 1 D14710 2.25 AF305224 Acc. No. 2.59 1 frequency D63465 1 Ratio* 2.05 4.57 AF035262 1 Novel cDNA Novel cDNA 9.11 Novel cDNA GenBank identity Novel cDNA Novel cDNA Novel cDNA Acc. No.

Oncogene SCLC gene expression profiling CS Bangur et al 3820 Table 3 Summary of candidate genes dierentially over-expressed in SCLC Candidate GenBank Recovery Fold over-expression in SCLC name Identity Acc. No. frequency Microarray Real-Time RT ± PCR

L978P SOX4 X70683 31 5.3 26.3 L580S SOX2 Z31560 4 3.2 6 L1435P SOX21 AF107044 1 2.1 5.4 L1437P Ubiquitin-conjugating enzyme E2 homolog AF161499 8 5.6 20.4 L1438p Pituitary tumor transforming gene protein 1 AF095287 7 8.6 12.8 L1439P NUF2R AF326731 1 6.9 25.9 L986P ESTs AW956024 2 4.3 6.9 L1423P CRMP-1 D78012 2 9.7 8.7 L1424P Novel cDNA 1 2.5 11.6

Figure 3 Microarray analysis of genes dierentially over-expressed in SCLC. cDNA clones were PCR ampli®ed, arrayed onto 32 replicate glass slides, and analysed using 32 pairs of Cy3 or Cy5 labeled cDNA samples as indicated. The ¯uorescence intensity for each sample is shown as pseudocolor image and corresponds to the level of expression of a given gene in the tissue represented by that sample. Also, listed are the fold dierential expression values for the Cy3 labeled sample over the Cy5 labeled sample in case of the SCLC sample pairs (Sample Pair #23 ± 30)

pro®le (Figure 3). L1437P, L1438P, and L1439P are RT ± PCR analysis of L1424P shows that it is expressed expressed in majority of the SCLC tumors (Figure 4B). not only in the atypical carcinoid as observed by However, expression of these genes in the metastatic microarray analysis (Figure 4C), but also in one of the atypical carcinoid tumor was at a very negligible level. primary SCLC tumors which was not detected by L1438P and L1439P expression can also be detected in microarray analysis (Figure 3). This may be due to the bone marrow, which is again consistent with the higher sensitivity of real-time RT ± PCR technology in microarray analysis results. Similar to the microarray detecting low abundant messages. results, the real-time RT ± PCR analysis showed that L986P and L1423P are over-expressed not only in SCLC tumors but also in the atypical carcinoid tumor Discussion (Figure 4C). Signi®cant levels of L986P and L1423P message are also detected in normal neurological We have been successful in identifying 209 genes that tissues such as brain and pituitary gland. Real-time are dierentially over-expressed in SCLC tumors. This

Oncogene SCLC gene expression profiling CS Bangur et al 3821 was achieved using a combination of technologies, strasser, 1996). The ubiquitin proteolytic pathway plays which included suppression subtractive hybridization an essential role in the regulation of key cellular for cDNA subtraction, high-density cDNA microarray functions including the normal progression of cell cycle and quantitative real-time RT ± PCR. Combination of (Hochstrasser, 1996; Tannoch et al., 2000). Several subtractive hybridization and high-density cDNA tumor suppressors such as p53 and oncogenes are also microarrays has been well documented for identifying turned over by this pathway (Spataro et al., 1998). It tumor speci®c genes in a variety of cancers (Kitahara will be interesting to see which proteins this new et al., 2001). A good indication for the success of this member of the E2 enzyme family targets for degrada- approach is the fact that several of the genes including tion and whether up-regulation of L1437P expression GRP, ASH1, TOP2A and CEA are known to be highly causes tumor transformation. L1438P or pituitary expressed in SCLC tumors (Ball et al., 1993; Giaccone, tumor-transforming gene (PTTG) functions as an 1994; Kim et al., 1992; Yamaguchi et al., 1983). inhibitor of sister-chromatid separation (Zhang et al., Another indication of the success of this approach was 1999; Zou et al., 1999). Over-expression of PTTG in the identi®cation of a large number of genes that have NIH 3T3 cells induces transformation in vitro and been previously described to be associated with cancer. tumor formation in vivo (Zhang et al., 1999). This These include genes involved in cell ± cycle regulation, oncogenic activity is attributed to its function as over- apoptosis, protein synthesis and degradation, transcrip- expression of PTTG disrupts sister-chromatid separation factors, and oncogenes among others (Table 2). tion and causes aneuploidy (Yu et al., 2000; Zou et al., Quantitative real-time RT ± PCR was used to con®rm 1999). PTTG is over-expressed in multiple tumors, the expression of some of the target genes identi®ed by including pituitary adenomas, leukemia, lymphoma, microarray analysis. This was used as an independent lung, colon, breast and ovarian carcinomas (Heaney et but complimentary approach to microarray expression al., 2000; Puri et al., 2001; Zhang et al., 1999), making analysis and provided us with an accurate quantitative it an ideal therapeutic target for treating cancer. picture of the expression of these genes in the target L1439P is the human homologue of the yeast NUF2 tissues. gene, which encodes a protein that is localized to the L978P, L580S and L1435P belong to the SOX nuclear face of the spindle pole body (Osborne et al., protein family (Wegner, 1999). These proteins contain 1994). Disruption of NUF2 function in yeast results in a conserved DNA binding motif known as the HMG cells arrested in mitosis, with shortened mitotic spindle box, which is present in a large number of non-histone and fully replicated DNA (Osborne et al., 1994), chromosomal proteins part of the HMG box super- suggesting a speci®c defect in chromosome segregation. family. SOX proteins are involved in various aspects As both L1438P and L1439P are required for the of early development during embryogenesis such as sex proper progression of cell division through mitosis it determination, neural development, lens development, will be interesting to see if over-expression of these two chondrogenesis, and hemopoiesis (Wegner, 1999). To genes contributes to the high rate of aneuploidy found our knowledge this is the ®rst report that describes the in SCLC (Jackson-York et al., 1991; Oud et al., 1989; dierential expression of SOX genes in SCLC tumors. Travis et al., 1991). However, it should be noted that Gure et al. (2000) L1423P or collapsing response mediator protein-1 have earlier identi®ed SOX2, SOX3 and SOX21 as (CRMP-1) is one of the ®ve known members of the highly immunogenic SCLC tumor antigens by serolo- CRMP intracellular signaling phosphoproteins (Hama- gical expression cloning approach. They showed jima et al., 1996; Quinn et al., 1999). CRMP proteins presence of high-titer antibodies to SOX2, SOX3 and are involved in axonal guidance and neuronal dier- SOX21 in SCLC patient sera and suggested that these entiation by mediating semaphorin/collapsing induced proteins could be potential cancer diagnostics and growth cone collapse (Quinn et al., 1999). Several vaccine targets. It has also been recently reported that studies have shown the presence of autoantibodies ectopic expression of SOX3 in chicken embryo against CRMP-3 and CRMP-5 in SCLC patients ®broblasts can cause oncogenic transformation of (Honnorat et al., 1999; Yu et al., 2001). It will be these cells in culture (Xia et al., 2000). These results interesting to see if autoantibodies against CRMP-1 are suggest that members of SOX protein family can also present in SCLC patients. Recently it was reported induce aberrant cell growth and may have the that CRMP-1 might be involved in cancer cell invasion potential to induce oncogenic transformation in (Shih et al., 2001). Speci®cally it was shown that there SCLC. is an inverse correlation between CRMP-1 expression L1437P, L1438P and L1439P are all genes that are and the invasive ability of lung adenocarcinoma cells. either directly or indirectly involved in the regulation It was also reported that CRMP-1 is dierentially of cell cycle. It has been well documented that expressed in NSCLC tumors and the expression is disruptions in the normal process of cell cycle related to the tumor stage, lymph node metastasis, and progression is a common feature shared by all cancers. survival of the patients (Shih et al., 2001). L986P and L1437P is a novel member of the ubiquitin-conjugating L1424P represent two novel cDNA sequences. Full- enzyme E2 gene family. E2 enzymes are part of the length cloning of these genes and further characteriza- ubiquitin-proteasome pathway for protein degradation tion of their function will help us in determining their and are a large family of ubiquitin-carrier proteins with usefulness as cancer therapeutic targets and their role dierent functions and substrate speci®cities (Hoch- in the mechanisms of SCLC tumorigenesis.

Oncogene SCLC gene expression profiling CS Bangur et al 3822 By using cDNA subtraction and microarray expres- Materials and methods sion pro®ling we have been able to take a comprehensive look at the changes in expression pattern of genes Tissue and RNA sources in SCLC. We report the identi®cation of several hundred dierentially expressed genes and the initial Tumor and normal tissues samples were obtained from Cooperative Human Tissue Network (CHTN), National characterization of nine genes that are preferentially Disease Research Interchange (NDRI), and other clinical expressed in SCLC tumors. Further characterization of sources. SCLC tumor cell lines were obtained from American these nine genes and other genes identi®ed in this study Type Culture Collection (ATCC). Total RNA was isolated will be helpful in de®ning a panel of diagnostic and from the tissue samples using Trizol total-RNA isolation kit therapeutic targets for SCLC and better our under- (Invitrogen, Carlsbad, CA, USA). Poly(A)+ RNA puri®ca- standing into the biology of this malignancy. tion was carried out using Qiagen mRNA Puri®cation Kit

Oncogene SCLC gene expression profiling CS Bangur et al 3823 (Qiagen Inc. Valencia, CA, USA). Some normal tissue total- subtractive hybridization of adapter-ligated testers with RNA and poly(A)+ RNA were purchased from Clontech drivers was performed according to the PCR-Select cDNA (Palo Alto, CA, USA) and Invitrogen (Carlsbad, CA, USA). Subtraction kit protocol, using a tester to driver ratio of 1 : 60. Two rounds of PCR ampli®cation were performed to complete the subtraction and amplify the subtracted cDNA Subtracted cDNA libraries fragments. The ®rst ampli®cation was performed as follows: To enrich for genes preferentially expressed in SCLC four 5 min adapter extension at 758C, then 948C/30 s, 668C/30 s subtracted libraries were constructed using the PCR-Select and 728C/90 s for 27 cycles. The second PCR was performed cDNA Subtraction kit (Clontech Inc., Palo Alto, CA, USA). with conditions of 948C/30 s, 688C/30 s and 728C/90 s for 12 The SCL1 subtracted library was constructed using poly(A)+ cycles. The primers used for PCR were supplied with the RNA from a single SCLC primary tumor sample as tester PCR-Select cDNA Subtraction kit. The SCL2 subtracted and a pool of eight normal tissue (lung, brain, kidney, liver, library was constructed using the procedure essentially pancreas, skin, heart and spleen) poly(A)+ RNA in equal described above for the construction of the SCL1 library. amounts as driver. The SMART cDNA ampli®cation kit Except that the poly(A)+ RNA used for generating the tester (Clontech Inc., Palo Alto, CA, USA) was used to synthesize cDNA was derived from an independent SCLC tumor and amplify the tester and driver cDNA. Two hundred ng of sample. Also, the cDNA was synthesized straight from 2 mg poly(A)+ RNA for both tester and driver were used to poly(A)+ RNA using reagents included in the PCR-Select synthesize the cDNA and ampli®cation was carried out for 24 cDNA Subtraction kit, without the initial cDNA ampli®ca- cycles. The ampli®ed cDNA was puri®ed and 2 mgof tion that was carried out in case of the SCL1 library ampli®ed cDNA for both tester and driver were then used construction. The SCL3 and SCL4 libraries were constructed for the construction of the SCL1 library using the PCR-Select using the procedure essentially as described above for the cDNA Subtraction kit. The tester and driver cDNAs were construction of the SCL2 library. Except that the driver blunt-ended with T4 polymerase and then digested with a cDNA pool was spiked with cDNAs for gastrin-releasing mixture of four endonucleases including MscI, PvuII, DraI, peptide (GRP) and achaete-scute homolog 1 (ASH1), two and StuI. The digested tester cDNAs were ligated to adapter highly abundant genes present in the SCL2 library, during 1 and adapter 2R, provided with the PCR-Select cDNA the hybridization step. For the SCL3 library the hybridiza- Subtraction kit, in separate ligation reaction. The ligation tion was carried out at a tester to driver ratio of 1 : 60 and for reaction mixtures were incubated overnight at 168C. The the SCL4 library the ratio was 1 : 80.

Figure 4 Quantitative real-time RT ± PCR analysis of genes dierentially over-expressed in SCLC. (A) Real-time PCR analysis of L978P, L580S and L1435P. (B) Real-time PCR analysis of L1437P, L1438P, and L1439P. (C) Real-time PCR analysis of L986P, L1423P, and L1424P. Each gene was analysed using an identical set of 36 cDNA samples representing (1) SCLC primary tumor, (2) SCLC primary tumor, (3) SCLC cell line NCI-H69, (4) SCLC cell line NCI-H128, (5) SCLC cell line DMS 79, (6) SCLC cell line HTB 171, (7) SCLC cell line HTB 173, (8) SCLC cell line HTB 175, (9) metastatic atypical carcinoid, (10) pool of four lung adenocarcinomas, (11) pool of four primary lung squamous cell carcinomas, (12) normal lung, (13) normal lung, (14) normal brain, (15) normal pituitary gland, (16) normal adrenal gland, (17) normal thyroid gland, (18) normal pancreas, (19) normal heart, (20) normal liver, (21) normal skeletal muscle, (22) normal stomach, (23) normal kidney, (24) normal small intestine, (25) normal colon, (26) normal bladder, (27) normal esophagus, (28) normal skin, (29) normal salivary gland, (30) normal trachea, (31) normal bone marrow, (32) PBMC, (33) normal spleen, (34) normal lymph node, (35) normal thymus, and (36) normal spinal cord. The expression of each gene in each cDNA sample was normalized to the internal b-actin levels and is reported as copies per 1000 copies of b-actin

Oncogene SCLC gene expression profiling CS Bangur et al 3824 cDNA elements deposited at multiple locations on the Cloning and sequencing analysis of the subtracted cDNA library. microarray. Secondary PCR products generated from the subtracted cDNAs were subcloned into the pCR2.1 TOPO T/A cloning Quantitative real-time RT ± PCR vector (Invitrogen, Carlsbad, CA, USA). The ligated cDNAs were then transformed into DH10B electro-competent cells Quantitative real-time RT ± PCR reactions were performed by electroporation (Invitrogen, Carlsbad, CA, USA). Forty- on a GeneAmp1 5700 sequence detection system using the eight to 96 clones were randomly picked from each library SYBR1 Green I dye method for Real-Time PCR (Perkin for sequence analysis. Plasmid DNAs were prepared using a Elmer/Applied Biosystems Division, Foster City, CA, USA). QIAprep 96 Turbo Miniprep kit (Qiagen Inc. Valencia, CA, To compare the relative level of gene expression in multiple USA) according to the manufacturer's protocol. DNA tissue samples, a panel of 36 cDNAs was constructed using sequencing was performed at Corixa using an ABI377 total RNA from SCLC tumors, SCLC cell lines, metastatic automated sequencer (Perkin Elmer/Applied Biosystems atypical carcinoid, lung adenocarcinomas (pool of four tumor Division, Foster City, CA, USA), with M13 forward and samples), squamous cell lung carcinomas (pool of four tumor reverse primers. DNA homology database searches were samples), normal lung, and other normal tissues. Total RNA performed using the BLAST program. was treated with DNase I (Invitrogen, Carlsbad, CA, USA) to remove genomic DNA contamination before cDNA synthesis. Mock cDNA synthesis reactions were performed, Gene expression analysis by cDNA microarray without reverse transcriptase, using 5 mg of DNase treated A total of 2400 cDNA inserts from the SCL1, SCL3 and total RNA in a ®nal volume of 100 ml. Two ml of the mock SCL4 subtracted cDNA libraries were PCR ampli®ed with cDNA reaction was used in a 25 ml b-actin PCR reaction to M13 forward and reverse primers. The ampli®ed cDNA verify the absence of genomic DNA contamination. DNase inserts were analysed visually on 1% agarose gels to ensure treated total RNA samples veri®ed to be genomic DNA free quality and quantity of the DNA prior to deposition onto 32 were then used for the cDNA synthesis, which was carried replicate glass slides (microarrays) using a Genetic Micro- out using 25 mg of DNase treated total RNA, by oligo-dT systems GMS 417 Arrayer according to the manufacturer's priming and SuperScript II reverse transcriptase (Invitrogen, recommendations. Gene expression pro®les were obtained Carlsbad, CA, USA), in a ®nal volume of 500 ml. from 32 pairs of ¯uorescently-labeled cDNA samples Quantitative real-time RT ± PCR reactions were performed synthesized from poly(A)+ RNA isolated from various lung using gene speci®c primers to quantify the copy number of tumors, normal lung and other normal tissues. The the gene of interest (GOI) in each cDNA sample. The real- ¯uorescently-labeled samples for hybridization were gener- time PCR reactions were performed in 25 ml volumes that ated using a two-step process as described earlier (Hughes et include 2.5 ml of SYBR green buer, 2 ml of cDNA template al., 2001). Brie¯y, 1 mg of poly(A)+ RNA was ®rst converted and 2.5 ml each of the forward and reverse primers for the into aminoallyl-modi®ed cDNA (aa-cDNA) during reverse GOI. Mock reactions without cDNA template were also transcription with SuperScript II reverse transcriptase (In- included to check for non-speci®c signal due to PCR artifacts vitrogen, Carlsbad, CA, USA) and then reacted with N- such as primer dimerization. The cDNAs used for the PCR hydroxysuccinimide esters of Cy3 or Cy5 (Amersham reactions were diluted 1 : 5 for each GOI and 1 : 50 for the b- Pharmacia Biotech, Piscataway, NJ, USA). In each case, as actin control. To quantitate the expression level of GOI in an internal control for reverse transcription (RT) and tumor and normal tissue samples, standard curves were microarray hybridization sensitivity, a synthetic mRNA generated using known copy numbers (20 to 26106) of GOI transcript (pE1a, Hughes et al., 2001) was spiked into each and b-actin cDNA, and GOI expression in each cDNA sample prior to reverse transcription. Likewise, as a coupling sample was normalized to the internal b-actin levels. The ®nal eciency control, an aminoallyl-modi®ed transcript (derived real-time RT-PCR results are reported as copy number of from pCAT; Promega, Madison, WI, USA) was spiked into GOI per 1000 copies of b-actin. each aa-cDNA sample. Hybridization of samples, imaging of microarrays and quanti®cation of microarray features were done as described earlier (Marton et al., 1998), except that a GenePix 4000A (Axon Instruments, Union City, CA, USA) Acknowledgments microarray scanner was used. Hybridization intensity of each We thank Dhileep Sivam for coordinating the ampli®cation spot was normalized to the mean intensity of all non-control of cDNA inserts for microarray analysis, Dianne Boyle for spots for each channel. Quality of hybridization data was preparation of RNA and Dr Jiangchun Xu for coordinat- evaluated using standard metrics such as mean signal to noise ing the microarray analysis with Rosetta Inpharmatics. ratio (SNR) for all cDNA spots, SNR for each individual Some tissue samples used in this study were obtained from spot, spot morphology and ability to detect spiked in the National Disease Research Interchange and from the transcripts. The reproducibility of this method was ensured Cooperative Human Tissue Network, which is funded by by including 16 ubiquitously expressed genes as control the National Cancer Institute.

References

Anbazhagan R, Tihan T, Bornman DM, Johnston JC, Saltz Ball DW, Azzoli CG, Baylin SB, Chi D, Dou S, Donis-Keller JH, Weigering A, Piantadosi S and Gabrielson E. (1999). H, Cumaraswamy A, Borges M and Nelkin BD. (1993). Cancer Res., 59, 5119 ± 5122. Proc. Natl Acad. Sci. USA, 90, 5648 ± 5652. Backert S, Gelos M, Kobalz U, Hanski ML, Bohm C, Mann Diatchenko L, Lau YF, Campbell AP, Chenchik A, B, Lovin N, Gratchev A, Mansmann U, Moyer MP, Moqadam F, Huang B, Lukyanov S, Lukyanov K, Riecken EO and Hanski C. (1999). Int. J. Cancer, 82, Gurskaya N, Sverdlov ED and Siebert PD. (1996). Proc. 868 ± 874. Natl. Acad. Sci. USA, 93, 6025 ± 6030.

Oncogene SCLC gene expression profiling CS Bangur et al 3825 Feld R, Sagman U and Lederer G. (2000). Staging and Ono K, Tanaka T, Tsunoda T, Kitahara O, Kihara C, prognostic factors for small cell lung cancer. Pass HI, Okamoto A, Ochiai K, Takagi T and Nakamura Y. Mitchell JB, Johnson DH, Turrisi AT and Minna JD (eds). (2000). Cancer Res., 60, 5007 ± 5011. Lippincott Williams & Wilkins: Philadelphia, USA, Osborne MA, Schlenstedt G, Jinks T and Silver PA. (1994). pp. 612 ± 627. J. Cell. Biol., 125, 853 ± 866. Giaccone G. (1994). Anticancer Res., 14, 269 ± 276. Oud PS, Pahlplatz MM, Beck JL, Wiersma-Van Tilburg A, Ginsberg RJ, Vokes EE and Raben A. (1997). Non-small cell Wagenaar SJ and Vooijs GP. (1989). Cancer, 64, 1304 ± lung cancer. DeVita Jr VT, Hellman S and Rosenberg SA 1309. (eds). JB Lippincott Co.: Philadelphia, USA, pp. 858 ± 911. Parker SL, Tong T, Bolden S and Wingo PA. (1997). CA Girard L, Zochbauer-Muller S, Virmani AK, Gazdar AF Cancer J. Clin., 47, 5 ± 27. and Minna JD. (2000). Cancer Res., 60, 4894 ± 4906. Puri R, Tousson A, Chen L and Kakar SS. (2001). Cancer Gray JW and Collins C. (2000). Carcinogenesis, 21, 443 ± 452. Lett., 163, 131 ± 139. GuineeDG,Jr.,FishbackNF,KossMN,AbbondanzoSL Quinn CC, Gray GE and Hock®eld S. (1999). J. Neurobiol., and Travis WD. (1994). Am.J.Clin.Pathol.,102, 406 ± 41, 158 ± 164. 414. Sgroi DC, Teng S, Robinson G, Le Vangie R, Hudson Jr. JR Gure AO, Stockert E, Scanlan MJ, Keresztes RS, Jager D, and Elkahloun AG. (1999). Cancer Res., 59, 5656 ± 5661. Altorki NK, Old LJ and Chen YT. (2000). Proc. Natl. Shih JY, Yang SC, Hong TM, Yuan A, Chen JJ, Yu CJ, Acad. Sci. USA, 97, 4198 ± 4203. Chang YL, Lee YC, Peck K, Wu CW and Yang PC. Hamajima N, Matsuda K, Sakata S, Tamaki N, Sasaki M (2001). J. Natl. Cancer Inst., 93, 1392 ± 1400. and Nonaka M. (1996). Gene, 180, 157 ± 163. Spataro V, Norbury C and Harris AL. (1998). Br. J. Cancer, Heaney AP, Singson R, McCabe CJ, Nelson V, Nakashima 77, 448 ± 455. M and Melmed S. (2000). Lancet, 355, 716 ± 719. Tannoch VJ, Hinds PW and Tsai LH. (2000). Adv. Exp. Med. Hochstrasser M. (1996). Ann. Rev. Genet., 30, 405 ± 439. Biol., 465, 127 ± 140. Honnorat J, Byk T, Kusters I, Aguera M, Ricard D, Travis WD, Linnoila RI, Tsokos MG, Hitchcock CL, Cutler Rogemond V, Quach T, Aunis D, Sobel A, Mattei MG, Jr. GB, Nieman L, Chrousos G, Pass H and Doppman J. Kolattukudy P, Belin MF and Antoine JC. (1999). Eur. J. (1991). Am. J. Surg. Pathol., 15, 529 ± 553. Neurosci., 11, 4226 ± 4232. Villaret DB, Wang T, Dillon D, Xu J, Sivam D, Cheever MA Hughes TR, Mao M, Jones AR, Burchard J, Marton MJ, and Reed SG. (2000). Laryngoscope, 110, 374 ± 381. Shannon KW, Lefkowitz SM, Ziman M, Schelter JM, Wang T, Hopkins D, Schmidt C, Silva S, Houghton R, Meyer MR, Kobayashi S, Davis C, Dai H, He YD, Takita H, Repasky E and Reed SG. (2000). Oncogene, 19, Stephaniants SB, Cavet G, Walker WL, West A, Coey E, 1519 ± 1528. Shoemaker DD, Stoughton R, Blanchard AP, Friend SH Wegner M. (1999). Nucleic Acids Res., 27, 1409 ± 1420. and Linsley PS. (2001). Nat. Biotechnol., 19, 342 ± 347. Williams CL. (1997). Chest Surg. Clin. N. Am., 7, 1±19. Ihde DC. (1992). N. Engl. J. Med., 327, 1434 ± 1441. Wistuba II, Behrens C, Virmani AK, Mele G, Milchgrub S, Ihde DC, Pass HI and Glatstein E. (1997). Smallcelllung Girard L, Fondon JW, III, Garner HR, McKay B, Latif F, carcinoma. DeVita Jr VT, Hellman S and Rosenberg SA Lerman MI, Lam S, Gazdar AF and Minna JD. (2000a). (eds). JB Lippincott Co.: Philadelphia, USA, pp. 911 ± 941. Cancer Res., 60, 1949 ± 1960. Jackson-York GL, Davis BH, Warren WH, Gould VE and Wistuba II, Berry J, Behrens C, Maitra A, Shivapurkar N, Memoli VA. (1991). Cancer, 68, 374 ± 379. Milchgrub S, Mackay B, Minna JD and Gazdar AF. Kalemkerian GP. (2000). Growth Factors. Pass HI, Mitchell (2000b). Clin. Cancer Res., 6, 2604 ± 2610. JB, Johnson DH, Turrisi AT and Minna JD (eds). Xia Y, Papalopulu N, Vogt PK and Li J. (2000). Cancer Res., Lippincott Williams & Wilkins: Philadelphia, USA, 60, 6303 ± 6306. pp. 181 ± 198. Xu J, Stolk JA, Zhang X, Silva SJ, Houghton RL, KimJ,KayeFJ,HensleeJG,ShivelyJE,ParkJG,LaiSL, Matsumura M, Vedvick TS, Leslie KB, Badaro R and Linnoila RI, Mulshine JL and Gazdar AF. (1992). Int. J. Reed SG. (2000). Cancer Res., 60, 1677 ± 1682. Cancer, 52, 718 ± 725. Yamaguchi K, Abe K, Kameya T, Adachi I, Taguchi S, Kitahara O, Furukawa Y, Tanaka T, Kihara C, Ono K, Otsubo K and Yanaihara N. (1983). Cancer Res., 43, Yanagawa R, Nita ME, Takagi T, Nakamura Y and 3932 ± 3939. Tsunoda T. (2001). Cancer Res., 61, 3544 ± 3549. Yang GP, Ross DT, Kuang WW, Brown PO and Weigel RJ. Lengauer C, Kinzler KW and Vogelstein B. (1998). Nature, (1999). Nucleic Acids Res., 27, 1517 ± 1523. 396, 643 ± 649. Yu R, Heaney AP, Lu W, Chen J and Melmed S. (2000). J. Marton MJ, DeRisi JL, Bennett HA, Iyer VR, Meyer MR, Biol. Chem., 275, 36502 ± 36505. Roberts CJ, Stoughton R, Burchard J, Slade D, Dai H, Yu Z, Kryzer TJ, Griesmann GE, Kim K, Benarroch EE and Bassett DE, Hartwell LH, Brown PO and Friend SH. Lennon VA. (2001). Ann. Neurol., 49, 146 ± 154. (1998). Nat. Med., 4, 1293 ± 1301. Zhang X, Horwitz GA, Prezant TR, Valentini A, Nakashima Nacht M, Ferguson AT, Zhang W, Petroziello JM, Cook BP, M, Bronstein MD and Melmed S. (1999). Mol. Endocri- Gao YH, Maguire S, Riley D, Coppola G, Landes GM, nol., 13, 156 ± 166. Madden SL and Sukumar S. (1999). Cancer Res., 59, Zou H, McGarry TJ, Bernal T and Kirschner MW. (1999). 5464 ± 5470. Science, 285, 418 ± 422. Nocito A, Bubendorf L, Maria TE, Suess K, Wagner U, Forster T, Kononen J, Fijan A, Bruderer J, Schmid U, Ackermann D, Maurer R, Alund G, Knonagel H, Rist M, Anabitarte M, Hering F, Hardmeier T, Schoenenberger AJ, Flury R, Jager P, Luc FJ, Schraml P, Moch H, Mihatsch MJ, Gasser T and Sauter G. (2001). J. Pathol., 194, 349 ± 357.

Oncogene