Transcriptomic analysis of hepatocellular carcinoma reveals molecular features of disease progression and tumor immune biology.

Hepatocellular carcinoma (HCC) develops in the context of chronic inflammatory liver disease and has an extremely poor prognosis. An immunosuppressive tumor microenvironment may contribute to therapeutic failure in metastatic HCC. Here, we identified unique molecular signatures pertaining to HCC disease progression and tumor immunity by analyzing genome-wide RNA-Seq data derived from HCC patient tumors and non-tumor cirrhotic tissues. Unsupervised clustering of gene expression data revealed a gradual suppression of local tumor immunity that coincided with disease progression, indicating an increasingly immunosuppressive tumor environment during HCC disease advancement. IHC examination of the spatial distribution of CD8+ T cells in tumors revealed distinct intra- and peri-tumoral subsets. Differential gene expression analysis revealed an 85-gene signature that was significantly upregulated in the peri-tumoral CD8+ T cell-excluded tumors. Notably, this signature was highly enriched with components of underlying extracellular matrix, fibrosis, and epithelial-mesenchymal transition (EMT). Further analysis condensed this signature to a core set of 23 genes that are associated with CD8+ T cell localization, and were prospectively validated in an independent cohort of HCC specimens. These findings suggest a potential association between elevated fibrosis, possibly modulated by TGF-ß, PDGFR, SHH or Notch pathway, and the T cell-excluded immune phenotype. Indeed, targeting fibrosis using a TGF-ß neutralizing antibody in the STAM™ model of murine HCC, we found that ameliorating the fibrotic environment could facilitate redistribution of CD8+ lymphocytes into tumors. Our results provide a strong rationale for utilizing immunotherapies in HCC earlier during treatment, potentially in combination with anti-fibrotic therapies.

Integrated genomic analysis of colorectal cancer progression reveals activation of EGFR through demethylation of the EREG promoter.

Key molecular drivers that underlie transformation of colonic epithelium into colorectal adenocarcinoma (CRC) are well described. However, the mechanisms through which clinically targeted pathways are activated during CRC progression have yet to be elucidated. Here, we used an integrative genomics approach to examine CRC progression. We used laser capture microdissection to isolate colonic crypt cells, differentiated surface epithelium, adenomas, carcinomas and metastases, and used gene expression profiling to identify pathways that were differentially expressed between the different cell types. We identified a number of potentially important transcriptional changes in developmental and oncogenic pathways, and noted a marked upregulation of EREG in primary and metastatic cancer cells. We confirmed this pattern of gene expression by in situ hybridization and observed staining consistent with autocrine expression in the tumor cells. Upregulation of EREG during the adenoma-carcinoma transition was associated with demethylation of two key sites within its promoter, and this was accompanied by an increase in the levels of epidermal growth factor receptor (EGFR) phosphorylation, as assessed by reverse-phase protein analysis. In CRC cell lines, we demonstrated that EREG demethylation led to its transcriptional upregulation, higher levels of EGFR phosphorylation, and sensitization to EGFR inhibitors. Low levels of EREG methylation in patients who received cetuximab as part of a phase II study were associated with high expression of the ligand and a favorable response to therapy. Conversely, high levels of promoter methylation and low levels of EREG expression were observed in tumors that progressed after treatment. We also noted an inverse correlation between EREG methylation and expression levels in several other cancers, including those of the head and neck, lung and bladder. Therefore, we propose that upregulation of EREG expression through promoter demethylation might be an important means of activating the EGFR pathway during the genesis of CRC and potentially other cancers.

A role for FOXO1 in BCR-ABL1-independent tyrosine kinase inhibitor resistance in chronic myeloid leukemia.

Chronic myeloid leukemia (CML) patients who relapse on imatinib due to acquired ABL1 kinase domain mutations are successfully treated with second-generation ABL1-tyrosine kinase inhibitors (ABL-TKIs) such as dasatinib, nilotinib or ponatinib. However, ~40% of relapsed patients have uncharacterized BCR-ABL1 kinase-independent mechanisms of resistance. To identify these mechanisms of resistance and potential treatment options, we generated ABL-TKI-resistant K562 cells through prolonged sequential exposure to imatinib and dasatinib. Dual-resistant K562 cells lacked BCR-ABL1 kinase domain mutations, but acquired other genomic aberrations that were characterized by next-generation sequencing and copy number analyses. Proteomics showed that dual-resistant cells had elevated levels of FOXO1, phospho-ERK and BCL-2, and that dasatinib no longer inhibited substrates of the PI3K/AKT pathway. In contrast to parental cells, resistant cells were sensitive to growth inhibition and apoptosis induced by the class I PI3K inhibitor, GDC-0941 (pictilisib), which also induced FOXO1 nuclear translocation. FOXO1 was elevated in a subset of primary specimens from relapsed CML patients lacking BCR-ABL1 kinase domain mutations, and these samples were responsive to GDC-0941 treatment ex vivo. We conclude that elevated FOXO1 contributes to BCR-ABL1 kinase-independent resistance experienced by these CML patients and that PI3K inhibition coupled with BCR-ABL1 inhibition may represent a novel therapeutic approach.

Predicting clinical benefit in non-small-cell lung cancer patients treated with epidermal growth factor tyrosine kinase inhibitors.

Erlotinib and gefitinib are small-molecule inhibitors of the epidermal growth factor tyrosine kinase. Erlotinib is approved for the treatment of locally advanced or metastatic non-small-cell lung cancer after failure of at least one prior chemotherapy regimen. Although it is active in unselected patients, clinical characteristics and tumor molecular markers associated with enhanced benefit have been identified. Notably, never-smoker status or a positive EGFR FISH test has been consistently predictive of greater erlotinib benefit. Other markers, such as EGFR mutations and EGFR protein expression, as determined by immunohistochemistry, and KRAS mutation status have not proven to be consistently associated with differential benefit.

Gene expression profiling for prognosis using Cox regression.

Given the promise of rich biological information in microarray data we will expect an increasing demand for a robust, practical and well-tested methodology to provide patient prognosis based on gene expression data. In standard settings, with few clinical predictors, such a methodology has been provided by the Cox proportional hazard model, but no corresponding methodology is available to deal with the full set of genes in microarray data. Furthermore, we want the procedure to be able to deal with the general survival data that include censored information. Conceptually such a procedure can be constructed quite easily, but its implementation will never be straightforward due to computational problems. We have developed an approach that relies on an extension of the Cox proportional likelihood that allows random effects parameters. In this approach, we use the full set of genes in the analysis and deal with survival data in the most general way. We describe the development of the model and the steps in the implementation, including a fast computational formula based on a subsampling of the risk set and the singular value decomposition. Finally, we illustrate the methodology using a data set obtained from a cohort of breast cancer patients.

Dysregulated expression of androgen-responsive and nonresponsive genes in the androgen-independent prostate cancer xenograft model CWR22-R1.

Treatment of metastatic prostate cancer with androgen-ablation often elicits dramatic tumor regressions, but the response is rarely complete, making clinical recurrence inevitable with time. To gain insight into therapy-related progression, changes in gene expression that occurred following androgen-deprivation of an androgen-dependent prostate tumor xenograft, CWR22, and the emergence of an androgen-independent tumor, CWR22-R, were monitored using microarray analysis. Androgen-deprivation resulted in growth arrest of CWR22 cells, as evidenced by decreased expression of genes encoding cell cycle components and basal cell metabolism, respiration and transcription, and the induced expression of putative negative regulatory genes that may act to sustain cells in a nonproliferative state. Evolution of androgen-independent growth and proliferation, represented by CWR22-R, was associated with a reentry into active cell cycle and the up-regulation of several genes that were expressed at low levels or absent in the androgen-dependent tumor. Androgen repletion to mice bearing androgen-independent CWR22-R tumors induced, augmented, or repressed the expression of a number of genes. Expression of two of these genes, the calcium-binding protein S100P and the FK-506-binding protein FKBP51, was decreased following androgen-deprivation, subsequently reexpressed in CWR22-R at levels comparable with CWR22, and elevated further upon treatment with androgens. The dysregulated behavior of these genes is analogous to other androgen-dependent genes, e.g., prostate-specific antigen and human kallikrein 2, which are commonly reexpressed in androgen-independent disease in the absence of androgens. Other androgen-responsive genes whose expression decreased during androgen-deprivation and whose expression remained decreased in CWR22 were also identified in CWR22-R. These results imply that evolution to androgen-independence is due, in part, to reactivation of the androgen-response pathway in the absence of androgens, but that this reactivation is probably incomplete.

Identification and characterization of novel genes located at the t(1;15)(p36.2;q24) translocation breakpoint in the neuroblastoma cell line NGP.

The distal portion of chromosome 1p is frequently deleted in several human cancers, suggesting the presence of one or more putative tumor suppressor genes on this chromosomal arm. In human neuroblastoma, a consistently deleted region at 1p36.1-p36.2 has been defined by comparison of molecular loss of heterozygosity (LOH) analyses. Recently we described the identification of a yeast artificial chromosome, YAC 927G4, that spans a translocation/duplication breakpoint within the minimally defined LOH region at 1p36.1-p36.2 in the neuroblastoma cell line NGP. Here we describe the identification of two overlapping P1 artificial chromosomes comprising 220 kb at the distal end of YAC 927G4, which we have used as hybridization probes under modified conditions to screen a composite, normalized cDNA library (IMAGE cDNA library). Hybridization screening resulted in the rapid and comprehensive identification of partial cDNAs of which a portion comprised two novel candidate genes, termed DNB1/ARPh and DNB5, which encode putative proteins of 1011 and 447 amino acids, respectively. The DNB1/ARPh gene, which was found to be ubiquitously expressed in human adult and fetal tissues, is highly related to the DRPLA gene, in which expansion of a CAG triplet appears to be causal in the dentatorubral and pallidolysian atrophy disease phenotype. The DNB5 sequence, in contrast, which is predominantly expressed in brain tissues and fetal kidney, failed to show any similarity to sequences in the public domain. A preliminary assessment of transcription and sequence of both genes in several neuroblastoma cell lines does not, thus far, support a causal role in neuroblastoma. However, further analyses are required to confirm these results.

Secretory type II phospholipase A2 (PLA2G2A) expression status in colorectal carcinoma derived cell lines and in normal colonic mucosa.

There is good evidence now that the secretory type II phospholipase A2 (Pla2g2a) gene represents the Mom1 locus, a genetic modifier of tumor resistance in the multiple intestinal neoplasia (Min) mouse. Previously we have mapped the human homolog PLA2G2A to 1p35-36.1 within a region that is the target of frequent deletions in sporadic colorectal tumors. Here we show 64% loss of heterozygosity (LOH) at the PLA2G2A locus in primary tumors. We studied PLA2G2A expression in both colorectal tumor cell lines and normal mucosa. Most of the lines lacked detectable PLA2G2A transcripts by Northern analysis. Large differences in expression were seen among normal mucosa of different patients with sporadic tumors. We analysed the coding region of PLA2G2A in eight colorectal cancer cell lines with hemizygous deletion at 1p35-36/PLA2G2A, in none we did detect a mutation. Biallelic expression of PLA2G2A was observed in a cell line heterozygous for an exon 3 polymorphism, rendering unlikely that imprinting is a pathway participating in the loss of PLA2G2A function. It remains uncertain if PLA2G2A, in particular its apparent lack of expression in tumor cells, might be a factor in human colorectal tumorigenesis.

Variations in the expression of the adenomatous polyposis coli (APC) tumor suppressor gene in human cancer cell lines of different tissue origin.

The adenomatous polyposis coli (APC) tumor suppressor gene APC is mutated in familial adenomatous polyposis and in most sporadic colorectal tumors. Through its interaction with beta-catenin the APC protein may play a role in a signal transduction pathway regulating cell proliferation. Despite the fact that APC is ubiquitously expressed, mutations leading to truncated proteins are restricted to tumors of the digestive system. To determine further alterations not resulting in protein truncation, but possibly influencing the signaling, we compared the relative expression level of the APC protein and transcripts in 24 human colorectal cancer cell lines and in additional 17 lines of noncolorectal tissue origins, which have not previously been studied. By Western analysis, the highest levels of full-length APC protein were found in a subset of neuroblastoma and retinoblastoma cell lines. In contrast, in five noncolorectal lines it was not detectable. Truncated APC was exclusively found in 18 of the 24 colorectal cancer cell lines, but was never detected in any cell line derived from other tissues. In most colorectal cancer cell lines the protein level of full-length or mutated APC was reduced. By the more sensitive immunoprecipitation analysis, weak expression of full-length APC could be shown even in those noncolorectal cancer cell lines where it was not detectable by Western blotting. In addition, APC transcript expression was found in all cell lines, the level in colorectal cancer cell lines being reduced.

Genomic instability in 1p and human malignancies.

Both cytogenetic and molecular genetic approaches have unveiled non-random genomic alterations in 1p associated with a number of human malignancies. These have been interpreted to suggest the existence of cancer-related genes in 1p. Earlier studies had employed chromosome analysis or used molecular probes mapped by in situ hybridization. Further, studies of the various tumor types often involved different molecular probes that had been mapped by different technical approaches, like linkage analysis, radioactive or fluorescence in situ hybridization, or by employing a panel of mouse x human radiation reduced somatic cell hybrids. The lack of maps fully integrating all loci has complicated the generation of a comparative and coherent picture of 1p damage in human malignancies even among different studies on the same tumor type. Only recently has the availability of genetically mapped, highly polymorphic loci at (CA)n repeats with sufficient linear density made it possible to scan genomic regions in different types of tumors readily by polymerase chain reaction (PCR) with a standard set of molecular probes. This paper aims at presenting an up-to-date picture of the association of 1p alterations with different human cancers and compiles the corresponding literature. From this it will emerge that the pattern of alterations in individual tumor types can be complex and that a stringent molecular and functional definition of the role that Ip alterations might have in tumorigenesis will require a more detailed analysis of the genomic regions involved.

The DDX1 gene maps within 400 kbp 5' to MYCN and is frequently coamplified in human neuroblastoma.

Human neuroblastoma cells frequently show amplification of the oncogene MYCN, which maps to 2p24. Previous studies have localized the DEAD box motif gene DDX1 to the same chromosome band and demonstrated coamplification of DDX1 and MYCN in two retinoblastoma cell lines. Recently, a high frequency of coamplification of DDX1 and MYCN has been shown in human neuroblastoma cells. We have determined the physical distance between the two genes by pulsed field gel electrophoresis in normal tissue and have found that DDX1 maps to a position at a maximum distance of 400 kbp 5' to MYCN. Two neuroblastoma cell lines with coamplification of DDX1/MYCN showed a similar topographic relationship of the two genes. In contrast, in two cell lines with high copy number, the DDX1 gene was not present in all amplified units recognized by MYCN and had changed its position in the amplified DNA relative to MYCN from 5' to 3', presumably by rearrangement during the amplification process. Our data show that the high frequency of DDX1 coamplification is due to its close physical distance to MYCN. Although amplification has resulted in an elevated expression of DDX1 the significance of overexpression for neuroblastoma remains unclear.

Human homologue of a candidate for the Mom1 locus, the secretory type II phospholipase A2 (PLA2S-II), maps to 1p35-36.1/D1S199.

Mice heterozygous for the dominant Min mutation in their Apc gene develop multiple intestinal neoplasia. Analogously, family members from familial adenomatous polyposis kindreds inheriting mutations in their human APC homologue develop a similar phenotype. Quantitative trait loci studies have identified the Mom1 locus (for modifier of Min-1), which is responsible for part of the genetic variability in polyp number found among inbred mouse strains. The secretory type II phospholipase [nonpancreatic Pla2s (type II Pla2s or Pla2s-II)] has been demonstrated to be a candidate for Mom1, and a mutation in Pla2s-II in mice carrying the Min mutation has been proposed to account for an increased polyp number compared to mice without the Pla2s-II mutation. In this study, we have mapped the chromosomal position of the human homologue of Pla2s-II. We have identified 3 mega-yeast artificial chromosomes that carry PLA2S-II and localized one of them by fluorescence in situ hybridization to the border between 1p35 and 1p36.1. The presence of the microsatellite marker D1S199 in all three clones integrates PLA2S-II into different genetic maps. This highly polymorphic CA repeat D1S199 has previously been shown by us to identify loss of heterozygosity in 48% of sporadic colorectal tumors, indicating that the human homologue of the Pla2s-II/Mom1 locus might be related to human colorectal cancer.

The 1p deletion is not a reliable marker for the prognosis of patients with neuroblastoma.

Human neuroblastoma cells often have deletions of the distal short arm of chromosome 1 (1p). Earlier studies using chromosome analysis had suggested that the 1p deletion is correlated with a poor survival chance for the patient. We have reevaluated this possibility by analyzing 51 neuroblastomas for loss of heterozygosity (LOH) at 1p. We detected LOH in 32% of the cases. LOH did not correlate with the age of the patients at diagnosis or with tumor stage but was correlated significantly with amplification of the MYCN proto-oncogene. Nine of 10 MYCN-amplified tumors had deletions in 1p (P < 0.001). Survival chances of patients with tumors carrying MYCN amplification together with the deletion at 1p were decreased significantly (eight of nine affected patients died) compared with a patient group without any of these aberrations (P < 0.001). However, the deletion of 1p alone without MYCN amplification was not associated with a poor outcome compared with patients who had neither deletion nor amplification (only two of eight affected patients died; P = 0.803). From these data we conclude that 1p deletions are not reliable markers to determine a patient's prognosis. They may, however, identify a subgroup of neuroblastomas in which MYCN is amplified readily, resulting in rapid tumor progression.

Deletion mapping defines different regions in 1p34.2-pter that may harbor genetic information related to human colorectal cancer.

Cytogenetic and molecular analyses of colorectal cancer cells have revealed deletions at 1p as prominent alterations, suggesting that genetic information on the short arm of chromosome 1 has a role in tumorigenesis. In this study we have used 33 microsatellite markers to fine map deletions at 1p in primary colorectal carcinomas. We found 1p-deletions in 84% of the cases (31/37). High frequencies of loss of heterozygosity (LOH), often the result of small independent interstitial deletions in the same tumor, defined three regions, that may harbor genetic information relevant for colorectal cancer: (i) region A between D1S243 and D1S468 (7cM; 1p36.3); (ii) region B between D1S436 and D1S199 (7cM; 1p35.1-36.31) and (iii) region C between D1S496 and D1S255 (1cM; 1p34.2-35). In addition we identified seven cell lines with LOH at 1p, all of which have deletions that span at least from the distal border of region A to the proximal border of region C.

Non-syntenic amplification of MDM2 and MYCN in human neuroblastoma.

Amplification of the MYCN gene is a well documented genetic alteration of aggressively growing human neuroblastomas. Through cytogenetic studies we have identified neuroblastoma cell lines which, in addition to amplified MYCN, carry amplified DNA not harbouring MYCN. In situ hybridization of biotinylated total genomic DNA to metaphase chromosomes of normal human lymphocytes by reverse genomic hybridization revealed the amplified DNA to be derived from chromosome 12 band q13-14. Subsequent filter analyses showed a 20- to 40-fold amplification of the MDM2 gene, located at 12q13-14, both in three cell lines and in an original tumor, in addition to amplified MYCN. As the apparent consequence of amplification abundant MDM2 protein was present, a part of which was complexed with p53.

A reciprocal translocation (1;15) (36.2;q24) in a neuroblastoma cell line is accompanied by DNA duplication and may signal the site of a putative tumor suppressor-gene.

Cytogenetic analyses and molecular deletion studies of human neuroblastomas have indicated the chromosomal bands 1p36.1-1p36.2 as a location of genetic information which may be involved in tumorigenesis. To define this putative neuroblastoma locus in more detail we have analysed cell lines with alterations of distal 1p. Here we show, by fluorescence in situ hybridization (FISH), that cell line NGP has a reciprocal 1;15 translocation. Loci D1S214/D1S96 could be shown to map telomeric/distal, D1S228 centromeric/proximal to the break. We have identified yeast artificial chromosomes (YACs) that cover the break and map to D1S160 and D1S244. This chromosomal position is within the smallest region of overlap (SRO) found in neuroblastoma tumors (Weith et al., 1989; Caron et al., 1993; Schleiermacher et al., 1994) and within the region of a constitutional interstitial deletion of a neuroblastoma patient (Biegel et al., 1993). Mapping studies with FISH revealed that the translocation is associated with duplication of DNA. It appears, as if the subchromosomal region we describe here is a good candidate for harboring the postulated neuroblastoma suppressor-gene.

Cytogenetic evolution of MYCN and MDM2 amplification in the neuroblastoma LS tumour and its cell line.

Amplification of the MYCN gene is frequently seen either in extrachromosomal double minutes (DMs) or in homogeneously staining regions (HSRs) of aggressively growing neuroblastomas. Total genomic DNA from cell line LS, from early passages of the same line and from original tumour material was biotinylated and hybridised to metaphase chromosomes of normal human lymphocytes. The reverse genomic hybridisation revealed the amplified DNA to be derived both from chromosome 2p23-24, which is the position of MYCN, and from chromosome 12 band q13-14. The MDM2 gene, located at 12q13-14, was found amplified both in early and late passages of LS, in addition to amplified MYCN. Amplification units of MYCN and MDM2 appear first to develop within DMs, which then integrate into different chromosomes to develop to HSRs.

Reciprocal translocation at 1p36.2/D1S160 in a neuroblastoma cell line: isolation of a YAC clone at the break.

Band 1p36.1-1p36.2 is frequently involved in chromosomal aberrations of neuroblastoma cells, and therefore thought to harbour genetic information which may be involved in tumorigenesis. To map this putative neuroblastoma locus, we screened neuroblastoma cell lines for reciprocal translocations at 1p36.1-2 which may signal the site of an affected gene. We identified a reciprocal 1;15 translocation in cell line NGP by fluorescence in situ hybridisation (FISH). As a strategy to clone the translocation breakpoint, we isolated yeast artificial chromosomes (YACs) specific for loci at 1p36. Screening of cell line NGP by FISH identified a YAC, 1050 kbp in size, which hybridised to both derivative 1;15 and 15;1 chromosomes. We conclude that this YAC, which maps to D1S160, covers the break. This chromosomal position is within the smallest region of overlap (SRO) found in neuroblastoma tumours and within the region of a constitutional interstitial deletion of a neuroblastoma patient. The YAC we describe here should serve as a DNA source for gene cloning approaches towards the isolation of candidates for the putative neuroblastoma suppressor gene.

MYCN is retained in single copy at chromosome 2 band p23-24 during amplification in human neuroblastoma cells.

Amplification of the human N-myc protooncogene, MYCN, is frequently seen either in extrachromosomal double minutes or in homogeneously staining regions of aggressively growing neuroblastomas. MYCN maps to chromosome 2 band p23-24, but homogeneously staining regions have never been observed at this band, suggesting transposition of MYCN during amplification. We have employed fluorescence in situ hybridization to determine the status of MYCN at 2p23-24 in five human neuroblastoma cell lines. All five lines carried, in addition to amplified MYCN in homogeneously staining regions or double minutes, single-copy MYCN at the normal position. In one line there was coamplification of MYCN together with DNA of the host chromosome 12, to which MYCN had been transposed. Our results suggest a model of amplification where MYCN is retained at its original location. They further sustain the view that either the initial events of MYCN amplification or the further evolution of amplified MYCN copies follow mechanisms different from those leading to amplification of drug-resistance genes.