Identification and temporal expression pattern of genes modulated ...

ã

Oncogene (1999) 18, 3546 ± 3552 1999 Stockton Press All rights reserved 0950 ± 9232/99 $12.00 http://www.stockton-press.co.uk/onc

SHORT REPORT

Identi®cation and temporal expression pattern of genes modulated during irreversible growth arrest and terminal dierentiation in human melanoma cells Fei Huang1,2,4, Jennifer Adelman1, Hongping Jiang1,3,4, Neil I Goldstein1 and Paul B Fisher*,2,3,4 1

GenQuest Incorporated, New York, NY 10032, USA; 2Department of Neurosurgery, Herbert Irving Comprehensive Cancer Center, Columbia University, College of Physicians and Surgeons, New York, NY 10032, USA; 3Department of Pathology, Herbert Irving Comprehensive Cancer Center, Columbia University, College of Physicians and Surgeons, New York, NY 10032, USA; 4Department of Urology, Herbert Irving Comprehensive Cancer Center, Columbia University, College of Physicians and Surgeons, New York, NY 10032, USA

Abnormalities in dierentiation are common occurrences in human cancers. Treatment of human melanoma cells with the combination of recombinant human ®broblast interferon (IFN-b) and the antileukemic compound mezerein (MEZ) results in a loss of tumorigenic potential that correlates with an irreversible suppression in proliferative ability and induction of terminal dierentiation. It is hypothesized that this is associated with the dierential expression of genes that may directly regulate cancer cell growth and dierentiation. To de®ne the relevant gene expression changes that correlate with and potentially control these important cellular processes a dierentiation induction subtraction hybridization (DISH) scheme is being used. A temporally spaced subtracted dierentiation inducer treated (TSS) cDNA library was constructed and dierentially expressed DISH clones were isolated and evaluated using a high throughput microchip cDNA (Synteni) array screening approach. Veri®cation of dierential gene expression for speci®c cDNAs was con®rmed by Northern blotting. The temporal kinetics of regulation and the expression pattern of DISH genes were also evaluated by microchip cDNA array screening. Using this approach with 1000 DISH cDNA clones (*10% of the DISH library) has resulted in the identi®cation and cloning of both 26 known and 11 novel cDNAs of potential relevance to growth control and terminal dierentiation in human melanoma cells. Keywords: melanoma dierentiation; high throughput gene expression arrays; dierentially expressed cDNAs

Hallmarks of cancer cells include a potentially unlimited lifespan, a loss of normal proliferative control and an inability to undergo normal programs of dierentiation (Fisher, 1984; Jiang et al., 1994). Dierentiation therapy provides a promising approach for reversing these properties in neoplastic cells, thereby, resulting in a loss of proliferative ability and the reexpression of dierentiated traits (Waxman,

*Correspondence: PB Fisher Received 8 October 1998; revised 23 November 1998; accepted 13 January 1999

1995). The mechanism by which dierentiation inducing agents cause phenotypic changes in tumor cells is believed to involve the selective activation of de®ned sets of genes that negatively control cell proliferation and the suppression of genes facilitating expression of the cancer state (Jiang et al., 1994). By de®ning the spectrum of genes that are modi®ed as a consequence of induction or irreversible growth arrest, terminal cell dierentiation and loss of tumorigenic potential, it should be possible to identify potentially important target genes and molecules for therapeutic intervention in cancer. Treatment of metastatic human melanoma cells with a combination of IFN-b+MEZ results in a loss of tumorigenic potential, irreversible growth arrest, antigenic modulation, enhanced melanin synthesis, profound changes in gene expression and terminal cell dierentiation (Fisher et al., 1985; Guarini et al., 1992; Jiang and Fisher, 1993; Jiang et al., 1993, 1995a,b,c). To de®ne the gene expression changes associated with induction of irreversible growth arrest and terminal dierentiation in metastatic human melanoma cells, HO-1 cells were treated with IFNb+MEZ, mRNAs were isolated that span the induction and commitment phases of dierentiation (2 ± 24 h), cDNA libraries were constructed from the pooled mRNAs and subtraction hybridization was used to produce a temporally spaced subtracted dierentiation inducer treated (TSS) cDNA library (Jiang and Fisher, 1993). Initially, 70 random clones were isolated and 23 clones (*33%) were found to be dierentially expressed in dierentiated versus undifferentiated human melanoma cells and seven of the dierentially expressed clones (*30%) initially represented novel sequences not reported in then current DNA databases (Jiang and Fisher, 1993; Jiang et al., 1994). The 16 known dierentially regulated genes included genes involved in stem cell proliferation (GOS-19-1, GOS-19-3), the cytoskeleton (vimentin), cytokine signaling and response (IFP-53, interleukin-8, ISG-56, ISGF-3 p91/84) and RNA processing (hnRNP core protein A1). Of the seven originally novel cDNAs, termed melanoma dierentiation associated (mda), mda-2 is homologous to the germ-cell-speci®c transcriptional repressor Tctex-1 (Roux et al., 1994; O'Neill and Artzt, 1995), mda-4 is a novel member of the human interferon-inducible gene family associated with

Monitoring gene expression during melanoma differentiation F Huang et al

control of tumorigenicity in a model of human melanoma (DeYoung et al., 1997), mda-6 is the universal cyclin dependent kinase inhibitor p21 (Jiang and Fisher, 1993; Xiong et al., 1993; Jiang et al., 1996a), mda-7 is a novel cancer growth suppressor gene (Jiang et al., 1995c, 1996b; Su et al., 1998b), mda-9 is a dierentiation associated and interferon gamma inducible gene (Lin et al., 1996, 1998) and mda-1 and mda-5 (Jiang and Fisher, 1993) still remain novel with unknown functions. The isolation of a high proportion of novel genes related to growth control and tumor suppression suggests that the DISH library could provide a valuable resource for the identi®cation and cloning of additional genes of importance to cancer development and progression. We have presently screened *10% (1000 random cDNA clones) of the TSS DISH cDNA library using high throughput robotic screening of immobilized cDNAs on glass chips (Ramsey, 1998; Marshall and Hodgson, 1998) to identify genes with elevated expression as a consequence of treatment of HO-1 cells with IFN-b+MEZ. These strategies have resulted in the identi®cation of both 26 known genes and 11 novel dierentially expressed genes with potential relevance to human melanoma growth, dierentiation and tumorigenicity. Treatment of HO-1 human melanoma cells with the combination of IFN-b+MEZ for 24 h followed by growth for 72 h in the absence of these inducers results in an irreversible loss of growth potential and terminal cell dierentiation in 495% of the HO-1 cell population (Fisher et al., 1985; Jiang et al., 1993; Kang et al., 1998). On the basis of this observation, it is hypothesized that the critical changes in gene expression that mediate terminal dierentiation are induced within the ®rst 24 h of treatment with IFN-b+MEZ (Fisher et al., 1985; Jiang and Fisher, 1993; Jiang et al., 1993, 1994). To de®ne those changes in gene expression that correlate with and that may provoke these modifications in the growth and physiology of human melanoma cells we are using a dierentiation induction subtraction hybridization scheme, the DISH approach (Jiang and Fisher, 1993). This strategy uses a temporally spaced (2 through 24 h) subtracted IFN-b+MEZ treated HO-1 (TSS) cDNA library to identify cDNAs that may vary in their expression rates during the process of terminal dierentiation, i.e., increases occur with distinct temporal kinetics. One thousand random clones have been isolated from this TSS cDNA library and analysed using the Synteni microchip cDNA array technology (Shalon et al., 1996) (Figure 1). This approach permits the sensitive and rapid identi®cation of dierentially expressed cDNAs and an evaluation of the temporal kinetics of expression of DISH genes with potential relevance to growth control and dierentiation in human melanoma cells. The Synteni approach provides an accurate and systematic evaluation of the magnitude of gene expression changes between two target RNA populations (Shalon et al., 1996; Schena et al., 1995, 1996; Heller et al., 1997). A limitation of the Synteni procedure is the requirement for cDNAs5500 bp (Marshall and Hodgson, 1998). The cDNAs to be analysed are microarrayed on a platform glass chip that permits screening with ¯uorescent probes prepared from dierent polyA+ RNA samples. In our studies RNA were prepared from untreated logarithmically growing

HO-1 cells (control) or HO-1 cells temporally treated with IFN-b (2000 units/ml), MEZ (10 ng/ml) or IFNb+MEZ (2000 units/ml+10 ng/ml). Using dual channel ¯uorescence, it was possible to compare relative temporal RNA expression between control and differentiation inducer treated cells. Thirty-seven DISH cDNA clones displaying 5®vefold increase in expression following treatment of HO-1 cells with IFNb+MEZ have been identi®ed using the Synteni procedure (Table 1). By probing the same samples with labeled polyA+ RNA from HO-1 cells treated with IFN-b or MEZ alone, the Synteni approach permits an automated analysis and identi®cation of genes upregulated by speci®c growth and dierentiation modulating agents and the temporal kinetics of the gene expression changes (data not shown). The Synteni scheme permitted the identi®cation of a number of known genes, including cDNAs that are interferon inducible (IFN 6 ± 16, interferon-inducible 71 kD 2'5' A synthetase, interferon-inducible 56 K protein, interferon induced cellular resistance mediator protein interferon-inducible peptide gene and interferon-inducible protein p78), immune system associated (MHC Class I HLA-C-a-2-chain, MHC-HLA-B27, MHC-HLA-BW62, MHC-HLA-B and Ia-associated invariant g-chain gene), dierentiation and growth regulated (p21 (mda-6), mda-7 and heat shock protein-70), transcription regulating (transcription factor ISGF-3 and clk-2 kinase), cytoskeleton related (®bronectin), mitochondrial associated (mitochondrial NADH dehydrogenase subunit 2, mitochondrial cytochrome oxidase subunit II, mitochondrial genes for tRNA and mitochondrial genomic DNA) and ®ve miscellaneous sequences (without obvious relationship to the growth arrest or dierentiation program of HO1 cells) (Table 1). Redundancy in gene identi®cation was apparent using the Synteni approach, including the independent isolation of two interferon-inducible 56 K protein clones, four MHC Class I HLA-C-a-2 chain clones, 11 transcription factor ISGF-3 clones, nine ®bronectin clones, 52 mitochondrial associated clones and six monocarboxylate transporter monolog MCT-6 clones (Table 1). In addition, 11 novel DISH genes upregulated by the combination of IFN-b+MEZ were identi®ed without homology to currently reported sequences in various DNA databases (Table 1). The redundancy in gene identi®cation can be reduced by prescreening for overlapping clones prior to preparing cDNA microarray grids (data not shown). The Synteni approach can be used to de®ne the temporal kinetics of regulation of DISH genes as a function of treatment with IFN-b+MEZ (Figure 2). Two basic patterns of altered gene expression are found in HO-1 cells treated with the terminal dierentiation inducing agents. These include, genes displaying maximum enhancement within the ®rst 12 h of treatment and genes exhibiting maximum induction or enhancement after 24 or 48 h treatment. Genes such as DISH-548 (novel), DISH-516 (IFN 6 ± 16 gene) and DISH-340 (ISGF-3) respond with a rapid kinetics of induction that is apparent within the ®rst 12 h of treatment. In the case of DISH-548 (novel), expression is maximum (8.8-fold increase over control) at 12 h posttreatment, expression decreases by 24 h (2.8-fold increase over control) and remains elevated (2.5-fold increase over control) for an additional 48 h. For

3547


3548

Figure 1 An example of microarray analysis of several DISH cDNA clones using the Synteni protocol. Temporally spaced cDNA libraries were constructed as previously described using pooled mRNAs from control actively proliferating HO-1 cells collected 2, 4, 8, 12 and 24 h after a DMEM-10 medium change or HO-1 cells grown for 2, 4, 8, 12 and 24 h with DMEM-10 containing IFNb+MEZ (2000 units/ml+10 ng/ml) (Jiang and Fisher, 1993). Subtraction hybridization between the control temporal cDNA library and the IFN-b+MEZ temporal cDNA library resulted in the creation of a TSS dierentiation inducer treated cDNA library, the DISH cDNA library (Jiang and Fisher, 1993). The DISH cDNA library was plated on LB-ampicillin plates with IPTG and X-Gal. White colonies were randomly selected and grown in 100 ml LB-ampicillin medium at 378C in 96-well microtiter plates. Five ml of cultures grown overnight were diluted to 50 ml and denatured at 958C for 5 min. Five ml was used as a template in a PCR reaction to amplify cDNA clone inserts using 5' amino terminus modi®ed primers. PCR products were veri®ed by gel electrophoresis and puri®ed with Qiaquick 96-well puri®cation kit (Qiagen). PCR reactions were used to amplify DISH cDNA clone inserts. Puri®ed PCR products with insert sizes of 5500 bp were immobilized on glass substrates using previously described protocols (Schena et al., 1995, 1996; Shalon et al., 1996). Poly(A)+ RNA were isolated from control and IFN-b+MEZ treated HO-1 cells using Ambion Micro poly(A) kits (Ambion). The ¯uorescent probes were synthesized using the GEMbright kit (Synteni, CA, USA). Two hundred ng of Poly(A)+ RNA was used in each reaction. Following the reverse transcriptase step, the RNA template was degraded by adding 2.5 ml of 0.5 M NaHC03, pH 9.2 and heating at 858C for 20 min. The probes were puri®ed using two successive gel ®ltration spin columns (Clonetech). The Cy3 and Cy5-labeled probes were pooled and used for hybridization. (a) The cDNA DISH clones were hybridized with Cy3 or Cy5 labeled poly(A)+ RNA isolated from temporally spaced control or IFN-b+MEZ treated HO-1 cells, respectively. Changes in the degree of ¯uorescence between RNA samples were then quantitated. (b) Documents the foldincrease in mRNA expression of DISH cDNA clones in HO-1 cells treated with IFN-b+MEZ versus untreated control cultures

DISH-516 (IFN 6 ± 16 gene), signi®cant elevated expression (4.4-fold over control) is apparent by 4 h, peak stimulation (13-fold increase over control) occurs at 12 h posttreatment, and reduced expression versus control is apparent by 72 h posttreatment. DISH-340 (ISGF-3) exhibits a narrow window of stimulation in HO-1 cells with upregulation being apparent only at 8

(6.8-fold increase versus control), 12 (7.7-fold increase versus control) and 24 h (5.2-fold increase versus control). Other genes, such as DISH-550 (novel), DISH-529 (novel), DISH-1496 (novel), DISH-170 (®bronectin) and DISH-241 (MCT-6) display optimum induction 24 or 48 h posttreatment. Maximum increase in expression of DISH-525 (novel) is apparent by 24 h


Table 1 DISH genes displaying elevated expression in HO-1 human melanoma cells induced to terminally dierentiate by treatment with IFN-b+MEZ Redundancya

DISH genes

Identity

DISH-35 DISH-174 DISH-180 DISH-303 DISH-516 DISH-933

Interferon-inducible interferon-inducible-mRNA (cDNA6 ± 16) interferon-inducible 71 kD 2'5' A synthetase interferon-inducible 56K protein interferon-induced cellular resistance mediator protein interferon-inducible peptide (6 ± 16) gene interferon-inducible protein p78

1 1 2 1 1 1

DISH-26 DISH-37 DISH-1428 DISH-1370 DISH-88

Immune system associated MHC Class I HLA-C-a-2 chain MHC-HLA-B27 MHC-HLA-BW62 MHC-HLA-B Ia-associated invariant g-chain gene

4 1 1 1 1

DISH-1883 DISH-2229 DISH-222

Dierentiation and growth regulated p21 (mda-6) mda-7 heat shock protein-70

1 1 1

DISH-340 DISH-950

Transcription regulation and signal transduction transcription factor ISGF-3 clk-2 kinase

DISH-170

Cytoskeleton related ®bronectin

DISH-287 DISH-39 DISH-149 DISH-140

Mitochondrial associated mitochondrial NADH dehydrogenase subunit 2 mitochondrial cytochrome oxidase subunit II mitochondrial genes for tRNA mitochondrial genomic DNA

4 25 5 18

DISH-169 DISH-168 DISH-668 DISH-241 DISH-244

Miscellaneous Elongation factor 1-a insulin induced protein I KM-102-derived reductase-like factor monocarboxylate transporter homolog (MCT-6) PMS2 related gene

1 1 1 6 1

DISH-192 (1) DISH-520 (1) DISH-990 (1) a

DISH-218 (1) DISH-529 (1) DISH-1431 (1)

11 1 9

Novel DISH genes DISH-232 (1) DISH-548 (1) DISH-1496 (1)

DISH-338 (1) DISH-550 (1)

Redundant clone: indicates the number of independent clones with the same sequence

posttreatment and maximum elevated expression of DISH-550 (novel), DISH-1496 (novel) and DISH-170 (®bronectin) results 48 h posttreatment. The strategy used to identify melanoma differentiation associated (mda) genes employs a combination of two inducing agents, IFN-b and MEZ (Fisher et al., 1985; Jiang and Fisher, 1993). Based on this protocol, it is anticipated that four classes of mda genes will be identi®ed. These include Type I mda genes upregulated by IFN-b and IFN-b+MEZ, Type II mda genes upregulated by MEZ and IFN-b+MEZ, Type III mda genes upregulated by IFN-b, MEZ and IFNb+MEZ and Type IV mda genes upregulated predominantly by the combination of IFN-b+MEZ (Jiang and Fisher, 1993; Jiang et al., 1994). Using the Synteni approach, three of the four dierent classes of mda genes have been identi®ed. The representative Northern blot of six DISH cDNAs (Figure 3) exempli®es this. DISH-516 (IFN 6 ± 16) and DISH529 (Novel) are representative of Type I mda genes with elevated expression being apparent after treatment with IFN-b or IFN-b+MEZ. DISH-5 (MCT-6) and

DISH-1496 (Novel) are representative of Type II mda genes that display increased expression after exposure to MEZ of IFN-b+MEZ. DISH-170 (®bronectin) and DISH-2229 (mda-7) represent Type IV mda genes with elevated expression predominantly in HO-1 cells treated with IFN-b+MEZ. With all six of these DISH genes, elevated expression versus control is apparent in HO-1 cells after both 24 and 48 h treatment with IFN-b+MEZ (Figures 2 and 3). In Figure 3, based on GAPDH hybridization, the level of RNA loaded from HO-1 cells treated for 48 h with IFN-b+MEZ (lane 6) is approximately 2.5-fold lower than the untreated 48 h control RNA sample (lane 5). Treatment of HO-1 human melanoma cells with the combination of IFN-b+MEZ results in terminal dierentiation with a concomitant loss of proliferative ability and tumorigenic potential in athymic nude mice (Fisher et al., 1985; Jiang and Fisher, 1993; Jiang et al., 1994). These ®ndings imply that the genes necessary for regulating growth, inducing terminal dierentiation and suppressing oncogenicity remain functional in these tumor cells. To de®ne the spectrum of genes

3549


3550

Figure 2 Temporal kinetics of expression of speci®c DISH genes. Using the Synteni approach (outlined in Figure 1), the relative level of expression of the various DISH genes was quantitated (as fold-increase over untreated cells) using RNAs isolated from temporally spaced (2, 4, 6, 12, 24, 48 and 72 h) control and IFN-b+MEZ (2000 units/ml+10 ng/ml) treated HO-1 cells

regulated during the process of commitment to and induction of irreversible growth arrest and terminal dierentiation in human melanoma cells we are using subtraction hybridization (DISH) combined with gene expression microarray cDNA microchip analysis. A TSS cDNA library was generated from HO-1 human melanoma cells treated with IFN-b+MEZ fro 2, 4, 8, 12 and 24 h (Jiang and Fisher, 1993). Since the ®rst 24 h after exposure to the inducing agents appear to be critical for induction and commitment to differentiation in HO-1 cells (Fisher et al., 1985; Jiang and

Fisher, 1993; Jiang et al., 1994, 1995b), this approach permits the direct identi®cation of dierentially expressed cDNAs associated with and potentially mediating terminal dierentiation and the associated phenotype changes in melanoma cells. In the present study, 1000 random clones isolated from HO-1 TSS cDNA library representing *10% of the library, were screened and evaluated by high throughput cDNA microarray (Synteni) (Shalon et al., 1996; Schena et al., 1995, 1996; Heller et al., 1997) analysis for upregulation by INF-b+MEZ. This


Figure 3 Northern blot analysis of several known and novel DISH cDNAs as a function of treatment with dierentiation and growth modifying agents. HO-1 cells were cultured in Dulbecco's modi®ed Eagle's medium (DMEM) supplemented with 10% fetal bovine serum and penicillin/streptomycin (100 U/100 mg/ml) at 378C in a 5% CO2 95% air-humidi®ed incubator. Cells were untreated (control) (lane 1) or treated for 24 h with IFN-b (2000 units/ml) (lane 2), MEZ (10 ng/ml) (lane 3) or IFNb+MEZ (2000 units/ml+10 ng/ml) (lane 4). Alternatively, RNA was isolated from untreated HO-1 control (lane 5) or HO-1 cells treated for 48 h with IFN-b+MEZ (2000 units/ ml+10 ng/ml) (lane 6). Total RNA was isolated, electrophoresed (15 mg/lane), and transferred to nylon membranes. Filters were hybridized with random printed 32P-dCTP-labeled DISH cDNA fragments. Membranes were washed and exposed for autoradiography. Probing with GAPDH was used to quantify RNA loading. The quantity of RNA in lane 6 is approximately 2.5-fold lower than in lane 5

approach resulted in the cloning of both known genes (26) and novel genes (11) displaying expression in HO1 human melanoma cells committed to terminally dierentiate by treatment with IFN-b+MEZ. These results suggest that the DISH approach combined with automated microarray cDNA screening will permit the identi®cation and cloning of the complete spectrum of genes in the TSS cDNA library that are dierentially expressed in human melanoma cells as a consequence of induction of loss of tumorigenic potential, irreversible growth arrest and terminal cell differentiation. In this context, the DISH strategy represents a sensitive approach for de®ning genes associated with and potentially regulating complex phenotypic changes in tumor cells, such as growth control, dierentiation and the cancer phenotype. Based on an initial evaluation of 70 random clones from the HO-1 TSS cDNA library, resulting in the identi®cation and cloning of 23 dierentiation regulated genes (Jiang and Fisher, 1993; Jiang et al., 1994), it was predicted that a number of genes previously identi®ed in this library should also be recognized using the Synteni cDNA microarray approach. This was the case as indicated by the subsequent identification of ISG-56 (two independent clonal isolates), ISGF-3 (11 independent isolates), p21 (mda-6) (one

clonal isolate) and mda-7 (one clonal isolate) in the HO-1 TSS cDNA library (Table 1). Additionally, previous studies indicated that ®bronectin expression in HO-1 melanoma cells is upregulated following treatment with IFN-b+MEZ (Su et al., 1995) and nine independent ®bronectin cDNA clones were identi®ed using the Synteni screening approach (Table 1). Temporal kinetic studies provide direct documentation of the validity of our TSS cDNA library screening strategy for identifying and cloning genes displaying dierential patterns of expression during the various phases of the dierentiation process (Figure 2). If a single time point had been chosen for constructing libraries and generating a subtraction library then the probability of identifying speci®c dierentially expressed cDNA clones with restricted expression patterns would be diminished signi®cantly. For example, if a single time-point was chosen for constructing TSS cDNA libraries the likelihood of cloning genes such as DISH-548 (novel), DISH-340 (ISGF-3), DISH-516 (IFN 6 ± 16) and mda-9 (Lin et al., 1996) would be greatest in subtraction libraries prepared from cultures treated for 8 or 12 h with IFNb+MEZ, whereas the cloning of genes such as DISH550 (novel), mda-7 and DISH-170 (®bronectin) would be greater using libraries prepared from cultures treated for 24 or 48 h with IFN-b+MEZ. The present ®ndings support the concept of using TSS cDNA libraries when one is attempting to identify and clone dierentially expressed genes, which display complex patterns of temporal expression. In fact, this appears to be the situation for all of the DISH genes analysed to date, i.e., the range of gene expression change varies as a function of time after initial exposure to IFNb+MEZ (Figure 2 and data not shown). Perhaps the most important and pertinent questions that remain to be answered are the roles of the numerous dierentially expressed DISH genes in the process of melanoma growth, dierentiation and tumorigenesis. With notable exceptions, such as tyrosinase (the rate limiting enzyme in melanin synthesis), it is not possible a priori to know if a speci®c DISH gene is associated with or causative of a distinct cellular phenotype in human melanoma cells treated with IFN-b+MEZ. Although the present studies cannot directly address these signi®cant issues, they do provide a starting point for further analyses. In the case of melanoma dierentiation, this process can be divided into three contiguous stages, induction of dierentiation, commitment to terminal dierentiation and maintenance of terminal dierentiation (Jiang et al., 1993; Kang et al., 1998). Genes that display altered kinetics within the ®rst 24 h of treatment with the dierentiation inducers may represent genes of importance in the process of induction and commitment to terminal dierentiation, which occurs within this time window. Possible genes serving this role could include DISH-548 (Novel), DISH-340 (ISGF-3) and DISH-516 (IFN 6 ± 16 gene). Genes displaying elevated expression predominantly in the middle range of the dierentiation process, e.g., after 24 to 48 h treatment, followed by a decline in expression may represent transition genes necessary for conversion of committed melanoma cells to a terminally dierentiated state. Genes possibly providing this function could include

3551


3552

DISH-241 (MCT-6) and DISH-550 (novel). In contrast, those genes that remain elevated in terminally dierentiated cells, even at 72 or 96 h posttreatment with the inducing agents, may represent genes that are relevant to maintenance of terminal cell dierentiation. Potential genes serving this function could include DISH-525 (novel), DISH-170 (®bronectin), DISH-1883 (p21, mda-6) and DISH-2229 (mda-7). A number of approaches can be used to address the potential functional relevance of the various DISH genes in human melanoma growth, dierentiation and tumorigenicity. If any or all of these genes are direct regulators of dierentiation, growth or the tumorigenic process then ectopically expressing these genes, using sense or inducible sense expression constructs (Jiang et al., 1995a,c, 1996b, 1997), or blocking expression of these genes, using antisense or ribozyme based technologies (Alama et al., 1997; Su et al., 1998a), may directly modify these phenotypes in melanoma cells. Further studies designed to characterize the presently identi®ed DISH genes and to identify the

remaining gene expression changes regulated as a function of induction of de®ned phenotypic changes in human melanoma cells are well warranted. These investigations will not only provide a more precise understanding of the factors that regulate growth, dierentiation and oncogenic potential of human melanoma cells, but they will also identify potentially important novel genes and their products than can serve as targets for pharmacologically modifying cancer growth and aggressiveness.

Acknowledgments This research was supported in part by National Cancer Institute Grant CA35675, a sponsored research award from GenQuest Incorporated, the Samuel Waxman Cancer Foundation and the Chernow Research Endowment. PB Fisher is the Michael and Stella Chernow Urological Cancer Research Scientist in the Departments of Neurosurgery, Pathology and Urology.

References Alama A, Barbieri F, Cagnoli A and Schettini G. (1997). Pharmacol. Res., 36, 171 ± 178. De Young KL, Ray ME, Su YA, Anzick SL, Johnstone RW, Trapani JA, Meltzer PA and Trent JM. (1997). Oncogene, 15, 453 ± 457. Fisher PB. (1984). Tumor Promotion and cocarcinogenesis In Vitro, Mechanisms of Tumor Promotion. Slaga TJ (ed). CRC Press Inc.: Boca Raton, Florida, pp. 57 ± 83. Fisher PB, Prignoli DR, Hermo Jr H, Weinstein IB and Pestka S. (1985). J. Interferon Res., 5, 11 ± 22. Guarini L, Graham GM, Jiang H, Ferrone S, Zucker S and Fisher PB. (1992). Pigment Cell Res. Suppl., 2, 123 ± 131. Heller RN, Schena M, Chai A, Shalon D, Bedillon T, Gilmore J, Wooley DE and Davis RW. (1997). Proc. Natl. Acad. Sci. USA, 94, 2150 ± 2155. Jiang H and Fisher PB. (1993). Mol. Cell Dierent., 1, 285 ± 299. Jiang H, Lin J and Fisher PB. (1994). Mol. Cell. Dierent., 2, 221 ± 239. Jiang H, Lin JJ, Su Z-z and Fisher PB. (1996a). Mol. Cell. Dierent., 4, 67 ± 89. Jiang H, Lin JJ, Su Z-z, Goldstein NI and Fisher PB. (1995c). Oncogene, 11, 2477 ± 2486. Jiang H, Lin J, Su Z-z, Kerbel RS, Herlyn M, Weissman RB, Welch D and Fisher PB. (1995a). Oncogene, 10, 1855 ± 1864. Jiang H, Lin JJ, Tao J and Fisher PB. (1997). Oncogene, 14, 473 ± 480. Jiang H, Lin J, Young S-m, Goldstein NI, Waxman S, Davila V, Chellappan SP and Fisher PB. (1995b). Oncogene, 11, 1179 ± 1189. Jiang H, Su Z-z, Boyd J and Fisher PB. (1993). Mol. Cell. Dierent., 1, 41 ± 66.

Jiang H, Su Z-z, Lin JJ, Goldstein NI, Young CSH and Fisher PB. (1996b). Proc. Natl. Acad. Sci. USA, 93, 9160 ± 9165. Kang D-c, Motwani M and Fisher PB. (1998). Intl. J. Oncology, in press. Lin JJ, Jiang H and Fisher PB. (1996). Mol. Cell. Dierent., 4, 317 ± 333. Lin JJ, Jiang H and Fisher PB. (1998). Gene, 207, 105 ± 110. Marshall A and Hodgson J. (1998). Nature Biotech., 16, 27 ± 31. O'Neill MJ and Artzt K. (1995). Development, 12, 561 ± 568. Ramsay G. (1998). Nature Biotech., 16, 40 ± 44. Roux AF, Rommens J, McDowell C, Anson-Cartwright L, Bell S, Schappert K, Fishman GA and Musarella M. (1994). Hum. Mol. Genet., 3, 257 ± 263. Schena M, Shalon D, Davis RW and Brown PO. (1995). Science, 270, 368 ± 369. Schena M, Shalon D, Heller R, Chai A, Brown PO and Davis RW. (1996). Proc. Natl. Acad. Sci. USA, 93, 10614 ± 10619. Shalon D, Smith JS and Brown PO. (1996). Genome Res., 6, 639 ± 645. Su Z-z, Goldstein NI and Fisher PB. (1998a). Proc. Natl. Acad. Sci. USA, 95, 1764 ± 1769. Su Z-z, Jiang H, Waxman S, Goldstein NI and Fisher PB. (1995). Mol. Cell. Dierent., 3, 225 ± 239. Su Z-z, Marireddi MT, Lin JJ, Young CSH, Kitada S, Reed JC, Goldstein NI and Fisher PB. (1998b). Proc. Natl. Acad. Sci. USA, 95, 14400 ± 14405. Waxman S. (1995). Dierentiation Therapy. Waxman S (ed). Serono Symposium Publications, Rome, Italy. Xiong Y, Hannon GJ, Casso D, Kobayashi R and Beach D. (1993). Nature, 366, 701 ± 704.