Advertisement

Melanocyte and Keratinocyte Carcinogenesis: p53 Family Protein Activities and Intersecting mRNA Expression Profiles

      Melanocytes and keratinocytes were analyzed for potential roles of p53, p73, and p63 tumor suppressor family proteins and of malignancy-specific gene expression changes in the etiology of multi-step cancer. Melanocytes expressed ΔNp73α, two p63 isoforms and p53. Although p21 and Noxa mRNA levels increased following DNA damage, p53 family member binding to p21 and Noxa DNA probes was undetectable, suggesting p53 family-independent responses. In contrast, keratinocytes expressed multiple isoforms each of p73 and p63 that were induced to bind p21 and Noxa DNA probes after ionizing (IR) or after ultraviolet B (UVB) irradiation, correlating with p21 and Noxa mRNA induction and with apoptosis. Interestingly, IR-resistant malignant melanocytes and keratinocytes both exhibited Noxa mRNA induction after UVB treatment, correlating with DNA binding of p53 family proteins to the Noxa probe only in keratinocytes. To uncover other malignancy-specific events, we queried mouse initiated keratinocyte clones for early changes that were exacerbated in malignant derivatives and also differentially expressed in human advanced melanoma versus normal melanocytes. Using a new method for ranking and normalization of microarray data for 5000 probe sets, 27 upregulated and 13 downregulated genes satisfied our query. Of these, the majority was associated with late-stage human cancers and six were novel genes. Thus, clonal lineage mouse models representing early through late cancer progression stages may inform the focus on early, potentially causal events from microarray studies of human cancers, facilitating prognosis and molecular therapy.

      Keywords

      IR
      ionizing radiation
      PRNP
      positional ranking and normalization procedure
      SCC
      squamous cell carcinoma
      UVB
      ultraviolet B
      A fraction of non-melanoma (less than 0.5%) and melanoma (less than 13%) or approximately 10,600 skin cancers per year turn deadly and, at present, we have no way to distinguish them. In the spirit of the Montagna Symposium 2004 theme “Keratinocyte and melanocyte cancers of skin: interactive pathways,” we compared pathways to cancer in non-melanoma and melanoma skin cancers in two respects: (1) p53 family member proteins' presence and activities within in vitro/in vivo models in the mouse and (2) global gene expression profiles between a mouse squamous cell carcinoma (SCC) carcinogenesis model and human melanoma. With the underlying hypothesis that the deadliest examples of each cancer type might have certain genes or pathways in common, we explored the possibilities for common pathways in cancer development between these different skin cell types and used mouse models encompassing early and late stages of cancer to inform selection of candidate genes from microarray analysis of late-stage human cancer.

      p53 Family in Melanocytes versus Keratinocytes: Introduction

      The p53 gene is generally wild-type in human melanomas of skin. Whereas p53 genes can be mutant in up to a quarter of advanced human melanomas, recent functional studies of p53 mutant protein activity in various tumors have revealed that over 70% of the mutations in the p53 gene naturally occurring in human melanoma retain ≥50% of wild-type p53 protein function (
      • Soussi T.
      • Kato S.
      • Levy P.P.
      • Ishioka C.
      Reassessment of the tp53 mutation database in human disease by data mining with a library of tp53 missense mutations.
      ). Yet, inactivation of the p53 pathway is believed to be required in essentially all human cancers, and studies of mouse melanoma models indicate that p53 pathway inactivation is required for melanoma even at early stages. Besides p53 gene mutation complementary to Ha-Ras mutations, inactivation of p53 protein can occur by INK4A loss and the resulting increase of the Mdm2 protein that degrades p53 (
      • Bardeesy N.
      • Bastian B.C.
      • Hezel A.
      • Pinkel D.
      • DePinho R.A.
      • Chin L.
      Dual inactivation of rb and p53 pathways in ras-induced melanomas.
      ). INK4A/ARF and CDK4 loss or mutation are prevalent in sporadic human melanoma and are responsible for 25%–40% of hereditary disposition in human melanoma patients (
      • Ghiorzo P.
      • Bianchi Scarr G.B.
      Genetics of melanoma susceptibility.
      ). Loss of APAF gene expression is a common mechanism of inactivating the p53 effector (downstream) signaling pathway in melanoma (
      • Soengas M.S.
      • Capodieci P.
      • Polsky D.
      Inactivation of the apoptosis effector APAF-1 in malignant melanoma.
      ). The discovery of other p53 family genes p63 and p73, each with multiple isoforms and activities, and their regulation of many of the same downstream genes as p53, has opened other avenues for modulation of p53 activities. Isoform expression and activities in a given cell type are particularly important because of multiple levels of potential regulation among p53 family proteins. TA isoforms are transcriptionally active whereas ΔN isoforms lack an N terminal transcriptional activation domain (
      • Tuve S.
      • Wagner S.N.
      • Schittek B.
      • Putzer B.M.
      Alterations of delta-p 73 splice transcripts during melanoma development and progression.
      ). The ΔN isoforms of p73 mRNA are positively regulated by transcriptionally active p53 and p73 and negatively regulate p73, p63, and p53 TA isoforms, establishing a feedback regulation. Wild-type p63 and p73 isoforms can oligomerize with each other and with mutant p53 (
      • Chan W.M.
      • Siu W.Y.
      • Lau A.
      • Poon R.Y.
      How many mutant p53 molecules are needed to inactivate a tetramer?.
      ). As noted above, p73 and p63 genes are wild-type in a broad sampling of human tumors, including melanoma. As yet, there are few reports of p73 and p63 expression in melanoma. The isoform specificity of expression in human tumors is just beginning to be studied and none of these studies have investigated all three family members and their isoform expression in the same samples. In one report, total p63 was expressed very little in normal melanocytes and only in a small fraction (less than 2%) of human melanomas on a tissue microarray compared with almost 12% with p53 expression (
      • Brinck U.
      • Ruschenburg I.
      • Di Como C.J.
      Comparative study of p63 and p53 expression in tissue microarrays of malignant melanomas.
      ). The importance of p53 and p73 pathways in melanoma, however, is suggested by findings that apoptosis induced by oligonucleotide homologous to the telomere 3′ overhang sequence (T-oligo) in tumors from human melanoma cell lines acts through p53 and p73 (
      • Puri N.
      • Eller M.S.
      • Byers H.R.
      • Dykstra S.
      • Kubera J.
      • Gilchrest B.A.
      Telomere-based DNA damage responses: A new approach to melanoma.
      ).

      p53 Family in Melanocytes versus Keratinocytes: Results and Discussion

      We sought to determine the status of three p53 family member proteins, their DNA-binding activities, and p53 downstream gene expression in melanocytes and melanoma, compared with a keratinocyte carcinogenesis model. Mouse models of melanoma and SCC development provide abilities for manipulation of gene expression and sampling of tissue stages during multi-step cancer development in vitro and in vivo, which is not possible in humans Figure 1. The keratinocyte model used in this study is comprised of clonally derived non-transformed, initiated and malignant cell derivatives (SCC) from mouse epidermis (
      • Kulesz-Martin M.
      • Yoshida M.A.
      • Prestine L.
      • Yuspa S.H.
      • Bertram J.S.
      Mouse cell clones for improved quantitation of carcinogen-induced altered differentiation.
      ,
      • Kulesz-Martin M.
      • Blumenson L.
      • Lisafeld B.
      Retinoic acid enhancement of an early step in the transformation of mouse epidermal cells in vitro.
      ,
      • Kulesz-Martin M.F.
      • Penetrante R.
      • East C.J.
      Benign and malignant tumor stages in a mouse keratinocyte line treated with 7,12-dimethylbenz[à]athracene in vitro.
      ). These cells lack Ha-Ras mutations (
      • Schneider B.L.
      • Bowden G.T.
      • Sutter C.
      • Schweizer J.
      • Han K.A.
      • Kulesz-Martin M.F.
      7,12–dimethylbenz[a]anthracene-induced mouse keratinocyte malignant transformation independent of Harvey ras activation.
      ) and have wild-type p53 genes, as do at least half of human cancers. Although the p53 gene remains wild-type, p53 protein function is defective at malignant conversion (
      • Knights C.D.
      • Liu Y.
      • Appella E.
      • Kulesz-Martin M.
      Defective p53 post-translational modification required for wild type p53 inactivation in malignant epithelial cells with mdm2 gene amplification.
      ). The non-transformed, transformed, and malignant melanocyte lines were independently derived from transgenic mouse lines as reported previously (
      • Broome P.M.
      • Gause P.R.
      • Hyman P.
      • Gregus J.
      • Lluria-Prevatt M.
      • Nagle R.
      • Bowden G.T.
      Induction of melanoma in tpras transgenic mice.
      ;
      • Gause P.R.
      • Lluria-Prevatt M.
      • Keith W.N.
      • Balmain A.
      • Linardopolous S.
      • Warneke J.
      • Powell M.B.
      Chromosomal and genetic alterations of 7,12-dimethylbenz[a]anthracene-induced melanoma from tp-ras transgenic mice.
      ). The malignant cells were derived from transgenic mice expressing mutant human Ha-Ras genes in melanocytes. The tumors in the melanoma model also lack p16 (INK4A) expression (
      • Gause P.R.
      • Lluria-Prevatt M.
      • Keith W.N.
      • Balmain A.
      • Linardopolous S.
      • Warneke J.
      • Powell M.B.
      Chromosomal and genetic alterations of 7,12-dimethylbenz[a]anthracene-induced melanoma from tp-ras transgenic mice.
      ). We analyzed this melanoma model for expression of p53, p63, and p73 proteins, for the DNA-binding capacity of these proteins to probes containing the cell cycle-associated p21 and apoptosis-associated Noxa mouse promoter sequences and for upregulation of p21 protein and Noxa mRNA in response to ionizing radiation (IR) or ultraviolet B (UVB) light.
      Figure thumbnail gr1
      Figure 1Cellular relationships within the mouse clonal model of epidermal carcinogenesis and the mouse TP-Ha-ras melanoma model. Non-transformed keratinocyte clonal cell strain (291), derived from the epidermis of normal Balb/C newborn mice was treated with 7,12-dimethylbenz[α]anthracene (DMBA) to derive initiated clone 291.03C (03C) based on a defect in the terminal differentiation response to extracellular Ca2+. Clone 03C produced squamous cell carcinoma (SCC) in mice at low frequency, from which SCC-producing cell line 291.03R (03R) was derived. The 03R line produces skin SCC in 100% of grafts to athymic nu/nu mice and these metastasize to the lung and regional lymph nodes. YMM (non-transformed) melanocytes were derived from the skin of 5-d-old yellow-coated mice on a C57BL/6 background. JMM (transformed) from TP-vjun mice, and 1984-1 melanoma from TP-Ras mice treated with DMBA.

      Expression of p53 family isoforms in mouse melanoma model

      The total steady-state levels of expression of p73, p63, and p53 proteins were detected by immunoblotting as shown in Figure 2. Non-transformed and initiated keratinocytes (initiated 03C is in lane 1 shown for comparison with the melanocyte model) expressed three isoforms of p73, transcriptionally active form TAp73α, dominant-negative ΔNp73α, and an unidentified lower molecular weight p73 isoform. Keratinocytes expressed four isoforms of p63, TAp63α, ΔNp63α, TAp63β (all identified by apparent molecular weight and isoform-specific RT-PCR) and a lower molecular weight form likely to be either a TAp63γ or ΔNp63β. Melanocytes expressed only one p73 isoform; dominant-negative ΔNp73α. Untreated YMM and transformed JMM (lanes 2 and 5) expressed ΔNp73α at similar basal levels. Untreated melanoma 1984-1 (lane 8) expressed elevated ΔNp73α compared with non-transformed YMM and initiated JMM melanocytes. Although p53 expression is elevated in about 25% of human melanomas, consistent with abnormal regulation of degradation of mutated or defective wild-type p53 proteins, elevated steady-state levels of wild-type p73 and p63 proteins have been found in a variety of human cancers (
      • Moll U.M.
      • Slade N.
      P63 and p73: Roles in development and tumor formation.
      ), consistent with the elevated expression seen in 1984-1 mouse melanoma cells. Melanocytes expressed two isoforms of p63, the TAp63β and either TAp63γ or ΔNp63β (yet to be determined by isoform-specific RT-PCR), and these were not elevated in melanoma 1984-1 cells. The predominant p63 form varied among the melanocyte lines, with the lower molecular weight form predominant in non-transformed YMM and melanoma 1984-1 and the higher molecular weight form predominant in transformed JMM. None of the p63 or p73 protein forms were induced by treatment with IR or UVB. p53 was expressed in YMM with little response to treatment (IR 1.7-fold untreated, UVB 1.3-fold). Basal p53 expression was reduced in transformed JMM and in melanoma 1984-1 relative to YMM but was restored weakly in response to IR in JMM (1.8-fold untreated JMM) and restored to at least normal expression in JMM and 1984-1 exposed to UVB (16-fold untreated JMM and 12-fold untreated 1984-1, respectively). A regulator of p53, Mdm2, has opposing effects on p53 (degradation) and p73 (stabilization) (
      • Ongkeko W.M.
      • Wang X.Q.
      • Siu W.Y.
      Mdm2 and mdmx bind and stabilize the p53-related protein p73.
      ). Although Mdm2 was not examined in the melanocytes in our study, increased levels or activity of Mdm2 in transformed or malignant melanocytes is one possible explanation for the observed increase in the basal levels of p73 protein correlated with decreased p53 protein in this model. The ability of IR or UVB to induce p53 in these cells suggests that a p53-stabilizing pathway in response to stress is retained in spite of reduced basal levels.
      Figure thumbnail gr2
      Figure 2p53 family expression in non-transformed YMM, transformed JMM, and melanoma cell line 1984-1. Melanocytes were cultured in RPMI 1640 plus 10% fetal bovine serum, and 1% antibiotic–antimycotic (Invitrogen, Carlsbad, California). YMM and JMM cells were further supplemented with 200 nM 12-O-tetradecanoyl-phorbol-13-acetate (TPA) and 200 pM cholera toxin. Cells were untreated (UT) or harvested 3 h following treatment with ionizing radiation (IR) (4 Gy), or 5 h following treatment with ultraviolet B radiation (UVB) (135 J per m2). Whole cell lysate aliquots of 40 μg (for p53 and p73) or 10 μg (for p63) were separated by 8% SDS-PAGE and transferred to nitrocellulose for immunoblotting as reported previously (
      • Wu Y.
      • Huang H.
      • Miner Z.
      • Kulesz-Martin M.
      Activities and response to DNA damage of latent and active sequence-specific DNA binding forms of mouse p53.
      ). Proteins were reacted with specific antibodies to p53 (pAb122), p63 (4A4 monoclonal from Santa Cruz, Santa Cruz, California, detects all isoforms), p73 (monoclonal 5B1288 from Imgenex (San Diego, California), detects all isoforms), or actin (C-2 -Santa Cruz) as loading control and visualized by chemiluminescence (ECL [Perkin Elmer, Boston, Massachusetts] Isoforms of p73 and p63, as well as p53 in melanocytes are shown in comparison with untreated initiated epidermal keratinocyte line 03C (lane 1). Putative p73 isoform identification is based upon apparent molecular weight alone (labeled in parentheses). p63 isoform identification is based, in addition, upon RT-PCR using primers specific to exon/intron junctions of α, β, and γ forms for the 3′ primer, and TA versusΔN specific junctions for the 5′ primers (Dennis Roop, personal communication) as follows: M63ΔN.F (TTGTACCTGGAAAACAATG), M63TA.F (TCGCAGAGCACCCAGACA), 63SEQ.2F (CCACCACAGGTTGGCACTG), 63SEQ.3F (CCAGATGATGAGCTGCTGTACC), M63-2R (GCATCGTTTCACAACCTCG), M634R (TAGTCCAGGCATGATGAG), M63-βR (CAGACTTGCCAAATCCTGAC), and M63γR (GGCTCCACAAGCTCATTC). To detect the shorter N terminal or C terminal p63 fragments, we used 40 cycles of 92°C for 30 s, 56°C for 1 min, and 72°C for 1 min with a 15-min extension time at 72°C. To detect the long fragments (i.e., TAα, TAβ, etc.), we used 40 cycles of 92°C for 1 min, 56°C for 2 min, and 72 °C for 3 min with a 15-min extension time at 72°C. Myc-tagged p63 plasmids of the various isoforms are used as positive controls for the RT-PCR reactions.
      In order to determine whether stabilization of total p53 family proteins leads to increased DNA binding and whether ΔN isoforms might negatively regulate transactivation of p53 downstream genes by binding to their promoters, we assessed the capability of the p53 family proteins for DNA binding to probes containing endogenous DNA sequences of the cell cycle-associated promoter p21 and apoptosis-associated promoter Noxa.

      DNA-binding activity of p53 family proteins in melanocytes

      In keratinocytes, all three p53 family members demonstrated activity for DNA binding to the p21 probe Figure 3. Selective binding of p63 to the Noxa probe was correlated with keratinocyte apoptosis (data not shown). In spite of the expression of ΔNp73α, TAp63β, TAp63γ or ΔNp63β and p53 in non-transformed YMM melanocytes, none of these endogenous proteins showed specific DNA-binding capability to p21 or Noxa DNA probes in the DNA-binding assays, either from untreated cells or cells treated with IR or UVB (data not shown). DNA binding of p53 protein to p21 but not Noxa probes was robust in JMM cells, unchanged after IR treatment, and elevated after UVB treatment (data not shown), consistent with total protein as shown in Figure 2. Although at least one isoform of all three p53 family members was present in the 1984-1 melanoma, only the dominant-negative ΔNp73α was detected specifically bound to DNA probes of downstream genes p21 and Noxa, and only in cells treated with IR (Figure 3, lanes 11 and 12, compared with non-specific binding to control DNA probe lacking p53-binding sites in lane 10). The occupation of target gene promoters by a dominant-negative form of p73 could inhibit activation of p21 or Noxa by IR in these cells. To address this issue we examined p21 and Noxa mRNA levels.
      Figure thumbnail gr3
      Figure 3DNA-binding capability of endogenous p53 family proteins from mouse melanoma 1984-1, and 03C keratinocyte cell lines assayed by DNA affinity immunoblotting (DAI). DAI of whole-cell lysates incubated with biotinylated DNA probe in vitro has been described (Liu et al, 2001;
      • Knights C.D.
      • Liu Y.
      • Appella E.
      • Kulesz-Martin M.
      Defective p53 post-translational modification required for wild type p53 inactivation in malignant epithelial cells with mdm2 gene amplification.
      ). Aliquots of 200 μg lysate from untreated (UNT) cells or cells harvested 3 h following treatment with ionizing radiation (IR) (4 Gy), or 5 h following treatment with ultraviolet B radiation (UVB) (135 J per m2) were incubated with specific biotinylated probe DNA sequences of the mouse p21 or noxa promoter or to probe sequences lacking p53-binding sites (Con). DAI probes were synthesized by PCR using a mouse keratinocyte genomic DNA template, and the primers indicated and probe sequences were verified as wild-type by sequencing at the OHSU Cancer Institute sequencing core facility.

      Activation of p53 downstream genes in melanocytes

      YMM cells demonstrated responses to both IR and UVB by induction of p21 mRNA as detectable by means of quantitative real-time PCR (Figure 4, left panel). Activation of these promoters in the cells would be consistent with lack of DNA binding of the negative regulatory forms of p73 or p63 in YMM cells to the p21 DNA probe. But, the induction of p21 mRNA in YMM cells by IR and by UVB occurred without DNA binding of any isoforms of p73, p63, or p53 in YMM. The lack of Noxa mRNA induction by IR or UVB in YMM cells correlated with lack of p53 family member DNA binding. Inactivity of the Noxa promoter in response to IR or UVB, however, was not because of negative regulation by ΔNp73 or ΔNp63 forms. The induction of p21 mRNA by IR without DNA-binding activity of a p53 family member in YMM cells, and the finding reported by Brian Nickoloff in this volume and by
      • Qin J.Z.
      • Stennett L.
      • Bacon P.
      P53-independent noxa induction overcomes apoptotic resistance of malignant melanomas.
      that notch transcription factor induces Noxa independent of p53 suggest that induction of Noxa is not only p53 independent in melanocytes but also p53 family independent.
      Figure thumbnail gr4
      Figure 4mRNA expression of p53 target genes p21 and Noxa RNA was isolated from YMM (control) and 1984-1 (melanoma) cells 24 h after treatment with ionizing radiation (IR) (4 Gy), or ultraviolet B radiation (UVB) (135 J per m2). The cDNA was generated using AMV reverse transcriptase (Roche, Indianapolis, Indiana) and random hexamer primers (Integrated DNA Technologies). Relative expression levels of p53 target genes p21 and Noxa were analyzed by quantitative real time PCR using SYBR-Green I reporter dye in a 7900HT thermocycler (Applied Biosystems Inc., Foster City, California) as described previously (
      • Horn E.J.
      • Albor A.
      • Liu Y.
      Ring protein trim32 associated with skin carcinogenesis has anti-apoptotic and e3-ubiquitin ligase properties.
      ). Gene-specific primers for p53 target genes were mouse p21 [(CCATGTCCAATCCTGGTGATG) and (CGAAGAGACAACGGCACACTT)] and mouse Noxa [(ACTGTGGTTCTGGCGCAGAT) and (TGAGCACACTCGTCCTTCAAGT)], and these data were normalized to expression levels of 18S RNA [(CGGCTACCACATCCAAGGAA) and (CCTGTATTGTTATTTTTCGTCACTACCT)]. Data are shown relative to gene expression levels in untreated YMM and untreated 1984-1 cells, respectively. Each sample was run in triplicate on three separate plates. Error bars depict standard deviation from the mean.
      Thus, DNA binding of ΔN forms of p73 or p63 is not required to negatively regulate activation of Noxa in the melanocytes. It is possible, however, that the dominant-negative forms ΔNp73α and potentially ΔNp63β that predominate in YMM negatively regulate Noxa or other p53 downstream genes without requiring complex formation with DNA. Further, it is possible that a necessary transcription factor for Noxa is absent. The lack of DNA binding by p73 or p63 isoforms in transformed melanocytes JMM, and lack of inducibility of the observed p53 DNA binding to the p21 probe in these cells are consistent with loss of function of p53 family members in JMM. This was confirmed by lack of induction of p21 or Noxa RNA expression after treatment of JMM cells with IR or UVB (data not shown).
      DNA-binding studies in 1984-1 melanoma supported regulation of p21 and Noxa transcription by p53 family members. Where ΔNp73α-specific binding to the p21 DNA probe was induced after IR treatment of 1984-1 melanoma cells (Figure 3, lane 11), p21 mRNA was not induced (Figure 4, right panel). Under the same conditions, ΔNp73α in YMM cells was not bound to p21 DNA probe and p21 RNA was induced (Figure 4, left panel). Consistent with negative regulation of Noxa and p21 response to IR in 1984-1 melanoma cells, ΔNp73α bound to Noxa and p21 DNA probes and neither p21 nor Noxa mRNA were induced by IR (Figure 4, right panel, second and 5th bar from left). Where p21 mRNA was strongly induced by UVB in 1984-1 melanoma cells (Figure 4, right panel, third bar from left), the ΔNp73α was not specifically bound to the p21 DNA probe (Figure 3, lane 15 indistinguishable from lane 14 control), and p53-specific binding to the p21 DNA probe was strongly induced (Figure 3, lane 15).

      Summary, Conclusions, and Future Studies of p53 Family in Melanocytes

      Keratinocytes and melanocytes of mouse cell culture models expressed different patterns of p53 family protein isoforms, consistent with the tissue specificity of isoform expression, as summarized for mouse keratinocyte and melanocyte models in Figure 5. Both cell types expressed ΔNp73α. In melanocytes, this was the only p73 isoform expressed, whereas keratinocytes also expressed a transcriptionally active form TAp73α and an additional unidentified lower molecular weight isoform. Both keratinocytes and melanocytes expressed TAp63β and a lower molecular weight isoform, either TAp63γ or ΔNp63β, but in YMM non-transformed and 1984-1 melanoma, this lower molecular weight form predominated, whereas TAp63α and TAp63β isoforms predominated (in that order) in keratinocytes. In initiated 03C keratinocytes, induction of p21 and Noxa mRNA correlated with induction or maintenance of p63 and p53 DNA binding to their respective DNA probes, suggesting p63- and p53-dependent transactivation of these genes in response to IR and UVB. Loss of induction of p63 and p53 DNA binding by IR in 03R SCC derivatives of 03C-initiated cells correlated with loss of respective mRNA induction and with failure to undergo growth arrest or apoptosis (data not shown). In both 1984-1 melanoma and 03R SCC derivatives, however, Noxa mRNA was induced by UVB. In the SCC-producing keratinocytes 03R, this correlated with p63 and p53 DNA binding to the Noxa probe and with apoptosis (data not shown). In the melanoma line, however, Noxa mRNA induction occurred without any p63 isoform or p53 protein binding to the Noxa DNA probe. UVB induction of Noxa mRNA may have been facilitated by lack of a dominant-negative ΔNp73α isoform DNA binding to the Noxa promoter. Although UVB induction of p21 RNA in YMM cells was independent of p53, p73, or p63 binding to DNA, UVB induction of p21 RNA in the 1984-1 melanoma cells correlated with the overall stabilization of p53 protein and DNA binding of p53 to the p21 probe.
      Figure thumbnail gr5
      Figure 5Schematic representation of the p53 family member response to ionizing radiation (IR) or ultraviolet radiation B (UVB) in 03C and 03R keratinocytes, and YMM and 1984-1 melanocytes. Reading from top to bottom, the schematic shows which p53 family member proteins are expressed before and after treatment with IR or UVB, followed by which p53 family members bind DNA in untreated cells or cells treated with IR or UVB (showing only those p53 family members that changed). The corresponding cellular responses in terms of p21 and noxa gene transactivation, and cell behavior follow. The complete p53 family member expression profile can be seen in for melanocytes and 03C keratinocytes. “All isoforms” for p73 include TAα, and ΔNα, and p63 includes TAα, ΔNα, TAβ, and (TAγ or ΔNβ). The 03R keratinocyte expression profile is similar to 03C except that p63 (all isoforms) are slightly reduced and p73 (all isoforms) are dramatically reduced.
      The UVB induction of Noxa in the UVB-treated 1984-1 melanocytes without detectable binding of endogenous p73, p63, or p53 proteins from melanocytes to the Noxa promoter sequence DNA probe is consistent with the findings of the Nickoloff laboratory that Noxa induction can be p53 independent in melanocytes (
      • Qin J.Z.
      • Stennett L.
      • Bacon P.
      P53-independent noxa induction overcomes apoptotic resistance of malignant melanomas.
      ). Further, the presence of p73, p63, and p53 in non-transformed YMM melanocytes without any detectable DNA binding to either the p21 or Noxa DNA probes leads us to conclude that p21 induction in these non-transformed melanocytes is not only p53 independent but also p73 and p63 independent.
      These studies show cell type specificity of p53 family proteins expressed in melanocytes at different stages of melanoma progression, the DNA-binding capability of the expressed isoforms for a cell cycle versus apoptosis-related promoter, and the responsiveness to IR or UVB radiation. Chromosome immunoprecipitation, however, will be necessary to determine specifically which p53 downstream promoters are occupied in cells by each p73, p63, and p53 isoform for selective regulation of p53 family targets. From the current results, it is possible that the dominant-negative forms ΔNp73α and potentially ΔNp63β predominating in melanocytes play a differential negative role in Noxa mRNA induction in YMM and/or that a necessary transcription factor for Noxa is absent. It is still possible, however, that p53 family members, such as TAp73 or TAp63 isoforms, play a role in the induction of these or other p53 downstream genes if expressed in melanocytes at other stages or conditions beyond the scope of this study.
      Both γ and δ p73 variants are reportedly expressed in human primary keratinocytes and different tumor cell lines, including melanoma (
      • De Laurenzi V.
      • Costanzo A.
      • Barcaroli D.
      Two new p73 splice variants, gamma and delta, with different transcriptional activity.
      ). The ΔN-p73 generated from the second promoter was the predominant isoform in benign nevi, similar to our findings in the YMM non-transformed cells. In the same study, however, TA-p73Δ isoform increases were associated with melanoma progression (
      • Tuve S.
      • Wagner S.N.
      • Schittek B.
      • Putzer B.M.
      Alterations of delta-p 73 splice transcripts during melanoma development and progression.
      ). The lack of induction of p73 by UVB in our study supports a previous report that expression of p73 and other cellular proliferation-related proteins was not changed by the UVB irradiation (
      • Zhang H.
      • Rosdahl I.
      Ultraviolet a and b differently induce intracellular protein expression in human skin melanocytes—a speculation of separate pathways in initiation of melanoma.
      ). The expression of ΔNp63 has been described in the basal cell compartment of squamous epithelium. The counterbalance between TAp63 and ΔNp63 isoforms appears to play an essential role in epidermal stratification and differentiation (
      • Yang A.
      • Kaghad M.
      • Wang Y.
      P63, a p53 homolog at 3q27–29, encodes multiple products with transactivating, death-inducing, and dominant-negative activities.
      ;
      • Koster M.I.
      • Kim S.
      • Mills A.A.
      • DeMayo F.J.
      • Roop D.R.
      P63 is the molecular switch for initiation of an epithelial stratification program.
      ). The occupation of target gene promoters by a dominant-negative form of p73 observed in the current study is an intriguing plausible mechanism for failure of IR to activate p21 or Noxa and for notorious melanoma IR resistance.

      Comparative Gene Expression Profiling of Cancer in Keratinocytes and Melanocytes: Introduction

      A global approach is needed to uncover changes in gene expression that characterize progression or make good prevention or therapeutic targets for melanoma or SCC. Because limited samples of human cancer progression are available, often late in the process, models of cancer development that encompass early as well as late stages are needed. These offer advantages to uncover potentially causative early changes that are also essential in late-stage disease, and thus potentially good targets for molecular therapy. We previously reported differences in gene expression between a non-transformed keratinocyte clone and two clonal derivatives: 03C, a precursor to SCC that metastasizes to lung and lymph node and 09C, a precursor to benign (papilloma) (
      • Wang Z.
      • Liu Y.
      • Mori M.
      • Kulesz-Martin M.
      Gene expression profiling of initiated epidermal cells with benign or malignant tumor fates.
      ). Since then, we have extended the study of gene expression in this keratinocyte clonal lineage model to the metastatic SCC derivative of 03C, 03R. As with most microarray experiments, however, selecting likely gene candidates for prediction of outcomes and potential molecular therapeutic targets is a challenging task.
      The results from two microarray experiments, one using the keratinocyte clonal lineage carcinogenesis model in the mouse (
      • Wang Z.
      • Liu Y.
      • Mori M.
      • Kulesz-Martin M.
      Gene expression profiling of initiated epidermal cells with benign or malignant tumor fates.
      ) and the other studying primary cultures of human melanoma (
      • Hoek K.
      • Rimm D.L.
      • Williams K.R.
      Expression profiling reveals novel pathways in the transformation of melanocytes to melanomas.
      ), were reanalyzed to isolate a subset of changes in common to melanoma and non-melanoma skin cancers. This reanalysis makes use of mouse models in which early stages are available to inform potential causative gene changes in human cancers for which only a single time point is generally available. Underlying our query design is the rationale that gene expression changes that occurred early (as can be detected in the mouse model) and were maintained or further increased/decreased in the malignant stage (in both mouse and human models) were more likely to be causal in nature. We posit that (1) as in the case of Bcr-Abl in chronic myelogenous leukemia in blood cancers (
      • Deininger M.
      • Buchdunger E.
      • Druker B.J.
      The development of imatinib as a therapeutic agent for chronic myeloid leukemia.
      ), these causal events make the best targets for novel molecular-targeted therapy development and (2) certain of these cancer initiating/causal events will occur in common in malignant skin cancers between keratinocytes and melanocytes.
      A new method of ranking and normalization of the genes from a microarray experiment was used to compare the mouse keratinocyte carcinogenesis model with human melanoma and normal melanocyte cell cultures. The new strategy includes steps for ranking genes and for normalizing data, and is thus called the “Positional Ranking and Normalization Procedure (PRNP).” A full description of this method will be presented in another publication (Pelz and Sears, in preparation).
      Pelz R, Kulesz-Martin M, and Sears, R: Non-linear normalization of microarray data using invariant probe sets. In preparation.
      Briefly, PRNP was implemented using “R” (
      • Team RDC
      R: A language and environment for statistical computing. R Foundation for Statistical Computing.
      ), an open source statistical computing environment. For each sample, the method first sorts the probe sets in ascending order based on the expression level reported by Affymetrix MAS 5.0 software. Each probe set is given a rank number based on its sorted order in the sample. Next, PRNP makes pairwise comparisons between samples by assigning positive or negative positional “shifts” to each probe set based on the difference in that probe set's rank between the two samples. Probe sets with small or no shifts represent the invariant genes between samples that we used to normalize the data. When calculating fold changes, we first normalize all samples to an idealized reference sample (normally the average of the control samples) using these invariant probe sets. Probe sets with large positive shifts represent the most upregulated genes, and probe sets with large negative shifts represent the most downregulated genes. When ranking the most up/downregulated genes between groups of biological replicates, a consensus of all possible pairwise comparisons between the biological replicates is used to produce a robust ranking.

      Comparative Gene Expression Profiling of Cancer in Keratinocytes and Melanocytes: Results and Discussion

      New aspects of this analysis of microarray data were combining the analysis of gene expression data from a genetically related mouse model of multi-step cancer with the analysis of human melanoma data, and using the PRNP to prioritize candidate genes. Application of the PRNP method to the mouse SCC model samples and the human melanoma data is illustrated in Figure 6. Up/downregulated genes were selected based on the magnitude of their positional shifts found using PRNP Figure 6a. The list of relevant up/downregulated genes was refined by requiring that genes be consistently up/downregulated in each mouse keratinocyte-initiated cellular precursor clone of SCC, and in each of the malignant SCC cellular derivatives, as well as being consistently up or downregulated in each human melanoma Figure 6b. After eliminating genome duplications between mouse and human and probe set redundancies, 27 genes were found to be upregulated and 13 downregulated according to all three of the comparisons above. The 40 retrieved genes that fit these criteria are shown in Table I. Of 27 upregulated genes, 13 are cancer associated as referenced in Pubmed. In addition, three of the remaining 27 genes interact with cancer-related proteins or are homologues of other cancer-related genes, resulting in 16 of 27 retrieved genes being cancer related. Five of the upregulated known genes have not been reported as cancer related. Six of the upregulated genes are unknown (indicated in Table I by difference in rank order from greatest to least). Of the 13 downregulated genes, all known genes, nine are referenced in other cancer-related publications.
      Figure thumbnail gr6
      Figure 6Schematic representation of the Positional Ranking and Normalization Procedure (PRNP) used to discover overlapping up- and downregulated genes in the mouse model of squamous cell carcinoma (SCC) and in human melanoma. A detailed description of this method will be presented in another publication (Pelz and Sears, in preparation). (A) PRNP is based on comparing the relative order of genes between two samples. As shown, the genes in a 291 (normal) sample and a 03C (initiated) sample are sorted based on expression level. Each gene in the comparison sample (03C) is given a rank offset by comparing its rank order in 03C with its rank order in 291. The genes are then sorted by these rank offsets to give downregulated genes on the left, upregulated genes on the right, and invariant genes in the middle (bottom schematic). One pairwise comparison is shown, but a consensus of all possible pairwise comparisons between biological replicates was used to produce a robust ranking of up/downregulated genes between 291 and 03C samples. (B) Venn diagram showing the strategy for finding overlapping upregulated genes between the early and late stages of the mouse model of SCC and the human melanoma data. PRNP as illustrated in (A) was applied to compare 291 to 03C, 03C to 03R and human normal to melanoma. The intersection of all three of the resulting ranked gene lists was determined. This resulted in a ranked list of genes that are upregulated in the early stage (03C) of the mouse model, further upregulated in the late stage of the mouse model (03R), and also upregulated in the human melanoma data. The same process was used to identify the ranked list of downregulated genes.
      Table IUp and downregulated genes in mouse-initiated keratinocyte precursors 03C and SCC derivatives 03R that are in common with up and downregulated genes in human melanoma cells, as extracted by the PRNP method
      Expression levelGene name
      Both known and unknown genes are listed. Genes are ordered based on PRNP ranking from greatest to least.
      Accession no.Fold change (291 to O3C)
      Fold changes are the average of three biological replicates.
      Fold change (291 to O3R)
      Fold changes are the average of three biological replicates.
      Fold change (normal to melanoma)Related cancersReferences
      UpregulatedPcolceBB2508112.716.451.5Uterine leiomyomata (UL), human brain tumor cells
      • Ligon A.H.
      • Scott I.C.
      • Takahara K.
      • Greenspan D.S.
      • Morton C.C.
      Pcolce deletion and expression analyses in uterine leiomyomata.
      ,
      • Mott J.D.
      • Thomas C.L.
      • Rosenbach M.T.
      • Takahara K.
      • Greenspan D.S.
      • Banda M.J.
      Post-translational proteolytic processing of procollagen c-terminal proteinase enhancer releases a metalloproteinase inhibitor.
      Hspg2NM_0083052.85.325.2FGF-2-induced tumor growth
      • Zhou Z.
      • Wang J.
      • Cao R.
      Impaired angiogenesis, delayed wound healing and retarded tumor growth in perlecan heparan sulfate-deficient mice.
      Igfbp2
      Genes noted inHoek et al (2004).
      AK0117844.072.041.1Human hepatoma, leukemia, colon carcinoma, prostate cancer
      • Akmal S.N.
      • Yun K.
      • MacLay J.
      • Higami Y.
      • Ikeda T.
      Insulin-like growth factor 2 and insulin-like growth factor binding protein 2 expression in hepatoblastoma.
      ,
      • Riikonen R.
      • Vettenranta K.
      • Riikonen P.
      • Turpeinen U.
      • Saarinen-Pihkala U.M.
      Cerebrospinal fluid insulin-like growth factor (igf-1) and insulin-like growth factor binding protein (igfbp-2) in children with acute lymphoblastic leukemia.
      • Fan C.W.
      • Chan C.C.
      • Chao C.C.
      • Fan H.A.
      • Sheu D.L.
      • Chan E.C.
      Expression patterns of cell cycle and apoptosis-related genes in a multidrug-resistant human colon carcinoma cell line.
      • Chi K.N.
      • Gleave M.E.
      Antisense approaches in prostate cancer.
      Col9a3
      Genes noted inHoek et al (2004).
      BG0744561.87.299.9(Includes all type IX collagens) pleomorphic adenoma, cartilaginous tumors, swarm rat chondrosarcoma
      • Luna M.G.
      Immunohistochemical localization of type i, ii and ix collagens in pleomorphic adenoma of human salivary glands.
      • Arai K.
      • Uehara K.
      • Nagai Y.
      Simultaneous expression of type ix collagen and an inhibin-related antigen in proliferative myoepithelial cells with pleomorphic adenoma of canine mammary glands.
      ,
      • Arai M.
      • Yada T.
      • Suzuki S.
      • Kimata K.
      Isolation and characterization of type ix collagen-proteoglycan from the swarm rat chondrosarcoma.
      UnknownBM1178273.06.732.5
      UnknownAV0246621.73.032.5
      Fn1
      Genes noted inHoek et al (2004).
      BM2343601.62.610.4Lymphoma, neurofibromas, papillary thyroid carcinoma, acute myeloid leukemia
      • Lossos I.S.
      • Czerwinski D.K.
      • Alizadeh A.A.
      • Wechser M.A.
      • Tibshirani R.
      • Botstein D.
      • Levy R.
      Prediction of survival in diffuse large-b-cell lymphoma based on the expression of six genes.
      ,
      • Holtkamp N.
      • Mautner V.F.
      • Friedrich R.E.
      Differentially expressed genes in neurofibromatosis 1-associated neurofibromas and malignant peripheral nerve sheath tumors.
      ,
      • Huang Y.
      • Prasad M.
      • Lemon W.J.
      Gene expression in papillary thyroid carcinoma reveals highly consistent profiles.
      ,
      • Arai Y.
      • Kyo T.
      • Miwa H.
      • Arai K.
      • Kamada N.
      • Kita K.
      • Ohki M.
      Heterogenous fusion transcripts involving the nup98 gene and hoxd13 gene activation in a case of acute myeloid leukemia with the t(2;11)(q31;p15) translocation.
      Tm4sf13
      Genes noted inHoek et al (2004).
      NM_0253591.86.027.3Thyroid cancer, lung cancer, colon cancer
      • Chen Z.
      • Mustafa T.
      • Trojanowicz B.
      Cd82, and cd63 in thyroid cancer.
      ,
      • Kao Y.R.
      • Shih J.Y.
      • Wen W.C.
      Tumor-associated antigen l6 and the invasion of human lung cancer cells.
      ,
      • Hashida H.
      • Takabayashi A.
      • Tokuhara T.
      Clinical significance of transmembrane 4 superfamily in colon cancer.
      UnknownBB3691915.214.78.6
      Itpr1NM_0105853.613.25.0Prostate cancer, gastric cancer
      • Vanden Abeele F.
      • Lemonnier L.
      • Thebault S.
      Two types of store-operated ca2+channels with different activation modes and molecular origin in lncap human prostate cancer epithelial cells.
      ,
      • Sakakura C.
      • Miyagawa K.
      • Fukuda K.
      Possible involvement of inositol 1, 4, 5-trisphosphate receptor type 3 (ip3r3) in the peritoneal dissemination of gastric cancers.
      VdrBC0067162.75.43.5Breast cancer, prostate cancer, colon cancer
      • Conde I.
      • Paniagua R.
      • Fraile B.
      • Ruiz A.
      • Arenas M.I.
      Expression of vitamin d3 receptor and retinoid receptors in human breast cancer: Identification of potential heterodimeric receptors.
      ,
      • Maistro S.
      • Snitcovsky I.
      • Sarkis A.S.
      • da Silva I.A.
      • Brentani M.M.
      Vitamin d receptor polymorphisms and prostate cancer risk in Brazilian men.
      ,
      • Palmer H.G.
      • Larriba M.J.
      • Garcia J.M.
      The transcription factor snail represses vitamin d receptor expression and responsiveness in human colon cancer.
      PscdbpBC00714480.2983.03.0None found; however, it is homologous to cancer-related pleckstrin
      Birc2NM_0074651.528.72.7Esophageal cancer
      • Fukuda K.
      • Sakakura C.
      • Miyagawa K.
      Differential gene expression profiles of radioresistant oesophageal cancer cell lines established by continuous fractionated irradiation.
      UnknownAV3225732.47.72.2
      PcafNM_0200052.03.33.0Prostate cancer, leukemia, human carcinoma
      • Wang X.
      • Yeh S.
      • Wu G.
      Identification and characterization of a novel androgen receptor coregulator ara267-alpha in prostate cancer cells.
      ,
      • Yang X.J.
      The diverse superfamily of lysine acetyltransferases and their roles in leukemia and other diseases.
      ,
      • Tamaki S.
      Effect of mutation in the p300 transcription coactivator on transcriptional response and cell proliferation of human carcinoma cell lines.
      UnknownNM_1331872.338.413.7
      PrkczNM_0088602.15.517.5(All protein kinase C's) lung cancer, colorectal cancer
      • Lahn M.
      • Su C.
      • Li S.
      Expression levels of protein kinase c-a in non-small-cell lung cancer.
      ,
      • Marshall J.L.
      • Eisenberg S.G.
      • Johnson M.D.
      A phase ii trial of isis 3521 in patients with metastatic colorectal cancer.
      Ssx2ipBB3963743.75.41.8Melanoma
      • dos Santos N.R.
      • Torensma R.
      • de Vries T.J.
      Heterogeneous expression of the ssx cancer/testis antigens in human melanoma lesions and cell lines.
      ClmnAB0596442.527.24.0None found
      Trim2NM_0307061.85.43.0None found—however, other trim proteins are cancer related
      • Horn E.J.
      • Albor A.
      • Liu Y.
      Ring protein trim32 associated with skin carcinogenesis has anti-apoptotic and e3-ubiquitin ligase properties.
      FcmdAB0773833.56.81.9Colon cancer, squamous carcinoma, pancreatic adenocarcinoma
      • Danes B.S.
      • Boyle P.D.
      • Traganos F.
      • Melamed M.R.
      Hyperdiploidy of cultured dermal fibroblasts from patients with aerodigestive squamous carcinoma.
      ,
      • Danes B.S.
      • Boyle P.D.
      • Traganos F.
      • Melamed M.R.
      In vitro evidence for genetic predisposition in some human pancreatic adenocarcinoma.
      ,
      • Danes B.S.
      • De Angelis P.
      • Traganos F.
      • Melamed M.R.
      Tetraploidy in cultured dermal fibroblasts from patients with heritable colon cancer.
      PhipBB5230302.211.61.8None found; homologous to cancer-related pleckstrin
      UnknownAY0338994.012.83.5
      Dbn1AB0643212.95.12.0None found
      Dip3bBC0022321.62.62.2None found
      Sf3b1NM_0311791.32.02.4None found
      Oxr1NM_1308852.411.31.7None found
      DownregulatedPou3f1NM_011141-1.8-6.6-14.7Other POU transcription factors
      • Theil T.
      • McLean-Hunter S.
      • Zornig M.
      • Moroy T.
      Mouse brn-3 family of pou transcription factors: A new aminoterminal domain is crucial for the oncogenic activity of brn-3a.
      Il24AF333251-2.6-34.2-6.2Lung cancer, induced by the RET/PTC3 oncoprotein
      • Ramesh R.
      • Ito I.
      • Gopalan B.
      • Saito Y.
      • Mhashilkar A.M.
      • Chada S.
      Ectopic production of mda-7/il-24 inhibits invasion and migration of human lung cancer cells.
      ,
      • Shinohara S.
      • Rothstein J.L.
      Interleukin 24 is induced by the ret/ptc3 oncoprotein and is an autocrine growth factor for epithelial cells.
      Npepl1BG076209-1.5-5.2-2.6None found
      LynM64608-1.5-3.3-3.0Sarcoma
      • Munshi N.
      • Ganju R.K.
      • Avraham S.
      • Mesri E.A.
      • Groopman J.E.
      Kaposi's sarcoma-associated herpesvirus-encoded g protein-coupled receptor activation of c-jun amino-terminal kinase/stress-activated protein kinase and lyn kinase is mediated by related adhesion focal tyrosine kinase/proline-rich tyrosine kinase 2.
      Cbr3AK003232-4.1-14.5-4.6(All carbonyl reductases) ovarian cancer, lung cancer
      • Umemoto M.
      • Yokoyama Y.
      • Sato S.
      • Tsuchida S.
      • Al-Mulla F.
      • Saito Y.
      Carbonyl reductase as a significant predictor of survival and lymph node metastasis in epithelial ovarian cancer.
      ,
      • Soldan M.
      • Nagel G.
      • Losekam M.
      • Ernst M.
      • Maser E.
      Interindividual variability in the expression and nnk carbonyl reductase activity of 11beta-hydroxysteroid dehydrogenase 1 in human lung.
      Adora2bNM_007413-1.3-2.3-2.5Astroglioma, pineal gland tumor
      • Fredholm B.B.
      • Altiok N.
      Adenosine a2b receptor signalling is altered by stimulation of bradykinin or interleukin receptors in astroglioma cells.
      ,
      • Suh B.C.
      • Kim T.D.
      • Lee J.U.
      • Seong J.K.
      • Kim K.T.
      Pharmacological characterization of adenosine receptors in pgt-beta mouse pineal gland tumour cells.
      LxnNM_016753-1.7-89.3-6.7None found
      Il1rnNM_031167-2.3-9.6-6.3Colorectal carcinoma, gastric cancer, prostate carcinoma
      • Allgayer H.
      • Nicolaus S.
      • Schreiber S.
      Decreased interleukin-1 receptor antagonist response following moderate exercise in patients with colorectal carcinoma after primary treatment.
      ,
      • Garza-Gonzalez E.
      • Hold G.
      • Perez-Perez G.I.
      • Bosques-Padilla F.J.
      • Tijerina-Menchaca R.
      • Maldonado-Garza H.J.
      • el-Omar E.
      Role of polymorphism of certain cytokines in gastric cancer in Mexico. Preliminary results.
      ,
      • Ricote M.
      • Garcia-Tunon I.
      • Bethencourt F.R.
      • Fraile B.
      • Paniagua R.
      • Royuela M.
      Interleukin-1 (il-1alpha and il-1beta) and its receptors (il-1ri, il-1rii, and il-1ra) in prostate carcinoma.
      Il1raM57525-1.5-3.8-6.3Colorectal carcinoma, gastric cancer, prostate carcinoma
      • Allgayer H.
      • Nicolaus S.
      • Schreiber S.
      Decreased interleukin-1 receptor antagonist response following moderate exercise in patients with colorectal carcinoma after primary treatment.
      ,
      • Garza-Gonzalez E.
      • Hold G.
      • Perez-Perez G.I.
      • Bosques-Padilla F.J.
      • Tijerina-Menchaca R.
      • Maldonado-Garza H.J.
      • el-Omar E.
      Role of polymorphism of certain cytokines in gastric cancer in Mexico. Preliminary results.
      ,
      • Ricote M.
      • Garcia-Tunon I.
      • Bethencourt F.R.
      • Fraile B.
      • Paniagua R.
      • Royuela M.
      Interleukin-1 (il-1alpha and il-1beta) and its receptors (il-1ri, il-1rii, and il-1ra) in prostate carcinoma.
      Ccne1NM_007633-1.3-1.9-2.1Ewing's tumors, rhabdomyosarcomas, breast cancer, bladder cancer
      • Zhang J.
      • Hu S.
      • Schofield D.E.
      • Sorensen P.H.
      • Triche T.J.
      Selective usage of d-type cyclins by Ewing's tumors and rhabdomyosarcomas.
      ,
      • Peters M.G.
      • Vidal Mdel C.
      • Gimenez L.
      Prognostic value of cell cycle regulator molecules in surgically resected stage i and ii breast cancer.
      ,
      • Kawamura K.
      • Izumi H.
      • Ma Z.
      Induction of centrosome amplification and chromosome instability in human bladder cancer cells by p53 mutation and cyclin e overexpression.
      TmlheAY033513-1.7-10.6-4.4None found
      CflarU97076-1.3-3.5-1.6Ewing's sarcoma, thyroid cancer, cervical cancer
      • Narayan G.
      • Pulido H.A.
      • Koul S.
      Genetic analysis identifies putative tumor suppressor sites at 2q35-q36.1 and 2q36.3-q37.1 involved in cervical cancer progression.
      ,
      • Kumar A.
      • Jasmin A.
      • Eby M.T.
      • Chaudhary P.M.
      Cytotoxicity of tumor necrosis factor related apoptosis-inducing ligand towards Ewing's sarcoma cell lines.
      ,
      • Ahmad M.
      • Shi Y.
      Trail-induced apoptosis of thyroid cancer cells: Potential for therapeutic intervention.
      Dusp9/MKP4
      Genes verified by RT-PCR. PRNP, Positional Ranking and Normalization Procedure; Igfbp2, insulin-like growth factor-binding protein 2; Col9a3, procollagen, type IX, α3; Fn1, fibronectin 1; Tm4sf13, transmembrane 4 superfamily member 13; Vdr, vitamin D receptor; Pcaf, p300/CBP-associated factor; Phip, pleckstrin homology domain interacting protein; Il24, interleukin 24; Il1ra, interleukin 1 receptor antagonist; Ccne1, cyclin E1; Cflar, CASP8- and FADD-like apoptosis regulator; SCC, squamous cell carcinoma.
      AV295798-3.5-18.6-1.9Human kidney carcinoma
      • Cheburkin Iu V.
      • Kniazeva T.G.
      • Peter S.
      Molecular portrait of human kidney carcinomas: The gene expression profiling of protein-tyrosine kinases and tyrosine phosphatases which controlled regulatory signals in the cells.
      Initiated keratinocytes were compared with non-transformed progenitor clone 291 and malignant keratinocytes were compared with initiated keratinocytes. Human melanomas were compared with normal human melanocytes. All data sets were analyzed using the PRNP method represented schematically in Figure 6. The upregulated genes that were found in all three comparisons are reported first followed by the downregulated genes that were found in all three comparisons. Genes with the largest rank offset (see Figure 6) are listed first. Fold changes were calculated from PRNP-normalized data where the invariant genes between samples (see Figure 6) were used to calibrate the normalization. Gene names and sequences were obtained from the Affymetrix online database. A PubMed search on the gene name revealed that many of the known genes have previously been associated with cancer. The six unknown, upregulated genes are listed.
      a Both known and unknown genes are listed. Genes are ordered based on PRNP ranking from greatest to least.
      b Fold changes are the average of three biological replicates.
      c Genes noted in
      • Hoek K.
      • Rimm D.L.
      • Williams K.R.
      Expression profiling reveals novel pathways in the transformation of melanocytes to melanomas.
      .
      d Genes verified by RT-PCR. PRNP, Positional Ranking and Normalization Procedure; Igfbp2, insulin-like growth factor-binding protein 2; Col9a3, procollagen, type IX, α3; Fn1, fibronectin 1; Tm4sf13, transmembrane 4 superfamily member 13; Vdr, vitamin D receptor; Pcaf, p300/CBP-associated factor; Phip, pleckstrin homology domain interacting protein; Il24, interleukin 24; Il1ra, interleukin 1 receptor antagonist; Ccne1, cyclin E1; Cflar, CASP8- and FADD-like apoptosis regulator; SCC, squamous cell carcinoma.
      Several of the higher-ranking upregulated genes in the current PRNP analysis were also extracted by the standard fold change method reported previously in
      • Hoek K.
      • Rimm D.L.
      • Williams K.R.
      Expression profiling reveals novel pathways in the transformation of melanocytes to melanomas.
      . These include insulin-like growth factor-binding protein 2 (Igfbp2), procollagen, type IX, α3 (Col9a3), fibronectin 1 (Fn1), and transmembrane 4 superfamily member 13 (Tm4sf13). Igfbp2 is involved with the regulation of cell growth and maintenance. Col9a3 plays an integral role in cell adhesion and is a structural constituent of the extracellular matrix. Fn1 is involved with the transforming growth factor-β signaling pathway and affects a variety of cellular activities including cell adhesion, metabolism, and wound healing. Tm4sf13 is an integral plasma membrane protein involved in signal transduction regulating cell development, activation, growth, and motility.
      Several other genes not previously identified from these models were highlighted using PRNP. Upregulated ones include: vitamin D receptor (Vdr), a DNA-binding transcription factor involved in calcium ion transport and homeostasis, p300/CBP-associated factor (Pcaf), a histone acetyltransferase involved in transcription regulation, and pleckstrin homology domain interacting protein (Phip), a protein-binding protein in the insulin receptor signaling pathway. The majority of the downregulated genes identified have already been discussed previously in multiple cancer publications. They include: interleukin 24 (Il24) and interleukin 1 receptor antagonist (Il1ra), cell surface signal transduction proteins involved in the immune response, CASP8- and FADD-like apoptosis regulator (Cflar), a caspase involved in the induction and regulation of apoptosis, and cyclin E1 (Ccne1), a kinase involved in regulating DNA replication initiation during the cell cycle. The novel perspective gained from this study is that these genes are altered early in the development of cancer, and their altered gene expression is either retained or exacerbated in metastatic cancer (see fold-change reports in Table I).

      Summary, Conclusions, and Future Microarray Studies

      The current approach combined a biologically well-characterized model and novel computational methods of microarray data analysis to greatly reduce the number of candidate genes for further analysis as cancer diagnostic or therapeutic targets. The genetically related cells of the mouse model, comprising non-transformed and early and late stages of cancer progression, helped narrow the field of human melanoma-associated genes from thousands of probe sets to 40. Of the 27 upregulated and 13 downregulated genes known previously, 16 and nine, respectively, have been previously related to cancer either directly (all but three) or indirectly through their family member proteins. Retrieving a high percentage of cancer-related genes by the comparison of mouse and human cancer samples suggests that PRNP is an informative method for analyzing microarray data and selecting cancer genes.
      The ranking method has the advantage that it is not affected by hybridization discrepancies. Combining this method with other statistical applications such as Principal Component Analysis will allow multi-dimensional statistical separation of “normal,” benign, and malignant cells from the mouse model. A complete statistical analysis of the mouse model of SCC will provide a basis for comparison that can be applied to microarray data from individual human tumors in order to inform their indolent versus acute course and consequently, to help plan the strategy of therapy.
      Fifteen percent of the genes extracted in this experiment are unknown or have never been previously related to cancer. These are listed by accession number in Table I. An investigation of these genes may reveal new players in the pathways of cancer. Knowledge of genes in common among different types of cancers may elucidate common cancer targets for future therapy.
      At least with respect to occurring early and being maintained or exacerbated in malignancy, the genes highlighted by PRNP fit the paradigm for promising cancer molecular therapeutic candidates, as discussed at the 2004 Montagna Symposium by the keynote speaker Brian Druker. The genes represent changes that occur early (like Bcr/Abl activation causal in CML) and are required to sustain the cancer. In the future, such targets alone or in combination promise to permit the same types of remissions in human solid tumors or epithelia as achieved to date in CML.

      ACKNOWLEDGMENTS

      This work was supported by NIH PHS grants to M. K. M. CA98893 and CA106195 Training Grant in Molecular Skin Pathobiology, to the OHSU Cancer Institute CA69533, to M. B. P. CA090897 and CA27502, and to R. H., CA44542, and AR41942 Yale Skin Diseases Research Center. Microarray data analysis and C. P. were supported by an NIH/NCI grant to R. S., K01 CA086957.

      References

        • Ahmad M.
        • Shi Y.
        Trail-induced apoptosis of thyroid cancer cells: Potential for therapeutic intervention.
        Oncogene. 2000; 19: 3363-3371
        • Akmal S.N.
        • Yun K.
        • MacLay J.
        • Higami Y.
        • Ikeda T.
        Insulin-like growth factor 2 and insulin-like growth factor binding protein 2 expression in hepatoblastoma.
        Hum Pathol. 1995; 26: 846-851
        • Allgayer H.
        • Nicolaus S.
        • Schreiber S.
        Decreased interleukin-1 receptor antagonist response following moderate exercise in patients with colorectal carcinoma after primary treatment.
        Cancer Detect Prev. 2004; 28: 208-213
        • Arai K.
        • Uehara K.
        • Nagai Y.
        Simultaneous expression of type ix collagen and an inhibin-related antigen in proliferative myoepithelial cells with pleomorphic adenoma of canine mammary glands.
        Jpn J Cancer Res. 1995; 86: 577-584
        • Arai M.
        • Yada T.
        • Suzuki S.
        • Kimata K.
        Isolation and characterization of type ix collagen-proteoglycan from the swarm rat chondrosarcoma.
        Biochim Biophys Acta. 1992; 1117: 60-70
        • Arai Y.
        • Kyo T.
        • Miwa H.
        • Arai K.
        • Kamada N.
        • Kita K.
        • Ohki M.
        Heterogenous fusion transcripts involving the nup98 gene and hoxd13 gene activation in a case of acute myeloid leukemia with the t(2;11)(q31;p15) translocation.
        Leukemia. 2000; 14: 1621-1629
        • Bardeesy N.
        • Bastian B.C.
        • Hezel A.
        • Pinkel D.
        • DePinho R.A.
        • Chin L.
        Dual inactivation of rb and p53 pathways in ras-induced melanomas.
        Mol Cell Biol. 2001; 21: 2144-2153
        • Brinck U.
        • Ruschenburg I.
        • Di Como C.J.
        Comparative study of p63 and p53 expression in tissue microarrays of malignant melanomas.
        Int J Mol Med. 2002; 10: 707-711
        • Broome P.M.
        • Gause P.R.
        • Hyman P.
        • Gregus J.
        • Lluria-Prevatt M.
        • Nagle R.
        • Bowden G.T.
        Induction of melanoma in tpras transgenic mice.
        Carcinogenesis. 1999; 20: 1747-1753
        • Chan W.M.
        • Siu W.Y.
        • Lau A.
        • Poon R.Y.
        How many mutant p53 molecules are needed to inactivate a tetramer?.
        Mol Cell Biol. 2004; 24: 3536-3551
        • Cheburkin Iu V.
        • Kniazeva T.G.
        • Peter S.
        Molecular portrait of human kidney carcinomas: The gene expression profiling of protein-tyrosine kinases and tyrosine phosphatases which controlled regulatory signals in the cells.
        Mol Biol (Mosk). 2002; 36: 480-490
        • Chen Z.
        • Mustafa T.
        • Trojanowicz B.
        Cd82, and cd63 in thyroid cancer.
        Int J Mol Med. 2004; 14: 517-527
        • Chi K.N.
        • Gleave M.E.
        Antisense approaches in prostate cancer.
        Expert Opin Biol Ther. 2004; 4: 927-936
        • Conde I.
        • Paniagua R.
        • Fraile B.
        • Ruiz A.
        • Arenas M.I.
        Expression of vitamin d3 receptor and retinoid receptors in human breast cancer: Identification of potential heterodimeric receptors.
        Int J Oncol. 2004; 25: 1183-1191
        • Danes B.S.
        • Boyle P.D.
        • Traganos F.
        • Melamed M.R.
        Hyperdiploidy of cultured dermal fibroblasts from patients with aerodigestive squamous carcinoma.
        Cytometry. 1987; 8: 584-589
        • Danes B.S.
        • Boyle P.D.
        • Traganos F.
        • Melamed M.R.
        In vitro evidence for genetic predisposition in some human pancreatic adenocarcinoma.
        Pancreas. 1987; 2: 584-588
        • Danes B.S.
        • De Angelis P.
        • Traganos F.
        • Melamed M.R.
        Tetraploidy in cultured dermal fibroblasts from patients with heritable colon cancer.
        Dis Markers. 1988; 6: 151-161
        • De Laurenzi V.
        • Costanzo A.
        • Barcaroli D.
        Two new p73 splice variants, gamma and delta, with different transcriptional activity.
        J Exp Med. 1998; 188: 1763-1768
        • Deininger M.
        • Buchdunger E.
        • Druker B.J.
        The development of imatinib as a therapeutic agent for chronic myeloid leukemia.
        Blood. 2005; 105: 2640-2653
        • dos Santos N.R.
        • Torensma R.
        • de Vries T.J.
        Heterogeneous expression of the ssx cancer/testis antigens in human melanoma lesions and cell lines.
        Cancer Res. 2000; 60: 1654-1662
        • Fan C.W.
        • Chan C.C.
        • Chao C.C.
        • Fan H.A.
        • Sheu D.L.
        • Chan E.C.
        Expression patterns of cell cycle and apoptosis-related genes in a multidrug-resistant human colon carcinoma cell line.
        Scand J Gastroenterol. 2004; 39: 464-469
        • Fredholm B.B.
        • Altiok N.
        Adenosine a2b receptor signalling is altered by stimulation of bradykinin or interleukin receptors in astroglioma cells.
        Neurochem Int. 1994; 25: 99-102
        • Fukuda K.
        • Sakakura C.
        • Miyagawa K.
        Differential gene expression profiles of radioresistant oesophageal cancer cell lines established by continuous fractionated irradiation.
        Br J Cancer. 2004; 91: 1543-1550
        • Garza-Gonzalez E.
        • Hold G.
        • Perez-Perez G.I.
        • Bosques-Padilla F.J.
        • Tijerina-Menchaca R.
        • Maldonado-Garza H.J.
        • el-Omar E.
        Role of polymorphism of certain cytokines in gastric cancer in Mexico. Preliminary results.
        Rev Gastroenterol Mexico. 2003; 68: 107-112
        • Gause P.R.
        • Lluria-Prevatt M.
        • Keith W.N.
        • Balmain A.
        • Linardopolous S.
        • Warneke J.
        • Powell M.B.
        Chromosomal and genetic alterations of 7,12-dimethylbenz[a]anthracene-induced melanoma from tp-ras transgenic mice.
        Mol Carcinogen. 1997; 20: 78-87
        • Ghiorzo P.
        • Bianchi Scarr G.B.
        Genetics of melanoma susceptibility.
        Forum (Genova). 2003; 13: 114-126
        • Hashida H.
        • Takabayashi A.
        • Tokuhara T.
        Clinical significance of transmembrane 4 superfamily in colon cancer.
        Br J Cancer. 2003; 89: 158-167
        • Hoek K.
        • Rimm D.L.
        • Williams K.R.
        Expression profiling reveals novel pathways in the transformation of melanocytes to melanomas.
        Cancer Res. 2004; 64: 5270-5282
        • Holtkamp N.
        • Mautner V.F.
        • Friedrich R.E.
        Differentially expressed genes in neurofibromatosis 1-associated neurofibromas and malignant peripheral nerve sheath tumors.
        Acta Neuropathol (Berlin). 2004; 107: 159-168
        • Horn E.J.
        • Albor A.
        • Liu Y.
        Ring protein trim32 associated with skin carcinogenesis has anti-apoptotic and e3-ubiquitin ligase properties.
        Carcinogenesis. 2004; 25: 157-167
        • Huang Y.
        • Prasad M.
        • Lemon W.J.
        Gene expression in papillary thyroid carcinoma reveals highly consistent profiles.
        Proc Natl Acad Sci USA. 2001; 98: 15044-15049
        • Kao Y.R.
        • Shih J.Y.
        • Wen W.C.
        Tumor-associated antigen l6 and the invasion of human lung cancer cells.
        Clin Cancer Res. 2003; 9: 2807-2816
        • Kawamura K.
        • Izumi H.
        • Ma Z.
        Induction of centrosome amplification and chromosome instability in human bladder cancer cells by p53 mutation and cyclin e overexpression.
        Cancer Res. 2004; 64: 4800-4809
        • Knights C.D.
        • Liu Y.
        • Appella E.
        • Kulesz-Martin M.
        Defective p53 post-translational modification required for wild type p53 inactivation in malignant epithelial cells with mdm2 gene amplification.
        J Biol Chem. 2003; 278: 52890-52900
        • Koster M.I.
        • Kim S.
        • Mills A.A.
        • DeMayo F.J.
        • Roop D.R.
        P63 is the molecular switch for initiation of an epithelial stratification program.
        Genes Dev. 2004; 18: 126-131
        • Kulesz-Martin M.
        • Blumenson L.
        • Lisafeld B.
        Retinoic acid enhancement of an early step in the transformation of mouse epidermal cells in vitro.
        Carcinogenesis. 1986; 7: 1425-1429
        • Kulesz-Martin M.
        • Yoshida M.A.
        • Prestine L.
        • Yuspa S.H.
        • Bertram J.S.
        Mouse cell clones for improved quantitation of carcinogen-induced altered differentiation.
        Carcinogenesis. 1985; 6: 1245-1254
        • Kulesz-Martin M.F.
        • Penetrante R.
        • East C.J.
        Benign and malignant tumor stages in a mouse keratinocyte line treated with 7,12-dimethylbenz[à]athracene in vitro.
        Carcinogenesis. 1988; 9: 171-174
        • Kumar A.
        • Jasmin A.
        • Eby M.T.
        • Chaudhary P.M.
        Cytotoxicity of tumor necrosis factor related apoptosis-inducing ligand towards Ewing's sarcoma cell lines.
        Oncogene. 2001; 20: 1010-1014
        • Lahn M.
        • Su C.
        • Li S.
        Expression levels of protein kinase c-a in non-small-cell lung cancer.
        Clin Lung Cancer. 2004; 6: 184-189
        • Ligon A.H.
        • Scott I.C.
        • Takahara K.
        • Greenspan D.S.
        • Morton C.C.
        Pcolce deletion and expression analyses in uterine leiomyomata.
        Cancer Genet Cytogenet. 2002; 137: 133-137
        • Lui Y.
        • Asch H.
        • Kulesz-Martin M.F.
        Functional quantification of DNA-binding proteins p53 and estrogen receptor in cells and tumor tissues by DNA affinity immunobloting.
        Cancer Res. 2001; 61: 5902-5906
        • Lossos I.S.
        • Czerwinski D.K.
        • Alizadeh A.A.
        • Wechser M.A.
        • Tibshirani R.
        • Botstein D.
        • Levy R.
        Prediction of survival in diffuse large-b-cell lymphoma based on the expression of six genes.
        N Engl J Med. 2004; 350: 1828-1837
        • Luna M.G.
        Immunohistochemical localization of type i, ii and ix collagens in pleomorphic adenoma of human salivary glands.
        Kokubyo Gakkai Zasshi. 1996; 63: 122-130
        • Maistro S.
        • Snitcovsky I.
        • Sarkis A.S.
        • da Silva I.A.
        • Brentani M.M.
        Vitamin d receptor polymorphisms and prostate cancer risk in Brazilian men.
        Int J Biol Markers. 2004; 19: 245-249
        • Marshall J.L.
        • Eisenberg S.G.
        • Johnson M.D.
        A phase ii trial of isis 3521 in patients with metastatic colorectal cancer.
        Clin Colorectal Cancer. 2004; 4: 268-274
        • Moll U.M.
        • Slade N.
        P63 and p73: Roles in development and tumor formation.
        Mol Cancer Res. 2004; 2: 371-386
        • Mott J.D.
        • Thomas C.L.
        • Rosenbach M.T.
        • Takahara K.
        • Greenspan D.S.
        • Banda M.J.
        Post-translational proteolytic processing of procollagen c-terminal proteinase enhancer releases a metalloproteinase inhibitor.
        J Biol Chem. 2000; 275: 1384-1390
        • Munshi N.
        • Ganju R.K.
        • Avraham S.
        • Mesri E.A.
        • Groopman J.E.
        Kaposi's sarcoma-associated herpesvirus-encoded g protein-coupled receptor activation of c-jun amino-terminal kinase/stress-activated protein kinase and lyn kinase is mediated by related adhesion focal tyrosine kinase/proline-rich tyrosine kinase 2.
        J Biol Chem. 1999; 274: 31863-31867
        • Narayan G.
        • Pulido H.A.
        • Koul S.
        Genetic analysis identifies putative tumor suppressor sites at 2q35-q36.1 and 2q36.3-q37.1 involved in cervical cancer progression.
        Oncogene. 2003; 22: 3489-3499
        • Ongkeko W.M.
        • Wang X.Q.
        • Siu W.Y.
        Mdm2 and mdmx bind and stabilize the p53-related protein p73.
        Curr Biol. 1999; 9: 829-832
        • Palmer H.G.
        • Larriba M.J.
        • Garcia J.M.
        The transcription factor snail represses vitamin d receptor expression and responsiveness in human colon cancer.
        Nat Med. 2004; 10: 917-919
        • Peters M.G.
        • Vidal Mdel C.
        • Gimenez L.
        Prognostic value of cell cycle regulator molecules in surgically resected stage i and ii breast cancer.
        Oncol Rep. 2004; 12: 1143-1150
        • Puri N.
        • Eller M.S.
        • Byers H.R.
        • Dykstra S.
        • Kubera J.
        • Gilchrest B.A.
        Telomere-based DNA damage responses: A new approach to melanoma.
        Faseb J. 2004; 18: 1373-1381
        • Qin J.Z.
        • Stennett L.
        • Bacon P.
        P53-independent noxa induction overcomes apoptotic resistance of malignant melanomas.
        Mol Cancer Ther. 2004; 3: 895-902
        • Ramesh R.
        • Ito I.
        • Gopalan B.
        • Saito Y.
        • Mhashilkar A.M.
        • Chada S.
        Ectopic production of mda-7/il-24 inhibits invasion and migration of human lung cancer cells.
        Mol Ther. 2004; 9: 510-518
        • Ricote M.
        • Garcia-Tunon I.
        • Bethencourt F.R.
        • Fraile B.
        • Paniagua R.
        • Royuela M.
        Interleukin-1 (il-1alpha and il-1beta) and its receptors (il-1ri, il-1rii, and il-1ra) in prostate carcinoma.
        Cancer. 2004; 100: 1388-1396
        • Riikonen R.
        • Vettenranta K.
        • Riikonen P.
        • Turpeinen U.
        • Saarinen-Pihkala U.M.
        Cerebrospinal fluid insulin-like growth factor (igf-1) and insulin-like growth factor binding protein (igfbp-2) in children with acute lymphoblastic leukemia.
        Pediatr Blood Cancer. 2004; 43: 110-114
        • Sakakura C.
        • Miyagawa K.
        • Fukuda K.
        Possible involvement of inositol 1, 4, 5-trisphosphate receptor type 3 (ip3r3) in the peritoneal dissemination of gastric cancers.
        Gan To Kagaku Ryoho. 2003; 30: 1784-1787
        • Schneider B.L.
        • Bowden G.T.
        • Sutter C.
        • Schweizer J.
        • Han K.A.
        • Kulesz-Martin M.F.
        7,12–dimethylbenz[a]anthracene-induced mouse keratinocyte malignant transformation independent of Harvey ras activation.
        J Invest Dermatol. 1993; 101: 595-599
        • Shinohara S.
        • Rothstein J.L.
        Interleukin 24 is induced by the ret/ptc3 oncoprotein and is an autocrine growth factor for epithelial cells.
        Oncogene. 2004; 23: 7571-7579
        • Soengas M.S.
        • Capodieci P.
        • Polsky D.
        Inactivation of the apoptosis effector APAF-1 in malignant melanoma.
        Nature. 2001; 409: 207-211
        • Soldan M.
        • Nagel G.
        • Losekam M.
        • Ernst M.
        • Maser E.
        Interindividual variability in the expression and nnk carbonyl reductase activity of 11beta-hydroxysteroid dehydrogenase 1 in human lung.
        Cancer Lett. 1999; 145: 49-56
        • Soussi T.
        • Kato S.
        • Levy P.P.
        • Ishioka C.
        Reassessment of the tp53 mutation database in human disease by data mining with a library of tp53 missense mutations.
        Hum Mutat. 2005; 25: 6-17
        • Suh B.C.
        • Kim T.D.
        • Lee J.U.
        • Seong J.K.
        • Kim K.T.
        Pharmacological characterization of adenosine receptors in pgt-beta mouse pineal gland tumour cells.
        Br J Pharmacol. 2001; 134: 132-142
        • Tamaki S.
        Effect of mutation in the p300 transcription coactivator on transcriptional response and cell proliferation of human carcinoma cell lines.
        Kokubyo Gakkai Zasshi. 2002; 69: 139-151
        • Team RDC
        R: A language and environment for statistical computing. R Foundation for Statistical Computing.
        Vienna, Austria2004
        • Theil T.
        • McLean-Hunter S.
        • Zornig M.
        • Moroy T.
        Mouse brn-3 family of pou transcription factors: A new aminoterminal domain is crucial for the oncogenic activity of brn-3a.
        Nucleic Acids Res. 1993; 21: 5921-5929
        • Tuve S.
        • Wagner S.N.
        • Schittek B.
        • Putzer B.M.
        Alterations of delta-p 73 splice transcripts during melanoma development and progression.
        Int J Cancer. 2004; 108: 162-166
        • Umemoto M.
        • Yokoyama Y.
        • Sato S.
        • Tsuchida S.
        • Al-Mulla F.
        • Saito Y.
        Carbonyl reductase as a significant predictor of survival and lymph node metastasis in epithelial ovarian cancer.
        Br J Cancer. 2001; 85: 1032-1036
        • Vanden Abeele F.
        • Lemonnier L.
        • Thebault S.
        Two types of store-operated ca2+channels with different activation modes and molecular origin in lncap human prostate cancer epithelial cells.
        J Biol Chem. 2004; 279: 30326-30337
        • Wang X.
        • Yeh S.
        • Wu G.
        Identification and characterization of a novel androgen receptor coregulator ara267-alpha in prostate cancer cells.
        J Biol Chem. 2001; 276: 40417-40423
        • Wang Z.
        • Liu Y.
        • Mori M.
        • Kulesz-Martin M.
        Gene expression profiling of initiated epidermal cells with benign or malignant tumor fates.
        Carcinogenesis. 2002; 23: 635-643
        • Wu Y.
        • Huang H.
        • Miner Z.
        • Kulesz-Martin M.
        Activities and response to DNA damage of latent and active sequence-specific DNA binding forms of mouse p53.
        Proc Natl Acad Sci USA. 1997; 94: 8982-8987
        • Yang A.
        • Kaghad M.
        • Wang Y.
        P63, a p53 homolog at 3q27–29, encodes multiple products with transactivating, death-inducing, and dominant-negative activities.
        Mol Cell. 1998; 2: 305-316
        • Yang X.J.
        The diverse superfamily of lysine acetyltransferases and their roles in leukemia and other diseases.
        Nucleic Acids Res. 2004; 32: 959-976
        • Zhang H.
        • Rosdahl I.
        Ultraviolet a and b differently induce intracellular protein expression in human skin melanocytes—a speculation of separate pathways in initiation of melanoma.
        Carcinogenesis. 2003; 24: 1929-1934
        • Zhang J.
        • Hu S.
        • Schofield D.E.
        • Sorensen P.H.
        • Triche T.J.
        Selective usage of d-type cyclins by Ewing's tumors and rhabdomyosarcomas.
        Cancer Res. 2004; 64: 6026-6034
        • Zhou Z.
        • Wang J.
        • Cao R.
        Impaired angiogenesis, delayed wound healing and retarded tumor growth in perlecan heparan sulfate-deficient mice.
        Cancer Res. 2004; 64: 4699-4702

      Linked Article