The RET Receptor Is Linked to Stress Response Pathways

Home , Clusterin, DNAJC3, DOK5, HSPA1B, HSPA1L, Integrin

[CANCER RESEARCH 64, 4453–4463, July 1, 2004] The RET Receptor Is Linked to Stress Response Pathways

Shirley M. Myers and Lois M. Mulligan Division of Cancer Biology and Genetics, Queen’s Cancer Research Institute, Queen’s University, Kingston, Ontario, Canada

ABSTRACT viewed in Ref. 13). Furthermore, rearrangements of RET, resulting in juxtaposition of the RET kinase domain with a dimerization domain RET is a transmembrane receptor required for the development of from any of several other proteins, occur somatically in papillary neuroendocrine and urogenital cell types. Activation of RET has roles in cell growth, migration, or differentiation, yet little is known about the gene thyroid carcinoma (14). In each case, these mutations result in acti- expression patterns through which these processes are mediated. We have vation of the RET receptor, leading to inappropriate and/or increased generated cell lines stably expressing either the RET9 or RET51 protein RET-mediated signal transduction and resultant cell proliferation and isoforms and have used these to investigate RET-mediated gene expres- tumorigenesis. sion patterns by cDNA microarray analyses. As seen for many oncogenes, Activation of RET, either by ligand or through specific mutations, we identified altered expression of genes associated generally with cell– results in phosphorylation of multiple tyrosine residues that, in turn, cell or cell-substrate interactions and up-regulation of tumor-specific interact with specific adaptor molecules to trigger downstream sig- transcripts. We also saw increased expression of transcripts normally naling. Four of the tyrosines phosphorylated on RET activation, associated with neural crest or other RET-expressing cell types, suggesting these genes may lie downstream of RET activation in development. The tyrosines 905, 1015, 1062, and 1096, have been well characterized, most striking pattern of expression was up-regulation of stress response and are known to interact with a number of adaptor molecules to genes. We showed that RET expression significantly up-regulated the stimulate signaling through phosphatidylinositol kinase (PI3K)/AKT, genes for heat shock protein (HSP) 70 family members, HSPA1A, PLC-␥, RAS/extracellular signal-regulated kinase (ERK), p38MAPK, HSPA1B, and HSPA1L. Other members of several HSP families and c-Jun NH2-terminal kinase (JNK), and ERK5 pathways (15–19). In HSP70-interacting molecules that were associated with stress response response to these signals, cell type-specific responses are initiated that protein complexes involved in protein maturation were also specifically implement the varied functional roles of RET. An additional level of up-regulated by RET, whereas those associated with the roles of HSP70 in protein degradation were down-regulated or unaffected. The major mech- complexity of RET signaling is added by alternative splicing of the anism of stress response induction is activation of the heat shock tran- RET gene (20), which leads to functionally distinct RET isoforms, scription factor HSF1. We showed that RET expression leads to increased termed RET9 and RET51. These isoforms differ in their COOH- HSF1 activation, which correlates with increased expression of stress terminal amino acids, having either 9 or 51 unique residues, and have response genes. Together, our data suggest that RET may be directly distinct transforming and differentiative potentials in vitro (21, 22). In responsible for expression of stress response proteins and the initiation of transgenic mice, animals expressing only the RET9 isoform are viable stress response. and appear normal, whereas monoisoformic RET51 animals have kidney dysplasia and lack enteric ganglia (23). These differences INTRODUCTION likely reflect differences in signaling potential of these RET isoforms. An additional phosphotyrosine, Y1096, present only in RET51, has The RET proto-oncogene encodes a receptor tyrosine kinase re- been shown to bind GRB2, activating PI3K and RAS/MAPK path- quired for normal development of the kidney, peripheral, and central ways (24–27). In addition, RET9 and RET51 differ in protein inter- nervous systems, and of spermatogonia (1–3). RET has been impli- actions with phosphotyrosine 1062, which is the last amino acid cated in cell type-specific processes including cell proliferation, mi- common to both isoforms and lies in different amino acid contexts in gration, and differentiation, and likely plays each of these roles in each protein (20). Tyrosine 1062 has been shown to act as a binding specific cells and at specific developmental time points. In neural cell site for multiple adaptor proteins including SHC, docking protein lineages, RET also plays an important role in cell survival, particu- (DOK)1, DOK4, DOK5, fibroblast growth factor receptor substrate 2 larly in response to environmental stresses (4–11). (FRS2), and insulin receptor substrate 1 (IRS1) (15, 26, 28–30); In normal cells, the RET receptor is activated by the binding of both however, the relative binding of these molecules to RET9 and to a circulating ligand and a cell surface-bound coreceptor. RET ligands RET51 varies both quantitatively and qualitatively. For example, SHC are members of the glial cell line-derived neurotrophic factor (GDNF) binds both RET9 and RET51 but, whereas RET51 binds only to the family (reviewed in Ref. 12). These molecules interact directly with SHC-PTB domain, RET9 can also bind through the SHC-SH2 domain coreceptors of the GDNF family receptors ␣ (GFR␣) proteins, which potentiating differences in the specific downstream interactions and/or are linked to the cell surface by glycosylphosphatidyl-inositol linkage, the magnitude of the signals transduced (26). and the resultant complexes bind to the RET receptor to activate Although many of the pathways through which RET transduces downstream signaling events. In addition, activating point mutations extracellular signals have been identified or predicted, little is known of RET have been identified in patients with multiple endocrine neoplasia type 2, an inherited cancer syndrome characterized by of the gene targets that are specifically modulated in response to medullary thyroid carcinoma and pheochromocytoma and are found receptor activation. In this study, we have used gene expression in a large proportion of sporadic medullary thyroid carcinoma (re- microarray analyses to evaluate targets of RET activation in an embryonic kidney-derived cell line. In addition to predictable targets involved in cell–cell interaction, cell proliferation, and neuroendo- Received 11/17/03; revised 3/17/04; accepted 4/30/04. Grant support: This work was supported by grants from the National Cancer Institute crine differentiation, our data suggest that activation of the RET of Canada and the Canadian Institutes of Health Research. receptor may specifically target genes for proteins integral to induc- The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with ible cellular stress response. This study may provide a direct link 18 U.S.C. Section 1734 solely to indicate this fact. between the up-regulation of heat shock proteins (HSPs), seen in Requests for reprints: Lois M. Mulligan, Queen’s Cancer Research Institute, Bot- terell Hall, Room 329, Queen’s University, Kingston, Ontario, Canada, K7L 3N6. Phone: many primary tumor types and as a neuroprotective event, and the (613) 533-6310; Fax: (613) 548-1348; E-mail: [email protected]. stimulation of receptor tyrosine kinase activity. 4453

MATERIALS AND METHODS quence tags (ESTs) were obtained from the University Health Network Mi- croarray Centre. The EST content of the 1.7K array partially overlapped that Expression Constructs. Full-length human cDNAs encoding either the of the 19K array.1 Arrays were probed with differentially labeled cDNAs using 1072 amino acid (RET9) or the 1114 amino acid (RET51) isoforms of RET optimized protocols.2 Briefly, in a 40-␮l reaction, 10 ␮g of total RNA was (31) were cloned into CH269, an episomal expression vector derived from reverse transcribed using Superscript II according to the manufacturer’s in- vector pCEP4 (Invitrogen, Burlington, ON, Canada) under the control of the ␮ Ј Ј structions (Invitrogen) with 3.75 M Anchored-T primer (5 -T20VN-3 ); cytomegalovirus promoter. The sequence of each construct was verified by dATP, dGTP, and dTTP (500 ␮M each); and 50 ␮M dCTP, 10 mM DTT, 1 ng restriction digestion, and direct sequencing (Mobix, Hamilton, ON, Canada). of control RNA (artificial Arabadopsis transcripts), and either 25 ␮M Cy3- The GFR␣1 expression construct has been described previously (32, 33). dCTP or Cy5-dCTP (Mandel-NEN, Guelph, ON, Canada) at 42°C for 2–3h. Cell Culture and Transfection Experiments. E293, a transformed em- RNA was hydrolized and Cy5 and Cy3-labeled cDNA were combined, iso- bryonic kidney cell line, was maintained in DMEM, supplemented with 10% propanol precipitated, and resuspended in water. Labeled cDNAs were hybrid- fetal bovine serum, penicillin, streptomycin, and G418. For transient transfec- ized to cDNA microarrays in a medium consisting of DIG Easy Hyb solution tions, E293 cells were seeded into six-well plates and grown to approximately (Roche Applied Science, Laval, QC, Canada) containing 50 ␮g of yeast tRNA 70–80% confluence. Cells were transfected with 1 ␮g of expression construct (Invitrogen) and 50 ␮g of sheared calf thymus DNA (Sigma, Oakville, ON, or empty vector, with the FUGENE6 reagent (Roche Applied Science, Laval, Canada) per 100 ␮lina37°C dark humid chamber over night. Slides were QC, Canada) according to the manufacturer’s instructions, and were incubated ϫ ϫ for 48 h. In RET activation experiments, cells stably or transiently transfected washed with 1 SSC, 0.1% SDS at 50°C, rinsed in 1 SSC, and spun dry. with RET constructs were cotransfected with 1 ␮goftheGFR␣1 expression Arrays were scanned with a ScanArray 4000 scanner using ScanArray soft- construct, serum-starved for 1 h, and treated with 100 ng/ml of GDNF ware, and intensities were quantified using QuantArray software (Perkin- (Promega, Madison, WI) for 10 min, before harvesting. Elmer Life Sciences, Boston, MA). The expression ratios of cDNAs, for each Cell lines stably expressing RET9 and RET51 were generated by transfec- comparison, were calculated using background-corrected hybridization inten- tion of E293 cells with the RET constructs, as described above, and by sities normalized to corresponding intensity averages for the whole array. selection of RET-expressing clones with 400 ␮g/ml hygromycin. Stable cell Artifacts were removed from the data sets after visually inspecting spots on the lines were maintained in 100 ␮g/ml hygromycin and were regularly screened array images. Low-intensity hybridization signals (Յ500) were excluded from for RET expression. our analyses. After exclusion of artifacts and low-intensity hybridization, we Immunoprecipitations and Western Blotting. RET expression was con- were able to analyze more than 80% of array spots. MA plots of these firmed by Western analysis. Cells were washed three times with ice-cold PBS normalized data suggested that differences in gene expression detected were and lyzed in 20 mM Tris-HCl (pH 7.8), 150 mM NaCl, 1 mM sodium orthovana- not due to intensity-dependent dye-label effects between the compared samples date, 1% Igepal, 1 mM phenylmethylsulfonyl fluoride, 10 ␮g/ml aprotinin, 10 (Fig. 2). Paired t tests were performed on log-transformed data (35, 36) and the ␮g/ml leupeptin, and 2 mM EDTA, as described previously (21). Lysates were significance of gene expression differences (P Ͻ 0.1) is indicated in Tables cleared at 12,000 ϫ g,4°C for 15 min. Protein concentration was determined 1–3. Relative expression values are averages of ratios of normalized values. by BCA Protein Assay (Pierce, Rockford, IL.). For direct analysis, lysates were Validation by Northern and Quantitative Real-Time PCR Analyses. combined at 1:1 with reducing Laemmli buffer. For RET immunoprecipitation, Northern blots were prepared and hybridized using standard methods. The lysates were combined with a 1:50 dilution of antibodies specific to the COOH full-length RET coding sequence, described above, was used as a probe to terminus of either the RET9 isoform (C-19; Santa Cruz Biotechnology, Santa detect RET expression. All other hybridization probes were generated by PCR Cruz, CA) or the RET51 isoform (C-20; Santa Cruz Biotechnology), for2hat using specific primers selected from the EST sequences found on our microar- 4°C, with shaking. Antibody complexes were collected with Protein A-Sepha- rays and used to generate PCR products. For all other genes, primers were rose CL-4B (Amersham Biosciences, Baie d’Urfe´, QC, Canada) or Protein G selected from longer published cDNA sequences. The relative differences in PLUS Agarose (Santa Cruz Biotechnology), depending on the primary anti- expression levels of some transcripts were also confirmed by quantitative body host species, and were washed with cold lysis buffer, taken up in real-time reverse transcription (RT)-PCR (qRT-PCR) using the LightCycler Laemmli buffer, and frozen. Immunoprecipitates or total protein lysates were System with the QuantiTect SYBR Green RT-PCR kit (Qiagen Inc, Missis- boiled 1 min and separated on either 6 or 10% SDS/PAGE gels. Protein was sauga, ON, Canada) with 200 ng total RNA as template according to manu- transferred to nitrocellulose membrane (Bio-Rad, Mississauga, ON, Canada) facturer’s instructions at an annealing temperature of 55°C. All of the PCR and were blocked overnight in 5% nonfat milk in Tris-buffered saline–Tween products were designed to detect only cDNA if possible. Each assay was 20. RET expression was detected using the C-19 or C-20 antibody and RET repeated at least three times. To confirm the specificity of product, melt curves tyrosine phosphorylation was detected using the 4G10 antibody (Upstate were generated over a 60°C–95°C range and a negative control, without Technologies, Lake Placid, NY). Coimmunoprecipitation of RET-associated cDNA, was run with each assay. Relative copy number was calculated using proteins SHC, and GRB2 was detected using appropriate antibodies (Upstate the crossing threshold method and assuming an efficiency of 2 [relative copy Technologies). Heat shock transcription factor 1 (HSF1) was detected on number ϭ 2dCT(37)]. Western blots of whole cell lysates using a specific antibody (Stressgen, Primers used to make hybridization probes for genes shown here included: Victoria, BC, Canada). Proteins of interest were detected by incubation with HSPA1L (5Ј-GAGCTCGATTTGAAGAGTTG-3Ј/5Ј-ATTGTGGGGCCTGT- the appropriate primary antibody (1–2hat37°C with shaking), washing with GGCAGG-3Ј), HSPA1A (5Ј-TGTGCTCCGACCTGTTCCGA-3Ј/5Ј-AATG- Tris-buffered saline–Tween 20, and incubating with horseradish peroxidase- GCCTGAGTTAAGTGTA-3Ј), HSPA1B (5Ј-TGTGCTCCGACCTGTTC- conjugated secondary antibodies. Antibody binding was visualized using an Ј Ј Ј Ј Enhanced Chemiluminescence Detection system (Amersham Biosciences, CGA-3 /5 -TACATTCCCAGCCTTTGTAG-3 ), STIP1 (5 -AGCGG ACG- Ј Ј Ј Ј Baie D’Urfe´, QC, Canada). GA TTCGATTCAA-3 /5 -AGGAGTTGCCAATTCGAGCA-3 ), STUB1 (5 - Ј Ј Immunokinase Assay. For in vitro kinase assays, equivalent amounts of GAGATGGAGAGCTATGATGA-3 /5 -AAAGCGATGCTGAGAGGGGA- Ј Ј Ј Ј cell lysates were immunoprecipitated with appropriate anti-RET serum in a 3 ), RNF19 (5 -TCATCTGTGAGCTTGCCTTC-3 /5 -ACATCCTTGCCT- Ј kinase lysing buffer [30 mM Tris-HCl (pH 8.0), 1% Triton X-100, 150 mM TCATAGCG-3 ). Primers used for qRT-PCR of genes shown here included: RET (5Ј-AATTTGGAAAAGTGGTCAAGGC-3Ј)/(5Ј-CTGCAGGC- NaCl, 1 mM EDTA, 0.3 mM Na3VO4], as described above. Precipitated antibody complexes were washed with kinase lysing buffer, then were incu- CCCATACAAT-3Ј), {186 bp}; DNAJC3 (5Ј-AACAGAGCCAAGCATT- ␮ GCTG-3Ј)/(5Ј-GGTTCCATCTGTAAAACTTC-3Ј), {125 bp}; RNF19 (5Ј- bated for 20 min (30°C) in 10 mM Tris-HCl (pH 7.4), 5 mM MgCl2, with 2 g myelin basic protein (MBP) as substrate and 10 ␮Cof[␥-32P]ATP (34). AACTAACACAGCTGTAGACA-3Ј)/(5Ј-TCACTCAGGTTGTCTCGGAT- Reactions were stopped by adding Laemmli buffer, and denatured samples 3Ј) {152 bp}; and HSPA1B (5Ј-GGTCCCAAGGCTTTCCAGAG-3Ј)/(5Ј- were assayed on 10% PAGE gels. Gels were fixed and dried and 32P-labeled ATGCCGGTGCCCTGCTCTGTGGGCTCCGCT-3Ј), {156 bp}. Primer bands were detected, and signal intensity quantified, by phosphorimager. sequences used to amplify other EST sequences are available on request. Microarray Analyses. Total RNA was isolated from cultured cells using TRIZOL reagent, according to the manufacturer’s instructions (Invitrogen). 1 See website: http://www.microarrays.ca/. cDNA microarrays containing 19,000 (19K) or 1700 (1.7K) expressed se- 2 See website: http://www.microarrays.ca/support/proto.html. 4454

RESULTS Isoform-Specific RET Cell Lines. We generated constructs containing the full-length RET9 or RET51 sequence under control of a minimal cytomegalovirus promoter. Constructs were transfected into the E293 embryonic kidney cell line, which does not express detectable RET protein, and cell lines stably expressing either RET9 or RET51 were generated (Fig. 1). Both the E293ϩRET9 (RET9) and ϩ E293 RET51 (RET51) lines express a Mr 155,000 and a Mr 175,000 protein detectable with RET isoform-specific antisera (Fig. 1B). These correspond to the previously reported partially and fully glycosylated Fig. 2. Representative MA plot. MA plot showing the intensity-dependant ratio of our RET protein forms, respectively. As predicted, the Mr 175,000 protein normalized microarray data. The log expression ratio for each spot was plotted against the isoform showed significant autophosphorylation, which was increased average of the log intensities of hybridization for each spot. Our normalized data clustered further in the presence of RET coreceptor GFR␣1 on addition of the around zero, suggesting minimal systemic differences (e.g., dye-label incorporation, detection bias) affecting our gene expression predictions. RET-ve, RET negative. RET ligand, GDNF (Fig. 1B). Immunoprecipitated RET9 and RET51 proteins from our stable cell lines also efficiently phosphorylated an exogenous substrate, myelin basic protein, in in vitro immunokinase both transient transfections and stable lines, we demonstrated that assays (data not shown), indicating that our cell lines expressed RET9 and RET51 bound SHC, whereas GRB2 could be coimmuno- functional RET kinases. precipitated only with the RET51 isoform (Fig. 1C). These data are To confirm that they were fully functional, we investigated the consistent with previous observations that SHC binds RET at phos- ability of our RET protein isoforms to bind normal RET substrates. In photyrosine 1062, a position common to both protein isoforms, whereas GRB2 interacts directly with phosphotyrosine 1096, which is found only in the RET51 isoform (reviewed in Refs. 13 and 38). RET-Induced Gene Expression. RET expression can stimulate cell division, transformation, migration, or differentiation, yet little is known about the differences in gene expression through which these processes are mediated. We used gene expression microarrays to investigate downstream target genes with altered expression when RET-mediated signaling occurs. Because previous studies of transgenic animals suggest that RET9 expression may have the most profound effects on cells (23), we focused initially on gene expression related to this isoform. We performed multiple replicates of compar- isons using E293 [RET negative (RET-ve)] and RET9 RNAs with both 1.7K (n ϭ 4 arrays) and 19K (n ϭ 3 arrays) cDNA microarrays1 (Fig. 2). RET9- and RET51-expressing lines were also directly compared with four repetitions using each of these arrays. Differences in gene expression between the E293 (RET-ve) and RET51 cells were confirmed with the 1.7K array. Individual cDNAs were represented on these arrays from two to six times per array. The relative expression ratios between pairs of RNA sources for each cDNA spot were calculated, as described above, and data for all of the arrays for a given comparison were pooled. Threshold-intensity ratios of Ն2 for up-regulation and Յ0.5 for down-regulation were used to identify significant expression differences. ESTs were chosen for further evaluation if relative expression differences exceeded the threshold values for a minimum of three array spots, based on normalized ratios, and if differences in log-transformed data between RNA sources were significant in paired t tests (P Ͻ 0.1) (35, 36). We chose these Fig. 1. RET isoform expression. A, schematic representation of the expressed RET stringent selection criteria to minimize the number of false-positive isoforms showing the alternatively spliced COOH-terminal sequences: SP, signal peptide; expression differences in our study. z TM, transmembrane domain. , COOH-terminal amino acids encoding the RET9-specific We identified 99 ESTs reproducibly differentially expressed in sequences; , COOH-terminal amino acids encoding the RET51-specific sequences. B, Western blot analysis showing RET expression and phosphorylation in E293 cells stably RET9-expressing cells as compared with the parental E293 cell line expressing the RET9 or RET51 isoform and in the parental E293 cell line. Cells were (RET-ve). Eighty ESTs, including 43 defined genes, were up-regu- either untreated (-) or transiently transfected with a GFR␣1 coreceptor expression construct and treated with 100 ng/ml GDNF for 48 h and 10 min before harvesting, lated in the presence of RET9. An additional 19 ESTs (nine genes) respectively (ϩ). RET proteins were immunoprecipitated (IP) with isoform-specific RET had lower expression in the presence of RET. Here, we have focused antisera and were immunoblotted (IB) either with the same antibodies (␣ RET) or with an exclusively on ESTs representing defined genes or predicted genes antiphosphotyrosine antibody (␣ pY). C, interaction of RET9 and RET51 with adaptor proteins SHC and GRB2. Expression of SHC isoforms (p46SHC, p52SHC) and GRB2 was that encode proteins for which a function is known or inferred (Table detected in lysates from E293 transiently transfected with RET9 or RET51 expression 1). Validation of our microarray data were performed by more sen- constructs or empty vector (top 2 panels). Equal amounts of cell lysates from these sitive Northern and/or qRT-PCR analyses. As we anticipated, the gene transfected cells were immunoprecipitated with RET isoform-specific antibodies and interaction with SHC and GRB2 was confirmed by immunoblotting with appropriate differing most significantly in expression on our microarrays was antibodies (bottom 2 panels). The RET9-specific antibody was used for immunoprecipi- RET, which was more than 7.8-fold overexpressed in our RET9- tation of the empty vector control. An IgG band detected in RET9 immunoprecipitates Ͻ results from the detection of the anti-RET9 antibody used in immunoprecipitations by the expressing stable cell line (P 0.001). These differences were rabbit secondary antibody used to detect SHC. In B and C, kDa, Mr in thousands. consistent with our Western, Northern, and qRT-PCR analyses of RET 4455

Table 1 Genes differentially expressed in response to RET isoforms and for which a protein of known or predicted function has been defined Genes with known or predicted roles in stress response are indicated in bold. Gene names represent the most current designations according to the Human Genome Organization (HUGO) Gene Nomenclature Committee (www.gene.ucl.ac.uk/nomenclature/). Identification of one or more potential HSEs within the proximal promoter of each gene is indicated (ϩ). Data are sorted on the ratio of gene expression in the presence of RET9 over RET51 or RET-ve. Average ratios represent the total available data and may not demonstrate a two-fold difference in expression overall. However, within this data set for each gene, a minimum of three spots did show a two-fold expression difference. All of the genes shown had a significant difference in expression (P Ͻ 0.1) based on paired t tests of log-transformed data. Relative Expression

Gene Symbol Gene Name/Description Ratio P HSE

RET9:RET-ve

RET9ϾRET-ve RET RET proto-oncogene 7.82 Ͻ0.001 PER3 Period homolog 3 (Drosophila) 3.51 Ͻ0.001 MAPKAPK2 Mitogen-activated protein kinase-activated protein kinase 2 3.39 0.003 SYMPK Symplekin; Huntingtin interacting protein I 3.35 0.002 ϩ KCNJ10 Potassium inwardly-rectifying channel, subfamily J, member 10 3.14 Ͻ0.001 HSPA1A Heat shock 70 kD protein 1 3.05 Ͻ0.001 ϩ DRLM Down-regulated in liver malignancy 2.98 0.001 EEA1 Early endosome antigen 1, 162 kD 2.93 0.005 CLDN5 Claudin 5 (transmembrane protein deleted velocardiofacial syndrome) 2.88 0.001 MRPL23 Ribosomal protein L23-like 2.88 Ͻ0.001 SDK1 Sidekick homolog 1 (chicken) 2.85 Ͻ0.001 HSPA1B Heat shock 70 kDa protein 1B 2.83 Ͻ0.001 ϩ POLR2I Polymerase (RNA) II (DNA directed) polypeptide I, 14.5 kDa 2.60 0.001 CROC4 Transcriptional activator of the c-fos promoter 2.55 Ͻ0.001 SUPT5H Suppressor of Ty (Saccharomyces cerevisiae) 5 homolog 2.48 Ͻ0.001 ATP2C1 ATPase, Caϩϩ transporting, type 2C, member 1 2.47 0.001 FADS3 Fatty acid desaturase 3 2.42 Ͻ0.001 INSM1 Insulinoma-associated 1 2.38 Ͻ0.001 LOC149420 Casein kinase 2.35 Ͻ0.001 DKFZp727A071 Similar to tRNA synthetase class II 2.27 Ͻ0.001 CLU Clusterin 2.21 0.015 ABO26190 Kelch motif containing protein 2.20 Ͻ0.001 CYB5R1 Cytochrome b5 reductase 1 (B5R.1) 2.17 Ͻ0.001 FKBP4 FK506-binding protein 4 (59 kD) 2.16 0.007 ϩ SLC16A3 Solute carrier family 16 (monocarboxylic acid transporters)-3 2.14 Ͻ0.001 SCARA3 Scavenger receptor class A, member 3 2.11 Ͻ0.001 JTB Jumping translocation breakpoint 2.07 0.037 HSPA1L Heat shock 70 kD protein-like 1 2.06 Ͻ0.001 ϩ DUSP8 Dual specificity phosphatase 8 2.04 0.052 ENDOG Endonuclease G 2.03 0.022 NOV Nephroblastoma overexpressed gene 2.02 0.003 AKAP9 A kinase (PRKA) anchor protein 9 2.01 Ͻ0.001 MOCS3 Molybdopterin synthase sulfurylase 2.01 Ͻ0.001 DDX9 DEAD/H box polypeptide 9 (RNA helicase A, leukophysin) 1.81 0.026 DELGEF Deafness locus associated putative guanine nucleotide exchange factor 1.80 0.001 GRP Gastrin-releasing peptide 1.80 Ͻ0.001 CTSC Cathepsin C 1.76 0.038 SCHIP1 Schwannomin interacting protein 1 1.76 0.013 S100A10 S100 calcium-binding protein A10 ͓annexin II ligand, calpactin I, light 1.73 0.006 ϩ polypeptide (p11)͔ HEY1 Hairy/enhancer-of-split related with YRPW motif 1 1.60 0.094 PFKL Phosphofructokinase, liver 1.56 0.071 CD151 CD151 antigen 1.51 0.043 MPP3 Membrane protein, palmitoylated 3 (MAGUK p55 subfamily 3) 1.45 0.010 RET-veϾRET9 RNF19 Ring finger protein 19, E3 ubiquitin ligase 0.43 0.001 CIT Citron (rho-interacting, serine/theorine kinase 21) 0.45 Ͻ0.001 PCBP2 Poly(rC)-binding protein 2 0.47 Ͻ0.001 THH Trichohyalin 0.50 Ͻ0.001 C6orf32 Chromosome 6 open reading frame 32 0.52 Ͻ0.001 HIC2 Hypermethylated in cancer 2 0.53 Ͻ.001 CGI-67 CGI-67 protein 0.54 Ͻ0.001 PSG11 Pregnancy specific ␤1-glycoprotein 11 0.71 0.026 RPL4 Ribosomal protein L4 0.72 0.001

RET9:RET51

RET9ϾRET51 SYN1 Synapsin 1 1.73 0.015 BPAG1 Bullous pemphigoid antigen 1 1.63 0.074 ELK4 ETS-domain protein 1.48 0.059 TSC Tescalcin 1.28 0.034 RET51ϾRET9 DNAJC3 DnaJ (Hsp40) homolog 0.52 0.008 AQP1 Aquaporin 1 0.77 0.012 RNF19 Ring finger protein 19, E3 ubiquitin ligase 0.79 0.020

4456

Fig. 3. Expression of HSP70 gene family in response to RET. Northern analysis of RET and HSP70 family genes HSPA1A, HSPA1B, and HSPA1L in cell lines stably expressing RET9, RET51 (two independent clones) or a hygromycin resistance vector RETϪve and in untransfected E293 cells. Expression for each transcript was normalized to GAPDH and is indicated as fold expression relative to E293. Variation in RET transcript sizes reflect differences in the full-length cDNAs of RET9 (4.08 kb) and RET51 (3.53 kb) because of different 3ЈUTR sequences.

expression (Figs. 1, 3, and 4), which identified no RET expression in the heat shock 70 (HSP70) family. Specifically, we found that RET9 our untransfected E293 cells. Interestingly, our Northern and qRT- expression reproducibly up-regulated expression of three stress induc- PCR analyses indicated that the relative expression of RET9 was ible members of the HSP70 gene family, HSPA1A, HSPA1B, and much greater than predicted by microarray analysis (Ͼ130-fold and HSPA1L, by more than 2-fold on microarray analysis (Tables 1 and Ͼ680-fold increase, respectively), suggesting the relatively lower 2). Using Northern analyses, we confirmed that each gene was up- sensitivity of the microarray analysis, and confirming the necessity for regulated in both RET9 and RET51 cell lines, but not in parental E293 validation of these results. cells nor in an E293 cell line expressing only an empty hygromycin Our analyses suggest that genes with a broad range of functions resistance vector (RET-ve) but not RET (Figs. 3 and 4). This sug- may be modulated specifically in response to RET expression. In gested that this expression pattern was not a nonspecific response to particular, we noticed increased expression of transcripts that are transfection or selection of our stable lines but was directly related to normally expressed in neural or neuroendocrine cell types but are not RET expression. Using gene-specific probes from the 5Ј and 3Ј normally highly expressed in E293 cells, including PER3, SDK1, untranslated sequences of each gene, we showed that HSPA1A and SUPT5H, INSM1, CLU, DUSP8, and AKAP9 (Table 1). An overlap- HSPA1B were expressed at similar levels and that both were more ping group of genes with expression primarily associated with tumors highly expressed than HSPA1L, although all three were up-regulated of neuroendocrine or other RET-expressing cell types such as INSM1, on RET expression (Fig. 3). Although all three genes form a single CLU, NOV, and GRP, were also up-regulated. Increased expression of gene cluster on chromosome 6p21.3 (39), other genes, such as BAT3, several genes associated with cytoskeletal interactions and formation which lie in the same region (40) showed no difference in expression of tight junctions was detected including SYMPK, CLDN5, and in response to RET (Table 2), confirming the specificity of the CD151 (Table 1). observed expression pattern. RET-Mediated Expression of Stress Response Genes. The most The HSP70 proteins are one of several HSP families involved in strikingly altered pattern of gene expression identified in response to formation of protein complexes that target cellular proteins either for RET was up-regulated expression of genes encoding inducible mem- refolding and maturation or for ubiquitinization and degradation (re- bers of the stress response protein families. In preliminary analyses, viewed in Refs. 41 and 42). A variety of HSPs, cochaperones, and a we noted increased expression of genes encoding several members of number of interacting molecules form part of these complexes, which

Fig. 4. Real-time PCR quantitation of gene expression in response to RET. Quantitations were performed by the crossing threshold (CT) method (37). A, amounts of RET expression in RET-ve, RET9, and RET51 cells were compared after PCR amplification and fluorescence monitoring of SYBR green for each cycle. B, crossing thresholds and relative fold differences in expression for genes modulated by RET were averaged over four to five experiments. RET itself and HSPA1B are used for comparison. Relative differences in expression calculated from microarray analyses are also shown.

4457

Table 2 Expression of stress response-related genes in response to RET, analyzed by microarray Genes were selected based on established relationships with stress response. Data are sorted on the ratio of gene expression in the presence versus the absence of RET expression (RET9:RET-ve). Relative expression

Gene Symbol Gene Name/Description Ratio P HSEa

RET9: RET-ve MAPKAPK2 Mitogen-activated protein kinase-activated protein kinase 2 3.39 0.003b ST13 Suppression of tumorigenicity 13, Hsp70 interacting protein (HIP) 3.21 0.007b ϩ HSPA1A Heat shock 70 kDa protein 1 3.05 Ͻ0.001b ϩ HSPA1B Heat shock 70 kDa protein 1B 2.83 Ͻ0.001b ϩ DNAJB2 DnaJ (Hsp40) homolog, subfamily B, member 2 2.55 0.046b FKBP4 FK506-binding protein 4 (immunophilin family) 2.16 0.007b ϩ SCARA3 Scavenger receptor class A, member 3 2.11 Ͻ0.001b HSPA1L Heat shock 70 kDa protein-like 1 2.06 Ͻ0.001b ϩ HSPCA Heat shock 90 kDa protein 1␣ 1.79 0.022b ϩ HSPH1 Heat shock 105/110 kDa protein 1 1.65 0.011b ϩ DNAJA1 DnaJ (Hsp40) homolog, subfamily A, member 1 1.47 Ͻ0.001b ϩ HSPA8 Heat shock 70 kDa protein 8, constitutive 1.45 Ͻ0.001b ϩ UBQLN2 Ubiquilin 2 (CHAP1) 1.45 0.015b AUH AU RNA-binding protein/enoyl-Coenzme A hydratase 1.44 0.254 HSPD1 Heat shock 60 kDa protein 1 (chaperonin) 1.34 0.021b ϩ BAT3 HLA-B associated transcript 3, (Scythe) 1.17 0.131 HSF1 Heat shock transcription factor 1 1.17 0.510 HSPA2 Heat shock 70 kDa protein 2 1.15 0.432 CDC37 CDC37 cell division cycle 37 homolog (S. cerevisiae) 1.14 0.030b ϩ BAG1 BCL2-associated athanogene 1.07 0.201 DNAJB1 DnaJ (Hsp40) homolog, subfamily B, member 1 1.04 0.630 ϩ RNF19 Ring finger protein 19, E3 ubiquitin ligase 0.43 0.001b DNAJC3 DnaJ (Hsp40) homolog, subfamily C, member 3 (P58IPK) 0.45 0.002b HSPA9B Heat shock 70 kDa protein 9B (mortalin-2), constitutive 0.79 0.002b STCH Stress 70 protein chaperone, microsome-associated, constitutive 0.83 0.151 HSPE1 Heat shock 10 kDa protein 1 (chaperonin 10) 0.97 0.395 ϩ APG-1 Heat shock protein (hsp110 family) 0.98 0.674 a HSE, heat shock element; ϩ, presence of predicted HSEs within the proximal promoter region of the gene. b Genes for which the ratio of relative expression was significant in paired t tests (P Ͻ 0.1); are variable in size and composition, depending on the processes and stringent threshold ratio was used (e.g., Ն 1.3 fold difference and interactions they promote. To investigate whether RET induced these P Ͻ 0.1) an additional 6 genes with significant expression differences HSP70-related processes, we reviewed the genes in Table 1 to deter- on microarrays and one additional gene on Northern analysis (Fig. 5) mine whether other genes with known or predicted roles in cellular were recognized. This included members of the HSP40, HSP60, stress responses, were also modulated on RET expression. We iden- HSP70, HSP90, and HSP100 families (HSPCA, HSPH1, DNAJA1, tified four additional genes that were modulated by RET and are HSPA8, and HSPD1) as well as the genes for HSP70 interacting known to be expressed in response to stress and/or to interact with proteins UBQLN2 (CHAP1) (Table 2) and STIP1 (Fig. 5). The only HSPs. Three of these, MAPKAPK2, FKBP4, and SCARA3, were additional gene identified as down-regulated in our full microarray up-regulated by more than 2-fold when RET was expressed. Ring data set, the HSP40 gene DNAJC3, was reduced by 2.3 fold in the finger protein 19 (RNF19), an E3 ubiquitin ligase that is localized in presence of RET9 but was not reduced on RET51 expression (Table Lewy bodies in Parkinson’s disease (43, 44) was significantly down- 2, Fig. 4). Further, the STUB1 gene, which encodes the ubiquitin regulated on RET9 expression (Tables 1 and 2). ligase CHIP, and was not represented on our microarrays, was down- Because our conditions for identifying genes modulated through regulated moderately in the presence of RET, particularly RET51, on RET were stringent, and because many genes associated with cellular Northern analyses (Fig. 5). Additional members of the HSP70 and stress responses are not highly expressed, we reviewed our total microarray data set for alterations in expression of other HSP genes, cochaperones, and additional proteins with an established relationship with HSP70. In particular, we reviewed these data for expression of any additional members of HSP27, HSP40, HSP60, HSP70, HSP90, and HSP100 gene families. We were able to identify 27 genes with known or predicted roles in stress response for which microarray data were available (Table 2). In addition, we investigated several genes not found on our microarrays but with known important roles in stress response by Northern or qRT-PCR analyses (e.g., Fig. 5). Our data suggest that expression of a number of additional stress- related genes, but not all such genes, are significantly increased in our RET-expressing cells, although these did not meet our initial stringent selection criteria (e.g., expression levels below our thresholds, fewer than three spots with ratios Ն2). In addition to the seven genes described above, our microarray analyses identified three additional genes, including ST13 (also known as HIP), and two members of the Fig. 5. Expression of stress-related proteins in cell lines stably expressing RET. A, Northern analysis of STIP1, STUB1, and RNF19 expression in RET-ve, RET9, and RET51 HSP40 gene family, DNAJB2 and DNAJC3, with more than 2-fold cells. Expression in each sample was normalized to GAPDH and relative expression, differences in expression (P Ͻ 0.1) on RET expression. If a less compared with the vector control, is indicated. 4458

moter region of each of our candidate genes using TFSearch (45) to predict the presence of HSEs, the recognition sequence for HSF1 binding. As we would predict, HSEs were associated with the genes for 10/14 of the stress response proteins up-regulated in response to RET (Table 2). Only two additional genes up-regulated by RET, SYMPK, and S100A10, contained HSEs. We found no HSEs associated with the genes down-regulated by RET expression or differentially expressed between RET isoforms (Table 1). Evaluation of Predicted RET Targets. Previous studies of gene expression patterns associated with RET activation have been limited. Many of these have relied on predicted candidates known to lie downstream of other receptor tyrosine kinases or to be up-regulated in response to neuro-differentiative agents (46–54). In one study, Wa-

Fig. 6. Induction of HSP70 transcripts by RET through HSF1 activation. Western tanabe et al. (48) used a relatively insensitive method, differential analyses showing that, in the presence of RET51, inactive HSF1 (4) becomes phospho- display analysis, to identify candidate genes expressed in NIH 3T3 rylated (bracket). Northern blot showing HSPA1B expression increases only in the cells on RET activation, although the RET isoform was not specified presence of RET51 and increased HSF1 phosphorylation. RET-ve, RET negative. and few of these were confirmed on rigorous validation (55). Here, we used sensitive microarray analyses to evaluate whether these candi- HSP40 gene families and other stress response related genes, showed dates were also modulated by RET in our E293 cell model system. no consistent pattern of expression in response to RET or were clearly Using our total microarray data, we were able to evaluate expression not regulated through RET activation (Table 2). These data further of 21 of the candidate genes previously predicted to show altered suggest that the stimulation of stress response genes we have observed expression in response to RET (Table 3). None of these showed Ն Ͼ is specific and not a generalized phenomenon affecting all heat shock 2-fold difference in expression. Nine genes had 1.3-fold increased family genes or even all of the members of a single heat shock family. expression in our RET9 cell line as compared with RET-ve cells, but HSF1 and RET Expression. Interestingly, the heat shock tran- this difference was significant (P Ͻ 0.1) for only four of these, the scription factor, HSF1, known to be a primary inducer of stress related immediate early genes FOS and JUN, the translation initiation factor expression of HSP proteins, and particularly of HSP70 genes, did E1F4G3, and CFL1 (Table 3). not differ in expression, irrespective of RET expression (Table 2). RET Isoform-Related Gene Expression Differences. The RET9 However, in the presence of RET, we detected an increase in phos- and RET51 protein isoforms have been shown to differ in normal phorylated, activated forms of HSF1, which correlated with increased expression and function, as well as in their ability to transform (21, 23, transcription of HSP70 genes (Fig. 6). Our data suggest that RET- 56). Here, we used gene expression microarray analysis to investigate mediated HSP70 gene expression is induced by activation of HSF1. If the differences in gene expression stimulated by these two RET increased HSF1 phosphorylation is a direct result of RET activation, isoforms. We identified eight ESTs (four genes) more highly ex- we would predict that other genes with heat shock elements (HSEs), pressed in the presence of the RET9 isoform (Table 1). Four ESTs the recognition sequence for HSF family transcription factors, in their (three genes) were more highly expressed in the presence of the promoters, might also be regulated through this mechanism, in re- RET51 isoform (Table 1; Figs. 4 and 5). Although we found that the sponse to RET. To investigate this, we examined the proximal pro- differences in the relative expression of RET in these lines was not

Table 3 Assessing relative expression of previously predicted RET target genes by microarray analysis For each gene, expression in the presence of RET9 relative to expression in RET-negative cells was evaluated, and genes were sorted on relative expression ratio. Relative expression

Gene symbol Gene name/description Ratio P

RET9:RET-ve Referencea

FOS v-fos FBJ murine osteosarcoma viral oncogene homolog 1.96 0.001b (44, 45) EIF4G3 Eukaryotic translation initiation factor 4␥, 3 1.84 0.005b (46) JUN v-jun avian sarcoma virus 17 oncogene homolog 1.53 0.016b (45) LOX Lysyl oxidase 1.46 0.135 (46) ITGA6 Integrin, ␣6 1.38 0.142 (46) CCND3 Cyclin D3 1.38 0.102 (47) CFL1 Cofilin, non-muscle isoform 1.37 0.021b (46) CDKN1B Cyclin-dependent kinase inhibitor 1B (p27, Kip1) 1.35 0.235 (47) CCND1 Cyclin D1 1.34 0.510 (46) CTSL Cathepsin L 1.27 0.125 (46) VEGF Vascular enothelial growth factor precursor 1.24 0.359 (48–51) ENO2 Enolase 2(␥, neuronal) 1.17 0.030b (48) EGR1 Early growth response protein 1 1.16 0.673 (44, 48, 49, 50) CXCL12 Chemokine ligand 12, stromal cell-derived factor 1 precursor 1.09 0.067 (46) GJA1 Gap junction protein, ␣1, 43 kDa (connexin 43) 1.08 0.835 (52) PTN Pleiotrophin precursor 1.06 0.900 (46) PLOD2 Procollagen-lysine, 2-oxoglutarate 5-dioxygenase (lysine hydroxylase) 2 1.05 0.701 (46) STC1 Stanniocalcin 1.01 0.903 (46) DCN Decorin 0.99 0.525 (46) TIMP3 Tissue inhibitor of metalloproteinase 3 0.90 0.073b (46) ANXA4 Annexin IV 0.86 0.848 (46) a Previous studies showing altered gene expression in response to RET. b Genes for which the ratio of relative expression was significant in paired t tests (P Ͻ 0.1). 4459

Downloaded from cancerres.aacrjournals.org on September 29, 2021. © 2004 American Association for Cancer Research. RET AND THE STRESS RESPONSE significant (P ϭ 0.76), we cannot exclude the possibility that differ- RET expression, was the up-regulation of proteins associated with ences observed were related to subtle variation in the levels of RET stress response, and in particular, inducible members of the HSP70 expressed. family. Initiation of a complex pattern of protein expression, termed the stress response, occurs as a result of a variety of environmental DISCUSSION cues including heat shock, heavy metals, oxidative stress, inflamma- tion, and others (41, 68). These proteins form complexes that recog- Although the RET receptor has important roles in cell growth, nize misfolded or immature peptide chains and either assist in their differentiation, and survival, and has been shown to be a critical appropriate folding and maturation, or target these peptides for ubiq- determinant of the development of kidney and neural crest lineages, uitinization and proteosomal degradation (41, 68; Fig. 7). HSP40, 70 little is known about the specific genes through which these effects are and 90 family members have roles in both of these processes and the mediated. Limited previous studies have focused on known or pre- ultimate fate of the immature peptide is dependent on the combination dicted candidate genes linked to other receptor tyrosine kinases (Table of chaperones, cochaperones, proteases, ubiquitin ligases and/or other 3) but have been less successful in predicting novel target genes proteins that comprise the specific stress response complex. related specifically to RET. Microarray analysis represents a powerful Members of five of the seven HSP gene families were represented tool for screening large numbers of ESTs simultaneously to identify on our microarrays (HSP100, HSP90, HSP70, HSP60, and HSP40). such target genes and to elucidate novel functions or interactions of We saw altered expression of multiple family members in response to RET. Here, we have generated a series of RET isoform-specific stable RET (Table 2) but particularly notable was significant up-regulation cell lines and have used these to address the nature of genes that lie in expression of the stress-inducible members of these gene families. downstream of RET activation. We identified 43 known genes up- For example, we saw increased expression of the inducible members regulated, and 9 genes down-regulated, in response to RET activation. of the HSP70 family, HSPA1A, HSPA1B, and HSPA1L (Fig. 3), As might be predicted as a result of proto-oncogene activation, the whereas constitutively expressed family members had no consistent up-regulated genes included a number of genes involved in cell–cell pattern of expression in the presence of RET (1HSPA8, 2HSPA9B, or cell–cytoskeletal interactions. Disruption of normal cell interac- no change HSPA2, STCH). The HSP40 gene, DNAJB2 was also tions is a common feature of oncogene activation. In fact, RET has up-regulated, whereas DNAJC3, an HSP40 family member that in- previously been associated with changes in phosphorylation of cy- hibits stress response (69), was down-regulated (Table 2). In addition, toskeletal molecules such as paxillin, p130cas, and focal adhesion our data suggest that a number of proteins associated with HSP70 kinase (57–59). Our data indicate that this change in phosphorylation protein complexes in peptide refolding, including HSP70/90-interact- is not accompanied by increased expression of either paxillin or focal ing proteins ST13, STIP1, and FKBP4, as well as HSP40 and 90 adhesion kinase (data not shown). However, we showed that the genes family members, seem generally to be increased in expression (Table for SYMPK, CLDN5, CD151, and MPP3, proteins associated with 2; Figs. 5 and 7). Conversely, genes more prominently associated with tight junctions or known to interact with integrins, were up-regulated, HSP-mediated peptide degradation, such as HSP70-interacting co- as were several proteins linked to lamina interaction and synapse chaperones BAG1 and BAT3, and ubiquitin ligases STUB1 and formation (e.g., SDK1, AKAP9). Previous studies have suggested that RNF19, either showed no change in expression or were relatively disruption of adhesion-dependent signaling may contribute to medul- decreased in the presence of RET (Table 2; Fig. 7). The induction of lary thyroid carcinoma (59) and that RET stimulation by its ligands stress proteins seen in our study was not a nonspecific response to has a role in regulating development of synapses (60). Our data overexpression of any exogenous protein, because our RET-ve cells, suggest that cells expressing RET may have altered interactions due to expressing the hygromycin resistance gene, did not have increased increased expression of a number of cell surface molecules that expression of these proteins, nor are these up-regulated generally in modulate cell interactions with its neighbors or with the substrate. other expression array studies. Changes in gene expression and in cellular interactions may, in Together, our data suggest that RET expression favors the produc- part, be explained by an increase in differentiation signals provided by tion of HSP complexes that target immature proteins for refolding or RET activation. In fact, we saw increased expression of a number maturation rather than for ubiquitinization and degradation in re- of neuroendocrine cell type-specific transcripts including PER3, sponse to cellular stresses. This preference is consistent with the KCNJ10, SDK1, SUPT5H, DUSP8, and AKAP9, as well as transcripts neuroprotective role found for RET and its ligand GDNF (70–72). In specific to other RET-sensitive tissue types such as developing sper- a number of neuronal lineages, the RET ligand GDNF has been matogonia (SYMPK, CLU, DDX9) or thyroid (JTB), suggesting that strongly linked to survival and blockage of apoptotic responses to these genes may be induced as part of the normal developmental roles ischemia, cell damage, nitric oxide, and metal ions (4–11). The of RET. RET has been shown to be required for neural crest cell inducible members of the HSP families are also up-regulated by migration, maturation of neural cell types of the peripheral and central similar stresses (68). For example, GDNF, RET, and HSP70s are all nervous systems, and for cell fate decisions in undifferentiated cells of up-regulated in brain in response to ischemia (70–72) and in adult the spermatogonia (61–63), and these genes may lie downstream of nerve cells in response to injury (73–75). In each case, these mole- RET in these processes. However, RET also has cell type-specific cules are thought to be critical to survival, and possibly regeneration, proliferative roles. RET activation can promote proliferation of some of the damaged neurons (73–75). Whereas RET- and HSP-mediated populations of neuroblasts, thyroid C-cells and adrenal chromaffin survival signals may be independent events, our data suggest that cells and is implicated in both medullary and papillary thyroid carci- GDNF effects are mediated through RET, at least in part, by enhanc- noma and pheochromocytoma (reviewed in Refs. 13 and 38). Con- ing or stimulating stress response that in turn triggers prosurvival or sistent with this role, we saw increased expression of several tran- antiapoptotic pathways. High levels of inducible HSP70s prevent scripts specifically associated with other endocrine or renal tumor stress-induced apoptosis and block caspase activity, mitochondrial types including INSM1 (64), NOV (65), and GRP (66). Overexpres- damage, and nuclear fragmentation (76, 77). RET can also have sion of other tumorigenesis-related genes, such as CLU, which confers proapoptotic activity in the absence of its ligand (no signaling) that is apoptosis resistance in a variety of cell types (67), was also detected blocked by the presence of GDNF and the induction of downstream in response to RET. signals (78). It is interesting to speculate that RET expression may The most intriguing gene expression pattern noted in this study on contribute to up-regulation of HSP proteins and, thus, acts as part of 4460

Fig. 7. Schematic diagram of HSP70-related proteins. The relationships of stress response proteins analyzed in this study are shown. Relative change in expression in response to RET of each transcript is indicated (1, up-regulated; 2, down-regulated; Ϫ, not significantly altered). Phosphorylated forms of HSF1 (P) are indicated. HSE, heat shock element. the normal stress response in some tissues and may even help to many of the genes modulated by RET are incomplete and future modulate it. analyses will be needed to evaluate the contribution of HSF1/HSE to Increased expression of HSPs, particularly members of the HSP70 RET-modulated gene expression. However, our present data indicate and HSP90 families, has also been frequently recognized in a variety that stimulation of the stress response through HSF1 activation rep- of tumor types including gastric, endometrial, and breast cancers, in resents only one of the mechanisms by which RET modulates gene which it is associated with poor prognosis and resistance to therapy expression, because 36 of the 42 genes up-regulated by RET do not (79, 80). Stress response is also thought to serve an antiapoptotic contain HSEs. function in these cell types, leading to clonal outgrowth. RET activa- As we would predict, we found more gene expression differences tion is increased in several neuroendocrine tumors including papillary between RET-positive and RET-ve cell lines than between lines and medullary thyroid carcinoma, pheochromocytoma, seminomas, expressing different RET isoforms. We detected only four genes and lung and renal tumors (reviewed in Refs. 13, 81, and 82). We, up-regulated and three genes down-regulated in RET9 cells as com- thus, might also predict that RET expression may enhance or regulate pared with RET51. This surprisingly small number of genes may, in HSP expression in these and other tumor types, as well as in neural part, be due to differences in the relative expression of RET9 and tissues. RET51 in our cell lines, which would have masked some variation. Our data suggest that increased expression of RET triggers phos- Previous studies have shown that RET9 and RET51 are functionally phorylation but not increased expression of HSF1 (Fig. 6), the major different, having distinct capacities for binding downstream adaptors vertebrate transcription factor associated with cellular and organismal and activating signaling pathways (reviewed in Refs. 13 and 38). stress responses. HSF1 expression is not usually up-regulated on Using monoisoformic RET transgenic mouse models, de Graaff et al. induction of environmental stresses. Instead, existing monomers of (23) have shown that RET9, but not RET51, is required for develop- HSF1 become activated by phosphorylation, oligomerization, and ment of the kidney and enteric nervous system. Conversely, RET51, translocation to the nucleus, in which they bind specifically to HSEs but not RET9, is required for metabolism and growth of mature and induce gene transcription (reviewed in Ref. 83). HSEs are found sympathetic neurons (85). Thus, it is likely that RET9 and RET51 in the promoters of most stress-inducible genes but also have impor- may be associated with different gene expression patterns in kidney tant roles in normal developmental gene expression (84). We identi- and neural cell types. In this study, we have used kidney-derived cell fied HSEs close to the transcription start site of a number of stress lines to investigate gene expression associated with either RET9 or response genes up-regulated by RET (Table 2). However, very few of RET51, but it will be interesting to expand these studies to compare the genes identified in our initial expression microarray screen were gene expression in other cell types, such as neural lineages. predicted to be regulated through HSF1, because only two additional Our data show that RET leads to expression of a broad range of genes, SYMPK and S100A10, contained HSEs. As we would expect, genes including genes not previously linked to receptor tyrosine if RET does activate HSF1, none of the genes down-regulated by RET kinases and others with known or predicted roles in cell proliferation expression contain HSEs. The data available on the promoters of and differentiation. Interestingly, our data show a strong and specific 4461

Downloaded from cancerres.aacrjournals.org on September 29, 2021. © 2004 American Association for Cancer Research. RET AND THE STRESS RESPONSE up-regulation of genes associated with stress response and, in partic- 25. Arighi E, Alberti L, Torriti F, et al. Identification of Shc docking site on Ret tyrosine ular, the inducible members of the HSP70 family. These findings may kinase. Oncogene 1997;14:773–82. 26. Lorenzo MJ, Gish GD, Houghton C, et al. RET alternative splicing influences the suggest a direct role for RET in stress responses and may, in part, interaction of activated RET with the SH2 and PTB domains of Shc, and the SH2 explain the neuroprotective function of RET and of GDNF in response domain of Grb2. Oncogene 1997;14:763–71. 27. Borrello MG, Pelicci G, Arighi E, et al. The oncogenic versions of the ret and trk to brain or nerve injury. tyrosine kinases bind Shc and Grb2 adaptor proteins. Oncogene 1994;9:1661–8. 28. Grimm J, Sachs M, Britsch S, et al. Novel p62dok family members, dok-4 and dok-5, ACKNOWLEDGMENTS are substrates of the c-Ret receptor tyrosine kinase and mediate neuronal differentiation. J Cell Biol 2001;154:345–54. We thank Drs. Harriet Feilotter and Scott Andrew, and Andrew Day and 29. Melillo RM, Carlomagno F, De Vita G, et al. The insulin receptor substrate (IRS)-1 recruits phosphatidylinositol 3-kinase to Ret: evidence for a competition between Shc Julie Shaw for helpful discussion. and IRS-1 for the binding to Ret. Oncogene 2001;20:209–18. 30. Melillo RM, Santoro M, Ong SH, et al. Docking protein frs2 links the protein tyrosine REFERENCES kinase ret and its oncogenic forms with the mitogen-activated protein kinase signaling cascade. Mol Cell Biol 2001;21:4177–87. 1. Avantaggiato V, Dathan NA, Grieco M, et al. Developmental expression of the RET 31. Myers SM, Eng C, Ponder BAJ, Mulligan LM. Characterization of RET proto- protooncogene. Cell Growth Differ 1994;5:305–11. oncogene 3Ј splicing variants and polyadenylation sites: a novel C terminus for RET. 2. Pachnis V, Mankoo B, Costantini F. Expression of the c-ret proto-oncogene during Oncogene 1995;11:2039–45. mouse embryogenesis. Development (Camb) 1993;119:1005–17. 32. Sanicola M, Hession C, Worley D, et al. Glial cell line-derived neurotrophic factor- 3. Attie-Bitach T, Abitbol M, Gerard M, et al. Expression of the RET proto-oncogene dependent RET activation can be mediated by two different cell-surface accessory in human embryos. Am J Med Genet 1998;80:481–6. proteins. Proc Natl Acad Sci USA 1997;94:6238–43. 4. Kitagawa H, Hayashi T, Mitsumoto Y, Koga N, Itoyama Y, Abe K. Reduction of 33. Myers S, Worley D, Cote G, Mulligan L. Analysis of GFRalpha1 polymorphism and ischemic brain injury by topical application of glial cell line-derived neurotrophic splice variants; effect on GFRalpha1/GDNF/RET complexes. Am J Hum Genet factor after permanent middle cerebral artery occlusion in rats. Stroke 1998;29: 1999;65:A377. 1417–22. 34. Kato M, Takeda K, Kawamoto Y, et al. Repair by Src kinase of function-impaired 5. Henderson C, Phillips H, Pollock R, et al. GDNF: a potent survival factor for RET with multiple endocrine neoplasia type 2A mutation with substitutions of motoneurons present in peripheral nerve and muscle. Science (Wash DC) 1994;266: tyrosines in the COOH-terminal kinase domain for phenylalanine. Cancer Res 2002; 1062–4. 62:2414–22. 6. Mograbi B, Bocciardi R, Bourget I, et al. The sensitivity of activated Cys Ret mutants 35. Forster T, Roy D, Ghazal P. Experiments using microarray technology: limitations to glial cell line-derived neurotrophic factor is mandatory to rescue neuroectodermic and standard operating procedures. J Endocrinol 2003;178:195–204. cells from apoptosis. Mol Cell Biol 2001;21:6719–30. 36. Knudsen S, Workman C, Sicheritz-Ponten T, Friis C. GenePublisher: automated 7. Mograbi B, Bocciardi R, Bourget I, et al. Glial cell line-derived neurotrophic analysis of DNA microarray data. Nucleic Acids Res 2003;31:3471–6. factor-stimulated phosphatidylinositol 3-kinase and Akt activities exert opposing 37. Bernard PS, Wittwer CT. Real-time PCR technology for cancer diagnostics. Clin effects on the ERK pathway: importance for the rescue of neuroectodermic cells. Chem 2002;48:1178–85. J Biol Chem 2001;276:45307–19. 38. Manie S, Santoro M, Fusco A, Billaud M. The RET receptor: function in development 8. De Vita G, Melillo RM, Carlomagno F, et al. Tyrosine 1062 of RET-MEN2A and dysfunction in congenital malformation. TIG 2001;17:580–9. mediates activation of Akt (protein kinase B) and mitogen-activated protein kinase 39. Milner CM, Campbell RD. Structure and expression of the three MHC-linked HSP70 pathways leading to PC12 cell survival. Cancer Res 2000;60:3727–31. genes. Immunogenetics 1990;32:242–51. 9. Oppenheim RW, Houenou LJ, Parsadanian AS, Prevette D, Snider WD, Shen L. Glial 40. Spies T, Blanck G, Bresnahan M, Sands J, Strominger JL. A new cluster of genes cell line-derived neurotrophic factor and developing mammalian motoneurons: reg- within the human major histocompatibility complex. Science (Wash DC) 1989;243: ulation of programmed cell death among motoneuron subtypes. J Neurosci 2000;20: 214–7. 5001–11. 41. Nollen EA, Morimoto RI. Chaperoning signaling pathways: molecular chaperones as 10. Oppenheim RW, Houenou LJ, Johnson JE, et al. Developing motor neurons rescued stress-sensing ’heat shock’ proteins. J Cell Sci 2002;115:2809–16. from programmed and axotomy-induced cell death by GDNF. Nature (Lond.) 1995; 42. Hohfeld J, Cyr DM, Patterson C. From the cradle to the grave: molecular chaperones 373:344–6. that may choose between folding and degradation. EMBO Rep 2001;2:885–90. 11. Lingor P, Unsicker K, Krieglstein K. GDNF and NT-4 protect midbrain dopaminergic 43. Niwa J, Ishigaki S, Doyu M, Suzuki T, Tanaka K, Sobue G. A novel centrosomal neurons from toxic damage by iron and nitric oxide. Exp Neurol 2000;163:55–62. ring-finger protein, dorfin, mediates ubiquitin ligase activity. Biochem Biophys Res 12. Airaksinen MS, Saarma M. The GDNF family: signalling, biological functions and Commun 2001;281:706–13. therapeutic value. Nat Rev Neurosci 2002;3:383–94. 44. Ito T, Niwa J, Hishikawa N, Ishigaki S, Doyu M, Sobue G. Dorfin localizes to Lewy 13. Hansford JR, Mulligan LM. Multiple endocrine neoplasia type 2 and RET: from bodies and ubiquitylates synphilin-1. J Biol Chem 2003;278:29106–14. neoplasia to neurogenesis. J Med Genet 2000;37:817–27. 45. Heinemeyer T, Wingender E, Reuter I, et al. Databases on transcriptional regulation: 14. Santoro M, Melillo RM, Carlomagno F, Fusco A, Vecchio G. Molecular mechanisms TRANSFAC, TRRD and COMPEL. Nucleic Acids Res 1998;26:362–7. of RET activation in human cancer. Ann N Y Acad Sci 2002;963:116–21. 46. Colucci-D’Amato GL, D’Alessio A, Califano D, et al. Abrogation of nerve growth 15. Chiariello M, Visconti R, Carlomagno F, et al. Signalling of the Ret receptor tyrosine factor-induced terminal differentiation by ret oncogene involves perturbation of kinase through the c-Jun NH -terminal protein kinases (JNKS): evidence for a 2 nuclear translocation of ERK. J Biol Chem 2000;275:19306–14. divergence of the ERKs and JNKs pathways induced by Ret. Oncogene 1998;16: 47. van Puijenbroek AA, van Weering DH, van den Brink CE, et al. Cell scattering of 2435–45. 16. Murakami H, Iwashita T, Asai N, et al. Enhanced phosphatidylinositol 3-kinase SK-N-MC neuroepithelioma cells in response to Ret and FGF receptor tyrosine kinase activity and high phosphorylation state of its downstream signalling molecules activation is correlated with sustained ERK2 activation Oncogene 1997;14:1147–57. mediated by ret with the MEN 2B mutation. Biochem Biophys Res Commun 48. Watanabe T, Ichihara M, Hashimoto M, et al. Characterization of gene expression 1999;262:68–75. induced by RET with MEN2A or MEN2B mutation. Am J Pathol 2002;161:249–56. 17. Borrello MG, Alberti L, Arighi E, et al. The full oncogenic activity of Ret/ptc2 49. Melillo RM, Barone MV, Lupoli G, et al. Ret-mediated mitogenesis requires Src depends on tyrosine 539, a docking site for phospholipase C gamma. Mol Cell Biol kinase activity. Cancer Res 1999;59:1120–6. 1996;16:2151–63. 50. Califano D, Monaco C, De Vita G, et al. Activated RET/PTC oncogene elicits 18. Hayashi H, Ichihara M, Iwashita T, et al. Characterization of intracellular signals via immediate early and delayed response genes in PC12 cells. Oncogene 1995;11: tyrosine 1062 in RET activated by glial cell line-derived neurotrophic factor. Onco- 107–12. gene 2000;19:4469–75. 51. Califano D, D’Alessio A, Colucci-D’Amato GL, et al. A potential pathogenetic 19. Hayashi Y, Iwashita T, Murakamai H, et al. Activation of BMK1 via tyrosine 1062 mechanism for multiple endocrine neoplasia type 2 syndromes involves ret-induced in RET by GDNF and MEN2A mutation. Biochem Biophys Res Commun 2001;281: impairment of terminal differentiation of neuroepithelial cells. Proc Natl Acad Sci 682–9. USA 1996;93:7933–7. 20. Tahira T, Ishizaka Y, Itoh F, Sugimura T, Nagao M. Characterization of ret proto- 52. Califano D, Rizzo C, D’Alessio A, et al. Signaling through Ras is essential for Ret oncogene mRNAs encoding two isoforms of the protein product in a human neuro- oncogene-induced cell differentiation in PC12 cells. J Biol Chem 2000;275:19297– blastoma cell line. Oncogene 1990;5:97–102. 305. 21. Rossel M, Pasini A, Chappuis S, et al. Distinct biological properties of two RET 53. Xing S, Furminger TL, Tong Q, Jhiang SM. Signal transduction pathways activated isoforms activated by MEN 2A/FMTC and MEN 2B mutations. Oncogene 1997;14: by RET oncoproteins in PC12 pheochromocytoma cells. J Biol Chem 1998;273: 265–75. 4909–14. 22. Ishiguro Y, Iwashita T, Murakami H, et al. The role of amino acids surrounding 54. D’Alessio A, De Vita G, Cali G, et al. Expression of the RET oncogene induces tyrosine 1062 in ret in specific binding of the shc phosphotyrosine-binding domain. differentiation of SK-N-BE neuroblastoma cells. Cell Growth Differ 1995;6:1387–94. Endocrinology 1999;140:3992–8. 55. Kurokawa K, Kawai K, Hashimoto M, Ito Y, Takahashi M. Cell signalling and gene 23. de Graaff E, Srinivas S, Kilkenny C, et al. Differential activities of the RET tyrosine expression mediated by RET tyrosine kinase. J Intern Med 2003;253:627–33. kinase receptor isoforms during mammalian embryogenesis. Genes Dev 2001;15: 56. Ivanchuk SM, Eng C, Cavenee WK, Mulligan LM. The expression of RET and its 2433–44. multiple splice forms in developing human kidney. Oncogene 1997;14:1811–8. 24. Alberti L, Borrello MG, Ghizzoni S, Torriti F, Rizzetti MG, Pierotti MA. Grb2 57. Murakami H, Iwashita T, Asai N, Iwata Y, Narumiya S, Takahashi M. Rho-dependent binding to the different isoforms of Ret tyrosine kinase. Oncogene 1998;17:1079– and -independent tyrosine phosphorylation of focal adhesion kinase, paxillin and 1087. p130Cas mediated by Ret kinase. Oncogene 1999;18:1975–82. 4462

58. Romano A, Wong WT, Santoro M, Wirth PJ, Thorgeirsson SS, DiFiore PP. The high 72. Wang Y, Chang CF, Morales M, Chiang YH, Hoffer J. Protective effects of glial cell transforming potency of erbB-2 and ret is associated with phosphorylation of paxillin line-derived neurotrophic factor in ischemic brain injury. Ann N Y Acad Sci 2002; and a 23 kDa protein. Oncogene 1994;9:2923–33. 962:423–37. 59. Kim LT, Fleming JB, Lopez-Guzman C, Nwariaku F. Focal adhesions and associated 73. Newfry GA, Jones KJ. Differential effects of facial nerve transection on heat shock proteins in medullary thyroid carcinoma cells. J Surg Res 2003;111:177–84. protein 70 expression in the developing and adult hamster facial nucleus. Metab Brain 60. Wang CY, Yang F, He XP, et al. Regulation of neuromuscular synapse development Dis 1998;13:253–7. by glial cell line-derived neurotrophic factor and neurturin. J Biol Chem 2002;277: 74. Naveilhan P, ElShamy WM, Ernfors P. Differential regulation of mRNAs for GDNF 10614–25. and its receptors Ret and GDNFR alpha after sciatic nerve lesion in the mouse. Eur 61. Schuchardt A, D’Agati V, Larsson-Blomberg L, Costantini F, Pachnis V. Defects in J Neurosci 1997;9:1450–60. the kidney and enteric nervous system of mice lacking the tyrosine kinase receptor 75. Trupp M, Belluardo N, Funakoshi H, Iba´n˜ez CF. Complementary and overlapping Ret. Nature (Lond.) 1994;367:380–3. expression of glial cell line-derived neurotrophic factor (GDNF), c-ret proto-onco- 62. Meng X, Lindahl M, Hyvonen ME, et al. Regulation of cell fate decision of gene, and GDNF receptor-␣ indicates multiple mechanisms of trophic actions in the undifferentiated spermatogonia by GDNF. Science (Wash DC) 2000;287:1489–93. adult rat CNS. J Neurosci 1997;17:3554–67. 63. Durbec PL, Larsson-Blomberg LB, Schuchardt A, Costantini F, Pachnis V. Common 76. Mosser DD, Caron AW, Bourget L, Denis-Larose C, Massie B. Role of the human origin and developmental dependence on c-ret of subsets of enteric and sympathetic heat shock protein hsp70 in protection against stress-induced apoptosis. Mol Cell Biol neuroblasts. Development (Camb) 1996;122:349–58. 1997;17:5317–27. 64. Lan MS, Russell EK, Lu J, Johnson BE, Notkins AL. IA-1, a new marker for 77. Buzzard KA, Giaccia AJ, Killender M, Anderson RL. Heat shock protein 72 modu- neuroendocrine differentiation in human lung cancer cell lines. Cancer Res 1993;53: lates pathways of stress-induced apoptosis. J Biol Chem 1998;273:17147–53. 4169–71. 78. Bordeaux MC, Forcet C, Granger L, et al. The RET proto-oncogene induces apopto- 65. Martinerie C, Perbal B. Expression of a gene encoding a novel potential IGF binding sis: a novel mechanism for Hirschsprung disease. EMBO J 2000;19:4056–63. protein in human tissues. C R Acad Sci III 1991;313:345–51. 66. Oremek GM, Sapoutzis N. Pro-gastrin-releasing peptide (Pro-GRP), a tumor marker 79. Jaattela M. Escaping cell death: survival proteins in cancer. Exp Cell Res 1999;248: for small cell lung cancer. Anticancer Res 2003;23:895–8. 30–43. 67. Trougakos IP, Gonos ES. Clusterin/apolipoprotein J in human aging and cancer. Int 80. Jaattela M. Heat shock proteins as cellular lifeguards. Ann Med 1999;31:261–71. J Biochem Cell Biol 2002;34:1430–48. 81. Devouassoux-Shisheboran M, Mauduit C, Tabone E, Droz JP, Benahmed M. Growth 68. Jolly C, Morimoto RI. Role of the heat shock response and molecular chaperones in regulatory factors and signalling proteins in testicular germ cell tumours. APMIS oncogenesis and cell death. J Natl Cancer Inst (Bethesda) 2000;92:1564–72. 2003;111:212–24. 69. Melville MW, Tan SL, Wambach M, Song J, Morimoto RI, Katze MG. The cellular 82. Futami H, Egawa S, Takasaki K, Tsukada T, Shiraishi M, Yamaguchi K. Allelic loss inhibitor of the PKR protein kinase, P58 (IPK), is an influenza virus-activated of DNA locus of the RET proto-oncogene in small cell lung cancer. Cancer Lett co-chaperone that modulates heat shock protein 70 activity. J Biol Chem 1999;274: 2003;195:59–65. 3797–803. 83. Pirkkala L, Nykanen P, Sistonen L. Roles of the heat shock transcription factors in 70. Arvidsson A, Kokaia Z, Airaksinen MS, Saarma M, Lindvall O. Stroke induces regulation of the heat shock response and beyond. FASEB J 2001;15:1118–31. widespread changes of gene expression for glial cell line-derived neurotrophic factor 84. Xiao X, Zuo X, Davis AA, et al. HSF1 is required for extra-embryonic development, family receptors in the adult rat brain. Neuroscience 2001;106:27–41. postnatal growth and protection during inflammatory responses in mice. EMBO J 71. Sarabi A, Chang CF, Wang Y, Tomac AC, Hoffer BJ, Morales M. Differential 1999;18:5943–52. expression of the cell line-derived neurotrophic factor (GDNF) receptor GFRalpha1 85. Tsui-Pierchala BA, Ahrens RC, Crowder RJ, Milbrandt J, Johnson EM Jr. The long in heterozygous Gfralpha1 null-mutant mice after stroke. Neurosci Lett 2003;341: and short isoforms of Ret function as independent signaling complexes. J Biol Chem 241–5. 2002;277:34618–25.

4463

Shirley M. Myers and Lois M. Mulligan

Cancer Res 2004;64:4453-4463.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/64/13/4453

Cited articles This article cites 84 articles, 37 of which you can access for free at: http://cancerres.aacrjournals.org/content/64/13/4453.full#ref-list-1

Citing articles This article has been cited by 4 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/64/13/4453.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/64/13/4453. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.