Identification of H-Ras, Rhoa, Rac1 and Cdc42 Responsive Genes

Identiﬁcation of H-Ras, RhoA, Rac1 and Cdc42 responsive genes

Hidemi Teramoto1, Renae L Malek2, Babak Behbahani2, Maria Domenica Castellone1, Norman H Lee2,3 and J Silvio Gutkind*,1

1Oral and Pharyngeal Cancer Branch, National Institute of Dental Research, National Institutes of Health, Bethesda, MD 20892- 4330, USA; 2Department of Functional Genomics, The Institute for Genomic Research, Rockville, MD 20850, USA; 3Department of Pharmacology, The George Washington University Medical Center, Washington DC 20037, USA

The superfamily of small GTP-binding proteins has Introduction expanded dramatically in recent years. The Ras family has long been associated with signaling pathways Small GTP-binding proteins of the Ras family are contributing to normal and aberrant cell growth, while essential components of biochemical pathways transdu- Rho-related protein function is to integrate extracellular cing proliferative signals from the vast majority of signals with specific targets regulating cell morphology, growth factor receptors (Bar-Sagi and Hall, 2000). At cell aggregation, tissue polarity, cell motility and cytokin- the molecular level, cycling between an active GTP- esis. Recent findings suggest that certain Rho proteins, bound form and an inactive GDP-bound form, Ras including RhoA, Rac1 and Cdc42, can also play a role in regulates the activity of a number of key biochemical signal transduction to the nucleus and cell growth control. routes, including a cascade of serine–threonine kinases However, the nature of the genes regulated by Ras and represented by Raf and Mek, which culminate with the Rho GTPases, as well as their contribution to their activation of two mitogen-activated kinases (MAPK numerous biological effects is still largely unknown. To kinases), ERK1 and ERK2. Activation of these MAP approach these questions, we investigated the global gene kinases, in turn, controls the activity of nuclear expression pattern induced by activated forms of H-Ras, transcription factors that are critical for cell growth RhoA, Rac1 and Cdc42 using cDNA microarrays (Bar-Sagi and Hall, 2000). On the other hand, Rho comprising 19 117 unique elements. Using this approach, family GTPases transduce extracellular signals to we identified 1184 genes that were up- or downregulated specific targets regulating cell morphology, cell aggrega- by at least twofold. Hierarchical cluster analysis revealed tion, tissue polarity, cytokinesis, cell motility and the existence of patterns of gene regulation both unique smooth muscle contraction (Narumiya, 1996; Takai and common to H-Ras V12, RhoA QL, Rac1 QL and et al., 2001). For example, when microinjected into Cdc42 QL activation. For example, H-Ras V12 upregu- Swiss 3T3 fibroblasts, activated Rho rapidly induces the lated osteopontin and Akt 1, and H-Ras and RhoA formation of actin stress fibers and focal adhesion stimulated cyclin G1, cyclin-dependent kinase 8, cyclin A2 contacts, while other members of the Rho family, such and HMGI-C, while Rac1 QL and Cdc42 QL upregu- as Cdc42 and Rac, regulate, respectively, the formation lated extracellular matrix and cell adhesion proteins such of filopodia, and membrane ruffles and lamellipodia as alpha-actinin 4, procollagen type I and V and (Nobes and Hall, 1995). neuropilin. Furthermore, H-Ras V12 downregulated by Interestingly, recent work has established that Rho >eightfold 52 genes compared to only three genes by proteins are also integral components of signaling RhoA QL, Rac1 QL and Cdc42 QL. These results pathways leading to transcriptional control. For exam- provide key information to begin unraveling the complex- ple, Rac and Cdc42 have been reported to activate c-Jun ity of the molecular mechanisms underlying the trans- amino-terminal kinase (JNK) thereby regulating the forming potential of Ras and Rho proteins, as well as the transcriptional activity of c-Jun (Coso et al., 1995), and numerous morphological and cell cycle effects induced by Rho has been shown to induce expression from the these small GTPases. serum responsive element (SRE) through the transcrip- ONCOGENOMICS Oncogene (2003) 22, 2689–2697. doi:10.1038/sj.onc.1206364 tional activation of the serum response factor (SRF) (Hill et al., 1995). However, the nature of the genes Keywords: foci formation; small GTPases; microarray whose expression is regulated by Ras and Rho family members as well as their contribution to the numerous biological functions of these GTPases, including cell *Correspondence: JS Gutkind, Oral and Pharyngeal Cancer Branch, transformation (Perona et al., 1993), is still largely National Institute of Dental and Craniofacial Research, National unknown. To begin addressing this issue, we used Institutes of Health, 9000 Rockville Pike, Building 30, Room 212, cDNA microarray mouse chips to examine the global Bethesda, MD 20892-4330, USA; E-mail: [email protected] pattern of gene expression in NIH 3T3 cells stably Received 1 November 2002; revised 31 December 2002; accepted 6 expressing activated alleles of H-Ras, RhoA, Rac and January 2003 This article is a ‘United States Government Work’ paper as defined by Cdc42. We observed that 1184 out of 19 117 unique the US Copyright Act. genes were up- or downregulated by at least twofold by Identification of the small GTPases responsive gen H Teramoto et al 2690 these GTPases. Furthermore, we found that H-Ras V12 a Control H-Ras V12 RhoA QL potently decrease by more than eightfold the expression of 52 genes compared to only three genes in association with the activated form of Rho family members. Of interest was the identification of similarly coregulated genes by H-Ras V12 and RhoA, possibly explaining some of the equivalent biological effects between these two small Gproteins.

Rac1 QL Cdc42 QL Results

To begin exploring the nature of the genes regulated by H-Ras, RhoA, Rac1 and Cdc42 that may contribute to cell growth control, we compared the ability of their activated alleles to induce focus formation in NIH 3T3 cells. We observed that H-RasV12 was the most potent oncogene, around 10 times more potent than RhoA QL. In turn, activated Rho A was much more potent than Rac1 QL and Cdc42 QL that are poorly transforming in these cells (Figure 1a, b). The shape of the H-RasV12 b 1200 foci, diffuse and extended, was also quite different from that induced by RhoA QL, Rac1 QL and Cdc42 QL, which displayed a punctuate morphology (Figure 1a). 800

As these results indicated that there is a hierarchy of g DNA) transforming ability among these GTPases, we next µ 100 sought to explore the nature of the genes regulated by (foci/ each of these small GTPases, using a cDNA microarray analysis approach with a comprehensive mouse cDNA 0 chip that includes 26 383 elements (19 117 unique genes). For these experiments, we examined the total RNA

from established mass cultures of NIH 3T3 cells V12

expressing H-RasV12, RhoA QL, Rac1 QL and Cdc42 Control Rac1 QL RhoA QL

QL. All GTPases were expressed at comparable levels Cdc42 QL (Figure 2a). For array analysis, cDNA was transcribed H-Ras from total RNA purified from each cell line and Figure 1 (a) Focus formation assay in NIH3T3 cells transfected with small Gproteins. NIH 3T3 cells were transfected by the indirectly labeled with the fluorescent Cy5 dye. The calcium-phosphate technique with pCEFL AU5 vector (control) or labeled cell line cDNA was mixed with an equal amount with the same expression vector carrying cDNAs for H-RasV12, of a reference probe consisting of Cy3 labeled cDNA RhoA QL, Rac1 QL and Cdc42 QL. Cells were cultured for 2–3 made in parallel from empty vector transfected cells weeks in 5% calf serum, fixed and then stained. (b) Similar results (AU5). Using a highly stringent criterion (see Methods), were obtained for each experimental plasmid in three independent experiments. Data represent the number of foci per mgof we observed that of the 26 383 elements represented on transfected DNA (average7s.e.m.) these mouse arrays, 1401 elements (1184 unique genes) varied more than twofold (up or down) in three independent experiments (Figure 2b) (see Supple- mentary Data for the complete list of genes). These Cluster 2 includes Rac1 QL and Cdc42 QL specific results indicated that around 7.3% of all genes upregulated genes (14 elements, 14 unique genes). The examined are affected by activated Ras and Rho complete list of genes and their relative changes can be GTPases. found in Supplemental Table 1. As shown in Figure 2b, the hierarchical gene cluster To examine in more detail the different gene expres- expression pattern indicates that genes regulated by sion patterns induced by H-Ras, RhoA, Rac1 and Rac1 and Cdc42 are more correlated with the expression Cdc42, we listed some of those genes whose expression pattern induced by RhoA than with the expression was highly up- or downregulated by each of these profile of Ras. This hierarchy of relatedness also reflects activated GTPases (Table 1). From this table, we can their structural similarity (Takai et al., 2001) (Figure 2b, observe that there are substantial differences among the Supplemental Table 1). Furthermore, we observed that genes affected by these small Gproteins. For example, H-Ras and RhoA could induce the expression of a set of expression of osteopontin, a multifunctional acidic genes different from those stimulated by Rac1 and phosphoglycoprotein, was dramatically increased by Cdc42 (Clusters 1 and 2) (Figure 2c). For example, H-Ras V12 (12.7-fold), but downregulated by Rac1 Cluster 1 represents H-Ras V12 and RhoA QL QL (À2.6-fold). Early growth response protein 1 upregulated genes (82 elements, 56 unique genes), and (Egr-1) was upregulated by all Rho GTPases, RhoA

Oncogene Identiﬁcation of the small GTPases responsive gen H Teramoto et al 2691

Figure 2 (a) Expression of each small Gprotein was confirmed by immunostaining using an anti-AU5 monoclonal antibody. ( b) Cluster image showing the different classes of gene expression profiles induced by H-Ras V12, RhoA QL, Rac1 QL and Cdc42 QL in NIH 3T3 stable cell line. mRNA from reference cells was used to prepare cDNA labeled with Cy3-dUTP, and mRNA harvested from H-Ras V12, RhoA QL, Rac1QL and Cdc42 QL cell line was used to prepare cDNA labeled with Cy5-dUTP. The two cDNA probes were mixed and simultaneously hybridized to the microarray. Gene expression changes in response to GTPases were clustered hierarchically into groups on the basis of similarity in their expression profiles. All the data are available in Supplemental Table 1. (b) Gene expression profiles in Clusters 1 and 2. Cluster 1 represents genes (82 elements, 56 unique genes) that were upregulated by H-Ras V12 and RhoA QL. Cluster 2 reveals Rac1 QL and Cdc42 upregulated genes (14 elements, 14 unique genes)

Oncogene Identification of the small GTPases responsive gen H Teramoto et al 2692 Table 1 Selected genes up- and downregulated by H-RasV12, RhoA QL, Rac1 QL and Cdc42 QL Genes upregulated Genes downregulated Name Fold change Name Fold change H-Ras V12 High mobility group protein (HGMI-C) 18.1 Pot. MSA-precursor À100 Osteopontin 12.7 Pleiotrophin precursor (PTN) À50 Cyclin-dependent kinase 8 (Cdk8) 11.5 Osteoblast specific factor 2 precursor À50 Ephexin 11.5 Stromal cell-derived factor 1 precursor À20 Connexin 43 9.9 Myristoylated alanine-rich C-kinase À20 Gprotein signaling regulator RGS16 4.3 Angiotensin II receptor À16.7 Phospholipase A2 group VII 3.6 Fibulin-5 precursor (FIBL-5) À12.5 Prostaglandin G/H synthase 1 precursor-1 3.6 ADAM-TS 1 precusor À10 Phosphoglycerate kinase 3.4 Glutathione S-transferase À10 Fibronectin receptor beta subunit precursor 3.2 Thrombospondin 1 precursor À9.1 Adipose differentiation-related protein 3.0 Alpha platelet-derived growth factor À9.1 Akt kinase transforming protein (Akt1) 2.9 Frizzled-1 À9.1 Transferrin receptor 2.9 Procollagen, type I, alpha 2 À8.3 Cathepsin b precursor 2.8 Growth-arrest-specific protein 1 À8.3 JNK-interacting protein-2 (JIP2) 2.7 Collagen 2(VI) chain precursor À7.7 M2-type pyruvate kinase 2.7 Tissue inhibitor of metalloproteinas (TIMP-3) À7.7 MITOGEN-ACTIVATED PROTEIN KINASE 6 (ERK3) 2.7 Sparc precursor À7.1 Epithelial membrane protein-1 2.7 Cadherin-11 À7.1 Annexin A7 2.6 Insulin receptor substrate-1 À5.0 Mdm2 2.6 Neuropilin À5.0

RhoA QL Transferrin receptor 5.5 Angiotensin II receptor À16.7 Cyclin-dependent kinase 6 inhibitor (p18-ink6) 3.5 Pleiotrophin precursor (PTN) À4.8 Guanine nucleotide-binding protein b 3.3 Tesmin-1 À3.7 High mobility group protein (HMGI-C) 3.2 Spectrin beta subunit À3.2 Cyclin A2 3.1 Vinculin À3.2 Apoptosis inhibitor Survivin 3.0 Fascin À2.9 Early growth response protein 1 (Egr-1) 2.8 Radixin À2.9 P8 protein 2.8 Glutathione S-transferase À2.6 Transforming growth factor beta 3 precursor 2.7 Nuclear calmodulin-binding protein À2.4 JNK-interacting protein-2 (JIP2) 2.1 Sprouty homologue 2 (SPRY-2) À2.4

Rac1 QL Neuropilin 5.0 Cyclin A2 À4.1 Early growth response protein 1 (Egr-1) 4.9 Cyclin-dependent kinase 6 inhibitor (p18-ink6) À3.1 Alpha-actinin 4 3.3 Semaphorin 6C À3.1 COLLAGEN ALPHA 1(III) CHAIN PRECURSOR 3.1 Orphan G protein-coupled receptor À2.9 Fibronectin 1 3.0 Ras GTPase-activating protein 3 (Rasa3) À2.9 Collagen a1(V) 2.6 Transcription factor BTEB À2.7 Tenascin precursor 2.6 GTP-binding protein like 1 (Wrch1) À2.6 AM2 receptor 2.6 Fibroblast growth factor receptor b À2.6 Filamin 2.1 Osteopontin À2.6 Thrombospondin 1 precursor 2.0 Growth arrest speciﬁc protein (gas3) À2.6

Cdc42 QL Early growth response protein 1 (Egr-1) 4.4 Transforming growth factor beta 2 precursor À3.8 Neuropilin 3.3 Cyclin A2 À3.3 ETS-related transcription factor ERF 3.1 p8 protein À2.8 Tenascin precursor 2.5 Growth arrest speciﬁc 6 À2.8 IGFBP-4 2.4 Apoptosis inhibitor Survivin À2.8 Alpha-actinin 4 2.4 Glycogenin-1 À2.7 Thrombospondin 1 precursor 2.3 Heme oxygenase 1 À2.6 Ephrin-b2 precursor 2.1 Ras GTPase-activating protein 3 (Rasa3) À2.5 Tubulin 2.1 Rho GDP-dissociation inhibitor 2 À2.5 Protein kinase C delta 2.0 GTP-binding protein like 1 (Wrch1) À2.5

QL (2.8-fold), Rac1 QL (4.9-fold) and Cdc42 QL (4.4- and 2.8-fold, respectively), but not by Rac1 QL and fold), but repressed by H-Ras V12 (À2.9-fold). Glu- Cdc42 QL, which did not affect the regulation of these tathione S-transferase was downregulated only by genes (Tables 1 and 2). H-Ras V12 (À10-fold) and RhoA QL (À2.6-fold). As a large number of genes are regulated differently Furthermore, HMGI-C, a small nonhistone chromoso- by H-Ras V12, RhoA QL, Rac1 QL and Cdc42 QL, mal protein regulating transcription, and p8, an HMG- these observations led us to compare the gene expression I/Y-like protein, were highly stimulated by H-Ras V12 changes provoked by each GTPase based on gene (18.1- and 2.5-fold, respectively) and RhoA QL (3.2- functions (Table 2). From this analysis, a pattern

Oncogene Identiﬁcation of the small GTPases responsive gen H Teramoto et al 2693 Table 2 Summary of genes regulated by H-Ras V12, RhoA QL, Rac1 QL and Cdc42 QL Function Name H-ras V12 RhoA QL Rac1 QL Cdc42 QL Cell adhesion proteins Alpha-actinin 4 À16.7 À1.5 3.29 2.36 Cadherin-11 precursor À7.14 À1.75 1.07 À1.56 Collagen a1(V) À5.26 À1.45 2.63 1.78 Collagen alpha 1 (III) chain precursor À16.7 À1.32 3.07 2.26 Fibronectin 1 À2.04 1.28 2.99 1.65 Fibronectin receptor beta subunit precursor 3.23 2.46 À1.01 À1.82 Integrin alpha-6 precursor (VLA-6) 2.44 1.00 À1.02 1.05 Procollagen type V alpha 2 11.1 À1.61 2.03 1.27 Procollagen, type I, alpha 2 À8.33 À2.33 1.25 1.65 Thrombospondin 1 precursor À9.09 À1.43 1.99 2.32 Type IV collagen alpha 5 chain À4.17 À1.45 1.06 1.16 Vinculin 1.14 À3.23 À1.27 1.35

Cell cycle proteins Cyclin G1 2.95 1.76 1.36 À1.23 Cyclin-dependent kinase 8 11.47 1.57 À1.72 À1.69 G2/mitotic-speciﬁc cyclin a2 1.17 3.14 À4.00 À3.13

GDP/GTP binding proteins Ras GTPase-activating protein 3 (Rasa3) À1.01 1.12 À2.86 À1.85 GTP-binding protein like 1 (Wrch1) À1.96 À1.22 À2.56 À2.44 Rho gdp-dissociation inhibitor 2 2.01 1.47 À2.17 À2.50

Growth factors Alpha platelet-derived growth factor a (PDGF alpha) À9.09 À2.00 À1.06 1.40 Ephrin-b2 precursor À1.61 À1.41 À1.41 2.07 Growth-arrest-speciﬁc protein (gas3) À2.78 À1.82 À2.56 À2.27 Growth-arrest-speciﬁc protein 1 (gas1) À8.33 À1.61 À1.79 À1.43 Insulin-like growth factor binding protein À1.82 2.10 1.95 1.20 Stromal cell-derived protein-1 À3.85 À1.72 À1.01 1.08 Transforming growth factor beta 2 protein (TGF-beta2) À2.33 1.14 À1.61 À3.85 Transforming growth factor beta 3 protein (TGF-beta3) N/a 2.66 1.55 1.10

Membrane receptors Angiotensin II receptor À1.52 À1.52 1.25 1.71 Frizzled-1 À9.09 À2.04 1.32 À1.23 Orphan Gprotein-coupled receptor 2.19 À1.27 À2.94 À1.67

Neurobiology Neuropilin À5.00 À1.35 3.38 2.74 Semaphorin 6C 1.32 2.04 À3.13 À1.96 Sharpin 2.16 1.86 À1.33 À1.69 Slit2 À4.00 À2.50 n/a 1.94

Protein kinases Akt kinase transforming protein (Akt1) 2.88 1.07 À1.35 À1.47 CaM-like protein kinase 2.27 À1.10 À1.30 IPL1- and aurora-related kinase 1 1.83 1.44 À3.13 À2.44 M2-type pyruvate kinase 2.68 1.73 À1.25 À1.39 Mitogen-activated protein kinase 6 (ERK3) 2.68 1.32 À1.34 À1.49 Myristoylated alanine-rich C-kinase À16.67 À5.88 À1.33 1.08 Phosphoglycerate kinase 3.39 2.72 À1.20 À1.39 Protein kinase C delta À1.32 À1.05 1.37 2.02 Raf proto-oncogene serine/threonine-kinase À4.55 À2.38 À1.75 À1.75

Protein phosphatases Protein phosphatase 2A À1.52 À1.49 À1.20 1.74 Signaling intermediates Annexin A7 2.58 À1.59 À1.22 À1.30 Cathepsin b precursor 2.78 1.35 À1.43 À1.56 JNK-interacting protein À2 {JIP2) 2.75 2.10 À1.59 À1.67 Moesin homolog 2.31 1.99 1.07 À1.39 Osteopontin 12.71 1.46 À2.56 À1.45 Radixin À2.86 À2.63 À1.22 1.39

Structural proteins Actin, beta 2.11 À1.10 À1.01 À1.27 Actin, gamma 1 2.12 À1.11 1.21 À1.35 Filamin À1.19 À1.72 2.08 1.68 Gamma adducin À5.00 À2.56 À1.59 1.26 Matrilin 2 À5.56 1.09 1.30 1.55 Spectrin beta subunit À4.00 À3.03 1.23 1.55

Transcription regulators Early growth response protein 1 (egr1) À2.86 2.84 4.77 3.00 ETS-related transcription factor ERF À2.78 À2.27 1.78 3.07 High mobility group protein hmgi-c {HMGI-C) 18.13 3.16 À1.41 À1.39 Nuclear factor I/X À3.85 À1.09 À1.14 À1.33 Transcription factor BTEB 1.23 À1.11 À2.70 À1.59 Mdm2 2.56 1.27 1.08 À1.28 P8 protein 2.46 2.79 1.27 À2.50

Apoptosis regulators Survivin 1.31 3.04 À1.72 À2.78

Oncogene Identiﬁcation of the small GTPases responsive gen H Teramoto et al 2694 5 16 Microarrays Microarrays 14 4 12

3 10

8 2

Fold change 6 Fold change 1 4 2 0 0 AU5 Ras V12 RhoA QL Rac1 QL Cdc42 QL AU5 RasV12 RhoA QL Rac1 QL Cdc42 QL

WB ERK2 WB osteopontin

5 5 Microarrays Microarrays

4 4

3 3

2 2 Fold change Fold change 1 1

0 0 AU5 Ras V12 RhoA QL Rac1 QL Cdc42 QL AU5 Ras V12 RhoA QL Rac1 QL Cdc42 QL

WB Akt 1 WB Egr-1

Figure 3 Relation between microarray results and Western blot analysis. Relative mRNA levels of the indicated genes (ERK2, osteopontin Akt1 and Egr-1) were determined by microarray hybridizations. Indicated genes were also tested for protein expression level by Western blotting of total cell extracts using anti-ERK2, Akt1, osteopontin and Egr-1 antisera

appears to emerge. For example, many genes encoding was elevated only in RhoA QL-, Rac1 QL- and Cdc42 cell adhesion proteins and extracellular matrix compo- QL-expressing cells. nents, such as alpha-actinin 4, collagen A1 and procollagen type I and V, are upregulated by Rac1 QL, and Cdc42 QL, but downregulated or not affected by active Ras and RhoA. Among this gene class, only expression Discussion of integrin beta 1 subunit and fibronectin receptor beta subunit was generally stimulated by H-Ras and RhoA A key role for Ras in cell growth control was predicted but not Rac1 and Cdc42. Conversely, cell cycle genes from the frequent occurrence of activating mutations in such as cyclin G1, cyclin-dependent kinase 8 and cyclin a large variety of human cancers (Bos, 1989). Further- A2 were in general stimulated by H-Ras and RhoA, more, the functional activity of Ras is now known to be but inhibited by Rac and Cdc42. Some protein required for mitogenic signaling by most growth factor kinases signaling molecules, such as Akt 1, or signaling receptors (Porter and Vaillancourt, 1998). In contrast, intermediates, including annexin A7 and osteopontin, the diversity of biological functions performed by small and structural proteins, including actin, beta and GTPases of the Rho family is just beginning to be fully gamma 1, were upregulated mainly by H-Ras V12. appreciated. In particular, the role of Rho-like proteins To examine whether mRNA abundance was reflected in the regulation of the actin-containing cytostructures in protein expression levels, we also examined the and their dynamic changes during cell migration and expression of a number of gene products by using adhesion, as well as their role in cellular growth and commercially available antibodies (Figure 3). ERK2 transformation have been recently characterized in great mRNA expression levels did not change (not shown) detail (Hall, 1998). However, the nature of the nuclear and, indeed, its protein levels were nearly identical in events regulated by these GTP-binding proteins is much vector control (AU5), H-Ras V12, RhoA QL, Rac1 QL less understood. and Cdc42 QL. In contrast, Akt 1 and osteopontin were In this study, we have examined which genes are highly expressed in H-Ras V12, but Egr-1 expression specifically regulated by activated forms of H-Ras,

Oncogene Identification of the small GTPases responsive gen H Teramoto et al 2695 RhoA, Rac1 and Cdc42 by a cDNA microarray analysis Ras is the most frequently mutated oncogene in approach. Using a mouse cDNA chip including 19 117 human cancer (Macara et al., 1996), and the mechanism unique genes, we found that around 7.3% of all genes underlying tumor promotion by Ras may depend on the examined are affected by these GTPases. The different combination of its effects on gene expression and gene expression patterns correlated with the focus- changes in post-translational processing of pre-existing forming ability of these GTPases. Indeed, there were and induced proteins. Among the former, upregulation many genes that were costimulated by H-Ras V12 and of many genes such as those detected here, including RhoA QL but not Rac1 QL and Cdc42 QL, reflecting HMGI-C, osteopontin, connexin 43, Akt, ERK3, and the higher transforming potential of H-Ras and RhoA. the downregulation of growth-inhibiting genes such as The hierarchical gene cluster expression pattern also Gas1 and ERF (Table 1, Figure 3) may contribute to the revealed that genes regulated by RhoA were more transforming activity of H-Ras. Indeed, expression of related to those of Rac1 and Cdc42 than to H-Ras. This HMGI-C by H-Ras was reported earlier (Zentner et al., hierarchy of relatedness of gene expression is also 2001), and truncated HMGI-C can alone induce suggested by their structural similarity, which suggests neoplastic transformation of NIH 3T3 cells (Fedele that their structural features may affect the down- et al., 1998). Osteopontin, a secreted phosphoprotein stream nuclear targets for these small GTPases. On associated with the process of neoplastic transforma- the other hand, as a large number of genes were tion, has been shown recently to be highly expressed in differently regulated by H-Ras V12, RhoA QL, Rac1 lung cancer, and that its overexpression is dependent on QL and Cdc42 QL, we compared these genes based the activated status of the ras oncogene (Zhang et al., on the function of their predicted protein product. 2001). An interesting observation is that only H-Ras Overall, the emerging pattern suggests that Rac1 QL stimulates the expression of Mdm2, a key molecule and Cdc42 QL upregulate many genes encoding mediating the degradation of p53 and thus inhibiting cell adhesion proteins and extracellular matrix compo- p53 tumor suppressor function (Michael and Oren, nents, which are either downregulated or not affected 2002). Downregulation of molecules by H-Ras may be by H-Ras V12 and RhoA QL. Cell cycle genes and also important for tumorigenicity. For example, the growth-promoting transcriptional regulators, in con- transcriptional events promoted by H-Ras, either by trast, were primarily stimulated by Ras and RhoA, and enhancing or decreasing gene expression, may explain in some cases the same genes were inhibited by Rac and the high transforming potential of active H-Ras alleles. Cdc42. It is likely that many of the changes in gene expression Although an extensive analysis of how distinct gene detected by this extensive gene array analysis will have expression patterns correlate with protein expression biological significance, thus warranting further investi- levels is beyond the scope of the present study, we gation to assess their potential contribution to Ras- nonetheless observed a good correlation between induced malignancies. mRNA abundance and protein expression in a few It is also interesting that some of the biological representative cases. For example, ERK2 gene expres- activities elicited by RhoA and resembling H-Ras- sion does not change among transfectants, and its mediated effects may depend on its ability to promote protein product was nearly identical in all cells. gene expression patterns similar to H-Ras but distinct Correlating also with their gene expression pattern, from those induced by Rac1 and Cdc42. Among them, Egr-1 levels were elevated only in RhoA QL-, Rac1 QL- the most dramatic is the high transforming activity of and Cdc42 QL-expressing cells, and Akt 1 and RhoA and its activated exchange factors, including PDZ osteopontin were expressed higher in H-Ras V12- RhoGEF, LARG and p115 (Fukuhara et al., 2001), transfected cells. These results suggest that the result which is in contrast with the poor transforming activity of microarray expression patterns and protein levels of Rac and Cdc42 and GEF promoting the accumula- may correlate well for a number of these proteins. In tion of their active form, such as Tiam (van Leeuwen addition, some of these events may be linked to each et al., 1995). Indeed, there are subsets of genes induced other. For example, it was reported that the expression (Cluster 1) or repressed (Cluster 2) by RhoA that are of Egr-1 can be stimulated by RhoA in NIH 3T3 cells also regulated in the same direction by H-Ras. These (Vara Prasad and Dhanasekaran, 1999), and this same genes are regulated, for the most part, in an transcription factor can transactivate the fibronectin opposite direction by activated forms of Rac1 and gene (Liu et al., 2000), which is induced by Rho Cdc42. For example, cyclin G1 is upregulated by GTPases. Fibronectin is a key component of the activated RhoA and H-Ras but not activated Rac1 extracellular matrix, and plays a crucial role in and Cdc42. Of interest, RhoA also promotes the determining how the microenvironment of the cells expression of the apoptotic inhibitor survivin, while affects a variety of morphogenetic processes, as it Rac1 and Cdc42 diminish its expression. These differ- promotes cell adhesion, migration and signal transduc- ences may now help explain the clearly more potent tion (Miyamoto et al., 1995). On the contrary, expres- growth-promoting activity of Rho. On the other hand, sions of both Egr-1 and fibronectin are down regulated the fact that Rac and Cdc42 selectively stimulate the by H-Ras V12. Loss of this matrix component often expression of extracellular matrix and cell adhesion accompanies oncogenic transformation, thus allowing molecules can be related to their ability to regulate changes in cell growth, morphology and tissue organiza- morphological changes and cell migration. In line with tion (Brenner et al., 2000). this notion, an unexpected observation was that Rac

Oncogene Identification of the small GTPases responsive gen H Teramoto et al 2696 and Cdc42 potently promote the expression of neuro- (Erk 2, 05-157; UBI, NY, USA), osteopontin (sc-10593; Santa pilin, a molecule initially characterized in axon guidance Cruz, CA, USA), Akt-1 (sc-1618; Santa Cruz, CA, USA), and (Chen and Tessier-Lavigne, 1998) but now known to be Egr-1 (sc-110; Santa Cruz, CA, USA) antibodies. Immuno- expressed in multiple cell types and to mediate cell complexes were visualized by enhanced chemiluminescence migration in response to gradients of semaphorins and detection (Amersham Corp., NJ, USA) using goat anti-mouse growth factors, such as VEGF (Soker et al., 1998). or goat anti-rabbit IgGs coupled to horseradish peroxidase as a secondary antibody (Cappel, NC, USA). In summary, we have conducted an extensive analysis of the gene expression patterns induced by activated forms of small GTPases of the Ras and Rho families. cDNA micoarrays, probes and hybridization From this analysis, a GTPase-specific fingerprint of gene Fabrication of the microarray has been described previously expression appears to emerge that correlates with their (Malek et al., 2002). Arrays composed of 26 383 elements diverse biological activities. Among them, similari- (19 117 unique genes) were derived from the Ko/NIH ties between H-Ras and RhoA, and their clear (15 247elements) (Tanaka et al., 2000) and brain molecular differences with Rac1 and Cdc42 may now help explain anatomy project (BMAP) cDNA mouse clone sets (11 136 their distinct ability to promote cell growth and elements) (http://brainest.eng.uiowa.edu/). For probe prepara- tion, total RNA was isolated from three to five independent transformation. mass cultures of confluent stable NIH 3T3 mouse cell lines (AU5, AU5 H-Ras V12, AU5 RhoA QL, AU5 Rac1 QL and AU5 Cdc42 QL) under serum-free conditions (24 h). Total Materials and methods RNA was prepared using the RNeasy Mini Kit as per the manufacturer’s protocol (QIAGEN, CA, USA). Labeled Cell lines and transfection cDNA was synthesized from 15 mg total RNA from the test cell lines (AU5 H-Ras V12, AU5 RhoA QL, AU5 Rac1 QL NIH 3T3 cells were maintained in Dulbecco’s modified Eagle’s and AU5 Cdc42 QL), and from an equal amount of total RNA medium (Life Technologies) supplemented with 10% calf from the reference cell line (AU5). Probe labeling, purification serum, and transfected by the calcium-phosphate precipitation and hybridizations were performed as described previously technique adjusting the total amount of DNA to 2 mg/plate (Yang et al., 2002). with vector alone. For microarray stable cell lines, mass populations of cells expressing the transfected gene were selected for their ability to grow in the presence of Geneticin Image scanning and cluster analysis (G418) (GIBCO BRL, NY, USA). Image scanning, fluorescence intensity measurements, back ground subtraction, data normalization and cluster analysis DNA constructs were performed as described previously (Malek et al., 2002). Plasmids expressing epitope-tagged GTPases, including For cluster analysis, we used TIGR Multi Experiment Viewer / S pCEFL AU5, pCEFL AU5-c-H-Ras V12 and pCEFL AU5- (TMEV), available at http:/www.tigr.org/softlab/ . Before RhoA QL, pCEFL AU5-Rac1 QL, pCEFL AU5-Cdc42 QL, clustering, logarithms (base2) were calculated for the measured were described previously (Teramoto et al., 1996). median fluorescence ratio for each gene. Our experimental noise, as measured by self-to-self hybridization of the AU5 cell line, was determined to be 0.2 (log2). Each array hybridization Western blots was repeated independently at least three independent times, Lysates containing approximately 50 mg of total cellular and differentially regulated genes were stringently set at four protein were analysed by Western blotting after SDS– standard variations from the mean (equivalent to 1.73-fold polyacrylamide gel electrophoresis using anti-MAP Kinase change) (Yang et al., 2002).

References

Bar-Sagi D and Hall A. (2000). Cell, 103, 227–238. Macara IG, Lounsbury KM, Richards SA, McKiernan C and Bos JL. (1989). Cancer Res., 49, 4682–4689. Bar-Sagi D. (1996). FASEB J., 10, 625–630. Brenner KA, Corbett SA and Schwarzbauer JE. (2000). Malek RL, Irby RB, Guo QM, Lee K, Wong S, He M, Tsai J, Oncogene, 19, 3156–3163. Frank B, Quackenbush J, Jove R, Yeatman TJ and Lee NH. Chen H, He Z and Tessier-Lavigne M. (1998). Nat. Neurosci., (2002). Oncogene, 21, 7256–7265. 1, 436–439. Michael D and Oren M. (2002). Curr. Opin. Genet. Dev., 12, Coso OA, Chiariello M, Yu JC, Teramoto H, Crespo P, Xu N, 53–59. Miki T and Gutkind JS. (1995). Cell, 81, 1137–1146. Miyamoto S, Teramoto H, Coso OA, Gutkind JS, Burbelo Fedele M, Berlingieri MT, Scala S, Chiariotti L, Viglietto G, PD, Akiyama SK and Yamada KM. (1995). J. Cell Biol., Rippel V, Bullerdiek J, Santoro M and Fusco A. (1998). 131, 791–805. Oncogene, 17, 413–418. Narumiya S. (1996). J. Biochem. (Tokyo), 120, 215–228. Fukuhara S, Chikumi H and Gutkind JS. (2001). Oncogene, Nobes CD and Hall A. (1995). Cell, 81, 53–62. 20, 1661–1668. Perona R, Esteve P, Jimenez B, Ballestero RP, Ramon Y, Hall A. (1998). Science, 279, 509–514. Cajal S and Lacal JC. (1993). Oncogene, 8, 1285–1292. Hill CS, Wynne J and Treisman R. (1995). Cell, 81, Porter AC and Vaillancourt RR. (1998). Oncogene, 17, 1343– 1159–1170. 1352. Liu C, Yao J, Mercola D and Adamson E. (2000). J. Biol. Soker S, Takashima S, Miao HQ, Neufeld Gand Klagsbrun Chem., 275, 20315–20323. M. (1998). Cell, 92, 735–745.

Oncogene Identiﬁcation of the small GTPases responsive gen H Teramoto et al 2697 Takai Y, Sasaki T and Matozaki T. (2001). Physiol. Rev., 81, Vara Prasad MV and Dhanasekaran N. (1999). Oncogene, 18, 153–208. 1639–1642 Tanaka TS, Jaradat SA, Lim MK, Kargul GJ, Wang X, Yang IV, Chen E, Hasseman JP, Liang W, Frank BC, Wang S, Grahovac MJ, Pantano S, Sano Y, Piao Y, Nagaraja R, Doi Sharov V, Saeed AI, White J, Li J, Lee NH, Yeatman TJ and H, Wood III WH, Becker KGand Ko MS. (2000). Proc. Quackenbush J. (2002). Genome Biol., 3, research 0062. Natl. Acad. Sci. USA, 97, 9127–9132. Zhang J, Takahashi K, Takahashi F, Shimizu K, Ohshita F, Teramoto H, Crespo P, Coso OA, Igishi T, Xu N and Gutkind Kameda Y, Maeda K, Nishio K and Fukuchi Y. (2001). JS. (1996). J. Biol. Chem., 271, 25731–25734. Cancer Lett., 171, 215–222. van Leeuwen FN, van der Kammen RA, Habets GG and Zentner MD, Lin HH, Deng HT, Kim KJ, Shih HM and Ann Collard JG. (1995). Oncogene, 11, 2215–2221. DK. (2001). J. Biol. Chem., 276, 29805–29815.

Oncogene