Unless otherwise indicated, the growth medium for the human foreskin-derived fibroblasts, i.e. MSU-1.1 and PH3MT strains and their derivatives, was Eagle's minimal essential medium supplemented with 0.2 mM L-aspartic acid, 0.2 mM L-serine, 1.0 mM sodium pyruvate, and 10% supplemented calf serum (SCS) (Hyclone Laboratories, Logan, UT). Penicillin, 100 units/ml, and streptomycin, 100 μg/ml were also included. All cell lines were cultured at 37°C in a humidified incubator containing 5% CO₂, 95% ambient air.

To determine the genetic changes required for malignant transformation of human fibroblasts, McCormick, Maher, and their colleagues developed the MSU-1 lineage of human fibroblasts, beginning with finite life span skin fibroblasts derived from the foreskin of a normal neonate [12]. The parental cell strain in that lineage (designated MSU-1.1) is a chromosomally-stable, near-diploid, telomerase-positive, infinite life span cell strain that has been shown to be capable of being transformed into malignant cells by transfection of numerous oncogenes including HRas, NRas and KRas [13-17]. For example, MSU-1.1 cells transfected with an over-expressed oncogenic T24 HRas protein containing an activating V12 mutation (HRas^V12) formed sarcomas when injected subcutaneously into athymic mice [15]. The malignant cell line, designated PH3MT, derived from such a tumor was used in the present study.

The pTet-tTAk and pTet-splice vectors were purchased from Invitrogen (Carlsbad, CA). This Tet-off system allows tetracycline-regulated expression of genes in mammalian cells. To facilitate selection, a pTet-tTAk vector expressing a gene coding for histidinol resistance was constructed and designated pTet-tTak^HisR. A pTet-splice vector containing a gene coding for puromycin resistance was also constructed and designated pTet^PuroR. We derived the Flag-Cdc42^N17 and myc-Rac1^N17 mutant cDNAs by PCR amplification using PFU polymerase (Stratagene, La Jolla, CA), and separately subcloned each into the pTet^PuroR vector. The cDNA templates used for PCR amplification of Cdc42^N17 and Rac1^N17 were kindly provided respectively by Dr. K. Gallo (Michigan State University) and Dr. G. Bokoch (Scripps Institute, La Jolla, CA). We verified the integrity of the resulting pTet-Flag-Cdc42^N17 and pTet-myc-Rac1^N17 vector constructs by automated DNA sequencing (Visible Genetics, Bayer, Toronto, ON, Canada). Lipofectamine (Invitrogen, Carlsbad, CA) was used to transfect the plasmids into human cells per manufacturers protocol.

PH3MT cells were transfected with the pTet-tTak^HisR vector, selected using histidinol (1 mM) (Sigma, St. Louis, MO), and designated PH3MT-tTak-C1. Such cells were subsequently transfected with the empty vector pTet^PuroR, or with the pTet^PuroR vector containing myc-Rac1^N17 cDNA, or with the pTet^PuroR vector containing FLAG-Cdc42^N17 cDNA, and selected for resistance to puromycin (0.5 μg/ml) (Sigma, St. Louis, MO). A puromycin resistant strain resulting from transfection of the empty pTet-tTAk vector was also prepared and was designated PH3MT-VC-C2. Two independent clones that exhibited tetracycline-regulated expression of myc-Rac1^N17 were chosen for study and designated PH3MT-Rac1^N17 (-C1 or -C2). Two independent clones with regulatable FLAG-Cdc42^N17 expression were similarly selected and designated PH3MT-Cdc42^N17 (-C1, and C2). A sixth cell strain, designated PH3MT-Rac1^N17/Cdc42^N17, expressing both myc-Rac1^N17 and FLAG-Cdc42^N17 dominant-negative proteins was also isolated.

To generate MSU-1.1 cell strains expressing GFP-Rac1^V12 or GFP-Cdc42^V12 fusion proteins, the GFP nucleotide sequence from the pCRUZ-GFP vector (Santa Cruz Biotechnology, Santa Cruz, CA) was isolated and ligated into the pcDNA6-V5-HisA vector (Invitrogen, Carlsbad, CA), which confers blasticidin resistance. Rac1^V12 and Cdc42^V12 cDNA template sequences were purchased from University of Missouri-Rolla Research Center (UMR) (Rolla, MO), PCR-amplified, and ligated downstream of the GFP nucleotide sequence to enable the transfectants to express N-terminally-labelled proteins. The resulting constructs were transfected into the parental, non-transformed MSU-1.1 cell strain, and selected for resistance to blasticidin (1 μg/ml) (Invitrogen, Carlsbad, CA) and screened for expression of the GFP-fusion protein by fluorescence microscopy. Identified clones were isolated, expanded, and screened by Western blotting for GFP-tagged Rac1^V12 or Cdc42^V12 protein expression.

Whole cell protein extracts were prepared using a lysis buffer consisting of 50 mM Tris-HCl, pH 7.2, 150 mM NaCl, 50 mM NaF, 0.5% NP-40, 1 mM Na₃V0₄, 200 mM benzamidine, 1 mM PMSF, 25 μg/ml aprotinin, and 25 μg/ml leupeptin. Total protein concentration was quantified using the Coomassie protein assay reagent (Pierce Biotechnology, Rockford, IL). Lysates were denatured in 5× Laemelli sample buffer, separated by either 10% or 12% SDS-PAGE, and transferred to PVDF membrane. The membrane was blocked for 2 hr with Tris-buffered saline containing 0.1% Tween-20 (TBST) and 5% (w/v) non-fat milk. For the majority of the studies, the membrane was probed overnight with the primary antibody at 4°C, then probed for 1 hr at room temperature with the appropriate horseradish peroxidase-linked secondary antibody (Sigma and Santa Cruz Biotechnology). Both antibodies were diluted in TBST containing 5% milk. The membrane was incubated with the Supersignal West Pico chemiluminescent horseradish peroxidase substrate (Pierce Biotechnology, Rockford, IL) and then exposed to film. The primary antibodies used were anti-FLAG, 1:1000 dilution (Sigma, St. Louis, MO), anti-myc 9E10, 1:500 dilution, and anti-GFP, 1:1000 dilution (Santa Cruz Biotechnology, Santa Cruz, CA).

The focus reconstruction assay was performed essentially as described [15]. Briefly, MSU-1.1 cells were plated at a density of 5 × 10⁴ cells per 100-mm diameter dish containing growth medium supplemented with 0.5% SCS and 20 mM HEPES (pH 7.4) in the presence or absence of tetracycline. On the following day, the cells to be assayed for focus reconstruction were plated (100-200 cells per dish) in these same dishes. Cells received fresh medium every fourth day. After 3 weeks of growth, the cells were fixed with neutral buffered formalin, stained with methylene blue, and then examined for quality and quantity of foci.

Cells were plated in growth medium containing 10% SCS into a series of 60-mm diameter dishes at a density of 10⁴ cells per dish and incubated for 48 hr. The cells were then washed twice and growth medium containing 0.5% SCS was added. The number of cells in three replicate dishes was counted at the indicated intervals using a Coulter counter. An equation derived from a best-fit exponential curve was used to determine the doubling time of cells in log-phase growth. This experiment was repeated three times.

Cells were suspended in 0.33% top agarose, then plated into 60-mm-diameter culture dishes at a 5 × 10³ cells per dish, and then covered with 2 mL of growth medium. The growth medium was replaced weekly. After three weeks, the cells were fixed with 2.5% glutaraldehyde. This experiment was repeated three times.

To test PH3MT derived cell strains, 10⁶ cells were injected subcutaneously into the right and left flank of a series of athymic Balb/c mice. To turn off the dominant-negative form of the proteins, tetracycline was added to the drinking water of half of the mice, at a concentration of 1 mg/ml. To mask the flavor of tetracycline, 5% sucrose was added to the drinking water of all mice. Tumor measurements were made weekly, and mice were sacrificed when a tumor with a volume of 0.5 cm³ developed on either flank. The tumorigenicity results were displayed in the form of Kaplan-Meier survival plots using MedCalc Version 8.2 http://www.medcalc.be. A log-rank test was used to determine statistical significance.

To determine whether the MSU-1.1 derivative cell strains expressing GFP-Rac^V12 or GFP-Cdc42^V12 were capable of forming tumors in athymic mice, a 1 cm³ absorbable gelatin sponge (Gelfoam size 50, Pharmacia) was placed in each flank. One week later, 10⁷ cells were injected directly into the sponges at each injection site. Such sponges were used to ensure that a high number of injected cells were retained at each injection site.

PolyA+RNA was extracted using the Micropure PolyA+RNA extraction kit according to the manufacturer's instructions (Ambion, Applied Biosystems, Austin, TX). To synthesize cRNA products from 3 μg of polyA+RNA, the procedures described in the Affymetrix expression analysis technical manual were followed. Personnel at the Genomics Technology Sequence Facility at Michigan State University carried out the hybridization to the Affymetrix HU95A human genome chip and probed, washed, and scanned the arrays as described in the Affymetrix manual. Such analysis was carried out in duplicate.

Total mRNA expression was calculated for each gene represented on the Affymetrix chip. Expression differences were calculated within each experiment as well as between experiments. This approach resulted in four separate data sets, i.e. both data sets collected from cells grown in the presence of tetracycline were compared to both data sets collected from cells grown in the absence of tetracycline. To ascertain differentially expressed genes, one-way parametric analysis of variance (ANOVA) tests were performed by using a Benjamini-Hochberg false discovery rate (BH FDR) for multiple testing correction, with a P value of 0.05 using MAS5.0 software. The significant changes in each of the four comparisons, as determined by the MAS5.0 software, were identified using Microsoft Access query tool.

To detect levels of secreted uPA or VEGF protein from PH3MT derived cells, cells were plated in 100 mm-diameter culture dishes in growth medium containing tetracycline (1 mg/ml). After 24 hr, the medium was removed, the cells were washed twice, and tetracycline-free medium was added. Cells were incubated for another 24 hr to allow for expression of dominant-negative genes. The cells were then washed twice, and given medium lacking serum. After 24 hr the cells were stimulated with growth medium, cobalt chloride (CoCl₂) (100 μM), deferoxamine (DFO) (100 μM) or were incubated in a hypoxic chamber (1% O₂) for 24 hr. To determine the level of secreted VEGF and uPA protein from MSU-1.1 derivative cell strains, 1.5 × 10⁵ cells were plated in a 100 mm-diameter tissue culture dish containing growth medium, and incubated for 24 hr. The cells were then washed twice with serum-free medium, and serum-starved for another 24 hr before being stimulated with growth medium containing 10% SCS. The media were collected, and centrifuged at 5000 × g for 5 min. To detect secreted VEGF protein, we carried out a sandwich ELISA using the human VEGF DuoSet (R&D Systems, Minneapolis, MN) ELISA kit following the manufacturer's protocol. To detect secreted levels of uPA protein, a similar sandwich ELISA was carried out using the uPA ELISA kit (Oncogene Science, Bayer, Cambridge, MA) according to the manufacturer's protocol.

For all ELISA analyses, after the removal of conditioned medium, whole cell protein extracts were made, quantified, and used to normalize the levels of uPA and VEGF protein detected by dividing the concentration of uPA or VEGF protein in the conditioned medium by the total amount of protein in the whole cell extract, expressed as percent of vector control, or as fold-change. Such experiments were repeated three times. Error bars indicate the standard deviation from the mean. To determine statistical significance, the student's two-tailed t-test was used.

To investigate whether the activity of Rac1, Cdc42, or both proteins, is required for HRas^V12-induced malignant transformation of human fibroblasts, we transfected PH3MT cells, a cell line that had been malignantly-transformed by transfection of HRas^V12, with plasmids encoding dominant negative forms of Rac1 (Rac1^N17) and/or Cdc42 (Cdc42^N17). It is possible that over-expression of these dominant negative proteins may alter other signalling pathways by sequestering exchange factors that are utilized by other members of the Rho GTPase family in addition to Rac1 and/or Cdc42. However, our use of these mutants allows for a direct comparison of studies completed in human fibroblasts presented here to those studies completed in rodent fibroblasts that used identical dominant-negative or constitutively-active mutants. In these cells, expression of dominant negative mutants is controlled in a Tet-off system, i.e., absence of tetracycline results in dominant negative protein expression. Regulated expression of dominant negative mutant proteins was observed in two independent clones of PH3MT cells transfected with plasmids containing either Rac1^N17 or Cdc42^N17 (Fig. 1a). Regulated expression of both Rac1^N17 and Cdc42^N17 was observed in one clone (Rac1^N17/Cdc42^N17). As expected, dominant negative mutant expression was not observed in parental cell strains PH3MT-tTak-C1 or a vector control transfected cell strain (PH3MT-VC-C2).

To evaluate the effect of dominant negative protein expression on the transformed phenotype of PH3MT cells, commonly used transformation assays including: growth curve analysis in reduced serum, focus reconstruction, and tumor forming ability, were carried out. Growth curve analysis revealed a significant increase in the doubling time in all dominant negative protein expressing clones cultured in medium lacking tetracycline (dominant negative protein expression) relative to their growth in medium containing tetracycline (dominant negative protein expression suppressed) (p < 0.05) (Table 1). A moderate increase in the doubling time of the vector control in the absence of tetracycline was also observed indicating some toxicity of tTak expression (p < 0.05). However, the extent to which tTak expression alone affected doubling time was significantly less than what was observed in all cell strains expressing dominant negative proteins (p < 0.01). Focus reconstruction assays also revealed a significant decrease in the ability of dominant negative protein expressing clones to form foci on a lawn of non-transformed human fibroblasts (MSU-1.1 cells) (Fig. 1b). In general, expression of Rac1^N17 affected both growth in reduced serum, and focus reconstruction properties to a greater extent than did expression of Cdc42^N17. These results suggest that dominant negative protein expression does indeed affect the transformed properties of PH3MT cells, and suggest that the activities of both Rac1 and Cdc42 play a significant role in the ability of HRas^V12-transformed cells to grow in reduced serum conditions and to avoid contact inhibition.

To determine whether the activities of Rac1 and/or Cdc42 proteins are required for maintaining the ability of PH3MT cells to form tumors in athymic mice, cells expressing Rac1^N17, Cdc42^N17, or both mutant proteins, were injected subcutaneously into athymic mice and tumor growth was monitored. As a control, another set of mice was injected with the parental cell strain, PH3MT-tTAk-C1, or with vector control cell strain PH3MT-VC-C2. By 15 weeks, all mice that had been injected with the latter two cell strains, in the presence or absence of tetracycline, developed tumors and were sacrificed (Fig. 2a &2b). In contrast, dominant-negative interference with Rac1 activity resulted in decreased tumor-forming ability, and mice injected with these cells showed a significant prolongation of a tumor-free lifespan (p < 0.01) (Fig. 2c &2d). However, not all mice escaped tumor formation. We hypothesized that following injection, cells with levels of dominant negative protein expression sufficient to prevent tumor growth were selected against, while cells with low levels of mutant protein expression were able to proliferate and resulted in tumor formation. To examine this possibility, we derived cell lines from resected tumors, selected for injected cells using the puromycin resistance marker, and compared the level of Rac1^N17 expression relative to the level expressed in the cell strain prior to injection. We were unable to detect Rac1^N17 in cell strains derived from these tumors (Fig. 2e &2f). These results strongly suggest that Rac1 activity plays a critical role in the ability of the transformed fibroblasts to form sarcomas, and that when the dominant-negative protein expression is suppressed, these cells re-acquire the ability to form such tumors. Although there was no significant difference in the length of survival of mice injected with cells expressing Cdc42^N17, subsequent Western blotting of tumor-derived cell strains revealed similar results, i.e., these cell strains had lost detectable levels of expression of dominant-negative proteins (Fig. 3a-d).

Subcutaneous injection of PH3MT cells expressing both Rac1^N17 and Cdc42^N17 proteins resulted in significantly prolonged tumor-free survival (p < 0.0001) (Fig. 3e). Again, not all mice escaped tumor formation. Expression analysis of two tumor-derived cell lines revealed undetectable levels of dominant-negative protein expression (Fig. 3f). These data suggest that the activity of both Rac1 and Cdc42 is essential for HRas^V12-induced transformation of human fibroblasts.

To further address the roles of Rac1 and Cdc42 in the development of a transformed phenotype, we also evaluated the ability of cells expressing activated mutants to display transformed properties. Based on studies in rodent fibroblasts, we hypothesized that expression of Rac1^V12 or Cdc42^V12 in human fibroblasts would elicit a transformed phenotype [8,9]. To test this hypothesis, we stably transfected the parental MSU-1.1 cell strain with vectors encoding GFP-Rac1^V12 or GFP-Cdc42^V12 proteins (Fig. 4a), and assayed them for the ability to grow in medium with reduced serum, form large colonies in agarose, and develop into sarcomas in athymic mice. Expression of Rac1^V12, but not Cdc42^V12 resulted in an increased ability of MSU-1.1 fibroblasts to grow in medium with reduced serum (Fig. 4b). Logarithmic extrapolation revealed a doubling time of 19 hr in HRas^V12 expressing PH3MT cells and 20 hr in MSU-1.1 cells expressing Rac1^V12. This is in contrast to parental, vector-control, and Cdc42^V12 expressing cells which displayed doubling times of >30 hr.

The ability of cells to grow in reduced serum conditions typically correlates with the potential to form anchorage independent colonies in soft agar. To explore this possibility, these same cell strains were suspended in agarose and monitored for colony formation (Fig. 4c). We found that expression of Cdc42^V12 confers the ability for these cells to form large anchorage independent colonies, whereas expression of Rac1^V12 resulted in inconsistent small colony formation. Together, these data support the results of previous studies in mouse and rat fibroblasts indicating a role for Rac1 in proliferation in reduced serum conditions and Cdc42 in anchorage independent growth [8].

To determine whether Rac1^V12 or Cdc42^V12 expression in human fibroblasts results in malignant transformation, we subcutaneously injected these mutant expressing cell strains into the flanks of athymic mice. Surprisingly, neither Rac1^V12 nor Cdc42^V12 expression resulted in the ability for these cells to form tumors 28 weeks post injection (data not shown).

Because the activities of both Rac1 and Cdc42 play essential roles in mediating HRas^V12-induced transformation, we hypothesized that downstream effectors of these two G-proteins play similarly indispensable roles. To identify these genes, mRNA was harvested from the PH3MT-Rac1^N17/Cdc42^N17 cell strain grown either in the presence, or absence of tetracycline. Therefore, in this analysis, the same cell strain was used as both the control, i.e. normal Rac1 and Cdc42 signalling, and the experimental, i.e. inhibited Rac1 and Cdc42 signalling, groups. Using Affymetrix GeneChip technology, a total of 29 significant expression differences were identified (Table 2). The relatively small number and magnitude of significant gene changes identified in this analysis is likely attributable to the unique ability to compare transcriptome differences in the same cell line, thereby removing background "noise" typically observed in array analyses comparing transcriptome profiles in independent cell clones. Our criteria for choosing genes for follow-up experiments included those genes that: 1) were known to be affected by HRas mutations, and 2) had a known role in cancer. Two such genes, uPA (urokinase plasminogen activator) and VEGF (vascular endothelial growth factor) were of particular interest since previous work in our lab, and others, has shown that HRas transformed cells including PH3MT, as well as N-Ras and K-Ras transformed cells derived from the parental MSU-1.1 cell strain, secrete significantly elevated levels of both uPA and VEGF proteins compared to their non-transformed parental cell strain ([18,19] and data not shown). In the present study, our array analyses detected >1.5 fold decrease in both VEGF and uPA gene expression when both Rac1 and Cdc42 were inhibited, suggesting that Rac1 and/or Cdc42 play significant roles in uPA and VEGF expression downstream of oncogenic HRas.

To determine whether the difference in uPA mRNA levels corresponded to alterations in secreted levels of uPA protein, ELISA analyses were conducted using tissue culture supernatants (Fig. 5a). Inhibition of either Rac1 or Cdc42 in PH3MT cells resulted in a 60% and 70% reduction in secreted uPA protein levels, respectively. Interestingly, inhibition of both proteins resulted in an additive reduction. Although our transcriptome array analyses did not indicate whether or both Rac1 and/or Cdc42 affected uPA gene expression, this analysis indicates that the activation of both proteins is required for the full elaboration of increased uPA secretion.

In a parallel experiment, we also determined whether expression of Rac1^V12 or Cdc42^V12 could increase the levels of secreted uPA protein in the parental, non-transformed MSU-1.1 cell strain (Fig. 5b). Interestingly, expression of neither Cdc42^V12 nor Rac1^V12 resulted in increased levels of secreted uPA protein, indicating that although their activities are required to mediate the secretion of uPA in Ras^V12-transformed PH3MT cells, their activation alone is not sufficient to induce similar increases in expression. In fact, activation of Rac1 resulted in a small, but reproducible decrease in levels of secreted uPA protein.

To determine if Rac1 and/or Cdc42 regulate HRas^V12-induced VEGF secretion, we measured the amount of VEGF protein secreted from PH3MT cell strains stably expressing Rac1^N17 and/or Cdc42^N17 dominant-negative mutants. Inhibition of Rac1 alone, or both Rac1 and Cdc42, completely abrogated HRas^V12-induced secreted VEGF levels (Fig. 6a). However, inhibition of Cdc42 alone resulted in only a 40% reduction. These data indicate that in addition to Cdc42-mediated pathways, other signalling networks, dependent on Rac1 activity, play significant roles in VEGF secretion in the context of HRas^V12 activation.

Current studies indicate that both Rac1 and Cdc42 regulate hypoxia inducible factor (HIF)-induced VEGF expression in response to hypoxia in human hepatocellular carcinoma and gastric cancer cell lines [20,21]. For this reason, we hypothesized that both Rac1 and Cdc42 regulate secreted levels of VEGF protein from HRas^V12-transformed fibroblasts grown in hypoxic conditions. To address this hypothesis, we cultured PH3MT derivative cell strains expressing Rac1^N17 and/or Cdc42^N17 under hypoxic (1% O₂) conditions. Furthermore, we carried out parallel studies using two hypoxia mimetics, cobalt chloride (CoCl₂) or deferoxamine (DFO). Under each condition, inhibition of either protein resulted in a 50% - 60% reduction in the level of secreted VEGF protein, whereas when these cells were exposed to either CoCl₂ or DFO, inhibition of both Rac1 and Cdc42 completely eliminated detectable levels of VEGF protein (Fig. 6a). However, there was not an additive decrease in VEGF secretion in cells exposed to hypoxia. Therefore, these data indicate that both Rac1 and Cdc42 pathways are required for VEGF expression in HRas^V12-transformed human fibroblasts exposed to hypoxic conditions.

Because the activation of Rac1 and Cdc42 is required for the full elaboration of VEGF expression downstream of oncogenic HRas, we hypothesized that introduction of Rac1^V12 or Cdc42^V12 protein in MSU-1.1 cells results in an induction of VEGF protein secretion (Fig. 6b). However, we found that expression of Rac1^V12 did not induce VEGF expression in human fibroblasts. In contrast, expression of Cdc42^V12 induced a significant increase in expression of VEGF (6-fold; p < 0.05). This suggests that in human fibroblasts, Rac1 regulates the levels of secreted VEGF in the context of HRas^V12-induced mitogenic signalling, whereas the activation of Cdc42 is sufficient to increase levels of secreted VEGF, independent of other oncogenic HRas-mediated effector pathways.