Germ Line Gain of Function with SOS1 Mutation in Hereditary Gingival Fibromatosis*

Mutation of human SOS1 is responsible for hereditary gingival fibromatosis type 1, a benign overgrowth condition of the gingiva. Here, we investigated molecular mechanisms responsible for the increased rate of cell proliferation in gingival fibroblasts caused by mutant SOS1 in vitro. Using ectopic expression of wild-type and mutant SOS1 constructs, we found that truncated SOS1 could localize to the plasma membrane, without growth factor stimuli, leading to sustained activation of Ras/MAPK signaling. Additionally, we observed an increase in the magnitude and duration of ERK signaling in hereditary gingival fibromatosis gingival fibroblasts that was associated with phosphorylation of retinoblastoma tumor suppressor protein and the up-regulation of cell cycle regulators, including cyclins C, D, and E and the E2F/DP transcription factors. These factors promote cell cycle progression from G1 to S phase, and their up-regulation may underlie the increased gingival fibroblast proliferation observed. Selective depletion of wild-type and mutant SOS1 through small interfering RNA demonstrates the link between mutation of SOS1, ERK signaling, cell proliferation rate, and the expression levels of Egr-1 and proliferating cell nuclear antigen. These findings elucidate the mechanisms for gingival overgrowth mediated by SOS1 gene mutation in humans.

Hereditary gingival fibromatosis (HGF) 2 is a genetic condition characterized by a slowly progressive, benign fibrous enlargement of keratinized gingiva (1)(2)(3). Genetic studies demonstrate locus heterogeneity for HGF, but etiologic mutations have only been identified in the Son of Sevenless-1 gene (SOS1) (4). An SOS1 g.126,142-126,143insC insertion mutation causes a frameshift and early termination of the protein, yielding a chimeric 1,105-amino acid protein that consists of 1,083 SOS1 N-terminal amino acids followed by 22 novel amino acids. This truncation abolishes four C-terminal proline-rich motifs, which are required for Grb2 binding. The form of HGF due to SOS1 mutation is designated HGF1. SOS1 functions as a guanine nucleotide exchange factor (GEF) that couples receptor tyrosine kinases to the Ras signaling pathway and controls cell proliferation, differentiation, vesicle trafficking, and regulation of the actin cytoskeleton (5). Under the control of two classes of regulatory proteins, GEF and GTPase-activating proteins, Ras functions as a molecular switch between GDP/GTP cycling (6). Three Ras-GEFs; SOS, guanine nucleotide-releasing factor, and guanyl nucleotide-releasing protein, have been characterized in controlling Ras activation by catalyzing GDP release and association with GTP (7)(8)(9)(10)(11). Two regulatory regions govern the GEF activity of SOS1: a catalytic site that interacts with nucleotide-free Ras and an allosteric site that enhances exchange activity with the binding of nucleotide-Ras (12,13). Upon activation of receptor tyrosine kinase, Grb2-SOS1 complexes are recruited to the plasma membrane leading to the exchange of GDP for GTP and Ras activation (14). Although the Grb2-SOS1 complex functions exclusively as an Ras activator, SOS1 can also function as a GEF that is specific to the GTPase Rac1. These two distinct catalytic functions of SOS1 are mutually exclusive and reciprocally related (7).
The activation of Ras signal stimulates downstream signaling pathways, including the mitogen-activated protein kinase (MAPK) family, a ubiquitous signal transduction pathway (15,16). Differences in the duration, intensity, spatial distribution, and temporal qualities of ERK signaling govern distinct biological responses (17)(18)(19). For example, transient induction of ERK signaling by EGF leads to proliferation of PC12 cells, whereas sustained signaling induces differentiation (20 -25). In response to extracellular stimuli, MAPK signaling exerts control of cell cycle progression. Translocation of activated ERK from the cytoplasm to the nucleus is necessary for activation and stabilization of Elk1, c-Jun, c-Myc, and c-Fos. These transcription factors then regulate the expression of genes, such as cyclin D1 and p21 WAF1/CIP1 , which are critical for the progression from G 1 to S phase (26,27).
Increased cell proliferation has been reported for HGF1 gingival fibroblasts in both monolayer and three-dimensional matrix cultures (28). Here we report that mutant SOS1 contributes an increased and sustained activation of ERK signaling in HGF1 fibroblasts under serum-starved conditions. Sustained ERK signaling leads to increased expression of cell cycle regulators and transcription factors. RNA interference-mediated SOS1 depletion was used to confirm the association between SOS1 mutation, ERK activation, and gingival fibroblast proliferation. These findings document a gain-of-function SOS1 mutation and provide a molecular mechanism for gingival overgrowth in HGF1.

EXPERIMENTAL PROCEDURES
Isolation of Fibroblasts and Cell Cultures-Human gingival tissues were obtained with informed consent from three normal subjects (control) and three patients (HGF1). All patients were heterozygous carriers of the SOS1 g.126,142-126,143insC mutation. Three sets of gingival fibroblasts from HGF1 patients and age-matched normal controls (ages 20 -24 years) were isolated and grown in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum and 1ϫ antibiotic-antimycotic solution (growth medium) at 37°C in a 5% CO 2 -humidified incubator and maintained up to 10 passages (28). HeLa cells were obtained from ATCC (Manassas, VA).
Plasmid Construction, Transfection, and Subcellular Fractionation-The full-length human SOS1 expression plasmid (pCGN-HASOS1) was a gift from D. Bar-Sagi (State University of New York, Stony Brook, NY). The vector control plasmid was generated by HindIII digestion to release the HA-SOS1 insert and self-ligated. The QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA) was used with oligonucleotide (5Ј-AGCATCTGCACCAAATTCTTCCcAAGAACACCGTTA-ACACCTCC-3Ј, GenBank TM accession number L13857) to generate the mutated pCGN-HASOS1 construct that carried the HGF1 mutation (small case and cytosine insertion). The mutation was confirmed by DNA sequencing. For transient transfection, 15 ϫ 10 4 cells/well were seeded in 6-well plates a day before experiments. Expression constructs (2 g) were transfected into primary gingival fibroblasts using JetPEI reagent (ISC Bioexperss, Kaysville, UT) or into HeLa cells by Lipofectamine 2000 (Invitrogen). Transfection was terminated 48 -72 h post-transfection, and total cellular lysates were obtained (29). To study the subcellular distribution of SOS1, 100 ϫ 10 4 HeLa cells were plated in 10-cm dishes 24 h before transfection. Eight micrograms of indicated expression constructs were used, and transfected cells were harvested 48 h post-transfection. Cellular fractions were isolated using the ProteoExtract Subcellular Proteome Extraction kit (EMD Biosciences, La Jolla, CA).
Cell Proliferation and Ras Activation Assays-For proliferation assays, 1 ϫ 10 4 cells/well of primary gingival fibroblasts were seeded in 48-well plates. After 24 h the growth medium was replaced with Dulbecco's modified Eagle's medium containing 20 mM HEPES (starving medium) for overnight. The medium was then replaced with growth medium and maintained for 9 days. Cells from triplicate wells were trypsinized at each time point, and the total cell number was determined by using a Coulter Counter (Beckman Z series). To monitor Ras activation, cells were seeded in 150-mm culture dishes, grown to 85% confluence, and serum-starved for 16 h. After treatment with EGF (25 ng/ml) for 0.5-6.0 h at 37°C, cellular extracts were collected in lysis buffer (25 mM Tris-HCl, pH 7.5, 150 mM NaCl, 10 mM MgCl 2 , 1 mM EDTA, 1% Nonidet P-40, 0.5% sodium deoxycholate, 10% glycerol, 2.0 mM phenylmethanesulfonyl fluoride, 20 M Na 3 VO 4 , 10 M NaF, 1 g/ml pepstatin A, and 10 g/ml aprotinin plus protease inhibitor tablet (Complete tablet, Roche Applied Science). Ras activity was measured using a Ras Activation kit (Upstate, Charlottesville, VA). For each reaction, 600 g of whole cell lysate was incubated with Raf-1 Ras binding domain-agarose (15 g) for 1 h at 4°C. The complexes were collected by centrifugation and washed five times with buffer (25 mM Tris-HCl, pH 7.5, 150 mM NaCl, 10 mM MgCl 2 , 1 mM EDTA, 10% glycerol, 20 M NaF, and 1% Nonidet P-40). Protein complexes were released with SDS sample buffer, separated by 4 -12% NuPAGE, and transferred to a polyvinylidene difluoride membrane. Proteins were detected by mouse anti-Ras antibody (0.05 g/ml) and goat anti-mouse horseradish peroxidase-conjugated secondary antibody (Bio-Rad, 1:5000).
Results were adjusted for loading controls. Immunofluorescent staining was conducted as previously described (30). After transfection, gingival fibroblasts were maintained in either growth or starving medium for 16 h and fixed with paraformaldehyde (3.7%) for 15 min at room temperature. Cells were incubated with anti-HA (1:1000) for 1 h at room temperature. After washing, the cells were incubated with goat anti-mouse conjugated with TRITC IgG antibody for 30 min. After wash-ing, the cell nuclei were stained with 4Ј,6-diamidino-2-phenylindole, dihydrochloride (Sigma) for 5 min at room temperature and washed three additional times. Slides were examined with fluorescence microcopy (Olympus, IX71), and images were processed using Adobe Photoshop CS.
Real-time PCR-Total RNA from equivalent cell densities of control and HGF1 fibroblast cultures maintained either in starving medium for 16 h or switched to growth medium for 1 day were prepared by TRIzol solution (Invitrogen) and the "RNeasy Mini kit" (Qiagen). RNA quality and quantity were determined using a Bioanalyzer (Agilent 2100, Wilmington, DE). Reverse transcription-PCR experiments were conducted as described previously (32) with modification. TaqMan probes and PCR primers were purchased from ABI Biosystems (Foster City, CA). These included SOS1 (Hs00362308_m1), E2F1 (Hs00153451_m1), E2F2 (Hs00 231667_m1), transcription factor DP1 (TFDP1, Hs00955488_m1), transcription factor DP2 (TFDP2, Hs00232366_m1), cyclin E1 (CCNE1, Hs00233356_m1), and cyclin E2 (CCNE2, Hs00180319_m1). The RNase P or TATA-box-binding protein (TBP) was used as endogenous control for normalization. QPCR Human Reference Total RNA (Stratagene) was used as a calibrator in all quantitative reverse transcription-PCR experiments. Relative levels of the indicated transcripts in each sample were calculated as 2 Ϫ⌬⌬Ct , where ⌬⌬C t ϭ [target gene's C t Ϫ TBP C t ] sample Ϫ [target gene's C t Ϫ TBP Ct] calibrator . Experiments were performed at least three times with triplicate samples.
RNA Interference Knock-down and Focus Oligoarray Analysis-Duplex RNA of SOS1 target sequences flanking the cytosine insertion were synthesized with the Silencer siRNA Construc-tion Kit (Ambion). Transient transfection experiments were conducted as previously described (30). Briefly, 18 ϫ 10 4 cells/well of gingival fibroblasts were plated in 6-well plates 24 h before transfection. Double-stranded siRNA alone or together with SOS1 expression constructs (1 g/35-mm) was introduced into HeLa cells by Lipofectamine 2000 (Invitrogen). The siRNA for Luciferase (5Ј-CTTACG-CTGAGTACTTCGA) (31) was used as a control. Total cellular lysates were collected after 48-h transfection and subjected to Western blot analyses. To monitor the effects of knock-down on proliferation, fibroblasts were seeded at 8000/well into 48-well plates. Cultures were transfected with indicated siRNA (50 nM) overnight and replaced with fresh growth medium (day 0). At each time point, the total number of cells was determined as described above. Experiments were performed at least three times with triplicate samples. To monitor gene expression in response to EGF treatment, normal (control), and HGF1 fibroblast cultures were serumstarved for 16 h and treated with EGF (50 ng/ml) for 24 h. Total RNA from EGF-treated and untreated cultures was prepared as described above. The complementary RNA probe was synthesized (TrueLabeling-Am kit, SuperArray Bioscience, Frederick, MD) together with biotin-16-UTP labeling. The complementary RNA probes were incubated with cell cycle microarray membranes, and the expression levels of each gene were detected with chemiluminescence using a copalyl diphosphatestar substrate (SuperArray Bioscience). Membranes were exposed to x-ray film (Kodak, XAR film) and signal analyzed by the GEArray Expression Analysis Suite (SuperArray Bioscience) with subtraction of background and normalization with housekeeping genes of glyceraldehyde-3-phosphate dehydrogenase and ␤-actin.
Statistical Analyses-For comparisons of proliferation assays, real-time PCR experiments and quantification of phospho-ERK levels, the Student's t test was used to compare the mean values between HGF1 and controls. Values of p Ͻ 0.05 were considered significant.

Expression of Mutant SOS1 in Gingival Fibroblasts-
The insertion mutation of SOS1 in HGF1 results in an early termination as illustrated in Fig. 1A. To verify the presence of truncated SOS1 in HGF1 fibroblasts, antibody with the epitope against the N-terminal region of SOS1 (SOS1(N)), detected one band (ϳ170 kDa) in controls and two bands (170 and 130 kDa) in HGF1 patients (Fig. 1B, upper panel). The 170-kDa band The asterisk indicates the mutation site found in HGF1 patients with insertion of cytosine (1083) followed by 22 residues (1083-1105) with frameshift region before early termination. B, total cellular extracts (50 g) of cultured fibroblasts from three control and three HGF1 individuals were subjected to SDS-PAGE and blotted with SOS1(N) and ␤-actin antibody (upper panel). Control and HGF1 fibroblasts were transfected with indicated SOS1 constructs and probed with SOS1(N) and ␤-actin antibody (lower panel). The full-length (a) and truncated (b) SOS1 bands are indicated. C, total RNA from fibroblasts of three controls and four HGF1 patients were subjected to reverse transcription-PCR analysis. After normalization against RNase P, the relative levels of SOS1 (solid bar) and TBP (striped bar) transcripts are presented as mean Ϯ S.D. from three separate experiments with triplicate samples. The asterisk indicates significant difference between HGF1 and control (p Ͻ 0.05).
corresponds to the full-length SOS1 and the smaller ϳ130-kDa band approximates the calculated molecular weight of the truncated mutant SOS1. To confirm this, expression constructs of wild-type and mutant SOS1 were transfected into primary gingival fibroblasts. As shown in Fig.  1B (lower panel), the expressed wild type (arrow a, lanes 2 and 5) and mutant (arrow b, lanes 3 and 6) bands were detected at the same positions as the endogenous fulllength (lanes 1 and 3) and truncated SOS1 (lanes 4 and 5). Western blotting results revealed the intensity of the mutant (lower) band was 40% less than the full-length (upper) band in HGF1 samples (Fig. 1B). The lower amount of mutant SOS1 could result from the instability of either its transcript or its protein product. Real-time PCR experiments were conducted to monitor the total levels, of both wild-type and mutant SOS1 transcript in HGF1 fibroblasts. Overall levels of SOS1 transcript were 40% lower in HGF1 fibroblasts than in control fibroblasts (Fig. 1C). The TBP transcript served as an internal control. The lower levels of mutant SOS1 protein in HGF1 fibroblasts likely reflect a less stable mutant SOS1 transcript. Because all three sets of controls and HGF1 patients showed similar results, only results from one set of control and HGF1 fibroblasts are presented.
Sustained Activation of MAPK in HGF1 Gingival Fibroblasts-Because SOS1 plays a critical role in the Ras/MAPK/ERK signaling pathway (22), the effect of the SOS1 mutation on Ras signaling was evaluated. Although Ras activity was low in serum-starved control fibroblasts, it increased rapidly after EGF treatment and subsequently deceased, approaching basal levels by 30 min (Fig. 2A). In contrast, Ras activity was 5-fold higher in HGF1. Upon EGF stimulation, Ras activity in HGF1 fibroblasts showed higher levels than control in all respective time intervals suggesting mutant SOS1 remained active even under serum-starved conditions, leading to higher Ras activity in response to growth factor stimuli. We next studied how the SOS1 mutation altered signal transduction through the MAPK/ERK pathway. As shown in Fig. 2B, panel a (lanes 2-5), transient activation of ERK1/2 with increasing concentrations of EGF resulted in a dose-related increase in both control and HGF1 fibroblasts. The magnitude of ERK1/2 activation was at least 30% greater in HGF1 than in control fibroblasts at each respective dosage of EGF treatment (Fig. 2B, panel b). In untreated cells, whereas the ERK1 signal was weak but detectable only in HGF1 samples, the ERK2 signal alone showed ϳ3-fold greater intensity in HGF1 than in control fibroblasts (Fig. 2B, panel a, lane 1). These data suggest Ras remains active and leads to a sustained activation of ERK1/2 signal in serum-starved HGF1 fibroblasts. Activation of ERK1/2 by EGF was evaluated in the presence of the selective pharmacological inhibitors AG1478 and PD98059 (32,33). The presence of PD98059 or AG1478 resulted in a similar degree of reduction on phospho-ERK signal in both cell types (Fig. 2C,  panel a, lanes 1-4). The overall phospho-ERK1/2 level remained higher in HGF1 than in control fibroblasts (Fig. 2C,  panel b). The duration of ERK activation was studied through the continued induction of EGF. In the presence of EGF, the phospho-ERK signal increased and peaked at the 1-h time point in both cell types (Fig. 2D). The signal gradually returned to basal levels after 6 h (Fig. 2D, panel a). Although the pattern of phospho-ERK response to EGF induction was similar in both cell types, the level was 50% higher in HGF1 than in control fibroblasts (Fig. 2D, panel b).
The effect of PD98059 on the duration of EGF-induced ERK signaling is shown in Fig. 2E. Activation of ERK signaling was reduced in the presence of MEK inhibitor in both cell types (Fig.  2E, panel a). However, the pattern of reduction was different in HGF1 compared with control cells (Fig. 2E, panel b). After 20-min incubation with inhibitor, 30% of phospho-ERK remained in control fibroblasts, whereas Ͼ80% of phospho-ERK remained in HGF1 fibroblasts. The phospho-ERK signal was sustained up to 2 h in HGF1 but not in control fibroblasts. At 4 h, phospho-ERK decreased to basal levels in controls, however, significant levels of active ERK signal remained in HGF1 fibroblasts. These results demonstrate that, in the absence of growth factors, Ras together with its downstream ERK1/2 signaling remained active and sustained in HGF1 fibroblasts. Transient induction by EGF elicited a stronger response, both in magnitude and duration, indicating the functional consequence of the SOS1 mutation in HGF1 fibroblasts.

Subcellular Localization of Mutant SOS1 in Gingival
Fibroblasts-Because binding of growth factors to their receptors triggers the recruitment of the SOS1-Grb2 complex from the cytoplasm to plasma membrane (34), we examined the distribution of SOS1 in subcellular fractions. In HGF1 fibroblasts, the endogenous full-length SOS1 was found in both cytosol and membrane fractions, with only trace amounts of mutant SOS1 detected in the membrane fraction (data not shown). This is due to the low expression level of the endogenous mutant SOS1 in HGF1 fibroblasts (Fig. 1B). Therefore, the HA-tagged fulllength and truncated SOS1 constructs were expressed in HeLa cells, which have higher transfection efficiency than primary gingival fibroblasts (data not shown). Under starving conditions, Ͻ40% of full-length SOS1 was distributed in the membrane/organelle fraction (Fig. 3A, lane 4), whereas Ͼ60% of truncated SOS1 was found in membrane/organelle fraction (lane 6). The SOS1(C) antibody, which only detects full-length SOS1, detected Ͻ30% of endogenous full-length SOS1 was distributed in the membrane/organelle fraction (compare lanes 1 and 2 with lanes 5 and 6). The identity of the additional (upper) band found in the membrane/organelle fraction (lanes 2 and 6) remains unknown. These results demonstrate the relative ratio of distribution is higher for truncated SOS1 in the membrane/ organelle fraction. Immunofluorescence staining revealed that, under growth conditions, both wild-type and mutant SOS1 localized in the cytoplasm and cellular membrane (Fig. 3B, panels a and b, arrows). However, in serum-starved fibroblasts, whereas wild-type SOS1 restricted in the cytoplasm (Fig. 3B,  panel c), mutant SOS1 localized in both the cytoplasm and plasma membrane (Fig. 3B, panel d, arrows). These observations were supported by the immunostaining on the distribution of endogenous SOS1 in serum-starved fibroblasts (Fig. 3C). These results suggest that, in the absence of growth factors, truncated SOS1 localized in the plasma membrane even though it lacks the carboxyl-proline-rich domains.
Mutant SOS1 Leads to Stronger ERK Signaling in HGF1 Fibroblasts-To test if the presence of mutant SOS1 leads to increased ERK signaling, SOS1 expression constructs were transfected into gingival fibroblasts, and ERK was monitored. Transfection of increasing amounts of the mutant SOS1 construct into serum-starved control fibroblasts resulted in a dosedependent increase of ERK signaling (Fig. 4, lanes 3-6). Levels of phospho-ERK signal were 50% greater in fibroblasts transfected with truncated construct than those transfected with wild-type construct (lanes 2 and 6). Levels of phospho-ERK in  2-4). Cellular extracts were probed with antibodies of phospho-ERK1/2 and total ERK1/2 as a control. B and C, panel b, quantification of data from a. The data of control (solid bar) and HGF1 fibroblasts (striped bar) were normalized with respective ERK1/2 and expressed as the -fold changes over control (lane 1). D, a, cellular extracts of serum-starved fibroblasts treated with EGF (50 ng/ml) for indicated time intervals were probed with anti-phospho-ERK1/2 and ␤-actin antibodies. b, quantification of data from (a). The data of control (dashed line) and HGF1 fibroblasts (solid line) were normalized with each respective ␤-actin and expressed as the -fold changes over control fibroblasts. E, panel a, serum-starved fibroblasts were treated with EGF (50 ng/ml) for 10 min, and medium was replaced with fresh medium containing PD 98059 (10 M) (lanes 3-6 and lanes 9 -12) and harvested at the indicated time points. Cellular extracts were probed with antibodies of phospho-ERK1/2 and total ERK1/2 as a control. Panel b, quantification of data from a. The data of control (dashed line) and HGF1 fibroblasts (solid line) were normalized with each respective ⌭⌹⌲1/2 and expressed as the -fold changes over control (lane 1).
wild-type-transfected fibroblasts were similar to fibroblasts transfected with three times less mutant construct (lanes 2 and  4). These results provide evidence that the presence of mutant SOS1 leads to higher levels of phospho-ERK signal in gingival fibroblasts.
Selective Knock-down of Wild-type and Mutant SOS1 by siRNA-RNA interference experiments were conducted to see if eliminating the endogenous mutant SOS1 reduced ERK signaling and affected fibroblast proliferation. To achieve the spe-cific knock-down on mutant transcript, the siRNA target sequences were designed to flank the mutation site. Three sets of siRNAs were designed to include the extra cytosine base together with either upstream or downstream sequences around the mutation site (Fig. 5A,  mS, mR, and mL). To evaluate the specificity and effectiveness of each siRNA, the duplex RNAs were cotransfected with indicated SOS1 expression constructs into HeLa cells. Each siRNA showed different degrees of knock-down effect (Fig.  5B). Almost all expressed full-length SOS1 was depleted in wS siRNAtreated samples (lane 3). Trace levels of SOS1 may represent endogenous SOS1, because it was equivalent to that in the vector control (lanes 1 and 3). Whereas wS showed ϳ100% depletion, mS exhibited ϳ60% knock-down effect on wild-type SOS1 (lane 3 and 4). As for the other duplex RNAs, whereas mL reduced about half (lane 6), mR did not have any effect on knockdown of wild-type SOS1 (lane 5). In the presence of expressed truncated SOS1 transcript, the wS and mL siRNAs showed slight or no knockdown effect when compared with controls (lanes 7, 8, and 11). In contrast, both mS and mR siRNAs inhibited all of the expressed truncated SOS1. The wS siRNA was therefore chosen to target the wild-type SOS1 transcript, and mR siRNA was chosen to target mutant transcript due to its nominal effect on wild-type SOS1. To confirm the above observations, different dosages of mR siRNA were co-transfected with either wild-type or mutant SOS1 expression construct into HeLa cells. Fig. 5C shows the level of wild-type SOS1 was gradually reduced as the dosage of mR increased, however, it was still detectable even when 100 nM of mR was applied (lane 6). In contrast, mutant SOS1 transcript was depleted completely when as little as 25 nM of mR was used (lane 9). These results demonstrate that mR duplex RNA could specifically and effectively knock-down the expressed truncated SOS1 with little effect on the expressed full-length SOS1.
The knock-down effects of wS and mR siRNA on endogenous SOS1 and ERK signaling are shown in Fig. 5D. Wild-type and mutant SOS1 were depleted 60 and 30%, respectively, by wS siRNA (lane 2). Although the mR siRNA did not affect the level of endogenous wild type, it showed a dose-response reduction on the endogenous mutant SOS1 with almost complete depletion at 50 nM (lane 6). The functional effects of SOS1 knock-down were demonstrated through the level of phospho-ERK signaling (Fig. 5, D and E). When transfected HGF1 fibroblasts were maintained under serum-starved conditions, acti-  arrow (a, b, and d). Bars, 20 m. C, serum-starved control (a) and HGF1 (b) fibroblasts were immunostained with SOS1(N) antibody. Arrows indicate endogenous SOS1 distribution at cell peripheral. Bar, 5 m.

FIGURE 4. Presence of mutant SOS1 results in an increase of ERK1/2 signal.
Control gingival fibroblasts were transfected with indicated SOS1 expression constructs. The pCGN vector construct was added (lanes 3-5) to ensure equal amounts of total plasmid were used for transfection. Culture medium was replaced with serum-starved medium for 16 h before harvest. Cellular extracts were subjected to SDS-PAGE and probed with antibodies against SOS1, phospho-ERK1/2, and total ERK1/2 as control.
vated ERK signaling was reduced ϳ50% following wS treatment (lane 2) and displayed a dose-dependent decrease in phospho-ERK with increasing amounts of mR siRNA (lanes 3-6). Similar reductions in phospho-ERK signal were observed when 50 nM wS or 5 nM mR was applied; however, only 10% of the phospho-ERK signal remained compared with the control when 50 nM mR was used (Fig. 5E). Reduction in phospho-ERK by wS was most likely due to partial depletion of mutant SOS1 (Fig. 5D,  lane 2). More importantly, although the mR siRNA showed little effect on wild-type SOS1 depletion, the phospho-ERK signal decreased as the concentration of mR increased (lanes 3-6) indicating that, under serum-starved culture conditions, mutant SOS1 contributed to sustained phospho-ERK signaling in HGF1 fibroblasts. These results demonstrate that mR siRNA can specifically target endogenous mutant transcript, with little if any effect on endogenous wild-type transcript of SOS1.

SOS1 Depletion Results in Decreased Fibroblast Proliferation-
To study the consequences of SOS1 depletion on gingival fibroblast proliferation, fibroblasts were transfected with the indicated siRNAs, and cell proliferation was monitored. Proliferation of control fibroblasts treated with wS (Fig. 6A, striped bar) decreased 20% on day 2 compared with untreated controls (solid bar). On day 4, cell numbers were further decreased. Although cell numbers increased at both days 6 and 8, the number of wS-treated fibroblasts was still ϳ30% lower than the untreated control fibroblasts. As expected, mR siRNA did not show any effect on the growth of control fibroblasts (unfilled bar). A similar trend appeared when wS and mR duplex RNAs were used together (dotted bar), although the reduction in cell proliferation was greater compared with the cultures treated with wS alone (striped bar). Although depletion of wild-type SOS1 reduced the proliferation of control fibroblasts, after day 4, cell proliferation started to recover, indicating the knock-down effects last ϳ4 days. When the same experiment was conducted for HGF1 fibroblasts in the presence of either wS or mR siRNAs, cell proliferation also decreased (Fig. 6B). Depletion of wild-type SOS1 (striped bar) resulted in 35% decrease in cell proliferation compared with control HGF1 fibroblasts (solid bar) from day 2 to day 6; however, no significant difference was observed by day 8. Knock-down of mutant SOS1 (unfilled bar) resulted in a Ͼ50% reduction in cell proliferation compared with HGF1 controls; however, the cell growth remained low and did not recover on day 8 as for the wild-type SOS1 (wS) knock-down. When HGF1 fibroblasts were treated with both wS and mR siRNAs, cell proliferation decreased ϳ70 and ϳ80%  7). Cellular extracts were blotted with antibody of SOS1(N) and ␤-actin. C, HeLa cells were co-transfected with indicated SOS1 construct (pSOS1) and mR siRNA (0, 10, 25, 50, and 100 nM), and the cellular extracts were blotted with SOS1(N) antibody and ␤-actin served as loading control. D, HGF1 fibroblasts were transfected with luciferase siRNA (control), wS (50 nM) or mR (lanes 3-6: 5, 10, 25, and 50 nM) for 48 h, and cellular extracts were resolved in SDS-PAGE and probed with antibodies of SOS1(N), phospho-ERK1/2, and total ERK1/2 as control. E, quantification of phospho-ERK1/2 from D. The data of phospho-ERK1/2 were normalized with each respective ERK1/2 and expressed as -fold changes of mean Ϯ S.D. over normal control. on days 2 and 4, respectively, compared with the HGF1 control cells (dotted versus solid bars). These data suggest 1) targeted knock-down of the mutant SOS1 transcript reduced proliferation of HGF1 fibroblasts and 2) knock-down of mutant SOS1 had a greater effect on reduction of fibroblast proliferation than the knockdown of wild-type SOS1. These findings suggest mutant SOS1 is responsible for the increase of fibroblast proliferation seen in HGF1.
Up-regulation of Cell Cycle Regulatory Proteins in HGF1 Gingival Fibroblasts-Sustained ERK signaling in fibroblasts is associated with cell cycle control (35,36). To test if prolonged ERK signaling affects cell cycle progression in primary gingival fibroblasts, we used a focused microarray containing 112 genes involved in cell cycle regulation. Under serum-starved conditions, HGF1-and EGF-treated control fibroblasts represent sustained and transient ERK signaling, respectively. To monitor the effects of either transient or sustained ERK signaling on the expression level of cell cycle regulators, two experimental groups were studied: 1) EGF-treated versus untreated control fibroblasts and 2) HGF1 versus control fibroblasts. As shown in Table 1, most transcripts showing a Ͼ2-fold increase in HGF1 over control fibroblasts were involved in the cell cycle progression and transition for G 1 phase and from G 1 into S phase. In most cases, the relative increase in differential expression was greater in the HGF1 versus control group than in the EGFtreated versus untreated control group, particularly, for cyclins C, D1, D2, E1, and E2 and RBBP8 (CTIP) ( Table 1). Although the transcription factors E2F1 and E2F2 were increased ϳ2-fold, their dimerization partners DP1 and DP2 were upregulated Ͼ4and 7-fold, respectively. These results reveal that, even under serum-starved conditions, key proteins involved in cell cycle regulation continued to be expressed, suggesting that these genes may be downstream targets of the sustained ERK signaling in HGF1 fibroblasts.
To validate the microarray findings real-time PCR and Western blotting were performed. Cyclin E-Cdk2 and Rb-E2F-DP complexes are known to promote cell cycle transition from G 1 into S phase in mammalian cells (37,38). When gingival fibroblasts cultures were serum-starved overnight, more phosphorylated Rb isoforms were detected in HGF1 fibroblasts than in controls (Fig. 7A, lanes 1 and 3). After switching to growth medium, a major phospho-Rb, most likely p130, was detected in both cell types, and the overall phospho-Rb was higher in HGF1 fibroblasts (lanes 2 and 4) indicating sustained phospho-ERK signaling in HGF1 fibroblasts resulted in stronger Rb phosphorylation. Expression of cyclin E1 (CCNE1) was 3-fold greater in HGF1 (Fig. 7B, unfilled bar) than in control fibroblasts (solid bar) under serum-starved conditions, consistent with microarray results. Similar increases were found in both control (striped bar) and HGF1 (dotted bar) fibroblasts grown in growth medium. Interestingly, cyclin E2 (CCNE2) expression was 2.5-fold higher in HGF1 fibroblasts (unfilled bar) than in control fibroblasts (solid bar) under serum-starved conditions. After switching to growth medium, levels of cyclin E2 increased significantly: 15-and 46-fold higher in control (striped bar) and HGF1 (dotted bar) fibroblasts, respectively. In addition, cyclin E2 levels were 8-fold higher in HGF1 (dotted bar) than in control fibroblasts (striped bar) under growth medium suggesting cyclin E2 plays a role in the overgrowth of HGF1.
Expressions of E2F1, E2F2, TFDP1, and TFDP2 were also evaluated, because activation of the E2F family is the key step for cell cycle progression from G 1 into S phase. Under serumstarved conditions, E2F1 transcript levels were similar for control and HGF1 fibroblasts (Fig. 7C, solid and unfilled bars). After switching to growth medium, E2F1 expression increased in control (10-fold) and HGF1 (40-fold) fibroblasts, respectively. Interestingly, the relative levels of E2F2 were 80-fold greater than E2F1 in both types of fibroblasts under serumstarved conditions. TFDP1 expression levels were 30% higher in HGF1 than in control fibroblasts in both serum-starved and growth conditions (Fig. 7D). No differences were found for TFDP2 expression between the two cells types regardless of culture conditions; however, after switching from starving to growth condition, it decreased 75% in both cell types. Taken together, these data suggest that 1) cyclin E2, E2F1, and TFDP1 play a major role in cell cycle regulation in gingival fibroblasts and 2) the cell cycle is more active in HGF1 fibroblasts than in control fibroblasts and may underlie the higher proliferation rates in HGF1 fibroblasts.
One of the immediate-early gene products induced by ERK signaling is the early growth response gene (Egr-1) (39). Under serum-starved conditions, trace amounts of Egr-1 were detected in control fibroblasts (Fig. 8A, lane 1), whereas signif- icant levels of Egr-1 were found in HGF1 fibroblasts (lane 2). After switching to growth medium, Egr-1 levels increased slightly in both control and HGF1 fibroblasts (lanes 3 and 4), however, Egr-1 levels remained 5-fold higher in HGF1. The presence of wS, mR, or both siRNA duplexes reduced both Egr-1 and PCNA, another marker for cell proliferation, in HGF1 fibroblasts compared with controls (Fig. 8B). Interest-ingly, higher levels of Egr-1 and PCNA were detected in the presence of wS compared with mR (lanes 2 and 3). Compared with the control sample (lane 1) Ͻ10% of Egr-1 and PCNA remained when both wS and mR were used together (lane 4). These results are consistent with the finding that greater ERK signal was observed in the knockdown by wS than by mR siRNA in HGF1 fibroblasts (Fig. 5D). These observations provide further evidence that sustained ERK signaling leads to the up-regulation of regulators that promote cell cycle progression from G 1 to S phase in HGF1 fibroblasts.

DISCUSSION
In this study, we have elucidated the mechanism of gingival overgrowth in HGF1 with SOS1 mutation. Our studies demonstrate that the overall level of SOS1 transcript is lower in HGF1 fibroblasts suggesting the mutant transcript is significantly less than the wild-type transcript. Although the relative level of wild-type SOS1 transcript appears to be higher than the mutant transcript in HGF1 fibroblasts (Fig. 1B), targeting mutant transcript by mR siRNA produced more profound effects on reduction of cell proliferation than wS (Fig. 6B), suggesting that the increased proliferation of HGF1 fibroblasts was chiefly dependent upon the function of mutant SOS1.
Both the N-and the C-terminal domains are involved in down-modulation of SOS activity (40). SOS1 lacking the C-terminal domains and targeted to the plasma membrane exhibits increased GEF activity and triggers the Ras signaling cascade without external stimuli (41)(42)(43)(44)(45). In HGF1 fibroblasts, without growth factors stimuli, we observed the truncated SOS1 distributed in the membrane/organelle faction. Although such subcellular fractionation analyses do not rule out the possibility that mutant SOS1 could associate with endomembranes, immunostaining of endogenous SOS1 and HA-tagged SOS1 showed that mutant protein could distribute to the plasma membrane in serum-starved fibroblasts. The mechanism for the translocation remains unknown. It is unlikely that the novel 22 C-terminal residues resulting from the frameshift mutation play a role in membrane targeting, because it failed to direct green fluorescent protein to the plasma membrane (data not shown). Recently, SOS1 substitution mutations found in some Noonan syndrome patients (46,47) are critical in the maintenance of SOS1 autoinhibition and lead to elevation of Ras activity. In HGF1, the truncation mutation abolishes the C-terminal pro-  line-rich domains and could also release the autoinhibition state of SOS1.
Mutant SOS1 in HGF1 fibroblasts is capable of sustained activation of Ras/MAPK signaling in the absence of growth factors stimuli. Activation of ERK signaling was not blocked by AG1478 suggesting mutant SOS1 was already present in the plasma membrane, contributing to sustained activation of Ras signaling. The presence of PD98059 also reduced but did not completely block ERK signaling. These findings suggest that mutant SOS1 may activate ERK signaling through other pathways or that HGF1 cells reduced their sensitivity to the kinase inhibitor through an unknown mechanism.
Mutations in key components of the MAPK pathway that result in sustained ERK activation correlate with tumor formation (48,49). Furthermore, high transformation activity in NIH3T3 fibroblasts and skin tumor formation in transgenic mice were shown by artificially constructed C-terminal truncated forms of SOS1 (43,44,50). It is believed that the strong promoter activity used in these constructs produces high levels of mutant protein expression, which leads to strong ERK signaling and transformation activity. Therefore, a threshold level of expression might be necessary to transform cells. We speculate that the benign form of overgrowth in HGF1 can be explained as following: the low level of SOS1 mutant with partial release from its autoinhibition state could result in sustained activation with attenuated ERK signal, which is sufficient to lead to higher proliferation but not strong enough to cause transformation.
In primary fibroblasts, overexpression of the key players in the MAPK pathway such as Ras and MEK results in sustained Ras/ERK signaling and cell growth arrest (51,52). When we compared sustained (HGF1) versus transient (control treated with EGF) activated ERK signaling in gingival fibroblasts, the higher expression of cyclin C, D, and E family members in HGF1 fibroblasts (Table 1) suggests these cyclin proteins are the downstream targets of the sustained ERK signaling. Interestingly, the expression of cyclin E2 is low or undetectable in non-transformed cells and elevates significantly in tumor-derived cells (53). We found that levels of cyclin E2 increase markedly in both cell types after medium switch indicating cyclin E2 plays a major role in transition from G 1 to S phase in gingival fibroblasts (Fig. 7B).
E2F and TFDP also play a pivotal role in the control of cell cycle progression (54,55). Different E2F-DP complexes bind to different pRb family members, pRb, p107, and p130 (53). Upon hyperphosphorylation of Rb, the E2F-DP complex is released and activates their downstream target genes (56). We observed higher levels of phospho-Rb in HGF1 fibroblasts under both serum-starved and growth conditions. When serum-starved fibroblasts were switched to growth conditions, E2F1 and TFDP1 levels, not E2F2 and TFDP2, were significantly increased (Fig. 7, C and D). It is likely that E2F1 plays a major role in cell cycle control in response to the sustained ERK signal. Both transient and sustained activation of ERK induce expression of immediate-early genes such as fos, jun, and myc; however, only sustained ERK signaling leads to phosphorylation and stabilization of immediate-early gene-encoded proteins (17,18). We observed that the increase of Egr-1 expression in HGF1, and the depletion of either wild-type or mutant SOS1, reduced the level of both Egr-1 and PCNA, one of the target genes of the E2F-DP complex (Fig. 8).
The findings in this study reveal a Grb2-independent function of mutant SOS1 and suggest a mechanism for HGF1 fibroblast proliferation (see supplemental Fig. S1). Unlike SOS1 mutations in Noonan syndrome, where no gingival overgrowth was reported (46,47), the truncated SOS1 in HGF1 results clinically in gingival overgrowth. It is unclear why different SOS1 mutations result in different clinical phenotypes. The SOS1 insertion mutation in HGF1 results in a truncated, chimeric protein, whereas the Noonan syndrome mutations are substitution mutations that alter a single amino acid. In addition, why the overgrowth phenotype of HGF1 is restricted to gingival tissues remains unclear. At least four alternative isoforms of SOS1 have been reported in different tissues and cell lines (49). We have observed two SOS1 isoforms (I and II) expressed in gingival fibroblasts, with the majority consisting of isoform II, which contains an extra 15 C-terminal residues that are not present in isoform I (57), (data not shown). It is unlikely these alternative isoforms contribute to the tissue-specific phenotype observed in HGF1, because the early termination of SOS1 occurs upstream of the 15-amino acid difference and would therefore affect both isoforms. It is possible that specific E2F-DP-Rb complexes together with their target genes may be present in gingival fibroblasts and contribute to the localized gingival phenotype. In conclusion, the SOS1 mutation in HGF1 appears to be a gain of function that drives an increase in gingival fibroblast proliferation. Human HGF1 offers a unique model and opportunity to study the SOS1/Ras/MAPK signaling pathway and its biological outcome in a disease state, and the application of RNA interference on allele-specific depletion of SOS1 also provides a potential tool for controlling the gingival overgrowth.