Activation state-dependent interaction between Gαq subunits and the Fhit tumor suppressor

Background The FHIT tumor suppressor gene is arguably the most commonly altered gene in cancer since it is inactivated in about 60% of human tumors. The Fhit protein is a member of the ubiquitous histidine triad proteins which hydrolyze dinucleoside polyphosphates such as Ap3A. Despite the fact that Fhit functions as a tumor suppressor, the pathway through which Fhit inhibits growth of cancer cells remains largely unknown. Phosphorylation by Src tyrosine kinases provides a linkage between Fhit and growth factor signaling. Since many G proteins can regulate cell proliferation through multiple signaling components including Src, we explored the relationship between Gα subunits and Fhit. Results Several members of the Gαq subfamily (Gα16, Gα14, and Gαq) were found to co-immunoprecipitate with Fhit in their GTP-bound active state in HEK293 cells. The binding of activated Gαq members to Fhit appeared to be direct and was detectable in native DLD-1 colon carcinoma cells. The use of Gα16/z chimeras further enabled the mapping of the Fhit-interacting domain to the α2-β4 region of Gα16. However, Gαq/Fhit did not affect either Ap3A binding and hydrolysis by Fhit, or the ability of Gαq/16 to regulate downstream effectors including phospholipase Cβ, Ras, ERK, STAT3, and IKK. Functional mutants of Fhit including the H96D, Y114F, L25W and L25W/I10W showed comparable abilities to associate with Gαq. Despite the lack of functional regulation of Gq signaling by Fhit, stimulation of Gq-coupled receptors in HEK293 and H1299 cells stably overexpressing Fhit led to reduced cell proliferation, as opposed to an enhanced cell proliferation typically seen with parental cells. Conclusions Activated Gαq members interact with Fhit through their α2-β4 region which may result in enhancement of the growth inhibitory effect of Fhit, thus providing a possible avenue for G protein-coupled receptors to modulate tumor suppression.


Background
The chromosomal localization of FHIT (Fragile Histidine Triad) in the common fragile region of the human genome suggests a positive correlation between the loss or inactivation of the FHIT gene and carcinogenesis. As predicted for a tumor suppressor, the Fhit protein is absent or markedly reduced in most human cancers [1]. The role of FHIT in tumor suppression is perhaps best exemplified by studies performed with FHIT-deficient mice. Transgenic mice carrying one or two inactivated Fhit alleles are viable and long-lived, but they show increased rates of spontaneous and carcinogen-induced cancers [2,3]. Encouragingly, the development of carcinogeninduced tumors in these mice can be prevented by administration of Fhit-expressing viral vectors [4]. Moreover, Fhit overexpression enhances the susceptibility of many types of cancer cells to exogenous inducers of apoptosis.
Fhit is one of the HIT (histidine triad) superfamily members, which share an HxHxHxx motif (where x is a hydrophobic residue) for nucleotide binding. Human Fhit can hydrolyze dinucleoside polyphosphates, preferably Ap 3 A (to AMP and ADP). Despite numerous attempts to elucidate the function of Fhit in tumor suppression, the biological action of Fhit remains elusive. Current evidence based on Fhit mutants with impaired substrate binding (L25W and I10W/L25W mutants) or hydrolytic activity (H96D mutant) supports the notion that the formation and stability of the Fhit-Ap 3 A complex is crucial in growth inhibition and apoptosis [5][6][7]. There is also evidence to suggest that the intracellular concentration of Ap 3 A [8] or its abundance relative to other dinucleoside polyphosphates [9] may be correlated with Fhit-mediated apoptosis. The hypothesis that the Fhit-Ap 3 A complex could be an important signaling molecule is an interesting possibility, but it has yet to be confirmed biochemically.
A number of important cancer-related genes and pathways have recently been linked to Fhit. In colon cancer cell lines, Fhit inhibits cell growth by attenuating the signaling mediated by NFκB [10]. Fhit also inhibits the activity of Akt, a key effector in the phosphatidylinositol 3-OH kinase (PI3K) pathway [11], and serves as a physiological target of the Src tyrosine kinase [12]. Src is a crucial cytoplasmic tyrosine kinase downstream of several growth factor receptors, including those of the EGF receptor family, which are often overexpressed and activated in human breast and ovarian carcinomas. Indeed, activation of EGF receptor family members induces Fhit degradation via the proteasome pathway which purportedly depends on Src-mediated Fhit phosphorylation at Tyr 114 [13]. However, biochemical data suggest that phosphorylation favors the formation and persistence of the Fhit-Ap 3 A complex [14]. Additionally, the mitochondrial Fhit can sensitize cells to apoptosis by binding and stabilizing ferredoxin reductase [15], which is important for the production of reactive oxygen species, and by enhancing mitochondrial Ca 2+ -uptake capacity [16]. These reports help us to better understanding the mechanism of tumor suppression by Fhit, but it remains unclear as to how one can restore Fhit levels in the tumor cells for cancer treatment.
Many signaling pathways operated by growth factors are similarly modulated by the heterotrimeric G proteins, which are critical players in many aspects of cellular function including cell proliferation, differentiation and apoptosis. These signaling pathways include the mitogen-activated protein kinases (MAPKs) [17], PI3K/ Akt [18], tyrosine kinases [19], and transcription factors such as STAT3 and NFκB [20,21]. Gα subunits of heterotrimeric G protein are classified into four subfamilies (Gα s , Gα i , Gα q , and Gα 12 ) [22]. It is noteworthy that some Gα subunits can directly activate tyrosine kinases such as Bruton's tyrosine kinase (Btk) [19]. Interestingly, Src has also been shown to be activated by members from all four subfamilies of G proteins [23][24][25][26] and this may provide a link to regulate Fhit phosphorylation. Constitutively activating mutations of the Gα subunits that lock these signaling molecules in their GTP-bound active state have been found to be associated with several types of tumor [27]. Sustained stimulation of the G q and G 12 pathways often leads to mitogenesis in various cell types [28]. As a continuing effort to understand the functions of G proteins in cell growth and proliferation, we have explored the notion that G proteins can modulate Fhit. Surprisingly, we discovered that several α subunits of G q family members can associate with Fhit only in their active state.

Constitutively active Gα q mutants stimulate Fhit phosphorylation at Tyr 114 through Src
Src is known to be activated by Gα q subunits [20,25] and thus it is conceivable that stimulation of G q -coupled receptors may lead to Fhit phosphorylation. To facilitate the detection of Fhit phosphorylation, we raised an antiphospho-Fhit Tyr 114 antiserum which can detect Srcinduced Fhit Tyr 114 phosphorylation with high sensitivity ( Figure 1A); overexpression of Src was sufficient to induce Fhit phosphorylation in transfected HEK293 cells due to the increase in activated Src (P-Src in Figure 1A). We then began the study by examining the ability of the G q -coupled type 2 bradykinin receptor (BK 2 R) to stimulate Fhit phosphorylation by using a previously characterized HEK293 cell line stably expressing BK 2 R (293/ BK 2 R cells) [29]. 293/BK 2 R cells transiently expressing Flag-Fhit were stimulated with or without 100 nM bradykinin for various durations and then assayed for Fhit phosphorylation. Bradykinin-induced Fhit phosphorylation was hardly detected at short treatment times (data not shown) but was reproducibly observed albeit weakly with cells treated for 24 h (~2.5-fold of basal; Figure 1B, DMSO control). As shown in Figure 1B, bradykinin-induced Fhit Tyr 114 phosphorylation was significantly suppressed by pretreatment of the cells with Src inhibitors (10 μM PP1 or 25 μM PP2). As HEK293 cells endogenously express the G q -coupled muscarinic M 3 receptor [30], we examined whether receptor activation can induce Src-mediated Tyr 114 phosphorylation of endogenous Fhit. In contrast to 293/BK 2 R cells overexpressing Flag-Fhit, we could not detect carbachol-induced phosphorylation of endogenous Fhit in native HEK293 cells unless the cells were treated with 100 μM Na 3 VO 4 , a tyrosine phosphatase inhibitor ( Figure 1C); this suggests that phosphorylated Fhit may undergo dephosphorylation and thereby making its detection extremely difficult when the level of phospho-Fhit is limiting. Nevertheless, the carbachol-induced phosphorylation of endogenous Fhit was sensitive to Src inhibition by PP1 ( Figure 1C). In order to confirm that G q signals can lead to Fhit phosphorylation, we made use of constitutively active mutants of Gα q subunits as well as Fhit Y114F, a previously characterized non-phosphorable mutant [12,13]. The constitutively active Gα mutants harbor a point mutation at a conserved arginine or glutamine (e.g., Gα q R183C or Gα q Q209L) which abolishes the GTPase activity of the Gα subunit and maintains them in the GTP-bound active state. Transient co-expression of constitutively active Gα q mutants with Fhit should lead to increased phosphorylation of wild-type Fhit but not Fhit Y114F. Interestingly, co-expression of constitutively active mutants of Gα q or Gα 14 (another member of the Gα q family) with Fhit resulted in increased levels of the latter (Additional file 1), a phenomenon similar to that seen with bradykinin-treated 293/BK 2 R cells (cf lanes 1 and 2 of the Flag-Fhit immunoblot in Figure 1B). After adjusting the expression level of Fhit between the various transfectants, Fhit phosphorylation was clearly detected in cells co-expressing the constitutively active Gα q RC or Gα 14 QL ( Figure 1D). Transfectants coexpressing the wild-type Gα subunits exhibited little or no Fhit phosphorylation while no phospho-Fhit could be detected in cells co-expressing Fhit Y114F ( Figure 1D).
As tyrosine kinases such as Btk can be directly activated by Gα q [19], we examined whether Src can form complexes with Fhit and/or Gα q . Because activated Gα 16 (GNA15, another member of Gα q subfamily with 85% sequence identity to its mouse isoform Gα 15 [31]) has previously been shown to stimulate Src phosphorylation at Tyr 416 [21], we transfected HEK293 cell with different combinations of Flag-Fhit, Src, Gα 16 and Gα 16 QL and then subjected the cell lysates to co-immonuprecipitation assays using an anti-Flag affinity gel ( Figure 1E). Both Src and Gα 16 QL were detected in the immunoprecipitates of Flag-Fhit when all three proteins were co-expressed simultaneously ( Figure 1E, lane 4); note that the Src-specific band (marked by an asterisk) ran just above a non-specific IgG band. Control experiments omitting either Src or Gα 16 QL demonstrated that both proteins were able to interact with Flag-Fhit independently or endogenous levels of interacting proteins (including Src and Gα q subunits) were not limiting (cf lanes 1 and 6 in Figure 1E). Compared to Gα 16 QL, wild-type Gα 16 exhibited a much weaker ability to associate with Flag-Fhit (cf lanes 3 and 5 versus 4 and 6 in Figure 1E). Yet again, co-expression of Gα 16 QL, but not wild-type Gα 16 or Src, increased the levels of Fhit in the transfectants ( Figure 1E, lanes 4 and 6). Taken together, these results suggest that Fhit may associate with Gα subunits in a GTP-bound statedependent and Src-independent manner.

Several Gα q members interact with Fhit in an activity-dependent manner
The preceding experiments suggest that members of the Gα q subfamily may interact with Fhit upon binding GTP. To assess if this interaction is specific to Gα q subunits, we performed co-immunoprecipitation assays using Flag-Fhit and various Gα subunits. HEK293 cells were co-transfected with Flag-Fhit or Flag-vector in combination with a selected Gα subunit in its wild-type or constitutively active form. The expressions of Flag-Fhit and Gα subunits between different groups were adjusted to comparable levels prior to co-immunoprecipitation with an anti-Flag affinity gel or anti-Gα antiserum. Constitutively active mutants of Gα q , Gα 14 , and Gα 16 , but not their wild-type counterparts, formed complexes with Flag-Fhit as predicted ( Figure 2A). However, despite being a member of the Gα q subfamily, the constitutively active mutant of Gα 11 failed to interact with Flag-Fhit ( Figure 2A). Representative members (Gα s , Gα i2 and Gα 13 ) from each of the remaining Gα subfamilies were also subjected to co-immunoprecipitation assays with Flag-Fhit. As shown in Figure 2A, both wild-type and Figure 2 Fhit interacts with activated Gα q subunits but not with Gβ, small GTPases or RGS proteins. A, HEK293 cells were co-transfected with either pFlag-CMV2-Fhit (F) or pFlag-CMV2 vector (V) and in combination with individual construct encoding wild-type or the constitutively active mutant (RC for G q , QL for the others) of different Gα proteins: G q , G 11 , G 14 , G 16 , G s , G i2 and G 13 . After 24 h overexpression, cell lysates were prepared and subjected to immunoprecipitation (IP) with anti-Flag agarose affinity gel. Total cell lysates (TCL) and the immunoprecipitates were analyzed by Western blotting (WB). B, HEK293 cells were transfected with pcDNA3 (Vector) or pFlag-CMV2-Fhit together with Gα 16 or Gα 16 QL. Transfectants were subjected to IP with anti-Gα 16 antiserum and protein G agarose. C, DLD-1 colon carcinoma cell lysates were incubated without or with GDPβS or GTPγS for 30 min at 4°C and then subjected to IP with anti-Fhit antiserum and protein A agarose. D, HEK293 cells were co-transfected with either Flag-Gβ 1 or pFlag-CMV2 vector with HA-Gγ 2 and pcDNA3-Fhit. Transfectants were immunoprecipitated with anti-Flag agarose affinity gel. E, HEK293 cells were co-transfected with either pFlag-CMV2-Fhit or pFlag-CMV2 vector and in combination with individual HA-tagged construct encoding Ras, RGS19, Rap1A, or RGS16. Cell lysates were immunoprecipitated with anti-Flag agarose affinity gel. Data shown represent one of three or more sets of immunoblots; other sets yielded similar results.
constitutively active Gα s and Gα 13 were pulled down by Flag-Fhit, but not by the vector control, suggesting that Gα s and Gα 13 were capable of forming complexes with Flag-Fhit irrespective of their activation status. Neither wild-type nor constitutively active Gα i2 or Gα z was co-immunoprecipitated with Flag-Fhit, indicating that both Gα i2 and Gα z behaved like Gα 11 and could not associate with Fhit. To ascertain that Fhit can truly interact with activated members of Gα q , we examined the association between Gα 16 QL and Fhit by reciprocal co-immunoprecipitation using an anti-Gα 16 antiserum to pull down Fhit from lysates of HEK293 cells expressing wild-type Gα 16 or Gα 16 QL; Fhit was indeed co-immunoprecipitated along with Gα 16 QL, but not with wild-type Gα 16 ( Figure 2B). To further confirm their interaction in a native system, we screened for cell lines that endogenously express Fhit at a detectable level. Out of eight cell lines examined, DLD-1 colon carcinoma cells have relatively high levels of endogenous Fhit (data not shown) and they were used to examine the interaction between endogenous Fhit and Gα q . Cell lysates were incubated with non-hydrolysable GDPβS or GTPγS (100 μM each at 4°C for 30 min) to shift the endogenous G proteins to the basal or activated state, respectively. Cell lysates were subsequently subjected to co-immunoprecipitation with anti-Fhit antiserum and protein A sepharose. As compared to the controls, more Gα q was detected in the Fhit immunoprecipitate following GTPγS treatment ( Figure 2C). This result suggests that activated Gα q subunits can interact with Fhit in a native cellular environment.
Since other signaling components along the G protein pathway may also be involved in the Fhit/Gα q interaction, possible association of Fhit with Gβγ, regulators of G protein signaling (RGS proteins), and monomeric GTPases were examined by co-immunoprecipitation assays. Many effectors such as adenylyl cyclase and phospholipase Cβ (PLCβ) can be simultaneously regulated by Gα and Gβγ subunits. It is thus worth investigating whether Fhit can also associate with Gβ 1 γ 2 , a Gβγ complex which is known to bind various effectors including tyrosine kinases [32]. We co-expressed Flag-tagged Gβ 1 and HA-tagged Gγ 2 with untagged Fhit. The Flag-Gβ 1 subunit was clearly capable of forming a complex with HA-Gγ 2 , yet it was unable to co-immunoprecipitate Fhit ( Figure 2D). As shown in Figure 2E, both RGS19 (also known as Gα-interacting protein, GAIP) and RGS16 did not co-immunoprecipitate with Flag-Fhit. RGS4, RGS10, and RGS20 also failed to interact with Fhit (data not shown). It should be noted that, under identical experimental conditions, RGS19 and Ras can interact efficiently with their known partners [33,34]. Monomeric small GTPases contain the same core domains for GTP-binding as the heterotrimeric Gα subunits. Hence, the ability of Flag-Fhit to form a complex with selected small GTPases was examined. Neither Ras nor Rap1A, which belong to the Ras family of the small GTPase superfamily, could be coimmunoprecipitated by Flag-Fhit ( Figure 2E), suggesting that small GTPases cannot form complexes with Fhit protein. These observations further support the notion that Gα q /Fhit interactions are specific and not shared by other signaling components along the G protein pathway.
Activated Gα 16 interacts with Fhit directly through its α2-β4 region To investigate whether Fhit is able to directly interact with activated Gα q members, we performed pull-down assays using purified GST, GST-tagged Fhit (GST-Fhit) and His-tagged Gα 16 (His-Gα 16 ). The purity of both GST-Fhit and His-Gα 16 proteins was estimated to be greater than 90% by Coomassie blue staining ( Figure 3A). Equal amounts of recombinant His-Gα 16 and GST-Fhit (or GST) were incubated at 4°C for 30 min in the presence of 100 μM GDPβS or GTPγS in order to stabilize His-Gα 16 in the inactive or active conformation. Although a small amount of His-Gα 16 appeared to be nonspecifically associated with the glutathione sepharose ( Figure 3B, lanes 1 and 2 of right panel), GTPγS-His-Gα 16 was clearly pulled down by GST-Fhit ( Figure 3B, lane 4 of right panel). In contrast, GDPβS-His-Gα 16 failed to associate with GST-Fhit. Collectively, these results suggest that Fhit can selectively associate with activated Gα q members except Gα 11 , and both purified Gα 16 and endogenous Gα q can interact with Fhit in their active states. Such activation state-dependent interactions are reminiscent of Gα/effector regulations.
In order to understand the molecular basis of the interaction between Gα q and Fhit, we mapped the Fhitinteracting regions on Gα 16 by using a series of chimeras in which discrete regions of Gα 16 were swapped with Gα z (a member of Gα i subfamily). These chimeras have Figure 4 The α2-β4 region is important for Gα 16 to interact with Fhit. A, Schematic representation of the N188, N210, N246, N266, C128 and C164 chimeras. Predicted secondary structures are illustrated as boxes (α helices) or ovals (β strands) above the chimeras. Sequences from human Gα z are shaded in grey while those from human Gα 16 are in black. B, HEK293 cells were transiently co-transfected with Flag tagged Fhit and the wild-type or constitutively active mutants of Gα 16 , Gα z , N188, N210, N246, N266, C128 or C164. Cell lysates were immunoprecipitated with anti-Flag agarose affinity gel (upper panels). Expression levels of Gα 16 , Gα z , Flag-Fhit and α-tubulin in the total cell lysate were detected by western blotting (lower panels). Data shown represent one of three or more sets of immunoblots; other sets yielded similar results. Figure 5 Interaction with Gα q is not dependent on the ability of Fhit to bind or hydrolyze Ap 3 A. A, HEK293 cells were co-transfected with either one of pFlag-CMV2 constructs encoding wild-type Fhit, Fhit-Y114F, Fhit-L25W, Fhit-I10W/L25W, Fhit-H96D or pFlag-CMV2 vector and in combination with wild-type Gα q or constitutively active Gα q mutant (Gα q RC). Cell lysates were immunoprecipitated with anti-Flag affinity gel. Numerical values shown above the blot represent the relative intensities of Gα q RC being co-immunoprecipitated as compared to their corresponding wild-type Gα q . The band intensity of wild-type Gα q pulled down by Flag-Fhit was set as 1.0. Data shown represent one of three or more sets of immunoblots; other sets yielded similar results. B, Ap 3 A (100 μM) was incubated in the absence (top) or presence of 1 μg GST protein (middle) or 1 μg GST-Fhit protein (bottom) at 37°C for 10 min and then subjected to HPLC analysis. The elution profiles were compared with HPLC elution profiles of nucleotide standards (data not shown) including been previously used to successfully determine the receptor and effector interacting domains of Gα 16 and Gα z [35,36]. Gα 16/z chimeras were preferred because of the lack of endogenous expression of either Gα 16 or Gα z in HEK293 cells. The differential ability of Gα 16 QL and Gα z QL to interact with Fhit ( Figure 2A) permits identification of Fhit-interacting regions on Gα 16 through gain of function analyses. Since the effector interacting domain is likely to reside in the carboxyl half of the Gα subunit [36,37], we have selected chimeras composed of Gα z backbones with their C-terminal regions increasingly replaced by Gα 16 sequences all the way up to the β2 domain ( Figure 4A); mirror images of selected chimeras were also included. Among the various chimeras examined, constitutively active N188QL and N210QL (N-terminal 188 or 210 amino acids from Gα z , respectively) were more efficiently pulled down by the anti-Flag affinity gel than their corresponding wild-types; both chimeras were as effective as, if not better than, Gα 16 QL ( Figure 4B). Constitutively active C128QL (C-terminal 128 amino acids from Gα z ) also showed higher affinity with Fhit than its wild-type ( Figure 4B). In contrast, N246QL, N266QL and C164QL failed to associate with Flag-Fhit and behaved like the negative control Gα z QL ( Figure 4B). These results demonstrate that the residues between 210 and 246 of Gα 16 , which represent the regions from α2 to β4, are required for interaction with Fhit. Based on the structures of active Gα q in the complex with p63RhoGEF and RhoA [PDB: 2RGN_A] as well as inactive Gα q with Gβγ complex [PDB: 3AH8], molecular modeling of Gα 16 predicted that the α2-β4 domain interacts with Gβγ in the inactive state but becomes exposed to the outer surface in the active state (Additional file 2). We have also attempted to determine the Gα q -interacting region on Fhit by constructing a series of Fhit truncation mutants with deletions at either the C-or N-terminus (Additional file 3). However, deletion at either terminus apparently impaired the stability of these mutants because their expressions were hardly detectable unless the transfected cells were treated with the proteasome inhibitor MG132 (Additional file 3). The inadequate expression of these truncation mutants precluded coimmunoprecipitation assays. Nevertheless, expressions of two mutants were enhanced upon co-expression of Gα q QL, but not Gα q (Additional file 3). This suggests that interaction with activated Gα q may stabilize Fhit.

Formation of the Gα q /Fhit complex is independent of Fhit's ability to bind Ap 3 A or be phosphorylated at Tyr 114
In an attempt to unveil the biological function of the Gα q /Fhit interaction, we asked if such association is affected by Fhit phosphorylation at Tyr 114 or Fhit's ability to bind Ap 3 A. Previous studies have shown that Fhit undergoes degradation upon phosphorylation by Src kinase at Tyr 114 [13] and activated Gα q can stimulate tyrosine kinases [25]. Many signaling molecules regulate their binding to protein partners through tyrosine phosphorylation. To test if this holds true for Fhit, we employed the Fhit Y114F mutant in co-immunoprecipitation assays. Since Flag-Fhit Y114F appeared to interact with constitutively active Gα q RC to an extent similar to Flag-Fhit ( Figure 5A), it suggests that phosphorylation of Fhit Tyr 114 is not a prerequisite for the formation of Gα q /Fhit complexes.
Ap 3 A is the substrate of Fhit, and binding of Ap 3 A to Fhit can affect the conformation of Fhit and hence its ability to associate with other proteins. I10W/L25W and L25W are Fhit mutants that exhibit 30-and 7-fold increase of K m , respectively [7]. Apparently, these mutants have a lower affinity to associate Ap 3 A although they can still hydrolyze Ap 3 A. On the other hand, H96D, the Ap 3 A hydrolytic dead mutant of Fhit does not hydrolyze Ap 3 A and stabilizes the Ap 3 A/Fhit conformation [6]. Therefore, the associations between Gα q and these mutants were assessed. As shown in Figure 5A, all three mutants effectively co-immunoprecipitated Gα q RC but not wild-type Gα q ; their interactions with Gα q RC were essentially similar to that observed with Flag-Fhit. Hence, the binding of Ap 3 A to Fhit has little or no effect on the formation of Gα q /Fhit complexes.
Since many activated Gα subunits can regulate the enzymatic activity of their effectors, constitutively active Gα q may modulate the hydrolase activity of Fhit. To test this possibility, we used purified GST-Fhit and His-Gα 16 proteins. The hydrolysis of Ap 3 A to AMP and ADP was monitored by HPLC as described previously [38]. Upon incubation with 1 μg GST-Fhit at 37°C for 10 min, 100 μM Ap 3 A was completely hydrolyzed to AMP and ADP ( Figure 5B). No hydrolysis was detected when Ap 3 A was incubated with GST alone or with heat denatured GST-Fhit ( Figure 5B). We then optimized the assay in order to cater for the detection of possible stimulatory effect on the hydrolase activity of Fhit. Upon reducing the amount of GST-Fhit in the reaction to 0.5 μg, approximately half of the Ap 3 A was hydrolyzed to AMP and ADP ( Figure 5C). To mimic the constitutively active Gα 16 QL, the recombinant His-Gα 16 protein was loaded with 100 μM GTPγS. His-Gα 16 protein loaded with GDPβS was used as a negative control. As shown in Figure 5C, the presence of GTPγS-bound or GDPβS-bound His-Gα 16 did not affect the ability of GST-Fhit to hydrolyze Ap 3 A. The extent of Ap 3 A Figure 7 Fhit does not affect G q -mediated TPR1 interaction or Ras activation. A, HEK293 cells were co-transfected with different combinations of pcDNA3 (Vector), Fhit, Gα 16 , Gα 16 QL, Flag-TPR1 and Flag tag constructs. One day later, cells were immunoprecipitated with anti-Flag agarose affinity gel and probed for the presence of Fhit, Gα 16 , and Flag-TPR1 in the immunoprecipitates using specific antibodies as indicated. B, HEK293 cells were co-transfected with either pcDNA3 (Vector) or Fhit in combination with pcDNA3, Gα 16 or Gα 16 QL, and subjected to Ras activation assay. The results shown on the bar chart represent relative band intensities of Ras from RBD-Raf-1 immunoprecipitation normalized against their corresponding amounts of total Ras (RBD-Raf-1/TCL). The value of vector-transfected basal control was set as 100%. * Transfection of Gα 16 QL significantly induced Ras activity as compared to vector control (Dunnett's t test, P < 0.05). Immunoblots shown represent one of three sets; two other sets yielded similar results.
hydrolysis by GST-Fhit was essentially identical under all three conditions ( Figure 5D). Neither guanine nucleotides interfered with the detection of the substrate or product; GDPβS was eluted after Ap 3 A while GTPγS could not be detected under our experimental conditions. These results suggest that activated Gα 16 does not regulate the hydrolase activity of Fhit. However, it remains possible that activated Gα 16 can indirectly modulate the enzymatic activity of Fhit in a cellular environment.
Fhit does not alter the signaling function of Gα q As members of the Gα q family are known to regulate mitogenic pathways [39], Fhit may exert its tumor suppressive effect by altering the functions of these Gα subunits. To test this postulation, we determined the effect of Fhit on the ability of Gα q and Gα 16 to regulate a panel of known effectors. We first examined the ability of Gα q RC and Gα 16 QL to stimulate PLCβ in the absence or presence of Fhit overexpression. HEK293 cells were cotransfected with various combinations of Fhit, Fhit mutants, and wild-type or constitutively active mutants of Gα q and Gα 16 . As predicted, both Gα q RC and Gα 16 QL were capable of stimulating the endogenous PLCβ and inducing the formation of IP 3 ( Figure 6A). Co-expression of Fhit or its mutants neither stimulated nor inhibited the ability of Gα q RC and Gα 16 QL to activate PLCβ ( Figure 6A). We have also examined whether Fhit affects the ability of endogenous G q -coupled histamine receptors to stimulate PLCβ activity in HeLa cells. As there are conflicting results on the Fhit expression level in HeLa cells [16,[40][41][42][43], we have confirmed that the HeLa cells used in our study do express endogenous Fhit to a level slightly higher than that seen with HEK293 cells (Additional file 4), which are known to express Fhit [8]. Variations in the reported Fhit levels in HeLa cells might be attributed to differences in the gene expression profiles of sublines. After knocking Figure 8 Overexpression of Fhit does not affect constitutively active Gα q or Gα 16 -induced phosphorylation of ERK, IKK, or STAT3, or the activity of NFκB. A, HEK293 cells were co-transfected with either pcDNA3 (Vector) or Fhit in combination with pcDNA3, Gα q or Gα q RC. Cell lysates were prepared and immunoblotted with antiphospho-ERK (P-ERK), anti-ERK (ERK), anti-phospho-IKK (P-IKK), or anti-IKK (IKK). B, HEK293 cells were transfected as in A except the Gα q constructs were replaced with those corresponding to Gα 16 . Cell lysates were probed with anti-phospho-Tyr 705 -STAT3 (P(Y)-STAT3), anti-phospho -Ser 727 -STAT3 (P(S)-STAT3) or anti-STAT3 (STAT3) antiserum. The vector transfection control was set as 100% control. * The level of ERK, IKK and STAT3 phosphorylation was significantly higher than the vector control (Dunnett's t test, p < 0.05). Immunoblots shown represent one of at least three sets; all other sets yielded similar results. C, HEK293 cells stably expressing the NFκB luciferase reporter gene, pNFκB-TA-luc, were cotransfected with either pcDNA3 (Vector) or Fhit in combination with pcDNA3, Gα 16 or Gα 16 QL. One day later, transfectants were subjected to luciferase assay. * Expression of Gα 16 QL significantly induced NFκB stimulation as compared to pcDNA3 control (Dunnett's t test, P < 0.05). down of Fhit by siRNA or overexpression of Fhit in HeLa cells ( Figure 6B), intracellular Ca 2+ mobilization was measured by a FLIPR device with 0.1, 1 or 10 μM histamine as agonist. Figure 6C showed typical Ca 2+ signals induced by 0.1 μM histamine. There was no significant difference among the maximal Ca 2+ responses induced by different concentrations of histamine in control, Fhit-deficient or Fhit-overexpressing cells ( Figure 6D). These observations suggest that Fhit does not affect the Gα q/16 /PLCβ pathway.
Apart from PLCβ, Gα q subunits are known to interact with TPR1 which associates with activated Ras [34,44]. This raises the question whether Fhit could interfere with Gα 16 QL/TPR1/Ras signaling. If Fhit and TPR1 compete for the same region on Gα 16 QL, Fhit will displace and prevent TPR1 from binding to Gα 16 QL. In coimmunoprecipitation assays, the ability of Flag-TPR1 to pull down Gα 16 16 QL could simultaneously bind to both Fhit and Flag-TPR1, thus forming a TPR1/Gα 16 QL/Fhit complex that can be immunoprecipitated by the anti-Flag antibody. The presence of such a complex implies that Fhit may be involved in regulating Gα 16 QL-mediated Ras activation. Ras activation assay was employed to investigate the effect of Fhit on Gα 16 QL-induced Ras activity. In agreement with a previous report [44], Gα 16 QL significantly induced Ras activation as compared to the vector control and wild-type Gα 16 ( Figure 7B). However, there was no significant elevation or attenuation of Ras activity when cells were co-transfected with Fhit ( Figure 7B).
In addition to PLCβ and Ras signaling, other cytoplasmic signaling molecules known to be regulated by Gα q and Gα 16 were examined in the presence or absence of Fhit expression. Phosphorylation states of various signaling molecules including ERK, STAT3 and IKK were examined using phospho-specific antibodies. Gα q RC significantly stimulated the phosphorylations of ERK and IKK and such responses were unaffected by the presence of Fhit ( Figure 8A). Similar results were obtained with Gα 16 QL (data not shown). Likewise, Gα 16 QL significantly stimulated STAT3 phosphorylation at both Tyr 705 and Ser 727 and these responses were not affected by the co-expression of Fhit ( Figure 8B); similar results were obtained with Gα q RC (data not shown).
Since phosphorylation of IKK results in activation of NFκB transcription, Gα 16 QL-stimulated NFκB transcriptional activity was also evaluated. As shown in Figure 8C, Gα 16 QL significantly induced NFκB luciferase activity as compared to pcDNA3 and Gα 16 control. Consistent with the phosphorylation profiles of IKK, expression of Fhit did not affect the Gα 16 QL-stimulated NFκB transcriptional activity.

G q activation enhanced the growth inhibitory effect of Fhit
As Fhit is a tumor suppressor, we asked whether the growth inhibitory effect of Fhit could be affected upon activation of G q -coupled receptors. HEK293 and H1299 cells were chosen for this part of the study because they endogenously express G q -coupled muscarinic M 1 and gastrin-releasing peptide receptors (GRPRs), respectively. We established 293/Fhit cells and H1299/Fhit cells stably expressing Fhit ( Figure 9A). Prolonged stimulation of G q -coupled receptors is often associated with mitogenesis [28], and thus treatment of 293/vector cells with 100 μM carbachol for 4 days or more significantly stimulated cell growth ( Figure 9B). In contrast, carbachol significantly inhibited the growth of 293/Fhit cells ( Figure 9B); it should also be noted that 293/Fhit cells exhibited reduced growth rate as compared to the 293/ vector cells ( Figure 9B). A similar effect was observed in H1299 cells. Bombesin has previously been shown to stimulate the proliferation of non-small lung cancer cells including H1299 cells [45,46]. In the present study, activation of GRPR by 100 nM bombesin for 4 days significantly increased the growth of H1299/vector cells but it suppressed the growth of H1299/Fhit cells ( Figure 9C). These data suggest that mitogenic responses elicited by G q activation are re-directed into growth suppressive signals when the level of Fhit is elevated. This switching of functional outcome is consistent with the notion that the tumor suppressive action of Fhit is correlated to its expression level [47].

Discussion
Receptors coupled to members of the Gα q subfamily mediate a wide range of diverse cellular responses, ranging from cell growth and proliferation to cell differentiation [39]. Established models indicate that the actions of G q -linked receptors are mediated by inositol lipid signaling, but growing evidence suggests that these pathways alone cannot account for all of the responses. Instead, the extensive list of diverse cellular events involving Gα q -linked signals suggests that Gα q subfamily members have multifaceted roles in signal transduction which are not yet fully appreciated. The present study has demonstrated that activated Gα q subunits can directly interact with Fhit, a tumor suppressor widely implicated in many types of cancer [1]. This is especially interesting in view of the ability of Gα q subunits to modulate cell growth and proliferation through regulating critical signaling pathways [48].
The interaction between Gα subunits and Fhit exhibits a high degree of selectivity as demonstrated by the lack of association of Fhit with Gβγ, monomeric GTPases, and RGS proteins. Among the four subfamilies of Gα subunits, at least three can interact with Fhit. Although Gα i2 is often regarded as a representative member of the G i subfamily, its inability to interact with Fhit does not necessarily indicate that the other eight Gα i members cannot be partners of Fhit. Likewise, one cannot exclude the possibility that some specific combinations of Gβγ can interact with Fhit unless all viable permutations have been tested. Since both the wild-type and constitutively active mutants of Gα s and Gα 13 associate with Fhit equally well, such interactions may not be subjected to dynamic cell signaling regulations. Far more interesting is the activation state-dependent interaction between Gα q subunits and Fhit. Activation of Gα q subunits by agonist-bound receptor is expected to drive the formation of Gα q /Fhit complexes. Our data suggest that Fhit can indeed interact with activated Gα q in a native cellular environment ( Figure 2C) and it can directly associate with activated Gα 16 in vitro ( Figure 3B). It is noteworthy that the Gα subunits are attached to the inner leaflet of the plasma membrane through fatty acylation and thus Fhit needs to be present at the plasma membrane in order to interact with Gα subunits productively. Analysis of Fhit protein expression in subcellular fractions of normal rat tissue suggests that it is localized at the plasma membrane and the nucleus [49]. Hence Fhit can be in close proximity to Gα q subunits for efficient interactions.
The inability of Gα 11 to interact with Fhit is rather surprising. The ubiquitously expressed Gα 11 exhibits 90% sequence homology to Gα q and is thus more closely related to Gα q than the primarily hematopoietic Gα 14 and Gα 16 [22], and yet the latter two could interact with Fhit as effectively as Gα q . No report has indicated any major difference between Gα 11 and Gα q both in terms of receptor coupling and effector regulation [39]. The ability of Fhit to distinguish Gα 11 from Gα q as well as Gα 14 and Gα 16 thus represents a unique feature of Fhit, but no immediate clue can be drawn as to why it does not form a complex with Gα 11 .
The use of Gα 16/z chimeras has enabled us to identify the α2-β4 region of Gα 16 as an Fhit-interaction domain (Figure 4). This region has been shown to interact with Gβγ complex in the GDP bound Gα q but it becomes available for effector interaction when Gα q adopts the active GTP-bound conformation (Additional file 2). In different Gα q members, this region associates with various effectors such as p63RhoGEF [50] and PLCβ [51]. The binding of Fhit to the α2-β4 region may thus account for the preference of Fhit for constitutively active Gα q mutants that are dissociated from the Gβγ dimers.
The interaction of Gα with Fhit opens a host of possibilities in terms of their biochemical and cellular consequences. Given the known functions of Gα subunits as signal transducers and that only activated Gα q/14/16 can interact with Fhit, perhaps the most logical prediction is that Fhit acts as an effector of Gα. If this hypothesis is correct, then activated Gα subunits may affect the localization, stability, or function of Fhit. However, there is a lack of effect of Gα 16 QL on the Ap 3 A hydrolase activity of Fhit. Because Fhit binds and hydrolyzes Ap 3 A in vitro [38], any model of Fhit function should take this into account. The ability of GST-Fhit to hydrolyze Ap 3 A into AMP and ADP was, however, unaffected by either GDPβS-or GTPγS-bound His-Gα 16 . Moreover, Fhit mutants with impaired affinity for Ap 3 A (L25W and I10W/ L25W) or a lack of hydrolase activity (H89D) formed complexes with activated Gα q subunits as effectively as wild-type Fhit ( Figure 5A). These results suggest that activated Gα q subunits have little effect, if any, on the enzymatic activity of Fhit. However, it should be noted that because the catalytic mechanism of Fhit requires leaving-group exit and water entry at the substrate-exposed surface of the dimeric enzyme, polypeptides that bind to the Fhit-Ap n A complex are expected to stabilize the complex and retard turnover [6]. Subtle changes in the K m and/or K cat of Ap 3 A hydrolysis by Fhit will require detailed kinetic studies.
Equally disappointing is that the formation of the Gα q / Fhit complex was unable to interfere with any of the  Figure 10 Distinct regulations of Fhit by G q -and EGF-dependent pathways. Agonist binding to G q -coupled receptor leads to Gα q activation and dissociation with Gβγ complex. Activated Gα q can interact with Fhit and stabilize it, which results in increased Fhit level and consequent enhancement of the growth suppressive effect of Fhit. On the other hand, activation of the EGF receptor stimulates Src-mediated phosphorylation of Fhit at the Tyr 114 site. The phosphorylated Fhit undergoes degradation which leads to a decrease in the Fhit protein level as well as the tumor suppressive effect of Fhit. Although activated Gα q also stimulates Src-mediated Fhit Tyr 114 phosphorylation, the overall Fhit protein amount is increased rather than decreased, indicating that either an additional signal is required for the induction of Fhit degradation (which is concomitantly generated by EGF but not by activated Gα q ; indicated as a dashed line) or activated Gα q can up-regulate Fhit via stabilization. known signaling pathways triggered by Gα q . The canonical effector molecules of activated Gα q subunits are the various isoforms of PLCβ. Despite the fact that PLCβ also binds to the Fhit-interacting α2-β4 region of Gα q [51], overexpression of wild-type Fhit or its mutants did not affect Gα q RC-or Gα 16 QL-induced PLCβ activity ( Figure 6A). Activated Gα q may have a higher affinity and preference for PLCβ, resulting in the almost instantaneous formation of IP 3 and mobilization of intracellular Ca 2+ (agonist-induced Ca 2+ mobilization peaks within 10-15 s; Figure 6C). The co-localization of Gα q and PLCβ in lipid rafts [52] helps to ensure the efficiency of the G q /PLCβ pathway. Fhit and other effectors may bind to the activated Gα q when the latter becomes dissociated from PLCβ. In this scenario, Fhit would not be able to compromise PLCβ signaling effectively. However, it should be noted that overexpression of p63RhoGEF can inhibit Gα 16 QL-induced PLCβ activity albeit only partially [53] and the presence of Fhit in lipid rafts remains to be confirmed. Fhit can apparently associate with the Gα 16 QL/TPR1 complex since it is detected in the Gα 16 QL/TPR1 immunoprecipitates but not in the absence of Gα 16 QL ( Figure 7A). The possible existence of an Fhit/Gα 16 QL/TPR1 complex suggests that Fhit binds to Gα 16 QL on a region distinct from that of TPR1, and this is in agreement with our mapping of the Fhitinteraction domain by using the Gα 16/z chimeras ( Figure 4) and the fact that TPR1 interacts with the β3 domain of Gα 16 [36]. The lack of effect of Fhit on Gα 16 QL-induced Ras activation further suggests that co-expression of Fhit would not affect the activities of signaling molecules downstream of Ras. This is indeed true for ERK, STAT3, IKK, and NFκB ( Figure 8).
Although the interaction of activated Gα q and Fhit is independent of the ability of Fhit to become phosphorylated or to bind and hydrolyze Ap 3 A, activation of Gα q could apparently increase Fhit Tyr 114 phosphorylation through Src ( Figure 1B), stabilize Fhit ( Figure 1B and Additional file 1, Additional file 3) and enhance the cell growth inhibition effect of Fhit (Figure 9). G q signals often lead to increased cell growth [28], but by forming a complex with Fhit which can stabilize Fhit, activation of Gα q may result in reduced cell growth (Figure 9). Given that activation of EGF receptors triggers the degradation of Fhit [13], and despite the demonstrated ability of activated Gα q to stimulate Fhit phosphorylation ( Figure 1B-C), it is rather puzzling to observe that activated Gα q can apparently increase the levels of Fhit ( Figure 1B and Additional file 1) and stabilize the truncation mutants of Fhit (Additional file 3). The divergent regulatory outcome of phosphorylated Fhit may be attributed to the differing signaling capacities of EGF-and G qdependent pathways, which could lead to conditional proteasomal degradation of Fhit ( Figure 10). An alternative explanation is that Fhit becomes less susceptible to degradation upon binding activated Gα q , and this might lead to an elevated level of Fhit ( Figure 10). Increased Fhit levels can lead to the suppression of cell proliferation ( Figure 9B and C; [4]), while the knock down of Fhit by siRNA increases the viability of DLD-1 cells [10]. If activation of Gα q can elevate the level of Fhit, this might account for the ability of G qcoupled receptors to inhibit cell proliferation ( Figure 9B and C; [48]). Further investigations are required to elucidate the mechanism by which activated Gα q regulates the level of Fhit. We are currently pursuing the notion that Gα q stimulates the translation of Fhit as we have preliminary data to suggest that the up-regulation of Fhit is blocked by cycloheximide. Since the expression level of Fhit may determine its functional outcome [47], it is tremendously important that quantification of Fhit should be carefully determined in any cellular system to be employed. It should also be noted that Fhit expression can enhance the effects of the p53 tumor suppressor [54] by modulating p53-regulated genes [55]. Hence, the functional relevance of Gα q /Fhit interaction should be revisited in experimental systems with different p53 status.

Conclusions
The present study provides multiple indications that several members of the Gα q family can bind to the tumor suppressor Fhit in their GTP-bound active state. The Fhit-interaction domain on the Gα subunit was identified as the α2-β4 region which would be occluded by the Gβγ dimer in the GDP-bound inactive heterotrimeric G q protein, thus accounting for the preference of Fhit to bind activated forms Gα q subunits. Neither the hydrolase activity of Fhit nor the signaling capacity of activated Gα q was affected by the formation of activated Gα q /Fhit complexes. In cells with elevated levels of Fhit, activation of G q -coupled receptors led to growth suppression rather than stimulation. Consistent with the tumor suppressive function of Fhit, these observations suggest that the formation of Gα q /Fhit complex may modulate cell proliferation.

Reagents
Human cDNAs of various Gα subunits were obtained from Guthrie Research Institute (Sayre, PA). Wild-type Fhit in pCMV-SPORT6 was purchased from Invitrogen (Carlsbad, CA). pRcCMV-Fhit Y114F was a generous gift from Dr. K. Huebner (Comprehensive Cancer Center and Department of Molecular Virology, Immunology, and Medical Genetics, Ohio State University). L25W, I10W/L25W, and H96D mutants of Fhit were kindly provided by Dr. C. Brenner (Department of Genetics and Biochemistry, Dartmouth Medical School). Cell culture reagents, including LipofectAMINE PLUS reagents were purchased from Invitrogen (Carlsbad, CA).Anti-Gα 16 and anti-Gα 14 were obtained from Gramsch Laboratories (Schwabhausen, Germany). Anti-Fhit antibody was from Invitrogen (Carlsbad, CA). Anti-G q/11 α-subunit antibody was purchased from Calbiochem (San Diego, CA). Anti-αtubulin antibody, anti-HA antibody, anti-Flag antibody and anti-Flag affinity gel were from Sigma-Aldrich (St. Louis, MO). Antisera against Gα s , Gα i2 and Gα 13 were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Anti-GST antibody was from Abcam (Cambridge, UK). Anti-phospho-Fhit-Tyr 114 antibody was raised in rabbits against a synthetic peptide corresponding to AA 106-122 of human Fhit containing the phosphorylated tyrosine residue and an additional N-terminally cysteine residue for coupling (C-DFHRNDSI(pY)EE LQKHDK). Antibodies were affinity-purified using the immunizing phospho-peptide coupled to SulfoLink® Agarose beads from Thermo Scientific (Rockford, IL) and subsequently cross-absorbed against the non-phosphorylated peptide. Specificity of antibodies was verified by Western blot using cell lysates prepared from HEK293 cells transiently transfected with cDNAs of Fhit or Fhit and Src. Other antibodies were purchased from Cell Signaling Technology (Danvers, MA). GDPβS and GTPγS were from Calbiochem (San Diego, CA). Protein G-agarose and dithiobis[succinimidylpropionate] (DSP) cross-linker were from Pierce Biotechnology (Rockford, IL). ECL kit and Glutathione Sepharose TM 4 Fast Flow beads were from Amersham Biosciences (Piscataway, NJ). Ni-NTA Agarose was obtained from Qiagen (Valencia, CA). Ras activation kit was a product of Upstate-Millipore (Billerica, MA).

Construction of G protein chimeras and truncation mutants of Fhit
The Gα chimeras (except C128) were constructed as described previously [36] by PCR method using human Gα 16 and Gα z cDNAs. Briefly, the N-terminal 188, 210, 246 and 266 amino acids or the C-terminal 128 and 164 amino acids of Gα 16 were swapped to the corresponding regions of Gα z to generate N188, N210, N246, N266, C128 and C164. Primers were designed to cover the overlapping regions of the chimeras, so that 5′ and 3′ fragments can be annealed together to obtain the full length chimeras by PCR. Then the full length PCR products were subcloned into the pcDNA3 vector. All chimeras were confirmed by dideoxynucleotide sequencing. Primer sequence for constructing C128 is 5′-GTG CCT GGA GGA GAA CAA CCA GAC AAG TCG GAT GGC AG-3′.
Flag-tagged Fhit truncation mutants, F131N, F95N, F50C and F27C, were constructed by PCR method using the human Fhit cDNA as a template. The primers were designed based on the secondary structure of Fhit. The outer forward and reverse primers of Fhit are 5′-CGA  AGC TTA TGG ACT ACA AAG ACG ATG ACG ACA  AGT CGT TCA GAT TTG GCC AAC ATC TC-3′ and 5′-CCT CGA GTC ACT GAA AGT AGA CCC GCA GAG CTG C-3′, respectively. The reverse primers of F131N and F95N are 5′-CCT CGA GTC ATG ATC TCC AAG AGG CAG GAA AGT C-3′ and 5′-CCT CGA GTC AAA CGT GCT TCA CAG TCT GTC CGG C-3′, respectively. The forward primers of F50C and F27C are 5′-CGA AGC TTA TGG ACT ACA AAG ACG ATG ACG ACA AGC TGC GTC CTG ATG AAG TGG CCG-3′ and 5′-CGA AGC TTA TGG ACT ACA AAG ACG ATG ACG ACA AGA ATA GGA AAC CTG TGG TAC CAG GAC-3′, respectively. All truncation mutants were confirmed by dideoxynucleotide sequencing.
For co-immunoprecipitation experiments, HEK293 cells were grown to 80% confluency in 100 mm tissue culture plates and then co-transfected with various combinations of cDNAs (3 μg/plate) using 15 μL PLUS and LipofectAMINE reagents in MEM. Serum was replenished 3 h after transfection. Cross-linking was performed one day after transfection; transfected HEK293 cells were washed with PBS twice and then treated with 0.5 mM DSP in PBS for 10 min at room temperature. Cells were then washed with PBS twice and maintained in quenching solution containing 50 mM glycine in PBS, pH 7.4, for 5 min. Cells were subsequently lysed in ice-cold RIPA buffer (25 mM HEPES at pH 7.4, 0.1% SDS, 1% Nonidet P-40, 0.5% sodium deoxycholate, 1 mM dithiothreitol, 200 μM Na 3 VO 4 , 4 μg/mL aprotinin, 100 μM phenylmethylsulfonyl fluoride, and 2 μg/mL leupeptin). Cell lysates were gently rocked with a primary antiserum at 4°C overnight, and then incubated in 30 μL protein G-agarose (50% slurry) at 4°C for 2 h. Alternatively, the cell lysates were incubated in 30 μL anti-Flag affinity agarose gel (50% slurry) at 4°C for 4 h. Immunoprecipitates were washed with icecold RIPA buffer (400 μL) for four times, resuspended in 50 μl RIPA buffer and 10 μl 6× sample buffer and then boiled for 5 min. Target proteins in the immunoprecipitates were analyzed by Western blots. Signal intensities of the immunoreactive bands were quantified using Image J software, version 1.38x (National Institutes of Health, USA).

Expression and purification of recombinant Gα 16 and Fhit proteins, and GST pull-down
Fhit and Gα 16 were subcloned into pGEX-4 T-1 and pET21a(+) expression vectors, respectively, and transformed into E. coli BL21 strain. 750 ml bacterial cultures were grown at 37°C until the OD 600 reached 0.6-0.8. The cultures were cooled down at 4°C for 20 min and 0.2 mM IPTG was added. The cultures were then grown at 18°C overnight (for GST-Fhit) or 30°C for 15 h (for His-Gα 16 ).
Cells were harvested by centrifugation for 15 min at 6,000 rpm and resuspended in 30 ml ice-cold lysis buffer for GST-tagged Fhit (50 mM Tris, pH 7.5, 500 mM NaCl, 2 mM EDTA, 1 mM dithiothreitol, 1 mM phenylmethylsufonyl fluoride, 2 μg/ml leupeptin) and lysed by three rounds of sonication. After addition of Triton X-100 to a final concentration of 1%, the lysate was incubated at 4°C for 10 min. Cell debris was removed by centrifugation at 18,000 rpm for 20 min. The cleared supernatant was then incubated with Glutathione Sepharose™ 4 Fast Flow beads at 4°C for 1.5 h with gentle rotation. The beads were spun down at 4,000 rpm for 1 min and washed four times with wash buffer (lysis buffer with 150 mM NaCl and 10% glycerol). The beads were then loaded into a chromatography column and GST-Fhit was eluted washing buffer containing 20 mM glutathione. Similar procedure was used for the purification of His-tagged Gα 16 except that Ni-NTA Agarose and a different lysis buffer was employed (PBS, pH 8.0, 300 mM NaCl, 10 mM imidazole, 1 mM dithiothreitol, 1 mM phenylmethylsufonyl fluoride, 2 μg/ml leupeptin and 20 mM 2-mercaptoethanol). His-Gα 16 was eluted in washing buffer containing a discontinuous gradient of imidazole (from 30 mM to 250 mM). Proteins eluted at fractions 6 and 7 were pulled. Purified GST or GST-Fhit were mixed with Gα 16 (2 μg each) in 500 μL pull-down buffer (50 mM Tris-HCl, pH 7.5, 100 mM NaCl, 40 mM NaP 2 O 7 and 5 mM MgCl 2 ) in combination with 1 μM GDPβS or GTPγS, and then the mixture was incubated at 4°C for 30 min. Glutathione sepharose was then added and the mixture was further incubated at 4°C for 2 h. After being washed with pull-down buffer twice, the beads were resuspended in sample buffer and subjected to Western blot analysis.
Assay for diadenosine triphosphate hydrolysis by recombinant Fhit 100 μM of Ap 3 A was incubated with or without recombinant GST-Fhit protein or GST protein in 50 mM HEPES-NaOH, pH 6.8, containing 0.5 mM MnCl 2 for 10 min at 37°C in a total volume of 100 μl. Reactions were stopped by heat inactivation (95°C, 10 min). 50 μl of nucleotide standards and assay solutions were then analyzed by HPLC with a Mono Q column, eluted with a gradient from 50 to 600 mM ammonium bicarbonate, pH 8.5, at a flow rate of 1 ml/min. Absorbance of nucleotides were detected at 254 nm. For reactions that required His-Gα 16 incubations, 0.5 μg His-Gα 16 was pre-incubated with either GDPβS or GTPγS (100 μM each) at 30°C for 30 min in GTP binding activation buffer (50 mM Hepes, pH 8, 10 mM MgCl 2 , 1 mM dithiothreitol, 1 mM EDTA, and 100 mM NaCl) prior to incubation with Fhit/Ap 3 A for 10 min. The extent of Ap 3 A hydrolysis by 0.5 μg GST-Fhit was measured in the absence or presence of His-Gα 16 , and was expressed as percentage of Ap 3 A hydrolyzed during the reaction based on the areas under the peaks of Ap 3 A before and after the hydrolysis reaction [38].

Ras activation assay
HEK293 cells were co-transfected with 200 ng Gα, 200 ng Flag-Fhit and 100 ng Ras cDNAs. After 1 day, transfectants were serum starved for 4 h. Cells were then washed twice with ice-cold PBS and lysed with the Mg 2+ lysis buffer (MLB; 125 mM HEPES at pH 7.5, 750 mM NaCl, 5% Nonidet P-40, 50 mM MgCl 2 , 5 mM EDTA, 10% glycerol, and appropriate protease inhibitors). Clarified cell lysates were immunoprecipitated with 20 μL Raf-1 RBD agarose for 45 min and subsequently washed three times with 400 μL ice-cold MLB. Eluted protein samples in 50 μL MLB and 10 μL 6× sampling dye were then resolved in SDS gels and analyzed using specific anti-Ras antibody.
Inositol phosphates accumulation assay HEK293 cells were seeded on a 12-well plate at 2 × 10 5 cells/well one day prior to transfection. Various cDNAs at a concentration of 0.5 μg/well were transiently transfected into the cells using Lipofectamine PLUS® reagents. One day after transfection, cells were labeled with inositol-free Dubecco's modified Eagle's medium (DMEM; 750 μL) containing 5% FBS and 2.5 μCi/mL myo-[ 3 H]inositol overnight. The labeled cells were then washed once with IP 3 assay medium (20 mM HEPES, 5 mM LiCl, serum-free DMEM) and then incubated with 500 μl IP 3 assay medium at 37°C for 1 h. Reactions were stopped by replacing the assay medium with 750 μL ice-cold 20 mM formic acid and the lysates were kept in 4°C for 30 min before the separation of [ 3 H]inositol phosphates from other labeled species by sequential ion-exchange chromatography as described previously [56].

Western blotting analysis
Protein samples were resolved on 12% SDS-polyacrylamide gels and transferred to Osmonics nitrocellulose membrane. Resolved proteins were detected by their specific primary antibodies and horseradish peroxidase-conjugated secondary antisera. The immunoblots were visualized by chemiluminescence with the ECL kit from Amersham, and the images detected in X-ray films were quantified by densitometric scanning using the Eagle Eye II still video system (Stratagene, La Jolla, CA, USA).

Measurement of intracellular Ca 2+ by FLIPR
The intracellular Ca 2+ was measured by using an optimized Fluorometric Imaging Plate Reader (FLIPR) protocol [58]. HeLa cells were seeded into clear-bottomed black-walled 96-well plates. The growth medium was replaced by 200 μL labeling medium containing 1:1 (v/v) ATCC-MEM medium: Hank's balanced salt solution, 2.5% (v/v) fetal calf serum, 20 mmol/L HEPES, pH 7.4, 2.5 mmol/L probenecid and 2 μmol/L Fluo-4 AM. Histamine was prepared as a 5× solution in Hank's balanced salt solution into another polypropylene 96-well plate. After 1 h labeling, cell and drug plates were placed in a FLIPR (Molecular Devices, Sunnyvale, CA, USA). Immediately after the addition of 50 μL of drug solution into the cell medium, changes in fluorescence were monitored over 120 s following excitation at a wavelength of 488 nm and detection at 510-560 nm.