Association of activated Gα_q to the tumor suppressor Fhit is enhanced by phospholipase Cβ

The fragile FHIT gene at the chromosomal fragile site FRA3B is often regarded as an early target of DNA damage in precancerous cells. Its gene product, the ubiquitously expressed Fhit (Fragile Histidine Triad) protein, is a member of the HIT (histidine triad) superfamily with three signature histidines in the conserved nucleoside binding motif. Fhit binds and hydrolyzes various dinucleoside polyphosphates (such as Ap₃A, Ap₄A, Ap₃G and Cp₃G) into two nucleotides where one is a nucleoside monophosphate [1]. The preferred substrate of Fhit is Ap₃A (diadenosine 5′,5‴-P1,P3-triphosphate) which is hydrolyzed to AMP and ADP. Interestingly, Fhit acts as a tumor suppressor and its down-regulation is associated with different tumors including lung cancers [2]. Re-expression of Fhit in Fhit-deficient tumor cells can notably suppress tumor development [3–5]. Several theories of tumor suppression have been proposed for Fhit with the overarching idea of Fhit acting as a genome “caretaker” [6, 7]. Reduced Fhit expression has indeed been shown to increase DNA replication stress and genome alterations as a result of a decreased intracellular thymidine triphosphate (dTTP) level [8]. Moreover, Fhit is apparently involved in suppressing lung tumor cell migration/invasion by down-regulating the expression of matrix metalloproteinase 2/9 [9]. Despite considerable efforts, the precise mechanism by which Fhit exerts its tumor suppressive function remains elusive. The dinucleoside polyphosphate hydrolase activity of Fhit seemingly plays a trivial role in tumor suppression [10].

A number of studies have revealed unsuspecting binding partners of Fhit that may provide linkages to processes that contribute to tumor eradication such as cellular oxidation and apoptosis. Fhit-interacting molecules include β-catenin [11], ferredoxin reductase [12], Src tyrosine kinase [13] and ubiquitin conjugating enzyme 9 [14]. More recently, we have demonstrated that Fhit can distinguish between inactive and active signal transducing Gα subunits of the G_q family [15]. This finding is intriguing as it may link Fhit to Gα_q-dependent signals that modulate a variety of cellular events. Fhit-mediated suppression of epithelial-mesenchymal transition in bronchial cells involves the epidermal growth factor receptor (EGFR), Src, and extracellular signal-regulated kinase (ERK) [16] that have all been shown to be activated or transactivated by Gα_q [17, 18]. In human colon cancer cell lines, Fhit inhibits cell proliferation by attenuating the nuclear factor κB (NFκB) pathway [19] which can be stimulated by G_q-coupled receptors [20]. It is also noteworthy that sustained activation of the G_q pathway often leads to mitogenesis in a variety of cell types [21] with disparate mechanisms of regulating cell cycle progression [22]. The opposing roles of Fhit and Gα_q tend to suggest that they may exert counteracting actions on each other. However, Fhit neither inhibits nor enhances Gα_q-induced signals [15] whereas its own expression becomes translationally up-regulated by activated Gα_q [23]. Given the links between G_q signaling and mitogenesis as well as those between Fhit and tumor suppression are well-established, it seems reasonable to expect that the binding of Fhit to activated Gα_q would impart functional consequences. Since the canonical signaling pathway of all Gα_q subfamily members (Gα_q, Gα₁₁, Gα₁₄ and Gα_15/16) is the activation of phospholipase Cβ (PLCβ), we further explored the influence of PLCβ on the formation of Fhit/activated Gα_q complexes. Here, we report that different isoforms of PLCβ can also associate with Fhit in the absence of Gα_q activation.

We have previously shown that Fhit directly interacts with activated members of the Gα_q family (Gα_q, Gα₁₄, and Gα₁₆) via their α2-β4 region without affecting Gα_q-induced PLCβ activation [15]. As PLCβ also interacts with the α2 region of the activated Gα_q [25], we asked whether PLCβ can compete with Fhit for the activated Gα_q in co-immunoprecipitation assays. HEK293 cells were co-transfected with Flag-tagged Fhit and wild-type or activated Gα₁₆ (Gα₁₆QL) with or without PLCβ1, PLCβ2 or PLCβ3. Because activated Gα_q signaling always increase the expression levels of Fhit [23], we adjusted the Fhit cDNA amount for transfection to obtain similar Fhit expression levels. In order to facilitate the assessment of expression and to minimize interference by endogenous Gα_q subunits, we have opted for using Gα₁₆ as a representative Gα_q member. In vector transfected cells, Fhit pulled down detectably more Gα₁₆QL than wild-type Gα₁₆ (Fig. 1), as reported previously [15]. Overexpression of PLCβ1, PLCβ2 or PLCβ3 increased the affinities of both wild-type and activated Gα₁₆ for Fhit (cf lanes 1, 2 and lanes 4, 5 in the three panels of the second row of Fig. 1). Thus PLCβs did not appear to compete with Fhit for activated Gα₁₆. Instead, the presence of PLCβs apparently enhanced or stabilized the association of Fhit and Gα₁₆. More interestingly, all three isoforms of PLCβ were also detected in the Fhit-immunoprecipitates (Fig. 1). To test if Fhit can form complexes with PLCβs, HEK293 cells were transfected with vector or PLCβ1-3 in combination with Flag-Fhit or Flag vector. All three isoforms of PLCβ were able to co-immunoprecipitate with Fhit (Fig. 2a). Since HEK293 cells endogenously express PLCβ1 (Fig. 2a, upper left panel), Flag-Fhit might be able to pull down endogenous PLCβ1. Longer exposure of the anti-PLCβ1 blot indeed revealed the presence of endogenous PLCβ1 in the immunoprecipitates of Fhit (Fig. 2b). Moreover, as compared to the Flag vector control, Fhit could pull down endogenous wild-type Gα_q when one of the PLCβ isoforms was overexpressed (cf lane 1 and lanes 3, 5, and 7 of the fourth row on the right in Fig. 2a). These findings indicate that PLCβs may interact with Fhit and increase the association between Fhit and Gα_q. We have previously reported that increased cell proliferation by Gα_q activation is suppressed in the presence of Fhit [15]. Not surprisingly, we also found that PLCβ3 overexpression alone was sufficient to trigger a higher cell growth rate and Fhit co-expression significantly decreased PLCβ3-induced cell proliferation (data not shown). Therefore, Fhit appears to be capable of suppressing G_q-PLCβ mediated cell proliferation.

If PLCβ binds to Fhit via activated Gα_q, other proteins known to associate with Gα_q may also co-immunoprecipitate with Fhit. Besides PLCβs, activated Gα_q also interacts with other proteins such as RGS2 [26] and p63RhoGEF [27]. Unlike PLCβs, Fhit did not interact with HA-tagged RGS2 in the co-immunoprecipitation assay using anti-Flag or anti-HA affinity gel (Fig. 3a). Similarly, in transfected HEK293 cells expressing various combinations of Flag-Fhit, myc-tagged p63RhoGEF, Gα_q and constitutively active Gα_qRC, myc-p63RhoGEF pulled down Gα_qRC but not Fhit (Fig. 3b). Moreover, the expression of Fhit did not affect the interaction between Gα_qRC and p63RhoGEF (Fig. 3b). These results suggest that the ability of PLCβs to associate with Fhit and increase the interaction between Fhit and Gα_q may be specific.

Since overexpression of PLCβs appeared to enhance the interaction between Fhit and Gα₁₆ (Fig. 1), we assessed whether Fhit can reciprocally enhance the interaction of PLCβ with Gα_q members. We performed co-immunoprecipitation assay with anti-PLCβ3 antiserum because it has the best specificity among the different anti-PLCβ antisera. Gα₁₆QL as well as Fhit was co-immunoprecipitated with PLCβ3 (Fig. 4). Co-expression of Fhit appeared to weaken the interaction between Gα₁₆QL and PLCβ3 (Fig. 4 cf lanes 2 and 4 of row two). The reduction of Gα₁₆QL in the PLCβ3-immunoprecipitates was not due to variations in the expression levels or pull down efficiency, as these parameters were essentially similar in the different samples (Fig. 4 rows one and five).

By using a series of Gα_16/z chimeras with discrete regions of Gα₁₆ replaced by those of Gα_z (a Gα_i subfamily member which does not interact with Fhit), we have previously identified the α2-β4 region of Gα₁₆ as critical for Fhit interaction [15]. Interestingly, the α2 region of Gα_q is seemingly involved in binding to PLCβ [25]. To further investigate the associations among Gα_q, PLCβ and Fhit, we constructed two new chimeras named zα2β4 (Gα₁₆ backbone with α2β4 region from Gα_z) and 16α2β4 (Gα_z backbone with α2β4 region from Gα₁₆), wherein the α2-β4 region of Gα₁₆ or Gα_z was swapped with each other (Fig. 5a). As shown in Fig. 5b, both wild-type and constitutively active mutants of the zα2β4 and 16α2β4 chimeras can be expressed in HEK293 cells to levels that were comparable to those of Gα₁₆, Gα_z, or C128 (a previously characterized chimera with the C-terminal 128 residues of Gα₁₆ swapped with the corresponding sequences of Gα_z). Because the Gα-specific antibodies are N-terminal targeting, the zα2β4 and 16α2β4 chimeras were recognized by anti-Gα₁₆ and anti-Gα_z antisera, respectively. When examined for their ability to stimulate PLCβ, activated zα2β4QL efficiently stimulated the formation of inositol phosphates to an extent similar to that of Gα₁₆QL (Fig. 5b). The constitutively active Gα_zQL did not activate PLCβ because Gα_z belongs to the Gα_i subfamily (Fig. 5b). Since neither 16α2β4QL nor C128QL was able to stimulate PLCβ (Fig. 5b), it indicated that the α2-β4 region of Gα₁₆ alone was not sufficient to stimulate PLCβ.

A hallmark of Fhit/Gα_q interaction is the enhanced association with the activated Gα_q subunits over their wild-type counterparts [15]. Activation state-dependent interaction with Fhit was reproducibly observed in the present study (Fig. 1 and Additional file 1: Figure S1) and this feature was therefore used as an indicator of Fhit association with the chimeras. In co-immunoprecipitation assays, both the wild-type and constitutively active mutant of zα2β4 were pulled down by Flag-Fhit to similar extents (Fig. 5c), despite the fact that zα2β4QL was fully capable of stimulating PLCβ (Fig. 5b). Thus the α2β4 region of activated Gα₁₆ is required by Fhit association but not by PLCβ activation. In the control groups, more Gα₁₆QL was detected in the Fhit-immunoprecipitates than wild-type Gα₁₆ whereas both Gα_z and Gα_zQL were hardly detected (Fig. 5c). These control groups produced the same results as in our previous report [15]. The presence of 16α2β4 or 16α2β4QL in the Fhit-immunoprecipitates was even weaker than those Gα_z (Fig. 5c), suggesting that this chimera could not be recognized by Fhit. It should be noted that activation state-dependent association with Fhit has previously been demonstrated for the C128 chimera [15], while C128QL was not able to stimulate PLCβ (Fig. 5b). According to the above results of zα2β4, 16α2β4 and C128, the capability of activated Gα₁₆QL to associate with Fhit did not affect its ability to stimulate PLCβ and it may explain why PLCβ enhanced the association of Fhit and Gα_q instead of competing with Fhit for Gα_q,

The diversity of pathways downstream of G_q has endowed mammalian cells with a complex signaling network for the delicate regulation of a multitude of biological effects, with some responses being cell type-specific. One example is the fact that activation of G_q leads to proliferation in some cells while it induces apoptosis in other cell types [22]. The tumor suppressor Fhit taps into the G_q signaling network through its ability to associate with activated Gα_q [15, 23]. Interestingly, Fhit suppresses G_q-mediated cell growth in H1299 lung cancer cells via an unknown mechanism [15, 23]. Here, we showed that the canonical Gα_q effector PLCβ can form a complex with activated Gα_q and Fhit, and can increase the overall association of the latter two proteins. Many tripartite or even higher order complexes involving Gα_q subunits are known to exist [28]. For instance, Gα_q can simultaneously bind to p63RhoGEF and RhoA [27], to G protein-coupled receptor kinase 2 (GRK2) and Gβγ [29], as well as to ADP-ribosylation factor 6 (ARF6) and ARNO (a GEF for ARF6) [30]. Hence, the existence of a Gα_q/PLCβ/Fhit complex seems plausible.

Given the prior demonstration that direct binding of Fhit to activated Gα_q does not affect PLCβ activity [15], the ability of PLCβ to form a complex with Fhit/Gα_q is rather surprising. All three PLCβ isoforms (PLCβ1, 2 and 3) tested as well as the endogenously expressed PLCβ1 could be detected in the immunoprecipitates of Fhit in the absence of Gα_q overexpression (Fig. 2). Therefore, it is possible that PLCβ can directly interact with Fhit. Although the PLCβ association with Fhit may also occur via binding to endogenous Gα_q subunits, the interaction between PLCβ and Fhit appears to be specific because other Gα_q effectors such as RGS2 and p63RhoGEF did not interact with Fhit even in the presence of activated Gα_q (Fig. 3). PLCβs have weak affinities for inactive Gα_q subunits (EC₅₀ at ~10 μM range) [31] but this basal interaction between PLCβ3 and inactive Gα_q could be detected (Fig. 4). The increased affinity of Fhit with inactive Gα_q upon overexpression of PLCβs (Figs. 1 and 2b) may result from the basal interactions between PLCβs and inactive Gα_q and the interaction between PLCβ and Fhit. It remains to be demonstrated if PLCβ can directly interact with Fhit.

The α2-β4 region of Gα_q is essential for the binding of Fhit [15] and other regions may also be required (Fig. 5b). According to the structure of Gα_q-PLCβ3 complex, PLCβ3 interacts with the α2 and α3 region of Gα_q by a helix-turn-helix domain [25] and a similar interaction domain is also present in the Gα_q/p63RhoGEF complex [27]. Different from PLCβ, p63RhoGEF could not interact with Fhit irrespective of whether activated Gα_q was present (Fig. 3b). This tends to suggest that when activated Gα_q is bound to an effector other than PLCβ, Fhit is precluded from interacting with Gα_q. The complete lack of evidence on Fhit and PLCβ competing for activated Gα_q further supports the existence of a tripartite complex of Fhit/Gα_q/PLCβ. A prerequisite for the simultaneous binding of Fhit and PLCβ to activated Gα_q is that the two molecules should not use identical regions on Gα_q for interaction. This notion is indirectly supported by the results pertaining to the zα2β4QL chimera, which stimulated PLCβ activity as robustly as Gα₁₆QL (Fig. 5b) but failed to interact with Fhit beyond that of wild-type zα2β4 (Fig. 5c). It would appear that the conformation of activated zα2β4QL can be recognized by PLCβ but not by Fhit. Although PLCβ has additional contact points (e.g., the α3-β5 region) on Gα_q [25], the lack of detrimental effect upon the replacement of the α2-β4 region in the zα2β4QL chimera is rather intriguing and warrant some discussion. In the α2-β4 region of Gα_q, nine of the ten residues (Q209-K215, H218, C219 and E221) for PLCβ3 interaction are the same with members of Gα_i subfamily while the exceptional residue on Gα_i subunits corresponding to R214 of Gα_q is identical to that of Gα₁₆ [25]. Hence, it is possible that substitution of α2-β4 region of Gα₁₆ (a Gα_q member) with the corresponding region of Gα_z (a Gα_i subfamily member) would still allow the zα2β4QL chimera to interact productively with PLCβ (Fig. 5b). In contrast, the mere presence of the α2-β4 region of Gα₁₆ in the 16α2β4QL chimera was insufficient to support efficient interaction with PLCβ or Fhit. Collectively, these results indicate that the substitution-induced conformational changes on the binding interface of Gα_q are tolerated by PLCβ but not by Fhit.

PLCβ is a key molecule in transducing activated G_q protein signal to its downstream signal pathways, and similar to G_q protein, it also plays complicated roles in regulating cell growth. PLCβ2 expression level is positively correlated with breast cancer [32] and it promotes mitosis and migration of breast tumor cells [33]. On the other hand, in human erythroleukemia cells, PLCβ1 suppresses proliferation probably through regulating cyclin D3 [34]. PLCβ3-deficiency in mice leads to lymphoma and other tumors [35]. The interaction between PLCβs and Fhit as well as the complex formation among Fhit, Gα_q and PLCβ reveals new pathway(s) of cell growth inhibition by Gα_q and PLCβ.

Activated Gα_q binds to Fhit through the α2-β4 region of Gα_q (Fig. 6a and [15]), and it binds to PLCβ through multiple regions of Gα_q including the α2 region (Fig. 6b and [25]). The overlapping PLCβ and Fhit binding domain on the activated Gα_q is the α2 region (Fig. 6c). As substitution of α2-β4 region of Gα₁₆ with the corresponding region of Gα_z did not affect the activated Gα₁₆-induced PLCβ activation (Fig. 5b), Fhit may bind to this overlapped interacting region on Gα_q and change the binding interface between Gα_q and PLCβ without affecting the PLCβ activity. The altered binding interfaces of activated Gα_q and PLCβ may trigger Gα_q and PLCβ to form a ‘clamp’ around Fhit. Beside protein interactions that were found between two protein pairs among Gα_q, PLCβ and Fhit, PLCβs increased the interaction between Gα_q and Fhit. But Fhit or Gα_q did not enhance the interaction between PLCβ and Gα_q or Fhit, respectively (Fig. 6d). One possibility for increased affinities of Fhit to Gα_q by PLCβ is that PLCβ may interact with and stabilize the complex of Fhit and Gα_q. Another possibility is that when binding to Gα_q, PLCβ may provide direct binding sites on itself for Fhit which also leads to the formation of a heterotrimeric protein complex. In both possibilities, the activated Gα_q recruits PLCβ which acts as a positive regulator for the association of Fhit with activated Gα_q. In the future, the involvement of PLCβ in the regulation of Fhit by Gα_q and their possible roles on cancer therapy should be demonstrated.

We showed that PLCβ could interact with Fhit, and the expression of PLCβ increased the interaction between Gα_q and Fhit. This regulatory effect appears to be unique to PLCβ because other Gα_q effectors such as RGS2 and p63RhoGEF could not interact with Fhit. Substitution of the α2-β4 region of Gα_q with Gα_i did not affect Gα_q-induced PLCβ activation but eliminated the interaction between Gα_q with Fhit. This new Gα_q/PLCβ/Fhit signaling complex represents a novel pathway of Gα_q regulation on tumor suppression.