RNA-induced silencing attenuates G protein–mediated calcium signals

Abstract

Phospholipase Cβ (PLCβ) is activated by G protein subunits in response to environmental stimuli to increase intracellular calcium. In cells, a significant portion of PLCβ is cytosolic, where it binds a protein complex required for efficient RNA-induced silencing called C3PO (component 3 promoter of RISC). Binding between C3PO and PLCβ raises the possibility that RNA silencing activity can affect the ability of PLCβ to mediate calcium signals. By use of human and rat neuronal cell lines (SK-N-SH and PC12), we show that overexpression of one of the main components of C3PO diminishes Ca²⁺ release in response to Gα_q/PLCβ stimulation by 30 to 40%. In untransfected SK-N-SH or PC12 cells, the introduction of siRNA(GAPDH) [small interfering RNA(glyceraldehyde 3-phosphate dehydrogenase)] reduces PLCβ-mediated calcium signals by ∼30%, but addition of siRNA(Hsp90) (heat shock protein 90) had little effect. Fluorescence imaging studies suggest an increase in PLCβ-C3PO association in cells treated with siRNA(GAPDH) but not siRNA(Hsp90). Taken together, our studies raise the possibility that Ca²⁺ responses to extracellular stimuli can be modulated by components of the RNA silencing machinery.—Philip, F., Sahu, S., Golebiewska, U., Scarlata, S. RNA-induced silencing attenuates G protein–mediated calcium signals.

G protein signaling is one of the most common ways cells respond to environmental cues. There are 4 families of mammalian G proteins. Our lab has focused on the Gα_q family, which is coupled to receptors for dopamine, bradykinin, acetylcholine, and angiotensin II as well as other hormones and neurotransmitters (1). Binding of these agents to their specific receptors allows Gα_q to activate its main effector, phospholipase Cβ (PLCβ). PLCβ is a soluble enzyme that binds strongly to membranes, where it catalyzes the hydrolysis of the signaling lipid phosphatidylinositol 4,5-bisphosphate (PIP₂), leading to release of the second messengers inositol 1,4,5-trisphosphate (IP₃) and diacylglycerol (2, 3). IP₃ diffuses to the endoplasmic reticulum, where it opens Ca²⁺ channels, resulting in an increase in intracellular Ca²⁺ and modulation of the conformation and activity of a variety of calcium-sensitive proteins. As the ligand-bound receptors are internalized and Gα_q becomes deactivated, the cellular Ca²⁺ concentration gradually returns to basal levels.

In conjunction with G protein signaling, cells have other mechanisms that allow them to quickly respond and adapt to environmental changes. One prominent mechanism is the down-regulation of gene expression by RNA silencing allowing cells to rapidly adjust the expression levels of particular proteins (4, 5). In this process, small duplex RNAs bind to the RNA-induced silencing complex (RISC) in association with a promoter called C3PO (component 3 promoter of RISC) (6). C3PO is an asymmetric octamer consisting of 2 subunits of the nuclease translin-associated factor X (TRAX) and 6 subunits of the nucleic acid binding protein translin (7). C3PO helps to degrade the passenger strand of the silencing small interfering RNA (siRNA), leaving the guide strand free to hybridize to its target mRNA, which is then degraded by Argonaut 2 of the RISC.

In cultured cells, the majority of PLCβ is localized on the plasma membrane in association with Gα_q (8). However, there is also a stable population in the cytoplasm that is constant throughout the Gα_q activation cycle. In a previous yeast 2-hybrid study, we found that PLCβ associated with the TRAX subunits of C3PO (9); we subsequently showed that PLCβ binds C3PO in the cytoplasm of cells (10). Solution studies using purified proteins have shown that the binding site of TRAX and Gα_q on PLCβ overlap (9). This overlap allows TRAX to directly compete with Gα_q for PLCβ binding and reduces the ability of Gα_q to activate the enzyme (9). Interestingly, our studies demonstrated that in cultured cells, overproduction of PLCβ inhibits C3PO activity and reverses RNA silencing by specific siRNAs (10).

It has been known that protein synthesis is linked to changes in intracellular Ca²⁺ in an indirect way. For example, some transcription factors are modulated by increased Ca²⁺ whose cellular levels can be mediated by PLCβ activity (11). These transcription factors can in turn affect G protein regulation by influencing the production of proteins involved in the signaling cascade. In this study, we tested the possibility that RNA silencing activity directly affects Gα_q-mediated Ca²⁺ signals through changes in the relative association of PLCβ/Gα_q and PLCβ/C3PO. We found that transfection with specific siRNAs changes the population of cytosolic PLCβ/C3PO and in turn affects Ca²⁺ signals mediated by PLCβ. Our studies demonstrated that these 2 separate cellular responses to environmental and adaptive cues are directly linked.

MATERIALS AND METHODS

Materials

SK-N-SH cells were the generous gift of L. Devi (Mt. Sinai Medical Center, New York, NY, USA). SK-N-SH cells were cultured in DMEM supplemented with 10% fetal bovine serum and 1% penicillin–streptomycin. PC12 cells were cultured in DMEM supplemented with 10% horse serum, 5% fetal bovine serum, and 1% penicillin–streptomycin. All the cells were maintained in the incubator at 37°C and 5% CO₂. siRNAs were purchased from Dharmacon (Lafayette, CO, USA) and were a combination of 4 different siRNAs targeting different regions of mRNA for efficient knockdown of genes. The morpholino designed to target HCN2 channels (TTGGTCCTCTCCCTGCCCCTCACCT) was the generous gift of P. Brink (Stony Brook University, Stony Brook, NY, USA).

Intracellular calcium measurement

SK-N-SH and PC12 cells at 80 to 90% confluence were washed twice with HBSS (Invitrogen, Carlsbad, CA, USA) containing Ca²⁺ and Mg²⁺ ions, supplemented with 1 mg/ml bovine serum albumin and 5 mM glucose. The cells were then loaded with a final concentration of 5 μM Fura-2-AM, cell permeant dye (Invitrogen), and incubated for 30 min at 37°C in the dark. After labeling, cells were washed twice with HBSS buffer to remove Fura-2-AM that did not enter the cells. Cells were counted using a hemocytometer with 0.1% trypan blue to check for live cells. The cell density was adjusted to a final concentration of ∼10⁶ cells/ml using HBSS buffer. The spectrophotometric measurements were performed with an ISS PCH spectrofluorometer (ISS, Urbana, IL, USA). Intracellular calcium was calculated by measuring the ratio of fluorescence intensity (R) at λ_ex/λ_em = 340/510 and 380/510 nm of ∼10⁵ cells taken in a microcuvette. A final concentration of ∼50 to 80 μM carbamoyl chloride (carbachol) (Sigma-Aldrich, St. Louis, MO, USA) was added to the cells in the microcuvette to stimulate a Ca²⁺ response. The ratios for maximum and minimum Ca²⁺ values (R_max and R_min, respectively) were obtained by disrupting the cell membrane with 10% Triton X-100, followed by treatment with 0.2 M EDTA. Intracellular Ca²⁺ concentration was calculated by the following formula (12):

where K_d is the effective dissociation constant and S_f2 and S_b2 are the fluorescence intensities at 510 nm when excited at 380 nm for Ca²⁺-free (with EDTA) and Ca²⁺-bound condition (with Triton X-100), respectively.

Single-cell calcium measurement

SK-N-SH cells were cultured in MatTek glass-bottom dishes. Cells were washed twice with HBSS with Ca²⁺ and Mg²⁺ ions. Calcium Green-AM (Thermo Scientific, Carlsbad, CA, USA) was added to the culture dish and incubated for 30 min at room temperature. Cells were washed 2 times with HBSS and incubated for 45 min before imaging. The fluorescence intensities of Calcium Green–labeled Ca²⁺ were measured over time with a Zeiss LSM 510 device (Carl Zeiss GmbH, Jena, Germany) upon Gα_q stimulation by carbachol addition. Changes in the intracellular calcium of individual cells were calculated after quantifying the intensities by ImageJ software (Image Processing and Analysis in Java; National Institutes of Health, Bethesda, MD, USA).

Effect of siRNA treatment

SK-N-SH and PC12 cells were transfected with a final concentration of 10 nM of glyceraldehyde 3-phosphate dehydrogenase (GAPDH) siRNA or heat shock protein 90 (Hsp90) siRNA using the DharmaFECT 1 transfection reagent (Dharmacon GE) following the manufacturer’s protocol. Three hours after transfection, intracellular Ca²⁺ measurements were performed either on cell ensembles or on individual cells.

Colocalization studies

SK-N-SH cells were transfected with enhanced cyan fluorescence protein (eCFP)-TRAX or eCFP plasmid with FuGene HD transfection reagent (Promega, Madison, WI, USA) following the manufacturer’s protocol. After 48 h of transfection, the cells were treated with 2 to 5 µM carbachol and washed twice with PBS buffer. Cells were fixed with 3.7% formaldehyde for 30 min. After fixing, the cells were washed with PBS and permeabilized in 0.2% Nonidet P-40 detergent for 5 min, followed by blocking with 4% goat serum for 1 h. The fixed cells were incubated with primary antibody for PLCβ for 1 h. The cells were washed 3 times for 15 min each, treated with Alexa Fluor 647–labeled secondary antibody for 1 h, and washed 3 times to remove unbound antibodies. The immunostained cells were imaged with a Zeiss LSM 510 device, and colocalization was calculated from the images using the Zeiss software.

Fluorescence resonance energy transfer between PLCβ1 and TRAX

eCFP-TRAX and PLCβ1–enhanced yellow fluorescence protein (eYFP) were transfected into SK-N-SH cells using FuGene HD transfection reagent (Promega). After protein expression, cells were either mock transfected (transfection reagent only) or transfected with 10 nM siRNA(GAPDH) or siRNA(Hsp90). Three hours after siRNA transfection, Forster resonance energy transfer (FRET) was measured by sensitized emission between eCFP-TRAX and eYFP-PLCβ1 in mock-transfected, siRNA(GAPDH)-transfected, and siRNA(Hsp90)-transfected cells in an Olympus confocal microscope (Olympus, Tokyo, Japan) with ×60 oil immersion objective using Fluoview 3000 software (Olympus). Spectral bleed-through was subtracted using cells transfected with only eCFP-TRAX or eYFP-PLCβ1.

Fluorescence correlation spectroscopy

PC12 cells were cultured for 24 h in 35 mm glass-bottom MatTek chambers to achieve 80% confluence. eYFP-PLCβ1 was overexpressed in these cells using Lipofectamine LTX according to the manufacturer’s protocol; 48 h after transfection, the cells were either transfected with GAPDH siRNA using DharmaFECT or mock transfected with buffer and DharmaFECT alone. Fluorescence correlation spectroscopy traces were recorded 2 h after transfection with a Zeiss confocor3 microscope and a ×60 oil immersion objective. eYFP was excited using 1% of the 514 nm line of an argon ion laser; time traces were recorded for 10 s in the 517 to 641 nm section of the Meta1 channel, and autocorrelation functions were computed by confocor3 software. The autocorrelation function for 3-dimensional diffusion was fit by QuickFit, v3.0 software (http://www.dkfz.de/Macromol/quickfit/), which was also used to compute diffusion coefficients. The confocal volume was calibrated with rhodamine 6G.

FRET with oligonucleotides

eCFP-TRAX and eYFP-PLCβ1 were transfected into SK-N-SH cells using FuGene HD transfection reagent (Promega) by following the manufacturer’s protocol. Forty-eight hours after transfection, 1 to 5 μM Alexa Fluor 532 labeled single-strand DNA (ssDNA) or Rhodamine Red–labeled ssDNA were microinjected with InjectMan NI2 with a FemtoJet pump (Eppendorf, Hamburg, Germany) mounted on a Zeiss Axiovert 200 M. FRET measurements were performed immediately after microinjection of the oligonucleotides with a Zeiss LSM 510 confocal microscope and the data analyzed by Zeiss software.

RESULTS

C3PO overexpression affects calcium signals

C3PO is an octamer consisting of 6 translin subunits that are responsible for the binding of specific oligonucleotides and 2 TRAX subunits that hydrolyze the bound oligonucleotides (7). We have found that PLCβ binds to an external site on one or both of the TRAX subunits in a 1:1 stoichiometric manner (13). Previously we observed that when we overexpressed TRAX in HEK293 cells that were overproducing both PLCβ1 and Gα_q, we could quench the rise in intracellular Ca²⁺ resulting from activation of Gα_q-PLCβ (10). This result suggests that at high levels TRAX could sequester PLCβ to reduce Ca²⁺ signals. Here we determined if this is the case in the human neuronal cell line SK-N-SH expressing endogenous levels of PLCβ and Gα_q and transfected with free eCFP or eCFP-TRAX. The purpose of using a CFP tag is that it allows us to monitor the transfection efficiency, which was between 65 and 70%, and to identify transfected cells in the single-cell studies. It is important to note that expression of TRAX and translin are closely linked (14) (unpublished data), and thus overproducing TRAX leads to a cellular increase of the translin and in turn the C3PO octamer.

In this first series of studies, we loaded the cells with a fluorescent calcium indicator, Fura-2-AM, and viewed an ensemble average of ∼10⁶ cells/ml (or ∼10⁵ cells) SK-N-SH cells in suspension and compared the Ca²⁺ response after carbachol stimulation of Gα_q. We found that the Ca²⁺ release appeared to be composed of at least 2 populations (Fig. 1), one with a narrow distribution that peaks at shorter times (∼20 s after carbachol addition; Fig. 1, inset, blue line) and a second with a much broader distribution that peaks at longer times (∼80 s after carbachol addition; Fig. 1 inset, red line). In cells transfected with TRAX, Ca²⁺ release is reduced by 30 to 40% (n = 3). If we assume that the shape and distribution of the peak remain unchanged with TRAX overproduction, then higher TRAX levels appears to primarily affect the second, broadly distributed peak that spans the entire timescale and contributes to the initial peak.

Figure 1.

TRAX overexpression reduces PLCβ1-mediated Ca²⁺ signals in cell suspensions. A total of 100 µl of SK-N-SH cells at 10⁶ cells/ml were loaded with Fura-2-AM, and Ca²⁺ levels were calculated as described in Materials and Methods (12). Change in fluorescence was monitored before and after stimulation with carbachol at 100 s. Black circles, mock-treated SK-N-SH cells; gray circles, cells transfected with eCFP-TRAX. Inset uses the same axes as full figure and depicts 2 proposed responses, narrow distribution (blue line) and broad distribution (red line); n = 3; sem is shown.

We then monitored the Ca²⁺ response via Gα_q-PLCβ stimulation in single SK-N-SH cells loaded with the calcium indicator Ca²⁺ Green. The traces for a large compilation of cells are shown in Fig. 2, in which we note that each cell was analyzed individually and reported as a normalized response because the change in signal of the indicator depends on the amount of dye loaded, the focus and imaging settings of the microscope, and cell volume and size. We found that each cell showed a distinct behavior that could be categorized into at least 2 response groups that were not apparent in ensemble measurements. Control cells (Fig. 2A) showed 2 distinct Ca²⁺ recoveries. The majority of the cells (∼80%, n = 78) had a smooth rise and recovery, similar to the broadly distributed behavior seen in ensemble measurements. The remaining 20% showed an initial spike in Ca²⁺ level followed by a slow and broad recovery containing many small oscillations. This oscillating behavior has been reported to arise from desynchronization of the intra- and extracellular Ca²⁺ release and recovery rates (15). Taking into account the percentage of the 2 populations, when combined, the resulting curve closely resembles the ensemble average of the Ca²⁺ response.

Figure 2.

Single-cell studies show populations of cells with different Ca²⁺ release behavior. SK-N-SH cells were loaded with fluorescent Ca²⁺ indicator (Ca²⁺ Green), and change in intensity of individual cells was recorded after addition of carbachol (green arrow). Each cell showed distinct normalized change in intensity. A, B) Cells that were transfected with eCFP or mock transfected gave identical results and were compiled together in control cell group (n = 78) (A) and cells overexpressing TRAX (n = 82) (B). Legends indicate percentages of population showing similar behavior, represented by different colors. C) SK-N-SH cells treated for 10 min with IP₃ receptor blocker xestospongin C (n = 54) before stimulation where different populations are shown (black and green). D) SK-N-SH cells overproducing TRAX and treated with xestospongin C (n = 37) where different populations are shown in black, green, and red, noting that green and red populations may not be distinct. In all studies, sem is shown.

We then carried out single-cell measurements in cells overexpressing eCFP-TRAX (Fig. 2B). We found that when TRAX is overproduced, 3 populations are seen. The primary population (52%, n = 82) shows a smooth rise and recovery that is similar to the major population of the control cells. However, the half-maximal value of the rise is shifted toward longer time periods compared to control (i.e., ∼50 to ∼90 s) and did not recover during the time of the measurements (5% recovery vs. 50% at 300 s). The remaining cells show a sharp Ca²⁺ peak with a rapid recovery. Although the response for most of this population was smooth, a significant fraction showed periodic oscillations. Because almost half of this latter population contributed to the Ca²⁺ response at very short times while the other half contributed to the Ca²⁺ at longer times, when the single-cell populations are combined, the overall total response resembles the ensemble behavior.

We postulated that the minor cell populations seen when eCFP-TRAX is overproduced arises from alternate Ca²⁺ entry pathways that do not involve PLCβ, such as Ca²⁺ channels induced by stretch or other mechanical forces. We attempted to test this idea by removing extracellular Ca²⁺ from the medium, but this resulted in distress and fairly rapid (i.e., within 1 h) detachment of cells from the substrate. Therefore, we isolated the contribution of Gα_q-PLCβ-mediated Ca²⁺ signals in healthy cells by blocking the immediate downstream IP₃ channels in the endoplasmic reticulum with xestospongin C (16, 17). In this study, cell suspensions were treated with xestospongin C and loaded with Fura-2-AM. These cells showed an 85% reduction in Ca²⁺, although we noted that ∼15% of cells were not affected by xestospongin C, which we attribute to be due to experimental methods or the quality of the reagent. When we viewed the effect of xestospongin C on single cells, we found 2 populations (Fig. 2C), one consisting of 2 sharp peaks with an interval of ∼200 s, which we attributed to non-PLCβ-mediated Ca²⁺ responses, and another showing behavior similar to the major population of mock-treated cells, suggesting that these cells were not affected by xestospongin C. We note that this latter behavior is identical to the Ca²⁺ response seen in cardiomyocytes when treated with PLC inhibitors (18). When we repeated xestospongin C measurements in cells overexpressing eCFP-TRAX, we found that only the smooth calcium response (i.e., similar to the major population of mock-treated cells) is affected (Fig. 2D).

Presence of siRNAs affects PLCβ-mediated Ca²⁺ signals

We have previously found that PLCβ reverses siRNA-induced knockdown of GAPDH but not knockdown of Hsp90 (10). We have also found that C3PO hydrolyzes siRNA(GAPDH) much more rapidly than siRNA(Hsp90), and that the specific effect of PLCβ on the activity of siRNA(GAPDH) appears to be due to its ability to reduce the relatively rapid rate of siRNA(GAPDH) hydrolysis to a level similar to siRNA(Hsp90) (13). Here we wanted to determine whether C3PO activity in turn affects PLCβ signals. We first verified that the transfection efficiencies of siRNAs are high (∼80–85%), as assessed by viewing cells after transfection with fluorescently labeled siRNAs and confirming protein down-regulation by Western blot analysis. Similar reductions in protein levels were observed for both fluorescently tagged and unlabeled siRNAs (10). We then used unlabeled siRNA for the studies described below and found that the observed effects are reduced 10 to 15% because of incomplete transfection.

We carried out ensemble Ca²⁺ measurements in SK-N-SH cells transfected with siRNAs after a short incubation period (2–3 h). We found a pronounced reduction (∼30%) in Ca²⁺ release in cells treated with siRNA(GAPDH) compared with mock-treated cells (Fig. 3A). Cells treated with siRNA(Hsp90) had a less pronounced reduction. To determine whether this same reduction in Gα_q-PLCβ-mediated Ca²⁺ response in the presence of siRNA(GAPDH) is seen in other cell lines, we repeated this study in PC12 cells (Fig. 3B). We found that siRNA(GAPDH) treatment showed a similar, pronounced reduction in Ca²⁺ signals as seen in SK-N-SH cells. Alternatively, siRNA(Hsp90) treatment did not affect the Ca²⁺ response.

Figure 3.

Treatment with siRNA(GAPDH) or siRNA(Hsp90) affects Gα_q/PLCβ-mediated Ca²⁺ responses. SK-N-SH (A) or PC12 (B) cell suspensions were treated with siRNA(GAPDH) (yellow circles) or siRNA(Hsp90) (orange triangles), or mock treated (brown circles) for 2 h; changes in intracellular Ca²⁺ with addition of carbachol were monitored by Fura-2-AM; n = 3; sem is shown.

We then carried out single-cell Ca²⁺ measurements for SK-N-SH cells transfected with siRNA(GAPDH) or siRNA(Hsp90). These cells showed 3 populations (Fig. 4). In ∼20% of the siRNA(GAPDH)-treated cells, the Ca²⁺ release was low enough such that we postulated that the rapid spike due to non-Gα_q-PLCβ-mediated Ca²⁺ entry was visible. It is notable that for siRNA(GAPDH), the rise of the curve was delayed so that the spike that occurred at short times from non-Gα_q/PLCβ-mediated calcium increases could be viewed as a small shoulder in the major population (Fig. 4B). This shift is reminiscent of the delay seen when TRAX is overproduced and may reflect a stronger association between C3PO and PLCβ with bound siRNA(GAPDH). The third population, which accounted for only ∼10% of the cells, showed oscillating Ca²⁺ fluxes. siRNA(Hsp90) transfected cells also showed behavior distinct from mock-transfected cells that was not apparent in the ensemble measurements.

Figure 4.

Single-cell studies showing effect of siRNAs on Gα_q/PLCβ-mediated Ca²⁺ responses. Normalized Ca²⁺ responses after addition of carbachol, as monitored by Ca²⁺ Green, of single SK-N-SH cells treated for 3 h with siRNA(GAPDH) (n = 97) (A), siRNA(Hsp90) (n = 82) (B), or morpholino RNA (n = 64) (C). Legends indicate percentages of population showing similar behavior, represented by different colors/indicators.

To determine whether the effects of siRNA on PLCβ-mediated Ca²⁺ release occur with other reagents, we repeated the studies using RNA morpholino targeting the human cAMP-gated HCN2 Na⁺/K⁺ channel, the actions of which are not expected to affect Gα_q/PLCβ-mediated Ca²⁺ responses (19). Morpholinos are 25 nt nonhydrolyzable RNAs that can result in nonproductive silencing complexes. We found that treatment with morpholino gives a pronounced attenuated and delayed Ca²⁺ response (Fig. 4C). This result also demonstrated that the effect of siRNA on calcium levels are not off target or indirect, such as through other calcium channels. Taken together, our results suggest that the effect of siRNAs on Ca²⁺ signals depends on the assembly of the C3PO-RISC more than the specific RNA.

Cellular localization of PLCβ is dynamic

The simplest explanation for the effect of siRNA activity on Gα_q-PLCβ-mediated Ca²⁺ signals is that the presence of siRNA shifts the dynamic distribution of the Gα_q-bound and C3PO-bound PLCβ populations. Support for this idea stems from our previous studies showing that in HEK293 cells overexpressing PLCβ1, the colocalization between PLCβ and TRAX decreases dramatically upon stimulation with carbachol (10). To determine whether this is also the case in SK-N-SH cells, we measured the change in colocalization between PLCβ1 and eCFP-TRAX with the addition of carbachol. These studies were carried out by overexpressing either eCFP-TRAX or free eCFP, and immunostaining PLCβ1. We found that colocalization between free eCFP and PLCβ1 decreased slightly with stimulation (0.69 vs. 0.61, P < 0.001, n = 24), possibly reflecting an increased membrane population with Gα_q stimulation. However, we found that carbachol stimulation significantly decreases the colocalization between eCFP-TRAX and PLCβ (0.68 vs. 0.53, P < 0.001, n = 24). Thus, under conditions where PLCβ1 is limiting (i.e., SK-N-SH cells overproducing TRAX), there is a net shift of PLCβ away from TRAX when the affinity between PLCβ and Gα_q increases upon stimulation.

We extended studies showing that the cellular localization of PLCβ depends on the relative levels of Gα_q and C3PO as well as their state of activation (i.e., carbachol stimulation and RISC assembly, respectively) in PC12 cells. Here we measured the relative amounts of eYFP-PLCβ1 in the membrane and in the cytosol in cells transfected with TRAX or treated with siRNA(TRAX) for 48 h. In TRAX-transfected cells, we found a substantial cytosolic population of PLCβ (0.68 ± 0.07 sd, n = 20) compared to the normalized membrane population (1.00 ± 0.05 sd, n = 20). However, when the level of TRAX is knocked down using siRNA, the cytosolic population is substantially reduced (0.21 ± 0.11 sd, n = 20) compared to the membrane (1.00 ± 0.08 sd, n = 20).

We wanted to determine whether the addition of siRNAs affected the physical interaction between PLCβ1 and TRAX. Thus, we carried out FRET studies to determine whether the distance between the eCFP and eYFP tags, which were both placed on the N terminus of the proteins, is within 30 Å. Decreases in the overall association between PLCβ and C3PO or conformational changes that increase the separation between the probes will reduce the FRET. We transfected cells with eCFP-TRAX and eYFP-PLCβ1 and measured the change in FRET when cells were transfected with siRNA (Fig. 5). Mock-transfected cells showed very low FRET values that, when normalized to positive and negative controls, equaled 0.013 ± 0.002 sem (n = 42) (Fig. 5). However, treatment of cells with siRNA(GAPDH) increased the distribution toward high FRET values to 0.213 ± 0.048 sem (n = 38), making the overall FRET significantly higher than control (P = 0.002). Alternately, transfection with siRNA(Hsp90) caused the FRET values to return to the same distribution as control (0.011 ± 0.001 sem, n = 49). Thus, transfection with siRNA(GAPDH) either shifts a population of PLCβ to increase its association with C3PO, or transfection with siRNA(GAPDH) changes the conformation of the PLCβ-C3PO complex to allow for better energy transfer.

Figure 5.

Treatment with siRNAs affects distribution of eCFP-TRAX/eYFP-PLCβ1 FRET efficiency. FRET values between eCFP-TRAX and eYFP-PLCβ1 expressed in SK-N-SH cells was determined by sensitized emission 3 h after cells were treated with transfection reagent only (mock treated, n = 42), siRNA(GAPDH) (n = 38), or siRNA(Hsp90) (n = 49). Although no significant differences are seen between mock-treated and siRNA(Hsp90), differences in FRET for siRNA(GAPDH) samples between mock-treated and siRNA(Hsp90) are significant (P = 0.002).

We followed changes in the mobility of eYFP-PLCβ1 of mock-transfected cells and cells transfected with siRNA(GAPDH) by fluorescence correlation spectroscopy. In mock-transfected cells, PLCβ1 showed 2 populations. The first had a diffusion coefficient of D = 89 ± 2 µm²/s, which we attribute to either monomeric or PLCβ1 in small complexes that have high mobility. The second had lower diffusion coefficient of D = 2.0 ± 0.2 µm²/s, which reflects PLCβ bound to the membrane or a larger complexes. When cells were transfected with siRNA(GAPDH), the mobilities of the 2 populations were unchanged but the size of the faster population increased from 0.519 ± 0.005 to 0.575 ± 0.006, reflecting an increase in the faster-diffusing cytoplasmic population. This increase is consistent with a reduction in membrane-associated PLCβ.

Cytosolic pool of PLCβ localizes with oligonucleotides

Although C3PO binds strongly to oligonucleotides, we have never been able to detect oligonucleotide binding to purified PLCβ in solution using fluorescent-labeled ssDNA or by electrophoresis. However, if PLCβ is bound to C3PO in the cytoplasm, then it may also be in proximity to oligonucleotides bound to C3PO. We tested this possibility by measuring FRET between PLCβ tagged with a FRET donor and microinjected ssDNA covalently labeled with a FRET acceptor. As a control, we first confirmed that microinjected ssDNA binds to C3PO in cells. We expressed eCFP-TRAX in SK-N-SH cells and measured the amount of FRET between eCFP-TRAX and microinjected Alexa Fluor 532–ssDNA (Alexa Fluor 532: 5′-CGCGCGCGCGCGCGC-3′) immediately after microinjection. We found a substantial amount of FRET between the ssDNA and eCFP-TRAX compared to free eCFP and Alexa Fluor 532–ssDNA (Fig. 6A, C), suggesting that a population of the microinjected oligonucleotide binds to C3PO. We then carried out a similar study in which we expressed eYFP-PLCβ1 in SK-N-SH cells and microinjected the same ssDNA labeled with the FRET acceptor Rhodamine Red. These studies revealed a significant amount of FRET between PLCβ1 and the oligonucleotide (Fig. 6B, C), suggesting that oligonucleotides can localize close to the cytosolic pool of PLCβ1, presumably through their common association with C3PO.

Figure 6.

PLCβ1 and TRAX are in proximity to microinjected ssDNA. A) SK-N-SH cells expressing eCFP-TRAX were microinjected with 3 µM ssDNA labeled with Alexa Fluor 532 (Alexa Fluor 532: 5′-CGCGCGCGCGCGCGC-3′) and normalized FRET images, determined by sensitized emission, were acquired. B) In similar study, SK-N-SH cells expressing eYFP-PLCβ1 were microinjected with 5 µM ssDNA (rhodamine: 5′-CGCGCGCGCGCGCGC-3′), and FRET was monitored. C) Box plots showing complied normalized FRET values among eCFP-TRAX and Alexa Fluor 532–ssDNA (n = 26), eCFP-X₁₂-eYFP (positive control, n = 12), and negative control eCFP and Alexa Fluor 532–ssDNA (n = 21) are shown at left. Similar data for eYFP-PLCβ1 and rhodamine-ssDNA (n = 23), Gα_q-eYFP, and rhodamine-ssDNA (negative control, n = 12) and bradykinin type 2 receptor–eYFP with added rhodamine-bradykinin positive control, n = 15) are shown. We also note that microinjection of free dyes (either Alexa Fluor 532 or Rhodamine Red) did not result in FRET to eCFP-TRAX or eYFP-PLCβ1, respectively.

DISCUSSION

PLCβ is primarily localized on the plasma membrane, where it binds to Gα_q to mediate Ca²⁺ signals. The calcium released by PLCβ in turn stimulates the highly active PLCδ as well as Ca²⁺-sensitive Ca²⁺ channels that open to allow the entry of external Ca²⁺ (20). Alternatively, the PLCβ in the cytosol has a population whose level varies with cell type, cell conditions, and levels of expressed proteins. This cytosolic localization is stabilized by C3PO, which competes with Gα_q binding to PLCβ, as indicated in previous studies in HEK293 cells (10), and this control of PLCβ localization by the levels of C3PO was found here for SK-N-SH and PC12 cells. The binding of PLCβ to C3PO dampens the ability of C3PO to rapidly hydrolyze certain microRNAs such as siRNA(GAPDH) but does not affect the much slower rate toward other microRNAs such as siRNA(Hsp90). Although the biophysical basis of C3PO’s action is not well understood, in purified form, C3PO hydrolyzes different oligonucleotides at different rates (13). This mechanism appears to carry over into cultured cells, where overexpression of PLCβ reverses siRNA-mediated down-regulation of siRNA(GAPDH) but not siRNA(Hsp90) (21). The ability of PLCβ to affect RNA silencing correlates well with our observations that PLCβ and C3PO interact in cells, as supported by colocalization, FRET, and pull-down assays (9, 10). Although we have never been able to detect oligonucleotide binding to PLCβ in solution, C3PO binds strongly and fairly nonspecifically to oligonucleotides (22). Oligonucleotide binding to fluorescent-tagged C3PO is readily seen in cells by FRET. Additionally, FRET between the fluorescent oligonucleotide and PLCβ is most easily explained by the binding of fluorescent nucleotides to C3PO that place them within FRET distance of PLCβ.

It is perplexing that although we see large and reproducible changes in the cytosolic population of PLCβ with TRAX levels in HEK293 (10), SK-N-SH, and PC12 cells, the level of PLCβ on the plasma membrane does not appear to change. One explanation is experimental. The membrane level of PLCβ may be so much higher than in the cytosol, due to its strong binding affinity (23), that changes in concentration that appear large in the cytosol are not significant on the membrane. This may be accentuated by the thickness of the bottom membrane attached to the substrate, which may include folds and invaginations that are not optically resolved. It is also important to note that previous confocal imaging studies could not detect either recruitment of PLCβ to the membrane with stimulation or changes in FRET between PLCβ and Gα_q (8). Although this result was unexpected, considering the large increase in PLCβ-Gα_q affinity with activation (24), it was noted that the local levels of the proteins on the plasma membrane complemented by other stabilizing factors work to keep the local concentrations of the proteins above their effective K_d. The net effect is that Gα_q-PLCβ in conjunction with Gβγ and receptors are part of a preformed signaling complex that allows for rapid cell signals and that—important for this work—the Gα_q binding sites for PLCβ are saturated before stimulation, and no changes the number of PLCβ-Gα_q complexes occur. Thus, changes in PLCβ localization may largely depend on C3PO levels rather than Gα_q.

Although our studies support the idea that PLCβ is recruited to the cytosol by C3PO, we were surprised to find that the interaction between PLCβ and C3PO is sensitive to C3PO activity. Specifically, we found that the low degree of FRET between PLCβ and C3PO in SK-N-SH cells is significantly increased soon after transfection with siRNA(GAPDH), showing a shift in the population toward a higher amount of PLCβ-C3PO complexes. This increase is accompanied by a net increase in more mobile PLCβ populations. This FRET result correlates well with the higher degree of PLCβ/TRAX colocalization with siRNA treatment in HEK293 cells (10). However, this increase in eYFP-PLCβ/eCFP-TRAX FRET with the addition of siRNA(GAPDH) is not seen with siRNA(Hsp90). These results are in accord with the idea that C3PO handles different siRNA differently, leading to the speculation that the assembly and activity of other components of the RISC similarly depend on the nature of the silencing RNA. While this interesting idea certainly deserves further study, our results here suggest that treatment with siRNA(GAPDH) promotes the association between PLCβ and C3PO. We also note that it is possible that siRNA(GAPDH) changes the orientation between PLCβ that is already bound to C3PO, allowing for more productive FRET.

Changes in the association between PLCβ and Gα_q and TRAX are functionally seen in the characteristics of Ca²⁺ release. In general, Ca²⁺ release due to Gα_q-PLCβ stimulation is some combination of the release of Ca²⁺ from intracellular stores and extracellular entry with concomitant recovery mechanisms. The resulting Ca²⁺ curve is typified by a skewed gaussian overlaid with oscillations resulting from nonsynchronous recovery mechanisms (25). In the cells used here, we found that ∼20% show this oscillatory behavior. Overproduction of TRAX, and in parallel C3PO, has profound effects on Gα_q/PLCβ-mediated Ca²⁺ release. Ensemble studies show a large reduction in the amount of Ca²⁺ released at shorter times, which has the net effect of extending the duration of the signal. The single-cell studies allowed us to better dissect different populations that contribute to ensemble behavior. Approximately half of the cells showed a smooth, skewed gaussian whose maximum was greatly shifted with TRAX overproduction compared to controls and whose recovery was too long to be monitored. The remainder of the cells showed behavior that was consistent with non-Gα_q-PLCβ-mediated Ca²⁺, as seen when cells are treated with xestospongin C. We note that the behavior seen for the xestospongin C–treated cells is similar to that seen in cardiomyocytes treated with a generic PLC inhibitor or ryanodine (18). These results support the idea that TRAX overproduction reduces the association between PLCβ and Gα_q as well as activation by Gα_q through direct competition for binding

It was surprising to find that siRNA(GAPDH) attenuates PLCβ/Gα_q-mediated Ca²⁺ release in both SK-N-SH and PC12 cells, while siRNA(Hsp90) had a much smaller (SK-N-SH cells) or insignificant (PC12 cells) effect. Although we collected data shortly after siRNA transfection and before translational processes, we cannot discount the possibility that siRNA affects calcium responses through other unknown, short-term events. The very pronounced changes in Ca²⁺ response when cells are transfected with nonhydrolyzable morpholino oligonucleotides suggests that these compounds bind strongly to the C3PO-RISC machinery and stabilize the conformation or conformations that enhance PLCβ association. While the nature of the C3PO-RISC morpholino complexes needs to be further investigated, the important point for this study is that their effects on PLCβ-mediated Ca²⁺ signals are not specific to siRNA(GAPDH) and its downstream targets. However, it is unclear why siRNA(GAPDH) has a much more profound effect on PLCβ-Gα_q-mediated Ca²⁺ release than siRNA(Hsp90). The simplest explanation is that the binding of PLCβ to C3PO is different when siRNA(GAPDH) is bound versus siRNA(Hsp90), as supported by previous solution studies (13), and/or that RISC-C3PO complexes differ with different oligonucleotides, as noted above. More detailed biophysical studies are now underway to better understand RISC-C3PO complexes.

In summary, we have shown that processing of silencing RNAs directly affects G protein–regulated calcium signals and provides strong evidence that these effects are mediated though changes in PLCβ-C3PO interactions. This direct linkage may allow cells to rapidly respond to environmental cues through changes in cellular protein content as well as changes in the activity of Ca²⁺-sensitive enzymes. Our studies suggest a reciprocal relationship between posttranscriptional gene regulation and G protein signaling.

Glossary

C3PO

component 3 promoter of RISC

eCFP

enhanced cyan fluorescence protein

eYFP

enhanced yellow fluorescence protein

FRET

Forster resonance energy transfer

GAPDH

glyceraldehyde 3-phosphate dehydrogenase

Hsp90

heat shock protein 90

IP₃

inositol 1,4,5-trisphosphate

PLCβ

phospholipase Cβ

RISC

RNA-induced silencing complex

siRNA

small interfering RNA

ssDNA

single-strand DNA

TRAX

translin-associated factor X