Elucidation of GPR55-Associated Signaling behind THC and LPI Reducing Effects on Ki67-Immunoreactive Nuclei in Patient-Derived Glioblastoma Cells

2.2. Treatment

Either 10,000 cells for Ki67 staining or 20,000 cells for NFAT labeling were seeded on sterile coverslips (Dr. Ilona Schubert Laborfachhandel, Leipzig, Germany, 01-0012) in 24-well plates (Greiner Bio-One™, Frickenhausen, Germany, 662160). Cells were allowed to adhere overnight. They were then treated for 24 h with a fresh medium containing THC or LPI (Table 1). Following treatment, the cells were fixed using a 4% (w/v) paraformaldehyde solution (AppliChem, Darmstadt, Germany, 1.414.511.211). To assess effectors of GPR55-mediated signaling, cells were pre-incubated (marked as ++) with the corresponding inhibitors (Table 1) for varying durations. Subsequently, the medium was replaced by a fresh medium containing THC or LPI with or without inhibitors for 24 h. All treatments were conducted in DMEM supplemented with 10% (v/v) FBS. Cytotoxic effects of all inhibitors in the absence or presence of THC or LPI were assessed by viability assay (Figures S1 and S2). The concentration of inhibitors was determined based on preliminary experiments in which a range of increasing concentrations of each inhibitor was evaluated (Figures S1–S3). Notably, in cases of high inhibitor concentrations, alternate signaling routes might become activated and induce additional off-target effects.

2.3. 3-(4,5-Dimethylthiazol-2-yl)-2,5-Diphenyl-Tetrazolium Bromide (MTT)-Viability-Assay

A total of 3000 cells were placed in 96-well plates (Greiner Bio-One™, Frickenhausen, Germany, 650160) and allowed to adhere overnight. Cells were treated with inhibitors at concentrations specified in Table 1 for 24 h. Four hours before termination of experiments, 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide solution (MTT, Invitrogen, Schwerte, Germany, M6494) at a concentration of 5 mg/mL was added and incubated for 4 h at 37 °C and 5% (v/v) CO₂. After removing the MTT solution, formazan crystals within the cell bodies were dissolved using 100 µL of dimethyl sulfoxide (DMSO, Sigma, Darmstadt, Germany D4540). Absorbance values (Ab) were measured at 540 nm and 720 nm using a microplate reader (SynergyTMMx, BioTek Instruments, Winooski, VT, USA). Control groups consisted of cells incubated in media free of inhibitors, whereas media without cells served as blanks. Cell viability was calculated as follows:

All experiments were performed independently at least three times with six technical replicas for each treatment group.

2.4. PCR

Total RNA extraction was performed using peqGOLD Trifast™ (Peqlab, Erlangen, Germany, 30-2010). Extracts were treated with a DNA-free™ Kit (Invitrogen, Schwerte, Germany, AM1906) according to the manufacturer’s instructions. RNA concentration was quantified using a Synergy™ Mx Microplate Reader (SynergyTMMx, BioTek Instruments, Winooski, VT, USA). PCR amplification was conducted with 4 µL generated cDNA (Reverse Transcription System, Thermo Fisher Scientific, Waltham, MA, USA, K1691) in 20 µL volume containing 10 µL PCR-MasterMix (Promega Inc., Madison, WI, USA, M7505), 0.5 µL forward and 0.5 µL reverse primer (25 pM), 0.25 µL of EvaGreen dye (Biotium, Fremont, CA, USA, 31000), and 4.75 µL of nuclease-free water (Promega Inc., Madison, WI, USA, P1193) The PCR reaction was performed for 40 cycles using a PCR cycler (7900HT Fast Real-Time PCR System, Applied Biosystems™, Thermo Fisher Scientific, Waltham, MA, USA). Conditions used were as follows: initial denaturation at 95 °C, followed by 40 cycles of denaturation at 94 °C (3 s), annealing at 60 °C, elongation at 72 °C (30 s), and fluorescence detection at 76 °C (GNA13, GNAQ, GNAI1, GAPDH, TPB) or 80 °C (GNA12, GNAO1, GNAI2, GNAI3, GNASS, GNASL, POLRR2A) (15 s). For visualization of PCR products, 30 PCR cycles were performed. Products were separated by electrophoresis on 1.5% (w/v) agarose gels (PeqLab, Erlangen, Germany, 35-1020) containing GelRed (Biotium, Fremont, CA, USA, 41003) and were visualized using BioTek Synergy Mix (BioTek Instruments, Winooski, VT, USA) (Figure S5). The primers used are listed in Table 2. Relative transcript levels were calculated using the 2^−∆∆Ct method. For normalization, RNA polymerase II subunit A (POLR2A) served as an internal reference. Normalization to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and TATA-binding protein (TBP) were performed to ensure validity and reproducibility as well as to show stable expression of used reference genes in the experimental setup (Figures S6 and S7).

2.5. Immunochemical Staining of Ki67

Endogenous peroxidase activity was blocked by 3% (v/v) H₂O₂ (Roth, Karlsruhe, Germany, 8070.2) in methanol (Roth, Karlsruhe, Germany, AE01.2) for 10 min. Subsequently, cells were washed three times with 0.02 M PBS containing 0.3% (v/v) Triton X-100 (AppliChem, Darmstadt, Germany, A1388.0500) (PBS/Triton). To minimize unspecific binding, normal goat serum (NGS) (Merck, Darmstadt, Germany, S26) was used at a dilution of 1:20. Cells were incubated overnight at room temperature with primary antibodies (Table 3) appropriately diluted in 0.02 M PBS containing 0.5% (w/v) bovine serum albumin (BSA) (Sigma, Darmstadt, Germany, A7906) and 0.3% (v/v) Triton X-100. After rinsing with PBS/Triton, cells were treated with a goat anti rabbit-specific biotinylated secondary antibody for 1 h at room temperature (Table 3). After incubation with ExtrAvidin^® peroxidase (Sigma, Darmstadt, Germany, E2886) for 1 h, diaminobenzidine (Sigma, Darmstadt, Germany, D8001) was added together with 0.05% (v/v) H₂O₂ for 3 min. Meyer’s hematoxylin (Hollborn und Söhne GmbH, Leipzig, Germany, H02-0500) was used to counterstain the cell nuclei. Cells were dehydrated in an ascending concentration gradient of ethanol, cleaned with xylene (Roth, Karlsruhe, Germany, 97.13.3), and embedded in Entellan (Merck, Darnmstadt, Germany, 107960).

Staining was visualized by using a Leica DMi8 microscope (Leica, Wetzlar, Germany) at 200× magnification. A total of five different regions on each coverslip were captured. Ki67-positive (Ki67⁺) and hematoxylin-positive cells were counted manually, and the ratios of Ki67⁺ nuclei were determined as previously described [16].

2.6. Immunofluorescence Labeling of NFAT

Fluorescence labeling was performed to determine subcellular localization of NFAT isoforms 1–4 after THC or LPI stimulation at different time points (5 min, 10 min, 30 min, 2 h, 4 h, and 24 h). Fixed cells were permeabilized with PBS/Triton for 10 min, followed by incubation with NGS (1:20 in PBS/Triton) for 30 min, before the primary antibody (Table 3) was diluted in 0.5% (w/v) BSA and PBS/Triton was applied overnight at room temperature. Cells were washed with PBS/Triton and treated for 60 min with Alexa488-labeled secondary antibody (Table 3). Nuclei were counterstained with DAPI (Sigma, Darmstadt, Germany, D5637) for 5 min. Coverslips were mounted on slides with a fluorescent mounting medium (Dako, Jena, Germany, S3023). Images of fixed cells were acquired at 400× magnification using a confocal laser scanning microscope (Leica TCS SPE, Wetzlar, Germany). The following excitation wavelengths were used: 405 nm for DAPI and 488 nm for NFAT isoforms. Emission was detected in the range of Δλ = 415–480 nm (DAPI) and Δλ = 500–570 nm (NFAT).

3.1. Cell-Type-Specific ROCK Signaling and a Strongly Involved PLC-IP3 Signaling

To further confirm GPR55-dependent signaling of THC and LPI, we first evaluated the involvement of known key effectors associated with GPR55 pathways. RhoA and its downstream effector ROCK have recently been implicated as key effectors in GPR55-mediated signaling [25]. As many of the cellular actions of RhoA are mediated by ROCK, the effects of ROCK inhibitor Y-27632 were investigated on THC and LPI responses. In GBM #10, the addition of Y-27632 markedly blocked THC- and LPI—mediated changes in the ratio of Ki67⁺ cells (Figure 1a,b, Figures S3a and S4a, Table S2). Interestingly, in GBM #4, ROCK was not involved since effects of THC or LPI were still observed in the presence of Y-27632 (Figure 1a,b, Figures S3a and S4a, Table S2). These data indicate a cell-type-specific impact of THC and LPI on the number of Ki67⁺ nuclei via RhoA/ROCK-dependent signaling.

PLC has been shown to be another important part of signaling pathways triggered by GPR55 [2,25,36,38]. Thus, cells were treated with PLC inhibitor U73122. U73122 significantly diminished THC effects in GBM #4 and GBM #10 (Figure 2a and Figure S3b, Table S3). Similar results were obtained for LPI, as LPI’s effects were no longer detected in the presence of U73122 (Figure 2b and Figure S4b, Table S3). The specificity of U73122 to block PLC was confirmed by using its inactive structural analogue U73343 at the same concentrations used for U73122. Indeed, treatment with U73343 failed to block the responses to THC and LPI in both glioblastoma cell lines (Figure 2a,b, Table S4). U73122 and U73343 themselves displayed no significant effects on the number of Ki67⁺ nuclei (Figure 2a,b, Figures S3b and S4b, Tables S3 and S4). These data indicate that PLC is strongly involved in the regulation of the number of Ki67⁺ nuclei by THC and LPI in both GBM #4 and GBM #10.

PLC converts membrane-bound PIP2 to DAG and IP3. Subsequently, IP3 acts as a second messenger and binds to IP3-sensitive receptors on the ER, resulting in an increase in intracellular Ca²⁺ concentrations. To assess whether IP3 and its receptor were involved in the reduction of Ki67⁺ nuclei upon THC or LPI exposure, we used 2-APB to antagonize IP3-sensitive receptors (Figure 3a,b, Figure S3c and S4c, Table S5). Pretreatment with 2-APB attenuated the responses to THC and LPI in both GBM #4 and GBM #10 (Figure 3a,b, Figures S3c and S4c, Table S5), making the participation of IP3-mediated signaling downstream of GPR55 evident and further confirming PLC-dependent signaling.

3.2. Characterization of Gα- and Gβγ-Subunits, Which Might Couple to GPR55 Signaling

In light of the diversity of G proteins and their functions, the expression of different G-protein α subunits and their isoforms were examined by qRT-PCR analysis of untreated glioblastoma cells. Our results demonstrated that Gα_o, Gα_i (isoforms 1, 2, and 3), Gα_s (small and large isoforms), Gα₁₂, Gα₁₃, and Gα_q were expressed by GBM #4 and GBM #10 at different levels. It is well known that GPR55 can be coupled to the Gα_12/13 and Gα_q proteins. Both subunits were found in GBM #4 and GBM #10 at different levels of transcription (Figure 4a,b, Figures S5 and S6). Remarkably, Gα_q was highly expressed by GBM #10 when compared to GBM #4, leading to an altered expression abundance of GNAQ compared to other transcripts of Gα subunits (Figure 4b and Figure S7). As PLC-dependent and RhoA/ROCK-independent signaling were observed in GBM #4 (Figure 1a,b, Figure 2a,b, Figures S3a–c and S4a–c, Tables S2–S4), Gα_q-mediated signaling by GPR55 was assumed. In contrast, in GBM #10, both RhoA/ROCK- and PLC-dependent signaling (Figure 1a,b, Figure 2a,b, Figures S3a–c and S4a–c, Tables S2–S4) were suggested via Gα_12/13 and/or Gα_q proteins.

According to heterodimerization with other GPCRs, including CB₁ and CB₂, GPR55 might also activate other intracellular G-proteins such as Gα_i/o indirectly. Notably, GBM #4 showed a significantly higher amount of Gα_o transcripts than GBM #10, leading to an altered expression abundance of GNAOI compared to other G-protein transcripts (Figure 4b). The involvement of Gα_i/o proteins as reported for CB₁/CB₂-dependent signaling was examined using the pertussis toxin (PTX), which inhibits the coupling of Gα_i/o proteins to their cognate GPCRs. PTX was unable to block actions of THC and LPI in GBM #4 and GBM #10. Notably, PTX (100 ng/mL) displayed by itself an altered number of Ki67⁺ nuclei (Figure 5a,b, Table S8). Furthermore, no additive agonistic or antagonistic effects of PTX and THC or LPI were observed (Figure 5a,b, Table S8). Since blocking the Gα_i/o proteins with PTX leads to unhindered activation of adenylyl cyclase (AC), the increased AC activity was assumed to be responsible for the reduced number of Ki67⁺ nuclei after PTX treatment. To clarify the role of activated AC, increasing concentrations of forskolin (FSK) were applied (Figure 5c, Table S9). FSK reduced the number of Ki67⁺ nuclei in a concentration-dependent manner in GBM #4 and GBM #10. The data confirm the participation of an increased AC activity in reducing the number of Ki67⁺ nuclei after PTX treatment that is independent of GPR55.

Since GPCRs have the capacity to produce signals through the actions of liberated Gα and Gβγ subunits, we further examined the involvement of Gβγ subunits. Targeting Gβγ subunits by gallein, a small molecule that binds to Gβγ and disrupts Gβγ signaling, significantly reversed the decreased number of Ki67⁺ nuclei elicited by THC in GBM #4 (Figure 6a and Figure S3d, Table S10). In contrast, THC’s effect on GBM #10 remained unaffected in the presence of gallein (Figure 6a and Figure S2d, Table S10). Similar results were obtained for LPI, as gallein attenuated the response to LPI in GBM #4, but not in GBM #10 (Figure 6b, Figure S4d, Table S10). The data suggest that Gβγ-dependent (GBM #4) and Gβγ-independent (GBM #10) signaling might follow an activation of GPR55 by THC or LPI.

4.1. The Involvement of PLC-IP3 and RhoA-ROCK Signaling Pathways

GPR55 is thought to bind to Gα_12/13 or Gα_q rather than Gα_s or Gα_i/o proteins [1,2,25]. Stimulation of Gα_q proteins is generally followed by an activation of PLC [2]. Studies on cell lines and primary cells of different origins revealed that pharmacological activation of GPR55 by LPI and other cannabinoids led to increased intracellular Ca²⁺ concentrations ([Ca²⁺]_i) via the activation of PLC and formation of IP3 [2,25,36,38]. In HEK293 cells and neurons of dorsal root ganglia, a PLC-IP3-dependent increase in [Ca²⁺]_i was observed after stimulation with 5 µM THC and 3 µM LPI [2]. The data of the present work indicate that the THC- and LPI-mediated reduction of Ki67⁺ nuclei might follow a PLC- and IP3-dependent mechanism via GPR55. PLC-IP3-dependent signal transductions through the activation of CB₁ and CB₂ receptors have so far not been reported for THC. In CB₁-transfected HEK293 cells, stimulation with 10 µM THC had no measurable effects on [Ca²⁺]_i, although CB₁-dependent Ca²⁺ release via WIN 55,212-2 was detected [39]. These results are highly consistent with CB₁/CB₂-independent effects of THC despite the presence of CB₁ receptors as reported previously [16] and support GPR55-specific signaling via PLC and IP3.

In addition, GPR55-mediated RhoA activation by LPI and cannabinoids was reported in HEK293 and PC12 cells [1,2,25,26,36]. RhoA-ROCK was shown to be activated downstream of Gα_12/13 and to a lesser extent Gα_q. Since many cellular processes, such as proliferation or migration, might be regulated by RhoA and ROCK, the involvement of RhoA via the inhibition of ROCK was here investigated. A cell-type-specific RhoA-ROCK signaling was observed, indicating differences in intracellular coupling of G proteins to GPR55. Interestingly, in GBM #10, the reduction of the Ki67⁺ nuclei after THC and LPI stimulation might be explained by the interplay between PLC and RhoA-ROCK signaling pathways. In endothelial cells of a rat mesenteric arterial bed, an interaction between PLC- and RhoA-ROCK-modulated [Ca²⁺]_i was found after GPR55 activation by LPI or AM251 [36]. The release of intracellular Ca²⁺ was characterized by two phases. Inhibition of PLC by U73122 blocked the initial phase of Ca²⁺ release, whereas the late phase was abolished by inhibiting ROCK via Y-27632, suggesting a biphasic mechanism [36]. Both phases were abolished by blocking IP3-sensitive receptors with 2-APB, indicating ROCK-induced IP3 generation by direct PLC activation [36]. These data suggest a similar mechanism in GBM #10 after THC or LPI stimulation, involving both PLC- and ROCK-dependent signaling.

The precise mechanism of ROCK-regulated PLC is not fully understood. In GPR55-expressing HEK293 cells, the actin cytoskeleton has been shown to play a pivotal role in the generation of GPR55-dependent Ca²⁺ signals. The family of Rho proteins, including RhoA, regulates protein biosynthesis and cell proliferation as well as the organization of the actin cytoskeleton by affecting actin stabilization through activation of LIM kinase or actomyosin contraction via MLC (modulator of VRAC current 1) [40]. RhoA activation and an intact actin cytoskeleton are required for the THC (5 µM) mediated rise in Ca²⁺ in HEK293 cells [2]. In contrast, pure PLC-mediated pathways as found upon the activation of muscarinic acetylcholine receptors of the parasympathetic nervous system were not affected by either RhoA inhibition or deteriorated actin cytoskeleton formation [40]. Therefore, an intact actin cytoskeleton might play a central role in modulating [Ca²⁺]_i when GPR55 is coupled to RhoA-ROCK signaling. In murine fibroblast cell line NIH 3T3, a disrupted cytoskeleton altered the spatial interaction between PLC and IP3-sensitive receptors, thus affecting PLC-dependent Ca²⁺ signaling [41]. Taken together, these data suggest that the RhoA-ROCK-signaling pathway may indirectly promote PLC-dependent reduction of Ki67⁺ nuclei in GBM #10 through alterations in the spatial location of PLC- and IP3-sensitive receptors. The requirement for direct ROCK activation is supported by the ROCK-independent signaling in GBM #4, making the involvement of basal-active ROCK not evident.

4.2. Cell-Type-Dependent Coupling of GPR55 to Gα_12/13 and/or Gα_q and the Role of Gβγ

The divergence in RhoA-ROCK involvement suggests cell-type-specific differences in coupling to intracellular G proteins. Generally, RhoA-ROCK signaling is known to be activated downstream of Gα_12/13, whereas PLC activation is responsive to Gα_q pathways [2]. Previous studies implicated intracellular coupling of GPR55 to Gα1_2/13 and/or Gα_q pathways [1,2,30]. Missing effects of the ROCK inhibitor in GBM #4 indicated a coupling of GPR55 signaling to Gα_q. In contrast, in GBM #10, GPR55-dependent signaling might be related to Gα_12/13 coupling or dual signaling via Gα_q and Gα_12/13 due to PLC- and ROCK-dependent signaling. Comparable observations were conducted in GPR55-expressing HEK293 cells [25], in neurons of the dorsal root ganglia [2], and in mesenteric arterial endothelial cells [36], where a PLC- and RhoA-ROCK-dependent increase in [Ca²⁺]_i was measured. Moreover, increased [Ca²⁺]_i in the neurons of dorsal root ganglia after GPR55 activation was attributed to dual Gα_12/13 and Gα_q signaling [2]. Some GPCRs functionally switch their associated subfamily of G proteins depending on the expression level of G proteins [39,42]. In the present work, differences in the expression of the relevant Gα_q and Gα₁₂ subunits were found but did not explain cell-type-specific signaling.

GPCRs can initiate signaling cascades not only via Gα subunits. Dissociation of the trimeric complexes after receptor activation also releases free and active Gβγ subunits that might be relevant for signal transduction events. Thus, the aspect of Gβγ-dependent signaling was further examined. A cell-type-specific Gβγ dependence was found. A correlation between Gβγ signaling and GPR55 has so far not been demonstrated in tumor cells. In murine slice cultures of the substantia nigra, GPR55 signaling induced by LPI or O-1602 was associated with the presynaptic release of [³H] gamma-amino-butyric acid (GABA) by a Gβγ-dependent mechanism [43]. In contrast to the present study, Gβγ-mediated AC activation with a subsequent cAMP accumulation was assumed to be an underlying mechanism [43], reflecting species- and cell-type-specific differences.

Although all isoforms of G proteins release one molecule of a Gβγ dimer per activation event, Gβγ signaling is strongly linked to Gα_i/o-derived Gβγ subunits [44]. It is known that GPR55 does not interact with Gα_o/i proteins. Alternatively, GPR55 is able to form heterodimers with Gα_o/i-coupled receptors, as described for CB₁ [45], CB₂ [46,47], and LPA2 [48], resulting in an altered activity and signal transduction of GPR55. To test the involvement of Gα_o/i proteins, PTX was used, as it prevents signal transduction of Gα_o/I by catalyzing irreversible ADP-ribosylation of the α subunit of trimeric Gα_o/iβγ complexes [49]. In our study, PTX was found to be unsuitable for analyzing Gα_o/i-dependent signaling because it directly affected the number of Ki67⁺ nuclei. The responses to PTX could be explained by an accumulation of cAMP due to an enhanced cAMP production via stimulatory GPCRs and an unhindered activation of cAMP-mediated signaling pathways. As an alternative mechanism, Gα_o/i-independent effects of PTX [50] should be considered, which occurred after using high concentrations of PTX. Direct stimulation of AC and cAMP production by FSK reduced the number of Ki67⁺ nuclei comparable to PTX, demonstrating unhindered activation of cAMP-mediated signaling pathways and confuting the Gα_o/i-independent effects of PTX. As PTX treatment showed high background activity, the role of Gα_o/i-derived Gβγ in GPR55 signaling in GBM #4 remains an open question.

Gβγ subunits are able to stimulate PLC directly or indirectly via PI3K [29,51,52,53,54]. Notably, Gα_q-derived Gβγ subunits also have the capacity for signaling but to a lesser extent than those derived from Gα_i/o and might generate signals via PI3K activation [55]. PI3K and PI3K-dependent PLC activation were described as mechanisms for GPR55-driven Ca²⁺ signaling in endothelial cells after AEA treatment [29]. Thus, in accordance with Gα_q signaling in GBM #4, Gα_q-derived Gβγ signaling via PI3K could be conceivable. Therefore, the involvement of PI3K should be investigated in further studies.

Alternatively, it was reported that Gβγ subunits might influence the extent of Gα_q-dependent signaling through direct interaction with PLC [51,53,54]. On the one hand, Gβγ subunits were able to directly stimulate the activity of PLC [51,53] and thus act synergistically with the Gα_q-dependent stimulation of PLC [52,53,54]. On the other hand, Gβγ subunits inhibit the Gα-GTPase activity of PLC, as PLC is both an effector of Gα_q and a Gα_q-selective GTPase-activating protein (GAP) [53,56]. Since prevention of Gβγ disrupts interaction with their downstream effectors, it is conceivable that stimulation of PLC by Gα_q alone was insufficient to exert effects after THC or LPI application. Therefore, in the case of GBM #4, a reduced number of Ki67⁺ nuclei by GPR55 might be determined by the duration and strength of PLC stimulation through interplay between Gα_q and functional Gβγ subunits. The divergence of signaling observed in GBM #10 might base on the coupling of GPR55, as GBM #10 Gα_12/13 and ROCK activation directly or indirectly contributed to a sufficient duration and strength of PLC activation and/or downstream signals to modulate the number of Ki67⁺ nuclei. As discussed previously, this is in line with a biphasic increase in [Ca²⁺]_i by PLC and ROCK signaling in rat mesenteric arterial bed endothelial cells stimulated with LPI or AM251 [36]. It remains to be determined whether the functionality of GPR55, its Gβγ, and RhoA-ROCK-dependent signaling and responsiveness to THC or LPI might be used as a positive prognostic marker.

The authors would like to thank Chalid Ghadban, Candy Rothgänger-Strube, and Christin Zöller for their excellent technical assistance. We acknowledge the financial support from the funding program Open Access Publishing by the German Research Foundation (DFG).