PREX1 integrates G protein-coupled receptor and phosphoinositide 3-kinase signaling to promote glioblastoma invasion

Abstract

A defining feature of the brain cancer glioblastoma is its highly invasive nature. When glioblastoma cells are isolated from patients using serum free conditions, they accurately recapitulate this invasive behaviour in animal models. The Rac subclass of Rho GTPases has been shown to promote invasive behaviour in glioblastoma cells isolated in this manner. However the guanine nucleotide exchange factors responsible for activating Rac in this context have not been characterized previously. PREX1 is a Rac guanine nucleotide exchange factor that is synergistically activated by binding of G protein αγ subunits and the phosphoinositide 3-kinase pathway second messenger phosphatidylinositol 3,4,5 trisphosphate. This makes it of particular interest in glioblastoma, as the phosphoinositide 3-kinase pathway is aberrantly activated by mutation in almost all cases. We show that PREX1 is expressed in glioblastoma cells isolated under serum-free conditions and in patient biopsies. PREX1 promotes the motility and invasion of glioblastoma cells, promoting Rac-mediated activation of p21-associated kinases and atypical PKC, which have established roles in cell motility. Glioblastoma cell motility was inhibited by either inhibition of phosphoinositide 3-kinase or inhibition of G protein βγ subunits. Motility was also inhibited by the generic dopamine receptor inhibitor haloperidol or a combination of the selective dopamine receptor D2 and D4 inhibitors L-741,626 and L-745,870. This establishes a role for dopamine receptor signaling via G protein βγ subunits in glioblastoma invasion and shows that phosphoinositide 3-kinase mutations in glioblastoma require a context of basal G protein-coupled receptor activity in order to promote this invasion.

Keywords: Abbreviations: GEF; guanine nucleotide exchange factor.

Figure 1. PREX1 expression in glioblastoma cells

A. PREX1 expression in glioblastoma cell line U87MG and glioblastoma cells isolated from patients under serum-free conditions (PriGO 7A, 8A, 9A and 17A) was analyzed by Western blotting. GAPDH was used as a loading control. B. PriGO8A cells were mock transfected or transfected with control siRNA duplex, or with two different siRNA duplexes targeting PREX1 (siPREX1a and siPREX1b). Two days after transfection cell lysates were collected and analyzed by Western blotting. C. PriGO17A cells were treated with 100ng/mL histone deacetylase inhibitor Trichostatin A for 24 hours after which cell lysates were collected and analyzed by Western blotting.

Figure 2. PREX1 expression in glioblastoma tumour xenografts

A. PriGO8A cells were analysed for PREX1 expression by immunohistochemistry. Left and right panels show low and high magnification images respectively. B. PriGO9A cells were injected intracranially in SCID/beige mice and were analysed for PREX1 expression by immunohistochemistry after 6 months of growth (right). SOX2 (left) marks cancer cells. C. PriGO7A cells were injected intracranially in SCID/beige mice and were analysed for PREX1 expression by immunohistochemistry after 6 months of growth. Stem121 was used to detect cancer cells. Cancer cell distribution and PREX1 expression is shown at the cortical subarachnoid space (bottom panels). Scale bars are 500μm and 200μm (panel A right image and panel C bottom images).

Figure 3. PREX1 expression in glioblastoma clinical cases

A. PREX1 was analysed by immunohistochemistry of a tissue microarray containing samples from thirty five glioblastoma patients, with adjacent brain tissue from two cases and normal brain tissue from three subjects. Representative stained cores from the two adjacent brain tissue cases and the three normal brain tissue subjects are shown, along with representative cores from five different glioblastoma cases. B. Intensity of staining (0 – negative, 1, 2 and 3) and frequency of positive staining (0 – no cancer cells positive, 1 – 0-33%, 2- 33-66% and 3 – 66%-100% cancer cells positive) were scored independently by MD and JW with scoring discrepancies averaged. C. PREX1 mRNA expression from the Cell 2013 TCGA RNA Seq V2 dataset was analysed by cancer subtype using cBioPortal [23, 24]. Expression in the classical subtype was significantly different from expression in the mesenchymal and neural subtypes when data were analyzed using Kruskal-Wallis One Way ANOVA on Ranks and All Pairwise Multiple Comparison Procedures (Dunn's Method). * indicates p<0.05.

Figure 4. PREX1 in glioblastoma invasion and motility

A. PREX1 levels were depleted in PriGO 8A cells using siRNA as shown in Figure 1B. In vitro invasion was assessed using Transwell chambers coated with Matrigel basement membrane matrix three days after siRNA-mediated knockdowns. A representative result from three independent experiments for Mock, siControl2, siControl3 and siPREX1a treatments is shown. Data are shown as the mean +/− SE. B. Three days after siRNA-mediated knockdowns live PriGO-8A cell counts were quantified by trypan blue exclusion from three independent experiments. Data are shown as the mean +/− SD. C. Cell motility was assessed three days after siRNA-mediated knockdowns by video microscopy. Cell movement per ten minute frame was quantified in ImageJ with the MTrackJ plug-in and displayed as the average of twenty cells per condition with each condition done in two to three independent experiments. Data are shown as the mean +/− SE. * P<0.05.

Figure 5. Upstream regulation of PREX1 activity in glioblastoma

A. PriGO8A cells were treated with 30μM gallein and/or 1μM BKM120 with DMSO used as solvent control. Cell motility was assessed 24 hours after treatment by time-lapse video microscopy as in Figure 4. B. U87MG cells were treated with 30μM gallein and 24h later cell motility was assessed by time-lapse video microscopy. C. PriGO8A cells transduced with a PTEN cDNA doxycycline inducible lentiviral vector were treated with doxycycline at 1μg/mL for 24 hours in media containing 5% of the standard EGF and FGF2 supplements. Cell motility was assessed by time-lapse video microscopy. D. PriGO8A cells were treated with a single dose of gallein, BKM120, haloperidol or DMSO at the concentrations used above. Cell growth was assessed by crystal violet signal intensity (OD 570nm). Data are shown as the mean +/− SD. E. PriGO8A cells were treated with the indicated concentrations of haloperidol and 24h later cell motility was assessed by time-lapse video microscopy. F. PriGO8A cells were treated with DMSO vehicle, 100 nM L-741,626, 100 nM L-745,870 or a combination of both inhibitors at 100 nM each and 24h later cell motility was assessed by time-lapse video microscopy. For A-C and E-F, data are shown as the mean +/− SE. * P<0.05.

Figure 6. Effect of PREX1 inhibition on PriGO7A and PriGO9A cells

A. PTEN levels in glioblastoma cells and human cortical tissue lysate were analysed by Western blotting. GAPDH was used as a loading control. B and E. In vitro invasion as in Figure 4A was performed on PriGO9A and PriGO7A taken as an average of two independent experiments done in duplicate. C and F. Cell motility of PriGO9A and PriGO7A cell cultures was assessed by time-lapse video microscopy as in Figure 4C three days after PREX1 knockdown by siRNA. D and G. Cell motility of PriGO9A and PriGO7A cell cultures was assessed by time-lapse video microscopy as in Figure 4C 24 hours after treatment. Data are shown as the mean +/− SE. * P<0.05.

Figure 7. PREX1 signals through RAC1 to regulate PAK and PKCι activity

A. Two days after PREX1 knockdown using siRNA phospho-PAK1/PAK2 levels and total PAK2 levels were analysed by Western blotting. B. PriGO8A cells were treated with DMSO, gallein, BMK120 and haloperidol as previously for 24 hours. Lysates were collected and analyzed for phospho-PAK1/PAK2 and total PAK2 levels by Western blotting. C. PriGO8A (left) and PriGO9A cells (right) were transduced with triple-flag-tagged Par6A (TFT-Par6). Lysates were collected under non-denaturing conditions and immunoprecipitation was performed using anti-flag M2-conjugated magnetic beads. Bound proteins were eluted using triple-flag-peptide. D-G. Phospho-PKCι (T555) levels were analysed by Western blot following expression of TFT-Par6 (D), knockdown of Rac1 (E and F) and knockdown of PREX1 (G).