https://orcid.org/0000-0001-7491-9405
Abstract
Cyclooxygenase-2 (COX-2) is a key regulator of inflammation. High constitutive COX-2 expression enhances survival and proliferation of cancer cells, and adversely impacts antitumor immunity. The expression of COX-2 is modulated by various signaling pathways. Recently, we identified the melastatin-like transient-receptor-potential-7 (TRPM7) channel-kinase as modulator of immune homeostasis. TRPM7 protein is essential for leukocyte proliferation and differentiation, and upregulated in several cancers. It comprises of a cation channel and an atypical α-kinase, linked to inflammatory cell signals and associated with hallmarks of tumor progression. A role in leukemia has not been established, and signaling pathways are yet to be deciphered. We show that inhibiting TRPM7 channel-kinase in chronic myeloid leukemia (CML) cells results in reduced constitutive COX-2 expression. By utilizing a CML-derived cell line, HAP1, harboring CRISPR/Cas9-mediated TRPM7 knockout, or a point mutation inactivating TRPM7 kinase, we could link this to reduced activation of AKT serine/threonine kinase and mothers against decapentaplegic homolog 2 (SMAD2). We identified AKT as a direct in vitro substrate of TRPM7 kinase. Pharmacologic blockade of TRPM7 in wildtype HAP1 cells confirmed the effect on COX-2 via altered AKT signaling. Addition of an AKT activator on TRPM7 kinase-dead cells reconstituted the wildtype phenotype. Inhibition of TRPM7 resulted in reduced phosphorylation of AKT and diminished COX-2 expression in peripheral blood mononuclear cells derived from CML patients, and reduced proliferation in patient-derived CD34+ cells. These results highlight a role of TRPM7 kinase in AKT-driven COX-2 expression and suggest a beneficial potential of TRPM7 blockade in COX-2-related inflammation and malignancy.
Introduction
The ubiquitously expressed melastatin-like transient-receptor-potential-7 (TRPM7) cation channel regulates immune system homeostasis, inflammation, and malignant progression.1–11 TRPM7 primarily localizes to the plasma membrane and consists of a channel pore conducting divalent cations (Mg2+, Ca2+, Zn2+), fused to an intracellular and constitutively active C-terminal serine/threonine α-kinase, which may convey separate signaling capacity.12–14 TRPM7 channel and kinase are interdependent in that Mg2+ enters through the channel, and the kinase domain requires Mg2+ to function. The TRPM7 protein is essential for embryonic development and thymopoiesis.15 TRPM7 kinase-inactivated mice are viable but show defects in T-cell, pancreatic beta-cell, and platelet functions, ultimately protecting mice from acute graft-versus-host disease and arterial thrombosis.3,9,16,17 Established TRPM7 kinase substrates include annexin A1, myosin II, phospholipase C gamma 2 (PLCγ2), Ras homology family member A (RHOA), and mothers against decapentaplegic homolog 2 (SMAD2).2,3,18–22 The phosphoinositide-3-kinase/AKT serine/threonine kinase (PI3K/AKT) and extracellular signal-related kinase (ERK1/2) signaling hubs are functionally connected,23–25 although direct phosphorylation via TRPM7 kinase was so far not reported. TRPM7 kinase regulates differentiation of murine T lymphocytes toward proinflammatory TH17 cells,3 and AKT/mTOR-dependent neutrophil recruitment in mice.4 TRPM7 is indispensable for lipopolysaccharide (LPS)-induced activation of murine macrophages,6 for phagocytosis26 and for differentiation to M1 over M2 macrophages.27 It is thus highly involved in proinflammatory mechanisms. Concomitantly, TRPM7 promotes cell proliferation, migration, invasion, and tumor growth in human malignancies and derived cell lines, predominantly due to its ion channel capacities.5,10,28 We recently established a link between TRPM7 kinase activity and RHOA-dependent growth of hepatocellular carcinoma,22 yet a specific role of TRPM7 kinase over ion channel function remains understudied in the context of malignancies. The functional relation to PI3K/AKT and ERK1/2 pathways further suggests a role of TRPM7 kinase in cancerogenic signaling.
Here, we investigated a possible mechanism of TRPM7 kinase in leukemic signaling. We identified COX-2 as proinflammatory and cancerogenic mediator downstream of TRPM7. COX-2 enhances survival and proliferation of malignant cells, while negatively influencing antitumor immunity.29,30 We investigated the genetic ablation, genetic inactivation, or pharmacologic inhibition of TRPM7 channel-kinase in CRISPR/Cas9-edited chronic myeloid leukemia (CML)-derived HAP1 cells, and in primary CML patient cells. We observed diminished constitutive expression of COX-2, and reduced cell proliferation. The results pinpoint this to TRPM7 kinase directly acting on AKT as an in vitro substrate, thereby inducing COX-2 expression downstream of AKT activation. Our data suggest a potential benefit for TRPM7 kinase inhibition in CML, chronic hematopoietic malignancies, or inflammatory conditions driven by AKT/COX-2 signaling.
Materials and Methods
Cell Culture
Wildtype (clone C631, WT) and TRPM7-deficient (clone 10940–04, Horizon Genomics, KO) HAP1 cells were reported previously.31,32 TRPM7-K1648R-mutated HAP1 cells (clone HZGHK431, KD) were generated by CRISPR/Cas-9 genome editing at Horizon Genomics (Vienna, Austria).22 Cells were cultured in Iscove’s modified Dulbecco’s medium (IMDM) containing 10% FBS. KO cells were supplemented with 6 m m MgCl2, removed 21 h prior to experiments. Cells were grown at 37°C in a humidified atmosphere with 5% CO2.
Primary Human CML Cells
Peripheral blood of patients diagnosed with CML (Table ST1) was collected by venipuncture. The study was conducted according to the guidelines of the Declaration of Helsinki, approved by the Institutional Review Board (Ethics Committee) of the Province of Salzburg (415-E2009/2-2016, issued to RiG). PBMCs were isolated of whole blood. CD34+ cells were sorted on an ARIA III (BD, USA). Cells were cultured in StemSpan SFEM II media supplemented with StemSpan CD34+Expansion Supplement and UM729 (all from Stem Cell Technologies, USA), according to the manufacturer’s instructions.
Antibodies and Inhibitors
Immunoblotting: α-NFATc1 (sc-13033), α-ERK1/2 (clone C-9), α-pERK (Tyr204) (sc-7383), α-pERK1/2 (Thr202/Tyr204) (clone D13.14.4E), α-pAKT (Thr308) (clone B-5), α-NFκB p65 (cloneA-12), and α-HSP90 (clone F-8) were from Santa Cruz Biotechnology. α-LaminB1 (PA5-19468) and α-GAPDH (PA1-987) were from Thermo Fisher Scientific. α-pAKT (Ser473) (clone D9E), α-AKT (clone C67E7), and α-pCREB (Ser133) (clone 87G3) were from Cell Signaling Technology. α-pTRPM7 (Ser1511) was described previously.33 HRP-linked secondary antibodies were from Thermo Scientific (goat anti-rabbit) or Cell Signaling Technology (horse anti-mouse). Inhibitors TG100-115 (S1352), IPI-549 (Eganelisib, S8330), and GSK2269557 (Nemiralisib, S7937) were from SelleckChem. Thapsigargin (T9033) was from Sigma–Aldrich. NS8593 (N-196) was from Alomone. Imatinib (SML1027) was from Sigma–Aldrich.
Inductively Coupled Plasma Mass Spectrometry
HAP1 cell pellets were analyzed using inductively coupled plasma mass spectrometry (ICP-MS, ALS Scandinavia), as reported previously.32 Briefly, cells were grown in 125 cm2 flasks until 50%–75% confluency, trypsinized, washed, and collected by centrifugation. Cell pellets were dried at 70°C before analysis. Elementary contents of Mg or Zn were normalized to sulfur.
Electrophysiology
TRPM7 currents were acquired via whole-cell patch clamp. A ramp from −100 to +100 mV over 50 ms acquired at 0,5 Hz and a holding potential of 0 mV was applied. Inward and outward current amplitudes were extracted at −80, respectively, +80 mV. Data were normalized to cell size measured after whole-cell break in (pA/pF−1). Capacitance was measured using the capacitance cancellation (EPC-10, HEKA). Mg2+-free extracellular solution (in m m): 140 NaCl, 3 CaCl2, 2.8 KCl, 10 HEPES-NaOH, 11 glucose (pH 7.2, 290–300 mOsm). Intracellular solution (in m m): 120 Cs-glutamate, 8 NaCl, 10 Cs-EGTA, 5 EDTA (pH 7.2, 290–300 mOsm).
Calcium Imaging
HAP1 cells were seeded into ibidi µ-dishes (Ibidi). Intracellular Ca2+ was monitored with Fura-2 AM (SantaCruz) using dual excitation at 340 and 380 nm, emission at 520 nm. Cells were loaded with 3 µm Fura-2 AM and 0.05% Pluronic F-127 (Invitrogen) in imaging buffer, 15 min at 37°C. Cells were washed with buffer (in m m): 140 NaCl, 2 CaCl2, 4 KCl, 0.4 MgCl2, and 10 HEPES-NaOH (pH 7.3). Time lapse images were acquired on an Anglerfish imager, using 5 µm thapsigargin (Thermo Fisher) to mobilize intracellular Ca2+. Images were analyzed with Fiji.
Nuclear Extraction
A Nuclear Extract Kit (Active Motif, 40010) was applied according to the manufacturer’s instructions.
Sodium Dodecyl Sulfate–Polyacrylamide Gel Electrophoresis and Western Blotting
Lysates were prepared in RIPA buffer and diluted with 4x Laemmli buffer (62.5 m m Tris/HCl, 20% (v/v) glycerol, 5% (v/v) β-mercaptoethanol, 4% (w/v) SDS, and 0.1% (w/v) bromophenol blue), heated to 60°C (analysis of TRPM7), respectively, 95°C (other proteins) for 5 min and separated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE). Proteins were transferred onto nitrocellulose or polyvinylidene fluoride membranes, blocked with 5% BSA or skim milk in TBST buffer. Membranes were incubated with respective antibodies according to standard procedures. Immune reactivity was quantified via densitometry (Bio-Rad). Samples were normalized to untreated and/or WT samples, and loading controls. Data were analyzed with ImageJ.
Bio-Plex Assay
For Bio-Plex ProTM Cell Signaling Assay (Bio-Rad), cells were preincubated with DMSO, NS8593 (30 µm), TG100-115 (20 µm), or IPI-549 (160 n m, Eganelisib) plus GSK2269557 (100 n m, Nemiralisib) for 30 min. Cells were lysed using Cell Signaling Reagent Kit (Bio-Rad). Samples were processed according to the manufacturer’s instructions, against targets: pAKT (Ser473, 171-V50001M), pERK1/2 (Thr202/Tyr204, Thr185/Tyr187, 171-V50006M), and pSMAD2 (Ser465/Ser467, 171-V50019M). Data were normalized to the total protein or beta-actin.
mRNA Isolation
mRNA was isolated from cell pellets, either by RNeasy Mini Kit (Qiagen) according to the instructions, or as follows: pellets were resuspended in 1 mL TRIzol (Thermo Fisher Scientific, 15596-026) and incubated for 5 min at RT. A volume of 200 µL trichloromethane/chloroform was added, samples shaken for 15 s, incubated for 3 min at RT, and centrifuged with 8000 g for 15 min at 4°C. The aqueous phase was transferred, mixed with 500 µL isopropanol and incubated for 10 min at RT. After centrifugation, supernatant (SN) was discarded and the RNA-pellet washed with 1 mL 75% ethanol. Centrifugation with 7500 rpm for 5 min at 4°C was followed by discarding the SN and drying the pellet for 30–40 min at RT. RNA samples were resuspended in 10–12 µL DEPC-treated H2O, heated to 55°C for 10 min, and concentration was determined via OD measurement.
cDNA Synthesis and Quantitative Real-Time Polymerase Chain Reaction (qRT-PCR)
For cDNA synthesis, up to 4 µg mRNA was diluted in H2O, mixed with 1 m m dNTP and 0.5 µg Oligo(dT)12–18 and incubated for 5 min at 65°C. On ice, 5x First-Strand Buffer, 1 m m dithiothreitol (DTT) and DEPC-treated H2O were added. After incubation at 42°C for 2 min, SuperScript II Reverse Transcriptase (Thermo Fisher Scientific) was added and incubated for 50 min at 42°C. Samples were heated to 70°C for 15 min. Transcripts were analyzed by specific primer pairs. Master mixes additionally contained cDNA diluted 1:20, and SYBR Green (Thermo Fisher Scientific or Sigma–Aldrich). Transcripts were measured in technical triplicates on a LightCycler 480 (Roche) or a CFX-96 (Bio-Rad): 50°C 2′, 95°C 10′ (preincubation), 95°C 15″, 62°C 30″, 72°C 30″, 40 cycles (amplification), 95°C 10″, 60°C 1′ (melting), 40°C 10′ (cooling). Primer pairs: (all human, 5′–3′): hCOX-2 (fw) TGTATGTATGAGTGTGGGATTTGAC and (rev) GATCATCTCTGCCTGAGTATCTTTG, hCOX-2 (fw) GAGTTATGTGTTGACATCCAGATCAC and (rev) GCTGCTTTTTACCTTTGACACCC, hCOX-1 (fw) CACAGTGCGCTCCAACCTTA and (rev) CTTTGGTTCCATGGGTGTG, hHPRT1 (fw) AAGCTTGCTGGTGAAAAGGA and (rev) AAGCAGATGGCCACAGAACT, or hHPRT1 (fw) CCCTGGCGTCGTGATTAGTG and (rev) TCGAGCAAGACGTTCAGTCC. COX-2 was acquired by two distinct primer pairs on the same samples, and pooled for statistical analysis. A minimum of three independent experiments were performed. CT values of housekeeping transcripts were subtracted from measured CT values, to calculate 2^(-ΔCT) values.
COX-2 Promoter Activity Assay
HAP1 cells were seeded into 24-well plates at equal cell numbers, and left to attach during 6 h incubation. Cells were then transfected with pDRIVE5-Lucia plasmid containing the hCOX-2 promoter to express secreted Lucia luciferase (InvivoGen). Transfection was performed using the TransIT-X2 system (Mirus Biologicals) on Opti-MEM carrier medium (Gibco), according to instructions. If applicable, drugs were added at this point. After 22 h, 20 uL supernatant were combined with 50 uL of 1x Quanti-Luc4 reagent (InvivoGen), and product was measured on a GloMax luminescence reader (Promega), with 0.1 s reading time. At least 4 independent experiments were performed. Data were normalized to respective wildtype or DMSO control readout of the individual experiments, and compared to GFP expression of a control plasmid. Untransfected wells were measured alongside all conditions.
COX Activity Assay
HAP1 cell pellets were lysed with ice cold lysis buffer [DPBS + 1% (w/v) NP40/Igepal + protease inhibitor (Roche)]. Samples were rotated for 30 min at 4°C, centrifuged with 12 500 rpm for 5 min at 4°C, supernatant was transferred into a new reaction tube on ice. COX Activity Assay (Abcam, ab204699) was performed according to the manufacturer’s instructions, in doublets. Peroxidase activity was read on a microplate reader.
In Vitro Kinase Assay
Purified recombinant human full-length inactive AKT1-GST (A16-14 G) and GST (G52-30 U) were from SignalChem. RBC HotSpot kinase assay was performed by Reaction Biology Corp. (Woolbridge, USA). Standard reaction buffer contained: 20 m m HEPES, pH 7.5, 10 m m CaCl2, 1 m m EGTA, 0.01% Brij35, 0.02 mg/mL BSA, 0.1 m m Na3VO4, 2 m m DTT, 1% DMSO, and 2 m m MnCl2. Assays were carried out with 4 µm33P-ATP, 4 µm protein substrate or GST control, and 20 µm mannose binding protein (MBP) control substrate. TRPM7 channel-kinase (Carna Biosciences, 10-111) was titrated from 50 n m in a 2-fold serial dilution. Incorporation of radioactive 33P was analyzed. Raw data were converted to n m substrate phosphorylated, based on readout for 4 u m 33P-ATP. Kinase alone was deducted as background. In addition, data were calculated as 33P incorporated per n m enzyme substrate. Reactions were carried out in duplicates, incubating for 1 h.
Fluorescence Microscopy
Localization of NFATc1 was acquired on a Zeiss LSM 780 microscope, using a 63x oil objective. Cells were stimulated with 5 µm thapsigargin for 15 min, or left unstimulated. Cells were permeabilized with 0.1% Triton X-100 for 5 min and stained for intracellular NFAT using anti-NFATc1 antibody (1:100, Santa Cruz, 7A6) in 0.2% BSA/1% normal goat serum in PBS, and secondary anti-mouse antibody AF647 (1:1000, Cell Signaling). Cells were counterstained with DAPI (0.2 µg/mL), and mounted onto glass coverslips using Antifade ROTIMount FluorCare. Zen 3.5 software was applied. Regions of interest were defined for nuclear outlines (DAPI signals). Nuclear NFATc1 was normalized to respective DAPI intensities.
Statistics
Data were plotted with GraphPad Prism 7 or higher. Statistical analysis used two-tailed unpaired t-test (two samples), one-way or two-way ANOVA with multiple comparisons (more samples). qRT-PCR and imaging data were analyzed with nonparametric Mann–Whitney ranks test (two samples) or multiple comparisons test with Dunnett’s correction (more samples). Western blot was analyzed by Ratio-paired t-test.
Illustration
The schematic was finalized with CorelDRAW X8.
Results
TRPM7-deficient HAP1 Cells Show Reduced NFAT Translocation and COX-2 Expression
To study the role of TRPM7 in cellular signaling of CML cells, the human CML-derived cell line HAP1 in which TRPM7 was ablated (TRPM7 KO) via CRISPR/Cas9-based genome editing was utilized.31,32 Whole-cell patch-clamp experiments confirmed functional channel inactivity in TRPM7 KO cells, indicated by lacking TRPM7 current density and diminished voltage-dependent ion currents (Figure 1A).31 Using ICP-MS, we confirmed a significant reduction in total Mg2+ and Zn2+ content of TRPM7 KO cells compared to WT (Figure 1B and C).31,32 As TRPM7 also conducts Ca2+ ions and has been implicated in Ca2+ signaling, we further analyzed intracellular free Ca2+ concentrations ([Ca2+]i) in TRPM7 WT and KO cells using Fura-2 AM-based ratiometric imaging. Basal [Ca2+]i was unaltered in TRPM7 KO cells compared to WT (Figure 1D); however, the Ca2+ mobilization from intracellular stores by depletion with thapsigargin was significantly reduced in KO cells (Figure 1E). Collectively, these data reinforce the notion that deletion of TRPM7 protein leads to cellular deprivation of divalent cations including Mg2+, Zn2+, and Ca2+.32 We investigated whether lack of TRPM7 affects the nuclear expression of Ca2+-dependent nuclear factor of activated T cells (NFAT) and demonstrated via Western blot that in TRPM7 KO cells the basal and thapsigargin-induced nuclear expression of NFATc1 was reduced (Figure 1F). Accordingly, imaging of NFATc1 nuclear translocation in response to thapsigargin showed a significant reduction in KO cells, compared to WT (Figure 1G). Analogously, nuclear factor “kappa-light-chain-enhancer” of activated B cells (NFκB) p65 was reduced in TRPM7 KO cells (Figure S1A). Analyzing the expression of inflammatory/protumorigenic genes downstream of Ca2+ and NFAT signaling highlighted a selective reduction in the constitutive expression of COX-2 mRNA in TRPM7 KO cells (Figure 1H), while COX-1 and MYC were unaltered (Figure S1B and C). COX-2 expression upon thapsigargin stimulation increased with treatment, compared to untreated WT cells, while KO cells showed reduced COX-2 throughout all conditions (Figure S1D). The lower expression of COX-2 mRNA was reflected by significantly reduced COX-2 promoter activity (Figure 1I) and diminished COX protein activity in TRPM7 KO cells (Figure S1E). Besides NFATc1, we examined the nuclear content of phosphorylated CREB, which can serve as calcium-driven signaling entity to induce COX-2 expression.34 Nuclear pCREB was comparable in TRPM7 KO and WT cells (Figure S1F). Basal nuclear expression of pCREB was generally high in HAP1 cells (Figure S1F). In conclusion, we show manifold alterations in cellular ion homeostasis and signaling in TRPM7 KO HAP1 cells, including reduced expression of the inflammation marker COX-2.
TRPM7 protein knockout results in reduced NFAT translocation and COX-2 expression in HAP1 cells. (A) Electrophysiological characterization of TRPM7 currents in TRPM7 WT (black, n = 11) and KO (red, n = 12) HAP1 cells using whole-cell patch clamp. Averaged current densities at +80 and –80 mV are plotted versus time (left, mean ± SEM) and representative current–voltage (I/V) relationships extracted at 200 s are depicted (right). Note the abolished TRPM7 channel activity in KO cells. (B–C) Scatter plots of Mg2+ (B) and Zn2+ (C) ion contents in TRPM7 WT and KO cells assessed via ICP-MS (n = 4). Concentrations are counts of ion over sulfur (S), in ppm. (D) Fura-2 AM-based ratiometric measurements of basal Ca2+ concentrations in resting WT (black, n = 149) and TRPM7 KO (red, n = 82) HAP1 cells. (E) Time course (left) and quantification (right) of Fura 2-AM-based imaging of cellular Ca2+ mobilized from intracellular stores by stimulation with 5 µm thapsigargin (arrow). WT HAP1 cells are shown in black and TRPM7 KO cells in red. Right panel shows area under the curve (AUC) of stimulation response. (F) Western blot (left) and semiquantification (right) of nuclear NFATc1 in lysates of resting or thapsigargin-stimulated TRPM7 WT and KO cells. Normalized to Lamin B1 (n = 13). (G). Intracellular localization of NFATc1 in TRPM7 WT and KO cells, by confocal microscopy. Cells were stimulated with thapsigargin, or left unstimulated (exemplary images, left panel). Quantification gives nuclear NFATc1 stain normalized over DAPI signal (right panels, n = 374–555 cells). Scale bar: 5 μm. (H) Relative COX-2 mRNA expression in WT (black) and TRPM7 KO (red) HAP1 cells, assessed by qRT-PCR. Data were related to respective housekeeping controls and depicted as 2−ΔCT (n = 8–9 triplicates). (I) Relative COX-2 promoter activity in HAP1 WT or TRPM7 KO cells, assessed by a luminescence reporter assay on secreted Lucia luciferase (n = 6). Statistics: Student’s t-test (B, C, F, I) or Mann–Whitney ranks test (D, E, G, H), *P < .05, **P < .005, ***P < .0005, ****P < .00005, and n.s.—not significant. Data are mean ± SD.
TRPM7 Kinase Regulates COX-2 Expression Independent of NFAT Activation
To understand the specific contribution of TRPM7 kinase moiety to cellular signaling and COX-2 expression, CRISPR/Cas-9-edited HAP1 cells with a point mutation at the TRPM7 kinase active site (TRPM7K1648R, kinase-dead, KD)22 were employed. Western blot of an auto-phosphorylation site of TRPM7 (pSer1511) confirmed the disrupted TRPM7 kinase activity (Figure S2A),22 and patch-clamp confirmed a functional TRPM7 channel in KD cells (Figure 2A).22 Using ICP-MS, we found unaltered Mg2+ and Zn2+ contents in TRPM7 KD cells compared to WT cells (Figure 2B and C). Analysis of basal [Ca2+]i using Fura-2 AM revealed no changes in TRPM7 KD cells (Figure 2D). Compared to TRPM7 KO cells, Ca2+ mobilization from intracellular stores resulted in a shifted, yet less-pronounced reduction of mobilized Ca2+ in TRPM7 KD cells (Figure 2E). Our data therefore confirm a modulation of store-operated calcium entry by TRPM7 kinase, as has been described previously in other cell types.9,35 However, nuclear expression of NFATc1 was unaffected in TRPM7 KD cells (Figure 2F), and nuclear NFATc1 translocation in response to thapsigargin was not altered compared to WT (Figure 2G). Importantly, we found a persistent reduction of COX-2 mRNA expression and COX-2 promoter activity in TRPM7 KD cells (Figure 2H and I), similar to our observations in TRPM7 KO cells. COX-1 mRNA was unaffected (Figure S2B). Analogously to NFATc1, nuclear NFκB p65 and pCREB levels were unaltered at high levels in TRPM7 KD cells (Figure S2C and D). Together, by investigating TRPM7 KD cells with its still intact ion channel but disrupted kinase signaling domain, we identified a functional role of TRPM7 kinase in inducing COX-2 expression, that acts separate from TRPM7 ion channel mechanisms.
Targeted inactivation of TRPM7 kinase reduces COX-2 expression without affecting NFAT translocation. (A) Electrophysiological characterization of TRPM7 currents in TRPM7 WT (black, n = 12) and KD (blue, n = 9) HAP1 cells using whole-cell patch clamp. Averaged current densities at +80 and –80 mV are plotted versus time (left, mean ± SEM) and representative current–voltage (I/V) relationships extracted at 200 s are depicted (right). (B–C) Scatter plots of Mg2+ (B) and Zn2+ (C) ion contents in TRPM7 WT and KD cells assessed via ICP-MS and analyzed as in Figure 1B (n = 4). (D) Fura-2 AM-based ratiometric measurements of basal Ca2+ concentrations in resting WT (black, n = 113) and TRPM7 KD (blue, n = 199) HAP1 cells. (E) Time course (left) and quantification (right) of Fura 2-AM-based imaging of Ca2+ mobilized from intracellular stores by stimulation with 5 µm thapsigargin (arrow). WT HAP1 cells are shown in black and TRPM7 KD cells in blue. Right panel shows AUC of stimulation response. (F) Western blot (left) and semiquantification (right) of nuclear NFATc1 in lysates of resting or thapsigargin-stimulated TRPM7 WT and KD cells. Normalized to Lamin B1 (n = 10). (G) NFATc1 localization in TRPM7 WT and KD cells, by confocal microscopy. Cells were stimulated with thapsigargin, or left unstimulated (exemplary images, left panel). Quantification gives nuclear NFATc1 stain normalized over DAPI signal (right panels, n = 331–359 cells). Scale bar: 5 µm. (H) Relative COX-2 mRNA expression in WT (black) and TRPM7 KD (blue) HAP1 cells, assessed by qRT-PCR. Data were related to respective housekeeping controls and depicted as 2−ΔCT (n = 8–9 triplicates). (I) Relative COX-2 promoter activity in HAP1 WT or TRPM7 KD cells, assessed by a luminescence reporter assay on secreted Lucia luciferase (n = 5). Statistics: Student’s t-test (B, C, F, I) or Mann–Whitney ranks test (D, E, G, H), **P < .005, ***P < .0005, ****P < .00005, and n.s.—not significant. Data are mean ± SD.
TRPM7 Kinase Induces COX-2 Expression Via Phosphorylation of AKT
Typically, COX-2 expression is regulated via NFAT, NFκB, and/or CREB transcription factors.34,36,37 We showed that TRPM7 kinase affects COX-2 expression independently of NFATc1, CREB, or NFκB in the HAP1 clones. SMAD2, AKT, and ERK1/2 signaling entities are also considered upstream of COX-2 induction38–40—each representing reported signaling factors downstream of TRPM7 kinase.3,24 We first analyzed phosphorylation of SMAD2 (Ser465/Ser467), a direct target of TRPM7 kinase downstream of TGF-ß, that induces inflammatory Th17 cells3 and lung fibrosis41 and may also be involved in TGF-ß-dependent induction of COX-2.40,42,43 pSMAD2 was significantly reduced in TRPM7 KO and KD cells compared to WT HAP1 cells (Figure S3A and B), confirming previous findings. Applying the TRPM7 channel blocker NS8593 (NS)44,45 to WT cells further confirmed the kinase-effect on pSMAD2, compared to controls (DMSO, and the SK-channel blocker Apamin as off-target control for NS) (Figure S3C). Due to kinases requiring magnesium ions to function, targeting TRPM7 with NS simultaneously also blocks TRPM7 kinase mechanisms, as shown previously.4,22 A kinase inhibitor targeting TRPM7 alongside PI3K kinase, TG100-115 (TG),33,46 showed no significant effect on pSMAD2 here, similar to controls [PI3K-targeting Eganelisib (EGA) and Nemiralisib (NEM)] (Figure S3D). Since SMAD activity has also been linked to activation of AKT,47 and PI3K/AKT signaling serves downstream of TRPM7,23,24 we investigated AKT signaling in the HAP1 clones. As depicted in Figures 3A and S4A, TRPM7 KO cells showed significantly reduced phosphorylation of AKT (Ser473), compared to WT. By applying the TRPM7 blocker NS to WT cells, we confirmed the reduction of phosphorylated AKT (Ser473) (Figure 3B) and an accompanying reduction of COX-2 expression (Figure 3C) and COX-2 promoter activity (Figure S3E), compared to controls or to COX-1 expression (Figure S3F). In line with our results on COX-2, in TRPM7 KD cells lacking TRPM7 kinase activity, we identified significantly reduced levels of pAKT (Ser473) (Figures 3D and S4B). This was further reflected by significantly reduced pAKT (Ser473) activation in response to serum stimulation, in KD cells (Figure S4C). Analogously, applying the TRPM7 kinase/PI3K inhibitor TG resulted in a significant reduction of pAKT (Ser473) (Figure S3E) and of COX-2 expression and promoter activity in WT cells (Figures 3F and S3G), whilst also affecting COX-1 expression (Figure S3H). Similar outcomes were observed with the PI3K-targeting controls EGA and NEM, acting upstream of AKT (Figure 3E and F). To answer the question if TRPM7 kinase may phosphorylate AKT via direct enzyme–substrate engagement, an in vitro kinase assay was performed. This approach has been successfully applied to identify SMAD2 and RHOA as TRPM7 kinase substrates in the past.3,22 Incubating purified recombinant TRPM7 protein and inactive AKT1 as kinase substrate, in vitro, TRPM7 kinase is able to phosphorylate AKT1 in a concentration dependent manner, compared to GST control (Figures 3G and S4D). To probe this interdependency in a cellular environment, we added the AKT activator SC7948 to TRPM7 KD cells. SC79 reconstituted AKT phosphorylation and TRPM7 kinase function on COX-2 expression, compared to control (Figure 3H and I). In addition, we investigated phosphorylation of AKT at residue Thr308, which is independent of serum stimulation and equally necessary for full AKT activation.49 Analogously, we found reduced pAKT (Thr308) in KD cells (Figure S4E) compared to WT cells. Contrarily, basal phosphorylation of ERK1/2, a master candidate in oncogenic signaling, was not affected in TRPM7 KO or KD cells compared to WT (Figure S4F and G). Together, the AKT-dependent induction of COX-2 expression in kinase-dead HAP1 cells, and the data obtained by the in vitro kinase assay strongly suggest AKT as major player in TRPM7-mediated COX-2 induction.
TRPM7 kinase induces COX-2 expression via AKT phosphorylation. (A) Representative Western blot (left) and semiquantification (right) of the basal phosphorylation of AKT (Ser473) in TRPM7 WT versus KO HAP1 lysates. Signals were normalized to respective GAPDH housekeeping controls, depicted as % of WT (n = 8). (B) pAKT (Ser473) levels in pharmacologically treated HAP1 WT cells, assessed by BioPlex technology (n = 4–5). Cells were treated with TRPM7 channel inhibitor NS8593 (NS, red, 30 µm), the K+ channel inhibitor Apamin as control (200 n m), or DMSO control. (C) Relative COX-2 mRNA expression in inhibitor-treated HAP1 WT cells, assessed via qRT-PCR (n = 4–5 triplicates). (D) Representative Western blot (left) and semiquantification (right) of basal pAKT (Ser473) in TRPM7 WT and KD HAP1 lysates. GAPDH was blotted as housekeeping control (n = 10). (E) pAKT (Ser473) levels in HAP1 WT cells treated with TRPM7 kinase/PI3K inhibitor TG100-115 (TG, blue, 20 µm), PI3K inhibitors Eganelisib (EGA, 160 n m) + Nemiralisib (NEM, 100 n m), or DMSO control. BioPlex technology was used (n = 6). (F) Relative COX-2 mRNA expression in inhibitor-treated HAP1 WT cells, assessed via qRT-PCR. Inhibitors as in C (n = 9–10 triplicates). (G) In vitro kinase activity assay of recombinant human TRPM7 protein (titrated) on recombinant inactive AKT1-GST fusion protein (gray) or GST control (white). Phosphorylation was measured after 1 h reaction and converted to relative amount of n m 33P-ATP substrate incorporated. Duplicate experiment. (H) Pharmacologic reconstitution of AKT activation on TRPM7 KD cells. Representative Western blot of pAKT (Ser473) in TRPM7 WT and KD HAP1 cells, cultured with the AKT activator SC79 or vehicle. Lysates were prepared after 21 h serum starvation followed by 1 h serum re-addition plus respective inhibitors, plotted against unstimulated control (Ctrl). GAPDH served as housekeeping control. (I) Pharmacologic reconstitution of COX-2 expression in TRPM7 KD cells. Relative COX-2 mRNA expression was analyzed in TRPM7 WT (black) and KD HAP1 cells (blue), treated with SC79 or DMSO vehicle for 2 h, respectively (n = 5–6). Statistics: Student’s t-test (A, D), one-way ANOVA (B, E), Kruskal–Wallis ranks test or multiple comparison with Dunnett’s correction (C, F, I), *P < .05, **P < .005, ***P < .0005, ****P < .00005, and n.s.—not significant. Data are mean ± SD.
Pharmacologic Blockade of TRPM7 Inhibits COX-2 Expression and Proliferation of CML Patient Cells
To investigate whether COX-2 expression was also targetable by pharmacologic modulation of TRPM7 in primary human CML cells, we analyzed constitutive COX-2 mRNA expression by qRT-PCR. CML patient peripheral blood mononuclear cells (PBMCs) were treated with the TRPM7 channel inhibitor NS, the TRPM7 kinase/PI3K inhibitor TG, or the PI3K inhibitors EGA + NEM as controls for TG. Preincubation of patient PBMCs with the TRPM7-inhibitors NS and TG resulted in significant reduction of constitutive COX-2 expression, while EGA + NEM reduced COX-2 nonsignificantly (Figure 4A). In line with the data obtained in the HAP1 cells, the TRPM7 inhibitors NS and TG resulted in a significant reduction of pAKT (Ser473) in CML patient PBMCs, confirmed by the PI3K-inhibitors EGA + NEM (Figure 4B). SMAD2 and ERK1/2 phosphorylation were again not significantly reduced with addition of NS, TG, or controls (Figure S5A and B). To investigate the effect of TRPM7 inhibition on CML cell proliferation, we examined proliferative cell counts of sorted CD34+ patient cells treated with NS, the CML drug Imatinib (IMA) targeting the oncogenic BCR::ABL kinase, or a combination of NS + IMA. An increase in dead cells, and a significant reduction of cell growth with addition of both NS or IMA was observed (Figure 4C and D). Combined treatment of cells with NS + IMA gave a combinatorial/additive effect on blocking CML cell proliferation (Figure 4C and D), suggesting that they act on distinct cellular pathways. HAP1 cells were already resistant to Imatinib (Figure S5C), but were sensitized by the addition of NS, as shown by significant reduction of cell viability upon treatment (Figure S5D). Taken together, our results identify TRPM7 as a proinflammatory regulator of COX-2, particularly affecting constitutive COX-2 expression via modulating AKT signaling, in primary CML cells and CRISPR/Cas9-engineered CML-derived HAP1 clones. We observe combinatorial effects on CD34+ CML cell proliferation via TRPM7 blockade together with Imatinib, a first-line candidate in the current CML treatment regimen. Thus, TRPM7 channel-kinase may provide a promising perspective as an alternative target in COX-2-driven malignant disease, that acts via BCR::ABL independent pathways.
TRPM7 blockade decreases COX-2 expression and proliferation of primary human CML cells. (A) Relative expression of COX-2 mRNA in CML patient-derived PBMCs cultured in the presence of inhibitors, analyzed via qRT-PCR. Inhibitors were compared to DMSO control: TRPM7 channel inhibitor NS8593 (NS, red, 30 µm), TRPM7 kinase/PI3K inhibitor TG100-115 (TG, blue, 20 µm), PI3K inhibitors Eganelisib (EGA, 160 n m) + Nemiralisib (NEM, 100 n m). (B) Basal pAKT (Ser473) levels in CML patient-derived PBMCs, assessed via BioPlex technology. Inhibitors as in A (n = 4–5). (C–D) Proliferation experiment of patient-derived CD34+ CML cells cultured in presence of indicated inhibitors (concentrations as in A). Imatinib was used at 1 µm. Cells were stained with ViaCount live/dead marker. For C, dead cell count was plotted after 3 d incubation. For D, live cell counts were plotted for each indicated day, normalized to respective day 0 (left, mean ± SEM). Cell proliferation was calculated as slope of respective data curves, and compared to DMSO control (D, right, n = 3–5). Statistics: one-way ANOVA (B, D), two-way ANOVA (C), or multiple comparisons test with Dunnett’s correction (A), * P < 0.05, **P < 0.005, ***P < 0.005, and n.s.—not significant. Data are mean ± SD.
Discussion
COX-2 has been studied extensively as a driver of inflammation and co-determinant of malignant progression. Its role as an anti-inflammatory target in clinical practice is undoubted. Chronic inflammation can drive the development and progression of cancer.50,51 In humans, both COX isoforms contribute to inflammatory mechanisms.52 The expression of COX-2, but not COX-1, is elevated in cancer entities including CML.29,53,54 COX-2 enhances survival and proliferation of malignant cells, while negatively influencing antitumor immunity.29,30 COX-2 was found significantly increased in CML patient cells compared to healthy controls, correlating with shorter survival.29 Leukemia cells potentially produce COX-2-dependent factors that enhance angiogenesis and inflammation, correlating with poor prognosis.30 Studies in humans indicate that therapy with COX-2 inhibitors might provide an effective approach to cancer treatment.25,53,54 Conventional nonsteroidal anti-inflammatory drugs (NSAIDs), which inhibit COX-1 and COX-2 are commonly used to treat acute inflammation and proinflammatory diseases.55 However, organ specific adverse effects may occur for more specific COX-2 inhibitors.56,57 Alternate approaches for the functional inhibition of COX-2 are desirable, and cellular mechanisms that modulate COX-2 expression, inflammation,and malignancy are widely relevant. A beneficial effect of COX-2-directed add-on therapy in combination with first-line chemotherapy remains a promising scenario in cancer management.58
TRPM7 is an essential player in immunity and inflammation, regulating differentiation of proinflammatory lymphocytes and macrophages. It also serves as modulator in development of new malignancies, facilitating cell growth, migration, and invasion.2,10 Takahashi et al. reported a connection between TRPM7-mediated Ca2+ entry and growth of the CML-derived cell line K562.28 We confirmed this in primary CML patient cells using the TRPM7 inhibitor NS8593, with a beneficial in vitro outcome on cell proliferation, when combined with CML-directed BCR::ABL kinase inhibitor Imatinib. The targeting of distinct signaling pathways may provide an advantage in combinatorial or alternative approaches to BCR::ABL inhibitors. The latter are prone to resistance mutations, posing a significant challenge to CML therapy59. Our data on the Imatinib-resistant HAP1 cells support these hypotheses. We further show that inhibition of TRPM7 channel-kinase in CML and HAP1 cells results in reduction of constitutive COX-2 expression, with an insignificant effect on COX-1. Previously, TRPM7 has been suggested to induce proinflammatory mediators, including COX-2, in bradykinin-stimulated vascular smooth muscle cells as well as in aldosterone-stimulated HEK-293 cells.60,61 By employing CRISPR/Cas9-edited TRPM7-deficient and kinase-mutated HAP1 cells, we were able to link COX-2 gene expression to TRPM7 kinase-mediated activation of AKT signaling. In murine and human neutrophils, we recently showed that TRPM7 kinase facilitates neutrophil chemotaxis and proinflammatory signatures, by inducing AKT phosphorylation.4 Here, we show that TRPM7 kinase drives cellular signaling in CML cells, via AKT and COX-2. This is independent of TRPM7 ion channel-mediated Ca2+ flux or NFATc1 translocation. We show that TRPM7 kinase can phosphorylate AKT in vitro, via direct phosphorylation. In comparison, we previously found a similar activity range against RHOA protein substrate, using the same assay.22 This is the first report of enzymatic activity of TRPM7 kinase on AKT protein, providing evidence for a potential point of action in the PI3K/AKT signaling pathway directly at the level of AKT.4,23,24 The phosphorylated residue(s) within AKT are yet to be confirmed, and our findings do not rule out additional interactions with PI3K, or other phosphoinositide phosphate-dependent mechanisms in cells. In 2008, Sahni et al. reported a growth arrest of TRPM7-deficient lymphocytes that is rescued by expression of constitutively active PI3K subunit p100.23 Our data indicate an enzymatic interaction of TRPM7 kinase with AKT, downstream of PI3K. For further proof of individual phospho-sites targeted within AKT, one would require follow-up experiments on purified TRPM7 and AKT proteins. While Western blot could give a first glimpse on the residues phosphorylated by TRPM7 kinase, one could also mutate the various phospho-sites of interest. Our Western blot data on cell lysates suggest that TRPM7 kinase affects both phospho-residues required for full AKT activation, that is, Ser473 and Thr308.62 This could position TRPM7 as unique AKT-activating kinase. In the context of leukemia, phosphorylation of AKT (Thr308) has been shown to correlate with high-risk AML,63 and a pronounced phosphorylation at Thr308 correlates with high AKT activity in human nonsmall cell lung cancer.64 Thus, via AKT signaling, TRPM7 kinase is functionally connected with expression of cancer-promoting genes. We have not ascertained if phosphorylation of the transcription factor SMAD2 may additionally mediate TRPM7 kinase-dependent induction of AKT/COX-2 further downstream.3,47
We show a potential benefit of TRPM7 blockade in enhancing CML drug sensitivity effects. A combinatorial effect with Imatinib therapy has been suggested for COX-2 inhibition.58 Beyond its role in CML signaling, inhibition of TRPM7 channel-kinase may positively affect anti-inflammatory signals in a range of pathophysiological conditions, as may be suggested for T-cell- and neutrophil-driven chronic inflammation.3,4 Also, TRPM7 kinase-mutated mice show protection from ischemic stroke,3,16 hence, its pharmacologic blockade may positively affect cardiovascular risks and cancer-associated thrombotic risk. It will be relevant to dissect further roles of TRPM7 in related tissues.
Acknowledgement
We thank Sheila Geiger, Dietmar Hillmann, and Nathalie Wacht for their excellent technical assistance. Michelle Duggan assisted in proofreading.
Author Contributions
B.H., W.N., and S.Z. wrote the manuscript. B.H., W.N. and S.H. performed main experiments. K.H., M.F., N.Z., A.M., V.S., L.F., and L.A. performed additional experiments. D.S., V.C., Ro.G., Ri.G., A.B., I.B., and T.G. provided valuable expertise and feedback. S.Z. conceived and supervised the study. All authors revised the manuscript and agreed on publishing.
Funding
This work was supported by the Deutsche Forschungsgemeinschaft (DFG) grant TRR-152, P15 (to V.C. and T.G.) and P14 (to S.Z.).
Conflict of Interest
The authors declare no conflict of financial interests. S.Z. holds the position of Editorial Board Member for FUNCTION and is blinded from reviewing or making decisions for the manuscript.
Data Availability
Data and materials may be requested by the corresponding author.
References
Author notes
Birgit Hoeger and Wiebke Nadolni contributed equally to this work.
