Inhibition of the Sodium-Potassium-Chloride Cotransporter Isoform-1 reduces glioma invasion

Abstract

Malignant gliomas metastasize throughout the brain by infiltrative cell migration into peritumoral areas. Invading cells undergo profound changes in cell shape and volume as they navigate extracellular spaces along blood vessels and white matter tracts. Volume changes are aided by the concerted release of osmotically active ions, most notably K(+) and Cl(-). Their efflux through ion channels along with obligated water causes rapid cell shrinkage. Suitable ionic gradients must be established and maintained through the activity of ion transport systems. Here, we show that the Sodium-Potassium-Chloride Cotransporter Isoform-1 (NKCC1) provides the major pathway for Cl(-) accumulation in glioma cells. NKCC1 localizes to the leading edge of invading processes, and pharmacologic inhibition using the loop diuretic bumetanide inhibits in vitro Transwell migration by 25% to 50%. Short hairpin RNA knockdowns of NKCC1 yielded a similar inhibition and a loss of bumetanide-sensitive cell volume regulation. A loss of NKCC1 function did not affect cell motility in two-dimensional assays lacking spatial constraints but manifested only when cells had to undergo volume changes during migration. Intracranial implantation of human gliomas into severe combined immunodeficient mice showed a marked reduction in cell invasion when NKCC1 function was disrupted genetically or by twice daily injection of the Food and Drug Administration-approved NKCC1 inhibitor Bumex. These data support the consideration of Bumex as adjuvant therapy for patients with high-grade gliomas.

Figure 1. NKCC1 is expressed in several glioma cell lines

A, Immunofluorescent images of D54 and U87 glioma cells labeled with an antibody against NKCC1. The staining was repeated in triplicate three independent times. Scale bar, 20 µm. B, Example Western blot showing NKCC expression in whole cell lysates of HEK293 (positive control), U87, and D54 cell lines with actin serving as a loading control. Full-length blots are presented in Supplemental Figure 1.

Figure 2. Inhibition of NKCC1 with bumetanide reduces 3-D migration when space is limited

A, Representative DIC images of D54 glioma cells that have migrated through an 8-µm or 3-µm Transwell barrier, for 5 or 12 h respectively, with or without bumetanide. Scale bar, 50 µm. B, B1, Quantification of D54 glioma cell relative percent migration of on 8- and 3-µm pore Transwell barriers. B2, Quantification of U87 glioma cell relative percent migration on 8- and 3-µm pore Transwell barriers. Unpaired t-test for 8- and 3-µm pore Transwells. Experiments were performed in triplicate and repeated five independent times. C, Representative images of U87 glioma cells at 0 and 8 h of 2-D migration in the presence or absence of 200 µM bumetanide. Scale bar, 100 µm. D, D1, Quantification of percent wound closure for both D54 and U87. The experiment was performed in duplicate and repeated at least three independent times. Unpaired t-test, p>0.05, for D54 and U87. D2, D54 glioma cell proliferation in the presence or absence of the NKCC1 inhibitor, bumetanide, at four different concentrations. The experiment was completed in triplicate and repeated three independent times. One-way ANOVA, p>0.05

Figure 3. NKCC1 localizes to the membrane on the leading edge of cells migrating across Transwell barriers

The first row contains representative merged and individual channel images obtained at 40×. White box indicates digital zoom images in the second and third rows. The second row shows little to no NKCC1 (green) and phalloidin (red) co-localization on the lagging edge of a migrating glioma cell while the leading edge of the same cell, as depicted in the third row, demonstrates NKCC1 expression on the plasma membrane. DAPI nuclear stain (blue). Arrows designate cross-sections of cells seen in three-view through the imaging plane.

Figure 4. Genetic knockdown of NKCC1 has no effect on proliferation but eliminates bumetanide-sensitive migration

A, Representative Western blot of whole cell lysates of HEK293 (positive control), U87 and D54 glioma cells, and D54 glioma cells stably transfected with NKCC1 knockdowns (NS, scrambled shRNA; 27, 141, 382, 662, 690, shRNAs each targeting a different portion of the SLC12A2 gene). Actin served as a loading control. B, Quantification of NKCC1-knockdown cell line protein levels normalized to loading controls. Experiments were performed in duplicate and repeated four independent times. One-way ANOVA, p<0.001, Tukey-Kramer post-hoc test. C, NKCC1-knockdown cell line proliferation during four days. Experiments were performed in triplicate and repeated five independent times. One-way ANOVA, p>0.05. D, Quantification of NKCC1-knockdown cell line relative percent migration across a 3-µm pore Transwell barrier. Experiments were performed in triplicate and repeated five independent times.

Figure 5. Functional RVI is inhibited by NKCC1 blockade with bumetanide or genetic knockdown

A, Left, Normalized MCVs (n=10,000-20,000 cells) of D54 glioma cells undergoing RVI after hyperosmotic (15 mM NaCl) challenge in the presence or absence of bumetanide. Experiments were performed four independent times. Right, D54 glioma cell normalized MCVs at 40 minutes post hyperosmotic challenge. B, Normalized MCVs for NS, 662, and 690 lines undergoing RVI after hyperosmotic challenge in the presence or absence of bumetanide. Experiments were performed four (662) or five (NS, 690) independent times. Bottom right, Knockdown cell line normalized MCVs at 40 minutes post challenge. One-way ANOVA, p<0.05, Tukey-Kramer post-hoc test.

Figure 6. Bumetanide or NKCC1 genetic knockdown inhibits in vivo glioma cell invasion but not tumor size

A, Examples of 30 µm brain slices with xenografted tumor tissue stained with H&E from vehicle or bumetanide treated mice. Upper panels show whole brain slice (black scale bar, 1 mm). Lower panels show 4× magnification routinely used for calculating volume (white scale bar, 100 µm). B, Immunofluorescent images of NS and 662 tumors in brain slices with cells that have migrated at various distances away from the main tumor mass. Scale bar, 100 µm. C, Quantification in vivo tumor volumes for D54-eGFP tumors (C1) and NKCC1-knockdowns (C2). Tumor volumes between vehicle (n=4) and bumetanide (n=7) treated tumors (C1) and between NS (n=7) and 662 (n=8) tumors (C2) were not significantly different. Unpaired t-test, p>0.05. D, Tumor cell average distance migrated from primary tumor mass in vehicle (n=390) and bumetanide (n=194) treated mice (D1) and knockdowns (D2), NS (n=4591) and 662 (n=4690). D3, Distribution of invaded D54-eGFP tumor cells from both vehicle and bumetanide treated tumors. The decay constants of the exponential fit curves were approximately 295 and 160 for vehicle and bumetanide treated groups respectively. D4, Distribution of invaded tumor cells from both NS and 662 tumors. The decay constants of the exponential fit curves were approximately 154 and 112 for NS and 662, respectively (For full equation values, see Supplementary Table 1).