LKB1 loss in melanoma disrupts directional migration toward extracellular matrix cues

Abstract

Somatic inactivation of the serine/threonine kinase gene STK11/LKB1/PAR-4 occurs in a variety of cancers, including ∼10% of melanoma. However, how the loss of LKB1 activity facilitates melanoma invasion and metastasis remains poorly understood. In LKB1-null cells derived from an autochthonous murine model of melanoma with activated Kras and Lkb1 loss and matched reconstituted controls, we have investigated the mechanism by which LKB1 loss increases melanoma invasive motility. Using a microfluidic gradient chamber system and time-lapse microscopy, in this paper, we uncover a new function for LKB1 as a directional migration sensor of gradients of extracellular matrix (haptotaxis) but not soluble growth factor cues (chemotaxis). Systematic perturbation of known LKB1 effectors demonstrated that this response does not require canonical adenosine monophosphate-activated protein kinase (AMPK) activity but instead requires the activity of the AMPK-related microtubule affinity-regulating kinase (MARK)/PAR-1 family kinases. Inhibition of the LKB1-MARK pathway facilitated invasive motility, suggesting that loss of the ability to sense inhibitory matrix cues may promote melanoma invasion.

Figure 1.

Loss of LKB1 does not affect invadopodia formation in melanoma cells. (A) Western blot showing lentiviral shRNA knockdown of a nontargeting sequence (NS) or LKB1 (knockdown [KD]) in human A2058 (BRAF^V600E/RB null) and mouse GR285 (Kras^G12D/p53 null/p16 null) melanoma. (B) Invadopodia formation on Alexa Fluor 647–gelatin. GFP is a marker of knockdown, and cells were stained with Alexa Fluor 568–phalloidin to label actin. (C) Percentage of cells with invadopodia shown as means ± SEM. (A2058: nontargeting sequence, n = 297; KD#1, n = 413; KD#2, n = 300. GR285: nontargeting sequence, n = 321; KD#1, n = 255; KD#2, n = 312.) (D) Live-cell invadopodia assay shows no difference in invadopodia formation over time. Dashed lines show raw data and solid lines are best-fit curves to the data. (Trial #1: nontargeting sequence, n = 60; LKB1 knockdown, n = 60. Trial #2: nontargeting sequence, n = 82, LKB1 knockdown, n = 101.) (E) Invadopodia assay on Alexa Fluor 405–gelatin of WM-266-4 (BRAF^V600D/PTEN null) Lifeact (LA)-GFP cells and tdTomato-labeled LKB498 (Kras^G12D/Lkb1 null) or LKB1 addback melanoma. Cells were stained with Alexa Fluor 647–phalloidin to label actin. Insets show magnified region of invadopodia. Bars, 20 µm.

Figure 2.

LKB1 limits single-cell and collective cell migration in melanoma. (A) Single-cell speed of LKB498 cells in 1 mg/ml 3D collagen. (null, n = 73; addback, n = 83.) (B) Spheroid invasion of LKB498 cells into 3.2 mg/ml 3D collagen. Cells were stained with Alexa Fluor 568–phalloidin to label actin and Hoechst 33342 to label nuclei. (C) Spheroid outgrowth of LKB498 cells into increasing concentrations of 3D collagen. (1 mg/ml: null, n = 22; addback, n = 24. 2 mg/ml: null, n = 21; addback, n = 27. 4 mg/ml: null, n = 30; addback, n = 29. 6 mg/ml: null, n = 20; addback, n = 23. 8 mg/ml: null, n = 28; addback, n = 23.) (D) Single-cell speed of LKB498 cells on increasing fibronectin concentrations. (0.1 µg/ml: null, n = 94; addback, n = 104. 1 µg/ml: null, n = 99; addback, n = 96. 10 µg/ml: null, n = 149; addback, n = 162. 100 µg/ml: null, n = 134; addback, n = 150.) (E) Montage images and edge outlines of time-lapse scratch wounds. (F) Relative wound area over time. (LKB498: null, n = 43; addback, n = 32. TKL2: null, n = 33; addback, n = 33.) Gray shaded area indicates times at which relative wound area shows significant differences with P < 0.01. Insets show magnified plots of time points at which pausing occurs (represented by the bars) at the wound boundary. (G) Scratch wound of LKB498 LKB1 addback monolayer on Cy3-fibronectin (FN). DIC, differential interference contrast. All data are shown as means ± SEM. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001 by two-tailed unpaired t test. Bars, 100 µm.

Figure 3.

LKB1 is necessary for haptotaxis but is not required for directional migration to PDGF. (A) Schematic of mixed fluorescent cells in microfluidics chamber system. (B) LKB1 is required for haptotaxis of LKB498 cells to fibronectin. (null, n = 99; addback, n = 75.) (C) Western blots showing phosphorylation of FAK in LKB498 cells upon suspension (Susp.) or adhesion (Adh.) to fibronectin. Add, addback. (D) Epifluorescence image of 3D haptotaxis chamber with Alexa Fluor 647–fibronectin (FN) gradient across 1 mg/ml 3D collagen. Inset shows a confocal image of fibronectin bound to collagen fibrils. (E) LKB1 is necessary for haptotaxis of LKB498 cells in 3D collagen. (null, n = 132; addback, n = 89.) (F and G) Spheroid invasion of TKL2 cells into 3D collagen is cell autonomous. (null, n = 118; addback, n = 120.) Data are shown as a box and whisker plots with the boxes extending from the 25th through 75th percentiles with the lines in the middle as the medians. Whiskers range from the minimum to maximum values. **, P < 0.01 by two-tailed unpaired t test. (H) LKB1 is dispensable for chemotaxis of LKB498 cells to PDGF. (null, n = 34; addback, n = 39.) (I) Western blots showing phosphorylation of AKT upon acute stimulation with 200 ng/ml PDGF. All haptotaxis and chemotaxis data are shown as mean ± 95% confidence intervals. ***, P < 0.001 by two-tailed unpaired t test. Bars, 100 µm. v, velocity; D/T, displacement (D) over the total path length (T).

Figure 4.

Membrane targeting and kinase activity of LKB1 are required for haptotaxis. (A) Sequence alignment of vertebrate LKB1 C terminus. Black shade indicates identical amino acids, and gray shade indicates similar amino acids. (B) Western blots of LKB1 reconstitution in LKB498 cells with wild type (WT) or prenylation (C430A), phosphorylation (S428A), and kinase-dead (K78I) mutants. Add, addback. (C) Membrane association and kinase activity are required for haptotaxis. (addback, n = 52; C430A, n = 37; S428A, n = 68; K78I, n = 42.) All data are shown as mean ± 95% confidence intervals. *, P < 0.05 by two-tailed unpaired t test. v, velocity; D/T, displacement (D) over the total path length (T).

Figure 5.

MARK family kinase activity is necessary for haptotaxis. (A) Several LKB1 substrates are not required for haptotaxis. Substrates not expressed were not tested in haptotaxis experiments and are listed as not available (N/A). (B) MARK family kinases are required for haptotaxis in LKB498 cells. (addback, n = 125; addback + MARK2, 3, and 4 knockdown [KD], n = 106.) (C) Western blots showing MARK inhibitor (MARKinh) 39621 (20 µM) validation in LKB498 cells. Add, addback; IP, immunoprecipitation; IB, immunoblot. (D) MARK inhibitor blocks haptotaxis in LKB498 cells. (addback, n = 133; MARK inhibitor, n = 92.) (E) Western blots validating effect of dominant-negative (DN) MARK3 overexpression in LKB498 cells. (F and G) DN-MARK3 blocks haptotaxis of LKB498 cells in 2D (F) and 3D (G). (2D: addback, n = 130; DN-MARK3, n = 131. 3D: addback, n = 48; DN-MARK3, n = 40.) All data are shown as mean ± 95% confidence intervals. *, P < 0.05; **, P < 0.01; ***, P < 0.001 by two-tailed unpaired t test. v, velocity; D/T, displacement (D) over the total path length (T).

Figure 6.

Haptotaxis does not require MAP4 or microtubules. (A) Western blots showing overexpression of active MAP4 in LKB498 cells. exp., exposure. (B and C) Active MAP4 does not affect haptotaxis of LKB498 cells in 2D (B) and 3D (C). (addback, n = 51; GFP-MAP4-S914A/S1046A [SA], n = 48.) (D) Nocodazole depolymerizes microtubules in LKB498 cells. Bar, 20 µm. (E) Nocodazole (noc) treatment of LKB498 LKB1 addback cells has no effect on haptotaxis. (640 nM, n = 37; 10.24 µM, n = 35.) (F and G) Washing of nocodazole alone (F) does not block haptotaxis (before wash in, n = 69; wash in, n = 51) but together with MARK inhibitor (G) blocks haptotaxis in IA32 fibroblasts. (before wash in, n = 155; wash in, n = 107.) All haptotaxis data are shown as mean ± 95% confidence intervals. **, P < 0.01; ***, P < 0.001 by two-tailed unpaired t test. v, velocity; D/T, displacement (D) over the total path length (T).

Figure 7.

MARK inhibition facilitates cell invasion. (A) Western blot of overexpression of DN-MARK3 in TKL2 cells shows no effect on AMPK phosphorylation. pACC, phosphorylated acetyl-CoA carboxylase. (B) Quantification of Western blot data (n = 3) in A is shown as means ± SEM. **, P < 0.01; ***, P < 0.001 by two-tailed unpaired t test. (C) DN-MARK3 enhances spheroid invasion into 4 mg/ml 3D collagen. Cells were stained with Alexa Fluor 568–phalloidin to label actin. Bar, 100 µm. (D) Quantification of data in C is shown as means ± SEM. (day 0–3: null, n = 14; addback, n = 9; addback + DN-MARK3, n = 22.) **, P < 0.01; ***, P < 0.001 by one-way analysis of variance (ANOVA) by Dunnett’s post-test, as compared with addback.