Compensation mechanism in tumor cell migration: mesenchymal-amoeboid transition after blocking of pericellular proteolysis

Abstract

Invasive tumor dissemination in vitro and in vivo involves the proteolytic degradation of ECM barriers. This process, however, is only incompletely attenuated by protease inhibitor-based treatment, suggesting the existence of migratory compensation strategies. In three-dimensional collagen matrices, spindle-shaped proteolytically potent HT-1080 fibrosarcoma and MDA-MB-231 carcinoma cells exhibited a constitutive mesenchymal-type movement including the coclustering of beta 1 integrins and MT1-matrix metalloproteinase (MMP) at fiber bindings sites and the generation of tube-like proteolytic degradation tracks. Near-total inhibition of MMPs, serine proteases, cathepsins, and other proteases, however, induced a conversion toward spherical morphology at near undiminished migration rates. Sustained protease-independent migration resulted from a flexible amoeba-like shape change, i.e., propulsive squeezing through preexisting matrix gaps and formation of constriction rings in the absence of matrix degradation, concomitant loss of clustered beta 1 integrins and MT1-MMP from fiber binding sites, and a diffuse cortical distribution of the actin cytoskeleton. Acquisition of protease-independent amoeboid dissemination was confirmed for HT-1080 cells injected into the mouse dermis monitored by intravital multiphoton microscopy. In conclusion, the transition from proteolytic mesenchymal toward nonproteolytic amoeboid movement highlights a supramolecular plasticity mechanism in cell migration and further represents a putative escape mechanism in tumor cell dissemination after abrogation of pericellular proteolysis.

Figure 1.

Proteolytic migration of HT-1080/MT1 fibrosarcoma cells in 3D collagen lattice. (A) Morphology of HT-1080/MT1 cell migrating through 3D collagen lattice monitored by video microscopy. (B) Reduction of migration speed (black bars) and induction of detached, nonmobile spherical morphology (inset) by adhesion-perturbing anti–β1 integrin antibody 4B4. (C) Confocal backscatter (matrix fibers) and fluorescence of MT1-MMP (red), β1 integrins (green), and colocalization (yellow; arrowheads) at fiber traction zone of the leading edge. (D) 3D reconstruction of a calcein-stained migrating cell by time-lapse confocal microscopy and (E) backscatter signal of the same cell from the central section (time, 60 min). Fiber bundling (white arrowheads), deposition of cell fragments (black arrowhead), and newly formed matrix defect (asterisk). Black arrows, direction of migration. Bars, 20 μm.

Figure 2.

Protease expression in HT-1080/MT1 cells and inhibition of collagenolysis. (A) mRNA expression of MMPs, ADAMs, cathepsins, and serine proteases detected by RT-PCR. Asterisks, proteases cleaving native type I collagen. (B) Degradation of 3D fibrillar collagen by HT1080/MT1 cells (500,000 cells) layered on top of a 1-mm-thick collagen matrix in the absence (1) or presence (2) of protease inhibitor cocktail. (B, 3) Confocal backscatter of lysis zone bordered by clumped collagen, as indicated by the square in (1). (B, 4) Negative control matrix overlaid with cell-free medium only. (C) Cell contact to FITC-labeled collagen fibers; confocal reflection (gray) and FITC fluorescence (green). Bar, 20 μm. (D) Migration-associated collagenolysis caused by HT1080/MT1 cells within 3D FITC–collagen lattices was quantified from the FITC release after 40 h of migration in the absence or presence of inhibitors. ***, P < 0.001; two-tailed t test for independent means, difference to control cells. As negative control, T cells did not release FITC above background levels (not depicted).

Figure 3.

Sustained migration in the presence of protease inhibitor cocktail. (A) Cell tracking analysis of the steady-state population speed ± SD and (B) mean speed for each individual cell (time, 20 h; n = 3 experiments, 120 cells).

Figure 4.

Transition of spindle-shaped (mesenchymal) to more spherical (amoeboid) migration in HT1080/MT1 and MDA-MB-231 cells in the presence of protease inhibitor cocktail. (A) Conversion of elongated (left) toward spherical shape (right) in HT1080/MT1 cells, and (B) higher magnification of an amoeboid migrating cell in the presence of inhibitor cocktail. Time in B, 117 min. (C) Median elongation (calculated from length divided by width) in the absence and presence of protease inhibitor cocktail (n = 3; 170 cells; ***, P < 0.0001). (D) Inhibition of collagen degradation by MDA-MB-231 cells by protease inhibitor cocktail (n = 3; P < 0.05, unpaired two-tailed t test). (E) Conversion from spindle shaped (left) to more spherical morphology (right), and (F) reduced median elongation in the presence of protease inhibitors in MDA-MB-231 cells (n = 3; 200 cells; ***, P < 0.0001). (G) Frequency of mesenchymal and amoeboid shape in actually migrating cells in the absence (▪) and presence (□) of protease inhibitor cocktail (HT-1080/MT1 cells, n = 3, 100 cells; **, P < 0.001 for difference to untreated control; two-tailed t test for independent means). Cells of indeterminate morphology (15–40%; for details see the Materials and methods) were excluded from analysis. Bars: (A and E) 100 μm; (B) 20 μm.

Figure 5.

Cellular mechanism of nonproteolytic movement within 3D collagen matrix and related changes in β1 integrin, MT1-MMP, and F-actin distribution in HT-1080/MT1 cells. (A) Induced amoeboid migration lacking fiber degradation. Alignment of cell body along a fiber strand (white arrowheads) and intact individual collagen fiber at its original position after cell detachment (black arrowhead). (B) Migratory alignment of the cell depicted in A along the preexisting fiber scaffold. The outline of the cell edge at 2.5-min time intervals (blue lines) was superimposed onto the 3D reconstruction of the transmigrated matrix structure. Bright pixels indicate colocalization of cell boundary and fibers (arrowheads). (C) Migration through a narrow gap bordered by fibers (black arrowhead) resulting in morphological adaptation and the formation of a constriction ring. (D) Reduced F-actin and β1 integrin focalization at fiber binding sites in an amoeboid HT1080/MT1 cell, compared with F and Fig. 1 C. Because of constriction caused by a perpendicular collagen fiber, this cell contains a lobulated main body. Black arrowhead, uropod. (E) Loss of clustered MT1-MMP and β1 integrins from interactions with fibers in induced amoeboid migration. Arrowheads indicate two simultaneous constriction rings bordered by perpendicular collagen fibers. (F) F-actin and β1 integrins in an untreated control cell of elongated shape. Time: (A) 35 min; (B) 65 min; (C) 20 min. Bars, 20 μm. Black arrows, direction of migration.

Figure 6.

In vivo translocation and morphology of HT-1080/MT1 cells in the mouse dermis. Intravital microscopy of ECM structure (light brown) and labeled cells 3 h (A–D) and 8 h (B) after injection. (A) 3D reconstruction of injection site (asterisk), more distant cells passively scattered around the injection site, and matrix structure at low magnification (10× objective). Control cells (green) and cells pretreated with protease inhibitor cocktail (red). Asterisk, center of injection site. Black arrows, multicellular cord. (B) Position change of control cells (top) and cells pretreated with protease inhibitor cocktail (bottom) from the injection site depicted in A. False-color reconstruction was obtained for each fluorescence channel 3 h (green) and 8 h (red) after injection. Orthotopic superimposition was controlled by the position of coinjected fluorescent beads (circles) as well as the multicellular cord. White arrowheads, regions of cell translocation (right). (C) High resolution imaging of cell morphology and (D) elongation of control cells (green) and cells pretreated with protease inhibitor cocktail (red) at least 2 h after injection (40× objective). Elongation was calculated from 170 cells, six independent experiments; ***, P < 0.0001. Bars: (A and B [left]) 200 μm; (B [right] and C) 100 μm.

Figure 7.

Concept on mesenchymal–amoeboid transition. A program of morphodynamic and molecular changes after abrogation of pericellular proteolysis by protease inhibitors provides a supramolecular mechanism for persistent nonproteolytic migration in 3D fibrillar collagenous tissues.