Physical limits of cell migration: control by ECM space and nuclear deformation and tuning by proteolysis and traction force

Abstract

Cell migration through 3D tissue depends on a physicochemical balance between cell deformability and physical tissue constraints. Migration rates are further governed by the capacity to degrade ECM by proteolytic enzymes, particularly matrix metalloproteinases (MMPs), and integrin- and actomyosin-mediated mechanocoupling. Yet, how these parameters cooperate when space is confined remains unclear. Using MMP-degradable collagen lattices or nondegradable substrates of varying porosity, we quantitatively identify the limits of cell migration by physical arrest. MMP-independent migration declined as linear function of pore size and with deformation of the nucleus, with arrest reached at 10% of the nuclear cross section (tumor cells, 7 µm²; T cells, 4 µm²; neutrophils, 2 µm²). Residual migration under space restriction strongly depended upon MMP-dependent ECM cleavage by enlarging matrix pore diameters, and integrin- and actomyosin-dependent force generation, which jointly propelled the nucleus. The limits of interstitial cell migration thus depend upon scaffold porosity and deformation of the nucleus, with pericellular collagenolysis and mechanocoupling as modulators.

Figure 1.

3D collagen lattices of varying origin and cross-link status: Assembly speed, stiffness, and pore sizes at different polymerization conditions. (A) Assembly speed and (B) organization of acid-extracted pepsinized bovine dermal compared with acid-extracted rat tail collagen matrices after in vitro reconstitution (1.7 mg/ml). (A) Real-time confocal reflection microscopy of fibril formation, represented as kymograph (top panel; B, bovine dermal collagen; R, rat tail collagen) and time-dependent signal intensity curves normalized to the end-point of complete polymerization (bottom panel, horizontal solid line; n = 3 ± SD, shaded areas). Dashed lines, half-maximal polymerization after 8 min (bovine) and 30 s (rat tail, also see inset). (B) Collagen fibril diameter and network geometry, detected by transmission and scanning electron microscopy (TEM, SEM) of fixed dehydrated lattices and confocal reflection microscopy (horizontal, xy; vertical, xz) of native hydrated samples. Marked areas in xz represent signal-free regions referred to as pores. Brackets (TEM) indicate representative fibril calibers. (C) Determination of elastic modulus of bovine and rat tail collagen lattices by AFM. Left, force curves while probing bovine dermal or rat tail collagen lattices (1.7 mg/ml). Dotted curves, medians of all values fitted by the Hertz formula. Oscillatory example curves represent each one of 5–35 repeats. Vertical lines, deformation depth at 1,500 pN applied force. Right, stiffness for different collagen types and concentrations. Unless stated otherwise, polymerization temperature was 37°C. ***, P < 0.0001; **, P < 0.01; *, P < 0.05; n.s., not significant; data points represent 5–35 measurements from different locations as average of 2–20 repetitive tappings/position. Gray horizontal lines, median. (D–F) Pore cross sections quantified by confocal reflection xz sections of bovine dermal or rat tail collagen lattices of different polymerization conditions. Horizontal lines, medians. Data show 1–2 representative experiments from n = 3–4; 21 data points each. ***, P < 0.001; **, P < 0.01; *, P < 0.05; n.s., nonsignificant. (F) Right column, SEM from rat tail collagen lattices assembled at different temperatures. Bars: (A; B, confocal reflection) 10 µm; (B, F, SEM) 1 µm; (B, TEM) 0.1 µm.

Figure 2.

Collagen pore size determines physical migration limits of HT/MT1 cells in 3D collagen lattices. (A, C, and E) Single-cell migration rates in bovine dermal or rat tail collagen lattices of varying concentration or polymerization temperature (1.7 mg/ml) using anchored 3D matrices in migration chambers (shown in Fig. S1 A), monitored by time-lapse microscopy and analyzed by single-cell cell tracking. Migration paths (A and C, left) and averaged single-cell speeds (A and C, right; E) after 24 h of observation in the absence or presence of MMP inhibitor GM6001. Horizontal lines and boxes and whiskers show the medians, 25th/75th, and 5th/95th percentile (50–180 cells, 19–24 h of migration analyzed, 3–6 independent experiments). Dashed lines indicate references for 100% MMP-independent movement and 90% inhibition. ***, P < 0.0001; n.s., not significant. (B, D, and F) Correlation between median pore cross section (from Fig. 1, D–F) and proteolytic and MMP-independent cell speed (A, C, and E) for each collagen condition. Symbols show medians, and whiskers the 5th/95th percentiles. R² (describing how well the regression line approximates real data points), slopes, and significance between slopes from untreated and GM6001-treated populations were as follows: (B) bovine dermal collagen, untreated control 0.998/0.044; GM6001-treated population 0.991/0.040/n.s.; (D) rat tail collagen, untreated control 0.851/0.010, GM6001-treated cells 0.999/0.017/*, P < 0.05 (P = 0.045); (F) rat tail collagen with varied polymerization temperature, untreated control 0.960/0.007, GM6001-treated cells 0.972/0.017/*, P < 0.05 (P = 0.01). Dashed lines, 90% inhibition of migration. Bars: (A and C) 100 µm.

Figure 3.

Requirement for deformation of the nucleus during MMP-independent migration at rate-limiting pore conditions. (A) Time-lapse sequence (left) and population speed (right) of leading edge and main cell body in bovine dermal and rat tail collagen (1.7 mg/ml) in the presence of GM6001. Tracking of the leading extension (white dotted lines) and body (green dotted lines) from starting (*) to end (triangle) position after 5 h. Speed is depicted as median with box blot and whiskers (5th/95th percentile; 45 cells, 3 independent experiments). ***, P < 0.0001; n.s., nonsignificant. (B) Nuclear deformation in moving or migration-arrested nuclei in HT1080 cells expressing H2B/EGFP and cytoplasmic DsRed2 in 6.6 mg/ml bovine dermal collagen in the presence of protease inhibitor cocktail (PI). Overview image from movie sequence (top; Video 5) highlighting moving (green arrowheads) and immobilized nuclei (red arrowheads). Middle: time-lapse example sequences of moving and immobilized nuclei from cells marked in the top image (numbers, time in min). Bottom: diameters of deformed nuclei from independent cells during migration or immobilization. 10 cells from one representative experiment out of n = 2; *, P < 0.05. (C) Morphology of cell body and nucleus during migration in 3D bovine dermal or rat tail collagen matrix (numbers, collagen concentration in mg/ml) in the absence or presence of GM6001. Cultures were fixed 10–16 h after polymerization and stained as indicated, including collagen cleavage neoepitope detection (cyan signal). Nuclear deformation imposed by collagen structures in moving cells in the presence of GM6001 (diameters 4–5 µm; empty arrowheads); conditions of immobilization induce nuclear rounding with occasional thin prolapse (white arrowheads). Arrows, direction of migration and protrusion, respectively. (D) Classification of nuclear morphologies under conditions of cell migration (middle) and immobilization (right). Left: representative xz cross section of a nondeformed HT/MT1 cell (50 µm²). Arrows, nuclear diameters used for quantification in B and E. Numbers show arbitrary ranges for diameters and corresponding cross sections for each phenotype. Red symbols (top-left corner) represent population phenotypes marked in E. (E) Nuclear diameters from cells cultured in collagen matrices of varying pore sizes in the absence or presence of GM6001. After 16 h, samples were fixed, stained with DAPI and phalloidin, and cells with polarized morphologies were analyzed by confocal microscopy for nuclear diameters (compare C). Red dashed ovals, subset of deformed nuclei with hourglass shape associated with movement (middle gray zone) or local prolapse in largely arrested cells (dark gray zone). Black dashed ovals highlight round nuclei of arrested phenotypes (light gray zone). Horizontal lines represent the medians (of each 21 nuclei per group; 2–3 independent experiments). Tables indicate the frequency of hourglass and prolapse phenotypes for each condition. Bars: (A) 50 µm; (B–D) 10 µm.

Figure 4.

Modulation of migration efficiency and nuclear deformation in dense collagen lattices by integrin- and actomyosin-mediated force generation. (A) Dose-dependent reduction of pMLC^T18S19 content in the presence of ROCK inhibitor Y-27632 in HT/MT1 cells. Signal intensity was calculated by densitometry and normalized to the total MLC signal. (B) Dose-dependent inhibition of collagen contraction by anti–β1 integrin mAb 4B4 (left) or Y-27632 (right). Cell-free and cell-containing collagen lattices treated with GM6001 (20 µM), and mAb 4B4, control IgG1 (3 µg/ml), or Y-27632 at indicated concentration. Matrix contraction was measured from gel areas (top images) after 48 h (left) or 24 h (right) as triplicates (means and SD from one representative experiment). Dashed horizontal lines mark 0, 50, and 100% gel contraction. (C) Mean population speed from single-cell analysis after 19–24 h of cell tracking in bovine collagen (1.7 mg/ml) in the absence or presence of MMP inhibitor GM6001 (20 µM) and mAb 4B4 or Y-27632. Medians and boxes and whiskers from 60–150 cells (n = 2–3 for Y-27632; n = 3–5 for 4B4). Dashed lines indicate 100% reference for MMP-independent movement and 90% inhibition. ***, P < 0.0001; **, P < 0.01; n.s., not significant. (D and H) Median elongation (cell body length divided by width) after 10 h of MMP-independent migration (50–90 cells from n = 2; ***, P < 0.0001; n.s., not significant). Horizontal line, median. (E) xy and xz image cross sections of confocal reflectance from bovine collagen fibrils (top) and quantification of pore cross sections (1.7 vs. 0.8 mg/ml concentration). (F and I) Population speed from single-cell analysis in bovine dermal collagen in the presence of 1 µg/ml mAb 4B4 (F) or 2 µM Y-27632 (I) and, where indicated, 20 µM GM6001. 60–108 cells from 2–3 independent experiments; ***, P < 0.0001; **, P < 0.01; *, P < 0.05; n.s., nonsignificant. (G and J) Nuclear diameter analysis as described in the legend of Fig. 3 E. Red dashed ovals, subset of deformed nuclei with hourglass (“h”) shape associated with movement (middle gray zone) or local prolapse (“p”) in largely arrested cells (dark gray zone). r/e, round or ellipsoid nuclei. Arrowheads, deformation of nuclei; arrows, migration direction. Cells migrated in bovine collagen in the presence of either 1 µg/ml 4B4 antibody or 2 µM Y-27632 and, when indicated, 20 µM GM6001. 11–27 cells each; n = 2. Bars: (B) 5 mm; (D) 25 µm; (E, G, and J) 10 µm.

Figure 5.

Pore size–dependent migration and deformation of the nucleus in mononuclear breast cancer cells, T-blasts, and PMN. (A) Lamin A/C content in distinct cell types. Lanes were loaded with whole-cell lysates normalized to GAPDH content and immunoblotted for lamin A/C. (B, D, and F) Migration efficiencies of the indicated cell types in rat tail collagen of different density in the absence or presence of GM6001. Steady-state speeds of single cells monitored over 24 h (MDA/MT1) or 2 h (T-blasts, PMNs). Dashed horizontal lines, subtotal (90% inhibition) and absolute (99% inhibition) migration limit (dashed top lane indicating 100% of MMP-independent migration). Box and whiskers show the medians, 25th/75th and 5th/95th percentiles (50–200 cells, 2–6 independent experiments). ***, P < 0.0001; n.s., not significant. (C, E, and G) Moving and immobilized phenotypes of MDA/MT1 cells, T-blasts, and PMN in the presence of GM6001 in rat tail collagen of different pore sizes (numbers, collagen concentration in mg/ml). Cultures were fixed at the end-point (16 h, MDA/MT1; 2 h, leukocytes) and stained as indicated. Insets, DAPI. Arrows, direction of migration and protrusion, respectively. Tables (C and E), ranges of nuclear cross sections and diameters and their frequencies at different collagen density (numbers in percent, full dataset shown in Fig. S4, C and D) associated with intact or abrogated migration (17–30 independent cells). Insets (G), schematics of different nuclear shapes, including rounded (immobilized), fully or partially unfolded. (H) Change of nuclear morphology in migrating PMN (rat tail collagen, 1.7 mg/ml) over 600-s time period (full sequence shown in Fig. S4 E, example 1; and Video 8). Insets, DAPI signal (left) and schematics of nuclear shape (right). Arrowhead, saltatory migration phase. (I) Nuclear elongation index (top graph) and distance between nucleus and leading extension (pseudopod–nucleus distance) plotted over time. Measurement as indicated by black arrows. Graphs show one representative example of a moving cell (red lines; compare to Fig. S4 E, example 2) and immobilized cell (black lines; compare to Fig. S4 F, example 1) out of n = 3. Bars: (C) 10 µm; (E, G, and H) 5 µm.

Figure 6.

Correlation of migration rates with pore cross section, but not scaffold stiffness. (A) Migration rates as a function of pore cross section in mononuclear tumor cells, T-blasts, and PMN in rat tail collagen of different density in the presence of GM6001. Speed was normalized to the maximum speed in low-density matrix (0.3 mg/ml) as 100% reference, and the subtotal and absolute migration limits are indicated (dotted lines). Data points depict medians from Figs. 1 E, 2 D, 5 (B, D, and F); S2 G, and S4 A. For T-blasts and PMN, data points were included for additional pore size conditions, n = 2–4. Subtotal limits were reached at median pore cross sections of 4–7 µm² for tumor cells and 3–4 µm² for leukocytes, and absolute limits at 2 µm² for leukocytes. (B) Association of migration rates and elastic modulus of 3D collagen lattices. Speed medians were used from Fig. 2 (A, C, and E) obtained for cells migrating in the presence of GM6001 in lattice conditions for which the elastic modulus was measured (compare Fig. 1 C). R² for linear regression (black line) was 0.2339 (nonsignificant).

Figure 7.

Physical migration limits of mono- and polymorphonuclear cell types in transwell filter assay. (A) Transmigration of different cell types through polycarbonate membranes of 5-, 3-, 1-, and 0.45-µm pore diameter after 20 h (HT/MT1, HT/wt, and MDA/MT1 cells), 5 h (T-blasts) or 3 h (PMN). The cell number in the bottom compartment was normalized to transmigration through 5-µm pores as 100% reference. Solid and dashed vertical lines represent the pore cross section at 90 and 99% inhibition, respectively, of migration for each cell type (black lines, mononuclear cells; gray area between black lines depicts a range of subtotal migration limits; red lines, PMN). Subtotal limits were reached at 7–10 µm² for tumor cells and T-blasts, and 4 µm² for PMNs; and absolute limits at 5 µm² for tumor cells and T-blasts and 1–2 µm² for PMNs (n = 3–5; means, SD). (B) Polycarbonate membranes with pore sizes ranging from 8 to 1 µm in diameter (top row). Representative images of nuclear deformation in transmigrating dual-color HT1080 cells and (C) PMNs during passage through pores of decreasing size. Filter membranes were fixed and, in C, stained as indicated (n = 2). Arrows, direction of transmigration. Bars: (B) 10 µm, (C) 5 µm.

Figure 8.

Tissue porosity and nuclear deformation jointly determine subtotal and absolute limits of cell migration. Cell migration efficacy is a joint function of substrate porosity and nuclear deformability, with MMP activity and mechanocoupling as modulators. (A) Minor deformation of nuclei and optimal velocity during migration through tissue of sufficient to high porosity, with mononuclear shapes in ellipsoid and polymorphonuclear nuclei in partly unfolded state. (B) Significant deformation of the nucleus and migration delay in cells moving through dense ECM with pore cross sections much below the nondeformed nucleus size. Mononuclear nuclei are compressed in their entirety and adopt hourglass-like or cigar-like shapes, whereas polymorphonuclear deformation consists of unfolding with transient pearl chain–like configuration as maximum. Maximal deformation is reached when the nucleus deforms to ∼10% of its original cross section, reaching a subtotal migration limit. (C) Migration arrest during confrontation with pore cross sections that exhaust the deformation capability of the nucleus. Both mono- and polymorphonuclear nuclei of migration-arrested cells retain a roundish, collapsed morphology with nonproductive transient prolapse, the cross section of which matches the geometry of the pore (∼5–10% of the original nuclear cross section). Modulators of migration rates near the physical limit include (1) pericellular proteolysis by MMPs that widens pore cross sections and (2) mechano-coupling toward ECM determining the force with which the nucleus is transported. CS, cross section; CR, compression ratio (degree of nuclear deformation versus nondeformed state); Tu, tumor cells; T, T blast; PMN, polymorphonuclear neutrophil. White arrows, direction of migration.