In Vitro Characterization and Micromechanics of Tumor Cell Chemotactic Protrusion, Locomotion, and Extravasation

Materials

Soluble CIV was obtained from Collaborative Research (Bedford, MA). For chemotactic assays, the collagen was diluted to various concentrations in Dulbecco’s modified Eagle’s medium (DMEM) (Biofluids, Gaithersburg, MD) with 0.1% bovine serum albumin (BSA) (Sigma, St. Louis, MO), and the solution was brought to a pH of 7.4 and an osmolality of 300 mmol/kg. Mouse monoclonal antibodies to human CD11a (LFA-1), CD11b (Mac-1), and CD-54 (ICAM-1) were purchased from CalTag Laboratories (Burlingame, CA). A mouse fibroblast L-cell line transfected with both human E selectin and ICAM-1 molecules (“EI cells”)⁵ was provided by Dr. Simon at UC Davis, CA.

Cell Culture

A2058 and C8161 human melanoma cell lines were provided by the National Cancer Institute (MD) and Penn State Hershey Medical Center (PA), respectively. Cells were cultured in DMEM (for A2058 cells) or DMEM-F12 (for C8161 cells) supplemented with 10% fetal bovine serum (FBS) (Biofluids) within an atmosphere of 5% CO₂/95% air at 37 °C. Cells for assays were detached while subconfluent by brief exposure to 0.05% trypsin/0.02% EDTA and allowed to recover for 1 h in a tissue culture medium containing DMEM/10% FBS.

T-leukemic Jurkat cells (provided by Dr. Lawrence, University of Virginia, VA) were cultured in suspension of RPMI 1640 medium supplemented with 10% FBS, L-glutamine (0.292 mg/ml), penicillin (100 U/ml), and streptomycin sulfate (100 µg/ml) (Gibco Laboratories, Grand Island, NY). For assays, Jurkat cells were resuspended in serum-free medium with 0.1% BSA. To characterize the effect of an increase in LFA-1 avidity on Jurkat cells, cells were pretreated with 50 ng/ml phorbol 12-myristate 13-acetate (PMA) (Sigma) for 20 min at 37 °C.

Dual Micropipette Assay

Experiments were performed with individual A2058 tumor cells. Pipettes were pulled to desired dimensions using a pipette puller (Narishige, Japan) and a micro-forge (Stoelting, Wood Dale, IL). A single A2058 cell (15 µm in diameter ±1.0 µm) was retained in a larger pipette [14 µm inner diameter (i.d.) ±1.5 µm] containing FBS-free DMEM/0.1% BSA. The chemotactic solution with CIV was prepared to desired concentrations (100–200 µg/ml) and filled into a smaller pipette, in which the pressure gradient at the pipette tip was controlled at zero. This small pipette (10 µm outer diameter ±1.5 µm) was then inserted into the larger pipette at a controlled distance (δ) between the tip of the small pipette and the cell front surface (Fig. 2). An initial concentration of the chemoattractant C₀ and δ provided a gradient to influence cell chemotaxis. Cell images were taken by a video camera and recorded for off-line analyses. Measurements on cell pseudopod protrusion length (L_p; a two-dimensional projected distance from the pseudopod tip to the base) and locomotion velocity (dL_c/dt; velocity of the cell center) were made at a later time. The pseudopod protrusion frequency (f_p), defined as the total number of peaks in protrusion length per given time period, was also measured.

Chemotactic Diffusion

The CIV chemotactic source (at x=0) was assumed to be constant (C₀) (for all time t≥0). CIV diffuses from x=0 toward the cell surface at x=L(t) when t>0. Let C(x,t) be the spatial-temporal distribution of the CIV concentration between the tip of the small pipette (x=0) and the moving cell surface [x=L(t); Fig. 2], the diffusion equation for C(x,t) was given by

where D is the diffusion constant for macromolecules within the medium. A steady chemotactic source was assumed at the small pipette tip with a constant concentration C₀, such that the initial condition became

and the boundary conditions were

in which K_r is a constant, representing the binding rate of the CIV molecules bound to the melanoma cell surface receptors. A moving boundary L(t)=δ−L_C(t) was assumed to satisfy the following condition:

in which K_g is a constant correlating the speed of a moving boundary with the chemotactic concentration gradient. Due to the moving boundary at x=L(t), the concentration C{L(t),t} became a key variable in determining the actual chemotactic process of cells at subsequent times.

Collagen IV Binding to Melanoma Cells

Detached A2058 cells were incubated in DMEM/0.1% BSA at 37 °C with a chosen concentration of CIV At the end of a chosen time, cold medium (DMEM/0.1% BSA; 1–2 °C) was immediately added to the cell suspension to minimize the CIV reaction with the cells, followed by a centrifugation at 200 g for 3 min (1–2 °C). After decanting the supernatant, the cells were resuspended at 1 × 10⁷ cells/ml in PBS/CS/GS (phosphate buffered saline/5% calf serum/2% goat serum/0.1% N_aN₃). Primary antibody, 1 µg of antimouse CIV was added per 1 × 10⁶ cells and incubated for 25 min at 4 °C. Cells were washed and tagged with phycoerythrin-conjugated antirabbit immunoglobulin antibody (1 µg/1×10⁶ cells/100 µl) (Jackson Immuno Research Laboratories, West Grove, PA) for 25 min at 4 °C. The cells were rewashed and fixed in PBS containing 1% paraformaldehyde. The number of cells binding detectable levels of CIV was determined by flow cytometry. Appropriate controls of cells incubated with secondary antibody alone or cells incubated with primary and secondary antibodies but without CIV were included. Under saturating antibody conditions, approximately 100% of the cells which had been incubated with CIV were detected. To determine the number of CIV molecules bound per cell absolute fluorescent values were calculated from a standard curve generated by running Sphero Rainbow 30-5 calibration particles (Spherotech, Libertyville, IL), a mix of five beads with known fluorescence values. In different experiments, the conditions were varied such that the concentrations of CIV ranged from 0 to 200 µg/ml and the time of incubation with collagen ranged from 1 to 50 min. However, because the washes took at least 3 min, we were unable to test actual binding times below about 3 min.

Adhesion Molecule Detection

The cells of interest were detached and prepared as described earlier. Samples were treated with murine anti-human CD Marker primary antibodies (LFA-1, Mac-1, ICAM-1; CalTag Laboratories) for 25 min at 4°C (1 µg mAb/10⁶ cells). The cells were then treated with secondary antibody, FITC-conjugated goat antimouse IgG F(ab)₂ fragment, for 25 min at 4 °C (Jackson ImmunoResearch, 1 µg/10⁶ cells). The samples were fixed with 2% formaldehyde (Sigma) and at least 10,000 cells were analyzed per trial using Coulter EPICS XL (Coulter Corp.) flow cytometer. The data are represented as the mean florescence divided by the background florescence with the error bar representing the standard error of the mean (SEM). Each data point represents at least five trials.

Flow-Migration Chamber Assay

In order to simulate the in vivo tumor extravasation process, a chemotaxis flow-migration chamber (Fig. 3) was developed to model a capillary and to characterize tumor cell adhesion to a cultured substrate-cell monolayer representing binding to a vessel wall. Under the influence of shear flow and cell-substrate adhesion, tumor-cell extravasation from the top compartment into the bottom compartment simulated cell migration from the capillary lumen into interstitial tissue space.

This in vitro flow-migration device is a modified-48 well chemotactic chamber. It was fabricated from polycarbonate [Fig. 3(A)]. The gasket, 0.02 in. PharmElast silicon (SF Medical, Hudson, MA), is an 11 cm×5.5 cm rectangle with a 7 cm×2 cm opening cut from the center separating the top and base plates, which creates a rectangular cross-sectional flow field with a large width (~2 cm)-to-height (~0.051 cm) aspect ratio [Fig. 3(B)]. Applying a prescribed shear rate (γ˙, s⁻¹) under a given volumetric flow rate Q:

where w and h are the width and height of the flow channel, respectively. The wall shear stress (τ_w; dyn/cm²) exerted on an adherent cell was calculated by

where μ is the viscosity of the suspension fluid. A flow loop consisting of two reservoirs and the chamber top plate (with inlets and outlets) was connected using Nalgene PVC tubing (~1/16 in. i.d.). A peristaltic pump (Cole Parmer) drove a constant flow through the flow loop. A sealed reservoir was added within the loop to eliminate pulsatility from the flow. The flow loop was housed in a self-designed incubator that maintained a 5% CO₂ and 37 °C environment.

EI cells were cultured on a fibronectin-coated polycarbonate filter (8 µm pore size) to form an “endothelial” monolayer. Prior to cell culture, the filter was first sterilized in 95% ethanol and exposed to ultraviolet light. The topside of the filter was coated with fibronectin (Collaborative Biomedical Products, MD) at a concentration 30 µg/ml in medium and incubated for 3 h. The sterilized filter was then transferred to a sterile Petri dish with the fibronectin-coated side facing up. A sterile glass block was placed on top of the filter and EI cells were seeded in the center section of the glass block, to ensure controlled cell growth on the topside of the filter [Fig. 3(C)]. Approximately 5 × 10⁵ cells were seeded per filter. A monolayer of EI cells was grown to 100% confluence on a filter 36–48 h after being placed onto the base plate of the chamber [Fig. 3(D)].

To stimulate cell migration, the center 12 wells (in the baseplate below the filter) were filled with soluble CIV as the chemoattractant, with a selected concentration (50–200 µg/ml) in RPMI 1640/0.1% BSA (Collaborative Biomedical Products, MD). All other wells were filled with non-stimulatory 0.1% BSA medium as a control [Fig. 3(B)].

Tumor cells were introduced to the flow channel, after which the chamber was assembled and connected to the flow loop. Initially a very slow flow (2 ml/min) was pumped through the system. The flow rate was then slowly increased to the desired rate (0–20 ml/min). Experiments were run at wall shear stresses in a range of 0–4 dyn/cm². After a selected time duration (1–4 h), the flow-migration chamber was removed from the flow loop and disassembled. The filter was gently removed and stained with Diff-Quik (Dade Behring Inc.). The C8161 cells on the bottom side of the filter were imaged using an inverted microscope and captured via NIH Image versus β4.0.2 with a PC. Three images of each migration filter were quantified and averaged for each data point. For each data point, a minimum of three filters was analyzed.

Statistical significance between cases tested in the flow-migration chamber were determined by the unpaired t test on Sigma Plot Version 4.01 (SPSS, Chicago, IL) with the corresponding P value reported for each case. All error bars on flow migration data histograms represent the SEM.

Directed Tumor Cell Migration

A series of experiments was performed with A2058 tumor cells to examine pseudopod protrusion and subsequent locomotion in response to soluble CIV stimulation under various concentration gradients. The initial chemotactic gradient was determined by C₀/δ (Fig. 2). There was no movement if the initial concentration of the chemoattractant was set zero or if the source pipette was filled with DMEM culture medium only. Measurements of pseudopod protrusion (L_p) and subsequent cell locomotion (L_c) for each cell were made as a function of time from the video replay. The time was displayed in “hours: minutes: seconds/tenths seconds,” with “00:00:00.0” being the time at which the cell was initially located in the pipette. The magnification of the system was calibrated before each experiment using a micrometer (50×2 µm) (Leica, Deerfield, IL).

In general, L_p has an initial lengthening phase over a short period of time (~3 min) followed by a fast shortening phase that seems to pull the whole cell forward by a distance L_c. These cyclic changes in L_p were found to be persistent, while L_c remained relatively idle during the lengthening phase of pseudopod protrusion in each cycle and increased only after every significant L_p lengthening peak. Pseudopod protrusion was, therefore, characterized by the protrusion frequency f_p as a function of time, defined by the total number of peaks in L_p divided by the time period. To obtain an average behavior of A2058 tumor cell locomotion, variables in dL_c/dt, L_p, and f_p were averaged (mean±SEM) over the total number of cells assayed at specified time points (Fig. 4). It is apparent that dL_c/dt, L_p, and f_p all increase with time during an early phase of cell crawling, followed by a relatively rapid decrease as a cell moves close to the chemotactic source.

Figure 5 represents time averages (mean±SEM) in $\begin{array}{cc}\overline{d{L}_c/dt},& \overline{L_p}\end{array}$, and $\overline{f_p}$ from the data shown in Fig. 4, respectively, in response to three different initial chemotactic concentration gradients. The $\overline{d{L}_c/dt}$ showed a greater magnitude under a higher initial concentration gradient, where the average cell locomotion velocities (µm/min) were 0.231±0.016 for G200:5 (C₀=200 µg/ml and δ=5 µm); 0.140±0.017 for G100:5 (C₀=100 µg/ml and δ=5 µm); and 0.133±0.016 for G200:10 (C₀=200 µg/ml and δ= 10 µm), respectively [Fig. 5(A)]. In comparison, $\overline{L_p}$ did not appear to correlate consistently with the changes in initial concentration gradient [Fig. 5(B)] where the average peak pseudopod protrusion lengths (µm) were 1.886±0.158 for G200:5; 1.337±0.127 for G100:5; and 1.710±0.132 for G200:10, respectively. However, $\overline{f_p}$ showed a strong correlation with the initial concentration gradient [Fig. 5(C)]. The mean peak pseudopod protrusion frequencies (1/min) were found to be 0.642±0.053 for G200:5; 0.533±0.025 for G100:5; and 0.517±0.020 for G200:10, respectively.

It should be noted that C₀/δ only represents the initial concentration gradient for a fixed diffusion domain (0<x<δ) (Fig. 2). Any subsequent cell locomotion generates a moving boundary at x=L(t)=δ−L_c(t), that will modify the attractant diffusion domain and concentration profile. Using Eqs. (1)–(5) with a set of model parameters, the spatial and temporal distributions in C(x,t) can be calculated. For example, if the attractant concentration at the cell front surface is denoted by C_d=C{L(t),t}, the actual chemotactic concentration gradient at a later time, t, will become {C₀−C_d}/L(t). Apparently, C_d increases while L(t) decreases. As a result of these temporal changes in L(t), the actual concentration C_d and gradient {C₀−C_d}/L(t) at the cell front surface could be solved [Figs. 6(A) and 6(B)], where T is the total locomotion time for a cell to crawl over a given distance δ. The model parameters were determined based on the best agreement between the experimental measurements on $\overline{d{L}_c/dt}$ and the calculated cell locomotion velocity dL/dt [Fig. 6(C)]. From Eqs. (1)–(5), two dimensionless parameters {k_r=δK_r/D and k_g=C₀K_g/D} were derived and found to be {3.42 and 1.18×10⁻¹⁰} for G200:5; {3.34 and 0.53×10⁻¹⁰} for G(100:5); and {2.89 and 0.45×10⁻¹⁰} for G200:10, respectively. Those kinetic parameters correlate with the concentration gradients of stimulus.

The dependence of CIV binding to A2058 cells on the local concentration of free CIV molecules was assayed by receptor staining and flow cytometry (Fig. 7). We found that CIV bound to cell surface receptors in a concentration-dependent manner [Fig. 7(A)]. However, the total amount of CIV bound to A2058 cells was apparently not influenced by time at least on the order of minutes [Fig. 7(B)]. We also performed CIV binding assay at a concentration up to 200 µg/ml [Fig. 7(A)] without seeing obvious saturation for concentration until about 100–200 µg/ml.

Dynamic Tumor Cell Extravasation

Experiments were performed to characterize in vivo tumor cell extravasation, a process of tumor cell dynamic adhesion to vascular endothelium under flow conditions and subsequent transendothelial migration in response to chemotactic stimulation from interstitial space (Fig. 1). To model this extravasation process, a novel in vitro flow-migration assay was developed (Fig. 3). C8161 cells were selected to characterize this new extravasation assay because they are invasive and metastatic melanoma tumor cells, which have been known to be highly motile.¹⁴

Tumor cell extravasation would normally occur in the microvasculature where fluid flows are steady and relatively slow. To model these in vivo conditions, flow-migration assays were run at a range of low shear stresses from 0 to 4 dyn/cm². Figure 8 shows that shear stresses can modulate the extent of C8161 cell extravasation. There was a significant initial drop (p=0.008) from the static migration case (0 dyn/cm²; 431.8±56.1 cells per mm²) to 1 dyn/cm² (234.83 ±25.1 cells per mm²), and a second decrease (p=0.0015) in migration from 1 to 2 dyn/cm² (122.2±12.8 cells per mm²). Subsequently, there was not a significant drop (p=0.86) in migration between 2 and 4 dyn/cm² (125.86±14.9 cells per mm²).

Standard static migration assays are commonly performed for the duration of 4 h.¹ It was necessary to determine if, under shear conditions, 4 h were sufficient for complete migration of tumor cells. To do this, a time curve of migration over time points from 1 to 4 h was assembled. It was found that, under 4 dyn/cm² shear stress, C8161 cell migration to CIV (100 µg/ml) increased as assay duration lengthened until about 3 h (p=0.006). Figure 9 demonstrates that migration between the time points of 3 and 4 h is not significantly different (p=0.99) and migration has reached a plateau by 4 h This guaranteed that migration assays were run to completion within 4 h and determined a critical parameter in developing the flow-migration protocol.

The concentration gradient of the chemoattractant is also an important parameter in characterizing cell migration. The concentration of CIV in the bottom wells (Fig. 3) was varied from C₀=50–200 µg/ml, and allowed to diffuse through the polycarbonate filter pores with an average filter thickness of δ~ 10 µm. Therefore, an initial chemotactic gradient was determined by C₀/δ, similar to that defined in the dual-pipette assay. A concentration of 100 µg/ml CIV was previously determined to be sufficient for maximum stimulation of A2058 tumor cell migration under static conditions.¹ In the flow-migration assay (under 4 dyn/cm² shear stress and 4 h migration time), a gradient G50:10 (C₀=50 µg/ml and δ=10 µm) was not sufficient for significant migration of the C8161 cells, as the migration was significantly below the case with a concentration gradient G100:10 (C₀=100 µg/ml and δ= 10 µm) (Fig. 10). Further increasing chemotactic gradient to G200:10 (C₀=200 µg/ml and δ= 10 µm) did not significantly increase migration above the case with G100:10.

The EI cell monolayer cultured on the flow-migration chamber substrate constitutively expresses ICAM-1 molecules.⁵ The level of ICAM-1 expression on EI cells (with an average mean fluorescence intensity 2.81±0.39) was comparable to that expressed on human umbilical vein endothelial cells (with an average fluorescence intensity 3.28±0.35) stimulated with cytokine interleukin IL-1β for 4 h.⁵ In order to extravasate, C8161 tumor cells need to adhere to this monolayer firmly enough to resist the shear stress imposed by the given fluid shear forces. A significant drop (p=0.002) in melanoma cell extravasation was found from a static case to the flow case under 4 dyn/cm² shear stress [Fig. 11(A)]. Flow cytometry analysis indicated that unstimulated-C8161 cells expressed relatively low level of Mac-1 and LFA-1, compared with its ICAM-1 expression [Fig. 11(B)]. In order to characterize the C8161 migration and elucidate which adhesion molecules might make contributions to tumor cell adhesion to the endothelium, monoclonal antibodies were used to selectively block Mac-1 and LFA-1 on C8161 cells, and ICAM-1 on EI cells, respectively. Antibody against Mac-1 or ICAM-1 apparently blocked C8161 extravasation in shear flow under 4 dyn/cm² shear stress, whereas blocking LFA-1 did not impact C8161 migration [Fig. 11(A)]. Figure 11 also illustrates that melanoma cells expressed much lower efficiency in extravasation under the flow conditions than the static conditions (p=0.002), which was correlated with a limited amount of functional Mac-1 (not much from LFA-1) on C8161 binding to ICAM-1 molecules on the substrate EI cell monolayer.

Figure 12 (A) illustrates that under the 4 dyn/cm² shear stress, C8161 extravasation was significantly enhanced (p<0.01) in the presence of PMA-activated Jurkat cells (1:1 tumor/Jurkat cell ratio), although the total migration of C8161 under flow conditions was still less than that under the static case. Adhesion of single C8161 cells or C8161/Jurkat cells emboli to ICAM-1-expressed EI cell monolayer was further examined over a range of increasing shear stresses [Fig. 12(B)]. By their ability to remain adhered at higher shear stresses, C8161/Jurkat pairs demonstrated stronger adhesion to ICAM-1 substrate compared to single C8161 cells. There was no evidence that Jurkat cells migrate toward CIV. This leads to an interesting model to study tumor-endothelium interactions in a shear flow in the presence of leukocytes.