Tumor cell invasion of collagen matrices requires coordinate lipid agonist-induced G-protein and membrane-type matrix metalloproteinase-1-dependent signaling

Abstract

Background: Lysophosphatidic acid (LPA) and sphingosine 1-phosphate (S1P) are bioactive lipid signaling molecules implicated in tumor dissemination. Membrane-type matrix metalloproteinase 1 (MT1-MMP) is a membrane-tethered collagenase thought to be involved in tumor invasion via extracellular matrix degradation. In this study, we investigated the molecular requirements for LPA- and S1P-regulated tumor cell migration in two dimensions (2D) and invasion of three-dimensional (3D) collagen matrices and, in particular, evaluated the role of MT1-MMP in this process.

Results: LPA stimulated while S1P inhibited migration of most tumor lines in Boyden chamber assays. Conversely, HT1080 fibrosarcoma cells migrated in response to both lipids. HT1080 cells also markedly invaded 3D collagen matrices (approximatly 700 microm over 48 hours) in response to either lipid. siRNA targeting of LPA1 and Rac1, or S1P1, Rac1, and Cdc42 specifically inhibited LPA- or S1P-induced HT1080 invasion, respectively. Analysis of LPA-induced HT1080 motility on 2D substrates vs. 3D matrices revealed that synthetic MMP inhibitors markedly reduced the distance (approximately 125 microm vs. approximately 45 microm) and velocity of invasion (approximately 0.09 microm/min vs. approximately 0.03 microm/min) only when cells navigated 3D matrices signifying a role for MMPs exclusively in invasion. Additionally, tissue inhibitors of metalloproteinases (TIMPs)-2, -3, and -4, but not TIMP-1, blocked lipid agonist-induced invasion indicating a role for membrane-type (MT)-MMPs. Furthermore, MT1-MMP expression in several tumor lines directly correlated with LPA-induced invasion. HEK293s, which neither express MT1-MMP nor invade in the presence of LPA, were transfected with MT1-MMP cDNA, and subsequently invaded in response to LPA. When HT1080 cells were seeded on top of or within collagen matrices, siRNA targeting of MT1-MMP, but not other MMPs, inhibited lipid agonist-induced invasion establishing a requisite role for MT1-MMP in this process.

Conclusion: LPA is a fundamental regulator of MT1-MMP-dependent tumor cell invasion of 3D collagen matrices. In contrast, S1P appears to act as an inhibitory stimulus in most cases, while stimulating only select tumor lines. MT1-MMP is required only when tumor cells navigate 3D barriers and not when cells migrate on 2D substrata. We demonstrate that tumor cells require coordinate regulation of LPA/S1P receptors and Rho GTPases to migrate, and additionally, require MT1-MMP in order to invade collagen matrices during neoplastic progression.

Figure 1

LPA stimulates and S1P inhibits migration of most tumor cell lines. Tumor cell migration analysis was performed on 8 μm polycarbonate gelatin-coated membranes and 20 μg/ml of fibronectin. LPA and/or S1P (1 μM) were added to the lower chambers and cells were allowed to migrate for 4 hours. Membranes were removed, stained, and migrating cells were quantitated using Scion^® software and Microsoft Excel^®. Data are expressed as mean number of migrating cells × 10³ (± S.D.) and represent the results of quadruplicate experiments.

Figure 2

LPA and S1P regulate tumor cell invasion of 3D collagen matrices. (A) HT1080, A2058, SKOV3 or HEK293 cells were induced to invade 3D collagen gels (3.75 mg/ml or 2.0 mg/ml) under serum-free conditions in the presence or absence of LPA alone (1 μM), S1P alone (1 μM), or both LPA and S1P (1 μM each). After 48 hours, cultures were fixed, stained, and images were acquired to demonstrate the effects of LPA and S1P on tumor cell invasion. Arrowheads indicate the position of the tumor cell monolayer at the initiation of invasion. Magnification = 20×. Scale bar = 50 μm. Inset (panel A) = higher power magnification of invading HT1080 cell. (B) Quantitation of tumor invasion from panel A. Data are expressed as mean numbers of invading cells per HPF (20×) (± S.D.) from a minimum of 20 fields.

Figure 3

Time-lapse analysis of HT1080 invasion: LPA and S1P induce ~700 μm of invasion of 3D collagen matrices. (A) HT1080 cells were seeded onto 3.75 mg/ml collagen gels in glass casings (see methods for details) and allowed to invade for 48 hours in the presence of DMEM, LPA, S1P, LPA and S1P (1 μM each) or LPA (1 μM) + GM6001 (5 μM). Digital images of representative fields were captured at ten minute intervals. Using Metamorph^®, invasion distance measurements at each time point from triplicate wells were obtained (n = 15). Magnification = 10×. Scale bars = 200 μm. (B) Data from panel A were plotted versus time and rates of invasion were calculated using linear regression (see Table 1). Data are expressed as mean invasion distance (n = 3 wells, ± S.D.) for each time point from triplicate experiments.

Figure 4

LPA₁ and S1P₁ regulate HT1080 invasion of 3D collagen matrices in response to LPA and S1P, respectively. (A) HT1080 cells were allowed to invade in the presence of either LPA or S1P (1 μM) in the presence (+) or absence (-) of 100 ng/μl of pertussis toxin (PTX). Cells were fixed, stained, and quantitated as described in Fig. 2. (B) HT1080 cells were transfected with siRNAs targeting the indicated genes and allowed to invade in the presence of either LPA or S1P (1 μM). Data are reported as mean numbers of invading cells per HPF (20×) (± S.D.) from a minimum of 20 fields. Statistical significance of means was determined using t-test where * = p < 0.01 relative to Luciferase (Luc) control for each lipid. (C) Real-time Quantitative PCR (RTQ-PCR) values for LPA₁ and S1P₁ mRNA normalized against 18S RNA were used to illustrate selective targeting of the designated siRNA.

Figure 5

Rac1 and Cdc42 are downstream effectors involved in LPA- and S1P-induced HT1080 invasion of 3D collagen matrices. (A) HT1080 cells were transfected with siRNAs targeting the genes indicated and allowed to invade in the presence either LPA or S1P (1 μM). Data are expressed as mean numbers of invading cells per HPF (20×) (± S.D.) from a minimum of 20 fields. Statistical significance of means was determined using t-test where * = p < 0.01 relative to Luciferase (Luc) control for each lipid. LMNA = Lamin A/C. (B) siRNA transfected HT1080 cell lysates were prepared for Western blot analysis. Lysates were probed for Lamin A/C, RhoA, Rac1, and Cdc42 to demonstrate siRNA specificity and probed for Actin as a loading control.

Figure 6

HT1080 cell motility using modified Boyden chambers does not require MMPs. HT1080 cells were seeded onto porous (8 μm) polycarbonate membranes coated with type I collagen (1 mg/ml) and allowed to migrate for four hours in the presence of 1 μM LPA or S1P in the presence or absence of synthetic MMP inhibitors (5 μM GM6001, 5 μM TAPI-0, or 5 μM TAPI-1). Membranes were removed, stained, and migrating cells were quantitated using Scion^® software and Microsoft Excel^®. Data are expressed as mean number of migrating cells × 10³ (± S.D.) and represent the results of quadruplicate experiments.

Figure 7

Time-lapse assessment of lipid agonist-induced HT1080 cell motility on 2D collagen substrates versus 3D collagen matrices. Nuc-GFP HT1080 cells were seeded on 2D plastic dishes coated with 50 μg/ml of collagen or embedded within a 3.75 mg/ml type I collagen matrix and their motility was analyzed using Metamorph^® software. Digital fluorescent images were acquired every 15 minutes and arranged in sequential order. A single cell is shown either moving on top of a dish (2D) or through 3.75 mg/ml collagen matrix (embedded in 3D) at 6 hour intervals for 24 hours. Tracings (generated by Metamorph^®) of the continuous movement of these cells over the 24 hour time period are shown by the blue lines. (Note: Metamorph^® creates a yellow line on a black background which is difficult to see. Therefore, the images were inverted to turn the yellow line blue and the black background white.) A blue (yellow) circle indicates the current position of the cell and a black (white) circle indicates the previous position. Subsequent tracings were converted to grayscale. Magnification = 20×. Scale bars = 50 μm.

Figure 8

Lipid agonist-induced tumor cell motility requires MMPs in 3D collagen matrices but not on 2D collagen substrates. Five individual cell tracings generated by Metamorph^® (grayscale) in the presence of the specified lipid with or without 5 μM GM6001 (see Table 2 for distance and velocity calculations). Magnification = 20×. Scale bars = 100 μm.

Figure 9

MT-MMPs are required for tumor cells to invade 3D collagen matrices. (A) HT1080 cells were induced to invade 3D collagen gels (3.75 mg/ml) as in Fig. 2 in the presence of 1 μM LPA and the synthetic MMP inhibitor GM6001 (5 μM), or TIMP-1, -2, -3, or -4 (5 μg/ml). Arrowheads indicate the position of the tumor cell monolayer at the initiation of invasion. Magnification = 20×. Scale bar = 50 μm. (B) Quantitation of HT1080 invasion from panel A. Data are expressed as mean numbers of invading cells per HPF (20×) (± S.D.) from a minimum of 20 fields.

Figure 10

HEK293 cells transfected with MT1-MMP cDNA invade 3D collagen gels in response to LPA. (A) Tumor cell lysates were prepared for Western blot analysis. Lysates were probed for MT1-MMP to assess protein expression in the four tumor cell lines. Lysates were probed for Actin as a loading control. (B) HEK293 cells were transfected with the pAdTrack-CMV plasmid as a control, or plasmids encoding MT1-MMP, MT2-MMP, or MT3-MMP cDNA 24 hours prior to placement in invasion assays. Cells were allowed to invade 2.0 mg/ml collagen gels in the presence or absence of 1 μM LPA. Data are expressed as mean numbers of invading cells per HPF (20×) (± S.D.) from a minimum of 20 fields. (C) Lysates from HEK293 cells transfected with cDNAs encoding the designated genes were prepared for Western blot analysis and probed for GFP, MT1-MMP, MT2-MMP, MT3-MMP, or Actin as a loading control. TRK = pAdTrack-CMV, MT1 = MT1-MMP, MT2 = MT2-MMP, MT3 = MT3-MMP.

Figure 11

siRNAs targeting MT1-MMP reduce protein levels and MMP-2 activity on gelatin zymography. (A) Invading HT1080 cells lysates were prepared for Western blot analysis and probed for MT1-MMP and Lamin A/C to demonstrate siRNA efficiency and specificity. Actin was used as a loading control. (B) Conditioned media from siRNA-treated HT1080 cells were collected after 48 hours of invasion and gelatin zymography was performed. Arrows denote pro-MMP-9, and pro- and active forms of MMP-2.

Figure 12

MT1-MMP is required for lipid-induced invasion of HT1080 cells in 3D collagen gels. (A) HT1080 cells were transfected with control siRNAs (Luciferase, Lamin A/C, or a scrambled MT1-MMP), siRNAs targeting soluble MMP-2 or MMP-9, or siRNAs targeting MT1-MMP (MT1-C, MT1-SP). Cells were allowed to invade 3D collagen gels in response to LPA or S1P for 48 hours prior to fixation and quantification of invasion as in Fig. 2. Data are expressed as mean numbers of invading cells per HPF (20×) (± S.D.) from a minimum of 20 fields per condition. Statistical significance of means was determined using t-test where * = p < 0.01 relative to Luciferase control. si = siRNA; Luc = Luciferase, LMNA = Lamin A/C, MT1-C = custom MT1-MMP, MT1-SP = SMARTpool^® MT1-MMP, MT1-sc = scrambled MT1-MMP. (B) siRNA transfected HT1080 cells from panel A were imaged as in Fig. 2 to demonstrate tumor cell invasion. Magnification = 20×. Scale bars = 50 μm. Arrowheads indicate the position of the monolayer at initiation of invasion.

Figure 13

The rate of HT1080 invasion is dependent on MT1-MMP. HT1080 cells transfected with siRNA targeting Luciferase (A), MT1-MMP (custom) (B), or scrambled MT1-MMP (C) were seeded onto 3.75 mg/ml collagen gels in glass casings (see methods for details) and allowed to invade for 48 hours in the presence 1 μM LPA. Data were collected as in Fig. 3 and were plotted versus time (D). Rates of invasion were calculated using linear regression (see Table 1). Data are expressed as mean rate of invasion (n = 3 wells, ± S.D.) for each time point from triplicate experiments. Magnification = 10×. Scale bars = 200 μm. Dashed line indicates front of tumor cell invasion. si = siRNA; Luc = Luciferase, MT1-C = custom MT1-MMP, MT1-sc = scrambled MT1-MMP.

Figure 14

MT1-MMP is required for tumor cell movement when cells are embedded within 3D collagen matrices. Representative individual 24 hour tracings (illustrated in Fig. 7) of Luciferase or custom MT1-MMP siRNA transfected Nuc-GFP HT1080 cells embedded in 3D collagen matrices are presented (grayscale). Images were acquired every 10 minutes. Magnification = 20×. Scale bars = 100 μm.

Figure 15

Schematic diagram summarizing the roles of LPA, S1P, RhoGTPases, and MT1-MMP in tumor migration and invasion of collagen matrices. LPA stimulates tumor cells migration while S1P inhibits migration of most tumor types on 2D matrix substrates. Lipid-induced invasion of 3D collagen matrices, but not migration, requires MT1-MMP. Only tumor cell invasion of 3D matrices is blocked by MMP inhibitors.