Conceived and designed the experiments: DH NK TS MS KI. Performed the experiments: DH. Analyzed the data: DH KI. Contributed reagents/materials/analysis tools: DH VQ AW. Wrote the paper: DH KI.
The authors have declared that no competing interests exist.
MT1-MMP is a potent invasion-promoting membrane protease employed by aggressive cancer cells. MT1-MMP localizes preferentially at membrane protrusions called invadopodia where it plays a central role in degradation of the surrounding extracellular matrix (ECM). Previous reports suggested a role for a continuous supply of MT1-MMP in ECM degradation. However, the turnover rate of MT1-MMP and the extent to which the turnover contributes to the ECM degradation at invadopodia have not been clarified. To approach this problem, we first performed FRAP (Fluorescence Recovery after Photobleaching) experiments with fluorescence-tagged MT1-MMP focusing on a single invadopodium and found very rapid recovery in FRAP signals, approximated by double-exponential plots with time constants of 26 s and 259 s. The recovery depended primarily on vesicle transport, but negligibly on lateral diffusion. Next we constructed a computational model employing the observed kinetics of the FRAP experiments. The simulations successfully reproduced our FRAP experiments. Next we inhibited the vesicle transport both experimentally, and in simulation. Addition of drugs inhibiting vesicle transport blocked ECM degradation experimentally, and the simulation showed no appreciable ECM degradation under conditions inhibiting vesicle transport. In addition, the degree of the reduction in ECM degradation depended on the degree of the reduction in the MT1-MMP turnover. Thus, our experiments and simulations have established the role of the rapid turnover of MT1-MMP in ECM degradation at invadopodia. Furthermore, our simulations suggested synergetic contributions of proteolytic activity and the MT1-MMP turnover to ECM degradation because there was a nonlinear and marked reduction in ECM degradation if both factors were reduced simultaneously. Thus our computational model provides a new in silico tool to design and evaluate intervention strategies in cancer cell invasion.
Prevention of invasion is important in cancer therapy. MT1-MMP is a membrane protein involved in degradation of ECM (extracellular matrix) that is highly expressed at invadopodia, which are small protrusions of cancer cells. ECM degradation by MT1-MMP at invadopodia is hypothesized as the initial step of cancer cell invasion. However, MT1-MMP is inhibited by the endogenous inhibitor TIMP-2, so continuous turnover of MT1-MMP at the surface of invadopodia would be required. In agreement, it has been reported that the blockade of vesicle transport, which is one mechanism involved in the turnover, blocked the ECM degradation. However, the turnover rate of MT1-MMP at invadopodia and the extent to which the turnover is critical for the degradation of ECM have not been clarified. In this report we measured the turnover rate of MT1-MMP at a single invadopodium and found rapid turnover rates with time constants of 26 s and 259 s, which primarily depended on the vesicle transport. A computational model was constructed based on the observed kinetics. If we blocked the rapid turnover, the ECM degradation was blocked both experimentally and in simulations. These results established the role of the rapid turnover of MT1-MMP in the ECM degradation at invadopodia.
Some matrix metalloproteinases (MMPs) are proinvasive and employed by motile and invasive cells to degrade extracellular matrix (ECM). Among the 23 MMPs in mammals, integral membrane type MMPs, especially MT1-MMP, are believed to provide major contributions to cancer cell invasion. In fact, specific inhibition of MT1-MMP activity or knockdown of its expression suppresses not only cancer cell invasion in vitro but also tumor growth in mice
MT1-MMP was reported to be enriched in invadopodia
Several reports have been published that support the importance of a continuous supply of MT1-MMP. It is reported that MT1-MMP is delivered to the invasion front of cells by vesicle transport in a 3D collagen matrix
In addition to the supply of MT1-MMP to the surface, the internalization of MT1-MMP was also reported
All these previous reports strongly suggest a crucial role of a continuous supply and internalization of surface MT1-MMP (i.e. the turnover of MT1-MMP), particularly by vesicle-mediated mechanisms. However, the turnover rate of MT1-MMP and the extent to which the turnover contributes to ECM degradation at invadopodia have not been clarified. In the present study, we attempted to measure experimentally the turnover rate of MT1-MMP at a single invadopodium, and to elucidate the role of the turnover in ECM degradation by computer simulations.
First, we analyzed the turnover rate of MT1-MMP at a single invadopodium by FRAP (Fluorescence Recovery after Photobleaching) experiments using a chimeric MT1-MMP with a pH-sensitive GFP protein (pHLuorin). The recovery of fluorescence was plotted in a double-exponential plot, which showed time constants of 26.0 s and 259 s. Based on these experiments, we constructed a computational model to analyze the role of the rapid MT1-MMP turnover. Simulations showed no appreciable ECM degradation when the rapid turnover was inhibited in our simulation experiments. This was also true in the actual experiments. In addition, simulations have shown a decrease in the ECM degradation concomitantly with an increased reduction of the turnover rate of MT1-MMP. Thus, we have quantified the rate of turnover of MT1-MMP at single invadopodia and shown its critical role in ECM degradation both experimentally and by computer simulations. Furthermore, our simulations have shown a synergetic effect on the inhibition of ECM degradation by simultaneous reductions of the MT1-MMP turnover rate and the vesicular content of it. Our computational model provides a new in silico tool to design and evaluate intervention strategies in cancer cell invasion.
Head and neck squamous cell carcinoma SCC61 cells express endogenous MT1-MMP (
(A) Protein levels of MT1-MMP were assessed by Western blot analysis using anti-MT1-MMP antibody. Actin as a loading control. (B) Representative images show Dylight 633-labeled fibronectin (Green) and actin (Red). (C) Quantification of fibronectin degradation area in
To visualize live-dynamics of MT1-MMP on the plasma membrane, we employed phLuorin-tagged MT1-MMP (MT1-Luo). The phLuorin, a pH-sensitive modified GFP, enabled us to monitor MT1-MMP exposed on the plasma membrane separately from that in transport vesicles, based on the different pH values in the compartments
MT1-MMP is thought to be supplied to invadopodia continuously by exocytosis. In order to test various transport pathways to invadopodia, we used bafilomycin A1, which inhibits lysosome function by inhibiting vacuolar-type H(+)-ATPase (V-ATPase) and thereby inhibiting the lysosomal vesicle transport pathway; brefeldin A, which inhibits transport of proteins from ER to the Golgi system; and dynasore, which inhibits dynamin function and thereby prevents both endocytosis and trans-Golgi exit
(A) MT1-MMP-phLuorin-expressing SCC61 cells cultured on Dylight 633-labeled fibronectin were subjected to FRAP- continuous photobleaching experiments. One half of the invadopodia area, indicated by the open box area in region 1, was the FRAP experimental area. Region 2 was a continuous photobleaching area. (B) Quantification of fluorescence recovery in the
The above data indicate that continuous transport of MT1-MMP into the plasma membrane is critical for maintenance of the invadopodia's ECM-degrading activity. However, it is unclear to what extent lateral diffusion of MT1-MMP contributes to localization of the protease at invadopodia. To evaluate this possibility, we monitored fluorescence recovery after photo-bleaching (FRAP) of MT1-Luo at invadopodia. One-half of an invadopodium area, indicated by the open box area in region 1 in
Since lateral diffusion is almost negligible in invadopodia, we then used FRAP to analyze the kinetics of vesicle transport to invadopodia. Photo-bleaching of MT1-Luo at invadopodia was 60% at 60 sec after having decreased once to 30% of the control level, but the recovery did not reach 100% even after 330 sec; instead, it approached a plateau at around 65% (
To determine the time constant of fluorescence recovery at invadopodia, we redrew experimental data in a semi-logarithmic scale (Figure S2 in
From our experiments and analyses we found that the recovery of MT1-MMP at invadopodia is composed of two independent processes. Since the contribution of lateral diffusion of MT1-MMP within invadopodia was negligible (
Two pools of MT1-MMP, pool X (PX) and pool D (PD), were assumed. In pool X, insertion was dependent on the surface density of MT1-MMP, and internalization proceeds at a constant rate. While in pool D, insertion proceeds at a constant rate, and internalization was dependent on the surface density.
Our experiment showed that the invadopodial area occupied by MT1-MMP before bleaching was not refilled completely even at the saturated level of fluorescence recovery. On the contrary, the newly inserted MT1-MMP resided at a different area in addition to the area occupied before bleaching. In addition, the invadopodial area occupied by MT1-MMP did not cover the whole invadopodial area at any time during the recovery. Thus, there seemed to be a limited number of sites available for MT1-MMP insertion in invadopodia (Figure S3 in
We derived the following equation for the surface density of MT1-MMP at invadopodia:
The reconstructed time courses for fluorescence recovery after photobleaching in the absence and presence of bafilomycin are shown by continuous curves in
We next constructed a model for the activation of MMP-2 and the degradation of ECM by MT1-MMP and MMP-2. On the surface of the invadopodial membrane, MT1-MMP is dimerized, bound to TIMP-2, and forms a quadruple complex by binding proMMP-2 (MT1-MMP.MT1-MMP.TIMP-2.MMP-2) as shown in
(A) MT-MMPs in the membrane are dimerized, bound with TIMP-2, and activate proMMP-2. Any MT1-MMP, which is TIMP-2-free, was assumed to degrade ECM. MT1-MMP and all of complexes were assumed to be internalized. (B) Complete diagram of interaction between MT1-MMP, TIMP-2 and proMMP-2. Insertion is shown by a thick arrow for MT1-MMP, and internalization of MT1-MMP is shown in broken-lined arrows with internalized species at small squares. MT1-MMP, TIP-2 and proMMP-2 are designated as M14, T2 and M2 for simplicity.
The state transition diagram of MT1-MMP is shown in
In the absence of TIMP-2 the quadruple complex cannot be formed, and in excess TIMP-2, the concentration of TIMP-2-free MT1-MMP is reduced and proMMP-2 processing will be reduced significantly. In both cases ECM degradation by activated MMP-2 will be reduced. The concentration of the MMP-2-ECM complex, which is a measure of ECM degradation activity by MMP-2, was maximum at a TIMP-2 concentration of 180 nM, and at both higher and lower concentrations, it was steeply decreased as was expected (Figure S5A in
(A) Simulated ECM degradation with the initial TIMP-2 concentration at 180 nM is shown as a black line. Degradation rate, which is the rate of degradation in µM/s, is shown in red. (B) The time course of ECM degradation with several turnover intervals. If the turnover interval is increased by the factors shown by small numbers, a longer time is required for the same degree of degradation, reaching the state of no turnover (red curve).
If we changed the turnover rate of MT1-MMP by changing the insertion (
Let the time to reach 50% degradation of ECM (
Next we sought the reason for the reduced ECM-degradation that was associated with the reduced turnover rate of MT1-MMP. We found that the ECM-degradation rate at a time point corresponding to 50% ECM degradation was decreased by a value corresponding to the increase in the reduction factor (
Activated MMP-2 is diluted by diffusion after it is released from the MT1-MMP complex, and diffusing TIMP-2, which replenishes free TIMP-2, increases the inhibition of MT1-MMP at the site of its inhibition by TIMP-2. All these effects reduce ECM degradation in a three-dimensional (3D) space. To visualize these possibilities, we ran spatiotemporal simulations of ECM degradation (Figure S7 in
In the case with the MT1-MMP turnover observed in our experiments, the ECM was greatly degraded at 800 s (upper panel of
(A) Simulated degradation of ECM. With rapid turnover, MT1-MMP causes degradation of ECM (top panel), while in its absence, no appreciable ECM degradation is seen (bottom panel). (B) The time course of ECM degradation in the absence and presence of the turnover of MT1-MMP. The importance of the turnover for the effective degradation of ECM is clearly seen.
As shown in Figure S8 in
As a reduction in the concentration of MT1-MMP will also lead to a reduced ECM degradation, we asked what would be more effective for the reduction in ECM degradation, a reduced turnover rate or the concentration of MT1-MMP As shown in
The degradation efficacy is defined as
Here we have revealed a rapid turnover of the surface MT1-MMP with time constants of 26.0 and 259 s at invadopodia by experiments. In addition, we have demonstrated both by experiments and simulations that ECM degradation was blocked by blocking of the rapid turnover of MT1-MMP. Thus we have revealed a critical role of the rapid turnover for the effective ECM degradation at invadopodia (Figure S9 in
ECM degradation at invadopodia provides room for initial protrusions. Further degradation provides additional space for the elongation and enlargement of invadopodia. These processes are thought to finally lead to the invasion of cancer cells
Since rapid delivery of vesicles carrying MT1-MMP to invadopodia is important for the ECM degrading activity, it is of particular interest how this delivery process is regulated. Bafilomycin inhibits acidification of lysosomes. Therefore, MT1-MMP appears to be immediately delivered to invadopodia via an endosome-lysosome pathway rather than by the Golgi apparatus to the plasma membrane, which is preferentially inhibited by brefeldin. Indeed, Steffen et al. reported that VAMP7, which associates with late endosome-lysosome vesicles and plays a role in membrane fusion, is required for MT1-MMP delivery to invadopodia and regulates invadopodia formation by supporting MT1-MMP-mediated ECM degradation. Colocalization of MT1-MMP-mCherry with GFP-VAMP7 at invadopodia accompanying ECM degrading was demonstrated
The internalization of ErbB receptors (ErbBRs) has been extensively studied, and its rate constant was reported to be in the range of 0.04–0.2/min depending on the ligand-occupancy
Endocytosis of receptors has been recognized as a mechanism for their down-regulation. For example, ErbB receptors are known to be internalized upon ligand binding, and this process down-regulated receptor activity
In our model we assumed the existence of pools X and D with different insertion and internalization mechanisms for MT1 MMP. However, the observed fluorescence recovery with two time constants can be explained by two pools D or two pools X. Nevertheless, our bafilomycin experiments indicated the existence of different mechanism for the two time-constant processes. Therefore we assumed different insertion and internalization mechanisms for pools X and D.
Mathematical models showing cancer cell invasion of tissue have been reported previously
Here we have found a rapid turnover of surface MT1-MMP at invadopodia with a time constant of 26.0 s. The reduction in the turnover rate according to both fast and slow time constants reduced the rate of ECM degradation as shown in
The SCC61 cell lines have been described previously
The matrix degradation assay was done as described by Chen et al. Briefly, fibronectin (BD Biosciences) was labeled with Dylight 633 (Fisher) by dialysis in borate buffer [0.17 mol/L borate, 0.075 mol/L NaCl (pH 9.3)]. The buffer was changed to PBS and dialyzed extensively for 3 to 4 days. To coat MatTek dishes, 2.5% gelatin/2.5% sucrose in PBS added to the dish, followed by crosslinking with 0.5% glutaraldehyde in PBS. 50 ug/mL solution of fluorescence-labeled fibronectin was incubated with the cross-linked gelatin in MatTek dishes in the dark for 1 h. The dish was sterilized with 70% ethanol, washed with PBS, and equilibrated with invadopodia medium [DMEM supplemented with 15% FBS and 5% Nu-Serum (BD)] for 30 min before the addition of cells. For invadopodia assays, 7×104 cells were suspended in 2 mL of invadopodia medium containing 100 uM EGF and added to the plate for 18 h (parental cells) or 5 h (MT1-Luo cells).
A combination of both FRAP and continuous photobleaching techniques was developed on a confocal laser-scanning microscope Nikon A1 microscope using Nice elements Software (Nikon). MT1-Luo cells cultured on Dylight 633 labeled fibronectin for 5 hours were then bleached and scanned at two different regions of interest (Region 1 and 2) using bleaching laser excitation settings. During the bleaching phase, the region 1 (FRAP region) was excited with regular imaging settings, whereas continuous bleaching settings were used at the continuous photobleaching region.
Cells were imaged using an inverted confocal laser-scanning microscope (LSM 510, Carl Zeiss MicroImaging). The 488-nm line of an argon laser was used to excite phLuorin. For all samples, a Zeiss Plan-Neofluar 63X/1.3 oil immersion lens was used for imaging. FRAP experiments were performed at room temperature. For confocal FRAP measurements squares were chosen as ROIs. Within the squares were photobleached for either 100 or 200 scan iterations using 100% transmission of the 488-nm-wavelength laser.
The models were constructed using A-Cell
A reaction-diffusion simulation program in C language was automatically generated by A-Cell from the constructed model, and compiled using Intel C++ Studio XE 2011 for Linux. The differential equations were numerically integrated by the fourth-order Runge-Kutta method. Simulations were run on a Linux-based system with Intel Xeon X5680 3.33 GHz.
Supplement figures S1, S2, S3, S4, S5, S6, S7, S8, S9 with legends, and texts for “Mathematical reconstruction of FRAP signals”, and “Detail of the Model”.
(PDF)
We thank Drs. Murakami, Y. and Sakurai, M in the University of Tokyo and Saito, T. in Osaka University and Dr. Kenworthy, A.K. in Vanderbilt University for their helpful discussion. We also thank to Dr. Miesenbock, G. in University of Oxford for the generous gift of the phLuorin plasmid.