Collective invasion of carcinoma cells

Abstract

Epithelial to mesenchymal transitions (EMT) have been suggested to be crucial during epithelial cancer cell invasion. However, in a three-dimensional “organotypic” invasion assay squamous cell carcinoma (SCC) cells that retain epithelial characteristics “hitch a ride” with carcinoma associated fibroblasts (CAFs) in order to collectively invade. Thus epithelial cancer cells can utilise the mesenchymal characteristics of CAFs without the need to undergo EMT themselves. This work provides new insight in cancer cell invasion and shows a new role for CAFs as a target for an anti-invasive therapy.

Cancer cell invasion and metastasis are the main causes of mortality in cancer patients. Understanding how cancer cells move and invade within the surrounding tissue is therefore a key issue. Stromal fibroblasts within a tumor play a crucial role in cancer cell proliferation, survival, angiogenesis as well as invasion (reviewed in ref. 1). In many cases stromal CAFs are able to produce a wide range of growth factors and cytokines that modulate tumor growth and invasion.^2,3 Their influence in cancer cell invasion and metastasis can also be mediated through the production of MMP's that promote extra-cellular matrix degradation.⁴

It has recently been shown that CAFs can play an unexpected role in SCC invasion.⁵ In a 3D ‘organotypic’ model of invasion that recreates the epidermal/dermal environment CAFs promote the collective invasion of SCC cells.⁶ 3D time-lapse confocal microscopy imaging showed that CAFs were always the leading cell of the invading cohort with the SCC cells following behind. These cohorts closely resembled invading clusters of SCC cells observed in human cancer samples.⁷ CAFs promoted SCC cells collective invasion by remodelling the matrix and making a path that SCC cells can use to invade. This process is clearly shown in Figure 1: a CAF (in red) leads the invasion of a collective chain of SCC cells (green) and makes a path in the surrounding matrix, visualized in grey using confocal reflectance microscopy. Two key experiments helped to understand the role of fibroblasts in this system. Firstly, the separation of the two cell populations by a thin layer of gel without fibroblasts completely abolished SCC invasion and so ruled out the possibility of long distance chemoattractant molecules inducing SCC invasion. Secondly, SCC cells were able to invade into a gel which had previously been remodelled by CAFs that had subsequently been removed. Together these experiments showed that tracks made by the fibroblasts are essential and sufficient to promote collective carcinoma cells invasion. Heterotypic cell contact between both populations was not required, as SCC cells can invade using tracks made by the CAFs even if the CAFs have been removed.

Figure 1.

Collective invasion of carcinoma cells led by fibroblast. Confocal time-lapse imaging of carcinoma associated fibroblast (red) leading the way of an invading chain of SCC cells (green) and making path into the surrounding matrix (grey). Panel is 80 x 80 mm and spans 300 minutes, scale are 20 um.

Interestingly, inhibition of Rho/ROCK signalling to the actomyosin cytoskeleton or MMPs using small molecule inhibitors blocked SCC invasion even when only CAFs where targeted. Blocking these pathways in carcinoma cells had little or no effect on their invasion. Moreover, inhibition of Rho function specifically in CAFs did not block their invasion into matrices but prevented SCC cells from following. These experiments showed the role of Rho/ROCK and MMPs molecular pathways in track generation by the CAFs and that targeting these pathways in CAFs, but not SCC cells, is critical for preventing cancer invasion. Strikingly, blockade of protease function after CAFs had remodelled the ECM had little effect on the ability of SCC cells to invade. This could explain the relative poor results obtained using MMPs inhibitors as anti-invasive therapies.⁸ Rho/ROCK function was dispensable in SCC cells; however, depletion of the small GTPase Cdc42 and its effector MRCK disrupted the acto-myosin cortex of carcinoma cells and blocked their capacity to invade in response to CAFs.

In order to invade and metastasise, carcinoma cells can switch from an epithelial state to a more mesenchymal phenotype.⁹ This process, called EMT, allows epithelial cancer cells to adapt their behaviour and confers the capacity to remodel the ECM on the cancer cells.¹⁰ However, in patient tissue samples, it has been observed that carcinoma cells can invade without undergoing an EMT, these cancer cells do not upregulate mesenchymal markers and retain cell to cell contact during their invasion.¹¹ This work explains how carcinoma cells that have not undergone EMT could invade a 3D matrix. These cells use the mesenchymal characteristics of the stromal fibroblasts to remodel the ECM and consequently follow behind invading fibroblasts. In tumours of mesenchymal origin CAFs are not required for invasion; work from Friedl and colleagues, clearly shows that HT1080 fibrosarcoma cells could lead collectively invading chains of cancer cells The authors showed how the leading cell of the collective chain remodels collagen fibres into tracks as it invades through the action of MT1-MMP (MMP14).¹²

In normal conditions, epithelial cells and dermal fibroblasts are in complete homeostasis and separated by a basement membrane (Fig. 2A). In addition, normal dermal fibroblasts are unable to promote SCC invasion. Understanding how CAFs are activated will be an important step forward. A desmoplastic response is observed in many tumours indicating a change in behaviour of fibroblasts.¹³ During wound healing or fibrosis, fibroblasts are in an active state that has been suggested to be similar to cancer activation.¹⁴ TGFβ has been shown to be a key player in fibroblasts activation and could support cancer progression.¹⁵ However, TGFβ was not responsible for SCC cells invasion since a TGFβ inhibitor had no effect in carcinoma cells collective invasion induced by the CAFs in the 3D invasion assay (Cedric Gaggioli and Steven Hooper, unpublished data). Interestingly, a probe that binds only to the active form of the small GTPase Rho showed that the activity of this protein was increased in CAFs compared to normal fibroblasts in tissue samples. Elevated expression of α5 integrin was also present in these cells and this has been implicated in Rho activation in a number of systems.^16–18 Consistent with this observation, depletion of integrin a5 in CAFs reduced their ability to promote the invasion of SCC cells. Alternatively, CAFs could also be derived from endothelial cells through a process called endothelial to mesenchymal transition¹⁹ (EndMT), or from cancer cells through EMT.²⁰ These processes could be responsible for CAFs generation in the tumor stroma resulting in matrix remodelling and tracks generation in order for the carcinoma cells to collectively invade the surrounding tissue and metastasize (Fig. 2B).

Figure 2.

Model of carcinoma cells collective invasion. (A) Schematic representation of a normal epithelium. Epithelial cells (light blue) and normal fibroblasts (pink) are separated by a basal membrane and are in a perfect homeostasis. Cross talk between both cell types occurs through adhesion and chemokine secretion. (B) Schematic representation of carcinoma cells collective invasion. CAFs (red) take the lead of a collective invading chain of SCC cells (brown). Invasion of CAFs is MMPs dependent but Rho/ROCK independent. However, track generation by CAFs is Rho/ROCK/MLC dependent. SCC cells require the small GTPase Cdc42 and its effector MRCK in order to collectively invade trough those tracks (black).

This study opens a new field of investigation for collective cancer cell invasion. This work highlights carcinoma associated fibroblasts as new therapeutic targets which will be a new direction in cancer cell invasion and metastasis therapy.

Abbreviations

CAF

carcinoma associated fibroblast

SCC

squamous cell carcinoma

ROCK

Rho kinase

MLC

myosin light chain

MMPs

matrix metalloproteases

MRCK

myotonic dystrophy kinase-related Cdc42-binding protein kinase

Addendum to: Gaggioli C, Hooper S, Hidalgo-Carcedo C, Grosse R, Marshall JF, Harrington K, Sahai E. Fibroblast led collective invasion of carcinoma cells with differing roles for RhoGTPases in leading and following cells. Nat Cell Biol. 2007;9:1392–1400.