the past 30 years. In its strongest (simplest) form, the EMT hypothesis is that a simple transcriptional switch drives a dramatic change
in cellular morphology and behavior, from a stationary epithelium
into individual, motile, mesenchymal cells [1]. EMT transcription
factors include Twist and Snail (first identified and named based on
their Drosophila gastrulation phenotypes), and Zebs 1 and 2. The
wholesale EMT includes loss of apicobasal polarity in favor of
leading/lagging migratory polarity, complete loss of cell-cell junctions, expression of mesenchymal instead of epithelial intermediate
filament proteins and adhesion molecules, expression of matrix
metalloproteinases, and resistance to apoptosis.
Since cancer cells coopt normal developmental mechanisms, an
elegant explanation for the acquisition of migratory and invasive
characteristics by carcinoma cells (cancers of epithelial origin) is that
they reactivate the embryonic EMT program. This simple idea was
satisfying, and a great deal of evidence accumulated in support of it
[1–4]. Excitement grew that inhibition of EMT would prevent
metastasis and thereby render most cancers curable.
2 Unwelcome Complexity
Cracks began to appear in the foundation of this simple line of
thinking. For example, pathologists rarely if ever observe individual, mesenchymal cells in or around tumors. In principle this might
be because EMT is a transient state that stimulates motility, making
it hard to catch in fixed samples. But technical improvements in live
imaging of embryonic and organ development also revealed that
the majority of cell movements do not involve complete EMT.
Rather, many movements are collective, meaning that some degree
of cell-cell contact and apicobasal polarity are maintained
[5, 6]. Cells commonly migrate in clusters, strands, sheets, elongating tubes, or in fluid-like masses. Plenty of migratory cells move
without activating the EMT program, and in some cases “EMT
transcription factors” such as Twist promote collective types of
motility independently of EMT [7]. Tumor cells, like normal
cells, are found in all of kinds of arrangements too. Moreover,
some circulating tumor cell clusters are 60–100 times more effective than single cells at seeding metastases in animal models [8–
10]. Mother Nature seems determined to defy our zealous pursuit
of simplicity.
Out of necessity then, the EMT concept morphed into partial
EMT, transient EMT, and partial/transient EMT. The idea
emerged that a spectrum of metastable intermediate states exists
along a continuum between the extremes of epithelial and mesenchymal morphologies [4, 11]. The need to account for diverse cell
morphologies and behaviors led to a somewhat nebulous concept
of epithelial plasticity [12, 13].
14
Denise J. Montell
in cellular morphology and behavior, from a stationary epithelium
into individual, motile, mesenchymal cells [1]. EMT transcription
factors include Twist and Snail (first identified and named based on
their Drosophila gastrulation phenotypes), and Zebs 1 and 2. The
wholesale EMT includes loss of apicobasal polarity in favor of
leading/lagging migratory polarity, complete loss of cell-cell junctions, expression of mesenchymal instead of epithelial intermediate
filament proteins and adhesion molecules, expression of matrix
metalloproteinases, and resistance to apoptosis.
Since cancer cells coopt normal developmental mechanisms, an
elegant explanation for the acquisition of migratory and invasive
characteristics by carcinoma cells (cancers of epithelial origin) is that
they reactivate the embryonic EMT program. This simple idea was
satisfying, and a great deal of evidence accumulated in support of it
[1–4]. Excitement grew that inhibition of EMT would prevent
metastasis and thereby render most cancers curable.
2 Unwelcome Complexity
Cracks began to appear in the foundation of this simple line of
thinking. For example, pathologists rarely if ever observe individual, mesenchymal cells in or around tumors. In principle this might
be because EMT is a transient state that stimulates motility, making
it hard to catch in fixed samples. But technical improvements in live
imaging of embryonic and organ development also revealed that
the majority of cell movements do not involve complete EMT.
Rather, many movements are collective, meaning that some degree
of cell-cell contact and apicobasal polarity are maintained
[5, 6]. Cells commonly migrate in clusters, strands, sheets, elongating tubes, or in fluid-like masses. Plenty of migratory cells move
without activating the EMT program, and in some cases “EMT
transcription factors” such as Twist promote collective types of
motility independently of EMT [7]. Tumor cells, like normal
cells, are found in all of kinds of arrangements too. Moreover,
some circulating tumor cell clusters are 60–100 times more effective than single cells at seeding metastases in animal models [8–
10]. Mother Nature seems determined to defy our zealous pursuit
of simplicity.
Out of necessity then, the EMT concept morphed into partial
EMT, transient EMT, and partial/transient EMT. The idea
emerged that a spectrum of metastable intermediate states exists
along a continuum between the extremes of epithelial and mesenchymal morphologies [4, 11]. The need to account for diverse cell
morphologies and behaviors led to a somewhat nebulous concept
of epithelial plasticity [12, 13].
14
Denise J. Montell