Meeting Report

Animal cells are asymmetrical. They have fronts and backs, centers and edges, or tops and bottoms, and they orient many of their structures according to this asymmetry. "Everybody who works in cell biology is interested in cell polarity for one reason or another, so whether you're interested in stem cells or in epithelial biology or neural biology or the immune system ... every cell is polarized," said Ian Macara of the University of Virginia School of Medicine, speaking at the June 29, 2009, meeting of the Academy's Cancer and Signaling Discussion Group.

As Macara and the meeting's other speakers explained, cell polarity isn't just involved in normal biology. Indeed, a change in cell polarity is one of the hallmarks of carcinogenesis, and understanding—and stopping—that transition is now one of the hottest topics in cancer research.

Not just a numbers game

Senthil Muthuswamy of Cold Spring Harbor Laboratory started the conference with a brief overview of the traditional view of tumor biology, in which cell proliferation is considered the primary problem. "It is thought in general [that] ... the loss of tissue architecture is a function of uncontrolled proliferation, so ... as long as you drive proliferation pathways really robustly in an uncontrolled manner, it'll eventually lead to a tissue architecture loss," said Muthuswamy.

Recently, though, some cracks have appeared in that theory. For example, some studies have shown that losing cell adhesion during carcinogenesis can disrupt tissue structure, even in the absence of major cell proliferation. Other work has revealed that a loss of cell polarity genes can cause excessive cell growth in the fly Drosophila melanogaster. Could disruptions in cell polarity alone really drive tumor formation?

Can cell polarity disruptions alone drive tumor formation?

To answer that question, Muthuswamy and his colleagues used a 3-dimensional cell culture system that mimics many aspects of mammary gland development. Culturing a mammary cell line in a gelatinous matrix, instead of a flat petri dish, allows the cells to form ball-like structures called acini. Cells in an acinus have apical and basal sides, and the structure itself looks like a duct from a human breast. "There is an uncanny resemblance in the way these cells organize themselves, giving us the hope that using this reductionist approach of culturing one cell type in a three-dimensional context may allow us to ask questions that could not be done by growing the same cells on a plastic dish," said Muthuswamy.

Activating the ErbB2 oncogene in the acinus system causes the growth of completely unorganized structures which proliferate faster than unmodified cells. Inactivating the Rb tumor suppressor, in contrast, causes rapid cell growth but allows the acinus to retain its duct-like structure. ErbB2, which is normally expressed at the apical-basal borders of the cells, seems to be serving some special purpose in the tissue's structure in addition to its role in cell proliferation.

Tracing the ErbB2 signaling pathway, the team soon discovered that the oncogene's two activities are separate: ErbB2 signals through one pathway to regulate tissue structure, and uses another pathway to regulate proliferation.

In subsequent experiments, the investigators found that disrupting the Scribble gene, which controls polarity but not proliferation, is sufficient to induce the dysplastic outgrowth of mammary cells in a mouse model. Disrupting both polarity control and proliferation control produces a synergistic signal, driving cells strongly toward tumor growth. "If you combine the two, what we found was there's a very strong synergy, and we got what looked like a neoplastic growth ... giving us the indication that maybe these two events actually cooperate in this context," said Muthuswamy.

The results support a two-component model of oncogenesis. Cells can experience either a hyperplastic event, which causes them to proliferate excessively, or a dysplastic event, which causes them to become disorganized. Neoplasia, or tumor formation, requires both events.

Skin deep

Enrique Rodriguez-Boulan of Weill Cornell Medical College works on epithelial cells, which are highly polarized. Each epithelial cell has clearly defined apical and basal sides, corresponding to the outer and inner sides of the epithelium. Epithelial tissue, which forms skin and the linings of guts and lungs, is the primary barrier between the body and the world. Partly because it encounters so many environmental insults, epithelial tissue is also the source of over 85% of human cancers.

After a brief review of the history of epithelial polarity research, Rodriguez-Boulan introduced his own work, which initially identified the trans-Golgi network and common recycling endosomes (CRE) as major components in the recycling and sorting of the proteins that distinguish the apical and basal sides of a cell. "These two systems were thought to operate separately for quite a few years, but in the last 10 years, it has become clear that they cooperate closely in the process of polarized sorting," said Rodriguez-Boulan.

How do the two systems coordinate their activities? To answer that, Rodriguez-Boulan and his colleagues performed a series of tagging and tracking experiments, producing time-lapse films of protein trafficking within epithelial cells.

The trans-Golgi network and recycling endosomes must cooperate.

The results suggest that the trans-Golgi network and CREs sort apically-targeted proteins using a complex set of tags, directing different proteins to cluster and move to the apical side of the cell. "Vesicles, all of them, move along linear paths, and those linear paths are really microtubules, so these vesicles ... have the ability to move along microtubules to the periphery of the cell," said Rodriguez-Boulan. Apical sorting signals can occur in any domain of a protein, and the apical sorting system may sometimes use small lipid rafts for targeting as well.

Basolateral sorting appears to be a simpler process; basolateral proteins carry signal sequences that are structurally similar to the signals that target cell-surface receptors to clathrin-coated pits for endocytosis. That implies that basolateral sorting could just be a modified version of endocytosis, but the evidence was indirect. "The suggestion was that clathrin might be involved in basolateral sorting, but even when we know that clathrin is in the Golgi, there was no evidence whatsoever that clathrin was involved in basolateral sorting," said Rodriguez-Boulan.

To test this hypothesis, the team knocked down clathrin expression in MDCK cells, a line of cultured, polarized cells. The knockdown selectively disrupts basolateral protein targeting, without affecting apical targeting, confirming that the basolateral pathway relies on clathrin-coated pits.

The genes that encode the protein sorting machinery, and those related to cell polarity in general, are often mutated in cancer cells, but their functions in carcinogenesis are still vague. "There's a clear correlation between cancer and disruption of the polarized machinery, which we don't understand yet very well," said Rodriguez-Boulan.

March of the macrophages

John Condeelis of Albert Einstein School of Medicine addressed one of the deadliest consequences of changes in cell polarity: metastasis. When a tumor metastasizes, cancerous cells migrate from the tumor to other parts of the body, often damaging other tissues and making the cancer much more difficult to treat. Indeed, in breast cancer, the most common malignancy in American women, metastasis is the major cause of death. "It turns out that metastasis is not about growth, it's about the spreading of the cells disseminating throughout the body, and it is one of the few pure cell motility diseases," said Condeelis.

Because motility is central to breast cancer pathogenesis, the best way to study the disease is light microscopy, which allows researchers to watch and measure single cell movement in real time. Unfortunately, traditional microscopes cannot see inside live animals, which is the best way to follow cell migration.

Condeelis and his colleagues have addressed that limitation with two-photon microscopy, which sends two long-wavelength photons into an organism's body on a collision course. The long wavelength is invisible, but it can penetrate deep tissue easily. When the photons collide, they produce a shorter, visible wavelength photon that the researchers can see, allowing them to watch the movements of individual cells anywhere inside a live animal's body.

Two-photon microscopy on whole animals shows metastasis in action.

Combining this powerful technique with fluorescently tagged proteins, the investigators were able to watch macrophages rapidly infiltrate tumors in a mouse model of breast cancer. The macrophages stimulate cells to move out of the tumor. "When these streams of cells get started, you'll have a tumor cell, macrophage, tumor cell, macrophage, tumor cell, and this line of cells will move out on collagen fibers and then into the blood vessels ... so it's a very efficient process for dissemination of these tumor cells all over the body," said Condeelis.

Why would macrophages, which are supposed to protect the body against disease, lead a parade of tumor cells to other tissues? Condeelis explained that the tumor is simply recapitulating a process that normally occurs during mammary development, when breast epithelium invades surrounding tissue to form mammary ducts: "This is an inherently invasive epithelium, and the invasion is triggered and led by the macrophages ... the breast tumor thinks it is a normal embryonic gland."

Next, the team constructed an artificial blood vessel by filling a permeable tube with growth factors. Inside a tumor, the tube acts like a lobster trap, luring metastatic cells to crawl into it. Gene expression profiles of the isolated cells revealed a set of 57 genes that correlate with metastasis, and clinical tests show that the same genes can predict metastasis in humans. "So this has very good predictive value ... The idea is to take the signature and run it through patient samples and determine if they have metastatic disease," said Condeelis.

Taking that result a step further, the researchers used the gene profile to identify the molecular pathways driving metastasis, and then developed a set of markers to predict which tumors were at the highest risk of migrating. In a retrospective study of breast cancer patients, this new test predicted distant metastasis of the tumor very reliably, which should make it useful in the clinic.

Speak, mammary

Ian Macara also works on breast cancer, a disease that is not only medically important but surprisingly convenient to study in the laboratory. Mammary development is mostly postnatal, the tissue is easily accessible, and a variety of cultured cell and transplant systems allow scientists to study it both in vitro and in vivo.

Macara's laboratory uses a mammary transplant model, in which they culture primary cells from a young mouse's mammary gland, transfect the cells with fluorescently labeled lentiviruses, then inject the transfected cells back into the fat pad of a host mouse, where they regenerate a glowing mammary gland. The procedure is much faster and more flexible than traditional transgenic and knock-out mouse techniques.

The mammary gland is a bilayered epithelial ductal system that consists of a luminal epithelial layer surrounded by a contractile myoepithelial layer. This organization is disrupted when Par3 is depleted. (Cytokeratin 14 (K14) staining shows the myoepithelial layer and E-cadherin staining shows the luminal layer.)

Using this system, the researchers found that knocking down expression of the Par3 polarity-determining gene derails mammary development: the epithelial layers are disorganized, and both proliferation and apoptosis, or programmed cell death, are unusually intense. "Instead of getting a nice end bud and mature duct, the entire mammary gland looks like one huge end bud," said Macara.

The results suggest that Par3 is necessary for determining cell fate, so that inhibiting Par3 prevents mammary progenitors from making the final step to become differentiated duct cells. Disrupting Par3 activity also causes the kinase aPKC to mislocalize, which may interfere with some kind of developmental switch, but the details are still obscure. "I think it's going to be very interesting to try and figure out this mechanism in detail—there's something very complicated going on," said Macara.

The Par3 gene helps determine mammary stem cell fate.

Macara and his colleagues are also looking at Par3's potential role in oncogenesis. So far, the lab has found that knocking down Par3 expression promotes metastasis in a mouse model of breast cancer, producing numerous metastatic tumors in the animals' lungs. The team is also employing a technique similar to the one Muthuswamy's group uses, culturing fragmented mammary glands in a three-dimensional matrix to observe the effects of Par3 and other genes more closely.

Taken together, the talks showed that researchers are finally getting the tools to connect all of the steps of the carcinogenic process, from observing molecular changes in cultured cells to predicting and treating metastatic cancer in patients. "I think this is going to be a very powerful way now that we can sort of go back and forth between in vivo and in vitro, and between mouse and human, and ask if we get the same phenotypes when we knock down polarity proteins or express oncogenes in human mammary glands in vitro versus mouse ones growing in vivo," said Macara.

Open Questions

What role is aPKC playing in determining mammary cell fate?

What are the primary signals macrophages use to draw tumor cells toward blood vessels?

Will drugs targeting newly identified metastasis factors forestall the process in human cancer patients?

Can misregulated apical and basolateral sorting serve as a marker or drug target for treating skin cancer?