Discovery Helps Show How Breast Cancer Spreads

Researchers at Washington University School of Medicine in St. Louis have discovered why breast cancer patients with dense breasts are more likely than others to develop aggressive tumors that spread. The finding opens the door to drug treatments that prevent metastasis.

Researchers at Washington University School of Medicine in St. Louis have discovered why breast cancer patients with dense breasts are more likely than others to develop aggressive tumors that spread. The finding opens the door to drug treatments that prevent metastasis. It has long been known that women with denser breasts are at higher risk for breast cancer. This greater density is caused by an excess of a structural protein called collagen. Collagen fiber alignment at the tumor boundary (dashed lines) is predictive of prognosis. Fibers that tend to be perpendicular to the tumor surface (top right, for example) encourage metastasis and indicate a poor prognosis. Fibers that run parallel to the tumor surface (bottom right) protect against cancer spreading. Tumors without DDR2 or SNAIL1 tend to show the protective parallel fiber alignment. (Credit: Nature Cell Biology)

It has long been known that women with denser breasts are at higher risk for breast cancer. This greater density is caused by an excess of a structural protein called collagen.

"We have shown how increased collagen in the breasts could increase the chances of breast tumors spreading and becoming more invasive," says Gregory D. Longmore, MD, professor of medicine. "It doesn't explain why women with dense breasts get cancer in the first place. But once they do, the pathway that we describe is relevant in causing their cancers to be more aggressive and more likely to spread."

The results appear online May 5 inNature Cell Biology.

Working in mouse models of breast cancer and breast tumor samples from patients, Longmore and his colleagues showed that a protein that sits on the surface of tumor cells, called DDR2, binds to collagen and activates a multistep pathway that encourages tumor cells to spread.

"We had no idea DDR2 would do this," says Longmore, also professor of cell biology and physiology. "The functions of DDR2 are not well understood, and it has not been implicated in cancer -- and certainly not in breast cancer -- until now.

At the opposite end of the chain of events initiated by DDR2 is a protein called SNAIL1, which has long been associated with breast cancer metastasis. Longmore and his colleagues found that DDR2 is one factor helping to maintain high levels of SNAIL1 inside a tumor cell's nucleus, a necessary state for a tumor cell to spread. Though they found it is not the only protein keeping SNAIL1 levels high, Longmore says DDR2 is perhaps the one with the most potential to be inhibited with drugs.

"It's expressed only at the edge of the tumor," says Longmore, a physician at Siteman Cancer Center at Washington University and Barnes-Jewish Hospital and co-director of the Section of Molecular Oncology. "And it's on the surface of the cells, which makes it very nice for developing drugs because it's so much easier to target the outside of cells."

Longmore emphasizes that DDR2 does not initiate the high levels of SNAIL1. But it is required to keep them elevated. This mechanism that keeps tumor cells in a state that encourages metastasis requires constant signaling -- meaning constant binding of DDR2 to collagen.

If that continuous signal is blocked, the cell remains cancerous, but it is no longer invasive. So a drug that blocks DDR2 from binding with collagen won't destroy the tumor, but it could inhibit the invasion of these tumors into surrounding tissue and reduce metastasis.

One possible way DDR2 may govern metastasis is its influence on the alignment of collagen fibers. If fibers are aligned parallel to the tumor's surface, the tumor is less likely to spread. While fibers aligned perpendicular to the surface of the tumor provide a path for the tumor cells to follow and encourage spreading. Tumors without DDR2 or SNAIL1 tend to show the parallel fiber alignment that is protective against spreading.

"This whole notion of fiber alignment and the tumor interface is a hot topic right now," Longmore says. "Our co-authors at the University of Wisconsin have developed a scoring method for collagen alignment that correlates with prognosis. And the bad prognosis disappears when you take away DDR2."

With the current emphasis on genetic mutations in cancer, Longmore is careful to point out that 70 percent of invasive ductal breast cancers show DDR2. But in 95 percent of these tumors the genes in this pathway -- from DDR2 to SNAIL1 -- are entirely normal, without mutations.

"If you did genomic sequencing, all of these particular genes would be normal," Longmore says. "You have to be careful not to just focus on mutations in cancer. This is an example of normal genes put together in an aberrant situation. The change in the environment -- the tumor and its surroundings -- causes the abnormal expression of these proteins. It is abnormal, but it's not caused by a gene mutation."

In early drug development efforts, Longmore and his colleagues have done some preliminary work looking for small molecules that may inhibit DDR2 binding to collagen.

"Currently there are no DDR2 specific inhibitors," Longmore says. "But there is great interest and work being done here and elsewhere to develop them."

Journal Reference:

1. Kun Zhang, Callie A. Corsa, Suzanne M. Ponik, Julie L. Prior, David Piwnica-Worms, Kevin W. Eliceiri, Patricia J. Keely, Gregory D. Longmore. The collagen receptor discoidin domain receptor 2 stabilizes SNAIL1 to facilitate breast cancer metastasis. Nature Cell Biology, 2013; DOI: 10.1038/ncb2743

Courtesy: ScienceDaily

Individual Brain Cells Track Where We Are and How We Move

Leaving the house in the morning may seem simple, but with every move we make, our brains are working feverishly to create maps of the outside world that allow us to navigate and to remember where we are.

Place cells in the real world (l) and in virtual reality. Place cells in the real world (left): The picture on the left shows the top-down view of a room in which a rat runs. There are four colorful walls and a check pattern on the floor. In the middle of the room, there is a narrow track (white line) on which the rat runs. Spikes fired by a neuron are shown by blue tick marks above and below the white track. Spikes fired when the rat runs from left-to-right are shown below the track and those spikes fired when the rat runs from right-to-left are shown above the track. These data were recorded in the real world. Notice that the spikes (blue tick marks) occur at the same absolute position hence these are called place cells. Place cells in virtual reality (right): These data were recorded in the virtual world. Note that the virtual world looks similar to the real world (other image). When the rat runs in this virtual world, the neurons fire at the same relative distance on the track in both movement directions, not at the same position as it does in the real world. Scientists hypothesize that the neurons are actually computing relative distances using cues on the walls and self-movement in the virtual world. In the real word, the presence of nonspecific sensory stimuli on the track, such as smells and textures, are hypothesized to make the neurons active at the same position. (Credit: Image courtesy of University of California - Los Angeles)

Take one step out the front door, and an individual brain cell fires. Pass by your rose bush on the way to the car, another specific neuron fires. And so it goes. Ultimately, the brain constructs its own pinpoint geographical chart that is far more precise than anything you'd find on Google Maps.

But just how neurons make these maps of space has fascinated scientists for decades. It is known that several types of stimuli influence the creation of neuronal maps, including visual cues in the physical environment -- that rose bush, for instance -- the body's innate knowledge of how fast it is moving, and other inputs, like smell. Yet the mechanisms by which groups of neurons combine these various stimuli to make precise maps are unknown.

To solve this puzzle, UCLA neurophysicists built a virtual-reality environment that allowed them to manipulate these cues while measuring the activity of map-making neurons in rats. Surprisingly, they found that when certain cues were removed, the neurons that typically fire each time a rat passes a fixed point or landmark in the real world instead began to compute the rat'srelative position, firing, for example, each time the rodent walked five paces forward, then five paces back, regardless of landmarks. And many other mapping cells shut down altogether, suggesting that different sensory cues strongly influence these neurons.

Finally, the researchers found that in this virtual world, the rhythmic firing of neurons that normally speeds up or slows down depending on the rate at which an animal moves, was profoundly altered. The rats' brains maintained a single, steady rhythmic pattern.

The findings, reported in the May 2 online edition of the journal Science,provide further clues to how the brain learns and makes memories.

The mystery of how cells determine place

"Place cells" are individual neurons located in the brain's hippocampus that create maps by registering specific places in the outside environment. These cells are crucial for learning and memory. They are also known to play a role in such conditions as post-traumatic stress disorder and Alzheimer's disease when damaged.

For some 40 years, the thinking had been that the maps made by place cells were based primarily on visual landmarks in the environment, known as distal cues -- a tall tree, a building -- as well on motion, or gait, cues. But, as UCLA neurophysicist and senior study author Mayank Mehta points out, other cues are present in the real world: the smell of the local pizzeria, the sound of a nearby subway tunnel, the tactile feel of one's feet on a surface. These other cues, which Mehta likes to refer to as "stuff," were believed to have only a small influence on place cells.

Could it be that these different sensory modalities led place cells to create individual maps, wondered Mehta, a professor with joint appointments in the departments of neurology, physics and astronomy. And if so, do these individual maps cooperate with each other, or do they compete? No one really knew for sure.

Virtual reality reveals new clues

To investigate, Mehta and his colleagues needed to separate the distal and gait cues from all the other "stuff." They did this by crafting a virtual-reality maze for rats in which odors, sounds and all stimuli, except distal and gait cues, were removed. As video of a physical environment was projected around them, the rats, held by a harness, were placed on a ball that rotated as they moved. When they ran, the video would move along with them, giving the animals the illusion that they were navigating their way through an actual physical environment.

As a comparison, the researchers had the rats -- six altogether -- run a real-world maze that was visually identical to the virtual-reality version but that included the additional "stuff" cues. Using micro-electrodes 10 times thinner than a human hair, the team measured the activity of some 3,000 space-mapping neurons in the rats' brains as they completed both mazes.

What they found intrigued them. The elimination of the "stuff" cues in the virtual-reality maze had a huge effect: Fully half of the neurons being recorded became inactive, despite the fact that the distal and gate cues were similar in the virtual and real worlds. The results, Mehta said, show that these other sensory cues, once thought to play only a minor role in activating the brain, actually have a major influence on place cells.

And while in the real world, place cells responded to fixed, absolute positions, spiking at those same positions each time rats passed them, regardless of the direction they were moving -- a finding consistent with previous experiments -- this was not the case in the virtual-reality maze.

"In the virtual world," Mehta said, "we found that the neurons almost never did that. Instead, the neurons spiked at the same relative distance in the two directions as the rat moved back and forth. In other words, going back to the front door-to-car analogy, in a virtual world, the cell that fires five steps away from the door when leaving your home would not fire five steps away from the door upon your return. Instead, it would fire five steps away from the car when leaving the car. Thus, these cells are keeping track of the relative distance traveled rather than absolute position. This gives us evidence for the individual place cell's ability to represent relative distances."

Mehta thinks this is because neuronal maps are generated by three different categories of stimuli -- distal cues, gait and "stuff" -- and that all are competing for control of neural activity. This competition is what ultimately generates the "full" map of space.

"All the external stuff is fixed at the same absolute position and hence generates a representation of absolute space," he said. "But when all the stuff is removed, the profound contribution of gait is revealed, which enables neurons to compute relative distances traveled."

The researchers also made a new discovery about the brain's theta rhythm. It is known that place cells use the rhythmic firing of neurons to keep track of "brain time," the brain's internal clock. Normally, Mehta said, the theta rhythm becomes faster as subjects run faster, and slower as running speed decreases. This speed-dependent change in brain rhythm was thought to be crucial for generating the 'brain time' for place cells. But the team found that in the virtual world, the theta rhythm was uninfluenced by running speed.

"That was a surprising and fascinating discovery, because the 'brain time' of place cells was as precise in the virtual world as in the real world, even though the speed-dependence of the theta rhythm was abolished," Mehta said. "This gives us a new insight about how the brain keeps track of space-time."

The researchers found that the firing of place cells was very precise, down to one-hundredth of a second, "so fast that we humans cannot perceive it but neurons can," Mehta said. "We have found that this very precise spiking of neurons with respect to 'brain-time' is crucial for learning and making new memories."

Mehta said the results, taken together, provide insight into how distinct sensory cues both cooperate and compete to influence the intricate network of neuronal activity. Understanding how these cells function is key to understanding how the brain makes and retains memories, which are vulnerable to such disorders as Alzheimer's and PTSD.

"Ultimately, understanding how these intricate neuronal networks function is a key to developing therapies to prevent such disorders," he said.

Other authors of the study included Pascal Ravassard, Ashley Kees and Bernard Willers, all lead authors, and David Ho, Daniel A. Aharoni, Jesse Cushman and Zahra M. Aghajan of UCLA. Funding was provided by the W.M. Keck foundation, a National Science Foundation career award grant and a National Institutes of Health grant (5R01MH092925-02).

Journal Reference:

1. P. Ravassard, A. Kees, B. Willers, D. Ho, D. A. Aharoni, J. Cushman, Z. M. Aghajan, M. R. Mehta. Multisensory Control of Hippocampal Spatiotemporal Selectivity.Science, 2013; DOI: 10.1126/science.1232655

Courtesy: ScienceDaily