A concrete solution

Cement materials, including cement paste, mortar, and concrete, are the most widely manufactured materials in the world. Their carbon footprint is similarly hefty: The processes involved in making cement contribute almost 6 percent of global carbon emissions.

The demand for these materials is unlikely to decline any time soon. In the United States, the majority of concrete bridges, buildings, and pavement-lined streets, erected in the 1960s and 1970s, were designed in an era with fewer environmental stresses to infrastructure and built to last 50 years at most.

Now, MIT researchers have discovered the beginnings of a new approach to producing concrete that is inspired by the hierarchical arrangements of simple building blocks in natural materials. The findings could lead to new ways to make concrete stronger and to use more sustainable, local materials as additives, to offset concrete’s greenhouse gas emissions.

In the new study, Oral Buyukozturk, a professor of civil and environmental engineering, and his colleagues analyzed a key property in concrete, at the level of individual atoms, that contributes to its overall strength and durability. The group developed a computer model to simulate the behavior of individual atoms which arrange to form molecular building blocks within a hardening material.

These simulations revealed that an interface within the molecular structure exhibited a “frictional” resistance under sliding deformation. The team then developed a cohesive-frictional force field, or model, that incorporates these atom-to-atom interactions within larger-scale particles, each containing thousands of atoms. The researchers say that accurately describing the forces within these assemblies is critical to understanding the way strength develops in concrete materials.

The team is now examining ways in which the cohesive and frictional forces of groups of atoms, or colloids in cement, are improved by mixing in certain additives such as volcanic ash, refinery slag, and other materials. The team’s computer model may help designers choose local additives based on the molecular interactions of the resulting mixtures. Through careful design at the microscopic level, he says, designers and engineers can ultimately build stronger, more environmentally sustainable structures.

“The conditions of the world are changing,” Buyukozturk says. “There are increased environmental demands, including from earthquakes and floods, and stresses on infrastructure. We need to come up with materials that are sustainable, with much longer design life and better durability. That is a big challenge.”

Buyukozturk and his colleagues, graduate student Steven Palkovic and Sidney Yip, professor emeritus in MIT’s Department of Nuclear Engineering, have published their results in the Journal of the Mechanics and Physics of Solids.

Strength from friction

Buyukozturk’s vision for revamped, locally sourced concrete is inspired, in part, by Roman construction. During the empire’s peak, the Romans erected temples, bath buildings, and amphiteaters in Pompeii, Ostia, and through Spain and the Middle East, including towns in Turkey, Libya, and Morocco. In each far-flung location, archaeologists have found that the Romans constructed their buildings from local materials — a technique that has helped preserve these structures for more than 2,000 years.

“They probably did this through intuition,” Buyukozturk says. “Ours is an effort to hopefully implement that kind of philosophy of using materials that are locally available, by understanding the underlying scientific principles within those materials.”

In their new paper, the scientists describe a computer model that is part of a computational framework that they have developed to analyze how the atomic structure of concrete affects engineering properties. These models simulate the sliding and movement of clusters of particles at molecular scales within concrete.

The researchers used their atomistic model to simulate mixtures containing Portland cement, the most common type of cement used in the world. Specifically, they simulated the mechanical response of the gel-like substance called calcium-silicate-hydrate (C-S-H), the main phase that forms when water reacts with Portland cement. The group modeled the movements of thousands of atoms in a C-S-H molecular building block, noting the influence of cohesive forces that cause particles to stick together, and the presence of a shear resistance as clusters of atoms slide past each other along a water-filled interface.

They then simulated how these molecular-scale properties control larger particles containing thousands of atoms, or colloids, at what they call the “mesoscale.” They discovered that the degree to which frictional properties resist the movement and separation of colloids at the mesoscale was the strongest factor in determining the strength of concrete at the centimeter scale.

Designers often use the properties of cement at the centimeter scale to predict the strength of a final, much larger-scale structure. The researchers thus implemented the results of their atoms-to-colloids simulations within computer models of the hardened microstructure, to allow for comparison with actual, centimeter-sized laboratory experiments. Buyukozturk found the team’s predictions matched with experimental outcomes better than predictions made with simulations that neglect frictional interactions.

“The material science of cement strength is still in its infancy regarding molecular-level descriptions and an ability to perform quantitative predictions,” Yip says.  “The issue of frictional force, addressed in our work, pertains to the mechanical behavior of cement that varies over time. This rate sensitivity is an aspect of the scientific challenges at the mesoscale, which is the research frontier where microscale concepts and models developed in several physical science disciplines are linked to macroscale properties for technological applications.” 

Buyukozturk adds, “We are confident that our new framework is opening a new era in concrete science.”

Additives in the mix

The group is now working on integrating various additives into their model, to investigate the effect of such materials on the atom-to-atom behavior of cement, and the resulting strength of the final, solidified concrete. From preliminary studies, they have observed that there is a chemical dependence of the friction value, or degree to which colloids resist sliding against each other. Future work will investigate how additives influence the chemical composition of these colloidal phases. This information could be used as part of a database to design and optimize new concrete materials with improved strength and deformation behavior.

“We know relatively little of what happens when additives are used in concrete,” Palkovic says. “We would not expect volcanic ash from Saudi Arabia to give the same performance as volcanic ash from Hawaii. So we need this greater understanding of the material, that starts at the atomistic scale and accounts for the chemistry of the material. That can give us greater control and understanding of how we can use additives to create a better material.”

This research was supported, in part, by the Kuwait Foundation for the Advancement of Sciences, as part of the MIT-Kuwait signature project on sustainability of Kuwait’s built environment.

de MIT News http://ift.tt/2fVHh7x

First open-access data from large collider confirm subatomic particle patterns

In November of 2014, in a first, unexpected move for the field of particle physics, the Compact Muon Solenoid (CMS) experiment — one of the main detectors in the world’s largest particle accelerator, the Large Hadron Collider — released to the public an immense amount of data, through a website called the CERN Open Data Portal.

The data, recorded and processed throughout the year 2010, amounted to about 29 terabytes of information, yielded from 300 million individual collisions of high-energy protons within the CMS detector. The sharing of these data marked the first time any major particle collider experiment had released such an information cache to the general public.

A new study by Jesse Thaler, an associate professor of physics at MIT and a long-time advocate for open access in particle physics, and his colleagues now demonstrates the scientific value of this move. In a paper published in Physical Review Letters, the researchers used the CMS data to reveal, for the first time, a universal feature within jets of subatomic particles, which are produced when high-energy protons collide. Their effort represents the first independent, published analysis of the CMS open data.

“In our field of particle physics, there isn’t the tradition of making data public,” says Thaler. “To actually get data publicly with no other restrictions — that’s unprecedented.”

Part of the reason groups at the Large Hadron Collider and other particle accelerators have kept proprietary hold over their data is the concern that such data could be misinterpreted by people who may not have a complete understanding of the physical detectors and how their various complex properties may influence the data produced.

“The worry was, if you made the data public, then you would have people claiming evidence for new physics when actually it was just a glitch in how the detector was operating,” Thaler says. “I think it was believed that no one could come from the outside and do those corrections properly, and that some rogue analyst could claim existence of something that wasn’t really there.”

“This is a resource that we now have, which is new in our field,” Thaler adds. “I think there was a reluctance to try to dig into it, because it was hard. But our work here shows that we can understand in general how to use this open data, that it has scientific value, and that this can be a stepping stone to future analysis of more exotic possibilities.”

Thaler’s co-authors are Andrew Larkoski of Reed College, Simone Marzani of the State University of New York at Buffalo, and Aashish Tripathee and Wei Xue of MIT’s Center for Theoretical Physics and Laboratory for Nuclear Science.

Seeing fractals in jets

When the CMS collaboration publicly released its data in 2014, Thaler sought to apply new theoretical ideas to analyze the information. His goal was to use novel methods to study jets produced from the high-energy collision of protons.

Protons are essentially accumulations of even smaller subatomic particles called quarks and gluons, which are bound together by interactions known in physics parlance as the strong force. One feature of the strong force that has been known to physicists since the 1970s describes the way in which quarks and gluons repeatedly split and divide in the aftermath of a high-energy collision.

This feature can be used to predict the energy imparted to each particle as it cleaves from a mother quark or gluon. In particular, physicists can use an equation, known as an evolution equation or splitting function, to predict the pattern of particles that spray out from an initial collision, and therefore the overall structure of the jet produced.

“It’s this fractal-like process that describes how jets are formed,” Thaler says. “But when you look at a jet in reality, it’s really messy. How do you go from this messy, chaotic jet you’re seeing to the fundamental governing rule or equation that generated that jet? It’s a universal feature, and yet it has never directly been seen in the jet that’s measured.”

Collider legacy

In 2014, the CMS released a preprocessed form of the detector’s 2010 raw data that contained an exhaustive listing of “particle flow candidates,” or the types of subatomic particles that are most likely to have been released, given the energies measured in the detector after a collision.

The following year, Thaler published a theoretical paper with Larkoski and Marzani, proposing a strategy to more fully understand a complicated jet in a way that revealed the fundamental evolution equation governing its structure.

“This idea had not existed before,” Thaler says. “That you could distill the messiness of the jet into a pattern, and that pattern would match beautifully onto that equation — this is what we found when we applied this method to the CMS data.”

To apply his theoretical idea, Thaler examined 750,000 individual jets that were produced from proton collisions within the CMS open data. He looked to see whether the pattern of particles in those jets matched with what the evolution equation predicted, given the energies released from their respective collisions. 

Taking each collision one by one, his team looked at the most prominent jet produced and used previously developed algorithms to trace back and disentangle the energies emitted as particles cleaved again and again. The primary analysis work was carried out by Tripathee, as part of his MIT bachelor's thesis, and by Xue.

“We wanted to see how this jet came from smaller pieces,” Thaler says. “The equation is telling you how energy is shared when things split, and we found when you look at a jet and measure how much energy is shared when they split, they’re the same thing.”

The team was able to reveal the splitting function, or evolution equation, by combining information from all 750,000 jets they studied, showing that the equation — a fundamental feature of the strong force — can indeed predict the overall structure of a jet and the energies of particles produced from the collision of two protons.

While this may not generally be a surprise to most physicists, the study represents the first time this equation has been seen so clearly in experimental data. 

“No one doubts this equation, but we were able to expose it in a new way,” Thaler says. “This is a clean verification that things behave the way you’d expect. And it gives us confidence that we can use this kind of open data for future analyses.”

Thaler hopes his and others’ analysis of the CMS open data will spur other large particle physics experiments to release similar information, in part to preserve their legacies.

“Colliders are big endeavors,” Thaler says. “These are unique datasets, and we need to make sure there’s a mechanism to archive that information in order to potentially make discoveries down the line using old data, because our theoretical understanding changes over time. Public access is a stepping stone to making sure this data is available for future use.”

This research was supported, in part, by the MIT Charles E. Reed Faculty Initiatives Fund, the MIT Undergraduate Research Opportunities Program, the U.S. Department of Energy, and the National Science Foundation.

de MIT News http://ift.tt/2fCcUpO

Developing sensors to defend aircraft against lasers

Laser strikes, the aiming of high-power laser pointers at aircraft, are a growing safety concern for pilots and aircraft passengers. They pose numerous dangers to pilots, including distraction during crucial moments in flight, temporary flash blindness, and in rare cases, permanent eye damage. Laser strikes have increased steadily in the last decade and can be criminally motivated, but they are more commonly pranks or unintentional incidents.

Although perpetrators of laser strikes can be punished under federal, state, and local laws in the United States, the lack of accurate and timely information for law-enforcement officials means less than 1 percent of them are ever caught. It is difficult for pilots to see where a laser beam is coming from, and even more difficult for police officers to pinpoint the perpetrator's location based on the pilot's report. Certain military aircraft are equipped with sensors that can estimate perpetrators' geographic location, or geolocation, but it is costly and unrealistic to have them installed on every airplane. Other existing defenses against laser strikes are merely passive devices, such as laser-blocking goggles or cockpit window films that can actually degrade the pilots' vision.

The only offensive measure of preventing laser strikes involves baiting perpetrators with police helicopters. In an area where laser strikes are frequent or anticipated, a police helicopter flies at a low altitude to deliberately attract laser strikes. When the helicopter is targeted, its pilots — who are equipped with night vision cameras — locate the perpetrators and alert ground law enforcement. However, this practice is not widely adopted, it requires significant time and effort, and the equipment involved can cost as much as $250,000.

"It's not ideal," says Richard Westhoff from the Laser Technology and Applications Group at the MIT Lincoln Laboratory. "You really have to put people in harm's way to do that."

To address the present lack of effective laser strike mitigation systems, the Laser Technology and Applications and Air Traffic Control Systems groups at Lincoln laboratory have teamed up to develop the Laser Aircraft Strike Suppression Optical System (LASSOS). LASSOS is a ground-based sensor system that can accurately identify the probable location of a perpetrator of a laser strike and immediately notify law enforcement.

"These sensors can provide persistent, automated protection for a high-risk volume of airspace, such as a final approach path, by quickly locating the origin of a laser strike and transmitting the coordinates to local law enforcement," says Tom Reynolds, a member of the development team and the associate leader of the Air Traffic Control Systems Group. "This technology will enable law enforcement to launch a rapid and targeted response to a laser strike event, greatly increasing their chance of apprehending and prosecuting perpetrators."

The system works by capturing side-scattered laser light and tracing it back to the perpetrator's location. When a laser is pointed into the sky, a small fraction of its light is scattered by air molecules and aerosols, forming a residual streak in the laser's path. Two or more high-sensitivity, low-noise, charge-coupled device cameras image the scattered light from different vantage points, providing the geometric diversity needed to digitally reconstruct the laser streak in three dimensions. The geographic coordinates of the laser's origin are calculated by tracing the laser streak down to a topographically accurate model of the Earth's surface.

A feature of LASSOS that makes it particularly effective is its integration with Google Earth. As soon as a laser is detected by the cameras, a digital reconstruction of the streak appears on a Google Earth map in real-time. This image summarizes the detection event, depicting the laser's point-of-origin and most probable path through the night sky. Within 30 seconds of the image being captured, LASSOS provides nearby members of law enforcement with the perpetrator's GPS coordinates, nearest address, and the time of the incident. This information allows officers to rapidly intervene.

LASSOS has the potential to diminish both immediate and future threats of laser strikes. Even if lasers strikes do not directly hit an aircraft's cockpit, police officers can use LASSOS to locate the perpetrators and detain them before they have the chance to cause serious harm. Furthermore, data gathered by LASSOS during an incident can be used as evidence in the prosecutions. The developers of LASSOS hope to increase air traffic safety by deterring future laser strikes.

"This technology will significantly increase laser strike origin detection and perpetrator apprehension," says Brian Saar, principal investigator in the laboratory's LASSOS team and an assistant group leader in the Laser Technology and Applications Group. "As culprits are readily apprehended and prosecuted, the appeal of laser strikes as a crime with low risk of detection will decline."

The system prototype has already demonstrated its speed and accuracy in several tests. For one test trial, LASSOS's geolocation ability was tested at a distance of nine nautical miles, to simulate the typical length of a final approach path when an aircraft is most vulnerable to lasing. One sensor was placed on top of the B Building at Lincoln Laboratory and another on the Flight Test Facility at Hanscom Air Force Base. Testers shone high-power laser beams of the type commonly used during lasing events from a baseball field nine nautical miles away in Tewksbury, Massachusetts, and were geolocated by LASSOS in less than 30 seconds. The system was so accurate that it could distinguish whether the laser beams came from first, second, or third base.

Considering LASSOS's promising performance and versatile capabilities, researchers believe that in the future the system could be used to protect other targets of laser strikes, including ships, automobiles, and even individual people.

de MIT News http://ift.tt/2yKeLOa

Bug-repair system learns from example

Anyone who’s downloaded an update to a computer program or phone app knows that most commercial software has bugs and security holes that require regular “patching.”

Often, those bugs are simple oversights. For example, the program tries to read data that have already been deleted. The patches, too, are often simple — such as a single line of code that verifies that a data object still exists.

That simplicity has encouraged computer scientists to explore the possibility of automatic patch generation. Several research groups, including that of Martin Rinard, an MIT professor of electrical engineering and computer science, have developed templates that indicate the general forms that patches tend to take. Algorithms can then use the templates to generate and evaluate a host of candidate patches.

Recently, at the Association for Computing Machinery’s Symposium on the Foundations of Software Engineering, Rinard, his student Fan Long, and Peter Amidon of the University of California at San Diego presented a new system that learns its own templates by analyzing successful patches to real software.

Where a hand-coded patch-generation system might feature five or 10 templates, the new system created 85, which makes it more diverse but also more precise. Its templates are more narrowly tailored to specific types of real-world patches, so it doesn’t generate as many useless candidates. In tests, the new system, dubbed Genesis, repaired nearly twice as many bugs as the best-performing hand-coded template system.

Thinning the herd

“You are navigating a tradeoff,” says Long, an MIT graduate student in electrical engineering and computer science and first author on the paper. “On one hand, you want to generate enough candidates that the set you’re looking through actually contains useful patches. On the other hand, you don’t want the set to include so many candidates that you can’t search through it.”

Every item in the data set on which Genesis was trained includes two blocks of code: the original, buggy code and the patch that repaired it. Genesis begins by constructing pairs of training examples, such that every item in the data set is paired off with every other item.

Genesis then analyzes each pair and creates a generic representation — a draft template — that will enable it to synthesize both patches from both originals. It may synthesize other, useless candidates, too. But the representation has to be general enough that among the candidates are the successful patches.

Next, Genesis tests each of its draft templates on all the examples in the training set. Each of the templates is based on only two examples, but it might work for several others. Each template is scored on two criteria: the number of errors that it can correct and the number of useless candidates it generates. For instance, a template that generates 10 candidates, four of which patch errors in the training data, might score higher than one that generates 1,000 candidates and five correct patches.

On the basis of those scores, Genesis selects the 500 most promising templates. For each of them, it augments the initial two-example training set with each of the other examples in turn, creating a huge set of three-example training sets. For each of those, it then varies the draft template, to produce a still more general template. Then it performs the same evaluation procedure, extracting the 500 most promising templates.

Covering the bases

After four rounds of this process, each of the 500 top-ranking templates has been trained on five examples. The final winnowing uses slightly different evaluation criteria, ensuring that every error in the training set that can be corrected will be. That is, there may be a template among the final 500 that patches only one bug, earning a comparatively low score in the preceding round of evaluation. But if it’s the only template that patches that bug, it will make the final cut.

In the researchers’ experiments, the final winnowing reduced the number of templates from 500 to 85. Genesis works with programs written in the Java programming language, and the MIT researchers compared its performance with that of the best-performing hand-coded Java patch generator. Genesis correctly patched defects in 21 of 49 test cases drawn from 41 open-source programming projects, while the previous system patched 11.

It’s possible that more training data and more computational power — to evaluate more candidate templates — could yield still better results. But a system that allows programmers to spend only half as much time trying to repair bugs in their code would be useful nonetheless.

de MIT News http://ift.tt/2fV5QkY

Researchers identify molecular motor that transforms chromosomes

A molecular “motor” that organizes the genome into distinct neighborhoods by forming loops of DNA has been characterized by researchers at MIT and the Pasteur Institute in France.

In a study published in 2016, a team led by Leonid Mirny, a professor of physics in MIT’s Institute for Medical Engineering and Sciences, proposed that molecular motors transform chromosomes from a loosely tangled state into a dynamic series of expanding loops.

The process, known as loop extrusion, is thought to bring regulatory elements together with the genes they control. The team also suggested that DNA is decorated with barriers — akin to stop signs — that limit the process of extrusion.

In this way, loop extrusion divides chromosomes into separate regulatory neighborhoods, known as topologically associating domains (TADs).

However, while the researchers suggested that a ring-like protein complex called cohesin was a likely candidate for these molecular motors, this had yet to be proven.

Now, in a paper published in the journal Nature, a team led by Mirny and Francois Spitz at the Pasteur Institute, have demonstrated that cohesin does indeed play the role of a motor in the loop extrusion process.

“Each of these machines lands on the DNA and starts extruding loops, but there are boundaries on DNA that these motors cannot get through,” Mirny says. “So as a result of this motor activity, the genome is organized into many dynamic loops that do not cross the boundaries, so the genome becomes divided into a series of neighborhoods.”

The researchers also discovered that a different mechanism, that does not use cohesin, is at work organizing active and inactive regions of DNA into separate compartments in the cell’s nucleus.

To determine the role cohesin plays in genome formation, the team first deleted a molecule known as Nipbl, which is responsible for loading cohesin onto DNA.

They then used an experimental technique known as Hi-C, in which parts of DNA that are close to one another in 3-D space are captured and sequenced, in a bid to measure the frequency of physical interactions between different spots along chromosomes.

This technique, which was pioneered by Job Dekker, a professor of biochemistry and molecular pharmacology at the University of Massachusetts Medical Center in Worcester, has previously been used to demonstrate the existence of TADs.

The team first used the Hi-C technique to assess the organization of chromosomes before removing the Nipbl molecule from mice. They then removed the molecule and performed the same measurement again.

They found that the neighborhoods had virtually disappeared.

However, the compartmentalization between active and inactive regions of the genome had become even more marked.

The team believes the cohesin motors allow each gene to reach out to its regulatory elements, which control whether genes should be switched on or off.

What’s more, it appears that the cohesin motors are stopped by another protein, CTCF, which demarcates the boundaries of each neighborhood. In a recent study in the journal Cell, the Mirny lab, in collaboration with researchers at the University of California at San Francisco and the University of Massachusetts Medical School has demonstrated that if this demarcating protein is removed, the borders between neighborhoods disappear, allowing genes in one neighborhood to talk to regulatory elements they should not be talking to in another neighborhood, and leading to misregulation of genes in the cell.

“Cohesin is central for gene regulation, and we emphasize that this is a motor function, so it is not just that they (genes and their regulatory elements) find each other somewhere randomly in space, but they were brought together by this motor activity,” Mirny says.

This paper provides important new molecular insights into the mechanisms by which cells fold their chromosomes, according to Dekker, who was not involved in the current study.

“In this work the Mirny and Spitz labs combine mouse models with genomic approaches to study chromosome folding to reveal that the machine that loads the cohesin complex is critical for TAD formation,” Dekker says. “From this and another previous study, a molecular mechanism is coming into view where TADs form by cohesin and Nipbl-dependent chromatin loop extrusion, which is blocked by sites bound by CTCF.”

The researchers are now attempting to characterize how the absence of the molecular motor would affect gene regulation. They are also carrying out computer simulations in a bid to determine how the cohesin-based loop extrusion takes place at the same time as the genome is undergoing the independent process of segregating into active and inactive compartments.

“It’s like two pianists playing on the same piano,” says Nezar Abdennur, a PhD student in the Mirny lab, who took part in the study alongside fellow PhD student Anton Goloborodko. “They interfere and put constraints on each other, but together they can produce a beautiful piece of music.”

Abdennur and Goloborodko are co-first authors of the paper, along with Wibke Schwarzer of EMBL. The research was supported, in part, by the National Institutes of Health, the National Science Foundation, and the MIT-France MISTI Fund.

de MIT News http://ift.tt/2xG03r4

Letter regarding Campus Sustainability Task Force report

The following email was sent today to the MIT community by Provost Martin Schmidt and Executive Vice President and Treasurer Israel Ruiz.

To the members of the MIT community,

Two years ago we convened the Campus Sustainability Task Force (CSTF), charged to shape a vision and plan of action for campus sustainability at MIT. The CSTF has now drafted its report, Pathway to Sustainability Leadership by MIT, which reflects input from students, faculty, staff, and alumni since the CSTF launch in 2015. In releasing the preliminary report, we are opening a comment period through November, during which we are actively seeking feedback from across the MIT community.

We invite you to attend a campus-wide forum to discuss the report on Tuesday, October 17, 12:00 pm–1:30 pm in the Millikan Room (E53-482). Lunch will be provided. Please RSVP if you would like to attend.

We encourage you to read the report, which lays out the five key elements of the pathway to sustainability leadership. It is important for all voices to be heard as Institute leadership considers the task force’s recommendations. We and task force co-chairs Andrea Campbell and Julie Newman are eager to hear your thoughts, and hope you will attend the open forum. You may also send comments to sustainablemit@mit.edu.

Sincerely,

Marty Schmidt
Provost

Israel Ruiz
Executive Vice President and Treasurer

de MIT News http://ift.tt/2wnfK5i

Biologists identify possible new strategy for halting brain tumors

MIT biologists have discovered a fundamental mechanism that helps brain tumors called glioblastomas grow aggressively. After blocking this mechanism in mice, the researchers were able to halt tumor growth.

The researchers also identified a genetic marker that could be used to predict which patients would most likely benefit from this type of treatment. Glioblastoma is usually treated with radiation and the chemotherapy drug temozolamide, which may extend patients’ lifespans but in most cases do not offer a cure.

“There are very few specific or targeted inhibitors that are used in the treatment of brain cancer. There’s really a dire need for new therapies and new ideas,” says Michael Hemann, an associate professor of biology at MIT, a member of MIT’s Koch Institute for Integrative Cancer Research, and a senior author of the study.

Drugs that block a key protein involved in the newly discovered process already exist, and at least one is in clinical trials to treat cancer. However, most of these inhibitors do not cross the blood-brain barrier, which separates the brain from circulating blood and prevents large molecules from entering the brain. The MIT team hopes to develop drugs that can cross this barrier, possibly by packaging them into nanoparticles.

The study, which appears in Cancer Cell on Sept. 28, is a collaboration between the labs of Hemann; Jacqueline Lees, associate director of the Koch Institute and the Virginia and D.K. Ludwig Professor for Cancer Research; and Phillip Sharp, an MIT Institute Professor and member of the Koch Institute. The paper’s lead authors are former MIT postdoc Christian Braun, recent PhD recipient Monica Stanciu, and research scientist Paul Boutz.

Too much splicing

Several years ago, Stanciu and Braun came up with the idea to use a type of screen known as shRNA to seek genes involved in glioblastoma. This test involves using short strands of RNA to block the expression of specific genes. Using this approach, researchers can turn off thousands of different genes, one per tumor cell, and then measure the effects on cell survival.

One of the top hits from this screen was the gene for a protein called PRMT5. When this gene was turned off, tumor cells stopped growing. Previous studies had linked high levels of PRMT5 to cancer, but the protein is an enzyme that can act on hundreds of other proteins, so scientists weren’t sure exactly how it was stimulating cancer cell growth.

Further experiments in which the researchers analyzed other genes affected when PRMT5 was inhibited led them to hypothesize that PRMT5 was using a special type of gene splicing to stimulate tumor growth. Gene splicing is required to snip out portions of messenger RNA known as introns, that are not needed after the gene is copied into mRNA.

In 2015, Boutz and others in Sharp’s lab discovered that about 10 to 15 percent of human mRNA strands still have one to three “detained introns,” even though they are otherwise mature. Because of those introns, these mRNA molecules can’t leave the nucleus.

“What we think is that these strands are basically an mRNA reservoir. You have these unproductive isoforms sitting in the nucleus, and the only thing that keeps them from being translated is that one intron,” says Braun, who is now a physician-scientist at Ludwig Maximilian University of Munich.

In the new study, the researchers discovered that PRMT5 plays a key role in regulating this type of splicing. They speculate that neural stem cells utilize high levels of PRMT5 to guarantee efficient splicing and therefore expression of proliferation genes. “As the cells move toward their mature state, PRMT5 levels drop, detained intron levels rise, and those messenger RNAs associated with proliferation get stuck in the nucleus,” Lees says.

When brain cells become cancerous, PRMT5 levels are typically boosted and the splicing of proliferation-associated mRNA is improved, ultimately helping the cells to grow uncontrollably.

Predicting success

When the researchers blocked PRMT5 in tumor cells, they found that the cells stopped dividing and entered a dormant, nondividing state. PRMT5 inhibitors also halted growth of glioblastoma tumors implanted under the skin of mice, but they did not work as well in tumors located in the brain, because of the difficulties in crossing the blood-brain barrier.

Unlike many existing cancer treatments, the PRMT5 inhibitors did not appear to cause major side effects. The researchers believe this may be because mature cells are not as dependent as cancer cells on PRMT5 function.

The findings shed light on why researchers have previously found PRMT5 to be a promising potential target for cancer treatment, says Omar Abdel-Wahab, an assistant member in the Human Oncology and Pathogenesis Program at Memorial Sloan Kettering Cancer Center, who was not involved in the study.

“PRMT5 has a lot of roles, and until now, it has not been clear what is the pathway that is really important for its contributions to cancer,” says Abdel-Wahab. “What they have found is that one of the key contributions is in this RNA splicing mechanism, and furthermore, when RNA splicing is disrupted, that key pathway is disabled.”

The researchers also discovered a biomarker that could help identify patients who would be most likely to benefit from a PRMT5 inhibitor. This marker is a ratio of two proteins that act as co-factors for PRMT5’s splicing activity, and reveals whether PRMT5 in those tumor cells is involved in splicing or some other cell function.

“This becomes really important when you think about clinical trials, because if 50 percent or 25 percent of tumors are going to have some response and the others are not, you may not have a way to target it toward those patients that may have a particular benefit. The overall success of the trial may be damaged by lack of understanding of who’s going to respond,” Hemann says.

The MIT team is now looking into the potential role of PRMT5 in other types of cancer, including lung tumors. They also hope to identify other genes and proteins involved in the splicing process they discovered, which could also make good drug targets.

Spearheaded by students and postdocs from several different labs, this project offers a prime example of the spirit of collaboration and “scientific entrepreneurship” found at MIT and the Koch Institute, the researchers say.

“I think it really is a classic example of how MIT is a sort of bottom-up place,” Lees says. “Students and postdocs get excited about different ideas, and they sit in on each other’s seminars and hear interesting things and pull them together. It really is an amazing example of the creativity that young people at MIT have. They’re fearless.”

The research was funded by the Ludwig Center for Molecular Oncology at MIT, the Koch Institute Frontier Research Program through the Kathy and Curt Marble Cancer Research Fund, the National Institutes of Health, and the Koch Institute Support (core) Grant from the National Cancer Institute.

de MIT News http://ift.tt/2yKoNyF