Thursday, March 15, 2018

Biologists unravel another mystery of what makes DNA go 'loopy'

Interior of a cell showing the nucleus with the chromatin fiber (yellow) arranged in the three-dimensional space by loops formed by the CTCF protein (shown in pink). DNA is represented by thin blue lines on the chromatin. Graphic by Victor Corces. 

By Carol Clark

Scientists discovered another key to how DNA forms loops and wraps inside the cell nucleus — a precise method of “packing” that may affect gene expression.

The journal Science published the research by biologists at Emory University, showing that a process known as hemimethylation plays a role in looping DNA in a specific way. The researchers also demonstrated that hemimethylation is maintained deliberately — not through random mistakes as previously thought — and is passed down through human cell generations.

“In order for a protein called CTCF to make loops in the DNA, we discovered that it needs to have hemimethylated DNA close by,” says Emory biologist Victor Corces, whose lab conducted the research. “Nobody had previously seen that hemimethylated DNA has a function.”

Chromatin is made up of CTCF and other proteins, along with DNA and RNA. One role of chromatin is to fold and package DNA into more compact shapes. Growing evidence suggests that this folding process is not just important to fit DNA into a cell nucleus — it also plays a role in whether genes are expressed normally or malfunction.

The Corces lab specializes in epigenetics: The study of heritable changes in gene function — including chromatin folding — that do not involve changes in the DNA sequence.

DNA methylation, for example, can modify the activity of DNA by adding methyl groups to both strands of the double helix at the site of particular base pairs. The process can be reversed through demethylation.

As cells divide they make a copy of their DNA. In order to do so, they have to untangle the two strands of DNA and split them apart. Each parental strand then replicates a daughter strand.

“When cells divide, it’s important that they keep the methylation the same for both strands,” Corces says, noting that altered patterns of methylation are associated with cancer and other diseases.

Hemimethylation involves the addition of a methyl group to one strand of the DNA helix but not the other. Some researchers observing hemimethylation have hypothesized that they were catching it right after cell division, before the cell had time to fully replicate to form a daughter strand. Another theory was that hemimethylation was the result of random mistakes in the methylation process.

Chenhuan Xu, a post-doctoral fellow in the Corces lab, developed new experimental methods for DNA methylome mapping to conduct the research for the Science paper. These methods allowed the researchers to observe hemimethylation on DNA in human cells in real-time before, during and after cell division. They also mapped it as the cells continued to replicate.

“If the parental DNA was hemimethylated, the daughter DNA was also hemimethylated at the same place in the genome,” Corces says. “The process is not random and it’s maintained from one cell generation to the next over weeks.”

The researchers found that hemimethlyation only occurs near the binding sites of CTCF — the main protein involved in organizing DNA into loops.

“If we got rid of the hemimethlyation, CTCF did not make loops,” Corces says. “Somehow, hemimethylation is allowing CTCF to make loops.”

And when CTCF makes a loop, it does so by binding ahead, going forward in the DNA sequence, they observed.

“Research suggests that some disorders are associated with CTCF binding — either mutations in the protein itself or with the DNA sequence where the protein binds,” Corces says. “It comes back to the story of how important these loops are to the three-dimensional organization of chromatin, and how that organization affects the gene expression.”

Small steps lead to big career
Teen scientists bloom in lab
Epigenetics zeroes in on nature vs. nurture

Monday, March 12, 2018

Biophysicists discover how small populations of bacteria survive treatment

"We showed that by tuning the growth and death rate of bacterial cells, you can clear small populations of even antibiotic-resistant bacteria using low antibiotic concentrations," says biophysicist Minsu Kim. His lab conducted experiments with E. coli bacteria (above).

By Carol Clark

Small populations of pathogenic bacteria may be harder to kill off than larger populations because they respond differently to antibiotics, a new study by Emory University finds.

The journal eLife published the research, showing that a population of bacteria containing 100 cells or less responds to antibiotics randomly — not homogeneously like a larger population.

“We’ve shown that there may be nothing special about bacterial cells that aren’t killed by drug therapy — they survive by random chance,” says senior author Minsu Kim, an assistant professor in the Department of Physics and a member of Emory’s Antibiotic Resistance Center.

“This randomness is a double-edged sword,” Kim adds. “On the surface, it makes it more difficult to predict a treatment outcome. But we found a way to manipulate this inherent randomness in a way that clears a small population of bacteria with 100 percent probability. By tuning the growth and death rate of bacterial cells, you can clear small populations of even antibiotic-resistant bacteria using low antibiotic concentrations.”

Jessica Coates, as a graduate student at Emory, and Bo Ryoung Park, a research associate in the Kim lab, are co-first authors of the paper. Additional authors are graduate student Emreh Simsek and post-doctoral fellows Dai Le and Waqas Chaudry.

The researchers developed a treatment model using a cocktail of two different classes of antibiotic drugs. They first demonstrated the effectiveness of the model in laboratory experiments on a small population of E. coli bacteria without antibiotic-drug resistance. In later experiments, they found that the model also worked on a small population of clinically-isolated antibiotic-resistant E. coli.

“We hope that our model can help in the development of more sophisticated antibiotic drug protocols — making them more effective at lower doses for some infections,” Kim says. “It’s important because if you treat a bacterial infection and fail to kill it entirely, that can contribute to antibiotic resistance.”

Antibiotic resistance is projected to lead to 300 million premature deaths annually and a global healthcare burden of $100 trillion by 2050, according to the 2014 Review on Antimicrobial Resistance. The epidemic is partly driven by the inability to reliably eradicate infections of antibiotic-susceptible bacteria.

For decades, it was thought that simply reducing the population size of the bacteria to a few hundred cells would be sufficient because the immune system of an infected person can clear out the remaining bacteria.

“More recently, it became clear that small populations of bacteria really matter in the course of an infection,” Kim says. “The infectious dose — the number of bacterial cells needed to initiate an infection — turned out to be a few or tens of cells for some species of bacteria and, for others, as low as one cell.”

It was not well understood, however, why treatment of bacteria with antibiotics sometimes worked and sometimes failed. Contributing factors may include variations in the immune responses of infected people and possible mutations of bacterial cells to become more virulent.

Kim suspected that something more fundamental was a factor. Research has shown unexpected treatment failure for antibiotic-susceptible infections even in a simple organism like the C. celegans worm, a common model for the study of bacterial virulence.

By focusing on small bacteria populations, the Emory team discovered how the dynamics were different from large ones. Antibiotics induce the concentrations of bacterial cells to fluctuate. When the growth rate topped the death rate by random chance, clearance of the bacteria failed.

The researchers used this knowledge to develop a low-dose cocktail drug therapy of two different kinds of antibiotics. They combined a bactericide (which kills bacteria) and a bacteriostat (which slows the growth of bacteria) to manipulate the random fluctuation in the number of cells and boost the probability of the cell death rate topping the growth rate.

Not all antibiotics fit the model and more research is needed to refine the method for applications in a clinical setting.

 “We showed that the successful treatment of a bacterial infection with antibiotics is even more complicated than we thought,” Kim says. “We hope this knowledge leads to new strategies to fight against infections caused by antibiotic-resistant bacteria.”

CDC funds Emory project to automate analysis of mixed strains of antibiotic-resistant bacteria
Brazilian peppertree packs power to knock out antibiotic-resistant bacteria

Mathematician works to improve artificial intelligence

Emory mathematician Lars Ruthotto is pioneering a new field that applies the logic of differential equations to refine the chaos of deep learning. (Emory Photo/Video)

By April Hunt
Emory Report

 If you’ve ever told Siri to call your friend Bob and she answers with, “Calling cops,” you’ve seen the instability of artificial intelligence (AI) in action.

Those mistakes are the limitation of the AI technology known as deep learning. They arise from the design of the deep neural network, as well as the network’s “training,” which applies mathematical optimization methods to massive amounts of data rather than hand-crafting rules to accomplish a specific task.

Emory mathematician Lars Ruthotto has dedicated his research to modeling and solving such 21st century problems with the innovative use of differential equations that date back to the late 1600s. The National Science Foundation has rewarded his efforts with a CAREER Award, its most prestigious honor for junior faculty.

Put simply, Ruthotto is pioneering a new field — combining applied math, engineering and computer science — that applies the logic of differential equations to refine the chaos of deep learning.

“Focusing on this research question can impact specific areas of deep learning now as well as emerging technology,” says Ruthotto, an assistant professor in mathematics and computer science. “It’s uncharted territory, and my students and I will be at the forefront exploring it.”

The award is recognition of the new knowledge Ruthotto and his students are creating in the emerging field and also a hint of what’s to come, says Vaidy Sunderam, chair of Emory's Department of Math and Computer Science.

 “This grant establishes Emory as a research and education pioneer in innovative methods for robust deep learning, a key technology in the coming AI decade,” Sunderam says.

Read more in Emory Report.

Emory team vies for best social bot via Amazon's Alexa Prize 
CDC funds Emory project to automate analysis of mixed strains of antibiotic-resistant bacteria