Houston Methodist is one of 25 academic groups and industry partners that will be collaborating on a multi-institutional, $104 million effort to study bacteria and antibiotic resistance led by Harvard Medical School researcher Johan Paulsson.


The work is funded by the U.S. Department of Health and Human Services’ newly established Advanced Research Projects Agency for Health (ARPA-H) in an effort to address an unfolding crisis of antibiotic resistance that is expected to get worse as more bacteria become impervious to existing drugs.


The Houston Methodist Research Institute will be a key clinical partner, having a role in translational and clinical development, and helping to develop the technology that results from this research partnership.


Under Paulsson’s leadership, scientists from 25 research groups in California, Delaware, Georgia, Maryland, Massachusetts, Oklahoma, Texas, Virginia, Wisconsin, and the United Kingdom will work to develop novel microscopy, microfluidics, single-cell assays, and machine learning tools into technology for identifying bacteria and understanding their behavior.


The researchers plan to use this technology to improve diagnosis of bacterial infections in the clinic, and to aid in the development of more effective antibiotics in the lab. If it succeeds, the research has the potential to drastically transform how bacterial infections are diagnosed and treated.


More broadly, the work will aim to unravel the mysteries of bacterial behavior and the biologic mechanisms of bacterial diseases.


“What makes this novel is not that 25 research groups are funded to study antibiotic resistance, but that 25 groups with expertise in optics, mathematics, microbiology, and medicine can come together and do that as one coordinated team,” said Paulsson, professor of systems biology in the Blavatnik Institute at HMS.


The slow-moving pandemic


Bacterial infections are a serious global health threat — implicated as the second leading cause of death worldwide — and bacteria that have developed resistance to antibiotics are a major part of the problem.


According to some estimates, in the past five years, antibiotic-resistant bacteria may have caused more deaths globally than COVID-19, leading some experts to describe antibiotic resistance as “the slow-moving pandemic.”


“It’s a pandemic with pretty high death rates that also makes an awful lot of people seriously ill. It affects almost everybody, but because it’s so slow, people have, in a sense, accepted it,” Paulsson said.


As clinicians’ arsenal of effective antibiotics shrinks, bacteria will become a growing threat. Under this scenario, bacterial infections such as tuberculosis and staph will run rampant and be harder to treat. Drug-resistant organisms will render even the most routine procedures such as dental surgeries, cesarean sections, and appendectomies much riskier, and more complex surgeries such as organ transplants nearly impossible.


“A lot of modern medicine relies on the ability to control bacteria, so if we lose that ability, many seemingly unrelated medical interventions won’t be possible anymore,” Paulsson said.


Despite the growing threat, the fight against antibiotic-resistant bacteria has stalled. One challenge is that current technologies for detecting and identifying bacteria in clinical settings aren’t fast, cheap, reliable, or broad enough to be effective.


Additionally, progress has been slow in developing new classes of antibiotics that work against resistant bacteria. The last novel class of antibiotics was discovered nearly 40 years ago, in 1987. Finally, research has largely focused on studying average bacterial cells rather than the outliers that tend to drive evolution and antibiotic resistance.


The project will tackle all of these challenges head-on.


A diagnostic bottleneck


A mathematician by training, Paulsson became interested in studying the simplest possible biologic systems that could be analyzed and studied mathematically, such as certain simple gene circuits in the bacterium E. coli.


As Paulsson started studying E. coli experimentally, he ran into a logistical challenge: The existing tools for handling bacteria under the microscope were not fast or rich enough to capture the rare events in a bacterium’s life that he was studying. To overcome this challenge, Paulsson developed new tools — and eventually, began using those tools to study bacterial pathogens.


A turning point came two years ago, when Paulsson developed an unknown infection, most likely from a tick on his dog. The infection quickly progressed to sepsis. Paulsson was surprised to find out that the standard clinical approach for identifying the bacterial culprit is to test the patient’s blood for one or a few types of bacteria at a time and treat the patient with a series of antibiotics — a cumbersome process that often leads to delays in giving the patient the correct drug.


“It got serious fast, and I was lucky that the third antibiotic worked. I asked them, why don’t you just look at the blood agnostically and see what’s physically in there, instead of identifying specific genes, and they said, oh no, we can’t do that, we can only ask specific questions,” Paulsson recalled.


Paulsson thought that the combination of microfluidic-based tools, modern microscopy methods, and machine learning his lab had been developing might be able to address this diagnostic bottleneck. The day he was released from the hospital, he reached out to some colleagues.


A multifaceted approach


The newly funded ARPA-H project will build on tools from all of the groups involved, and has four main components:


  • Microscopy: The use of microscopes to visually examine bacteria. The researchers will combine six modern microscopy techniques to make bacterial imaging faster, cheaper, and more detailed than what has been possible until now.
  • Microfluidics: a technique used to extract bacteria out of the blood and prepare them for imaging under a microscope.
  • Microfluidics will allow the researchers to design and run single-cell assays on bacteria. Among other things, the tests will identify bacterial cells, genotype them on an individual level, and assess their responses to antibiotics.
  • The researchers will be working with extremely large numbers of bacterial cells that must be differentiated based on small differences that cannot easily be detected by the human eye. Thus, they will rely on machine learning tools and mathematics to help them process the images and identify the cell types present.


The foundational work behind these advances came from curiosity-driven basic research that had no specific application as a goal post, Paulsson said.


“To make new discoveries, we need to let people pursue pure, basic, even esoteric research, which can often lead to discoveries that have initially unforeseen but transformative applications down the road,” he added.


Translating technology


A main goal of the project is to build this technology into an instrument that clinicians can use to quickly, accurately, and cheaply diagnose a bacterial infection from a blood sample. Such an instrument, Paulsson said, could dramatically speed up diagnosis because of its powerful ability to find and identify bacterial cells.


“Right now, we can’t diagnose people quickly or agnostically enough,” Paulsson said, noting that a patient sample must be grown in the lab for several days before it can be analyzed to make an accurate and definitive diagnosis.


“In principle, we think we can develop technology to carry out the whole process in minutes — starting the moment the sample is taken, and not waiting for bacteria to be cultivated and isolated,” he said.


A second goal is to apply the technology to antibiotic discovery — an effort that will be spearheaded by scientist Kim Lewis of the Antimicrobial Discovery Center at Northeastern University.


Most existing antibiotics come from natural sources such as plants, animals, and microorganisms, yet discovery process is slow, and there are many potential sources that scientists have not yet found or tested.


“It’s not that the world has been exhausted, it’s more that the looking is slow,” Paulsson said. “We’ve only mined a few percent of the natural diversity out there.”


The technology could speed up the process by allowing researchers to test potential antibiotics against bacteria at high-throughput and with greater resolution. Ultimately, Paulsson hopes that the project will identify several new leads for antibiotics.


In a similar vein, the technology could be used to systematically test combinations of antibiotics to find those that are most effective against different bacterial infections.


Finally, the researchers plan to use the technology to advance their basic biological understanding of bacteria and bacterial dynamics. In particular, they want to study the small fraction of outlier bacteria in any population that survive antibiotic treatment.


It is these ‘persister’ bacteria, Paulsson said, that often cause infections to come back, ultimately fueling further antibiotic resistance. However, they have been difficult to study because their survivalist abilities appear to be triggered by seemingly rare random events in their lives, rather than genetics.


With this new technology, the researchers will be able to study these bacteria at a larger scale, and in greater detail to further probe the mechanisms underlying their outlier behavior.


“These hard-to-eradicate bacterial cells are like the survivalists that build bunkers in the desert to survive some pending doom,” Paulsson said. “We want to understand why and how they become survivalists.”


Breaking down barriers


Although Paulsson is the lead investigator, the project involves research groups at 10 academic institutions, five hospitals, and four companies throughout the United States. In addition to Paulsson and Lewis, other researchers include Eleftherios Mylonakis, chair of the Department of Medicine at Houston Methodist Hospital and professor of medicine with Weill Cornell Medical College, as well as many of the country’s other leaders in antimicrobial resistance, optics, machine learning, and more. The project will receive $70 million up front, and an additional $34 million if the first phase succeeds.


The project is one of the first few funded by ARPA-H, a research funding agency created last year. ARPA-H supports biomedical and health breakthroughs by targeting areas where progress has been slow through traditional research and commercial activities. The first project, announced in August, focuses on developing new mRNA platforms that prime the immune system to fight cancer and other diseases.


Industry involvement is a requirement of ARPA-H funding — an effort to encourage greater cross-pollination between academia, industry, and government to take advantage of the unique strengths that each brings to the execution of such ambitious projects.


Four U.S. companies co-applied for the commercial side of the new grant. The overall industry effort will be led by Latham BioPharm Group. Harvard start-ups Bifrost Biosystems and BLASTID, which are commercializing technology developed in the Paulsson lab, will create instruments for screening and diagnostics. Another company, Gener8, will handle manufacturing.


“One idea behind ARPA-H is to break down institutional barriers to see what can be accomplished if people can work in the same direction regardless of affiliation,” Paulsson said.