Autism’s Gut Connection: Microbes Could Soon Lead to New Treatments

This story appeared in the November 2020 issue as "Bacteria and the Brain." Subscribe to Discover magazine for more stories like this.

It’s not always easy to convince people that the human gut is a sublime and wondrous place worthy of special attention. Sarkis Mazmanian discovered that soon after arriving at Caltech for his first faculty job 14 years ago, when he explained to a local artist what he had in mind for the walls outside his new office.

The resulting mural greets visitors to the Mazmanian Lab today. A vaguely psychedelic, 40-foot-long, tube-shaped colon that’s pink, purple and red snakes down the hallway. In a panel next to it, fluorescent yellow and green bacteria explode out of a deeply inflamed section of the intestinal tract, like radioactive lava from outer space.

The mural is modest compared with what the scientist has been working on since. Over the last decade or so, Mazmanian has been a leading proponent of the idea that the flora of the human digestive tract has a far more powerful effect on the human body and mind than we thought — a scientific effort that earned him a $500,000 MacArthur Fellowship “Genius Grant” in 2012. Since then, Mazmanian and a small but growing cadre of fellow microbiologists have amassed a tantalizing body of evidence on the microbiome’s role in all kinds of brain disorders, including schizophrenia, Alzheimer’s disease, Parkinson’s disease and depression.

But the results they’ve seen in autism could, in the end, prove the most transformative. Autism affects about 1 in 59 children in the U.S., and involves profound social withdrawal, communication problems, and sometimes anxiety and aggression. The causes of the brain disorder have remained speculative. Now, Mazmanian and other researchers are finding that autism may be inextricably linked to — or even caused by — irregularities in the gut microbiome.

At 47, Mazmanian — with his shaved head, flannel shirt and skinny jeans — resembles a young, urban hipster on his way to write at the local café. Originally, literary life was his plan. Born in Lebanon to two Armenian refugees, neither of whom had more than a first-grade education, Mazmanian landed in the class of an energetic high school English teacher in California’s San Fernando Valley, where his family first settled. The teacher recognized his gift for language and encouraged him to pursue a career in literature. Mazmanian enrolled at UCLA in 1990, planning to major in English.

Everything changed when he took his first biology class. Hunched over his new, thick textbook in the library, reading about basic biological concepts like photosynthesis, Mazmanian felt a vast new world opening up to him.

“For the first time in my life, I wanted to turn the page and see where the story was going to go,” he says. “I think I decided that minute to become a scientist.”

Mazmanian was most fascinated by the idea that tiny organisms, invisible to the naked eye, could function as powerful, self-contained machines — powerful enough to take over and destroy the human body. After graduating with a degree in microbiology, Mazmanian joined a UCLA infectious diseases lab and began studying bacteria that cause staph infections. As his dissertation defense approached, Mazmanian read a one-page commentary penned by a prominent microbiologist, highlighting the fact that our intestines are teeming with hundreds, if not thousands, of different species of bacteria. But it was still largely unknown what they are and how they affect the human body.

When Mazmanian dug further, he found that no one had yet answered what seemed to him to be the most obvious question: Why would the human immune system, designed to attack and destroy foreign invaders, allow hundreds of species of bacteria to live and thrive in our guts unmolested? To him, the bacteria’s survival implied that we had evolved to coexist with them. And if that were so, he reasoned, there must be some benefit to both the microbes and the human body — a symbiotic relationship. But what was it?

Mazmanian set out to study the link between gut microbes and the immune system. As a postdoctoral researcher, he joined the lab of Harvard University infectious disease specialist Dennis Kasper.

To start, Mazmanian examined how the immune systems of germ-free mice — lab mice completely protected, starting at birth, from all microbes — differed from those of mice with either few or normal levels of microbes. He expected this initial census would be just a first step in a long and arduous quest for scientific pay dirt. But when he went to examine a printout of his results in the lab, he realized immediately he might already be onto something big. The germ-free mice had a 30 to 40 percent reduction in a specific type of immune cell known as helper T-cells.

Since helper T-cells play a key role in coordinating attacks against invading pathogens, the finding suggested that the immune systems of the germ-free mice were far less robust than those found in peers with normal levels of microbes.

“That was exciting, right?” Mazmanian recalls. “Obviously I repeated it and tested it in a number of different ways. Then I asked the next question: ‘Can I restore the [immune] function in an adult animal?’ ”

Mazmanian colonized the guts of the immunocompromised, germ-free mice with microbes from standard lab mice. After receiving the fecal transplant, their T-cell counts shot up. Within a month, their numbers were identical to mice raised outside the germ-free bubble.

Resolving to identify the microorganisms causing this transformation, Mazmanian resorted to trial and error. One by one, he added strains of bacteria found in the guts of mice to the guts of germ-free mice.

He got nowhere with the first five or six species he examined. Then, simply because it was convenient, he decided to test one more that was readily available in his lab. Mazmanian’s adviser, Kasper, had been studying a gut microbe called Bacteroides fragilis. When Mazmanian implanted one of Kasper’s specimens into the gut of his germ-free mice, the results were dramatic: The T-cell numbers spiked to normal. Eventually, Mazmanian demonstrated he could reproduce this effect simply by adding a single molecule that these bacteria produce, called polysaccharide A, to their guts.

“There was no logic in the choice whatsoever,” Mazmanian recalls. “[B. fragilis] was available, it came from the gut.” In other words, he got lucky.

Mazmanian dug deeper and discovered that the biggest impact B. fragilis had was on the population of a subtype of helper T-cells called regulatory, or suppressor, T-cells. These cells play a key role in preventing the immune system from attacking its host body, protecting against autoimmune or inflammatory diseases. It was the first time any scientist had demonstrated that a single compound from a single microbe could reverse a specific problem with the immune system.

To Mazmanian, the finding, published in 2005 in the journal Cell, alluded to new approaches to treating a wide array of autoimmune, inflammatory and allergic disorders. What if it were possible to help a faulty immune system by tweaking a patient’s microbiome? It was with this exploration in mind that he arrived in Pasadena in 2006 to set up his lab at Caltech.

A few years later, Mazmanian was having lunch on campus with neuroscientist and colleague Paul Patterson. Patterson had been preoccupied with a mystery that had, for years, confounded those studying autism in humans: When pregnant mothers have a severe infection in the second trimester, their babies are much more likely to develop autism.

As Mazmanian tells it, Patterson was a man of few words, and at lunch Mazmanian was “going on and on” about his own work.

“You know,” Patterson interjected thoughtfully, “I think kids with autism have GI issues.” Patterson recalled reading that something like 60 percent of children with autism had some form of clinical GI problem, such as bloating, constipation, flatulence or diarrhea. Was it possible, he wondered, that there was a microbiome connection?

As they talked, Mazmanian’s excitement grew.

A few years earlier, Patterson had discovered that when he exposed pregnant mice to pathogens like the influenza virus, they gave birth to pups that grew up more likely to be startled by loud noises, to shy away from social contact and to groom themselves repetitively — symptoms that resemble those of autism. Patterson was in the process of comparing the brains of these autism-mimicking mice with their neurotypical cousins to see if he could detect any differences that might explain how the maternal immune system was somehow interfering with the pups’ brain development.

Mazmanian had a suggestion: The next time Patterson sacrificed one of his autistic mice to study their brains, what if he set the intestines aside for his colleague down the hall?

When the guts arrived in Mazmanian’s lab, he found that the intestines of the neurotypical mice looked normal. But the guts of the autism-mimicking offspring were almost uniformly inflamed. Could it be that the microbiome was the cause of this inflammation? And could that, in turn, be somehow connected to the behavioral symptoms?

Throughout the winter and spring of 2012, Mazmanian and Patterson continued their conversation. Mazmanian found distinct differences in the microbiomes of the mice. And, they noticed, the mice with the features of autism had leaky gut syndrome, an increased permeability of the gut lining that can allow pathogens and allergens to leach out. This condition had also been reported in children with autism.

So Mazmanian and Patterson turned their attention outside the gut. They took blood samples to see if any gut microbes, or the compounds they produce, were circulating in the rest of the body. They homed in on one molecule in particular, called 4-ethylphenyl sulfate, which was roughly 45 times as abundant in the mice that had symptoms of autism. And it looked familiar: Structurally, it was almost identical to a molecule recently found to be significantly elevated in human children with autism.

It was enough to take the next step. Every day for three weeks, Mazmanian injected the molecule, harvested from the mice with autism-like symptoms, directly into the bloodstream of 5-week-old normal lab mice (the age at which the autistic mice normally developed leaky gut). Then Mazmanian and his team gave them a series of behavioral tests. The mice were far more easily startled and were less comfortable in large empty spaces than their untreated peers, indications of an increase in anxiety-related behaviors commonly seen in the mice with autism-like symptoms. The researchers published their results in Cell in 2013.

Though surprising, the data made sense in some ways. Many drug companies rely on small-molecule drugs that can be taken orally, but still manage to cross the blood-brain barrier and affect behavior. It seemed entirely possible that small molecules, created by bacteria in the gut, could enter the bloodstream and reach the brain. And they don’t even have to leak out of the gut to do so.

Patterson died in 2014, at age 70, just six months after the publication of the duo’s groundbreaking Cell paper. Around the same time, a series of parallel experiments in a clinic hundreds of miles away was already paving the way forward. While Patterson and Mazmanian had been working in mice, Rosa Krajmalnik-Brown, a microbiologist at Arizona State University, had teamed up with Jim Adams, who directs the university’s autism and Asperger’s research program, to study humans.

The researchers were conducting a detailed analysis of the microbiome of human autism patients and found that the bacteria were far less diverse in the children with autism. Notably, several important species involved in the digestion of carbohydrates were severely depleted.

Krajmalnik-Brown and Adams launched a preliminary trial to test the effects of fecal transplants on 18 children between the ages of 7 and 16 with severe autism, who also had severe GI issues. The researchers administered powerful antibiotics to kill off the microbiomes of the children and followed them with a bowel cleanse. They then replaced the microbes with transplanted flora taken from the guts of healthy neurotypical adult volunteers.

The results were better than anyone could have expected. The procedure resulted in a large reduction in GI symptoms and increased the diversity of bacteria in the children’s guts. But more significantly, their neurological symptoms were reduced.

At the onset of the study in 2017, an independent evaluator found 83 percent of participants had severe autism. Two years after the initial trial, only 17 percent were rated as severely autistic. And 44 percent were no longer on the autism scale.

“[My child] did a complete 180,” says Dana Woods, whose then-7-year-old son Ethan enrolled in the initial study five years ago. “His ability to communicate is so much different now. He’s just so much more present. He’s so much more aware. He’s no longer in occupational therapy. He’s no longer in speech therapy. After the study, he tested two points away from a neurotypical child.”

In their first report on the trial in 2017, the team highlighted a number of distinct changes in the microbiome after the transplants, in particular a surge in the populations of three types of bacteria. Among them was a four-fold increase in Bifidobacterium, a probiotic organism that seems to play a key role in the maintenance of a healthy gut.

But figuring out what was happening on a cellular level — to really look inside some guts — would require another vehicle. The ASU team needed Mazmanian’s mice.

“At the end of the day, what we care about is healing people and how the microbiome affects people,” explains Krajmalnik-Brown. “That’s why we work with people. But with mice you can do things that are more mechanistic.”

Together, Krajmalnik-Brown, Mazmanian and their collaborators would uncover some tantalizing new insights that go a long way to solving the mystery. In May 2019, the team published another high-profile paper in Cell, after they transplanted stool samples from Krajmalnik-Brown’s severely autistic patients into the guts of Mazmanian’s germ-free mice. The offspring of these mice showed the autism-like symptoms, such as repetitive and compulsive behavior.

This time, the team dug even deeper into the biochemical processes playing out in the brain, looking not just at behavior but at the chemicals involved in creating it. The mice that developed autism-like behaviors had measurably lower levels of two substances called taurine and 5-aminovaleric acid (5AV). When they dug into the literature, the team learned that these two substances are known to mimic activity of a key signaling agent in the brain called gamma-aminobutyric acid (GABA) — a neurotransmitter that other studies have found is deficient in the brains of children with autism.

What’s more, some have speculated that the tendency of children with autism to experience sensory overstimulation may stem from the inability to tamp down overexcited neurons. A lack of GABA could lead to just that.

The scientists next orally administered high levels of taurine and 5AV to pregnant mice with the autistic children’s microbiomes. When their pups were born, the researchers continued to feed the young the substances until they reached adulthood. Compared with untreated animals, the second-generation mice had significantly fewer behavioral symptoms. Taurine reduced repetitive behavior, as measured by marble burying, increased the level of social interaction, and relieved anxiety. Mice administered 5AV were more active and social.

“We healed humans with behavioral problems,” says Krajmalnik-Brown. “[And we] transferred some of those deficits and behaviors to mice — basically the opposite. It’s huge.”

Mazmanian hopes to take the next step in the months ahead.

“I can flip a switch, turn on a light, I know that switch turns on that light. I don’t know the circuit, I don’t know where the wire is,” Mazmanian says. “Exactly how that’s happening … we just don’t understand that.”

This most recent study, by itself, hardly proves that dysregulated microbiomes cause the brain disorder — a point that plenty of other scientists skeptical of Mazmanian’s work are happy to make.

“The paper made a big splash, but trying to model psychiatric-related human conditions in mice, in my view, is a little bit of a stretch,” says Sangram Sisodia, a neurobiologist at the University of Chicago who studies the microbiome. “A mouse with autism?”

Nor was that the only criticism. Several researchers have suggested that the group didn’t give proper attention to one of their tests ­— one whose results conflicted with their thesis ­— while others found flaws in the statistical methods they used to assess their results. Mazmanian downplays these criticisms, but agrees the work is not yet conclusive.

Meanwhile, the ASU trial has also engendered skepticism, mainly due to its tiny sample size, the lack of a control group and the methods by which the children were assessed for autism severity. Krajmalnik-Brown and Adams say they stand by their results, but agree more research is needed. In recent months, they have launched two new studies that will address these issues.

Adams insists the work is already changing lives. “We followed up with every one of our 18 participants,” he says, referring to the children who received fecal transplants. “Sure enough, we found that most of the GI benefits had remained. And family after family said their child just slowly, steadily continued making more improvement.” They published the update in Scientific Reports in spring 2019.

“I’m not ready to say the case is closed,” says Mazmanian. “Healthy skepticism is a good thing. I believe the preclinical data, I believe the mouse data. But there’s a lot of studies that still need to be done.”

Prior to his fecal transplant at age 7, Ethan Loyola suffered from chronic and severe diarrhea, constipation and cramping, symptoms so extreme that to his mother, Dana, he sounded like “a bit like a woman in labor when he was trying to have a bowel movement.”

“It was just awful watching your child go through this,” she says, explaining that when she enrolled her autistic son in the Arizona State study, her “only goal was to fix his gut.”

Remarkably, Ethan’s agony began to disappear just a few weeks into the trial. But that was not the most dramatic difference. Before the transplant, Ethan’s speech was drawn out and slow, his language skills rudimentary. He seemed to live in his own bubble. He had frequent outbursts. For as long as Dana could remember, her mornings with Ethan had been marked by arguing, fighting, pushing and anger. But then one morning, something shocking happened.

“He woke me up one morning with his face right in my face with this big smile and he said, ‘Morning, Mom!’ ” she recalls. “And he was just excited and happy and ready to go about his day with this big smile. It choked me up to the point where I teared up because I had never experienced a happy kid in the morning.”

Later, Ethan carried over an iPad and opened an app with a talking cat that repeats back the words children speak aloud. He played back a video recording of himself from just a few weeks earlier.

“[He] looks me in the eye and says, ‘Mom, why did I talk like that? What is wrong with me?’ And as soon as he did that, I caught my breath. I had to compose myself and say, ‘I don’t know. But do you feel better? Do you feel different? Why do you think?’ ”

Ethan’s communication skills had already begun to improve. Within a year of the study, his speech therapist graduated him from speech therapy because he had met all his goals.

“He went from one end of the rainbow all the way to the other end of the rainbow,” she says. “Prior to the study, I was very afraid. My biggest fear was ‘how is he going to navigate the world when I’m not here?’ And I think I have a lot of hope now that he is going to be OK now on his own.”

Editor's note: An earlier version of this story had Ethan's last name incorrect. It's Loyola.