I’d always filed fiber under “food for your gut bacteria,” and I never asked the obvious follow-up: what do they eat on the days it doesn’t show up? I had vaguely assumed they idle until the next bowl of lentils. A pair of Princeton papers suggests they go after protein your own body pours into the gut, and the prime suspect is the mucus lining your colon.
Food packaging may eventually list Prif right below fiber
The same work produced a pitch for your grocery cart. Joshua Rabinowitz, who led it at the Princeton branch of the Ludwig Institute for Cancer Research, has a name for plant protein that slips past digestion and feeds colon bacteria (Prif) and a prediction for it. “Food packaging may eventually list Prif right below fiber,” he said in the institute’s release. His own mice complicate that sentence. In the PNAS paper behind it, published July 27 and back in circulation this week, a diet rich in Prif and empty of fiber produced what the authors call an order-of-magnitude increase, roughly tenfold, in p-cresol sulfate. The comparison was a starch-and-casein diet, every animal was a male mouse, and p-cresol sulfate is a compound the authors class as a uremic toxin associated with worse cancer outcomes.
Two amino acids explain most of what follows. When bacteria break down protein in the colon, tyrosine becomes phenol and p-cresol, which the body then sulfates into phenol sulfate and p-cresol sulfate, a pair the Ludwig release also ties to systemic toxicity in people with kidney disease. Phenylalanine takes a different road, to hippuric acid and 3-phenylpropionate, which the paper describes as correlating with healthy body weight and supporting the gut barrier. People on plant-heavy diets tend to carry more of the second pair and less of the first, and nobody had pinned down why.
So Jenna AbuSalim, Rabinowitz and colleagues tagged amino acids with stable, non-radioactive carbon isotopes and followed the label through the mouse. After a 60-hour infusion of labeled tyrosine into the bloodstream, the tag showed up in 85 percent of circulating phenol sulfate and 57 percent of p-cresol sulfate. Tyrosine in the blood has to get back into the gut somehow, and the authors read those numbers as bacteria digesting protein the host secretes there. Labeled phenylalanine infused the same way never reached hippuric acid or 3-phenylpropionate. When the mice ate labeled algal protein instead, essentially all of that second pair traced back to the meal.
| Group | Value (percent) |
|---|---|
| Phenol sulfate | 85 |
| P-cresol sulfate | 57 |
Which left me wanting to know why a microbe would bother with the host when dinner keeps arriving from upstairs. The answer is that the host’s supply never stops. The colon secretes mucus around the clock, a slick layer of mucin glycoproteins that holds trillions of microbes away from the cells of the gut wall, and when fiber runs short, the bacteria that can break down mucin have that supply mostly to themselves. The paper doesn’t prove mucin is the protein being eaten, but it leans that way. Across the mice, Oscillospiraceae, a family the authors note is capable of digesting mucin protein, tracked circulating p-cresol sulfate with a correlation of 0.69. The same family tracked free p-cresol in stool from 130 liver transplant patients, a messy group in which many were on broad-spectrum antibiotics and no blood was available to measure the sulfate itself.
Fiber reshuffles that menu. Pectin and inulin, two soluble fibers, lowered the tyrosine-derived pair in mice while leaving the phenylalanine pair alone, and pectin shrank the share built from host protein. Cellulose, the insoluble scaffolding of plant cell walls, changed nothing. An overnight fast pushed the other way: more phenol sulfate and p-cresol sulfate, less of the other two, and a strong rise in the amount made from the host’s own protein. A mouse’s night without food isn’t your sixteen-hour window, and nobody here measured fasting people, but I’d want that tested before anyone sells fasting as a gut reset.
Fiber can’t make the phenylalanine pair, though, because that pair comes from dietary protein, and most dietary protein never reaches the colon. In this study whey and casein were about 99 percent digested before bacteria got a look, while rice protein was only about 80 percent digestible. Every plant protein tested except fava bean left extra protein in the stool, and that escaped fraction is what the group calls Prif, short for “proteins imitating fiber.” On rice protein, mice made roughly seven times more 3-phenylpropionate than on casein. The team measured only these four metabolites, hasn’t identified the bacteria responsible, and couldn’t reproduce the effect with a single dose of protein, which suggests it builds through longer-term shifts in the microbiome.
The same rice protein raised p-cresol sulfate, too. When the team crossed fiber against Prif in a two-by-two design, with four mice per group, fiber tended to push p-cresol sulfate down and Prif tended to push it up, and the fiber-free, Prif-rich corner is where the order-of-magnitude jump landed. Undigested protein arriving in a fiber-starved colon is simply more food for the bacteria making the toxin-linked pair. The authors’ conclusion is careful on exactly this point: “Both fiber and plant proteins are important to bias microbial production away from tyrosine-derived phenols.” Both, not either.
That’s my problem with a Prif callout on a package. I can already picture the bar: rice protein isolate, Prif splashed on the front, next to no fiber on the back, which is the corner that went wrong in these mice. The animals ate purified powders mixed into controlled chow, so nothing here compared an isolate against beans or lentils, and no person has eaten these diets in a trial. When I say the whole-food version, protein and fiber arriving together, is the one I’d bet on, I’m reasoning past the data, though it’s the only version that delivers the “both” the paper says you need.
An older result makes me even warier of isolates. In 2016 a University of Michigan team led by Mahesh Desai and Eric Martens reported in Cell, according to the university’s release, that mice colonized with 14 human gut species thinned their colonic mucus as fiber disappeared and fared worse against the gut pathogen Citrobacter rodentium. A diet built on purified soluble prebiotic fiber eroded the mucus just as badly as no fiber at all. Princeton’s purified inulin did lower the toxin-linked pair, so the two labs measured different things, but Michigan saw the mucus damage that Princeton’s tracers only point toward. “If you don’t feed them, they can eat you,” Martens said.
A companion paper in Nature Metabolism, described in the Ludwig release, muddies the probiotic aisle. Tracing isotopes in mice, rats and human cells, the group found the body itself makes several indoles and phenols long credited to bacteria, and those held up after antibiotics disrupted the microbiome. P-cresol sulfate fell, so for that compound the microbes are the lever, which is why the fiber result has teeth.
The work was funded by the NIH, the Ludwig Institute and Princeton. Rabinowitz discloses that he is a founder, director and stockholder of Farber Partners, an advisor and stockholder in Empress Therapeutics, and an inventor on Princeton-held patents. Keep that in mind, even though what these papers point toward is a grocery list, not a prescription.
Most of us already skip the fiber half of that list. Only 5 percent of American men and 9 percent of women meet the daily fiber target, according to an analysis of NHANES data from 2013 to 2018 presented to the American Society for Nutrition, which its authors flagged as preliminary until peer-reviewed. Falling short of a target isn’t the zero-fiber chow these mice ate, and nobody measured anyone’s mucus. Still, the half of the paper’s “both” that’s missing from most American plates is fiber.
So I’m keeping my plant protein attached to its fiber, which means beans, lentils and whole grains rather than isolates. I won’t buy anything that advertises Prif on the front until the fiber number on the back is big enough to earn it.
Sources
- PNAS: AbuSalim et al., Digestion-resistant proteins support the healthy metabolite profiles associated with plant-based diets (2026)
- Ludwig Cancer Research: A new light on the microbiome and its metabolites (July 2026)
- ScienceDaily: When your gut bacteria run out of fiber, they may start eating your gut lining (November 2026)
- ScienceDaily, University of Michigan news release on Desai et al., Cell (2016)
- American Society for Nutrition: Most Americans are not getting enough fiber in our diets (NHANES 2013 to 2018)