Emulsifiers, sweeteners and food additives: what they may do in the gut

A team around Malin Johansson in Gothenburg studied the mucus layer in the colon of mice and rats in 2008, with the microscope and with an analysis of the proteins involved.

They found two layers on top of each other, together roughly 150 micrometres thick. At their core both consist of the same gel-forming protein, the mucin Muc2. The outer one is loose, movable and densely populated. The inner one is firmly attached, densely packed and, in the healthy state, practically free of bacteria. In animals lacking Muc2, the bacteria sat directly on the gut cells and deep in the crypts, matching the inflammation and tumour development seen in exactly those animals.

For you this means: the distance between bacteria and gut wall is not a random quantity but an actively built structure. And when it becomes thinner, who touches whom changes.

The same working group summarised the state of knowledge on the mucus system of the entire digestive tract in 2013.

The small intestine has a single, unattached layer. Stomach and colon have two. In the colon the outer layer is the living space of the commensals, the inner one is impermeable to bacteria and is renewed hourly by the goblet cells at the surface. Certain bacteria and parasites carry protein-cleaving enzymes with which they can break down mucins. Disturbances of the immune system can also make the inner layer permeable. If bacteria reach the epithelium, the inflammatory response starts up.

For you this means: a one-off exposure is rarely the issue. The question is what happens every day over years, and whether rebuilding keeps up with breakdown.

In 2015 Benoit Chassaing and Andrew Gewirtz gave mice relatively low concentrations of carboxymethylcellulose, that is E 466, or polysorbate 80, that is E 433, via the drinking water. Healthy animals, inflammation-prone animals, germ-free animals and animals after faecal transplantation were studied.

Both emulsifiers triggered low-grade inflammation in healthy animals, along with weight gain and a metabolic syndrome. In inflammation-prone animals a pronounced colitis developed. This was accompanied by a thinner mucus layer, by bacteria moving closer to the wall, by an altered species composition and by more pro-inflammatory potential. Germ-free animals and the faecal transfer showed that the change in the bacteria was necessary and sufficient for both effects.

For you this means: here is a plausible mechanism, cleanly worked through. And these are mice. That sentence belongs in the same place as the result, not in a footnote.

In 2017 the same group kept human gut bacteria alive in a bioreactor, that is without a body around them, and exposed them to carboxymethylcellulose or polysorbate 80. They then transferred the treated bacteria into germ-free mice.

Both substances directly increased the pro-inflammatory potential of the human bacterial community, measured as bioactive flagellin. With carboxymethylcellulose this happened quickly, within a day, via altered gene activity. With polysorbate 80 it was slower and ran via an altered species composition. Transfer into germ-free mice reproduced many of these changes.

For you this means: the effect apparently acts directly on the bacteria and needs no pre-existing inflammation. It still remains test tube plus mouse.

In 2021 Sabrine Naimi and colleagues tested twenty common emulsifiers in the same bioreactor system against the same human bacterial community. Density, composition, gene activity and pro-inflammatory potential were measured.

Carboxymethylcellulose and polysorbate 80 showed sustained adverse effects. Many of the other eighteen substances did so as well, most notably various carrageenans and several plant gums. Some substances, by contrast, did not meaningfully change the bacterial community in this model. Lecithin is named explicitly.

For you this means: the ingredient list can be sorted. Anyone who deletes every E number across the board also hits the unremarkable ones and does not end up with a better result.

A team around Benoit Chassaing and James Lewis carried out a double-blind controlled feeding study in 2022. Sixteen healthy adults lived for eleven days on fully controlled meals. Nine received a completely emulsifier-free diet, seven the same diet plus fifteen grams of carboxymethylcellulose per day.

The emulsifier group reported somewhat more abdominal discomfort after eating. The diversity of gut bacteria decreased. In the stool, short-chain fatty acids and free amino acids fell, that is exactly the substances the intestinal lining draws its energy from. In every participant, tissue samples were taken by sigmoidoscopy before and after the phase to measure how close the bacteria sit to the wall. On average this distance did not change in either group. In two of the seven people in the emulsifier group, however, bacteria were found in the normally sterile inner mucus layer, together with marked shifts in composition. In the control group this was the case in none of the nine people.

For you this means: something happens, and it does not happen to everyone. Five of seven did not show this encroachment.

In 2025 Héloïse Rytter and colleagues took the gut bacteria of the FRESH participants, exposed them again to carboxymethylcellulose in the bioreactor and then transferred them into germ-free mice with an increased tendency to inflammation.

The model reproduced the differing sensitivity of the people. Some control participants who had received no carboxymethylcellulose at all in FRESH also turned out to be sensitive in the model. Sensitivity could be recognised from a signature of the baseline microbiome. Bacterial communities classed as sensitive by the model triggered colitis in the mice after feeding, those classed as insensitive did not.

For you this means: sensitivity to emulsifiers appears to be a property of your own bacterial community. Outside research there is no test for it so far.

In 2014 Jotham Suez and Eran Elinav gave mice saccharin, sucralose or aspartame and added antibiotic experiments, germ-free animals and faecal transfer. On top of that came a small intervention group of healthy humans who received saccharin.

In the mice, glucose intolerance developed under sweeteners. Antibiotics abolished the effect. Faecal transfer from sweetener-fed animals into germ-free animals carried the glucose intolerance across. In the small human group, some people showed a similar shift, others did not react at all.

For you this means: this is where the idea arose that sweeteners might act via the gut bacteria. At the decisive point it was mouse, and the human group was tiny.

In 2022 the same group randomised one hundred and twenty healthy adults to saccharin, sucralose, aspartame, stevia, the carrier glucose alone or nothing at all. For two weeks, in sachets, in amounts below the acceptable daily intake.

All four sweeteners as a group changed the stool and oral microbiome as well as the metabolic profile in the blood. Saccharin and sucralose additionally worsened the glucose response measurably, aspartame and stevia did not in this design. Selected human microbiomes were then transferred into germ-free mice. The animals showed the glucose response of their respective donors, that is of the strongest and of the weakest responders.

For you this means: sweeteners are not neutral for the gut. The answer is strongly person-dependent, and in both directions. Two weeks is also short.

The WHO commissioned a systematic review comparing higher against lower sweetener intake in adults and children.

In short-term randomised studies, higher intake slightly lowered body weight and BMI. In long-term observational studies it was associated with higher BMI and a higher risk of obesity. The overall certainty of the evidence was rated low. The recommendation is conditional: not to use sweeteners as a means of weight control or of lowering the risk of non-communicable diseases. People with existing diabetes and sweeteners in personal care products such as toothpaste are explicitly excluded.

If you have existing diabetes, excluded means something very concrete: this recommendation does not apply to you. Please do not swap sweetener for sugar on your own initiative. Discuss it with the practice that looks after your blood sugar management. And any ongoing medication stays as it is.

For you this means: conditional means the WHO regards its own evidence base as uncertain. And the statement is not that sweeteners make you fat. It is that a long-term benefit for weight control is not established. That is a difference.

The food additives panel of the European Food Safety Authority fully re-evaluated erythritol, that is E 968, in December 2023. The new cardiovascular papers and an estimate of actual intake in Europe were included.

For the first time a numerical value was set: 0.5 grams per kilogram of body weight per day. The critical endpoint for it was diarrhoea, not cancer and not thrombosis. The estimate showed that high consumers and especially children can exceed this value. On the cardiovascular papers the panel concluded that the available data do not establish a causal relationship between erythritol as a food additive and an increased cardiovascular risk, and it called for further research.

For you this means: for a person weighing 70 kilograms this value corresponds to roughly 35 grams per day. The practically relevant effect remains the laxative one. The thrombosis question is open, not answered.

In 2017 Sumit Bhattacharyya and Joanne Tobacman trained people with ulcerative colitis in remission to follow a carrageenan-free diet. In addition, all of them received, in blinded fashion, either capsules containing carrageenan, in an amount below the usual daily intake from food, or placebo capsules. Follow-up every two weeks by telephone, until relapse or up to one year.

Twelve people completed the questionnaires. Three from the carrageenan group relapsed, none in the placebo group, with a p-value of 0.046. Interleukin-6 rose in the carrageenan group, faecal calprotectin showed a trend in the same direction. In the placebo group they did not.

For you this means: twelve people are a signal and not proof. And the signal concerns people with ulcerative colitis. Transferring it to healthy people is not supported. What else counts in inflammatory bowel disease is covered in the article on Crohn's disease and ulcerative colitis.

In 2019 Federica Laudisi and Giovanni Monteleone gave mice maltodextrin, propylene glycol or gelatine via the drinking water and then challenged the animals. Cell culture experiments on a mucus-producing human cell line ran in parallel.

Only maltodextrin worsened intestinal inflammation in both models, propylene glycol and gelatine did not. The mechanism ran via stress in the endoplasmic reticulum of the goblet cells, mediated by IRE1-beta and the p38 MAP kinase. The consequence: less mucin-2 and less mucus. The bacteria at the mucosa barely changed. An inhibitor of this cellular stress prevented the mucin loss and eased the colitis.

For you this means: here is a second, independent route to a thinner mucus layer, and it does not run via the bacteria but via the mucus-producing cells themselves. Mouse and cell culture. There is no human study on it so far.

In 2021 the food additives panel of the European Food Safety Authority re-evaluated all available data on titanium dioxide, that is E 171, with a focus on particle size, uptake and genetic damage.

The panel could not rule out genotoxicity after uptake of the particles and therefore saw no way to derive a safe intake level. Conclusion: E 171 can no longer be classed as a safe food additive. The regulatory consequence was Regulation 2022/63. Since 7 February 2022, titanium dioxide has no longer been authorised as a food additive in the EU, and the transition period for goods already produced ended on 7 August 2022.

For you this means: in the German supermarket, E 171 no longer appears on the ingredient list. In medicines and in toothpaste the substance can still occur. Anyone reading older lists online will still find it as a current topic. It is not one.

In 2019 Gabriela Pinget and Laurence Macia gave mice food-grade titanium dioxide via the drinking water and studied the composition and spatial distribution of the gut bacteria as well as the gene activity of the mucosa.

The species composition barely changed. What did change were the bacterial metabolites and the spatial distribution: biofilms formed more often. The activity of the mucin-2 gene fell, that of the beta-defensin gene rose. Alongside this there were shorter crypts, more immune cells and raised inflammatory messengers.

For you this means: historically relevant, practically superseded. And a good example of European assessment responding to new data.

In 2019 Lucia Hrncirova and Tomas Hrncir colonised germ-free mice with human gut flora, some with an intact and some with an altered genetic background. The animals then received a mixture of common antimicrobial preservatives in amounts that are actually reached in European populations.

The mixture triggered a dysbiosis: overgrowth of Proteobacteria, a decline in Clostridiales, falling diversity. The bacterial community in the altered genetic background reacted considerably more sensitively than in the normal one.

For you this means: preservatives are meant to inhibit microorganisms, and they do not distinguish between spoilage organisms and gut bacteria. Whether this means anything clinically in humans is open. What stands out is that the genetic make-up co-determines how sensitively the system reacts.

In 2019 Kevin Hall and his team had twenty weight-stable adults live for four weeks in the clinical centre of the American health agency. Two weeks of ultra-processed food, two weeks of unprocessed food, the order randomly assigned. The meals were matched for calories offered, energy density, macronutrients, sugar, sodium and fibre. Participants could eat as much or as little as they wanted.

On the ultra-processed diet, participants ate 508 kilocalories more per day, with a standard error of 106 and a p-value of 0.0001. Of these, 280 kilocalories came from carbohydrates and 230 from fat, while protein intake stayed the same. Weight rose by 0.9 kilograms on the ultra-processed diet and fell by 0.9 kilograms on the unprocessed one.

For you this means: this study is experimental rather than observational, which makes it strong. It shows an effect on the amount eaten, not on the mucus layer. And it says nothing about which component is responsible. Twenty people, four weeks.

In 2020 the International Organization for the Study of Inflammatory Bowel Diseases published a consensus document from its nutrition cluster, explicitly labelled as expert opinion on the basis of the best available evidence in each case.

For people with Crohn's disease and ulcerative colitis it recommends reducing the intake of certain additives, among them maltodextrin, carrageenan, carboxymethylcellulose and polysorbate 80, and limiting titanium dioxide and sulphites. Each recommendation carries its level of evidence, which in several cases is low.

For you this means: the only professional body with concrete additive recommendations issues them for people with inflammatory bowel disease, not for the general population. This distinction is the core of the topic and is constantly blurred online.

The S3 guideline on irritable bowel syndrome from DGVS and DGNM, updated in 2021 under AWMF registry number 021/016, has a markedly expanded nutrition chapter.

Its focus is on symptom-oriented approaches, above all the low-FODMAP diet in three phases. Far-reaching restriction diets without evidence of clinical effectiveness are not recommended. The guideline does not issue a general recommendation to systematically avoid food additives.

For you this means: this caution is well founded. What is simply missing is the intervention study that would show that avoiding additives eases symptoms in irritable bowel syndrome. Without that evidence, a recommendation would be a restriction without an established gain. What a sensible search for causes in IBS looks like is covered in the article on IBS and the search for causes.