An oral biologist, a computational protein engineer, and a food scientist walk into a bar. They grab a few pints of beer, snatch some pretzels, and head back to the lab to run experiments with their glutenous loot.

While listing them sounds like the start of a bad joke, researchers with this broad range of backgrounds have found a common application for their skillsets in developing treatments for celiac disease. This autoimmune disorder is characterized by an irregular immune response to ingested gluten, a protein present in wheat, barley, and rye, which damages the small intestine and manifests in gastrointestinal symptoms and systemic health problems such as anemia and osteoporosis (1). Celiac disease is among the most common autoimmune diseases, affecting more than one percent of people worldwide. Despite its prevalence, there are no approved drugs for the disease, leaving patients with the challenging feat of avoiding ubiquitous and often hidden sources of dietary gluten.  

To provide greater flexibility for people with celiac disease, this diverse lineup of researchers plans to disrupt the immune response to gluten at several critical crossroads. By building upon enzymes in natural organisms and leveraging the latest advances in protein design, scientists are developing enzymes that can degrade immunogenic gluten-derived peptides. They are designing these gluten-degrading enzymes to withstand the harsh conditions of the gastrointestinal tract or food manufacturing processes, allowing the proteins to be administered as drugs or directly incorporated into specialized food products. Researchers are also developing drugs that block the immunogenic form of the peptides and exploring an immune tolerization approach to reprogram the immune response to gluten. Together, they aim to ease the constant burden that the daily experience of eating carries for people with celiac disease. 

Mind your Ps and Qs

 Eva Helmerhorst, an oral biologist at Boston University, originally analyzed the peptide composition of saliva to study how long antimicrobial peptides can survive in salivary conditions. She stumbled into celiac disease research when she discovered that many of the peptides displayed a distinct pattern of cleavage after a proline-glutamine (PQ) motif. “We were interested in that particular cleavage site because PQ is also highly prevalent in gluten,” Helmerhorst said.

Since the major digestive enzymes struggle to cleave these amino acids, ingested gluten is partially broken down into PQ-rich peptides. These peptides penetrate into the lining of the small intestine, where they are modified by the enzyme tissue transglutaminase 2. In people with celiac disease, the peptides are then presented to T cells that launch an inflammatory attack at the small intestine. 

An enzyme that degrades these problematic peptides could prevent the harmful immune response, and Helmerhorst was prepared to sift through saliva to find it. To determine which enzyme was responsible for the PQ cleavage pattern she observed, her team cultured bacterial colonies from saliva samples and tested their activity toward gluten-derived substrates, eventually identifying a subtilisin enzyme in Rothia mucilaginosa and Rothia aeria  bacteria as the snipper (2,3). Their discovery revealed that microbes in the oral cavity may play an unappreciated role in initiating the digestion of gluten during its first exposure to the gastrointestinal tract. “We shouldn't underestimate the ability of enzymes to digest gluten while you chew the food,” Helmerhorst said.

For the subtilisin enzyme to show potential as a treatment for celiac disease, it needs to be administered in higher amounts and farther down the gastrointestinal tract where it can interact with food for much longer. Unfortunately, these orally optimal enzymes are no match for the stomach, where an acidic pH and pepsin enzymes destroy most proteins in their path. To shield their gluten-degrading subtilisin from this extreme environment, the researchers attached a stabilizing polymer to the enzyme and encapsulated it in an enteric coating (4). “It passes through the stomach, and then when it reaches the small intestine where the pH is higher, the enteric coating comes off, and the contents are released,” Helmerhorst said. 

What if we kill the bacterium in a way that doesn't kill the enzymatic activity associated with the bacterium? 
- Eva Helmerhorst, Boston University

While this strategy preserved the activity of the enzyme, an effective celiac disease therapeutic would not simply survive the journey through the stomach, but perform its full gluten-degrading activity in the stomach before the peptides can reach the small intestine and trigger the immune response (5). Rather than isolating the subtilisin enzyme from the bacteria, the researchers wondered if they could deliver the full bacterium as a probiotic. 

The larger bacterium provides a natural protective coating for subtilisin in the stomach, and the enzyme retains its ability to access and degrade the gluten-derived peptides in this form. Helmerhorst’s team administered R. aeria  to mice by mixing it into their food and analyzed their digestive contents. The bacteria reduced the levels of two major immunogenic gluten-derived peptides by approximately 20 and 33 percent in the stomach before they reached the small intestine (5). 

As naturally occurring microbes in the mouth, R. mucilaginosa and R. aeria  seem well suited for probiotic applications. However, these bacteria still raise concerns about colonizing the gut and causing infection, especially in immunocompromised people. “Even though it's a natural bacterium and already there, if you give it in a probiotic dose, which is very high numbers, all those questions could come up,” Helmerhorst said. “So we thought, what if we kill the bacterium in a way that doesn't kill the enzymatic activity associated with the bacterium?” The researchers exposed R. aeria  to ethanol and found that while the bacteria was completely deactivated, the enzyme retained more than 90 percent of its activity (5). As ethanol can be easily evaporated before administering the bacteria, this simple treatment could be used to prepare safe and effective enzymatic bacterial preparations. 

Although the subtilisin enzyme hasn’t yet demonstrated sufficient gluten-degrading activity for treating celiac disease, Helmerhorst hopes that it might find applications in less severe forms of gluten intolerance. “We came to realize that trying to treat celiac disease may be a tall order because of the very high sensitivity of celiac patients to these gluten peptides,” she said. “For those patients that are allergic, you may not have to bring it down to these extreme low levels. ...It’s a very large group of people that could benefit from a digestive enzyme or other approaches to reduce gluten.”

High score player

 In 2011, Ingrid Swanson Pultz, a computational protein engineer from the University of Washington’s Institute for Protein Design, advised a group of undergraduate students competing in the International Genetically Engineered Machine (iGEM) synthetic biology competition. The team had to design, build, and test a new biological molecule in just one summer. Pultz was interested in exploring how to apply the tools of protein design to a diverse range of challenges. When several undergraduate students who had friends with celiac disease suggested developing a gluten-degrading enzyme, the project was decided. 

 The team set out to design an enzyme to selectively degrade immunogenic gluten-derived peptides in the stomach. “The real kicker for why it has been so difficult to get naturally occurring enzymes to do this is that it really needs to be specific for gluten,” Pultz said. “If you have something that will act on any protein, and you eat meat protein, dairy protein, or any other protein that we eat, it will distract the enzyme.” Competition from other substrates in a complex meal setting decreases the enzyme’s activity toward gluten, which could allow intact immunogenic peptides to sneak into the small intestine. 

Pultz’s team took the opposite approach of Helmerhorst’s by starting with an enzyme that fared well in the harsh conditions of the stomach and engineering it to specifically cleave gluten-derived peptides. The researchers scoured an enzyme database and identified the kumamolisin-As peptidase from Alicyclobacillus sendaiensis, a bacterium that thrives in acidic soil, as their template. They used a protein structure game called Foldit to analyze and optimize the chemical interactions between the enzyme and a dominant immunogenic gluten-derived peptide. As the researchers mutated specific amino acids in the active site, Foldit calculated a score based on the favorability of the resulting interactions with the peptide. “Eliminating the clashes and creating hydrogen bonds or hydrophobic interactions between the protein and the substrate reduced the total energy,” Pultz said. “It's kind of like a video game because you're trying to get the best score. It's pretty fun.”

When the team tested their high scoring computationally designed enzyme in the lab, they found that it was 116 times more active and 877 times more specific for the immunogenic gluten-derived peptide relative to wild type kumamolisin-As (6). They also scored big at the iGEM competition, taking home the grand prize. 

Pultz wanted to develop the prototype into a commercially viable therapeutic for celiac disease. While the enzyme had been designed to recognize a PQLP sequence in a 33-amino acid peptide, another highly immunogenic 26-amino acid gluten-derived peptide contains the sequence PQQP. By iteratively tweaking the active site using advanced computational tools and expressing and testing the resulting enzyme, Pultz’s team identified a design that balanced maximal activity toward both peptides simultaneously. They also introduced mutations that optimized the enzyme’s overall kinetics with the goal of driving the gluten peptide levels below the threshold for toxicity, which is estimated at 50 milligrams per day, and doing so within a physiologically relevant timeframe (7). 

The researchers combined their remodeled enzyme with whole wheat bread under simulated gastric conditions and observed that it degraded more than 99 percent of the gluten load in 30 minutes, demonstrating its potential to meet the threshold requirement in vivo (8). They also found that the activity of the enzyme toward the gluten component gliadin, the primary source of immunogenic peptides, eliminated its ability to induce a T cell response in a cellular assay (8). 

We can degrade purified gliadin peptides or purified gluten all day long. ...But we needed to make sure that it was working in actual lifelike settings. 
- Ingrid Swanson Pultz, University of Washington 

Pultz and researchers at the pharmaceutical company Takeda tested this enzyme, called TAK-062, in a clinical study  that demonstrated its safety and effective gluten degradation in the human body (9). Healthy participants ingested TAK-062 along with a meal consisting of ice cream, egg whites, whole wheat bread, orange juice, lime juice, vanilla extract, and (in certain high gluten dose experiments) gluten powder, providing a rich mixture of gluten, egg, and dairy proteins. 

When the researchers aspirated and analyzed the participants’ stomach contents, they found that in meals containing one to six grams of gluten, the enzyme digested an average of at least 97 percent of the gluten by 20 to 35 minutes after drug administration. The activity of TAK-062 was not affected when participants were pretreated with proton pump inhibitors, which are commonly used by people with celiac disease and alter the acidity of the stomach. “We can degrade purified gliadin peptides or purified gluten all day long,” Pultz said. “But we needed to make sure that it was working in actual lifelike settings.”

The high degree of gluten degradation observed in the trial suggests that TAK-062 can effectively detoxify the couple hundred milligrams of gluten that are often inadvertently ingested by people striving for a gluten-free diet, according to Dan Leffler, a gastroenterologist and medical director at Takeda. “It gives you confidence that you can have a reliable degree of protection to drop gluten levels low enough that they become nonharmful for people with celiac disease,” he said. 

It’s a common misconception that the option to simply eat gluten-free eliminates the need for a celiac disease therapeutic. “The truth is that the gluten-free diet is both really, really burdensome and inherently impossible in some ways,” Leffler said. 

In a 2020 study in Alimentary Pharmacology & Therapeutics, Leffler and his coauthors reported that gluten exposure is all but inevitable for even the most diligent eaters (10). And in a 2022 survey conducted by Beyond Celiac, only about 12 percent of respondents reported that a gluten-free diet completely treats their condition (11). “The highest need right now in celiac disease is in people who are doing their best on a gluten-free diet, but are still not responding and still having issues,” Leffler said. The researchers are currently evaluating TAK-062 in this patient population using participant symptom surveys, traditional endoscopy and biopsy, and video capsule endoscopy technology, where a camera the size of a pill captures images throughout the small intestine to provide a picture of intestinal damage. 

TAK-062 is among the first computationally designed proteins and the first developed using Foldit to proceed to clinical trials. “It really is somewhat groundbreaking the way that we can now optimize an enzyme’s or a protein’s characteristics to do the job we're looking for,” Leffler said. 

Pultz sees the impact of TAK-062 as advancing both celiac disease treatment and the computational protein field. “I hope that this enzyme is successful because it's my baby and I'd love to see it get out there and fly,” she said. 

Baked in 

Rather than developing a drug to be taken with a meal, other researchers are exploring the possibility of incorporating gluten-degrading enzymes directly into specialized food products for people with celiac disease. While gluten-free foods can be made using naturally gluten-free grains such as rice and millet, scientists have mixed the enzymes into wheat, barley, and rye-based flour, where they can easily access and break down gluten.

Franziska Ersoy, a food chemist at Leibniz University Hannover, explores mushrooms as a natural source of enzymes for such purposes. “When you look at the mushrooms in their natural environment outside in the woods or wherever they're growing, they need a lot of enzymes in order to be able to digest whatever substrate they're growing on,” Ersoy said. “We see what kind of enzymatic activities they might have that might be interesting for food applications.” 

In a recent study in Catalysts  lead by Ersoy, researchers treated wheat, barley, and rye flour samples with a proline-cleaving enzyme they discovered in the Flammulina velutipes  fungus (12,13). The team observed that the enzyme cleaved the gluten-based proteins at more sites than a commercially available proline-cutting dietary supplement, although this result may depend on the experimental conditions used, Ersoy said. She plans to further purify the enzyme and then evaluate its ability to degrade the immunogenic gluten sequences in flour using a T cell assay.    

While breaking down the gluten in flours could theoretically expand the range of baked goods that people with celiac disease can eat, the final product would fall flat. “Gluten is like the glue of a flour,” Ersoy said. Gluten forms a complex protein network that is critical to the dough’s elasticity and the bread’s structure and texture. “People don't really understand what can be done in order to degrade it in advance without affecting the baking properties,” Ersoy said. Additionally, the peptide degradation products may affect the food’s taste since peptides are often used to change a food’s flavor profile. 

Fortunately, other foods are more amenable to treatment with gluten-degrading enzymes. Adding the enzymes to grain used to make wheat beer, for example, would circumvent the challenges inherent to the finicky, gluten-dependent baking process. While Ersoy hasn’t sampled her enzyme-treated flour herself, some proline-containing peptide fragments have a bitter taste (14). Beer, conveniently, already has a bitter flavor. Thanks to the work of these researchers, Oktoberfest could become a lot more celiac disease friendly. 

Sachin Rustgi, a crop biotechnology researcher at Clemson University, developed transgenic wheat lines expressing gluten-degrading enzymes (15). “One of the key drivers is to bring the cost down and then to make it more available to people by just having [the enzyme] baked in,” he said. 

In order for the wheat to be used to create baked goods that rely on gluten to maintain their integrity, the enzymes shouldn’t be active until after the food is ingested. Degrading gluten postingestion also takes care of gluten that may have snuck into the food anywhere along its journey to the mouth. To meet this need, Rustgi selected a combination of sequentially active, digestive system stable enzymes developed for celiac disease therapy. The first is a glutamine-cleaving enzyme featuring an internal peptide inhibitor that is removed when exposed to acidic pH, triggering the enzymatic reaction. The second is a proline-cleaving enzyme that only degrades the short gluten-based peptides generated by the first enzyme, allowing its activity to be indirectly pH-regulated. 

Before these enzymes can break down gluten in the gut, however, they must survive the protein-denaturing temperatures of the baking process. Rustgi’s team wondered if similar enzymes in natural heat lovers might provide some clues into the structural features that impart temperature stability. They searched the literature for peptidases in thermophilic organisms and identified the Pyrococcus  family of bacteria as a source of proline-cleaving enzymes. “We thought we could avoid the melting because that enzyme naturally can survive those high temperatures since the bacteria live in thermal vents,” Rustgi said. 

The researchers analyzed these proline peptidase sequences for conserved amino acids that were likely responsible for their thermal stability. They discovered that the enzymes feature unique residues that join together two components of a barrel-like protein structure. By introducing these mutations into their proline-cleaving enzyme, they developed a version that can withstand temperatures up to 90 °C and retain approximately 60 percent of its gluten-degrading activity at physiological temperatures (16). The team took a similar approach to evolve the glutamine-cleaving counterpart toward thermal stability, incorporating mutations from a thermostable protease in Ervatamia coronaria, the pinwheel flower native to South Asia. They found that this enzyme could survive temperatures up to 60 °C (16).  

Rustgi believes that additional structural characterization and mutagenesis, perhaps by switching out larger segments of amino acids rather than single residues, could further improve the enzymes’ thermostability. Although baking temperatures far exceed 90 °C, not every part of the bread experiences these extreme temperatures, and protein storage bodies in the grain provide some protection against denaturation. This work may also provide a generalizable strategy for making other gluten-degrading enzymes more stable against high heat. 

Rustgi hopes that gluten-detoxified forms of wheat will allow people with celiac disease to enjoy both the dietary freedom and the nutritional value this staple food source provides (17). “By not avoiding wheat, we will not be depriving ourselves of essential food fibers,” he said. 

Beyond the breakdown 

While breaking down immunogenic gluten-derived peptides is a promising and thoroughly explored strategy for treating celiac disease, researchers are also developing therapeutics to block the immune response to gluten downstream. “We are looking at different mechanisms of action for the treatment of celiac disease because nothing is approved, and we don't know what will be potentially effective for different populations,” Leffler said. “I think we will find out eventually in celiac disease that you won't have a one-size-fits-all approach.”

“One could also argue that some of these therapies may work together and that the ultimate outcome of successful treatment for celiac disease might be a combination,” said Tobias Freitag, a gastroenterologist and immunologist at the University of Helsinki. 

In probing the pathway of celiac disease, researchers wondered if the seemingly minor modification of the gluten-derived peptides by tissue transglutaminase 2, which removes an amide group from specific glutamine residues, might actually be critical to the peptides’ immunogenicity. As it turns out, the unmodified peptides are surprisingly harmless. When they are deamidated, however, they develop a negative charge that promotes their presentation to T cells. “By blocking tissue transglutaminase, you're able to keep gluten in its much less inflammatory and much less immunogenic form, and that blocks the whole chain reaction of celiac disease from going forward,” Leffler said. 

The team at Takeda is investigating TAK-227, an oral small molecule that inhibits tissue transglutaminase 2 activity in the small intestine, as a treatment for celiac disease. After confirming that the drug did not produce adverse off-target effects in a healthy population, the researchers administered it to people with celiac disease along with a biscuit containing 3 grams of gluten (18). “This was the first trial that's shown protection against both intestinal damage and symptoms in a gluten challenge setting, so that was really encouraging,” Leffler said. The researchers are further evaluating TAK-227’s efficacy for people struggling to manage their disease with a gluten-free diet. 

Scientists are also investigating whether or not celiac disease might benefit from immune tolerization therapy. “The holy grail of autoimmunity is to have antigen-specific immune tolerance, so you don't suppress the whole immune system, you just retrain the immune system to not attack that thing that you're not supposed to attack,” Leffler said. 

Since researchers have pinpointed gluten antigen-specific T cell activity as the driver of the disease and can feasibly introduce gluten into the body and measure the immune response, celiac disease is an ideal candidate for this approach. The basic premise is that “we can shuttle the gluten proteins to these places in the body that are regulatory more than inflammatory, activate those, and then convince the body that this isn’t a problem anymore and we don’t really need to worry about this protein,” Leffler said. 

Researchers at Takeda are evaluating an immune tolerization therapy for celiac disease called TAK-101. The drug consists of gluten protein shielded within a polymer-based nanoparticle, which is injected into the bloodstream and delivered to the spleen and liver (19). There, “it targets these special parts of the immune system in the spleen and liver that are responsible for tolerance and telling the body to not attack,” Leffler said. 

The researchers propose that the nanoparticle is taken up by macrophages and other antigen-presenting cells in these organs, which chop up the gluten protein and display the fragments on the cell surface. Presenting the antigen in this way regulates gluten-specific T cell activity to rewire the immune response to gluten. 

In a recent clinical trial, people with celiac disease received either TAK-101 or a saline placebo intravenously and then ingested gluten (20). The researchers collected blood samples from the participants, exposed the white blood cells to gluten-derived peptides, and measured the amount of gluten-responsive T cells via their production of cytokines. People with well-controlled celiac disease should have very few of these T cells in their blood, the same way someone without an active flu infection wouldn’t have circulating influenza-fighting cells. The researchers found that TAK-101 significantly reduced the number of gluten-responsive T cells in the blood following gluten ingestion relative to the placebo. “We could prevent these gluten-specific T cell responses with gluten exposure in people who've been treated with TAK-101 without suppressing the rest of the immune system, so that was really exciting,” Leffler said. 

This one really pushes the science and has the potential to be disease modifying in a way that almost nothing else in the celiac space right now is. 
- Dan Leffler, Takeda

The trial was the first to demonstrate that autoimmunity’s holy grail may be within reach. “This one really pushes the science and has the potential to be disease modifying in a way that almost nothing else in the celiac space right now is,” Leffler said. In future studies, the researchers plan to explore what a drug that targets underlying disease pathology means for patients and their dietary restrictions. “Getting better data on what TAK-101 does to symptoms with gluten exposure over the long term is a key thing we're going to need to study in the near future,” Leffler said. “Another big question is how long does the effect of this last, and when do we need to re-dose, and how often?” 

It’s possible that the effect could persist indefinitely, potentially providing a cure for celiac disease, according to Freitag, who helped develop TAK-101. “Our understanding of what triggers the disease in the first place is that there is one event that changes the interpretation of the antigen by the immune system,” he said. “It’s feasible to consider that the [opposite] could actually happen with a therapy that delivers the antigen in the appropriate fashion to tolerogenic cell populations. You could have that as the rebooting event for the immune system.” 

For Leffler, the motivation to pursue every possible path toward treating celiac disease comes from witnessing a growing dissatisfaction with lifestyle-based management. “Over the years, you continue to see patients saying is this all I have? Is there nothing else you can do for me?” he said. “People with celiac disease are still getting severe complications and missing work and missing school and all these things due to poorly controlled disease. To me, I'm passionate about it because I'd love to see a world where we have options for people to help improve disease control and quality of life.” 

References

1. Lindfors, K. et al. Coeliac disease. Nat Rev Dis Primers  5, 3 (2019). 
2. Helmerhorst, E.J., Zamakhchari, M., Schuppan, D., & Oppenheim, F.G. Discovery of a novel and rich source of gluten-degrading microbial enzymes in the oral cavity. PLoS ONE  5, e13264 (2010). 
3. Zamakhchari, M. et al. Identification of Rothia bacteria as gluten-degrading natural colonizers of the upper gastro-intestinal tract. PloS ONE  6, e24455 (2011). 
4. Darwish, G., Helmerhorst, E.J., Schuppan, D., Oppenheim, F.G., & Wei, G. Pharmaceutically modified subtilisins withstand acidic conditions and effectively degrade gluten in vivo. Sci Rep  9, 7505 (2019). 
5. Wei, G., Darwish, G., Oppenheim, F.G., Schuppan, D., & Helmerhorst, E.J. Commensal bacterium Rothia aeria degrades and detoxifies gluten via a highly effective subtilisin enzyme. Nutrients  12, 3724 (2020). 
6. Gordon, S.R. et al. Computational design of an α-gliadin peptidase. J Am Chem Soc  134, 20513-20520 (2012). 
7. Catassi, C. et al. A prospective, double-blind, placebo-controlled trial to establish a safe gluten threshold for patients with celiac disease. Am J Clin Nutr  85, 160-166 (2007). 
8. Wolf, C. et al. Engineering of Kuma030: a gliadin peptidase that rapidly degrades immunogenic gliadin peptides in gastric conditions. J Am Chem Soc  137, 13106-13113 (2015).
9. Pultz, I.S. et al. Gluten degradation, pharmacokinetics, safety, and tolerability of TAK-062, an engineered enzyme to treat celiac disease. Gastroenterology  161, 81-93 (2021). 
10. Silvester, J.A. et al. Exposure sources, amounts and time course of gluten ingestion and excretion in patients with coeliac disease on a gluten-free diet. AP&T  52, 1469-1479 (2020). 
11. Beyond Celiac. Your life with celiac disease: 2022 community survey results. At < https://www.beyondceliac.org/celiac-news/2022-community-survey/>.
12. Schulz, K., Giesler, L., Linke, D., & Berger, R.G. A prolyl endopeptidase from Flammulina velutipes for the possible degradation of celiac disease provoking toxic peptides in cereal proteins. Process Biochem  73, 47-55 (2018). 
13. Ersoy, F. et al. A Prolyl Endopeptidase from Flammulina velutipes Degrades Celiac Disease-Inducing Peptides in Grain Flour Samples. Catalysts  13, 158 (2023). 
14. Ishibashi, N. et al. Taste of proline-containing peptides. Agric Biol Chem  52, 95-98 (1988). 
15. Osorio, C.E. et al. Development of wheat genotypes expressing a glutamine-specific endoprotease from barley and a prolyl endopeptidase from Flavobacterium meningosepticum or Pyrococcus furiosus as a potential remedy to celiac disease. Funct Integr Genomics  19, 123-136 (2019). 
16. Osorio, C.E. et al. Directed-mutagenesis of Flavobacterium meningosepticum prolyl-oligopeptidase and a glutamine-specific endopeptidase from barley. Front Nutr  7, 11 (2020). 
17. Rustgi, S., Alam, T., Jones, Z.T., Brar, A.K., & Kashyap, S. Reduced-immunogenicity wheat and peanut lines for people with foodborne disorders. Chem Proc  10, 67 (2022). 
18. Schuppan, D., et al. A randomized trial of a transglutaminase 2 inhibitor for celiac disease. N Engl J Med  385, 35-45 (2021). 
19. Freitag, T.L. et al. Gliadin nanoparticles induce tolerance to gliadin in mouse models of celiac disease. Gastroenterology  158, 1667-1681 (2020). 
20. Kelly, C.P. et al. TAK-101 nanoparticles induce gluten-specific tolerance in celiac disease: a randomized, double-blind, placebo-controlled study. Gastroenterology  161, 66-80 (2021).