Some people are overtly allergic to wheat (from here on, “wheat” will cover all gluten-containing grains). Minutes to hours after exposure, these individuals develop symptoms such as rashes, headaches, diarrhea, or shortness of breath—a well-known example is baker’s asthma. This wheat allergy (Inomata, 2009) engages the part of our immune system that responds quickly against parasitic worms, fungi, and microorganisms. In some of us, however, gluten triggers immune-mediated reactions whose symptoms develop gradually, weeks to years after its introduction in the diet.

In about 1 person out of 100 this hypersensitivity is expressed as celiac disease, defined as a chronic immune reaction against one’s own small intestine. Over time, this reaction flattens the intestinal wall (which is normally covered with millions of finger-like protrusions), reducing its surface and thus its ability to absorb important nutrients for both body and brain. If gluten is not withdrawn during childhood, the growth of some cranial bones is altered as well. As a result, over 80% of adult celiacs have unusual facial proportions (Zanchi et al., 2013). Quite typical is an especially high forehead relative either to the middle third of the face or to the forehead of healthy people (see, respectively, Finizio et al., 2005; and Zanchi et al., 2013).

Most people with celiac disease do not know they have it. In a sample of over 5,000 Italian students, for example, the ratio of diagnosed to undiagnosed cases was 1 to 6 (Catassi et al., 1995). In the elderly, celiac disease often goes unrecognized as well, with a mean delay of 17 years from the onset of symptoms to diagnosis (Gasbarrini et al., 2001). Alarmingly, blood markers of the disease have quadrupled in the United States in the last 50 years (Rubio-Tapia et al., 2009) and doubled in Finland in the last 20 (Lohi et al., 2007). Measurements were taken all at once on blood samples collected and frozen decades apart, hence the recent surge in the disease cannot be due to better detection or more lenient diagnostic criteria. Markers also increase within the same group of individuals over time, showing that an abnormal immune response to gluten can suddenly emerge in adulthood (Catassi et al., 2010).

Some people do better on a gluten-free diet and worsen upon a gluten challenge (even under double-blind, randomized, placebo-controlled conditions: Biesiekierski et al., 2011) although they do not meet the criteria of either wheat allergy or celiac disease. This non-celiac gluten sensitivity is diagnosed by exclusion, because there are currently no laboratory tests for it. The gut permeability of these people is normal, unlike that of celiacs—but gluten makes it soar just as much as that of celiacs (Hollon et al., 2015). Symptoms emerge hours to days after gluten exposure and are largely extraintestinal; they include headache and eczema but also fatigue and “foggy mind” (Sapone et al., 2012). Other individuals report being sensitive to gluten but actually experience bloating and abdominal pain from wheat’s carbohydrates (Biesiekierski et al., 2013). Many studies on non-celiac gluten sensitivity have not controlled for the presence of these carbohydrates; they can also be found in various vegetables, however, and whether their effects can go beyond mere intestinal discomfort is debatable (for opposing views, see Fasano et al., 2015; De Giorgio et al., 2016).

Over 95% of celiacs carry a specific variant of a gene that is corresponsible for the regulation of the immune system, and about 5% carry another (Diosdado et al., 2005). Crucially, both genes are implicated in the ability of the immune system to distinguish self from non-self. These genes are also present in 30–40% of the general population, however, and of course not all of them develop celiac disease; even monozygotic twins on the same diet can be discordant for it (Greco et al., 2002). Other factors must thus be involved—possibly, simple environmental triggers. These have been shown to range from delivering a baby (Malnick et al., 1998) to contracting a virus or a parasite. In one study, for example, nearly 90% of celiacs, vs. 17% of controls, showed evidence of previous infection with adenovirus (Kagnoff et al., 1987). Because a protein coded by this virus is structurally similar to gluten, it is plausible that in genetically predisposed individuals the initial reaction to the virus may extend to gluten and then to some proteins in our own intestine that resemble both—a process called molecular mimicry (see Kasarda, 1997).

Unfortunately, gluten resembles some brain-relevant substances too. In vitro, antibodies against gluten removed from human blood attack cerebellar proteins and components of the myelin sheath that insulates nerves (Vojdani et al., 2014). They also attack an enzyme involved in the production of GABA—our prime inhibitory neurotransmitter, whose dysregulation is implicated in both anxiety and depression. In the blood of blood donors, antibodies against wheat or milk and antibodies against these brain-relevant substances have been found to be simultaneously elevated, consistent with the presence of a cross-reaction (Vojdani et al., 2014). Most of us escape it only because our gut and blood-brain barriers are intact—and only as long as they remain so. Antibodies against the brain, triggered by gluten, can cause severe neurological dysfunctions whether or not one is celiac (Hadjivassiliou et al., 2010). Similar antibodies have also been found in the blood of a subgroup of schizophrenia patients; some of them carried blood markers of celiac disease, but others did not (Cascella et al., 2013).

If wheat can affect the brain, it should come as no surprise that it can affect mental health too (for a review, see Jackson et al., 2012a). Exceptionally large epidemiological studies, each involving many thousands of patients, have found that celiac disease is associated with an increased risk of depression (Ludvigsson et al., 2007b) and psychosis (Ludvigsson et al., 2007a). Among individuals with a normal intestinal wall, those who carry blood markers of celiac disease are three times more likely to develop autism in the future, and five times more likely to have already been diagnosed with it (Ludvigsson et al., 2013).

Antibodies against gluten have been found much more often in schizophrenia and autism patients than in the general population or in controls, a result that has been replicated repeatedly (Jackson et al., 2012a). Some figures are stunning, such as a reported presence of antibodies against gluten in 87% of unmedicated autistic children vs. 1% of normal children (Cade et al., 2000).

The main gene that predisposes to celiac disease also changes the composition of the microbes in the gut; a notable finding, because we now know that these microbes (collectively known as gut microbiota) are directly capable of shaping our behavior (Dinan et al., 2015; Kramer and Bressan, 2015). Carriers and non-carriers of the gene produce stools with significant bacterial differences at 1 month of age already (Olivares et al., 2015). Among other things, carriers host more clostridia; clostridia tend to be overrepresented in the guts of children with autism (Louis, 2012), and it is suggestive to associate these findings to the epidemiological evidence, discussed earlier, of a larger risk of autism in celiacs.

Gut microbes even appear to play a part in when (and possibly whether!) carriers do develop celiac disease. Because the maturation of our immune system is co-driven by our microbial community (Kranich et al., 2011), it is crucial that the latter develops normally—which could be jeopardized by feeding babies inappropriate foods at an inappropriate time. The microbiota matures enormously in the first 12 months of life, hence it might be important to avoid gluten during this period (Fasano, 2009). Indeed, a double-blind study on young carriers of the celiac gene compared the relevance of early (6 months of age) vs. late (12 months) introduction of gluten in their diets. Early introduction promptly caused loss of tolerance to gluten and set off the development of autoimmunity, arguably via a change in the composition of the still immature microbiota (Sellitto et al., 2012). Indeed, whether or not transgenic mice with the celiac gene in them express the disease has recently been shown to be entirely determined by their guts. Eating gluten started celiac disease in the mice who had been raised without gut microbes, or whose microbiota included pathogens or had been perturbed by antibiotics right after birth—but not in the mice whose microbiota was healthy (Galipeau et al., 2015).

Changes in gut microbiota due to a sudden, massive exposure to wheat products have also been hypothesized to mediate the well-known relationship between immigrant status and schizophrenia (Severance et al., 2014). This might be, for example, the case of people moving to Europe from sub-Saharan Africa, where staple grains do not include wheat and are traditionally broken down via fermentation before being eaten. It is thus entirely possible that bread can be harmful to our mental health not only directly, via some of the proteins it contains; but also indirectly, via its effects on our gut microbes. The causal relationship between eating bread and harboring certain microbes could actually go both ways, as suggested by recent evidence that our craving for certain foods may be driven by the intestinal bacteria that feed on them. Bread is ultimately broken down to sugar, and plenty of microbes thrive on sugar. When not enough of it comes their way, microbes might be able to manipulate their host by inducing bad mood and other painful sensations—relieved only by eating the right stuff (Alcock et al., 2014).

Like morphine, exorphins bind to opioid receptors that are widely distributed throughout the body—in places as different as the gut, the lungs, the genitals, the various districts of the nervous system. Such receptors are of course meant for our own opioids, endorphins (of internal origin, “endo”). Our body may produce endorphins to reduce pain when we need to continue functioning in spite of injury or stress, as during labor or combat. The “runner’s high”, the state of euphoria experienced by long-distance runners, might capitalize on this mechanism (Boecker et al., 2008; but see also Fuss et al., 2015).

It has been intriguingly suggested that one major function of endorphins would be to protect the organism against starvation in times of stressful, prolonged food scarcity (Margules, 1979, 1988). We know that the same opioid can exert opposite effects depending on which receptor it binds to (Teschemacher, 2003); the key might be whether the receptor happens to sit in the body or in the brain (Margules, 1988). Bound to opioid receptors in the gut, in fact, morphine-like endorphins tend to conserve bodily resources (by inducing constipation and water retention), reduce motor activity, decrease pain, suppress both reproduction hormones and sexual desire. Bound to opioid receptors in the brain, on the contrary, they promote energy expenditure, boosting reactivity and (hyper)activity. The former could be interpreted as a passive, hibernation-like response to seasonal food shortage; the latter as an active, migration-inducing one (Margules, 1988; see also Guisinger, 2003). The potential connection between a malfunctioning opioid system and eating disorders such as anorexia has not gone unnoticed—and it is supported by several lines of evidence (see Yeomans and Gray, 2002). Notably, endorphins are produced on demand, but exorphins are generated at virtually every (modern) meal.

Food exorphins appear to perform their job largely or entirely from the gut. Thus, they should support energy sparing in all kinds of manners (see evidence for some of these in Teschemacher, 2003). Yet exorphins directly bind to the opioid receptors of the brain as well, if they can get there (Kostyra et al., 2004). The question is whether they pass through the intestinal and blood-brain barriers in meaningful amounts. Some authors argue that, if these barriers are healthy, they probably do not (Miner-Williams et al., 2014). This is hardly reassuring, though, given how easily the function of even healthy barriers can be disrupted—be it by stress (Söderholm and Perdue, 2001), dietary components (Ulluwishewa et al., 2011), alcohol (Purohit et al., 2008), or familiar over-the-counter drugs (e.g., Smale and Bjarnason, 2003). Indeed, radioactively labeled gluten proteins fed to rats by stomach tube are later found in the animals’ brains in the form of exorphins (Hemmings, 1978; for related evidence regarding dairy proteins, see Sun and Cade, 1999).

The manufacture of exorphins is incredibly efficient. The nutritionally insignificant intake of 1 g of casein (about two tablespoons of cow milk), for example, produces opioids in large enough amounts to exert physiological effects (Meisel and FitzGerald, 2000). This is remarkable in view of the facts that (a) the opioids from gluten are stronger than those from casein (Zioudrou et al., 1979), and (b) the daily average consumption of gluten in Europe is 10–20 g, with many people exceeding 50 g (Sapone et al., 2012). In the brain of rats, the opioids from casein have been shown to be 10 times more potent than morphine (Herrera-Marschitz et al., 1989). If all exorphins released in the gut made it to the brain, it is hard to see how we could keep functioning.

Opioids are involved in both the palatability and rewarding aspects of food, hence they play a major role in food cravings and food addiction (for a review, see Yeomans and Gray, 2002). The opioid antagonist naloxone drastically reduces the intake of preferred foods, but not of nonpreferred ones, in rats (Glass et al., 1996; see also Boggiano et al., 2005). Naltrexone, which is much like naloxone but it lasts longer and can be taken by mouth, suppresses binge eating in humans (Marrazzi et al., 1995). Indeed, people who have first ingested naltrexone (against placebo: Yeomans and Gray, 1997) rate a bowl of pasta as less pleasant, and eat less of it. Tellingly, naloxone is famed for its ability to counteract the effects of an overdose of heroin, a potent derivative of morphine, and naltrexone is used in the treatment of heroin dependence.

Foods that contain exorphins, such as wheat and dairy products, have indeed a reputation for being rewarding and people find it extremely hard to give them up. The addictive properties of milk were arguably designed by evolution to gratify suckling young. The gut of newborns is highly permeable—not only to the mother’s antibodies as an aid to their still immature immune system, but also to milk opioids (see Teschemacher, 2003). Yet, production of the enzyme for properly digesting milk is genetically programmed to stop after weaning. Regular intake of milk by adults is evolutionarily novel and only started with animal domestication; it was permitted by a mutation of this enzyme in populations that kept cattle. Interestingly and perhaps worryingly, the opioids in bovine milk are 10 times stronger than those in human milk (Herrera-Marschitz et al., 1989). This might not be extraneous to the fact that about half of children up to 4 years of age need their milk bottle to fall asleep at night (in Thailand: Sawasdivorn et al., 2008). Note that, as mentioned, the opioids in wheat are even stronger than those in bovine milk (Zioudrou et al., 1979).

Arguably, foodstuffs whose digestion releases exorphins are preferred exactly because of their drug-like properties. It has been speculated, in fact, that this chemical reward might have been one incentive for the initial adoption of agriculture (Wadley and Martin, 1993). Why cereals rapidly and extensively replaced traditional foods even though they were less nutritious and required more labor has been widely regarded as a puzzle. Also, cultivation of cereals continued even when the abundance of more easily processed foodstuffs—such as meat, tubers, and fruit—rendered it unnecessary (see Murphy, 2007). A clue could be the fact that all major civilizations, in every inhabited continent, arose in groups that practiced cereal agriculture and not in groups that only cultivated tubers and vegetables or had no agriculture at all. According to Wadley and Martin’s rather audacious hypothesis, daily opioid self-administration could have increased people’s tolerance of crowded sedentary conditions, of regular work, of subjugation by rulers. If so, cereals might have ultimately helped the development of civilization.

Not all individuals handle these substances the same way. For example, abnormally high levels of milk and/or wheat exorphins have been found in the urine (Hole et al., 1979) and blood (Drysdale et al., 1982) of schizophrenia patients and in the urine (e.g., Sokolov et al., 2014; but see Cass et al., 2008) of autistic children. When purified and injected in the brain of rats, these substances made the rats behave in strikingly odd ways—very restless at first and then inactive and hyperdefensive. Among other things, the rats paid no attention to a ringing bell, in suggestive similarity to the apparent deafness often observed in children with autism (Sun and Cade, 1999; Cade et al., 2000). Interestingly for the nonpatients among us, exorphins coming from healthy people’s blood had on rats effects that were weaker and briefer but otherwise similar (Drysdale et al., 1982).

Besides producing behavioral disorders similar to those seen in schizophrenia and autism (such as decreased social interaction, reduced pain sensitivity, uncontrolled motor activity: Sun and Cade, 1999), exorphins activate in rats the same brain regions that are affected in schizophrenia and autism. The disruptive effects they exert on visual and auditory areas are consistent with typical malfunctionings such as hallucinations in schizophrenia (Sun et al., 1999). So perhaps it is no coincidence that a recent case report of an adult patient has described complete resolution of highly unsettling visual and auditory hallucinations, experienced daily from early childhood, upon removal of gluten from the diet (Genuis and Lobo, 2014).

The effects of food exorphins on behavior (for a comprehensive review, see Lister et al., 2015) and on the brain (Sun et al., 1999) are reversed by treating the rats with opioid antagonists. Naloxone has also been shown to temporarily erase psychotic symptoms, especially hallucinations, in schizophrenia patients (Emrich et al., 1977; Jørgensen and Cappelen, 1982). Naltrexone benefits some children with autism (Roy et al., 2015), arguably by blocking a brain opioid activity that might be abnormally high in these children (Sahley and Panksepp, 1987). Attempts to eliminate excess exorphins from the blood of schizophrenia patients via weekly dialysis for 1 year have led to remarkable results as well, with 40% of patients vastly improving or fully recovering from schizophrenia. In some patients who did not get better, the continuous production and absorption of exorphins on a regular diet might have been so large that dialysis failed to reduce their concentration in the blood. Indeed, of the five patients that combined dialysis with a diet devoid of gluten and casein, all either improved significantly or became entirely normal (Cade et al., 2000).

In psychotic children (Gillberg et al., 1985), schizophrenia patients (Lindström et al., 1986), and women with postpartum psychosis (Lindström et al., 1984), larger-than-normal amounts of exorphins have been detected in the cerebrospinal fluid. Exorphins clearly do not belong there. In the presence of faulty barriers, though, they could migrate from the gut to the blood (prompting an immune reaction) and from there to the cerebrospinal fluid. In people with schizophrenia (unlike in healthy individuals) the more antibodies against gluten one finds in the blood, the more one finds in the cerebrospinal fluid (Severance et al., 2015). This correlation suggests a larger diffusion of antibodies from one place to the other in patients than in nonpatients, pointing to some barrier dysregulation—possibly subtle or transient. It is worth recalling that gluten comes prepackaged with the ability to cause such dysregulation itself.