Chronic diarrhea is the primary cause of morbidity in colonies of captive non-human primates 
. A number of infectious pathogens have been identified that can induce this condition 
, but the role of dietary antigens had not been previously investigated. We hypothesized that captive rhesus macaques exhibiting clinical diarrhea of non-infectious origin may be sensitive to gluten, a major source of protein in their formulated diet. A subpopulation of these animals presented with chronic diarrhea, stomach distention and blistering skin rashes, clinical symptoms that are also observed in classic human celiac disease, and in its dermatologic manifestation, dermatitis herpetiformis. In this study we characterized the clinical, histological, and serological characteristics of a small number of gluten-sensitive macaques. We hope to continue with similar, more extensive studies in the future.
Juvenile macaques appear to be especially prone to reacting adversely to dietary gluten. The majority of animals with chronic diarrhea of non-infectious origin had elevated levels of anti-gliadin IgG and IgA in their plasma. A number of these seropositive macaques exhibited marked villus blunting, crypt hyperplasia, and intraepithelial lymphocytosis in duodenal biopsies. Their clinical, histological and serological markers resolved upon administration of a gluten-free diet, and returned upon reintroduction of dietary gluten.
Previously proposed animal models for gluten sensitivity include non-human primates, gluten-fed rabbits, Irish setter dogs, and transgenic mice (). Prior documentation of celiac disease-like enteropathy in non-human primates is limited to two case reports, one in a single rhesus macaque necropsy 
, and another in a single cynomolgus monkey which improved on a gluten-free diet 
. Elevated anti-gliadin IgG are observed in a majority of laboratory rabbits fed gluten as part of their standard diet 
(MTB, unpublished results), but anti-gliadin IgA are not present (MTB, unpublished results) and these rabbits are apparently asymptomatic. Irish setter dogs with gluten-sensitive enteropathy are the best-studied natural animal model for gluten sensitivity, exhibiting both gluten-dependent diarrhea and histological lesions 
. However, the absence of anti-gliadin and anti-TG2 antibodies as well as the lack of MHC class II linkage with disease in these animals precludes their use as a model for celiac disease 
. Finally, transgenic mouse models have been engineered to mimic celiac disease, most notably the NOD Ab° DQ8+
mouse, which expresses human DQ8 in an endogenous MHC class II-deficient (Ab°), autoimmune-prone (NOD) background 
. These mice develop skin rashes with subcutaneous IgA deposits that are reminiscent of dermatitis herpetiformis, but have no gastrointestinal lesions or GI-related symptoms. Thus, rhesus macaques may represent a more complete model for studying celiac disease pathology than prior animal models.
Comparison of celiac disease with proposed animal models for gluten sensitivity
Two important issues must be addressed before the implications of this animal model can be fully appreciated. First, a Mamu class II association with simian gluten sensitivity must be identified, in analogy to the strong association of MHC class II haplotypes DQ2/DQ8 with human celiac disease. Testing for such a genetic association is currently underway. Preliminary results show that the two gluten-sensitive macaques studied extensively herein and in the accompanying report 
, FH09 and FH45, are of genotype DRB1*0303(12), DRB*1007 at the Mamu class II DRB(1) locus. Second, a humoral immune response directed against TG2 is enacted during active enteropathy in celiac disease patients, but is not observed in gluten-sensitive macaques. This may be because we used recombinant human TG2 as antigen in our ELISA to capture plasma antibodies raised against rhesus macaque TG2. Human and macaque TG2 share 95% protein sequence identity, but may present different epitope binding surfaces in three-dimensional space, attenuating antibody affinity in our assay. Another reason for this discrepancy might be that in macaques TG2-mediated deamidation of gluten epitopes is not required for their presentation by gluten-binding Mamu class II alleles. Finally, if the pathogenesis of macaque and human gluten-sensitivity are substantially shared, it may reflect the possibility that anti-TG2 antibodies do not have a significant pathogenic role in celiac disease. Regardless of the answers to the above unresolved issues, gluten-sensitive rhesus macaques are likely to be valuable for studying the pathogenesis and treatment of human gluten sensitivity.