Humanization of autoantigen Wataru Nishie 1,3 , Daisuke Sawamura 1,3 , Maki Goto 1 , Kei Ito 1 , Akihiko Shibaki 1 , James R McMillan 1 , Kaori Sakai 1 , Hideki Nakamura 1 , Edit Olasz 2 , Kim B Yancey 2 , Masashi Akiyama 1 & Hiroshi Shimizu 1 Transmissibility of characteristic lesions to experimental animals may help us understand the pathomechanism of human autoimmune disease. Here we show that human autoimmune disease can be reproduced using genetically engineered model mice. Bullous pemphigoid (BP) is the most common serious autoimmune blistering skin disease, with a considerable body of indirect evidence indicating that the underlying autoantigen is collagen XVII (COL17). Passive transfer of human BP autoantibodies into mice does not induce skin lesions, probably because of differences between humans and mice in the amino acid sequence of the COL17 pathogenic epitope. We injected human BP autoantibody into Col17-knockout mice rescued by the human ortholog. This resulted in BP-like skin lesions and a human disease phenotype. Humanization of autoantigens is a new approach to the study of human autoimmune diseases. Evolution of a complex immune system in mammals can paradoxically lead to an increased likelihood of autoimmune disease, which affects approximately 5% of the general population 1 . The immune system comprises two evolutionarily different responses: innate immunity and adaptive immunity. Vertebrates are capable of both adaptive and innate immunity, which is common to all metazoans, and are consequently exposed to autoimmune diseases—in which an aberrant, adaptive immune system recognizes a self-component as an autoantigen. To understand the pathogenesis of these diseases and to develop new therapies, animal models corresponding to human autoimmune diseases are essential 2,3 . A small number of experimental animal models of autoimmune diseases have been generated by the passive transfer of autoantibodies from individuals 4,5 , but this approach has not been universally successful because of limited interspecies recognition of the autoantigen by the adaptive immune system. In reality, most animal models that reflect human autoimmune diseases have been identified from spontaneously arising diseases or have been generated by repeated immunization using host candidate autoantigens 2,3 . Although these animal models may develop certain aspects of human autoimmune diseases, there is as yet no reliable method to produce models that identically and faithfully reproduce human autoimmune disorders. Recently, transgenic mice expressing disease- associated human HLA alleles and T-cell receptors have been gener- ated, which have the potential to provide further insight into the pathogenesis of many diseases 2,3,6,7 . Conversely, few advances have been made in humanizing autoantigens in animals. Our study therefore focused on this approach in BP, the most common antibody-mediated autoimmune blistering skin disease. Although the sera of individuals with bulbus pemphigoid (BP) are known to contain autoantibodies to the collagen XVII (COL17) autoantigen, the pathogenicity of COL17 autoantibodies has not been unequivocally proven as previous passive transfer experiments have failed 8,9 . In the native COL17 epidermal protein, the pathogenic epitope is restricted to the noncollagenous 16th-A (NC16A) domain 10 , which shows distinct diversity among different species 8,9 . Here, to assess the direct pathogenicity of human BP autoantibodies in a mouse model, we humanized the mouse BP autoantigen. Further- more, we used this new mouse to demonstrate that COL17 decoy peptides block the pathogenic activity of BP-IgG in vivo. RESULTS Generation of COL17-humanized (COL17 m–/–,h+ ) mice We first generated Col17-knockout (COL17 m–/– ) mice, in which the phenotypic features closely resembled the human disease non-Herlitz epidermolysis bullosa (OMIM: 226650) caused by null mutations in the COL17A1 gene (Fig. 1 and refs. 11–13). When COL17 m–/– mice were born, blisters and erosions at sites of trauma were easily created by applying mild friction, and some pups exhibited spontaneous blister formation on their paws (Fig. 1c). Other characteristic findings seen in adult COL17 m–/– mice were genital erosions, hemorrhagic blisters around the digits, and diffuse, nonpigmented hair growth associated with hair loss (Fig. 1d,e). COL17 m–/– mice showed growth retardation compared with wild-type littermates, and most COL17 m–/– mice died within 2 weeks of birth. Mortality rates of the COL17 m–/– and wild-type littermates at 8 weeks were 80.1% and 4.0%, respectively (Supplementary Fig. 1 online). COL17 m–/– mice skin showed subepidermal blistering (Fig. 1f ) associated with a lack of collagen XVII immunostaining (Fig. 1g), and ultrastructurally showed small and poorly formed hemidesmosomes (Fig. 1h) lacking promi- nent inner (IP) and outer (OP) plaques and keratin filament insertion compared to controls. Next, we rescued COL17 m–/– mice by mating them with C57BL/ 6Ncr mice expressing human COL17 under the control of a human keratin 14 promoter (COL17 h+ ) (data not shown). After crossing heterozygote COL17 m–/+ mice and human COL17 transgenic Received 14 August 2006; accepted 26 September 2006; published online 25 February 2007; doi:10.1038/nm1496 1 Department of Dermatology, Hokkaido University Graduate School of Medicine, Sapporo 060-8638, Japan. 2 Department of Dermatology, Medical College of Wisconsin, 8701 Watertown Plank Road, Milwaukee, Wisconsin 53226, USA. 3 These authors contributed equally to this work. Correspondence should be addressed to H.S. (shimizu@med.hokudai.ac.jp). 378 VOLUME 13 [ NUMBER 3 [ MARCH 2007 NATURE MEDICINE TECHNICAL REPORTS © 2007 Nature Publishing Group http://www.nature.com/naturemedicine