As a first step in determining whether rare coding variants in NHE9 contributed to the autism phenotype, we constructed a structural model of the membrane domain of NHE9, and its yeast ortholog Nhx1, based on the crystal structure of a distantly related bacterial ortholog, NhaA35. Previously, we used evolutionary analysis and a composite fold-recognition approach to propose a three-dimensional model-structure of NHE1, a prototype of the cation/proton superfamily36,37. Utilizing a similar methodology, we aligned yeast Nhx1 and mammalian NHE9 to NhaA (Figure 1A), as well as to NHE1. In accordance with phylogenetic clustering, the resulting alignments showed that both Nhx1 and NHE9 were significantly more closely related to NHE1 than to bacterial NhaA, with sequence identities of 30% and 32% for the alignments of Nhx1 and NHE9 to NHE1, respectively, whereas aligning Nhx1 and NHE9 to NhaA resulted in sequence identities of 15% and 14%, respectively. This allowed us to extend the structural model from NHE1 to NHE9 (Figure 1B and Figure 2A) and Nhx1 (Figure 1C). To evaluate the Nhx1 and NHE9 models, we examined characteristic traits of membrane
