Structural and Deep Learning Insights into Na+/I- Symporter (NIS)

The thyroid hormones, T3 and T4, play a vital role in both pre- and postnatal development, as well as in the ongoing process of intermediary metabolism in nearly all tissues throughout our lives. Iodide (I−), an essential component of these hormones, is actively transported into the thyroid gland via a specific transporter called the Na+/I- symporter (NIS). This transporter moves I- along with two Na+ ions across the basolateral membrane of thyroid cells, using the sodium gradient as its energy source. NIS also transports oxyanions (XO4-s), like perrhenate (ReO4-) and the environmental pollutant perchlorate (ClO4-). We determined the cryo-electron microscopy structures of this important plasma membrane glycoprotein [1]. The structures of Apo-NIS and its complexes with I- and with ReO4- along with Na+ showed the 13 transmembrane segments of NIS in a similar occluded state, ready to open toward the inside of the cell. These structures revealed that iodide and the oxyanion bound to a positively charged hydrophobic central cavity, with two and one sodium atoms, respectively. To deepen our understanding, we also determined the cryo-EM structures of mutants of the key glutamine 72 residue (Q72A) and a double mutant L253P/V254F—(PF)-NIS [2]. The latter mutant exhibits features relevant to thyroid cancer treatments, as it cotransports oxyanions exclusively. Although these structures hold a wealth of information, they (apo, holo, and mutants) were all captured in the same conformation, likely locked by the vitrification conditions. This has limited our understanding of the NIS transport cycle, especially since we lack structural data of the outwardly open conformation, which is crucial for its function. To bridge this gap and better understand the conformational changes NIS undergoes during transport, we performed a Molecular Dynamics simulation of the symporter with substrates bound. This simulation was performed in a realistic system that includes an asymmetric lipid membrane mimicking mammalian lipids, a double membrane system to establish a ionic gradient as observed in vivo, and a membrane potential. Moreover, we developed an easy-to-follow kinetic model based on microseconds-long simulations analyzed using a machine-learning approach, which helped us visualize the different stages of the transport cycle.