Abstracts

Rationale: Germline genetic variants in SLC35A2, which encodes a Golgi-localized UDP-galactose transporter (UDPGalT) essential for cellular galactosylation, have been implicated in one type of congenital disorder of glycosylation associated with intractable seizures and a rare X-linked developmental and epileptic encephalopathy (Kodera H, et al. Hum Mut 2013;34:1708-14,2; Ng BG, et al. Am J Hum Genet 2013;92:632-6). Recently, we have identified post-zygotically-acquired, de novo, loss-of-function SLC35A2 variants in the brain tissue of 17% of radiographically non-lesional focal epilepsy cases (Winawer MR, et al. Ann Neurol 2018;83:1133-46). Despite the clear link to intractable epilepsy, the mechanisms of SLC35A2 underlying epileptogenesis remain elusive. This study seeks to establish a human induced pluripotent stem cell (hiPSC) model to study how SLC35A2 variants lead to epilepsy and to investigate the potential for galactose treatment to reverse the phenotype. Methods: hiPSCs from a healthy male (713-5, isogenic control) were CRISPR-edited to harbor either a pathogenic missense variant (c.910T>C, p.Ser304Pro; SLC35A2-S304P) or a frameshift indel (SLC35A2-KO). These hiPSC lines (713-5, SLC35A2-S304P, SLC35A2-KO) were differentiated to cortical excitatory neurons using dual SMAD inhibition protocols (Chambers SM, et al. Nat Biotechnol 2009;27:275-80) and plated on mouse astrocytes for analysis of neural network activity via multi-electrode array (MEA). Expression and co-localization analyses were evaluated using immunofluorescent (IF) staining. Cellular glycosylation was evaluated on immunoblot using MALI, a lectin that binds to terminal sialic acid residues on glycoproteins. Results: IF staining of hiPSCs for NANOG and SOX2, validates the isogenic control, SLC35A2-S304P and SLC35A2-KO retain pluripotent characteristics. However, expression of pluripotent marker TRA1-81, which recognizes a galactose-containing keratan sulfate epitope on podocalyxin, was lost in SLC35A2-S304P and SLC35A2-KO compared to the isogenic control suggesting that SLC35A2 variants act through loss of SLC35A2 transporter activity. This was confirmed using a MALI lectin binding assay that showed reduced binding in SLC35A2-S304P and SLC35A2-KO protein lysates compared to the isogenic control. Loss of transporter activity is likely due to loss of UDPGalT expression in SLC35A2-S304P and SLC35A2-KO hiPSCs as shown by reduced IF staining compared to the isogenic control, which demonstrates clear expression and co-localization with GM-130 (Golgi marker). To evaluate if the model recapitulates the seizure phenotype associated with SLC35A2 variants, we used MEA to assess their impact on neural network activity and development. Compared to the isogenic control, SLC35A2-S304P develop synchronous activity earlier compared to the isogenic control as quantified by the spike train tiling coefficient (P-value=<0.01), mutual information (P-value=<0.01), and percentage of spikes in network spikes (P-value=0.33). Conclusions: Our data show that SLC35A2 variants in the hiPSC-derived neuronal model system influence the development of neural network connectivity through loss of SLC35A2 transporter activity. In depth studies of disease mechanism, including the effects of pathogenic SLC35A2 variants on single cell and network electrophysiology, neural development, and the ability of galactose to reverse the phenotypes are underway in this model system. Funding: No funding