Glycogen is the major glucose reserve in eukaryotes, and defects in glycogen metabolism and structure lead to disease. Glycogenesis involves interaction of glycogenin (GN) with glycogen synthase (GS), where GS is activated by glucose-6-phosphate (G6P) and inactivated by phosphorylation. We describe the 2.6 Å resolution cryo-EM structure of phosphorylated human GS revealing an autoinhibited GS tetramer flanked by two GN dimers. Phosphorylated N- and C-termini from two GS protomers converge near the G6P-binding pocket and buttress against GS regulatory helices. This keeps GS in an inactive conformation mediated by phospho-Ser641 interactions with a composite “arginine cradle”. Structure-guided mutagenesis perturbing interactions with phosphorylated tails led to increased basal/unstimulated GS activity. We propose that multivalent phosphorylation supports GS autoinhibition through interactions from a dynamic “spike” region, allowing a tuneable rheostat for regulating GS activity. This work therefore provides insights into glycogen synthesis regulation and facilitates studies of glycogen-related diseases.

The macromolecule glycogen is the major glucose reserve in eukaryotes and defects of glycogen metabolism and structure lead to glycogen storage diseases and neurodegeneration. Glycogenesis begins with self-glucosylation of glycogenin (GN), which recruits glycogen synthase (GS). GS is activated by glucose-6-phosphate (G6P) and inactivated by phosphorylation, but how these opposing processes are coupled is unclear. We provide the first structure of phosphorylated human GS-GN complex revealing an autoinhibited GS tetramer flanked by two GN dimers. Phosphorylated N- and C-terminal tails from two GS protomers converge to form dynamic “spike” regions, which are buttressed against GS regulatory helices. This keeps GS in a constrained “tense” conformation that is inactive and more resistant to G6P activation. Mutagenesis that weaken the interaction between the regulatory helix and phosphorylated tails leads to a moderate increase in basal/unstimulated GS activity, supporting the idea that phosphorylation contributes to GS inactivation by constraining GS inter-subunit movement. We propose that multivalent phosphorylation supports GS autoinhibition through interactions from a dynamic “spike” region, thus allowing a “tuneable rheostat” for regulating GS activity. Our structures of human GS-GN provide new insights into the regulation of glycogen synthesis, facilitating future studies of glycogen storage diseases.