Protein kinase A controls the hexosamine pathway by tuning the feedback inhibition of GFAT-1

The hexosamine pathway (HP) is a key anabolic pathway whose product uridine 5’-diphospho-N-acetyl-D-glucosamine (UDP-GlcNAc) is an essential precursor for glycosylation processes in mammals. It modulates the ER stress response and HP activation extends lifespan in Caenorhabditis elegans. The highly conserved glutamine fructose-6-phosphate amidotransferase 1 (GFAT-1) is the rate-limiting HP enzyme. GFAT-1 activity is modulated by UDP-GlcNAc feedback inhibition and via phosphorylation by protein kinase A (PKA). Molecular consequences of GFAT-1 phosphorylation, however, remain poorly understood. Here, we identify the GFAT-1 R203H substitution that elevates UDP-GlcNAc levels in C. elegans. In human GFAT-1, the R203H substitution interferes with UDP-GlcNAc inhibition and with PKA-mediated Ser205 phosphorylation. Our data indicate that phosphorylation affects the interactions of the two GFAT-1 domains to control catalytic activity. Notably, Ser205 phosphorylation has two discernible effects: it lowers baseline GFAT-1 activity and abolishes UDP-GlcNAc feedback inhibition. PKA controls the HP by uncoupling the metabolic feedback loop of GFAT-1.

T he hexosamine pathway (HP) converts fructose-6phosphate (Frc6P) to uridine 5′-diphospho-N-acetyl-Dglucosamine (UDP-GlcNAc) (Fig. 1a) 1 . In all, 2 to 3 % of cellular glucose enter the HP, as well as L-glutamine (L-Gln), acetyl-coenzyme A, and uridine 2 . Hence, the HP integrates sugar, amino acid, fatty acid, and nucleotide metabolism and is considered an important nutrient sensing pathway. The GlcNAc moiety of UDP-GlcNAc is used as a building block for several macromolecules, including peptidoglycans in bacteria, chitin in fungi and insects, or glycosaminoglycans such as hyaluronic acid in vertebrates [3][4][5][6][7][8] . Moreover, UDP-GlcNAc is an essential precursor for glycosylation reactions in eukaryotes. N-linked  glycosylation takes place in the endoplasmic reticulum (ER) and is crucial for proper protein folding and the solubility of proteins 9 . Mucin-type O-linked glycosylation occurs in the Golgi apparatus and is found on many cell surface and secreted proteins 10 . In protein O-GlcNAcylation that occurs in cytoplasm and nucleus, UDP-GlcNAc is the donor for the attachment of a single GlcNAc moiety to serine or threonine residues 11 . This process can fine-tune a protein's function akin to phosphorylation and often O-GlcNAcylation and phosphorylation compete for modification of the same sites 12,13 . HP activity is regulated by its first and rate-limiting enzyme glutamine fructose-6-phosphate amidotransferase (GFAT, EC 2.6.1.16) 2 . Two GFAT paralogs exist that primarily differ in their tissue-specific expression patterns 14 . GFAT is a modular enzyme composed of a glutaminase domain responsible for hydrolysis of L-Gln into L-glutamate (L-Glu) and an isomerase/transferase domain that catalyzes the isomerization of Frc6P to glucose-6-phosphate (Glc6P) as well as the transfer of ammonia to Glc6P to build glucosamine-6-phosphate (GlcN6P) 15 . The relative orientation of the two domains with their respective active sites is relevant for catalysis and affected by substrate binding. In the absence of Frc6P, the glutaminase domain is very flexible (Fig. 1b, 1) and no corresponding electron density is visible in the E. coli GFAT structure, although SDS-PAGE analyses of dissolved crystals indicate the presence of the full-length protein 16,17 . The reaction is initiated by binding of Frc6P to the isomerase domain, which triggers the flexible glutaminase domain to adopt a specific position relative to the isomerase domain and locks the two domains in this state 16 (Fig. 1b, 2). Subsequently, L-Gln binds to the glutaminase domains, closing the pocket of the glutaminase active site 18 . This is accompanied by a rotation of the glutaminase domain by 21°r elative to the isomerase domain 19 (Fig. 1b, 2). Finally, a solventinaccessible channel is formed that links the two active sites to allow the diffusion of ammonia [19][20][21] .
A disturbed function of GFAT-1 due to gfat-1 mutations cause limb-girdle congenital myasthenic syndrome with tubular aggregates 22,23 . This inherited disorder is characterized by defective neuromuscular transmission through impaired neuromuscular junctions, which transmit impulses from motor neurons to skeletal muscle fibers 22 . Moreover, many studies suggest a role of the HP and GFAT in the development and progression of diabetes or cancer [24][25][26] . Both disorders are characterized by an aberrant glucose metabolism, and the HP links altered metabolism with aberrant glycosylation. Especially elevated O-GlcNAcylation contributes to the pathogenesis of diabetes and cancer. Altered O-GlcNAcylation was also reported to play a critical role in neurodegenerative diseases, heart disease, and inflammation 27,28 .
Given that GFAT drives the metabolic flux of the HP, its regulation is of great physiological importance and in eukaryotes it occurs through UDP-GlcNAc feedback inhibition [29][30][31] and through phosphorylation 32,33 . In human GFAT-1, Ser205 and Ser235 are phosphorylated by cAMP-dependent protein kinase A (PKA) 32,34,35 . Ser205 phosphorylation has reported effects on GFAT-1 activity, although the published data are contradictory: On the one hand, phosphorylation of Ser205, and the corresponding Ser202 in human GFAT-2, were reported to increase activity 32,35 . On the other hand, it was described that Ser205 phosphorylation leads to decreased GFAT-1 activity 34 . Phosphorylation of the second PKA site Ser235 in GFAT-1, which is not conserved in all eukaryotes and is absent in GFAT-2, seems not to affect the enzymatic activity 34,36 . Ser243 was shown to be phosphorylated by adenosine monophosphate (AMP)activated protein kinase (AMPK) and calcium/calmodulin-dependent kinase II, but the effect on GFAT activity is not clear: depending on the study, phosphorylation at this site is reported to be either activating or inhibiting 33,37,38 .
Previously, in a Caenorhabditis elegans forward genetic mutagenesis screen we identified gain-of-function mutations in gfat-1 that suppress tunicamycin-induced proteotoxic stress 39 . These mutations result in single amino acid substitutions in GFAT-1 and increase UDP-GlcNAc concentrations in the worms that also show improved protein quality control as well as a significant lifespan extension 39 . Recently, we found that loss of regulation by UDP-GlcNAc feedback inhibition is one gain-of-function mechanism and identified a critical role of the interdomain linker in UDP-GlcNAc inhibition 40 . Although UDP-GlcNAc binds to the isomerase domain of GFAT-1, this interaction inhibits the glutaminase function and consequently the combined GlcN6P production 41 . We proposed that UDP-GlcNAc disturbs the tight coupling of the active sites by interacting with the interdomain linker and by interfering with the relative orientation of the two domains 40 (Figs. 1b, 3). Additional analyses of other GFAT-1 gain-of-function mutations might help to understand the molecular mechanisms of GFAT-1.
Here, we identify the GFAT-1 R203H gain-of-function mutation that interferes with UDP-GlcNAc inhibition and with PKAdependent phosphorylation at Ser205. Analyses of the phosphomimic S205D substitution resolved the controversially discussed effect of Ser205 phosphorylation. Our data demonstrate that PKA phosphorylation at Ser205 lowers GFAT-1 activity and simultaneously blocks UDP-GlcNAc inhibition. We propose a model in which Ser205 phosphorylation controls domain interactions in GFAT-1 to modulate activity and feedback inhibition.
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-22320-y ARTICLE NATURE COMMUNICATIONS | (2021) 12:2176 | https://doi.org/10.1038/s41467-021-22320-y | www.nature.com/naturecommunications C. elegans for mutants that are resistant to tunicamycin-induced proteotoxic stress 39 . From this screen, we obtained the mutant allele dh783, which was not characterized until now. While 10 µg/ ml tunicamycin was toxic for N2 wild type worms, the dh783 allele conferred strong tunicamycin resistance ( Fig. 1c and Supplementary Fig. 1a). To identify the causal mutation, we performed Hawaiian single nucleotide polymorphism (SNP) mapping 42 . The normalized linkage score identified SNPs in exons of five candidate genes on chromosome II including gfat-1 (Fig. 1d). To test an involvement of the HP in the resistance phenotype, we performed a developmental tunicamycin resistance assay and knocked down the expression of glucosamine-6phosphate N-acetyltransferase (Fig. 1a), the enzyme catalyzing the second HP reaction. HP inhibition by RNAi targeting the C. elegans homolog gna-2 reduced the developmental tunicamycin resistance observed in the dh783 mutants (Fig. 1e). Given that tunicamycin toxicity can be suppressed by elevated UDP-GlcNAc levels through gain-of-function mutations in gfat-1 39 , we quantified UDP-GlcNAc and its epimer UDP-GalNAc by mass spectrometry. Indeed, we found significantly elevated UDP-GlcNAc and UDP-GalNAc levels in whole worm lysates in mutant carriers of the dh783 allele ( Fig. 1f and Supplementary Fig. 1b). We conclude that tunicamycin resistance was caused by the gfat-1 (dh783) allele due to the GFAT-1 R203H gain-of-function substitution.
To decipher the effect of the R203H substitution, we characterized human GFAT-1 R203H in activity assays and found similar kinetics for L-Glu and GlcN6P production as in wild type GFAT-1 (Fig. 2a, b and Table 1). In contrast, UDP-GlcNAc dose response assays revealed an~6-fold higher IC 50 value of R203H GFAT-1 compared to wild type GFAT-1 (R203H: 231.7 −15.2/+16.3 µM; wild type: 41.4 −3.5/+3.9 µM) (Fig. 2c). Subsequently, we crystallized human GFAT-1 R203H to further elucidate the structural basis of the reduced UDP-GlcNAc sensitivity. The GFAT-1 R203H crystals diffracted to a resolution limit of 2.77 Å ( Table 2). . c Representative UDP-GlcNAc dose response assay of wild type (WT, black circle) and R203H (cyan square) GFAT-1 (mean ± SD, n = 3). Table: IC 50 UDP-GlcNAc values (mean ± SEM, n = 4, ***p < 0.0001, unpaired two-sided t test). d, e Position of the R203H mutation in the structure of GFAT-1. Proteins are presented as cartoons. Superposition of wild type GFAT-1 (light gray/dark gray, PDB ID 6R4E) and R203H GFAT-1 (green-cyan, teal) with an RMSD of 0.38 Åover 1280 main chain residues. Glc6P (yellow sticks), L-Glu (violet sticks), and UDP-GlcNAc (white sticks) are highlighted, as well as the position of R203H (black box). d Overview. R203H (sticks) is located at the glutaminase domain of GFAT-1 (dashed box). e Close-up view of the position of R203H focusing on residues in close proximity. The R203H mutation and residues in close proximity to the mutation are highlighted with sticks. Arg203 interacts with the neighboring loops (dashed lines). Source data are provided as a Source Data file.
Compared to wild type GFAT-1, the R203H substitution does not cause major structural changes (Fig. 2d, e and Supplementary  Fig. 2b). While Arg203 interacts with the backbone residues of Ile178 and Leu181, as well as with Asp279 in wild type GFAT-1, these interactions are abolished in the R203H GFAT-1 (Fig. 2e). In the mutant enzyme, however, the side chain of His203 protrudes between Asp278 and Asp279, without affecting the orientation of the neighboring loop (residues 277-280) (Fig. 2e). Together these findings suggest that blunted UDP-GlcNAc feedback inhibition is responsible for the gain-of-function of GFAT-1 R203H and that Arg203 has a functional role in UDP-GlcNAc inhibition.
The GFAT-1 R203H substitution interferes with PKA phosphorylation at Ser205. In addition to the observed gain-of-function caused by reduced feedback inhibition, the R203H substitution alters the PKA consensus sequence of Ser205 from RRGS to RHGS, suggesting a potential change of Ser205 phosphorylation and a subsequent loss of regulation (Fig. 3a). To test whether GFAT-1 R203H can be phosphorylated at Ser205, we purified wild type and R203H GFAT-1 and performed in vitro phosphorylation with PKA, followed by LysC protease digest and untargeted as well as targeted mass spectrometric analyses (Fig. 3b). In the untargeted analysis, we identified a number of phosphorylated sites, among them the PKA phosphorylation sites Ser205 and Ser235 32,34 (Supplementary Dataset 1). For the targeted analysis, heavy isotope labeled phosphorylated peptides of wild type or R203H GFAT-1 were spiked in, allowing more sensitive relative quantification of the Ser205 phosphorylation  L-Glu production D-GlcN6P production  Fig. 3b). Untreated insect cell-derived wild type and R203H GFAT-1 preparations showed low detectable levels of phosphorylation at Ser205. In vitro PKA treatment resulted in phosphorylation of wild type GFAT-1 at Ser205 (Fig. 3c, d). In contrast, the increase of Ser205 phosphorylation was 80-fold reduced in the presence of the R203H substitution (Fig. 3c, d).
The second known PKA phosphorylation site of GFAT-1 at Ser235 34 was used as a positive control. For this site, we found a significant increase in the abundance of phosphorylated peptides corresponding to Ser235 in both wild type and R203H GFAT-1, proving full activity of PKA in our setup ( Supplementary Fig. 3). Thus, disruption of the PKA consensus sequence by the R203H substitution prevents PKA-mediated phosphorylation at Ser205. Taken together, in addition to a lower sensitivity to UDP-GlcNAc inhibition, the GFAT-1 R203H substitution interferes with PKAdependent regulation by suppressing Ser205 phosphorylation.
PKA phosphorylation at Ser205 modulates UDP-GlcNAc inhibition of GFAT-1. The loss of PKA phosphorylation in the GFAT-1 gain-of-function R203H variant suggests that Ser205 phosphorylation might be inhibitory. Until now, the effect of phosphorylation at Ser205 has been controversially reported as either activating 32 or inhibiting 34 . We aimed to mechanistically understand the effect of Ser205 phosphorylation and generated a GFAT-1 S205D mutant to mimic the phosphorylation. Importantly, wild type insect cell-derived GFAT-1 showed almost no phosphorylation at Ser205 (Fig. 3c, d), permitting the direct comparison with the S205D variant in activity assays. We observed reduced k cat values for L-Glu and GlcN6P production resulting in lower catalytic efficiency (k cat /K m ) for both active sites (Fig. 4a, b and Table 1). Overall GlcN6P synthesis was decreased 5-fold in the phospho-mimic mutant (k cat /K m : 4.4 mM −1 sec −1 ) compared to wild type GFAT-1 (k cat /K m : 21.3 mM −1 sec −1 ) ( Table 1). Strikingly, the dose-dependent inhibition by UDP-GlcNAc was completely abolished in the S205D variant (Fig. 4c). These data suggest that PKA-dependent phosphorylation at Ser205 reduces GFAT-1 baseline activity in vitro. At the same time, the S205D variant remained constitutively active even in the presence of UDP-GlcNAc levels that strongly inhibit wild type GFAT-1. Thus, the presence of UDP-GlcNAc influences whether the phosphorylation is activating or inhibiting in vivo: at low UDP-GlcNAc concentrations up to~30 µM, the phosphorylation is inhibiting due to the reduced catalytic function, while at high UDP-GlcNAc concentrations above~30 µM, the phosphorylation is activating due to the loss of feedback inhibition.
Understanding the effect of Ser205 PKA phosphorylation on GFAT-1 activity, we next analyzed its relevance in vivo. For that purpose, we generated HEK293 cell lines that stably overexpress either wild type, R203H, or S205D GFAT-1 with an internal His 6tag ( Fig. 4d) and measured UDP-GlcNAc and UDP-GalNAc levels in the cell lysates ( Fig. 4e and Supplementary Fig. 4). The cells were cultured at high glucose concentrations of 25 mM, to avoid any glucose limitation. Strikingly, wild type GFAT-1 overexpression did not elevate UDP-GlcNAc or UDP-GalNAc concentrations. Likewise, R203H GFAT-1 cells did not show increased UDP-GlcNAc or UDP-GalNAc levels ( Fig. 4e and Supplementary Fig. 4). This suggests that UDP-GlcNAc feedback inhibition is effective in fully suppressing the activity of wild type and R203H GFAT-1 even during overexpression. In contrast, we found significantly elevated levels of UDP-GlcNAc and UDP-GalNAc when the phospho-mimic S205D variant of GFAT-1 was overexpressed ( Fig. 4e and Supplementary Fig. 4). These data support an interference of the S205D substitution with the UDP-GlcNAc inhibition, which rendered GFAT-1 constitutively active. To further investigate the consequence of the phospho-mimic mutation in a physiological context, we introduced the homologous S202D substitution by CRISPR/Cas9 in C. elegans. While the homozygous mutation was lethal, heterozygous carriers developed normally and were used for tunicamycin resistance assays. GFAT-1 S202D expression resulted in significant tunicamycin resistance, indicating elevated UDP-GlcNAc levels (Fig. 4f). In summary, our data suggest that PKA controls GFAT-1 activity by interference with its UDP-GlcNAc feedback inhibition.
PKA phosphorylation at Ser205 modulates GFAT-1 domain interactions. To analyze the molecular consequence of the S205D substitution in human GFAT-1, we solved the crystal structure at 2.22 Åresolution (Table 2). Further, we performed thermal shift assays to assess stability and obtained tryptophan fluorescence spectra to analyze potential domain movements. In the S205D crystal structure, we did not observe major structural differences compared to wild type GFAT-1 (Fig. 5a). Notably, the PKA target residue Ser205 is located in a loop of the glutaminase domain pointing towards the glutaminase active site and the interdomain cleft between the glutaminase and isomerase domains (Fig. 5a). Ser205 is thus optimally positioned to influence domain interactions upon phosphorylation. Next, we assessed the thermal stability of GFAT-1 in thermal shift assays. Full-length GFAT-1 showed two clearly distinguishable melting points at 53 and 65°C, which are termed "low" and "high" hereafter ( Fig. 5b and Supplementary Fig. 5a). Most likely, these two melting points represent the two domains, which possess different stabilities: the isolated isomerase domain melted at 64°C suggesting that the high melting point corresponded to the isomerase domain, while the low melting point was assigned to the glutaminase domain ( Fig. 5b and Supplementary Fig. 5a). To investigate their relative interactions, rising salt concentrations were then used to destabilize the domains. For full-length GFAT-1, high salt (0.5-1.0 M) lowered the melting point of the glutaminase domain by 3-4°C, and it lowered the melting point of the isomerase domain by 1.7-2°C (Fig. 5c). In contrast, the isolated isomerase domain showed a much stronger reduction of the melting temperature of up to −10°C (Fig. 5c). Together, these results show that in full-length GFAT-1 the isomerase domain is stabilized by the glutaminase domain, indicating domain interactions. We conclude that analysis of the melting points in GFAT-1 variants can reveal changes in the interaction between its domains. To independently test whether thermal shift assays can indeed indicate domain interactions, we generated the C-tail lock mutant L405R of human GFAT-1 and characterized it structurally, as well as in thermal shift assays and in kinetic measurements. The well-conserved C-tail (C-terminal residues 670-681) is the major mobile element between the isomerase and glutaminase domain of GFAT. It bears residues of both active sites and mediates interactions between glutaminase and isomerase domain 19 . In E. coli, its flexibility is restricted by a salt-bridge between Arg331 and Glu608 43 . Miszkiel and Wojciechowski showed in molecular dynamic simulations that the C-tail lock mutation, which introduces a salt-bridge similar as observed in E. coli, limits the C-tail's flexibility 43 . We solved the crystal structure (resolution 2.38 Å, Table 2) and confirmed that the L405R substitution indeed formed a salt-bridge with the Cterminal Glu681 and additionally interacted with Asp428 ( Supplementary Fig. 5b-d). The kinetics (L-Gln and Frc6P) and the UDP-GlcNAc inhibition remained unaffected by the C-tail lock mutation (Table 1 and Supplementary Fig. 5e-g). Importantly, the L405R mutant clearly showed a single melting point at 56.9 ± 0.1°C in thermal shift assays ( Supplementary Fig. 5h, i), indicating a changed stability of both domains, which can only be explained by the altered interactions between them. Therefore, thermal shift assays are a valid tool to analyze domain interactions of GFAT-1.
We used the thermal shift assay to test the role of the phosphomimic S205D substitution in domain-domain interactions of GFAT-1. At physiological pH, phosphorylation introduces a dianionic phosphate group that causes a repulsion with other negative charges and enables the formation of electrostatic interactions with positively charged residues. Compared to wild type GFAT-1, the S205D mutant showed strongly reduced thermal stability of the glutaminase domain, as indicated by the specific reduction of the lower melting point (Fig. 5d and Supplementary Fig 5j, k). The stability of the isomerase domain (high melting point) was not affected (Fig. 5d and Supplementary Fig 5j, k). Based on the crystal structure and circular dichroism (CD) measurements of wild type and S205D GFAT-1 ( Fig. 5a and Supplementary Fig. 5l), we exclude major structural changes between wild type and S205D GFAT-1. Moreover, the crystallographic B-factors within the glutaminase domain of the phospho-mimic and the wild type structures were comparable ( Supplementary Fig. 5m, n). Therefore, we rule out a destabilization of the glutaminase domain as a direct consequence of the S205D substitution. Instead, its reduced thermal stability is likely caused by altered interactions between the glutaminase and isomerase domains.
We further obtained fluorescence emission spectra to test if changed domain interactions or even a change in relative domain positions might occur in solution. While changes of the relative domain positions or stabilizing interactions were not detected in the S205D phospho-mimic crystal structure, they are conceivable in solution. Human GFAT-1 possesses two tryptophan residues: Trp50 is located at the surface of the glutaminase domain and Trp96 is positioned between the two domains at the glutaminase active site (Fig. 5e). Thus, the Trp96 fluorescence signal monitors the local environment within the interdomain cleft, potentially indicating changes in domain interactions. We mutated Trp50 to phenylalanine to eliminate its contribution to the fluorescence signal and measured the fluorescence spectra of wild type (W50F) or S205D (W50F/S205D) GFAT-1. The fluorescence signal of wild type (W50F) GFAT-1 was reduced compared to W50F/ S205D GFAT-1, pointing to a more solvent-exposed Trp96 in wild type GFAT-1 compared to the S205D variant ( Fig. 5f, g). These data support the notion that the Ser205 phosphorylation alters the interaction of the two GFAT-1 domains, affecting activity and UDP-GlcNAc feedback inhibition.

Discussion
Here we present the GFAT-1 R203H gain-of-function variant and decipher the two consequences of the substitution: first, GFAT-1 R203H displays strongly reduced PKA-dependent phosphorylation at Ser205 due to a disrupted consensus sequence and, second, shows weaker sensitivity to UDP-GlcNAc feedback inhibition. The current study resolves the contradicting reports regarding the effect of PKA phosphorylation at Ser205 in GFAT-1 and provides a model for phosphorylation-induced changes in domain interactions that control activity and regulation: the phospho-mimic S205D substitution lowered GFAT-1 activity while simultaneously abolishing UDP-GlcNAc inhibition. Importantly, we demonstrate a modulation of the UDP-GlcNAc feedback inhibition of GFAT-1 by PKA.
We describe two independent gain-of-function mechanisms of GFAT-1 R203H. First, when UDP-GlcNAc levels are low, the reduced Ser205 phosphorylation in GFAT-1 R203H results in higher GFAT-1 activity compared to wild type GFAT-1 that is inhibited by Ser205 phosphorylation. Second, when UDP-GlcNAc levels are high, wild type GFAT-1 is under stronger feedback inhibition than GFAT-1 R203H, resulting in elevated cellular UDP-GlcNAc levels in the mutant. The observation of high UDP-GlcNAc levels in whole worm lysates in the gfat-1 (dh783) R203H mutant points to the predominant role of the defective feedback inhibition as the relevant gain-of-function mechanism.
Reduced UDP-GlcNAc sensitivity of GFAT-1 R203H indicates a functional role of Arg203 in the UDP-GlcNAc feedback mechanism. In wild type GFAT-1, Arg203 is implicated in a saltbridge network, which might help to position the glutaminase and isomerase domains (Fig. 6a). Previously, we proposed that UDP-GlcNAc promotes a catalytically unfavorable domain orientation, inhibiting GFAT-1 40 . In general, domain movements are common in the family of glutamine amidotransferases and the function of GFAT depends on the relative orientation of the glutaminase and isomerase domains 44 . Mouilleron et al. reported a structure of the E. coli GFAT active site mutant C1A (PDB 3OOJ) with a drastically changed orientation of the glutaminase domain relative to the isomerase domain, where one monomer adopts an inactive orientation without forming the ammonia channel 45 . This structure underpins the high flexibility of the two domains relative to each other. In wild type E. coli GFAT, the glutaminase domain adopts a specific position relative to the isomerase domain after Frc6P binding 16 . The presence of Frc6P activates the glutaminase function of GFAT more than 100-fold in E. coli and 70-fold in C. albicans and it is very likely that this substrate-induced activation occurs in the human GFAT as well 18,41 . Also, L-Gln binding induces a rotation of the glutaminase domain relative to the isomerase domain in E. coli GFAT 19 . These domain movements upon substrate binding are required to enable specific interactions between the isomerase and glutaminase domains that are necessary for catalysis. It is well-established that phosphorylation can introduce conformational changes within loops or domains through altered electrostatic interactions 46 . A well-studied example for a phosphorylation-induced change in domain orientation is rabbit muscle glycogen phosphorylase, whose domains are shifted by~50°relative to each other after phosphorylation 47,48 . The phosphorylation of GFAT-1 at Ser205 would introduce, at physiological pH, a negatively charged phosphate group, potentially affecting ionic interactions in close proximity. Arg202 of the glutaminase domain and Glu425 of the isomerase domain likely form a salt-bridge between the two domains and this interaction might be affected upon Ser205 phosphorylation (Fig. 6a).
Although we did not observe major structural differences between wild type and S205D GFAT-1 in the crystal structures, thermal shift assays and fluorescence measurements, which are performed in solution, clearly distinguished the two variants. Thermal shift assays showed a destabilization of the glutaminase domain in the S205D mutant, indicating altered domain interactions. Further, the fluorescence spectra are consistent with changes in relative domain position upon phosphorylation. We propose that Ser205 phosphorylation suppresses UDP-GlcNAc inhibition by stabilizing the glutaminase and isomerase domains in a catalytically productive orientation (Fig. 6b). While this conformation does not permit the high catalytic rate of wild type GFAT-1, the fixed conformation of the domains would prevent a UDP-GlcNAc induced domain orientation that might be needed for inhibition.
Our data reveal a UDP-GlcNAc concentration-dependent effect on the activity of phospho-Ser205 GFAT-1. This explains the previous contradicting reports regarding GFAT 34 . In both studies, cells were treated with forskolin to activate PKA and the GFAT-1 activity was analyzed in cell lysates. These lysates contained unknown UDP-GlcNAc concentrations that are likely to have affected the results. Moreover, Chang et al. confirmed the inhibitory effect in vitro with GST-tagged GFAT-1 in the absence of UDP-GlcNAc. However, the data obtained from a GST-tagged GFAT-1 should be taken with caution, because tagging GFAT at the N-or C-terminus interferes with catalysis 34,49,50 .
Very few enzymes are known in which phosphorylation interferes with feedback inhibition 51 . The uncoupling of the feedback loop by GFAT-1 phosphorylation is an elegant mechanism to rapidly modulate its enzymatic activity after a stress-induced signal. The cAMP-PKA signaling pathway acts downstream of G protein-coupled receptors (GPCR) to mediate signals of neurotransmitters and hormones, such as glucagon or adrenaline [52][53][54][55] . The downstream protein targets of the cAMP-PKA signaling pathway regulate glucose homeostasis by inhibition of glycolysis and glycogen synthesis, as well as by triggering glucose release through stimulation of glycogenolysis and gluconeogenesis 55 . Our previous data indicated that GFAT-1 is under a constant UDP-GlcNAc inhibition in vivo 40 . Presumably, uncoupling of the feedback inhibition maintains GFAT-1 activity and ensures steady UDP-GlcNAc production when PKA activity is high. Given that UDP-GlcNAc is an essential building block and precursor for all glycosylation reactions in mammals, this mechanism is optimally positioned to ensure a constant UDP-GlcNAc supply. In all, our findings illuminate how the different means of GFAT-1 regulation, kinase signaling and feedback inhibition, are coordinated at the molecular level to fine-tune metabolic flux in the HP.
C. elegans developmental tunicamycin resistance assay. Eggs were transferred to NGM plates containing 10 µg/ml tunicamycin (Sigma-Aldrich) seeded with freeze-killed E. coli. Freeze-killed bacteria were obtained by three freeze/thaw cycles of pelleted overnight cultures. After 4 days, day 1 adults were scored, sick larvae were not counted. Throughout the experiment, strain identity was unknown to researchers.
Mutant Hawaiian SNP mapping and sequence analysis. Genomic DNA was prepared using the QIAGEN Gentra Puregene Kit according to the manufacturer's protocol. Whole genome sequencing was conducted on the Illumina HiSeq2000 platform. Paired-end 100 bp reads were used; the average coverage was larger than 16-fold. Sequences were analyzed using the CloudMap Hawaiian and Variant Discovery Mapping on Hawaiian Mapped Samples (and Variant Calling) Workflow_2-7-2014 pipeline on Galaxy 42,57 . The WS220/ce10 C. elegans assembly was used as reference genome.
RNAi experiments. HT115 E. coli carrying vectors for dsRNA of the target gene under an isopropyl β-D-1-thiogalactopyranoside (IPTG) inducible promotor were used. Bacteria were seeded on NGM plates containing 100 µg/µl ampicillin (Merck Millipore) and 1 mM IPTG (Roth), RNAi was done for two generations. RNAi targeting luciferase was used as control. All RNAi clones were obtained from Ahringer and Vidal RNAi libraries 58,59 . Clones were validated by Sanger sequencing.
Small molecule LC/MS/MS analysis. UDP-HexNAc concentrations were measured as described previously 39 . C. elegans or HEK293 cells were lysed in water by freeze/thaw cycles and subjected to chloroform/methanol extraction. Absolute UDP-HexNAc levels were determined using an Acquity UPLC connected to a Xevo TQ Mass Spectrometer (both Waters) and normalized to total protein content.
Site-directed mutagenesis. A pFL vector for the generation of baculoviruses for the expression of human GFAT-1 isoform 2 (hGFAT1) with internal His 6 -tag between Gly299 and Asp300 was cloned previously 40 . The mutations W50F, R203H, S205D, and L405R were introduced into pFL-hGFAT1-His299 by sitedirected mutagenesis as described previously 60 . Mutagenesis primers are listed in Supplementary Table 1.
Baculovirus generation and insect cell expression of full-length GFAT-1. Sf21 (DSMZ no. ACC 119) suspension cultures were maintained in SFM4Insect™ HyClone™ medium with glutamine (GE Lifesciences) in shaker flasks at 27°C and 90 rpm. GFAT-1 variants were expressed using the MultiBac baculovirus system 61 . GFAT-1 variants were integrated into the baculovirus genome via Tn7 transposition and maintained as bacterial artificial chromosome in DH10EMBacY E. coli. Recombinant baculoviruses were generated by transfection of Sf21 cells with bacmid DNA. Obtained baculoviruses were used to induce protein expression in Sf21 cells.
Bacterial expression of GFAT-1 isomerase domain. The isomerase domain of human GFAT-1 isoform 2 (residues 316-681) was integrated in the plasmid pET28a(+) using NdeI and HindIII restriction sites. Cloning primers are listed in Supplementary Table 1. This vector was used to recombinantly express the isomerase domain with N-terminal His 6 tag and a thrombin cleavage site under the control of the T7 promoter in Rosetta (DE3) E. coli. LB cultures were incubated at 37°C until an OD 600 of 0.4-0.6 was reached. Protein expression was induced by addition of 0.5 mM IPTG for 3 h at 37°C. After harvest, pellets were stored at −80°C.
GDH-coupled activity assay and UDP-GlcNAc inhibition. GFAT's amidohydrolyzing activity was measured in a coupled enzymatic assay using bovine glutamate dehydrogenase (GDH, Sigma-Aldrich G2626) as previously described 50 with small modifications. The reaction mixtures contained 6 mM Frc6P, 1 mM APAD, 1 mM EDTA, 50 mM KCl, 100 mM potassium phosphate buffer pH 7.5, 6.5 U GDH per well and (for L-Gln kinetics) varying concentrations of L-Gln. For UDP-GlcNAc inhibition assays the L-Gln concentration was 10 mM. The plate was pre-warmed at 37°C and after enzyme addition, activity was monitored continuously at 363 nm in a SynergyHT (BioTek) microplate reader using Gen5 software (BioTek). Synthetized APADH was calculated with ε (363 nm, APADH) = 9100 l mol −1 cm −1 . Reaction rates were determined by Excel (Microsoft) and K m , v max , and IC 50 were obtained from Michaelis Menten or dose response curves, which were fitted by Prism 7 or 8 software (Graphpad).
GNA-1 and GNA-1-coupled activity assays. Human GNA-1 activity was measured as described previously 68 . For kinetic measurements, the assay mixture contained 0.5 mM AcCoA, 0.5 mM DTNB, 1 mM EDTA, 50 mM Tris/HCl pH 7.5, and varying concentrations of D-GlcN6P. The plates were pre-warmed at 37°C and reactions were initiated by addition of GNA-1. Absorbance at 412 nm was followed continuously at 37°C in a SynergyHT (BioTek) microplate reader. The amount of produced TNB, which matches CoA production, was calculated with ε (412 nm, TNB) = 13,800 l mol −1 cm −1 . Typically, GNA-1 preparations showed a K m of 0.2 ± 0.1 mM and a k cat of 41 ± 8 sec −1 .
GFAT's D-GlcN6P production was measured in a GNA-1-coupled activity assay following the consumption of AcCoA at 230 nm in UV transparent 96 well microplates 68 . The assay mixture contained 10 mM L-Gln, 0.1 mM AcCoA, 50 mM Tris/HCl pH 7.5, 2 µg hGNA-1, and varying concentrations of Frc6P. The plates were incubated at 37°C for 4 min before adding L-Gln. Activity was monitored continuously at 230 nm and 37°C in a SynergyHT (BioTek) microplate reader. The amount of consumed AcCoA was calculated with ε (230 nm, AcCoA) = 6436 l mol −1 cm −1 . As UDP-GlcNAc absorbs light at 230 nm, the GNA-1-coupled assay cannot be used to analyze UDP-GlcNAc effects on activity.
Untargeted protein Mass Spectrometry. 20% of STAGE tip purified peptides were separated on a 25 cm, 75 μm internal diameter PicoFrit analytical column (New Objective) packed with 1.9 μm ReproSil-Pur 120 C18-AQ media (Dr. Maisch HPLC GmbH) using an EASY-nLC 1200 (Thermo Fisher Scientific). The column was maintained at 50°C. Buffer A and B were 0.1% formic acid in water and 0.1% formic acid in 80% acetonitrile. Peptides were separated on a segmented gradient from 6% to 31% buffer B for 57 min and from 31 to 44% buffer B for 5 min at 200 nl/min. Eluting peptides were analyzed on an Orbitrap Fusion Tribrid mass spectrometer (Thermo Fisher Scientific). Peptide precursor m/z was measured at 60,000 resolution in the 350 to 1500 m/z range. Precursors with charge state from 2 to 7 only were selected for HCD fragmentation using 27% normalized collision energy. The m/z values of the peptide fragments were measured at a resolution of 30,000 using an AGC target of 2e5 and 80 ms maximum injection time. Upon fragmentation, precursors were put on a dynamic exclusion list for 45 s.
Heavy synthetic peptides, corresponding to phosphorylated Ser205 were separated on a segmented gradient from 6 to 60% buffer B for 62 min at 200 nl/ min. Analysis was carried out as described above.
Raw data were analyzed with MaxQuant version 1.6.1.0 using the integrated Andromeda search engine 71,72 . Peptide fragmentation spectra were searched against manually created GFAT-1 fasta file. The database was automatically complemented with sequences of contaminating proteins by MaxQuant. Methionine oxidation, protein N-terminal acetylation, and Phospho (STY) were set as variable modifications; cysteine carbamidomethylation was set as fixed modification. The digestion parameters were set to "specific" and "LysC/P," The minimum number of peptides and razor peptides for protein identification was 1; the minimum number of unique peptides was 0. Protein identification was performed at a peptide spectrum matches and protein false discovery rate of 0.01. For the analysis of the raw data from the synthetic heave peptides, Lys8 was set as "Standard" and used as a variable modification.
Targeted protein Mass Spectrometry. The phosphorylated (Ser205, pS), heavily labeled (K*) reference peptides Maxi SpikeTides L (wild type: SVHFPGQAVGTRRG-pS-PLLIGVRSEH-K* and mutant R203H: SVHFPGQAVGTRHG-pS-PLLIGVRSEH-K*) were purchased from JPT Peptide Technologies (Berlin, Germany). Heavy peptides were dissolved in 50% acetonitrile, 0.05% formic acid in water and sonicated in a water bath for 1 min. The solution was further diluted using 0.1% formic acid in water for a final peptide concentration of 7 nM. For analysis, desalted GFAT-1 peptides were dissolved in 10 µl 0.1% formic acid in water and 2 µl of the peptide solution were combined with 2 µl of the tenfold diluted heavy peptide solution; 2 µl were analyzed by targeted mass spectrometry.
For targeted analysis, peptides were separated as described above. Targeted analysis was performed on an Orbitrap Fusion Tribrid mass spectrometer (Thermo Fisher Scientific). The m/z values for charge states 4 and 5 from the heavy and light peptide were chosen for targeted fragmentation across the entire LC run. The isolation width was 1.6 and collision energy was set to 27%. Fragment m/z values were measured in the Orbitrap in profile mode, at a resolution of 60 K, an AGC target of 5e4, and a maximum injection time of 118 ms.
Raw data was analyzed using Skyline 73 version 19.1.0.193. A library was built from the untargeted analysis of the heavy peptides. The library ion match tolerance was set to 0.05 m/z, the ten most intense product ions were picked. Ions with charges 1 or 2 and type y an b were used for quantification. An isotope modification of type "Label" was created using the following modification: 13C(6) 25N(2) (K). Isotope label type and internal standard type were set to "heavy". For quantification, MS level was set to 2.
Thermal shift assay. Protein thermal stability was analyzed by thermal shift (thermofluor) assays. Proteins were incubated with SYPRO orange dye (Sigma-Aldrich), which binds specifically to hydrophobic amino acids, elevating fluorescence at 610 nm when excited with a wavelength of 490 nm. The melting temperature is defined as the midpoint of temperature of the protein-unfolding transition 74 . This turning point of the melting curve was extracted from the derivative values of the RFU curve, where a turning point to the right is a minimum. Reactions were done in white RT-PCR plates and contained 5 µl SYPRO orange dye (1:500 dilution in ddH 2 O) and 5-10 µg protein in SEC buffer in a total volume of 50 µl. The melting curves were measured in a Bio-Rad CFX-96 Real-Time PCR machine at 1°C/min at the FRET channel and the data analyzed with CFX Maestro (Bio-Rad).
Circular dichroism (CD) spectroscopy. GFAT-1 was dialyzed in 10 mM potassium phosphate buffer pH 7.5, 0.5 mM TCEP, 0.5 mM Na 2 Frc6P, and 10% (v/v) glycerol, and the protein concentration was adjusted to 0.2 mg/ml. UV spectra in the range of 195-260 nm were recorded with a J-715 CD spectropolarimeter (Jasco, Gross-Umstadt, Germany) at 20°C using a quartz cuvette with 0.1 cm path length. The buffer baseline was recorded separately and subtracted from each sample spectrum. The obtained ellipticity (θ, deg) was converted to mean residue ellipticity [θ] using: [θ] = θ/(10 n c l) in deg cm 2 dmol −1 (n is the number of amino acids, c the protein concentration in mol/l, and l the path length of the cuvette).
Fluorescence measurements. Fluorescence emission spectra were measured with a FluoroMax ® -4 (HORIBA) spectral fluorimeter at a concentration of 1.5 mg/ml in SEC buffer without Frc6P at 20°C. Excitation wavelength was set to 295 nm and the spectra were recorded in a range of 305 to 450 nm in a quartz cuvette with 1.5 mm path length. The bandwidth for the excitation monochromator was 4 nm and for the emission monochromator 2 nm.
Cells stably overexpressing GFAT-1 variants were generated by transfection of HEK293 cells with pcDNA3.1 plasmids with human GFAT-1 containing an internal His 6 -tag between Gly299 and Asp300 (pcDNA3.1-hGFAT1-His299). Internally tagged GFAT-1 was subcloned from pFL-hGFAT1-His299 using NheI and HindIII restriction sites. For each variant, one well of a 6 well plate was transfected with 2 µg of plasmid DNA with Lipofectamine ® 2000 (LifeTechnologies™) according to the manufacturer's protocol. Selection was performed with 500 µg/ml G418 for several weeks. For Western blot analysis, cells were washed, and proteins were extracted with RIPA-buffer (120 mM NaCl, 50 mM Tris/HCl, 1% (v/v) NP40, 0.5% (w/v) deoxycholate, 0.1% (w/v) SDS; pH 7.5). Cell debris was removed by centrifugation, protein concentration of the supernatants were determined using Pierce™ bicinchoninic acid (BCA) protein assay kit (Thermo Scientific) according to the manufacturer's protocol. Two micrograms of cell lysate was separated by reducing SDS-PAGE and transferred to PVDF membranes. Primary antibodies against human GFAT1 and α-TUBULIN were used. Chemiluminescence of the appropriate secondary rabbit or mouse HRPconjugated antibodies after incubation with ECL HRP substrate (Immobilon™ Western HRP Substrate, Millipore) was detected using a ChemiDoc™ Quantity One ® system (Bio-Rad).
Reporting Summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.