CryoEM analysis of the essential native UDP-glucose pyrophosphorylase from Aspergillus nidulans reveals key conformations for activity regulation and function

ABSTRACT Invasive aspergillosis is one of the most serious clinical invasive fungal infections, resulting in a high case fatality rate among immunocompromised patients. The disease is caused by saprophytic molds in the genus Aspergillus, including Aspergillus fumigatus, the most significant pathogenic species. The fungal cell wall, an essential structure mainly composed of glucan, chitin, galactomannan, and galactosaminogalactan, represents an important target for the development of antifungal drugs. UDP (uridine diphosphate)-glucose pyrophosphorylase (UGP) is a central enzyme in the metabolism of carbohydrates that catalyzes the biosynthesis of UDP-glucose, a key precursor of fungal cell wall polysaccharides. Here, we demonstrate that the function of UGP is vital for Aspergillus nidulans (AnUGP). To understand the molecular basis of AnUGP function, we describe a cryoEM structure (global resolution of 3.5 Å for the locally refined subunit and 4 Å for the octameric complex) of a native AnUGP. The structure reveals an octameric architecture with each subunit comprising an N-terminal α-helical domain, a central catalytic glycosyltransferase A-like (GT-A-like) domain, and a C-terminal (CT) left-handed β-helix oligomerization domain. AnUGP displays unprecedented conformational variability between the CT oligomerization domain and the central GT-A-like catalytic domain. In combination with activity measurements and bioinformatics analysis, we unveil the molecular mechanism of substrate recognition and specificity for AnUGP. Altogether, our study not only contributes to understanding the molecular mechanism of catalysis/regulation of an important class of enzymes but also provides the genetic, biochemical, and structural groundwork for the future exploitation of UGP as a potential antifungal target. IMPORTANCE Fungi cause diverse diseases in humans, ranging from allergic syndromes to life-threatening invasive diseases, together affecting more than a billion people worldwide. Increasing drug resistance in Aspergillus species represents an emerging global health threat, making the design of antifungals with novel mechanisms of action a worldwide priority. The cryoEM structure of UDP (uridine diphosphate)-glucose pyrophosphorylase (UGP) from the filamentous fungus Aspergillus nidulans reveals an octameric architecture displaying unprecedented conformational variability between the C-terminal oligomerization domain and the central glycosyltransferase A-like catalytic domain in the individual protomers. While the active site and oligomerization interfaces are more highly conserved, these dynamic interfaces include motifs restricted to specific clades of filamentous fungi. Functional study of these motifs could lead to the definition of new targets for antifungals inhibiting UGP activity and, thus, the architecture of the cell wall of filamentous fungal pathogens.

IMPORTANCE Fungi cause diverse diseases in humans, ranging from allergic syndromes to life-threatening invasive diseases, together affecting more than a billion people worldwide. Increasing drug resistance in Aspergillus species represents an emerging global health threat, making the design of antifungals with novel mechanisms of action a worldwide priority. The cryoEM structure of UDP (uridine diphosphate)-glucose pyrophosphorylase (UGP) from the filamentous fungus Aspergillus nidulans reveals an octameric architecture displaying unprecedented conformational variability between the C-terminal oligomerization domain and the central glycosyltransferase A-like catalytic domain in the individual protomers. While the active site and oligomerization interfaces are more highly conserved, these dynamic interfaces include motifs restricted to specific clades of filamentous fungi. Functional study of these motifs could lead to the definition of new targets for antifungals inhibiting UGP activity and, thus, the architecture of the cell wall of filamentous fungal pathogens. additional nucleotide-sugars. For example, the UGP reaction product UDP-Glc can be converted to UDP-galactose and UDP-glucuronic acid, making UGP one of three NSPs that are essential for the synthesis of the major fungal cell wall components.
In this work, we determine that the galF gene encoding the enzyme is an essential gene in A. nidulans. Furthermore, we isolate native UGP from A. nidulans (AnUGP; previously known as GalF or Gal2; FungiDB ID AN9148), and report, to our knowledge, the first structure of a UGP based on cryoEM data. AnUGP shows an octameric architec ture with unprecedented conformational variability in the individual protomers. In combination with enzymatic activity measurements and computational analysis, we unveil the molecular basis of substrate recognition, specificity, and catalysis for AnUGP, providing the groundwork for the future exploitation of UGP as a potential antifungal target.

AnUGP mediates the biosynthesis of UDP-Glc in A. nidulans
Endogenous native AnUGP was purified owing to its ability to bind a Ni-NTA affinity column without any added tag. The elution profile of the protein from a gel filtration column (Fig. S1) indicated it was an intact octameric complex. The activity of the isolated AnUGP was measured in the direction of synthesis of the nucleotide-sugar and pyro phosphate (PPi), coupling the reaction with the hydrolysis of PPi and following the formation of Pi by malachite green (MG) assay (Fig. 1B). As expected, the enzyme was specific for UTP and unable to use ATP or GTP as donor substrate under the conditions used (Fig. 1B). The specific activity of the enriched enzyme, calculated per milligram of total protein, at 30°C, is 0.41 ± 0.06 µmol/mg·min. Altogether, the enzymatic activity measurements indicate that AnUGP is a bona fide UGP.

AnUGP, encoded by the galF gene, is essential for growth in A. nidulans
Early genetic screens identified a series of A. nidulans mutants unable to grow when galactose (Gal) was used as the sole carbon source for the cultures (22). Among those mutants, gal2 showed slow growth under these culture conditions and was thus classified as a Gal utilization mutant. The A. nidulans linkage map (https://www.fgsc.net/ Aspergillus/gene_list/locigh.html) renamed gal2 as galF, and the FungiDB database (23) annotated galF as encoding a putative UTP-glucose-1-phosphate uridylyltransferase; thus, we refer to the protein as A. nidulans UGP (AnUGP) (24). UGP is required for Gal utilization. Specifically, the product of UGP, UDP-Glc, is used by galactose-1-phosphate (Gal-1-P) uridylyltransferase as a substrate for the conversion of Gal-1-P to Glc-1-P in the Leloir pathway (GALT; EC 2.7.7.12) (25).
To assess the essentiality of the gene galF, we generated a linear DNA construct in which the selection marker pyrG Afum (the pyrG gene of A. fumigatus) was fused to the promoter region (5´-UTR) and the 3´-UTR region of galF (see Materials and Methods). Protoplasts of strain TN02A3 were transformed with this construct. The TN02A3 strain bears a deletion of the gene nkuA, the homolog of the human KU70 gene, which is essential for non-homologous end joining DNA in double-strand break repair and favors homologous recombination (26). Most of the colonies obtained on selective regenera tion medium (RMM) displayed irregular and slower radial growth compared to wild type (Fig. 1C) and were unable to grow on selective AMM (standard Aspergillus minimal medium). These characteristics suggest that they were heterokaryons (27). Haploid and homokaryotic knock-out transformants would likely display an increased branching pattern as reported for RNAi knock-down mutants of the ugp homolog of the Agaricomy cete fungus Ganoderma lucidum (28).
A few of the colonies obtained showed a stable phenotype and homogeneous radial growth, both on selective RMM and AMM culture media (Fig. 1C). In contrast to previ ously described gal2 mutants (22), these isolated colonies did not show any growth defect on medium supplemented with Gal as the sole carbon source (Fig. 1C). Moreover, diagnostic PCR reactions indicated the presence of the pyrG cassette and the galF gene, strongly suggesting that those transformants were diploids (Fig. S2). Overall, these results indicate that none of the two types of transformants obtained were haploid and homokaryotic galF deletion mutants, and that galF, encoding AnUGP, is essential for A. nidulans viability.

Visualizing native AnUGP by cryoEM
To understand the molecular mechanism of UDP-Glc biosynthesis at a structural and functional level, we determined the structure of AnUGP (Uniprot code Q5I6D1) by cryoEM ( Fig. 2; Tables S1 and S2; Fig. S3). Both the raw cryoEM images and the 2D class averages (Fig. S3A through C) show a particle with eight subunits arranged around a fourfold axis (top view) and a twofold axis (side view), suggesting it displays D4 symmetry. Each AnUGP subunit (57.6 kDa protein; 514 amino acids) is composed of three domains: (i) the α-helical N-terminal (NT) domain (residues 1-81, 199-227, 354-376), (ii) the catalytic glycosyltransferase A-like (GT-A-like) domain [residues 82-198, 228-353, 377-404 (29-31)] containing the active site, and (iii) the C-terminal (CT) oligomeriza tion domain (residues 405-514) containing a left-handed parallel β-helix (LβH; residues Leu461-His514; Fig. 2C). The eight subunits are readily seen in an unbiased ab initio cryoEM map and an unsymmetrized C1 map ( Fig. S1D and E). In the C1 map, the with Glc (row 1) or Gal (row 2) as the main carbon source, after 96 h of culture at 37°C on 5.5 cm plates. gal2 mutants were described to show a growth phenotype when Gal was used as the carbon source (22). Colony radii of the two diploids analyzed here are equal to that of the reference wild-type strain. The diploid nature of these two transformants is confirmed in Research Article mBio density corresponding to one subunit appears weaker or more fragmented than the others, suggesting one subunit has higher conformational or occupational heterogene ity. To take advantage of the particle's apparent symmetry and increase the resolution of the cryoEM map, we refined the data while applying D4 symmetry, resulting in a 3.98 Å structure ( Fig. 2A; Video S1). Analysis of the map indicates that it displays a broad local resolution range, with the CT LβH oligomerization domain showing up to 3.2 Å resolution, while peripheral secondary structure elements in the NT and catalytic domains show resolution >5 Å (Fig. S3F). The low local resolution indicates that the peripheral regions display conformational and compositional variability that breaks the D4 symmetry. To account for this flexibility and improve the interpretability of the map, we used symmetry expansion to generate a new particle data set where the projections are replicated but rotated using the point group symmetry such that all eight subunits of the complex are aligned on a single subunit. Afterward, a restrained local refinement, focusing on this single subunit, yielded a 3.5 Å cryoEM map for the AnUGP monomer with local resolution ranging from 3.0 to >6 Å ( Fig. 2B and C; Fig. S3H and I). Accurate tracing of the backbone and positioning of the sidechains were possible in the best regions of the map where the local resolution was high, whereas only placement of secondary structure elements (e.g., helices) was possible where the resolution was low (Fig. S3F, H (43)]. This structural similarity to NSPs, and specifically to other eukaryotic UGPs, reinforces the observation that AnUGP is a bona fide UGP (Fig. 1B).

AnUGP oligomerizes through the CT domain utilizing interfaces maintained in animal and fungal UGPs
UGPs exist in diverse oligomeric states, including monomers in protozoan parasites (38,44) and plants (35,(45)(46)(47), dimeric and tetrameric bacterial UGPs (48,49), and octameric species observed in the animal and fungal kingdoms (33,36,(50)(51)(52). Whereas plant UGPs, which are active as monomers, were described to be transiently inactivated by dimerization (35,(45)(46)(47), octamers appear to be the active species of fungal and human UGPs (33,52). Octameric AnUGP can be described as a tetramer of homodimers ( Fig. 3) held together by two interfaces on the CT LβH (53), which is built of short β-strands (β17-β23) oriented parallel to each other and describing a triangular prism ( Fig. 3A and B). The cryoEM map of the LβH region shows high local resolution (up to 3.2 Å; Fig. 3B; Fig.  S3F, G, H), allowing the oligomerization interface to be described. Two AnUGP subunits build the homodimer through "end-to-end" contacts between their C-terminal extended β-strands ( Fig. 3A and B), forming an extended intermolecular β-sheet and a joint β-helix ( Fig. 3C through E, I, K). This interface involves (i) antiparallel pairing of the final extended β23-strands, which creates an extensive potential hydrogen bond network between the backbones, and (ii) hydrophobic residues like Val504, Val505, Leu509, Ile511, and Leu512 ( Fig. 3D and E). Next, the dimers engage in "side-to-side" contacts via one face of the LβH, which includes uncharged and hydrophobic residues like Thr465, Thr481, Ile483, Val485, and Val504, to form a tetramer of dimers ( Fig. 3F through H, J, and K). Additionally, in the "side-to-side" interface ( Fig. 3F through H), the rings of His463 from both partners lie "face-to-face" (subunits A and E), while the two CT His514 from each partner's "end-toend" pair (subunits H and D) approach the His463 stack end on, such that four subunits participate in the "side-to-side" interface. Overall, this oligomerization mode is preserved in other fungal and human UGPs (33,36,54) (Fig. S5A and B). All residues known to participate in both the "end-to-end" and "side-to-side" contacts are either strictly or functionally conserved in AnUGP, likely supporting an equivalent function in ScUGP (33) and HsUGP (36,54) (Fig. 4). Interestingly, extensive mutagenesis performed on the HsUGP LβH domain (Table S3C) suggests that individual or simultaneous mutation of AnUGP S508 and/or L509, as well as truncation of the CT eight amino acids, would result in dissociation of the octameric complex, which in the case of HsUGP was associated with a dramatic drop in activity (52).

The AnUGP active site and enzymatic mechanism
UGPs catalyze the interconversion of UTP and Glc-1-P to UDP-Glc and PPi in a Mg 2+dependent reaction, following an ordered sequential Bi-Bi mechanism in which the uridine compound is the first to enter and the last to exit the active site ( Fig. 1A; Fig.  S6) (44,(54)(55)(56)(57). To identify the AnUGP active site and gain insight into the catalytic mechanism of AnUGP, we compared the binding pocket of AnUGP with HsUGP (Fig. 5). The UDP-Glc binding pocket is in the GT-A-like domain, which consists of one Rossmann fold (58). The UDP-Glc binding site faces toward the inner cavity of the complex with four pockets each in both the upper and lower rings ( Fig. 5A and B). As seen in Fig. 5C, mapping the conservation of AnUGP on the cryoEM map shows that the region around the UDP-Glc binding pocket is highly conserved. Based on protein sequence alignments ( Fig. 4) and previous structural/functional studies using X-ray crystallography (33,35,36,54,59,60), we propose that in AnUGP, (i) the Glc binding pocket is formed by Asn256, Asn329, and Asn331, (ii) the three conserved basic residues, Lys135, His228, and Lys403, coordinate the phosphate groups, (iii) Asp258 is involved in the coordination of a Mg 2+ (57,61), and (iv) the binding site for the uridine moiety is formed by Leu121, Gly123, and Gly124 in the nucleotide-binding (NB) loop, as well as Gln198 and Gly227. In addition to being strictly conserved in AnUGP (Fig. 4), these residues are positioned similarly to those in HsUGP (Fig. 5D), supporting a common substrate/product binding and catalysis for eukaryotic UGPs. As shown above, AnUGP displays a strong preference for UTP ( Fig.  1B) and, accordingly, in the structure, it is evident that the presence of small residues, Gly123 and Gly124, in the NB loop is vital for reducing steric clashes with the uridine moiety (Fig. 5D). Previous mutation of the residues corresponding to Gly123 and Gly124 in HsUGP led to the loss of UGP activity (54) in agreement with structural studies on LmUGP suggesting these positions are restricted to glycine for steric reasons (59).

Conformational flexibility in the AnUGP subunits
The AnUGP cryoEM maps do not show proper D4 symmetry, but instead, the decreased local resolution in the NT and catalytic domains (Fig. S3F, H, J) suggests that these regions exhibit local flexibility that breaks the symmetry. Similarly, if the cryoEM-derived model of AnUGP is compared with the X-ray models of closely related UGPs, it is evident that the relative arrangement of the NT/central and CT domain can vary (Fig. S5C). One power of cryoEM lies in the fact that before vitrifying the sample, the protein is maintained in   Research Article mBio a solution mimicking its native state where it can sample its accessible conformational space (62). Therefore, this distribution of molecular conformations remains after freezing, and the resulting reconstruction represents an amalgamation of preferred low-energy states (62).
To explore the variability in the AnUGP monomer captured during freezing, we applied 3D variability analysis (3DVA) (63) to the symmetry-expanded data set before local alignment while using a mask to focus the analysis on the single monomer. 3DVA describes the variability in the data by solving for a continuous family of 3D AnUGP NT residues and residues forming the loop between α10 and β4, the side of helix α9, and the short loop between β14 and β15.
Research Article mBio structures and assigns each particle to a position within the principal component (PC) space. In this analysis, we observed that the first two PCs described clearly interpretable conformational changes in AnUGP. Importantly, we did not observe discrete clusters in the PC space, indicating that the heterogeneity in AnUGP was continuous rather than discrete, namely, in the apo AnUGP structure, there are no dominant conformations present. To visualize the continuous conformational change, a "rolling window" is used to select particles along each PC axis. These particle subsets, termed intermediates, are reconstructed so movement along the PC axis can be described as a series of structures ( Fig. 6A through F; Video S2). In Fig. 6, structures from the extremes of each PC axis, intermediate 1 (solid) and intermediate 10 (transparent), are compared. Comparing the two intermediates of PC 1 ( Fig. 6B and C) shows that the NT/central domains of the monomer swing in a left-to-right motion in view 1 (Fig. 6B). The continuous movement along PC 1 axis can be seen in Video S2, where all 10 intermediates are shown. PC 2 shows an orthogonal swinging where in view 2 (Fig. 6F), the NT/central domains move left to right; this corresponds to an in-and-out movement in view 1 (Fig. 6E).
The orthogonal swinging motions alter the interface with the neighboring subunits as judged by the presence of continuous density bridges between two elements in the intermediate cryoEM maps (low-passed filtered to a common resolution of 8 Å). For example, a bridge is observed between regions of the intermediate 1 cryoEM map corresponding to loop Asp426-Pro434, which is termed latch loop owing to its overlap with the latch loop in HsUGP (36), of one monomer and loop Lys320-Lys326 [320-loop; encompassing the 309 loop in HsUGP (36)] of the neighboring monomer (Fig. 6G). Importantly, the swinging motion of the monomer breaks the bridge in the map of intermediate 10 ( Fig. 6G; Video S2), indicating that the interface, termed latch-to-320loop, is dynamic. The interaction is also seen in the less processed C1 (Fig. 6G) and D4 (not shown) cryoEM maps, where the superposition of the model clearly indicates residues of the latch-and 320-loop could be involved. A similar interaction was observed in the static crystal structure of HsUGP (36), and the authors correlate it with differences in enzymatic activity between octameric UGPs, a so-called latch effect.
In intermediate maps (Fig. 6H), a continuous density bridge also forms through the partial ordering of some unmodeled NT residues of AnUGP (1-31), extending the density at the N-terminus of α1 to reach the neighboring subunit at a junction between the catalytic and CT oligomerization domain. This potential interface on the neighboring subunit would be formed by the non-conserved loop between α10 and β4, the side of helix α9, and the short loop between β14 and β15 (Fig. 6H). This contact, termed NT-to-β14/15-loop, is also readily seen in the C1 (not shown) and D4 cryoEM maps (Fig. 6H). Although local resolution of the cryoEM does not allow reliable modeling of these NT residues, fewer than 10 residues would be required to span this bridge if intrinsically unstructured (Fig. 6H). AlphaFold (64,65) indicates that in AnUGP, residues 1-23 are unstructured, while 23-32 may have some α-helical character. The alignment in Fig. 4 shows that most UGPs do have similar extensions at the NT but the sequence is not conserved. Moreover, this NT interaction has not been observed in previous crystal structures of UGPs where the first 4 to 41 NT residues can be disordered (33-36, 38, 54, 59), potentially due to crystal contacts or NT affinity tags. In summary, our analysis indicates that individual monomers experience conformational freedom inside the octameric assembly, and this variability can certainly influence interactions with neighboring monomers creating a dynamic set of intermolecular interfaces in addition to the more static "end-to-end" and "side-to-side" interfaces that are responsible for holding the monomers together in the octameric assembly (Fig. 3K).

DISCUSSION
Fungi cause diverse diseases in humans, ranging from allergic syndromes to superficial/life-threatening invasive fungal diseases, which together affect more than a billion people worldwide (66). Very recently, the World Health Organization (WHO) provided the fungal priority pathogens list, which represents the first global effort to systematically prioritize fungal pathogens, considering their research and development needs and perceived public health importance (WHO report 2022; https://www.who.int/publica tions/i/item/9789240060241). The critical group includes Cryptococcus neoformans, Candida auris, Candida albicans, and A. fumigatus. To date, only four classes of systemic antifungal medicines-azoles, echinocandins, pyrimidines, and polyenes-are used in clinical practice, and only a few others are under development. The increased incidence of drug-resistant strains of Aspergillus species, especially for azole-resistant A. fumigatus infections, represents an emerging global health threat. Under such circumstances, the design of novel antimicrobial agents with mechanisms of action different from those of existing drugs has become a worldwide priority. In that context, enzymes participating in fungal cell wall biosynthesis and remodeling are attractive drug targets due to their essential biological roles. In this study, we demonstrate that in A. nidulans, galF (FungiDB ID AN9148) is an essential gene encoding a UTP-specific UGP, referred to as AnUGP, a central enzyme in the metabolism of carbohydrates and assembly of the cell wall. Similarly, UGP was shown to be essential in S. cerevisiae (67) and the two filamentous fungal species Ganoderma lucidum (28) and Grifola frondosa (68); in all studies, downre gulation of UGP activity was associated with cell wall defects and abnormal growth phenotypes.
We describe a cryoEM structure of a UGP and show that the fold of the AnUGP monomeric unit is similar to that seen in crystal structures of eukaryotic UGPs (Fig. S5). Importantly, we harness the strength of cryoEM and analyze the conformational variability of the enzyme. Namely, we employ 3DVA to describe the major modes underlying the movement of an individual subunit in the complex. This movement alters the interaction with the neighboring subunit, and we observe changes in the interface between the latch loop, the 320-loop, and the SB (sugar-binding)-loop (Thr287-Thr295), as well as an interface between the NT residues and the neighboring subunit near the junction of the catalytic and CT domains. The identified active site, oligomerization, and dynamic interfaces (latch loop, the 320-loop, and the SB-loop) are critically important for the structure and enzymatic activity of UGPs (Fig. 7). Interestingly, the active site and oligome rization interfaces are more highly conserved, while the dynamic interfaces show more sequence variation (Fig. S7). The study of the functional significance and sequence divergence in the dynamic interface could provide the opportunity to design organism-specific regulators/inhibitors of UGPs.

The catalytic mechanism of AnUGP
Based on the sequence and structural conservation of AnUGP with other eukaryotic UGPs ( Fig. 4 and 5), we propose the UGP activity of AnUGP follows the mechanism outlined in Fig. S6. The catalytic mechanism of LmUGP was studied in detail, revealing that the basic residues, corresponding to AnUGP Lys135, His228, and Lys403 (Fig. 5D), play crucial roles in catalysis (60). In mutagenesis studies, the LmUGP and HsUGP residues corresponding to Lys135 and Lys403 were essential for activity, whereas mutation of the homologous residue of His228 rendered LmUGP almost inactive and HsUGP insoluble (Table S3A) (54,59). Assuming that eukaryotic UGPs follow a similar mechanism, AnUGP Lys135 would form salt bridges with the β-and γ-phosphates of the substrate UTP, provide electrostatic stabilization of the transition state, and coordinate the byproduct PPi after the reaction (60). His228 likely coordinates the phosphate group of Glc-1-P before and during the reaction and the β-phosphate group of the product UDP-Glc (60). In HsUGP, the sidechain of the corresponding histidine is rotated by 90° between the unbound and the UDP-Glc bound states, which is necessary to allow the formation of an H-bond with β-phosphate (54). Lys403 is expected to coordinate both the Glc-1-P phosphate and UTP α-phosphate, provide electrostatic stabilization of the transition state, and form hydrogen bonds to the α-and β-phosphates of the product UDP-Glc (60). The AnUGP active site residue Asp258 is strictly conserved not only in eukaryotic but also in bacterial UGPs, where it was shown to be involved in coordination of a Mg 2+ ion located between the UDP-Glc phosphate groups (57,61). In LmUGP, the Mg 2+ ion was shown to facilitate the S N 2 nucleophilic attack of Glc-1-P on the α-phosphate group of UTP and cleavage of the PPi leaving group by increasing the polarization and electro static attraction of the reacting atoms as well as favorably affecting the local geometry of the reaction center (60). Mutation of the corresponding residue in HsUGP yielded an inactive enzyme, although the crystal structure of the HsUGP-UDP-Glc complex does not contain a Mg 2+ ion, and the aspartate appears to coordinate the ribose moiety of the product (54). Accordingly, AnUGP Asp258 is likely to be enzymatically important and play a role in Mg 2+ coordination and/or ribose binding.
In contrast to these residues, whose role in catalysis is strongly supported by mutation and structural studies in diverse systems including related octameric UGPs, other residues like AnUGP Lys292 and Glu307 have more uncertain roles since their counterparts have been primarily studied in monomeric UGPs. AnUGP Lys292, which is located within the SB-loop (Thr287-Thr295; Fig. 4), may play  (see Table S3 for more extensive data) and the corresponding residues shown on the AnUGP structure, highlighting (A) the active site, (B) the oligomerization interface, and (C) the dynamic latch-to-320-loop interface. The residues mentioned in the table are drawn with a sphere representation and colored according to percent identity (based on alignment in Fig. 4). The high conservation of these regions, in particular the active site and oligomerization interface, suggests that their significance is maintained in fungi (Fig. S7). In the table, numbering corresponds to the long HsUGP isoform (HsUGP*; 508 AA) or the short isoform (HsUGP # ; 497 AA). All activities are given as percentage of wild type, and fw corresponds to the forward reaction (UDP-Glc formation) and rev to the reverse reaction. The superscript 1 indicates that the HsUGP latch loop was exchanged for that of ScUGP.
Research Article mBio a role in coordinating the UDP-Glc β-phosphate, like its homologous residue in AtUGP (35). However, this is inconclusive as in the AnUGP cryoEM map, the sidechain of Lys292 is not well resolved, similar to the situation in apo HsUGP and ScUGP structures (33,36), mostly likely due to the absence of bound substrate or product. In contrast, in the HsUGP-UDP-Glc complex, the sidechain corresponding to Lys292 points toward UDP-Glc but is not oriented correctly to interact with it (54). In Solanum tuberosum UGP, a glutamine mutant of the residue correspond ing to AnUGP Lys292 displayed significantly decreased V max values and increased K m values for Glc-1-P and PPi (69). Similarly, the LmUGP residue corresponding to AnUGP Glu307 plays a primary role in binding of the Glc moiety, and conse quently, its mutation caused a near-complete loss of activity and increased the K m for Glc-1-P but not for UTP (60). Likewise, the homologous residue in AtUGP (Glu271) was found to coordinate the Glc moiety (35). However, in the HsUGP-UDP-Glc complex, Glu306 (corresponding to AnUGP Glu307) appears to be located too far away to interact with the sugar ring (54). Similarly, AnUGP Glu307 is placed 6 Å away from the nearest Glc atom when UDP-Glc is modeled in the binding pocket (Fig. 5D). It is therefore possible that the crucial role of these residues in Glc binding is limited to monomeric UGPs.

(Inter)lock mechanism
In monomeric LmUGP and AtUGP, UTP binding triggers the closing of the NB-loop (Fig. 4) on the nucleotide portion of the substrate. In line with the sequential ordered mechanisms, the SB-loop is found in a closed conformation only once the sugar-binding site is occupied (35,59,60). In AtUGP, these local conformational changes are coupled to a displacement of the entire CT β-helix domain toward the catalytic domain (35). In LmUGP, the global conformational changes are even more pronounced and enable the formation of hydrogen bonds between SB-loop residues Glu251 and Arg443 (located in the CT β-helix domain of the same subunit), which are more than 11 Å apart in the apo state. This contact stabilizes the SB-loop in the closed conformation of the enzyme and was termed "lock mechanism" (60). The structural rearrangements observed in the active site of octameric HsUGP upon UDP-Glc binding resemble those in monomeric LmUGP and AtUGP, but with smaller amplitude, and are not coupled to significant overall movements of the domains (54). Instead, in the product-bound state of the octameric HsUGP, the side-to-side dimer and octamer, respectively, are 6.3% and 14% more compact than in the apo form, and a mutual intermolecular interaction between SB-loop residues Arg287 (corresponding to LmUGP "lock residue" Glu251) and Asp456 from the CT domain of its "side-to-side" neighboring subunit was observed (54). This interaction was absent or weaker in apo HsUGP (36), suggesting that it is reinforced in substrate/product-bound states. Mutation of both involved residues in HsUGP did not affect oligomerization but caused a dramatic drop in activity (Table S3D), confirming the functional importance of this interaction (54), which was therefore termed "interlock mechanism" in reference to LmUGP. The two interlock residues are strictly or functionally co-conserved in octameric fungal and animal UGPs, but not in monomeric protozoan and plant UGPs (Fig. 4). In AnUGP, the corresponding positions are occupied by Lys288 (SB-loop) and Asp462 (LβH domain), respectively, and could engage in the same type of interaction, although density for the sidechains in this region is weak and interaction cannot be unequivocally ascertained.

Latch mechanism
Another set of intermolecular interactions within UGP octamers, with the potential to influence enzymatic function, was proposed by Yu and Zheng: in HsUGP, a so-called latch loop from one subunit is located between the SB-loop and a small adjacent loop (termed "309 loop") of its side-to-side connected subunit (36). According to the authors, the latch loop of HsUGP may sterically interfere with conformational changes associated with the catalytic cycle, and ultimately limit HsUGP activity. The authors argue that in ScUGP, this latch effect is not observed due to differences in the amino acid sequence of the latch loop, resulting in a higher activity of ScUGP compared to HsUGP. We hypothesize that at least the interaction between the latch-and 320-loops (the equivalent of the HsUGP "309-loop") is maintained in AnUGP but dependent on conformational changes within the complex. Namely, we describe two major modes to this variability ( Fig. 6; Video S2) that allow the UGP subunit to swing in orthogonal directions and change the interface with neighboring subunits. This changing interface imparts a dynamic nature to the latch mechanism, namely in AnUGP, the latch-to-320-loop interface has the potential to be formed or broken depending on the swinging movement of individual UGP subunits in the octameric complex ( Fig. 6; Video S2). Moreover, this interface is in proximity to the potential AnUGP interlock residue Lys288 (Fig. 6G) opening up the possibility that the aforementioned interlock mechanism and the latch mechanism are intertwined and dependent on conformational changes in the octameric UGP. As Yu and Zheng demonstrated that mutations in the HsUGP latch and 309-loops can affect the enzymatic activity of HsUGP (Table S3C) (36), we propose that the dynamic nature of the interaction would similarly alter activity, potentially by stabilizing/destabilizing specific conformations of the UGP-Glc binding pocket.

Role of the N-terminus
Across eukaryotic UGPs, the N-terminus shows the most diversity in terms of length and sequence, even between species as closely related as A. fumigatus and A. nidulans (Fig. 4). In crystal structures, this region forms a two-helix bundle, but typically the first 4 to 41 residues are poorly resolved and not traced. Our analysis, however, indicates that conformational changes in the monomer can lead to the partial ordering of the NT residues to make an interface, NT-to-β14/15-loop, with the neighboring monomer ( Fig. 6H and Video S2). In line with its exposed position at the periphery of the octa meric complex, the UGP NT domain has been proposed to be the subject of regulatory mechanisms (70,71). In S. cerevisiae, changes in environmental and nutrient conditions cause PAS kinase to phosphorylate ScUGP at Ser11 (AnUGP Ser25), which targets the enzyme to the cell periphery, where its product UDP-Glc can be preferentially utilized for cell wall glucan synthesis (70,71). HsUGP exists in two isoforms encoded by the same gene, with the shorter isoform 2 being N-terminally truncated by 11 amino acids and predominantly expressed in the brain (72). This tissue-specific expression may imply that the two isoforms exhibit functional differences and/or different regulatory potential. For example, phosphorylation could serve as a means of differential regulation of the two isoforms, as an abundance of studies detected phosphorylation of Ser13 of HsUGP isoform 1, but not of the corresponding residue Ser2 of HsUGP isoform 2 (phospho site.org entries "UGP2" and "UGP2 iso2, " respectively; accessed 10/28/2022). Mammalian UGPs are furthermore frequently phosphorylated at Tyr89, Tyr186, Tyr298, and Ser448 (HsUGP isoform 1 numbering), which, except for HsUGP Tyr186, are conserved in AnUGP (Tyr101, Tyr299, and Ser454, respectively). While no phosphorylation of AnUGP was observed in a recent phosphoproteomic study (personal communication), this does not exclude that under certain physiological conditions, phosphorylation or other types of posttranslational modifications may occur at the exposed NT domain or elsewhere in the protein and affect, for example, AnUGP function, localization, or interaction with other factors.
In conclusion, the AnUGP structure reveals extensive interconnectivity between the subunits, namely each subunit shares interfaces with four of the other seven subunits in the complex (Fig. 3K). There are relatively stable interfaces like the "end-to-end" and "side-to-side" interfaces (A-H and A-E; Fig. 3K), involving the LβH. At the same time, there are a dynamic set of interfaces (NT-to-β14/15-loop, latch-to-320-loop; Fig. 6G and H; Fig. 3K stars) whose formation is dependent on the swinging motion of the subunit within the octameric complex (Video S2). Interestingly, these two dynamic interfaces sit at the opposite ends of strand β15 and, thus, together, could influence AnUGP activity via the 320-loop, which is close in primary sequence to the sugar-binding residues Asn329 and Asn331. The higher sequence divergence in these interfaces (Fig. 4) and their potential to allosterically influence enzymatic activity (i.e., interlock, 309-loop, and latch loop mutants; Fig. 7) could make these regions suitable targets for future development of allosteric UGP inhibitors that target sites located remotely from the conserved active site. The potential for selective allosteric inhibition of UGPs is evidenced by the identification of a small molecule targeting the monomeric LmUGP (37). Increasing our understanding of allosteric mechanisms through structural studies like those presented here is opening doors for rational allosteric inhibitor design (73).

AnUGP isolation and purification
Native AnUGP (UniProt ID Q5I6D1; FungiDB ID AN9148) was co-purified from A. nidulans owing to its inherent ability to bind Ni-affinity columns (33). Specifically, A. nidulans was grown in yeast-extract-sucrose broth (2% w/v yeast extract, 6% w/v sucrose, pH 5.8) supplemented with Hunter's trace elements (https://www.fgsc.net/methods/ anidmed.html) at 30°C (150 rpm) for 4 d. The biomass was subsequently collected by filtration through Miracloth, dried by blotting with paper towels, and stored at −80°C. To lyse the cells, the pellet was first ground in a mortar (under liquid nitrogen), then the paste was resuspended in buffer A (50 mM Tris, 200 mM NaCl, 2 mM β-mercaptoe thanol; pH 7.8) supplemented with DNAse and sonicated. The lysate was cleared by centrifugation (JA 25.50 rotor, 20,000 rpm, 20 min), filtered through a 0.22 µm syringe filter, and loaded to a HisTrap HP 1 mL column equilibrated in buffer A. The protein was eluted early in a buffer B gradient (50 mM Tris, 200 mM NaCl, 500 mM imidazole, 2 mM β-mercaptoethanol; pH = 7.8). The collected fractions were pooled, and a 12% SDS-PAGE gel showed the presence of a ~55 kDa protein and a larger ~130 kDa band. These bands were cut from the gel and analyzed by mass spectrometry (CIC bioGUNE Bilbao, Spain), which identified the 55 kDa protein as AnUGP (UniProt ID Q5I6D1) and the larger band as pyruvate carboxylase. For cryoEM, the pooled fractions from the HisTrap column were loaded to a Superose 6 GL 10/300 equilibrated in buffer A ( Fig. S1A and B), and a peak was observed in the elution profile around 14.5 mL which corresponds to an apparent molecular weight of 490 kDa (expected: 460 kDa for an AnUGP homoctamer). The fractions corresponding to this peak were stored (−80°C) in buffer A at a concentration of 0.18 mg/mL. Initial screening at the CIC bioGUNE facility indicated that on the grid, the protein was of sufficient purity, monodispersed, and largely octameric, consistent with the oligomeric state of AnUGP, strongly suggesting the co-isolated pyruvate carboxylase did not readily adhere to the grids during vitrification and/or is monomeric and not easily seen owing to its smaller size (Fig. S1C).

MG phosphate assay
The enzymatic activity of AnUGP was measured using the MG phosphate assay (74). AnUGP catalyzes the condensation reaction between UTP and Glc-1-P to produce UDP-Glc and PPi, in the presence of the divalent metal cation Mg 2+ . This reaction is coupled to the conversion of PPi into orthophosphate (Pi), mediated by the inorganic pyrophosphatase from Saccharomyces cerevisiae (ScPPase). The green complex formed between Pi, molybdate, and the MG dye is photometrically quantified. The assay reactions are as follows: The reaction [1] is stopped by adding the chelating agent EDTA. The MG reaction [4] is stabilized by the addition of sodium citrate to avoid product precipitation. The samples contained 0. 5

Generation of recombinant strains
The DNA construct for the generation of the A. nidulans null mutant of galF was synthesized by fusing three PCR amplicons (75). First, a 1.5 Kb fragment correspond ing to the promoter region of galF was amplified using oligonucleotides galF-PP1 (GTCCAAGCTGGCGCTCTGGC) and galF-PP2 (CATTGTGTGTGTTTTTGATGTGTTGTATGCG), and wild-type genomic DNA as template. Second, the selection marker pyrG Afum of A. fumigatus was amplified using oligonucleotides galF-SMP1 (CGCATACAACACATCAAA AACACACACAATGACCGGTCGCCTCAAACAATGCTCT) and galF-GFP2 (GCTTCAGTGGCC AATTAATGCTCGAGGTCTGAGAGGAGGCACTGATGCG), using plasmidic DNA as a template (75). Third, a 1.5 Kb fragment corresponding to the 3´-UTR region of galF was amplified with oligonucleotides galF-GSP3 (CTCGAGCATTAATTGGCCACTGAAGC) and galF-GSP4 (GAGGGAACACAGCACGTGCC), and genomic DNA as template. The fusion-PCR reac tion was carried out with oligonucleotides galF-PP1 and galF-GSP4, and the correct generation of the DNA construct was verified by agarose electrophoresis. A high-fidelity DNA polymerase was used in these PCR reactions (PrimeStar, Takara Bio Inc, Kusatsu, Shiga, Japan). Protoplasts of A. nidulans strain TN02A3 (genotype: pyrG89; argB2; pyroA4, ΔnkuA::argB; veA1) (26) were obtained and transformed based on the procedures developed by Tilburn and colleagues, as well as Szewczyk and colleagues (76,77). Transformants were selected on a regeneration medium (RMM), a minimal medium (AMM; see below) supplemented with 1 M sucrose. This medium lacked uridine and uracil (the nutrients associated to the marker pyrG Afum ) to enable growth of only transformants. In general, strains were cultivated in AMM (78,79) adequately supplemen ted according to their genetic markers. Glc (2%) or Gal (1%) and ammonium tartrate (5 mM) were used as sources of carbon and nitrogen, respectively.

Diagnostic PCR reactions
Adequately supplemented liquid AMM was inoculated with 10 6 conidia mL −1 . Strains were cultured for 18 h at 37°C and 150 rpm. Mycelia were filtered using Miracloth paper and lyophilized overnight. Samples were then homogenized with 5 mm beads (Bullet Blender CE; Next Advance, Inc, Troy, NY), and DNA fractions were extracted using the phenol:chloroform:isoamyl alcohol (25:24:1) procedure (80). To verify the correct integration of the DNA constructs and the homokaryotic, heterokaryotic, or diploid nature of the recombinant strains generated, diagnostic PCR reactions were run using genomic DNA samples as template and three oligonucleotide pairs. A first set of PCR reactions was run using primers galF-sPP1 (GTGGCTCCCTGACAGGCTGG) and galF-sGSP4 (CATAGCTCACTATGCTGCTGCTGG). These two oligonucleotides correspond to 5´-and 3´-UTR regions located upstream and downstream of primers galF-PP1 and galF-GSP4, respectively, which were used to generate by fusion-PCR the transformation cassette (see previous section). The amplicons that resulted from wild-type or transformant DNA samples could not be differentiated due to the similar length of galF (1.98 Kb) and the pyrG Afum fragment (1.90 Kb) and, thus, this reaction was used as control (not shown). A new set of PCR reactions was then run with primer pairs galF-sPP1 (located approxi mately 125 nucleotides upstream of galF-PP1) and galF-insideRP (GCTCGATCTGACGGA CGGAC), or galF-sPP1 and pyrG-insideRP (GCCCGTAGCCAGCGATCC).

Sample vitrification and electron microscopy
Screening the sample for cryoEM suitability was done using facilities at the CIC bioGUNE (Bilbao, Spain) and initially imaged at the CNB-CSIC (Madrid, Spain; Instruct PID 3796). AnUGP in buffer A (0.18 mg/mL) was vitrified using a Vitrobot (FEI) onto Quantifiol R1.2/1.3 (Cu 300) grids. Automated high-resolution data acquisition (EPU software, V2.8.1, Thermo Fisher Scientific, Waltham, MA) was performed at NeCEN (Netherlands Centre for Electron Nanoscopy, Instruct -PID14910) with a Titan Krios microscope (Thermo Fisher Scientific) at 300 kV equipped with a K3 direct electron detector (Table  S1). Eight thousand one hundred thirty-five movies were collected, with each movie containing 50 frames at a pixel size of 0.545 Å (super-resolution) and a total dose of 60 e -/Å 2 . More detailed imaging conditions are presented in Table S1.

Image processing and structure determination
All image processing steps were performed within the CryoSPARC and CryoSPARC Live software package (81) apart from particle picking, which was done in RELION-3.1 (82) using crYOLO (83). Movies were imported to CryoSPARC, patch motion corrected, while binning by a factor of 2 to the physical pixel size (1.09 Å/pixel), and the con trast transfer function estimated. Subsequently, the patch-aligned micrographs were imported into RELION and particle picking was performed with crYOLO, yielding a total of 451,301 projection images. The particle coordinates were reimported into CryoSPARC and micrographs were manually curated based on estimated resolution, defocus values, and motion parameters. The particles were windowed out and downsampled to a box size of 128 pixels × 128 pixels (2.18 Å/pixel) for initial cleaning (Fig. S3A and  B). Several rounds of 2D classification were performed and in each round, the best classes resembling an octameric AnUGP were selected, finally corresponding to 204,062 particles. The selected 2D classes (Fig. S3C) were rebalanced, and these particles (81,156) were used for ab initio reconstruction (one class; Fig. S3D). Subsequently, the full set of particles (204,062) were refined, using C1 symmetry, to generate a 4.9Å cryoEM map (Fig. S3E). Classification at this stage did not indicate contamination with other particles like the pyruvate carboxylase. Phenix.map_symmetry (84) showed that the C1 map displayed D4 symmetry and, therefore, a non-uniform refinement was run while applying D4 symmetry resulting in a 4.5 Å structure (Nyquist is 4.38 Å). Note that refinements employing lower-order symmetry did not improve the interpretability or resolution of the map as compared to D4. Because the D4 refinements were converg ing to near the Nyquist frequency, the particles were re-extracted and recentered to their full pixel size (256 pixels × 256 pixels, 1.09 Å/px). Heterogenous refinement (three classes) was used to clean out poorly aligning, junky, or broken particles. The best two classes were joined and used in a non-uniform refinement to generate a cryoEM map at 3.98 Å resolution ( Fig. S3F and G). This map showed obvious signs of flexibility in the subunits that break the D4 symmetry, so we employed symmetry expansion, followed by signal subtraction and a local restrained refinement (5° gaussian prior over rotation and 3 Å gaussian prior over shifts) to generate a map for a single AnUGP monomer ( Fig. S3H and I). Sharpening was performed with DeepEMhancer using the non-uniform refinement half maps generated in cryoSPARC (85). Local resolution was estimated with CryoSPARC using the non-uniform refinement half maps (86). Conformational variability was performed by downsampling the symmetry-expanded particles (before subtracting and local alignment) to a box size of 128 pixels × 128 pixels (2.18 Å/px). 3D classification without alignment (10 classes) was first used to clean the data set further and remove poorly aligned or broken particles. The best particle images (nine classes) were joined and subjected to 3DVA in CryoSPARC, solving for 10 PCs. The first two components yielded easily interpretable conformation changes and therefore were used to generate intermediate maps for analysis.

Model building
The AlphaFold tool in ChimeraX (87) was used to generate an initial model from the AnUGP sequence (UniProt ID Q5I6D1). The resulting model was imported into Isolde (88) and refined using interactive molecular-dynamics flexible fitting into the unsharpened D4 map, the sharpened D4 map, and the locally refined sharpened map. Distance and torsion restraints were applied using the AlphaFold model, the ScUGP model, and the AfUGP model as references. The resulting model was subsequently subjected to Real Space Refinement in Phenix (89) using the parameters output from Isolde (isolde write phenixRsrInput) and the sharpened D4 non-uniform refinement map. Statistics for this model were calculated using phenix.validation_cryoEM and the sharpened D4 non-uniform refinement map and corresponding half maps (Table S2).

DATA AVAILABILITY
The cryoEM movies, cryoEM maps, and atomic coordinates have been deposited with the Protein Data Bank (PDB), EMDB, and EMPIAR accession codes 8C0B, 16357, and EMPIAR-11471. Data are available from the corresponding authors upon reasonable request.

ADDITIONAL FILES
The following material is available online.  Figure 2A. The grey regions of the map are density that is unmodeled (e.g., NT residues). Video S2 (mBio00414-23-S0003.mp4). Conformational flexibility in AnUGP. The movie shows the sequence of intermediate reconstructions (intermediates 1-10) for each principal component from two orthogonal views. The analyzed monomer is colored according to the domain structure defined in Figure 2A, while the remaining seven monomers are grey.