Skip to main content
NCBI home page
Protein Science : A Publication of the Protein Society logoLink to Protein Science : A Publication of the Protein Society
. 2023 Jun 1;32(6):e4648. doi: 10.1002/pro.4648

A metal‐dependent conformational change provides a structural basis for the inhibition of CTP synthase by gemcitabine‐5′‐triphosphate

Matthew J McLeod 1,4, Norman Tran 1, Gregory D McCluskey 2,5, Tom D Gillis 2, Stephen L Bearne 2,3,✉, Todd Holyoak 1,✉
PMCID: PMC10182726  PMID: 37106216

Abstract

CTP synthases (CTPS) catalyze the de novo production of CTP using UTP, ATP, and l‐glutamine with the anticancer drug metabolite gemcitabine‐5′‐triphosphate (dF‐dCTP) being one of its most potent nucleotide inhibitors. To delineate the structural origins of this inhibition, we solved the structures of Escherichia coli CTPS (ecCTPS) in complex with CTP (2.0 Å), 2′‐ribo‐F‐dCTP (2.0 Å), 2′‐arabino‐F‐CTP (2.4 Å), dF‐dCTP (2.3 Å), dF‐dCTP and ADP (2.1 Å), and dF‐dCTP and ATP (2.1 Å). These structures revealed that the increased binding affinities observed for inhibitors bearing the 2′‐F‐arabino group (dF‐dCTP and F‐araCTP), relative to CTP and F‐dCTP, arise from interactions between the inhibitor's fluorine atom exploiting a conserved hydrophobic pocket formed by F227 and an interdigitating loop from an adjacent subunit (Q114‐V115‐I116). Intriguingly, crystal structures of ecCTPS•dF‐dCTP complexes in the presence of select monovalent and divalent cations demonstrated that the in crystallo tetrameric assembly of wild‐type ecCTPS was induced into a conformation similar to inhibitory ecCTPS filaments solely through the binding of Na+‐, Mg2+‐, or Mn2+•dF‐dCTP. However, in the presence of potassium, the dF‐dCTP‐bound structure is demetalated and in the low‐affinity, non‐filamentous conformation, like the conformation seen when bound to CTP and the other nucleotide analogues. Additionally, CTP can also induce the filament‐competent conformation linked to high‐affinity dF‐dCTP binding in the presence of high concentrations of Mg2+. This metal‐dependent, compacted CTP pocket conformation therefore furnishes the binding environment responsible for the tight binding of dF‐dCTP and provides insights for further inhibitor design.

Keywords: CTP synthase, filament‐competent conformation, gemcitabine‐5′‐triphosphate binding, inhibition, metal–nucleotide complex structure

1. INTRODUCTION

Cytidine‐5′‐triphosphate (CTP) synthase (CTPS) catalyzes the de novo production of CTP which, in turn, regulates cellular CTP levels (Hatse et al., 1999). The enzyme utilizes uridine‐5′‐triphosphate (UTP), adenosine‐5′‐triphosphate (ATP), and l‐glutamine‐derived (nascent) ammonia as substrates to produce CTP; however in vitro, free ammonia can be directly used to replace the glutamine‐derived species (Zalkin, 1985). Mechanistically, CTP synthesis occurs via phosphoryl transfer of the γ‐phosphate of ATP to the 4‐position of UTP to yield an activated 4‐phospho‐UTP intermediate (von der Saal et al., 1985; von der Saal & Villafranca, 1985), for which structural information has recently become available (Zhou et al., 2021). This intermediate subsequently undergoes nucleophilic attack by ammonia to produce CTP and inorganic phosphate (Levitzki & Koshland, 1971; Figure 1a).

FIGURE 1.

FIGURE 1

Open in a new tab

Reactions catalyzed by CTPS (a) and a space‐filling representation of the ecCTPS tetramer (b) showing the location of bound ADP (yellow) and CTP (magenta) (Endrizzi et al., 2005).

CTPS is tightly regulated by various nucleotides owing to its central role in nucleotide metabolism and homeostasis. Guanosine‐5′‐triphosphate (GTP) acts as an allosteric activator at low concentrations (<150 μM) and as an inhibitor of NH3‐dependent CTP synthesis at high concentrations (>150 μM) (Bearne et al., 2022; Levitzki & Koshland, 1972a; MacDonnell et al., 2004). Furthermore, CTP, UTP, and other pyrimidine analogues regulate CTPS activity through binding at the amidoligase active site and inducing changes in the quaternary state of the enzyme (Anderson, 1983; Levitzki & Koshland, 1972b). The apo form of CTPS from Escherichia coli (ecCTPS) typically exists as an inactive dimer in dilute solution, which transitions to an active tetramer upon binding the nucleotide substrates ATP and UTP (Levitzki & Koshland, 1972b). However, CTP induces formation of inactive tetramers (Pappas et al., 1998) that can further polymerize to generate inhibitory filaments. UTP can depolymerize the enzyme to active tetramers by competing for binding at the CTP site (Barry et al., 2014). In contrast, the two human isozymes of CTPS (hCTPS1 and hCTPS2) form enzymatically active filaments in the presence of the substrate UTP (Lynch & Kollman, 2020). However, hCTPS2 retains the ability to also form conformationally distinct inhibited filaments upon binding CTP (Lynch & Kollman, 2020). Recent cryo‐EM studies have afforded insights into the global conformational changes that must occur for ecCTPS, the two human variants of CTPS, and CTPS from Drosophila melanogaster (DmCTPS) to transition between the tetrameric and filamentous states (Lynch et al., 2017; Lynch & Kollman, 2020; Zhou et al., 2019, 2021). In general, these global changes involve stacking of tetramers that is facilitated by interaction of α‐helical secondary structure elements (i.e., α‐helix 274–284 of the linker segment between the glutamine amidotransferase [GAT] and amidoligase domains, and α‐helix 330–336 of the GAT domain of ecCTPS) (Endrizzi et al., 2005; Lynch et al., 2017).

The crystal structures of ecCTPS in its unliganded form and with both CTP and ADP bound (Endrizzi et al., 2004; Endrizzi et al., 2005), in conjunction with the body of in vitro biochemical data, suggested that the CTP‐ and UTP‐binding sites partially overlap via a common triphosphate‐binding pocket, while the pyrimidine rings of the respective nucleotides are bound at unique sites. Both binding sites were wedged between the A‐B‐A′ interface of the tetramer (Figure 1b) (Endrizzi et al., 2005). This initial hypothesis regarding the partially overlapping nature of the two nucleotide binding sites was recently confirmed for the structure of hCTPS1 in complex with UTP and DmCTPS in complex with UTP, ATP, GTP, and 6‐diazo‐5‐oxo‐l‐norleucine using cryo‐EM (Lynch et al., 2017; Zhou et al., 2021).

Due to its essential role in maintaining the intracellular CTP pool, CTPS has been long recognized as a therapeutic target for developing drugs directed at protozoan parasites (Fijolek et al., 2007; Hofer et al., 2001; Lim et al., 1996), bacterial (Mori et al., 2015; Yoshida et al., 2012) and viral infections (Andrei & de Clercq, 1993; de Clercq, 1993; de Clercq et al., 1991; Kang et al., 1989), and cancers (McPartland et al., 1974; Toschi et al., 2005). One notable compound, gemcitabine‐5′‐triphosphate (2′,2′‐difluoro‐5′‐deoxycytidine‐5′‐triphosphate, dF‐dCTP, Table 1) has long been utilized as a therapeutic agent against solid tumors (Heinemann et al., 1995; Toschi et al., 2005). Inhibition studies revealed that dF‐dCTP is a potent inhibitor of ecCTPS (IC50 ~ 1.2 μM), binding with an apparent affinity that was approximately 80‐fold greater than that of CTP (Table 1; McCluskey et al., 2016) and analysis of mono‐fluorinated compounds suggested that the 2′‐F‐arabino group made the greatest contribution to the binding affinity (McCluskey et al., 2016). While these data suggest that dF‐dCTP may be a suitable starting point for the development of novel selective inhibitors of CTPS, structural data accounting for this potent inhibition are required to inform further inhibitor development. Consequently, we undertook the structural characterization of the ecCTPS complexes with dF‐dCTP and other 2′‐monofluorinated CTP analogues to determine the structural basis for the strong binding affinity of dF‐dCTP. Herein, we demonstrate that this strong binding affinity arises through dF‐dCTP binding being associated with a conformational change in wild‐type ecCTPS to the closed, filament‐forming state that had only previously been observed for the C268A ecCTPS variant in complex with CTP (Lynch et al., 2017). Interestingly, we demonstrate that inducing this closed state is dependent upon the formation of a metal–nucleotide complex that is not unique to dF‐dCTP but also occurs with the 2′‐F‐arabino fluorinated compound F‐araCTP as well as the CTP product. Importantly, the metal–nucleotide complex forms at lower total metal concentration with dF‐dCTP and F‐araCTP than with CTP due to the formation of a hydrophobic pocket upon closure that is exploited by the 2′‐F‐arabino group in its tight binding with ecCTPS.

TABLE 1.

IC50 values for the inhibition of Gln‐dependent CTP formation catalyzed by wild‐type and C268A ecCTPS variants by CTP and analogues.

Inhibitor Inline graphic ecCTPS variant
Wild‐type C268A
[UTP] = 0.05 mM [UTP] = 0.20 mM [UTP] = 0.05 mM [UTP] = 0.20 mM
R, R′ IC50, μM N a IC50, μM n IC50, μM n IC50, μM n
CTP H, OH (97) b ; 107 ± 4 (1.0) b ; 1.6 ± 0.1 (260) b ; 234 ± 21 (0.9) b ; 1.6 ± 0.1 131 ± 20 1.8 ± 0.5 259 ± 42 1.5 ± 0.1
Arabinocytidine‐5′‐triphosphate (araCTP) OH, H (>1000) b – (>1000) b – – – – –
2′‐fluoro‐2′‐deoxycytidine‐5′‐triphosphate (F‐dCTP) H, F (109) b (0.7) b (600) b (0.9) b – – – –

2′‐fluoro‐2′‐deoxyarabinocyti‐dine‐5′‐triphosphate

(F‐araCTP)

 F, H (35) b (1.2) b (43) b (1.0) b – – – –
gemcitabine‐5′‐triphosphate (dF‐dCTP) F, F (1.2) b ; 3.8 ± 0.2 (1.4) b ; 1.8 ± 0.2 (3.9) b ; 7.3 ± 0.3 (1.5) b ; 2.0 ± 0.1 3.2 ± 0.1 1.7 ± 0.1 9.6 ± 0.8 1.9 ± 0.2
Open in a new tab
a

The value n is the Hill coefficient as defined in Equation 1.

b

Values from McCluskey et al., 2016. All other values are from this study.

2. RESULTS AND DISCUSSION

Prior kinetic studies on the inhibition of ecCTPS by CTP analogues having fluorine substitutions at the 2′‐position of the ribose ring demonstrated a marked difference in their inhibitory potency (Table 1). In addition to dF‐dCTP being the most potent nucleotide‐based inhibitor of ecCTPS, it also induced filament formation at concentrations lower than CTP (McCluskey et al., 2016; McCluskey & Bearne, 2018). To gain further insight into the differences in binding of these inhibitors and the basis for the tight binding of dF‐dCTP and its propensity to promote filament formation, we solved the structures of the ecCTPS complexes with CTP, dF‐dCTP, dF‐dCTP with either ADP or ATP, and with other 2′‐monofluorinated CTP analogues (Table 2). All complexes were obtained under identical crystallization conditions and solved in the P21212 space group. In general, all structures were well‐modeled to the corresponding electron density; however, insufficient density was available to model the surface loop comprising residues 429–436 in all structures. While the final maps in the inter‐domain tunnel region containing H57 and Y44 suggest this region is conformationally dynamic, the modeled conformation presented in the structures here is consistent with the ammonia tunnel predominantly populating the hypothesized “constricted” state (Endrizzi et al., 2004; Endrizzi et al., 2005). The density assigned to the cysteine residues was consistent with their reduced state, as was previously observed in the ecCTPS holo crystal structure (PDB ID: 1S1M) (Endrizzi et al., 2004), but contrasts with the ecCTPS•CTP + ADP complex (PDB ID: 2 AD5) (Endrizzi et al., 2005), which possessed a disulfide bridge between C261 and C268. When modeling the nucleotide‐associated metal (or water), the CheckMyMetal server (Zheng et al., 2014), interaction distances (Table S1), crystal conditions, and anomalous Fourier difference maps were all taken into consideration. As expected, the interaction distances are shorter for the phosphate–metal coordination compared to phosphate–water interactions, which themselves are within optimal distances for H‐bonding (Table S1).

TABLE 2.

Ligation state of the various structures of ecCTPS determined in this work.

Crystal “soaking” condition PDB‐ID Occupancy of ATP site Occupancy of CTP site Metal associated with nucleotide at the CTP site Global conformational state of CTPS tetramer
2.4 M Na+‐malonate, 1 mM dF‐dCTP, 5 mM MgCl2 8FV6 – dF‐dCTP Na+ Closed
2.4 M Na+‐malonate, 1 mM dF‐dCTP, 10 mM ATP, 5 mM MgCl2 8FV7 ATP dF‐dCTP Na+ Closed
2.4 M Na+‐malonate, 1 mM dF‐dCTP, 10 mM ADP, 5 mM MgCl2 8FV8 ADP dF‐dCTP Na+ Closed
2.4 M Na+‐malonate, 1 mM F‐dCTP, 5 mM MgCl2 8FV9 – F‐dCTP – Open
2.4 M Na+‐malonate, 1 mM F‐araCTP, 5 mM MgCl2 8FVA – F‐araCTP Na+ Partially closed
2.4 M Na+‐malonate, 1 mM CTP, 5 mM MgCl2 8FVB – CTP – Open
2.4 M K+‐malonate, 1 mM dF‐dCTP, 5 mM MgCl2 8FVC – dF‐dCTP – Open
2.4 M K+‐malonate, 1 mM dF‐dCTP, 100 mM MgCl2 8FVD – dF‐dCTP Mg2+ Closed
2.4 M Na+‐malonate, 1 mM CTP, 20 mM MgCl2 8SBR – CTP – Open
2.4 M Na+‐malonate, 1 mM CTP, 100 mM MgCl2 8FVE – CTP Mg2+ Closed
Open in a new tab

Inspection of either ternary ecCTPS•dF‐dCTP+ADP (8FV8) or ecCTPS•dF‐dCTP+ATP (8FV7) complex with the binary ecCTPS•dF‐dCTP (8FV6) complex demonstrated that occupancy of the ADP/ATP‐binding site did not influence the CTP‐binding pocket (Figures S1–S3). Consequently, all structural comparisons discussed are restricted to the complexes of ecCTPS without occupancy of the ATP/ADP site.

2.1. Global conformational changes: dF‐dCTP induces a filament‐like conformation in wild‐type ecCTPS tetramers

To investigate the structural determinants of the differential inhibition of ecCTPS by CTP and its 2′‐fluorinated analogues, tetramers of ecCTPS in complex with CTP (8FVB), dF‐dCTP (8FV6), F‐araCTP (8FVA), and F‐dCTP (8FV9) were aligned by anchoring the A subunit of each structure. Superposition of these structures revealed three distinguishable global conformational states that appear to define a conformational trajectory between the normally observed “open” tetramer and the more “closed” filamentous tetramer depending on the nature of the bound CTP/fluorinated analogue (Figures 2a,b) (Lynch et al., 2017). Globally, the ecCTPS•CTP (8FVB) and ecCTPS•F‐dCTP (8FV9) complexes are superimposable and represent the normally observed “open” state. Most strikingly, however, both the global subunit conformation and the localized structure of the CTP‐binding pocket of the ecCTPS•dF‐dCTP (8FV6) complex adopt a “closed” conformation that resembles the conformation of the tetramers comprising filaments of wild‐type ecCTPS observed using cryo‐EM and the C268A ecCTPS•CTP complex crystal structure (5TKV) (Lynch et al., 2017). Previously, this “closed” tetramer conformation has only been demonstrated in the C268A variant; however, our observations demonstrate that this “closed” state can also be accessed by the wild‐type enzyme solely through the binding of dF‐dCTP at the CTP‐binding pocket. When compared to either the ecCTPS•CTP (8FVB) or ecCTPS•F‐dCTP (8FV9) complexes, the conformational change associated with the wild‐type ecCTPS•dF‐dCTP (8FV6) complex can be defined by a nearly purely rotational transformation of ~6° of the B subunit toward subunit A in the crystallographic dimer (Figure 2). The third conformational state, as represented by the ecCTPS•F‐araCTP (8FVA) complex, appears to be an intermediate along the conformational trajectory between “open” and “closed” states and is characterized by a smaller rotational component of ~2.5° but there is also a more significant translational component than that observed for the ecCTPS•dF‐dCTP (8FV6) complex.

FIGURE 2.

FIGURE 2

Open in a new tab

Superposition of the complexes of CTP (8FVB), F‐araCTP (8FVA), and dF‐dCTP (8FV6) with ecCTPS anchored at subunit A. (a) The relative positional changes between the CTP (aquamarine), F‐araCTP (orange), and dF‐dFCTP (pink) complexes after alignment while anchoring subunit A in the crystallographic dimer reveals three distinguishable global conformational states. (b) Change in viewing angle looking down at the point indicated by the arrow in (a) to highlight rotational and translational differences in the C‐terminal domain of subunit B. (c) Changes in the bound nucleotide conformations observed at the CTP‐binding site. Three ligand‐binding modes at the CTP‐binding site of the superposed (a) subunits are shown, along with the helices making up the CTP‐binding pocket. The background protein structures, and the carbon atoms of the bound nucleotide are colored as in panel (a). All other atom colorings are as in Figure 3. The bound cation has been removed from the dF‐dCTP structure for clarity.

2.2. Local conformation of the CTP‐binding site and interactions with the 5′‐triphosphate

The three distinct global conformational states observed resulted in three distinct local ligand‐binding conformations for CTP and its 2′‐fluorinated analogues (Figure 2c). However, for each complex with a given ligand, the structures of the four ligand‐bound CTP‐binding sites, residing at the intersection of three subunits of the tetramer (e.g., A‐B‐A′), were identical. The local conformation of the CTP‐binding site was the same in the presence of either CTP (Figure 3a, S4) or F‐dCTP (Figure 3b, S5), while the protein assumed two slightly different conformations in the presence of either bound F‐araCTP (Figure 3c, S6) or dF‐dCTP (Figure 3d, S7), coinciding with the movement of the three helices (A′115–130, A′152–161, and B218–228) that partially comprise the CTP‐binding pocket (Figures 2c and 3f).

FIGURE 3.

FIGURE 3

Open in a new tab

Interactions for wild‐type ecCTPS with CTP and its 2′‐fluorinated analogues and C268A ecCTPS with CTP. Changing binding interactions observed in the CTP (a, aquamarine, 8FVB), F‐dCTP (b, yellow, 8FV9), F‐araCTP (c, orange, 8FVA), dF‐dCTP (d, pink, 8FV6), and C268A‐CTP (e, firebrick red, 5TKV; Lynch et al., 2017) complexes are highlighted as black dashed lines and annotated. The blue mesh illustrates the electron density from the 2Fo–Fc map clipped to F227 or K223 (rendered at 1.7σ). Atoms are colored as green (carbon), red (oxygen), blue (nitrogen), magenta (phosphorous), fluorine (dark green), light green (sodium), and light gray (magnesium). Representations of the CTP‐binding sites from the ecCTPS•CTP (aquamarine, 8FVB) and ecCTPS•dF‐dCTP (pink, 8FV6) complexes are superposed in panel (f). The hydrophobic pocket (Q114‐I116, gray) contributing to the binding of the fluorines, the interaction between the side chain of F227 with the cytosine ring and the 2′‐F‐arabino group, the interaction of E149 with the 2′‐F‐ribo and 3′‐OH‐ribo groups, and the shortening of the interaction between N229 and R158 are shown. The movements of the helices accompanying compaction of the CTP‐binding site in the presence of dF‐dCTP are also illustrated. Interaction distances are in Å. Corresponding LigPlot+ interaction diagrams are shown in Figures S4–S7.

2.3. The 2′‐ribo substitution—Similar inhibition by CTP and F‐dCTP

We anticipated that the differences in binding affinities observed for ligands with differential fluorine substitutions at the 2′‐position of the ribose would manifest as local structural changes among those residues responsible for coordinating the altered sugar. Generally, the substitution of fluorine for the 2′‐OH‐ribo group results in few structural differences consistent with the kinetic data that the substitution had little effect on binding (Table 1) (McCluskey et al., 2016). Structurally, the effect of substitution of the 2′‐OH‐ribo group by fluorine revealed no significant changes in any of the sugar–ligand interaction distances except for the E149‐binding determinant (Table S2). When comparing the CTP (8FVB) and F‐dCTP (8FV9) complexes, there is a slightly increased distance of 0.4 Å from the 2′‐F‐ribo of F‐dCTP, when compared to the 2′‐oxygen atom of CTP (3.4 vs. 3.0 Å), which is correlated to a subtle repositioning of the ribose ring, while the positioning of the 5′‐triphosphate and cytosine moieties were essentially identical (Table S2). The subtle increase of interaction distance upon fluorine substitution is consistent with prior studies suggesting that E149 is protonated. If E149 were deprotonated, repulsion between the fluorine and the side chain carboxylate would likely result in a significant increase in distance between these two atoms and relatively weaker binding. Thus, the structures of the ecCTPS•CTP (8FVB) and ecCTPS•F‐dCTP (8FV9) complexes are consistent with prior studies and support the ability of fluorine to substitute effectively for the 2′‐ribo‐OH of CTP (Biffinger et al., 2004; McCluskey et al., 2016).

2.4. The 2′‐arabino substitution—Increased inhibition by F‐araCTP and dF‐dCTP

In contrast to the substitution of the 2′‐OH‐ribo group by fluorine in F‐dCTP, the replacement of the 2′‐H‐arabino by a fluorine atom in F‐araCTP (8FVA) and dF‐dCTP (8FV6) was found to significantly enhance inhibition of ecCTPS (Table 1) (McCluskey et al., 2016), regardless of the identity of the substituent at the 2′‐ribo position. The increase in apparent binding energy accompanying inhibition by dF‐dCTP or F‐araCTP likely arises from specific interaction(s) of the enzyme with the 2′‐F‐arabino group as the apparent H‐bond between the 2′‐F‐ribo group and E149 was similar to the same interaction in the F‐dCTP‐bound structure (Figure 3b,d). Since F‐araCTP lacks a 2′‐ribo OH or F substituent, no interaction with E149 was apparent (Figure 3c). A comparison of the complexes of ecCTPS with bound CTP (8FVB) or F‐dCTP (8FV9) to the F‐araCTP‐bound (8FVA) complex revealed that the H‐bond between S14 and the terminal oxygen of the α‐phosphate shortened by ~1.1 Å (Table S2). However, this shortening was also observed in the complex of CTP with the C268A ecCTPS variant (Figure 3e, 5TKV), which did not have a tighter association with CTP (vide supra, Table 1) (Lynch et al., 2017), suggesting that this interaction does not dramatically affect the overall binding affinity.

Interestingly, the 2′‐F‐arabino group present in both F‐araCTP (8FVA) and dF‐dCTP (8FV6) was positioned within a hydrophobic pocket formed by the highly conserved residues V115 and I116, and the methylene groups of the conserved side chain of Q114 on an interdigitating loop from the A subunit, as well as F227 from the B subunit (Figure 3f, https://www.ncbi.nlm.nih.gov/Structure/cdd/cddsrv.cgi?uid=COG0504). While F227 is disordered in both the ecCTPS•CTP (8FVB) and ecCTPS•F‐dCTP (8FV9) complexes (Figures 3a,b), both the ecCTPS•dF‐dCTP (8FV6) and ecCTPS•F‐araCTP (8FVA) complexes had clear electron density surrounding F227, a mobile element within the CTP‐binding site that is important for CTPS filament formation via interactions with the pyrimidine ring (Endrizzi et al., 2004, 2005; Lynch et al., 2017; Lynch & Kollman, 2020). In the F‐araCTP‐ (8FVA) and dF‐dCTP‐bound (8FV6) complexes (Figure 3c,d), F227 was rotated toward the cytosine ring arranged in a staggered, inclined geometry deviating by about 30° from a true parallel, staggered π–π interaction between the two rings (5.3 Å) (Zhao et al., 2015), as well as participating in a 4.3‐Å dispersion interaction with the 2′‐F‐arabino group (Figure 3f). Hence, the stabilized rotomer of F227, as well as the adjacent I116 residue (located 3.5 Å away from 2′‐F‐arabino group) and the methylene groups of the Q114 side chain, comprise a hydrophobic pocket at the CTP‐binding site that can be exploited by inhibitors bearing a 2′‐F‐arabino group through favorable dispersion and electronic interactions (Table S2). These observations are consistent with the very weak inhibition by the non‐fluorinated araCTP analogue since the presence of the polar hydroxyl group at this position would be disfavored by the hydrophobic pocket (Biffinger et al., 2004; Dalvi & Rossky, 2010; McCluskey et al., 2016).

In addition to the 2′‐F‐arabino substitution, dF‐dCTP can exploit the H‐bonding interaction between the 2′‐F‐ribo and E149, which had been shown to be a major contributor to the binding of CTP and F‐dCTP (McCluskey et al., 2016). This additive effect on binding is likely why dF‐dCTP induces a more “filament‐like” global conformation than does the binding of F‐araCTP. Specifically, both the local and global rearrangements allow for the H‐bond between E149 and the 2′‐F‐ribo substituent of dF‐dCTP (3.3 Å) to shorten modestly (~0.1 Å) compared to F‐dCTP (3.4 Å), leading to the enhanced interaction observed for dF‐dCTP (cf. Figure 3b,d) (Table S2).

While the structures of the 5′‐triphosphate‐binding site in the ecCTPS•CTP (8FVB) (Figure 3a) and ecCTPS•F‐dCTP (8FV9) (Figure 3b) complexes were similar, some changes between the 5′‐triphosphate‐binding sites in these complexes and those with the 2′‐F‐arabino analogues were observed (Figure S2). Based upon our analysis, while these subtle variations in interactions with the 5′‐triphosphate group among the four complexes were measurable, it is unlikely that these contribute substantially to different binding affinities of CTP and its 2′‐fluorinated derivatives since all ligands have a 5′‐triphosphate group in common.

2.5. Conformational change of F227 in filament formation

As discussed above, one of the most significant changes observed when comparing the structures of the “open” ecCTPS•CTP (8FVB) and ecCTPS•F‐dCTP (8FV9) complexes to the partially “closed” ecCTPS•F‐araCTP (8FVA) and “filament‐like” ecCTPS•dF‐dCTP (8FV6) complexes, respectively, was the ordering of the F227 side chain in a single rotomeric conformation presumably because of the substitution of the 2′‐H‐arabino by the 2′‐F‐arabino group. In these structures, as well as the “filament‐like” C268A ecCTPS•CTP complex (5TKV), F227 was localized and formed a π–π interaction with the cytosine ring (5.3 Å) (Figure 3f) (Lynch et al., 2017). In addition to its important role in mediating enhanced interaction with the 2′‐F‐arabino‐containing CTP derivatives described above, F227 had also been implicated in filament formation by ecCTPS since its change in location was coincident with a shift of the 218–228 helix (Figure 3f), thereby rearranging the CTP‐binding pocket and subsequently promoting polymerization (Endrizzi et al., 2004, 2005; Lynch et al., 2017; Lynch & Kollman, 2020). Previously, McCluskey and Bearne (McCluskey & Bearne, 2018) investigated the nature of this interaction using F227L and F227A variants of ecCTPS to study the role of F227 in dF‐dCTP binding and filament formation. F227A would expand the pocket, remove the π‐interaction, and slightly weaken any potential dispersion interactions, while F227L would remove π‐interactions while largely maintaining hydrophobicity. Inhibition studies on these enzyme variants revealed that inhibition by CTP was modestly increased by the alanine substitution (IC50 = 30 μM), while the leucine substitution decreased inhibition (IC50 = 340 μM) relative to wild‐type ecCTPS (IC50 = 81 μM). The inhibition of the F227 variants by dF‐dCTP was also studied; however, the substitutions had minimal effect on the observed inhibition (i.e., F227, IC50 = 1.2 μM; F227A, IC50 = 1.3 μM; and F227L, IC50 = 1.9 μM) (McCluskey & Bearne, 2018). Furthermore, examination of the ability of the F227 variants to undergo filament formation revealed that the F227A variant did not undergo polymerization, while the F227L filaments were shorter (McCluskey & Bearne, 2018). The present structural study affords insights into these observations by revealing that the mobility of F227 in the ecCTPS•CTP (8FVB) and ecCTPS•F‐dCTP (8FV9) complexes is markedly reduced when a CTP analogue bearing the 2′‐F‐arabino group is present (i.e., F‐araCTP and dF‐dCTP). By stabilizing the conformation of F227 and the 218–228 helix in the filament competent state, dF‐dCTP can promote filament formation more effectively than CTP. Further support for the notion that F227 acts as a polymerization “switch” but does not contribute to binding of CTP analogues that lack a 2′‐F‐arabino group comes from the structure of the C268A ecCTPS variant for which the tetramers preferentially assume a “filament‐like” conformation in crystallo (5TKV; Lynch et al., 2017) with the side chain of F227 in the same stabilized conformation as we observed for the ecCTPS•F‐araCTP and ecCTPS•dF‐dCTP complexes (Figure 3c,d). Interestingly, this reduction in the mobility of F227 in the C268A variant did not lead to an increase in the potency of product inhibition by CTP (Lynch et al., 2017) or inhibition by dF‐dCTP (Table 1). The data therefore suggest that the enhanced binding of analogues containing a 2′‐F‐arabino originates from the interaction between F227 and the 2′‐F‐arabino group through dispersion and electronic effects while in those lacking the substitution, a weaker edge‐on interaction with the pyrimidine base is present.

2.6. The role of the metal–nucleotide complex in forming the filament‐competent conformation

In accord with our finding that dF‐dCTP induced a “filament‐like”, “closed” conformation of wild‐type ecCTPS similar to that observed for the C268A variant with bound CTP, no significant differences in the binding modes of dF‐dCTP (8FV6) to wild‐type ecCTPS and CTP binding to C268A ecCTPS (5TKV) variant were observed (Figure 3d,e). In both cases, as well as in the ecCTPS•F‐araCTP (8FVA) complex, the nucleotides were bound as the metal–nucleotide complex, suggesting that the filament‐forming competent state is coupled to the formation of the metal–nucleotide complex since these are the only three reported structures of ecCTPS to have a cation bound to the 5′‐triphosphate moiety of their respective ligands (Lynch et al., 2017; Table S1). The current work further demonstrates that in the absence of an amino acid substitution, the 2′‐F‐arabino lowers the apparent free energy barrier for closure of the enzyme which in turn is correlated with the metal‐binding competent conformation of the nucleotide. However, the presence of fluorine at both the 2′‐arabino and 2′‐ribo positions is required for the full expression of the filament‐competent conformation since the ecCTPS•F‐araCTP (8FVA) complex does not completely shift the conformation toward the “filament‐like” end point under the crystallization conditions. Additionally, in the C268A complex with CTP, F227 underwent a transition from a disordered to ordered state thereby forming a π‐stacking interaction with the cytosine moiety of CTP as observed here for the ecCTPS•dF‐dCTP (8FV6) and ecCTPS•F‐araCTP (8FVA) structures (vide vide) (Lynch et al., 2017). Although one might anticipate that this rearrangement of the CTP‐binding site with the formation of the metal–nucleotide complex and concomitant changes in the lysine (K187, 189, 223)‐ and S14‐ligand interaction distances (vide supra) would play a role in the strong binding of dF‐dCTP, the lack of change in the potency of inhibition of the C268A variant by both CTP (Lynch et al., 2017) and dF‐dCTP (Table 1, Figures S8 and S9), relative to wild‐type ecCTPS, suggests that this is not the case (indeed, the kinetic parameters for C268A ecCTPS‐catalyzed Gln‐ and NH3‐dependent CTP formation are very similar to the wild‐type enzyme [Table S1, Figure S10] as first reported by Lynch et al., 2017).

As mentioned above, in addition to the observed “closure” of the product nucleotide pocket establishing a filament‐competent conformation of the tetramer, the binding of dF‐dCTP to ecCTPS is correlated with the formation of a metal•dF‐dCTP complex. This metal–nucleotide complex was previously observed in the closed C268A variant in complex with CTP and modeled as the Mg2+–CTP complex. In our structures, the metal coordination is subtly different and bioinformatic analysis (Zheng et al., 2014), consistent with the crystallization/cryoprotection conditions used for the complexes reported containing 5 mM MgCl2 and 2.4 M sodium malonate, supports a sodium‐bound rather than a magnesium‐bound dF‐dCTP complex (8FV6) (Figure 4). Interestingly, neither CTP nor F‐dCTP show a metal–nucleotide complex despite identical crystallization conditions. To verify that sodium is indeed complexed with dF‐dCTP (and by extrapolation, F‐araCTP), we substituted MnCl2 for MgCl2 (5 mM) and collected anomalous diffraction data. The anomalous difference Fourier maps indicated no presence of manganese thus confirming the sodium•dF‐dCTP complex. Because our observation of the monovalent cation–nucleotide complex was surprising, we cryoprotected crystals with potassium malonate with low magnesium (5 mM) and found that dF‐dCTP was demetalated and the inhibitor‐bound ecCTPS was now in the CTP‐like, “open” conformation with F227 disordered (8FVC). Subsequently, raising the concentration of MgCl2 (100 mM) while using potassium malonate recapitulated the closed conformation, now with magnesium bound to the active site (8FVD). Finally, we again substituted MnCl2 for MgCl2 (75 mM) and the anomalous difference Fourier map indicated the presence of manganese in complex with dF‐dCTP (Figure S11). Together, these observations reinforce the conclusion that the induction of the filament‐like “closed” conformation is in part due to metal association, and both sodium and magnesium/manganese metals can evoke this closure when dF‐dCTP is present. Since this metal–nucleotide coupled conformational change is present with dF‐dCTP, we wanted to examine if a similar mechanism governs other ecCTPS–nucleotide complexes, specifically CTP. Although it does not appear that sodium can bind CTP, when high concentrations of MgCl2 (100 mM, 8FVE), but not moderate concentrations (20 mM, 8SBR), were used during cryoprotection, we observed the formation of the ecCTPS•Mg‐CTP complex and the induction of the filament‐like “closed” conformation like that observed with the C268A mutant. F227, as in the “closed” conformation with dF‐dCTP, was found in the stabilized conformation, π‐stacking with CTP. This series of structures indicates that sodium binding to dF‐dCTP and F‐araCTP is likely artifactual due the high molar concentrations of Na+ in the cryoprotectant. Second, 5 mM magnesium in crystallo is insufficient to bind to any nucleotide and higher concentrations of magnesium complexing with the nucleotide, lead to a global conformational change of the enzyme to the filament competent state, which was previously only observed in the C268A•Mg‐CTP complex.

FIGURE 4.

FIGURE 4

Open in a new tab

Interactions between the 5′‐triphosphate moiety of the bound nucleotide and ecCTPS. (a) The structure of dF‐dCTP (8FV6) is shown with the 2Fo–Fc electron density rendered at 3σ for the bound nucleotide and cation illustrated as a blue mesh. Atoms are colored as in Figure 3. Except for ligands of the cation, water molecules are removed for clarity. The changing interactions between the 5′‐triphosphate portion of the nucleotide in the open and closed states of ecCTPS are shown for the (b) ecCTPS•CTP (8FVB) and (c) ecCTPS•dF‐dCTP (8FV6) complexes. Structural differences in the bound nucleotide conformation, metal ion coordination, and side‐chain interactions are illustrated. The bound cation in the ecCTPS•dF‐dCTP (8FV6) complex (c) has an octahedral coordination with one oxygen donated by each of the three phosphates and three water molecules. In panels (b) and (c), the predicted bonding interactions are shown as black dashes with lengths labeled (Å). Carbon atoms are colored as light blue/gray and other atoms are colored as described in Figure 3.

Interestingly, as noted above, the structure of the non‐metal‐bound dF‐dCTP (8FVC) complex was in the “open” conformation and lacked the interaction between F227 and the 2′‐F‐arabino group that we postulate is important in tight binding of dF‐dCTP. With the only difference between the induction of the two conformations being the presence or absence of metal, formation of the metal‐bound nucleotide complex was postulated to be the key switch that drives the transition from the “open” to the “closed”, high affinity, filamentous state. Based upon the structural data, we hypothesize that the F227•2′‐F‐arabino interaction lowers the energetic barrier for the conformational transition to the filament competent state, resulting in a bound nucleotide conformation that requires metal coordination to stabilize the subtly altered 5′‐triphosphate portion of the nucleotide co‐incident with the remodeled CTP binding pocket. In contrast, analogues lacking the ability to form favorable electronic and dispersive interactions with the F227 and relying solely on the π‐stacking between F227 and the cytosine base require higher metal concentrations to drive the conformational transition. This structure‐based hypothesis would suggest that in the absence of metal coordination to the nucleotide, dF‐dCTP potency would decrease as it would bind to the open state and lose the stabilizing F227•2′‐F‐arabino interaction. Testing this hypothesis is difficult due to the catalyzed reaction requiring millimolar concentrations of Mg for measurable activity, which limits the lower concentration of Mg to ~2 mM (Robertson & Villafranca, 1993). With that caveat, we tested the inhibition of dF‐dCTP as a function of MgCl2 concentration. Consistent with the hypothesis, at low mM Mg, the binding of dF‐dCTP was measurably weaker than at 10 mM Mg (2 mM MgCl2, IC50 = 11.5 ± 0.5 μM; 5 mM MgCl2, IC50 = 11.0 ± 0.9 μM; 10 mM MgCl2, IC50 = 7.3 ± 0.3 μM). Taken together, the structural and functional data both support a model in which the energetic barrier to transition to the tight binding, filament‐competent conformation is coupled to the presence of a metal–nucleotide complex, and this barrier is lessened due to the favorable interactions of the 2′‐F‐arabino group of dF‐dCTP with F227, ultimately leading to tighter observed binding of the dF‐dCTP analogue compared to CTP.

3. CONCLUSION

To date, dF‐dCTP is one of the most potent inhibitors of CTPS with a strong propensity for promoting filament formation by the enzyme (McCluskey et al., 2016; McCluskey & Bearne, 2018). Herein, we have delineated those molecular interactions that contribute to the high binding affinity exhibited by dF‐dCTP with ecCTPS. The 2′‐F‐ribo substitution affords little binding advantage over the 2′‐OH‐ribo group (McCluskey et al., 2016), seen in the similar global and local conformations of the ecCTPS•CTP (8FVB) and ecCTPS•F‐dCTP (8FV6) complexes. Consistent with the observation that the 2′‐F‐arabino substitution (i.e., F‐araCTP, 8FVA) affords only modestly more binding affinity relative to CTP, the loss of the H‐bond between E149 and the 2′‐ribo substituent is compensated for by interactions with the 2′‐F‐arabino substitution. Changes in other protein–ligand interaction distances arising from the 2′‐F‐arabino substitution (lysine coordination [K187, K189, and K223] and shortening of the S14‐α‐phosphate H‐bond), all of which affect the 5′‐triphosphate rather than the nucleoside moiety, appear to be inconsequential with respect to the binding affinity since the structure of the C268A ecCTPS variant with bound CTP exhibits the same structural changes yet the binding affinities of CTP (Lynch et al., 2017) and dF‐dCTP (Table 1) are not markedly altered by this substitution, relative to wild‐type ecCTPS (Table 1). Significantly, interaction with the 2′‐F‐arabino group stabilizes a conformational switch to a closed, filament competent state in which an ordered F227 side chain from the opposing subunit, combined with Q114, V115, and I116, comprise a conserved hydrophobic binding pocket that is exploited by the hydrophobicity of the 2′‐F‐arabino group, leading to the enhanced binding of analogues bearing this substitution with concomitant compaction of the CTP‐binding site (Figure 5). Importantly, the adoption of the closed, filament competent state is not a unique property of dF‐dCTP but is simply obtained at lower total metal concentrations than with CTP or the other analogues due to the coupling of the closure of the active site to stabilization of the F227 hydrophobic pocket; a conformational change that is more accessible in dF‐dCTP due to its ability to favorably interact via the 2′‐F‐arabino substitution.

FIGURE 5.

FIGURE 5

Open in a new tab

Schematic representation of the structural changes induced by dF‐dCTP binding. Relative to the binding of CTP (helices outlined in black), the binding of dF‐dCTP alters the conformation of the CTP‐binding site (solid helices) leading to compaction of the site and formation of a filament‐competent tetramer, which is consistent with the ability of dF‐dCTP to promote filament formation at a lower concentration than CTP (McCluskey & Bearne, 2018).

4. MATERIALS AND METHODS

4.1. Materials

Sodium malonate was purchased from Hampton Research. CTP and ATP were purchased from BioBasic, Inc. ADP was purchased from ChemImpex. Thrombin was purchased from BioPharm Laboratories. dF‐dCTP was purchased from Jena Biosciences. F‐dCTP and F‐araCTP were purchased from TriLink Biotechnologies. Synthetic oligodeoxynucleotide primers for sequencing and site‐directed mutagenesis were purchased from Integrated DNA Technologies. Plasmid preparations for mutagenesis and transformations were conducted by use of a QIAprep Spin Miniprep Kit from Qiagen. Commercial DNA sequencing was conducted by the Robarts Research Institute (London, ON, Canada).

4.2. Protein expression and purification for crystallography

The ecCTPS expression construct pET‐15b‐CTPS1 used for this study has been described previously (Bearne et al., 2001). The recombinant protein was produced in BL21(DE3) cells grown in LB medium (2 L) with expression induced by the addition of IPTG (1.0 mM) when the OD600 value was between 0.6 and 0.8. After an additional 3 h of growth at 37°C, cells were harvested by centrifugation and re‐suspended in Buffer 1 (B1: 25 mM HEPES, pH 7.9, 1.0 M NaCl, 20 mM imidazole). Cells were lysed by sonication on ice using five 20‐s pulses. All subsequent purification steps were conducted at 4°C. The resulting cell lysate was clarified by centrifugation (12,000 × g, 1 h, 4°C) and subsequently applied to a column containing nickel‐NTA resin pre‐equilibrated in B1 for 1 h. The column was then washed with B1 until the flow‐through reached an A280 < 0.1. The bound protein was eluted using Buffer 2 (B2: 25 mM HEPES, pH 7.9, 1 M NaCl, 250 mM imidazole) in 10‐mL fractions. These fractions were concentrated using an Amicon nitrogen gas concentrator with a 10‐kDa MWCO Millipore Ultracel© cellulose membrane. The buffer was then exchanged with Buffer 3 (B3: 25 mM HEPES, pH 7.9) by passing the concentrated protein sample through a gravity flow column containing P6‐DG resin (Bio‐Rad) pre‐equilibrated in B3. The protein was digested for 9 h using thrombin (0.5 U) to catalyze removal of the His6‐tag (longer periods of incubation led to proteolytic degradation of ecCTPS). During this cleavage step, approximately 75% of the ecCTPS precipitated. The digestion was terminated by treatment with 1 mM PMSF. The remaining digested protein was then added to fresh nickel‐NTA pre‐equilibrated in B3 and incubated for 1 h to remove the cleaved His6‐tag and undigested ecCTPS. The nickel resin slurry was applied to a gravity flow column and washed with B3. The purified ecCTPS was obtained in the flow‐through and subsequently concentrated to 15 mg/mL using an Amicon nitrogen gas concentrator with a 10‐kDa MWCO Millipore Ultracel© cellulose membrane. The final protein concentration was determined using a NanoDrop spectrophotometer and an A280,0.1% value of 6.7 as calculated using the ExPASy website (Gasteiger et al., 2005). The protein was frozen by direct immersion in liquid nitrogen in ~30‐μL drops and subsequently stored at −80°C.

4.3. Site‐directed mutagenesis

The pET‐15b‐CTPS1 plasmid (Bearne et al., 2001) containing the open reading frame encoding ecCTPS was used as the template for site‐directed mutagenesis, following a protocol similar to that described for the Quikchange kit (Agilent) but following an inverse PCR‐based method (Dominy & Andrews, 2003). Reactions were conducted using Phusion High‐Fidelity DNA Polymerase (New England Biolabs). The forward (F) and reverse (R) synthetic deoxyoligonucleotide primers used to incorporate the mutation into the open reading frame were 5′‐CTTAAACGCA CCGGAAGCGAATCTG‐3′ (F) and 5′‐CTGAATCGTTTACAAATATAATCGTC‐3′ (R), where the codon encoding residue 268 is underlined and the altered base is shown in boldface. Mutant plasmids were used to transform E. coli DH5α cells using heat shock (Green & Sambrook, 2012), which were used for plasmid maintenance. The entire plasmid open reading frame was sequenced to verify that no other mutations in the nucleotide sequence were introduced.

4.4. Expression and purification of recombinant C268A ecCTPS

E. coli BL21(DE3) cells transformed with the pET‐15b‐CTPSC268A plasmid were used to produce recombinant enzyme as described previously for wild‐type ecCTPS (Bearne et al., 2001; McCluskey et al., 2016), except that the His·Bind resin was first washed with a step‐gradient from 5 to 60 mM imidazole in 20 mM Tris‐Cl buffer, pH 7.9, containing 500 mM NaCl. The protein was then eluted using the same buffer but containing 250 mM imidazole. Protein concentrations were estimated using the Bio‐Rad protein assay (Bio‐Rad Laboratories, Mississauga, ON, Canada) with BSA standards.

4.5. Enzyme assays

The rates of ecCTPS‐catalyzed conversion of UTP into CTP and the associated kinetic parameters were determined as described previously (Bearne et al., 2001). Turnover numbers (k cat) were calculated using molecular weights for the His6‐tagged, recombinant enzyme monomers of 62,911 Da (wild‐type) and 62,878 Da (C268A) calculated from the primary sequence using the ExPASy ProtParam program (Gasteiger et al., 2005). IC50 values for the inhibition of the wild‐type and C268A ecCTPS variants by CTP or dF‐dCTP were determined by monitoring the initial rates for the conversion of UTP to CTP with Gln (6.0 mM) as the nitrogen source, together with fixed concentrations of UTP (0.05 mM or 0.20 mM) and ATP (1.0 mM), and variable concentrations of CTP (0–1000 μM) or dF‐dCTP (0–20 μM). The IC50 values were determined by fitting Equation 1 to the relative velocities (v i/v o) obtained in the presence of various concentrations of inhibitor using nonlinear regression analysis and the program KaleidaGraph v. 4.02 from Synergy Software. The IC50 value is the concentration of the inhibitor that yields v i/v o = 0.5, and n is the Hill coefficient (Copeland, 2000). All kinetic parameters were determined in triplicate, and average values are reported. The reported errors are the standard deviations.

vivo=IC50nIC50n+In (1)

4.6. Crystallization and data collection

Crystals were grown as previously described (Endrizzi et al., 2004). In brief, hanging‐drop vapor diffusion crystallization trays were set up at room temperature and subsequently transferred to an incubator where crystal growth occurred at 4°C. The crystallization mother liquor consisted of a volume of 700 μL of 0.1 M TRIS‐Cl, pH 8.0 (measured at room temperature), containing 5 mM MgCl2 and concentrations of (NH4)2SO4 varying between 1.15 and 1.40 M at a ratio of 2 μL of protein at 15 mg/mL to 2 μL of mother liquor. Crystals were cryoprotected by stepwise transfer into increasing concentrations of sodium malonate cryoprotectant (Holyoak et al., 2003). Crystals were initially placed in a solution of 1.5 M sodium malonate, pH 7.0, containing 5 mM MgCl2 and subsequently transferred to a similar solution at 1.9 M sodium malonate. Finally, crystals were transferred to a 2.4 M sodium malonate solution containing 5 mM MgCl2 and, when required, CTP and its analogues were present at a concentration of 1 mM, while ADP and ATP were present at a concentration of 10 mM. The crystals of the complexes were obtained by soaking in the malonate stabilizing solution at 4°C for 1 h with the desired nucleotide. After soaking, the crystals were cryopreserved by direct immersion into liquid N2. Data on the cryo‐cooled crystals were collected at the Canadian Light Source (Saskatoon, Saskatchewan) on the 08B1‐1 beamline or the ID72B beamline at CHESS (Ithaca, New York). The resultant data were indexed and scaled using HKL‐2000 or DIALS and final data statistics are presented in Table S4 (Otwinowski & Minor, 1997; Winter et al., 2018).

4.7. Structure determination

All structures were solved through molecular replacement using the structure reported for the C268A variant of ecCTPS (PDB ID: 5TKV; Lynch et al., 2017) and MolRep in the CCP4 package (Collaborative Computational Project 4, 1994; Vagin & Teplyakov, 1997). Iterative rounds of real space refinement and the addition of water molecules and heteroatoms were carried out with COOT, followed by refinement using REFMAC5 or phenix.refine (Afonine et al., 2012; Emsley & Cowtan, 2004; Karplus & Diederichs, 2015). Final MolProbity statistics were collected with the online server at http://molprobity.biochem.duke.edu/index.php and final model statistics are presented in Table S4 (Chen et al., 2010; Davis et al., 2007). Resolution limits were determined before substantial drop‐off in CC1/2 values in the highest resolution bin (Karplus & Diederichs, 2015). Predicted binding interactions were automatically generated using LigPlot+ along with additional manual curation (Laskowski & Swindells, 2011). Manual inspection of the potential interaction distances was performed using COOT.

AUTHOR CONTRIBUTIONS

Conceptualization: Stephen L. Bearne and Gregory D. McCluskey. Methodology: Matthew J. McLeod, Gregory D. McCluskey, Tom D. Gillis, and Todd Holyoak. Investigation: Matthew J. McLeod, Norman Tran, Tom D. Gillis, Stephen L. Bearne, and Todd Holyoak. Resources: Stephen L. Bearne, Gregory D. McCluskey, Matthew J. McLeod, Norman Tran, Tom D. Gillis, and Todd Holyoak. Writing—Original Draft: Matthew J. McLeod. Writing—Review & Editing: Matthew J. McLeod, Norman Tran, Gregory D. McCluskey, Stephen L. Bearne, and Todd Holyoak. Supervision: Stephen L. Bearne and Todd Holyoak.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest.

Supporting information

Data S1: Supporting Information.

Click here for additional data file. (5.6MB, docx)

ACKNOWLEDGMENTS

Todd Holyoak acknowledges support from the Natural Science and Engineering Research Council (NSERC) of Canada. Matthew J. McLeod was supported by NSERC through a PGS‐D scholarship. Stephen L. Bearne acknowledges support from the Canadian Institutes of Health Research (CIHR, grant no. 114895), a Scotia Support Grant from the Nova Scotia Heath Research Foundation (grant no. 140), and a grant from the Nova Scotia COVID‐19 Health Research Coalition (NSHA Research & Innovation [Nova Scotia Health], Dalhousie University, Dalhousie Medical Research Foundation, the QEII Health Sciences Centre Foundation, the IWK Foundation, the IWK Health Centre, the Dartmouth General Hospital Foundation and Research Nova Scotia). Gregory D. McCluskey was a trainee in the Cancer Research Training Program of the Beatrice Hunter Cancer Research Institute, with funds provided by Craig's Cause Pancreatic Cancer Society, a CIBC Graduate Scholarship in Medical Research, and The Terry Fox Research Institute. Tom D. Gillis was supported, in part, by a Mathematics of Information Technology and Complex Systems (MITACS) Inc. Research Training Award and the NSERC CREATE Training Program in BioActives (510963). We thank Mika Park for her assistance purifying C268A ecCTPS. Part of the research described in this paper was performed using beamline CMCF‐BM at the Canadian Light Source, a national research facility of the University of Saskatchewan, which is supported by the Canada Foundation for Innovation, NSERC, the National Research Council, (CIHR), the Government of Saskatchewan, and the University of Saskatchewan. Part of the work presented here is based upon research conducted at the Center for High Energy X‐ray Sciences (CHEXS), which is supported by the National Science Foundation under award DMR‐1829070, and the Macromolecular Diffraction at CHESS (MacCHESS) facility, which is supported by award 1‐P30‐GM124166‐01A1 from the National Institute of General Medical Sciences, National Institutes of Health, and by New York State's Empire State Development Corporation (NYSTAR).

McLeod MJ, Tran N, McCluskey GD, Gillis TD, Bearne SL, Holyoak T. A metal‐dependent conformational change provides a structural basis for the inhibition of CTP synthase by gemcitabine‐5′‐triphosphate. Protein Science. 2023;32(6):e4648. 10.1002/pro.4648

Review Editor: Jeanine Amacher

Contributor Information

Stephen L. Bearne, Email: sbearne@dal.ca.

Todd Holyoak, Email: tholyoak@uwaterloo.ca.

DATA AVAILABILITY STATEMENT

The data and coordinates for the structures presented are deposited in the Worldwide Protein Data Bank (wwPDB) under the accession codes 8FV6, 8FV7, 8FV8, 8FV9, 8FVA, 8FVB, 8FVC, 8FVD, 8SBR, and 8FVE.

REFERENCES

  1. Afonine PV, Grosse‐Kunstleve RW, Echols N, Headd JJ, Moriarty NW, Mustyakimov M, et al. Towards automated crystallographic structure refinement with phenix.refine. Acta Crystallogr D Biol Crystallogr. 2012;68(4):352–67. 10.1107/S0907444912001308 [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Anderson PM. CTP cynthetase from Escherichia coli: an improved purification procedure and characterization of hysteretic and enzyme concentration effects on kinetic properties. Biochemistry. 1983;22(13):3285–92. 10.1021/bi00282a038 [DOI] [PubMed] [Google Scholar]
  3. Andrei G, de Clercq E. Molecular approaches for the treatment of hemorrhagic fever virus infections. Antiviral Res. 1993;22(1):45–75. 10.1016/0166-3542(93)90085-W [DOI] [PubMed] [Google Scholar]
  4. Barry RM, Bitbol AF, Lorestani A, Charles EJ, Habrian CH, Hansen JM, et al. Large‐scale filament formation inhibits the activity of CTP synthetase. Elife. 2014;3:1–19. 10.7554/eLife.03638 [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Bearne SL, Guo CJ, Liu JL. GTP‐dependent regulation of CTP synthase: evolving insights into allosteric activation and NH3 translocation. Biomolecules. 2022;12(647):1–27. 10.3390/biom12050647 [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Bearne SL, Hekmat O, MacDonnell JE. Inhibition of Escherichia coli CTP synthase by glutamate γ‐semialdehyde and the role of the allosteric effector GTP in glutamine hydrolysis. Biochem J. 2001;356(1):223–32. 10.1042/0264-6021:3560223 [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Biffinger JC, Kim HW, DiMagno SG. The polar hydrophobicity of fluorinated compounds. Chembiochem. 2004;5(5):622–7. 10.1002/cbic.200300910 [DOI] [PubMed] [Google Scholar]
  8. Chen VB, Arendall WB III, Headd JJ, Keedy DA, Immormino RM, Kapral GJ, et al. MolProbity: all‐atom structure validation for macromolecular crystallography. Acta Crystallogr. 2010;66(1):12–21. 10.1107/S0907444909042073 [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Collaborative Computational Project 4 . The CCP4 suite: programs for protein crystallography. Acta Crystallogr D Biol Crystallogr. 1994;50(5):760–3. 10.1107/S0907444994003112 [DOI] [PubMed] [Google Scholar]
  10. Copeland RA. Enzymes: a practical introduction to structure, mechanism, and data analysis. 2nd ed. New York: Wiley‐VCH, Inc.; 2000. [Google Scholar]
  11. Dalvi VH, Rossky PJ. Molecular origins of fluorocarbon hydrophobicity. Proc Natl Acad Sci U S A. 2010;107(31):13603–7. 10.1073/pnas.0915169107 [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Davis IW, Leaver‐Fay A, Chen VB, Block JN, Kapral GJ, Wang X, et al. MolProbity: all‐atom contacts and structure validation for proteins and nucleic acids. Nucleic Acids Res. 2007;35:W375–83. 10.1093/nar/gkm216 [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. de Clercq E. Antiviral agents: characteristic activity spectrum depending on the molecular target with which they interact. Adv Virus Res. 1993;42:1–55. 10.1016/S0065-3527(08)60082-2 [DOI] [PubMed] [Google Scholar]
  14. de Clercq E, Murase J, Marquez VE. Broad‐spectrum antiviral and cytocidal activity of cyclopentenylcytosine, a carboxylic nucleoside targeted at CTP synthetase. Biochem Pharmacol. 1991;41(12):1821–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Dominy CN, Andrews DW. In: Casali N, Preston A, editors. E. coli plasmid vectors. Methods and applications. Volume 235. Totowa, NJ: Humana Press Inc; 2003. [Google Scholar]
  16. Emsley P, Cowtan K. Coot: model‐building tools for molecular graphics. Acta Crystallogr. 2004;60(12):2126–32. [DOI] [PubMed] [Google Scholar]
  17. Endrizzi JA, Kim H, Anderson PM, Baldwin EP. Crystal structure of Escherichia coli cytidine triphosphate synthetase, a nucleotide‐regulated glutamine amidotransferase/ATP‐dependent amidoligase fusion protein and homologue of anticancer and antiparasitic drug targets. Biochemistry. 2004;43(21):6447–63. 10.1021/bi0496945 [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Endrizzi JA, Kim H, Anderson PM, Baldwin EP. Mechanisms of product feedback regulation and drug resistance in cytidine triphosphate synthetases from the structure of a CTP‐inhibited complex. Biochemistry. 2005;44(41):13491–9. 10.1021/bi051282o [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Fijolek A, Hofer A, Thelander L. Expression, purification, characterization, and in vivo targeting of trypanosome CTP synthetase for treatment of african sleeping sickness. J Biol Chem. 2007;282(16):11858–65. 10.1074/jbc.M611580200 [DOI] [PubMed] [Google Scholar]
  20. Gasteiger E, Hoogland C, Gattiker A, Duvaud S, Wilkins MR, Appel RD, et al. Protein identification and analysis tools on the ExPASy server. The proteomics protocols handbook. Totowa, NJ: Humana Press Inc.; 2005. p. 571–608. 10.1385/1592598900 [DOI] [Google Scholar]
  21. Green MR, Sambrook J. In: Inglis J, Boyle A, Gann A, editors. Molecular cloning: a laboratory manual. Volume 1. 4th ed. New York: Cold Springs Harbor Press; 2012. [Google Scholar]
  22. Hatse S, de Clercq E, Balzarini J. Role of antimetabolites of purine and pyrimidine nucleotide metabolism in tumor cell differentiation. Biochem Pharmacol. 1999;58(4):539–55. 10.1016/S0006-2952(99)00035-0 [DOI] [PubMed] [Google Scholar]
  23. Heinemann V, Schultz L, Issels RD, Plunkett W. Gemcitabine: a modulator of intracellular nucleotide and deoxynucleotide metabolism. Semin Oncol. 1995;22(4):11–8. [PubMed] [Google Scholar]
  24. Hofer A, Steverding D, Chabes A, Brun R, Thelander L. Trypanosoma brucei CTP synthetase: a target for the treatment of African sleeping sickness. Proc Natl Acad Sci U S A. 2001;98(11):6412–6. 10.1073/pnas.111139498 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Holyoak T, Fenn TD, Wilson MA, Moulin AG, Ringe D, Petsko GA. Malonate: a versatile cryoprotectant and stabilizing solution for salt‐grown macromolecular crystals. Acta Crystallogr D Biol Crystallogr. 2003;59(12):2356–8. 10.1107/S0907444903021784 [DOI] [PubMed] [Google Scholar]
  26. Kang GJ, Cooney DA, Moyer JD, Kelley JA, Kim HY, Marquez VE, et al. Cyclopentenylcytosine triphosphate. Formation and inhibition of CTP synthetase. J Biol Chem. 1989;264(2):713–8. 10.1016/S0021-9258(19)85001-9 [DOI] [PubMed] [Google Scholar]
  27. Karplus PA, Diederichs K. Assessing and maximizing data quality in macromolecular crystallography. Curr Opin Struct Biol. 2015;34:60–8. 10.1016/j.sbi.2015.07.003 [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Laskowski RA, Swindells MB. LigPlot+: multiple ligand‐protein interaction diagrams for drug discovery. J Chem Inf Model. 2011;51(10):2778–86. 10.1021/ci200227u [DOI] [PubMed] [Google Scholar]
  29. Levitzki A, Koshland DE. Cytidine triphosphate synthetase. Covalent intermediates and mechanisms of action. Biochemistry. 1971;10(18):3365–71. 10.1021/bi00794a008 [DOI] [PubMed] [Google Scholar]
  30. Levitzki A, Koshland DE. Role of an allosteric effector. Guanosine triphosphate activation in cytosme triphosphate synthetase. Biochemistry. 1972a;11(2):241–6. 10.1021/bi00752a015 [DOI] [PubMed] [Google Scholar]
  31. Levitzki A, Koshland DE. Ligand‐induced dimer‐to‐tetramer transformation in cytosine triphosphate synthetase. Biochemistry. 1972b;11(2):247–53. 10.1021/bi00752a016 [DOI] [PubMed] [Google Scholar]
  32. Lim RLH, O'Sullivan WJ, Stewart TS. Isolation, characterization and expression of the gene encoding cytidine triphosphate synthetase from Giardia intestinalis . Mol Biochem Parasitol. 1996;78(1–2):249–57. 10.1016/S0166-6851(96)02635-7 [DOI] [PubMed] [Google Scholar]
  33. Lynch EM, Hicks DR, Shepherd M, Endrizzi JA, Maker A, Hansen JM, et al. Human CTP synthase filament structure reveals the active enzyme conformation. Nat Struct Mol Biol. 2017;24(6):507–14. 10.1038/nsmb.3407 [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Lynch EM, Kollman JM. Coupled structural transitions enable highly cooperative regulation of human CTPS2 filaments. Nat Struct Mol Biol. 2020;27:42–8. 10.1038/s41594-019-0352-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. MacDonnell JE, Lunn FA, Bearne SL. Inhibition of E. coli CTP synthase by the “positive” allosteric effector GTP. Biochim Biophys Acta. 2004;1699:213–20. 10.1016/j.bbapap.2004.03.002 [DOI] [PubMed] [Google Scholar]
  36. McCluskey GD, Bearne SL. Biophysical analysis of bacterial CTP synthase filaments formed in the presence of the chemotherapeutic metabolite gemcitabine‐5′‐triphosphate. J Mol Biol. 2018;430(8):1201–17. 10.1016/j.jmb.2018.02.019 [DOI] [PubMed] [Google Scholar]
  37. McCluskey GD, Mohamady S, Taylor SD, Bearne SL. Exploring the potent inhibition of CTP synthase by gemcitabine‐5′‐triphosphate. Chembiochem. 2016;17(23):2240–9. 10.1002/cbic.201600405 [DOI] [PubMed] [Google Scholar]
  38. McPartland RP, Wang MC, Bloch A, Weinfeld H. Cytidine 5′‐triphosphate synthetase as a target for inhibition by the antitumor agent 3‐deazauridine. Cancer Res. 1974;34:3107–11. [PubMed] [Google Scholar]
  39. Mori G, Chiarelli LR, Esposito M, Makarov V, Bellinzoni M, Hartkoorn RC, et al. Thiophenecarboxamide derivatives activated by EthA kill Mycobacterium tuberculosis by inhibiting the CTP synthetase PyrG. Chem Biol. 2015;22(7):917–27. 10.1016/j.chembiol.2015.05.016 [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Otwinowski Z, Minor W. Processing of x‐ray diffraction data collected in oscillation mode. Methods Enzymol. 1997;276:307–26. [DOI] [PubMed] [Google Scholar]
  41. Pappas A, Yang WL, Park TS, Carman GM. Nucleotide‐dependent tetramerization of CTP synthetase from Saccharomyces cerevisiae . J of Biol Chem. 1998;273(26):15954–60. 10.1074/jbc.273.26.15954 [DOI] [PubMed] [Google Scholar]
  42. Robertson JG, Villafranca JJ. Characterization of metal ion activation and inhibition of CTP synthetase. Biochemistry. 1993;32:3769–77. Available from:. https://pubs.acs.org/sharingguidelines [DOI] [PubMed] [Google Scholar]
  43. Toschi L, Finocchiaro G, Bartolini S, Gioia V, Cappuzzo F. Role of gemcitabine in cancer therapy. Future Oncol. 2005;1(1):7–17. 10.1517/14796694.1.1.7 [DOI] [PubMed] [Google Scholar]
  44. Vagin A, Teplyakov A. MOLREP: an automated program for molecular replacement. J Appl Cryst. 1997;30(6):1022–5. 10.1107/S0021889897006766 [DOI] [Google Scholar]
  45. von der Saal W, Anderson PM, Villafranca JJ. Mechanistic investigations of Escherichia coli cytidine‐5′‐triphosphate synthetase: detection of an intermediate by positional isotope exchange experiments. J Biol Chem. 1985;260(28):14993–7. 10.1016/s0021-9258(18)95692-9 [DOI] [PubMed] [Google Scholar]
  46. von der Saal W, Villafranca JJ. Cytidine‐5′‐triphosphate synthetase catalyzes the phosphorlyation of uridine 5′‐triphosphate by adenosine‐5′‐triphosphate. J Am Chem Soc. 1985;107(3):703–4. [Google Scholar]
  47. Winter G, Waterman DG, Parkhurst JM, Brewster AS, Gildea RJ, Gerstel M, et al. DIALS: implementation and evaluation of a new integration package. Acta Crystallogr D Struct Biol. 2018;74:85–97. 10.1107/S2059798317017235 [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Yoshida T, Nasu H, Namba E, Ubukata O, Yamashita M. Discovery of a compound that acts as a bacterial PyrG (CTP synthase) inhibitor. J Med Microbiol. 2012;61:1280–5. 10.1099/jmm.0.046052-0 [DOI] [PubMed] [Google Scholar]
  49. Zalkin H. CTP synthetase. Methods Enzymol. 1985;113:282–7. 10.1016/0003-2697(85)90038-7 [DOI] [PubMed] [Google Scholar]
  50. Zhao Y, Li J, Gu H, Wei D, Xu YC, Fu W, et al. Conformational preferences of π–π stacking between ligand and protein, analysis derived from crystal structure data geometric preference of π–π interaction. Interdiscip Sci. 2015;7(3):211–20. 10.1007/s12539-015-0263-z [DOI] [PubMed] [Google Scholar]
  51. Zheng H, Chordia MD, Cooper DR, Chruszcz M, Müller P, Sheldrick GM, et al. Validation of metal‐binding sites in macromolecular structures with the CheckMyMetal web server. Nat Protoc. 2014;9(1):156–70. 10.1038/nprot.2013.172 [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Zhou X, Guo CJ, Chang C‐C, Zhong J, Hu H‐H, Lu G‐M, et al. Structural basis for ligand binding modes of CTP synthase. PNAS. 2021;118(30):1–11. 10.1073/pnas.2026621118/ [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Zhou X, Guo CJ, Hu HH, Zhong J, Sun Q, Liu D, et al. Drosophila CTP synthase can form distinct substrate‐ and product‐bound filaments. J Genet Genomics. 2019;46(11):537–45. 10.1016/j.jgg.2019.11.006 [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Data S1: Supporting Information.

Click here for additional data file. (5.6MB, docx)

Data Availability Statement

The data and coordinates for the structures presented are deposited in the Worldwide Protein Data Bank (wwPDB) under the accession codes 8FV6, 8FV7, 8FV8, 8FV9, 8FVA, 8FVB, 8FVC, 8FVD, 8SBR, and 8FVE.

Articles from Protein Science : A Publication of the Protein Society are provided here courtesy of The Protein Society

RESOURCES