Structure and characterisation of CMP-Kdn synthetase from the haptophyte microalgae Prymnesium parvum

Abstract

Sialic acids – 9-carbon ulosonic acids – are implicated in many cell–cell and host–pathogen interactions due to their prevalent location at the non-reducing end of glycoconjugates. Sialic acids have recently been observed in microalgae, including the toxic bloom-forming Prymnesium parvum, which produces the deaminated sialic acid, ketodeoxynonulosonic acid (Kdn), through de novo biosynthesis. Here we report on the key CMP-sialic acid synthetase enzyme (CMAS), PpNeuA, which activates Kdn to its sugar nucleotide congener, CMP-Kdn. In the present study, the X-ray crystal structure of PpNeuA was determined to 1.8 Å resolution and shows that it adopts a similar overall fold to that of other sialic acid synthetase enzymes, with which it shares ca 30% amino acid sequence identity. PpNeuA specificity for Kdn is dependent upon Arg196, a hydrophilic residue that is only found in Kdn-specific sialic acid synthetases. R196L mutation switches the substrate preference of PpNeuA from Kdn to N-acetylneuraminic acid (Neu5Ac). Kinetic analysis shows that Arg196 plays both a role in substrate binding (impact on K_M) and catalysis (impact on k_cat). In the context of generating metabolic probes to identify the location and context (glycolipid vs glycoprotein) of Kdn in P. parvum, we also report on the ability of PpNeuA to accept both 5Az-Kdn and 9Az-Kdn as substrates.

---

Introduction

Sialic acids are a diverse family of acidic nine-carbon sugars that are widespread in nature; they can be highly modified, leading to at least 60 known variations.¹ Most often they are found at the non-reducing terminal end of glycoconjugates (glycoproteins and glycolipids), where their exposure at the cell surface presents them to pathogens.² In humans, the most common sialic acid is Neu5Ac, while variations at C5 lead to other sialic acids, for example, the hydroxylated variant N-glycolylneuraminic acid (Neu5Gc) in mammals other than humans and the deaminated variant Kdn in cold blooded vertebrates (Fig. 1).³ Variations at C5, C7, and C8 lead to pseudaminic acid (Pse), a non-mammalian sialic acid present in bacteria.⁴ 3-deoxy-D-manno-octulosonic acid (Kdo) is a related eight-carbon sugar present in the cell wall polysaccharides of Gram-negative bacteria.⁵

Fig. 1 Sialic acid structure and CMAS mechanism. (A) Structure of C9 sialic acids and structurally related C8 acidic sugar, Kdo. The C5 variable substituent of sialic acid is highlighted in red. (B) Reaction mediated by CMAS enzymes, to activate sialic acid into the CMP-activated sialic acid.

Human and avian flu viruses are well known to exploit sialic acids to attach and gain entry to the cell.⁶ Studies have also shown that Kdn may be a mediator of viral infection in another haptophyte microalgae, Emiliania huxleyi.^7,8 Phylogenetically related to E. huxleyi, Prymnesium parvum is a unicellular mixotrophic haptophyte microalga that is responsible for harmful algal blooms (HABs) across the globe.^9,10 These HABs result in large-scale fish kills, with devasting effects on both the local ecology and economy.¹¹ Algal viruses and their role in the regulation of algal blooms are well studied,¹² but the role of such viruses in regulation of Prymnesium blooms is less well explored.⁹ We previously reported that a lytic, double-stranded DNA virus, PpDNAV, contributes to the demise of P. parvum HABs in the Norfolk Broads, UK.^13,14 We also showed that PpDNAV only infects those strains of Prymnesium that produce Kdn.¹³ Although the molecular mechanism behind viral infection remains to be understood, the possibility of PpDNAV-Kdn interactions emulating those of the well-studied influenza virus-NeuAc interactions is an interesting prospect.

Kdn has been widely found in lower vertebrates and pathogenic bacteria, as well as in mammalian tissues, cancers and microalgae.^2,15 Kdn incorporation into glycoconjugates is a result of either a specific Kdn biosynthetic pathway,^16,17 or through promiscuity in the Neu5Ac biosynthetic pathway.¹⁸ The first step of incorporation of sialic acids into glycoconjugates involves the activation of sialic acid to the cytidine monophosphate diester by CMP-sialic acid synthetase (CMAS) in the presence of Mg²⁺ (Fig. 1).¹⁹ The activated sialic acid is then incorporated into glycoconjugates by sialyltransferases.¹⁹ In a similar way, Kdo is activated by CMP-Kdo synthetase (CKS), and Pse is activated by CMP-Pse synthetase (PseF) where it is then incorporated into lipooligosaccharides (LOS) and lipopolysaccharide (LPS), or glycoproteins of pathogenic bacteria (Fig. 1).^5,20 This activation strategy differs from the more common nucleoside diphosphate diester activation of other monosaccharides.²¹ The structures of several CMAS and CKS enzymes have been solved from both eukaryotic mammalian organisms²² and bacterial sources,^23–27 recently the first structure of PseF was published.²⁰ The X-ray crystallographic structures of CMAS, CKS and PseF show that they share a common 3D structure, and catalytic mechanism.^24,26 The enzyme transitions between an open conformation, when CTP is bound, to the active closed conformation, after sialic acid binding. In the closed conformation, the catalytic arginine, and two divalent cations are positioned for catalysis and the formation of CMP-sialic acid.²⁴

While the majority of characterised CMAS enzymes have a preference for Neu5Ac, they often show some level of substrate promiscuity.²⁸ However, there are also examples of CMAS with a Kdn preference – e.g. from Oncorhynchus mykiss,²⁹ Danio rerio,³⁰ Bacteriodetes thetaiotamicron¹⁶ and P. parvum.¹⁷ There is limited understanding of what governs the difference in substrate specificity between CMAS enzymes, and there is currently no published X-ray crystal structure for a Kdn-specific CMAS enzyme.

The production of sialic acid glycosides is very important for fundamental research and pharmaceutical chemistry, such as for determining structure–function relationships³¹ and in the development of anti-influenza drugs.³² However, the chemical synthesis is demanding; therefore enzymatic syntheses are often employed. The CMAS enzymes utilised in such syntheses typically show specificity for Neu5Ac.²⁴ Recently, Kdn has been utilised as a temporary protecting group in the efficient production of α2,6-Neu5Ac-glycoconjugates,³³ highlighting the importance of and the requirement for Kdn-specific CMAS enzymes.

Herein, we present the X-ray crystal structure of the Kdn-specific CMAS enzyme, PpNeuA, to underpin greater understanding of the strong preference of this enzyme for Kdn. We analyse the molecular basis for Kdn specificity, and through single site substitution mutants demonstrate the importance of Arg196 in CMAS substrate specificity. We also assess the potential of the PpNeuA enzyme to activate azide-substituted Kdn, which is central to metabolic Kdn engineering studies.

Materials and methods

Expression of CMP-Kdn synthetase and mutants

CMP-Kdn synthetase (CAMPEP_0191219004, with transmembrane domain truncation), PpNeuA-WT, was expressed as previously described.¹⁷ PpNeuA mutants were cloned using IDT's gBlock gene fragments. The PpNeuA-WT sequence with the desired mutation and overhangs for In-Fusion™ cloning into the pOPINF vector³⁴ were used for cloning. A full list of sequences used in this study can be found in the supporting information. gBlock fragments were cloned into pOPINF vectors using In-Fusion™ Snap Assembly kit (Takara) per the manufacturer's instructions.

The wild-type and mutant enzymes were expressed in BL21 (DE3)™ cells (NEB). 1 L cultures were grown at 37 °C in Luria-Bertani (LB) medium containing 100 µg mL⁻¹ ampicillin until mid-exponential phase. Protein expression was induced with 0.5 mM isopropyl β-D-thiogalactopyranoside, and the cells were incubated at 18 °C overnight. Cells were harvested by centrifugation and then lysed, using a cell disruptor (25 kpsi, Constant Systems) in buffer containing 50 mM Tris-HCl, pH 7.5, 0.5 M NaCl, 20 mM imidazole, protease inhibitor (SigmaFast™ Protease Inhibitor Cocktail Tables, Sigma) and 10 µg mL⁻¹ DNase (Roche). All proteins were purified using nickel affinity chromatography with a linear gradient of imidazole from 20 mM to 500 mM. Protein sample fractions judged to be >95% pure by SDS-PAGE were pooled and dialysed into TBS and concentrated using an Amicon Ultra 10 kDa MWCO filter.

For protein crystallisation, the His-tag was cleaved using HRV 3C protease (GenScript). The protein was then repurified using nickel affinity chromatography, as before, but the flow through was collected. The protein was subjected to a final purification polishing step using gel filtration with a HiLoad 16/600 Superdex 200 pg column (GE Healthcare), eluted with TBS, at 0.5 mL min⁻¹. Fractions containing protein were pooled and concentrated to 10 mg mL⁻¹ using an Amicon Ultra 15 10 kDa MWCO concentrator.

Circular dichroism

Data was collected on a ChiraScan V100. The CD signal for measurement in the far-UV region (190–260 nm) was recorded in a 1 mm path length cell with a protein concentration of 1 mg mL⁻¹. CD spectra were recorded with a bandwidth of 0.5 nm and a response time of 15 seconds, averaging three scans.

Crystallisation, X-ray data collection and structure solution

Recombinant PpNeuA with His-tag removed was crystallized at 10 mg mL⁻¹ in TBS. Crystallization conditions were identified using the PACT Premier screen (Molecular Dimensions). Crystals suitable for diffraction experiments were obtained by sitting drop vapour diffusion in 400 nL drops containing equal volumes of protein and reservoir solution, after two weeks at 4 °C. The reservoir solution comprised of 0.2 M CaCl₂, 0.1 M Tris, pH 8.0, 20% w/v PEG 6000. Prior to data collection crystals were cryoprotected with PEG 200 and flash-cooled in liquid nitrogen. All data were collected at Diamond Light Source (Harwell, UK). Data reduction was performed with Dials and the structure solved using a model derived from murine CMAS (PDB ID: 1QWJ). Iterative rounds of model building, and refinement were performed in COOT³⁵ and Phenix.refine³⁶ respectively. Validation with MOLPROBITY³⁷ and PDBREDO³⁸ were incorporated into the iterative rebuild and refinement process. The resolution cut off was validated using a paired refinement in PDBREDO. Data collection and refinement statistics are shown in Table S1. Pairwise structural alignments were made using the TM-align algorithm via RSCB protein Data Bank.³⁹

Molecular docking

CTP and CMP-Kdn were docked into a receptor derived from the apo PpNeuA X-ray crystal structure using Molsoft ICM-Pro. The crystal structure of the apo form of PpNeuA was in an open conformation, a model of the closed conformation was generated prior to docking studies. The structure of Neisseria meningitidis CMP-sialic acid synthetase (NmCMAS, PDB ID: 6CKL) was chosen as the basis for the closed model of PpNeuA. The two structures were superimposed before imposing the conformation of Chain B from 6CKL onto the structure of PpNeuA. This PpNeuA model of the closed conformation was used as the starting point for all subsequent docking studies. Additional molecular dynamics for R196 and subsequent energy minimisation steps were performed during the various docking procedures. A more detailed explanation of the process is included in the supplementary information.

³¹P NMR biochemical assay

5Az-Kdn⁴⁰ and 9Az-Kdn⁴¹ were prepared using variations on published methods. 10 mM Kdn, Neu5Ac, 5Az-Kdn or 9Az-Kdn was mixed with 8 mM CTP in 50 mM Tris-HCl (pH 9.0). ³¹P NMR spectra of each reaction was recorded before the addition of 20 or 1 µg CMP-sialic acid synthetase enzyme (t = 0 min), and after incubation of the enzymes in 30-minute intervals over 6 h at 298 K and 400 MHz. The reaction followed via the reduction of CTP signals (−5.79 ppm doublet; −10.71 doublet; and −19.38 singlet), and the appearance of CMP-sialic acid signals (−5.51 ppm singlet; −4.65 ppm singlet).

Kinetic assay

The CMAS enzyme activity was followed from the initial rates of pyrophosphate formation as detected using the EnzChek Phosphate Assay Kit (Invitrogen), following the manufacturer's instructions. The reaction was carried out in 50 mM Tris-HCl, pH 9.0 at 200 µL volume in triplicate. The reaction was initiated by the addition of 200–800 ng PpNeuA or mutants and followed by absorbance at 360 nm. The amount of pyrophosphate produced was calculated from comparison with a pyrophosphate standard curve. The initial rates of reaction were measured over a range of sialic acid (Neu5Ac, Kdn) concentrations with a fixed amount of CTP (0.5 mM). V_max, K_M and k_cat were calculated according to the Michaelis Menten equation using GraphPad Prism.

Results and discussion

Overall structure of PpNeuA

The X-ray crystal structure of ligand-free PpNeuA, without the C-terminal transmembrane domain, was solved to 1.8 Å resolution. PpNeuA belongs to space group P12₁1. The asymmetric unit (ASU) has dimensions 56.43, 47.78, 87.85 Å. Much like bacterial CKS and CMAS,^{22–24,27,42} the model consists of a homodimer, which is consistent with gel filtration analysis (Fig. S1). The two monomers superimpose with a low RMSD of 0.34 Å over 219 C^α atoms. The PpNeuA model (Fig. 2) consist of two domains, an N-terminal globular α/β domain (residues 9 to 143, 181 to 233), and a smaller dimerization domain (residues 144 to 180). The dimerization domain of each monomer forms tight interactions with the globular domain of the opposite monomer, through hydrogen bonds, salt bridges and hydrophobic contacts. The dimer interfacial area (3730 Ų) is large and helps generate a tightly entwined complex. The globular domain is an α-β-α sandwich, with the central β-sheet composed of six parallel β-strands (β1–5, 10) and one anti-parallel β-strand (β9) sandwiched between α-helices on both sides. The dimerization domain is composed of two β-strands (β6–7). The loop between β3 and αB (residues 81–88) is unmodelled due to low electron density.

Fig. 2 PpNeuA structure. (A) PpNeuA structure with monomer 1 is coloured by secondary structure and displayed in ribbon representation, α-helices in red and β-sheets in green; monomer 2 is coloured in orange and shown in ribbon representation. The protein surface of monomer 2 is shown in greyscale with the dimerization domain interactions highlighted. (B) Superposition of a single domain of PpNeuA (red), CMAS from Vibrio cholerae (PDB ID: 6IFD) (cyan) and CMAS from Mus musculus (PDB ID: 1QWJ) (yellow). Both structures are shown in ribbon representation and the variance in the dimerization domain orientation is highlighted.

The CMAS crystals structures with the lowest RMSD are from M. musculus (PDB ID: 1QWJ) and Vibrio cholera (PDB ID: 6IFD)^24,25 with RMSD of 1.45 Å and 1.41 Å, respectively. The PpNeuA active site is formed at the dimer interface, and includes the CTP binding pocket, and the sialic acid binding pocket. The apo-PpNeuA adopts the open conformation with no divalent metals bound, in the same way as other apo-CMAS crystal structures.

Substrate binding in PpNeuA

PpNeuA shares considerable overall structural similarity with previously published CMAS and CKS structures, however sequence identity ranges between only 25 to 40% (CMAS), 12 to 17% (CKS), and 26% (PseF). Structural and sequence alignments show that residues involved in metal ion binding and CTP binding are wholly conserved, whilst there is more variation in the amino acids that bind sialic acid (Fig. S2 and S3).

The sialic acid binding site of PpNeuA is most like Neu5Ac specific CMAS, as the only difference between Neu5Ac and Kdn is the substitution at C5, Neu5Ac contains the larger and more hydrophobic N-acetyl group, whilst Kdn has a smaller and more hydrophilic hydroxyl group. The C5 group of Neu5Ac is accommodated within CMAS in a hydrophobic pocket composed of at least three hydrophobic and bulky residues in an otherwise polar active site (Fig. S2).⁴³ However, in PpNeuA the hydrophobic pocket is formed of Leu109, Tyr183 and Arg196, the third hydrophobic residue is replaced by hydrophilic arginine. CMAS enzymes from organisms that produce only Kdn – Klebsiella ozaenae,⁴⁴ Brevibacterium sp.⁴⁵ and Streptomyces albus⁴⁶ – all contain arginine in the same position (Fig. S2), while the Kdn-specific CMAS from Danio rerio has the hydrophilic threonine in this position.³⁰ Interestingly, the CMAS from Oncorhynchus mykiss (omCMAS), has leucine in this position despite Kdn being the preferred substrate.²⁸ The Kdn-preference of omCMAS is likely due to decreased hydrophobicity of the C5 binding pocket, which is composed of fewer hydrophobic residues than bacterial CMAS (NmCMAS and Vibrio cholerae CMAS (VcCMAS)).^23,24 Compared to the eukaryotic murine CMAS (MmCMAS),²² omCMAS has a shorter loop in the sialic acid binding site, this is proposed to alter the positioning of hydrophobic residues that are required to bind the C5 substituent, and thus weaken interactions with Neu5Ac compared to MmCMAS.^22,28

Despite the differences in stereochemistry and substituents between Kdn and Pse (Fig. 1), there are similarities in the sialic acid binding site. In PseF, the hydrophobic pocket contains three additional hydrophobic residues, Phe161, Leu184 add Phe215, compared to PpNeuA, this is to bind the additional N-acetyl group at C7.²⁰ Arg196 (PpNeuA) is equivalent to Ile214 of PseF, the hydrophobic Ile can accommodate the C5 N-acetyl substituent of Pse.

To further understand the structural basis for Kdn-specificity of PpNeuA, co-crystallisation in the presence of CTP and Kdn was attempted. However, these attempts did not result in co-crystals. Therefore CTP, CMP-Kdn and CMP-Neu5Ac were docked into a receptor derived from the apo PpNeuA X-ray crystal structure using Molsoft ICM-Pro.

CTP binding residues

Previously characterised CMAS enzymes, NmCMAS²⁴ and MmCMAS,²² showed that CTP bound in the open configuration, whilst the VcCMAS²³ was found in a partially closed conformation when CDP was bound. The partially closed conformation seen for VcCMAS is facilitated by Gln177, this residue is not conserved in NmCMAS, MmCMAS and PpNeuA, therefore, CTP was docked into the open conformation of PpNeuA. Most CTP binding residues are invariant (Fig. S2) and the interactions are conserved within PpNeuA with the docked pose matching previously co-crystalised CMAS²⁴ and CKS^27,47 structures (Fig. 3). Arg78 hydrogen bonds with O2 and N3 of cytosine base, giving specificity for cytosine. An additional interaction between Thr82 and N4 is noted in PpNeuA that is also seen in eukaryotic MmCMAS, but not in bacterial CMAS (NmCMAS and VcCMAS). Asn29 donates a hydrogen bond to the 2′OH and 3′OH of the ribose moiety. CTP triphosphate interacts with PpNeuA through Arg19, Lys28, Ser22 to the α-, β- and γ-phosphates much in the same way as other CMAS enzymes. Interestingly, PpNeuA and other Kdn specific CMAS enzymes do not have a Lys residue that is invariant in Neu5Ac-specifc enzymes (Fig. S2), in PpNeuA the residue is Val23.²⁴ In characterised CMAS enzymes this Lys (Lys16 NmCMAS numbering), can interact through hydrogen bonding interactions with the backbone amino group to the γ-phosphate,^23,24 and also through the amino group to the β-phosphate.²³ It has been demonstrated that this Lys is not essential for CMAS activity, although mutation to alanine does reduce activity.⁴⁸

Fig. 3 Modelled binding interaction between CTP and PpNeuA. PpNeuA is shown in grey, CTP carbon atoms are shown in cream. Hydrogen bonds are shown as grey lines.

Sialic acid binding residues

In previously characterised CMAS enzymes, Neu5Ac makes various contacts with the active site, many of which are highly conserved (Fig. S2).^23,24,43 In silico docking simulations of PpNeuA and its product, CMP-Kdn, showed that many of the interactions are conserved in PpNeuA. In structurally characterised CMAS, there are three key interactions between enzyme and CMP-Neu5Ac: the interaction between O4 and a serine; the interaction between Gln and O8, and the nitrogen of the N-acetyl group; and the interaction between the hydrophobic pocket and the C5 N-acetyl group. The docked pose of CMP-Kdn is similar to CMP-Neu5Ac in co-crystalised CMAS structures,^22,24 and makes two of the conserved interactions. In PpNeuA Thr89 hydrogen bonds with the O4 atom, in other CMAS enzymes this threonine is substituted by serine, which maintains the same hydrogen bonding abilities. In the modelled closed structure of PpNeuA, the strictly conserved Gln110 does not make a hydrogen bonding interaction with O8 of CMP-Kdn (Fig. 4), however, the backbone of Leu164 hydrogen bonds with O7 on CMP-Kdn. The hydrophobic pocket in PpNeuA, contains Arg196 which is stacked against Lys165 in an extended position, in this position, Arg196 can hydrogen bond with the O5 of CMP-Kdn (Fig. 4). Attempts to dock CMP-Neu5Ac were unsuccessful in generating a plausible pose, due to a large steric clash between the C5 N-acetyl group of CMP-Neu5Ac and Arg196. This clash occluded CMP-Neu5Ac from binding and infers that Neu5Ac is not a substrate for PpNeuA, this is in agreement with experimental results from Wagstaff et al.¹⁷ As Neu5Ac active CMAS enzymes commonly have Leu instead of Arg in this position, this mutation was made in silico and CMP-Neu5Ac was docked successfully, thus, suggesting the role of Arg196 as gate-keeping residue for CMP-Kdn specificity.

Fig. 4 Modelled binding interaction between sialic acid products and PpNeuA. (A) Docked pose and interactions between PpNeuA and CMP-Kdn and CMP-Neu5Ac. CMP-Neu5Ac is modelled within the binding site based upon the atom positions of the docked CMP-Kdn, direct docking of CMP-Neu5Ac failed to yield any plausible docked pose. A modelled mutation of R196L is shown in magenta, this mutation was modelled in silico and subsequently energy minimised within the pocket. An associated dot surface for L196 is shown in magenta highlighting the removal of a potential steric clash between R196 and CMP-Neu5Ac. (B) Close up view of the potential steric clash between CMP-Neu5Ac and Arg196 of PpNeuA. PpNeuA residues belonging to chain A of the homodimer are shown in yellow, those belonging to chain B are shown in orange. CMP-Kdn is coloured in cream, and the atoms belonging to the C5 NHAc substituent of CMP-Neu5Ac are coloured cyan. Dot surfaces for the C5 NHAc of CMP-Neu5Ac is coloured red, and R196L mutation is coloured magenta are shown. Hydrogen bonds are shown in black dashed lines. The CMP-Kdn binding pocket is shown as a grey surface with overlaid cyan mesh.

Importance of sialic acid binding residues

To test the importance of Arg196 in Kdn specificity and binding, two mutants were made, with Arg196 mutated to alanine or leucine (Fig. S4). Circular dichroism was conducted and showed that the mutants did not destabilise the fold of PpNeuA (Fig. S5). ³¹P NMR was used to follow the reaction of PpNeuA and mutants (Fig. 5). The loss of the signals for CTP, corresponding to the α-, β-, and γ-phosphates, and the formation of signals for CMP-sialic acid and pyrophosphate followed the reaction's progress. As previously shown, the wild-type enzyme does not convert Neu5Ac to CMP-Neu5Ac, whilst conversion of Kdn occurs rapidly.¹⁷ Both mutants were able to utilise Neu5Ac as a substrate. PpNeuA_R196L showed greater conversion of Neu5Ac than PpNeuA_R196A. Neither mutant was able to convert Kdn at the concentrations tested.

Fig. 5 ³¹P NMR spectrum of reaction catalysed by PpNeuA and mutants. The reaction was initiated with the addition of 1 µg enzyme and monitored over 60 min. Reactions were carried out with Kdn (A)–(C) or Neu5Ac (D)–(F). Reactions with Kdn were only successful with PpNeuA_WT. Reactions with Neu5Ac only progressed with the mutants, PpNeuA_R196L and PpNeuA_R196A.

As other CMAS enzymes show substrate promiscuity,^28,30,49 reactions were carried out with twenty times more enzyme to assess substrate promiscuity. After 60 minutes of reaction, PpNeuA_WT converted less than 1% Neu5Ac to product, and both mutants converted less than 1% Kdn to product (Fig. S6).

To understand the role of Arg196, the steady state kinetic parameters for PpNeuA_WT and mutants were calculated from monitoring the release of pyrophosphate in a continuous spectrophotometric assay (Table 1 and Fig. S6). The K_M values for PpNeuA_WT and mutants are in a similar range to earlier literature values for CMAS enzymes with their preferred substrate. However, compared to Kdn-specific CMAS, the K_M values for PpNeuA are 10-fold lower than both the enzymes from O. mykiss²⁸ and B. thetaiotamicron.¹⁶ For both mutants, the K_M for Kdn is at least 50 times higher than the wild type, showing the importance of Arg196 in substrate binding. For Neu5Ac, PpNeuA_R196A has weaker binding compared to the larger and more hydrophobic PpNeuA_R196L mutant and demonstrates the importance of the larger hydrophobic residue in binding the N-acetyl group in the otherwise polar active site. Interestingly, the single residue mutation can change substrate specificity in PpNeuA. omCMAS despite having leucine in this position is still more active towards Kdn than Neu5Ac,²⁸ this highlights the complexity and intricacy of substrate specificity in CMAS. omCMAS is the only sialic acid synthetase within O. mykiss, and as O. mykiss produces both Kdn- and Neu5Ac-containing glycans it is understandable that omCMAS has a relaxed substrate specificity.²⁸

Table 1 Kinetic parameters for PpNeuA and mutants

	KM/mM		kcat/s^−1
	Kdn	Neu5Ac	Kdn	Neu5Ac
WT	0.12	4.3	4.7	0.0070
R196L	6.7	0.27	0.028	0.93
R196A	30	0.60	0.037	0.46

---

In addition to the role of Arg196 in binding, as evidenced by the increased K_M for the mutants, the k_cat values of the mutants are also significantly reduced, indicating that Arg196 may also play a role in transition state stabilisation and hence catalysis. The k_cat values of the mutants with Kdn are more than 100 times reduced. As well as being involved in substrate binding, Arg196 is proposed to hold the substrate in a productive position for catalysis through the hydrogen bonding network. The loss of Arg196 disrupts the hydrogen bonding network, Horsfall et al., proposed that Tyr179 (NmCMAS numbering) plays a similar role to Arg196, in substrate binding and positioning.⁴³

Substrate specificity with azide-substituted sialic acids

We next assessed the substrate scope of PpNeuA with Kdn with azide modification at C5 or C9. Previous studies with C9-azide-modified Neu5Ac, Neu5Gc and Kdn have shown significantly lower incorporation of 9Az-Kdn compared to 9Az-Neu5Ac and 9Az-Neu5Gc,⁵⁰ low incorporation is partly due to the lack of Kdn specificity of CMAS, but also the identity of the sialyltransferase, where some sialyltransferases show low activity toward CMP-Kdn compared to other CMP-sialic acids.⁵¹ Activity was measured by ³¹P NMR after a two-hour reaction. Compared to the unsubstituted Kdn, these modified substrates showed lower conversions, 9Az-Kdn showed 31% conversion, whilst 5Az-Kdn only showed 10% conversion (Fig. S8). Unsurprisingly, the larger azido group in the C5 position, much like the N-acetyl group of Neu5Ac, may sterically clash with Arg196. In the docked model of PpNeuA with CMP-Kdn, there appears to be a larger pocket surrounding O9, which better allows the 9Az-Kdn to bind in a productive orientation for catalysis compared to 5Az-Kdn. The activity of PpNeuA towards both 9Az-Kdn and 5Az-Kdn augurs well for the efficient in vitro production of azide-labelled CMP-Kdn and related glycoconjugates; it also encourages metabolic glycoengineering studies in P. parvum to better understand the location and role of this acidic sugar in microalgal biology and viral infection processes.

Conclusions

In the present study, we conducted a structural characterisation of CMP-Kdn synthetase from P. parvum, to understand its substrate specificity. Whilst all previously structurally characterised CMAS enzymes show Neu5Ac substrate specificity, PpNeuA has specificity for Kdn. Arg196 is shown to be a gate-keeper residue for Kdn specificity and play a role in both substrate binding and catalysis. The ability of PpNeuA to utilise azide-modified Kdn was demonstrated, which opens the way for PpNeuA use in the production of Kdn-containing glycoconjugates suitable for bio-orthogonal modification for both environmental, biotechnological and biomedical applications.

Author contributions

C.M.: experimental data, writing – original draft preparation. A.J.M. and B.A.W.: experimental data, review-editing. I.I, E.D., A.R., M.E.O. and C.W.L.: experimental data. R.A.F: writing – review, editing, funding and conceptualization. All authors contributed to the article and approved the submitted version.

Conflicts of interest

The authors declare no conflict of interest.

Data availability

The atomic coordinates and structure factors for PpNeuA have been deposited in the Protein Data Bank (PDB) under accession number 9T4Z. The data are publicly available at the RCSB PDB website via the following DOI: https://doi.org/10.2210/pdb9t4z/pdb.

Additional data, supporting this article have been included as part of the supplementary information (SI). Supplementary information is available. See DOI: https://doi.org/10.1039/d5cb00285k.

Acknowledgements

CM, AM-C and BW were supported by the University of Manchester. We gratefully acknowledge engagement and financial support from John Currie and colleagues of the Norfolk Pike Angling Club. The authors would like to thank Diamond Light Source for beamtime on i03 under proposal MX24447-63.

References

1. A. Varki, Trends Mol. Med., 2008, 14, 351–360.
2. S. Dedola, S. Ahmadipour, P. de Andrade, A. N. Baker, A. N. Boshra, S. Chessa, M. I. Gibson, P. J. Hernando, I. M. Ivanova, J. E. Lloyd, M. J. Marin, A. J. Munro-Clark, G. Pergolizzi, S. J. Richards, I. Ttofi, B. A. Wagstaff and R. A. Field, RSC Chem. Biol., 2024, 5, 167–188.
3. A. L. Lewis, X. Chen, R. L. Schnaar and A. Varki, in Essentials of Glycobiology, ed. A. Varki, R. D. Cummings, J. D. Esko, P. Stanley, G. W. Hart, M. Aebi, D. Mohnen, T. Kinoshita, N. H. Packer, J. H. Prestegard, R. L. Schnaar and P. H. Seeberger, Cold Spring Harbor, NY, 4th edn, 2022, pp. 185–204 DOI:10.1101/glycobiology.4e.15.
4. H. S. Chidwick and M. A. Fascione, Org. Biomol. Chem., 2020, 18, 799–809.
5. S. Jelakovic, K. Jann and G. E. Schulz, FEBS Lett., 1996, 391, 157–161.
6. J. E. Stencel-Baerenwald, K. Reiss, D. M. Reiter, T. Stehle and T. S. Dermody, Nat. Rev. Microbiol., 2014, 12, 739–749.
7. J. M. Fulton, H. F. Fredricks, K. D. Bidle, A. Vardi, B. J. Kendrick, G. R. DiTullio and B. A. Van Mooy, Environ. Microbiol., 2014, 16, 1137–1149.
8. S. L. Rose, J. M. Fulton, C. M. Brown, F. Natale, B. A. Van Mooy and K. D. Bidle, Environ. Microbiol., 2014, 16, 1150–1166.
9. B. A. Wagstaff, E. S. Hems, M. Rejzek, J. Pratscher, E. Brooks, S. Kuhaudomlarp, E. C. O'Neill, M. I. Donaldson, S. Lane, J. Currie, A. M. Hindes, G. Malin, J. C. Murrell and R. A. Field, Biochem. Soc. Trans., 2018, 46, 413–421.
10. S. R. Manning and J. W. La Claire, Mar. Drugs, 2010, 8, 678–704.
11. J. Sobieraj and D. Metelski, Toxins, 2023, 15, 403.
12. C. P. Brussaard, J. Eukaryot Microbiol., 2004, 51, 125–138.
13. B. A. Wagstaff, I. C. Vladu, J. E. Barclay, D. C. Schroeder, G. Malin and R. A. Field, Viruses, 2017, 9, 40.
14. B. A. Wagstaff, J. Pratscher, P. P. L. Rivera, E. S. Hems, E. Brooks, M. Rejzek, J. D. Todd, J. C. Murrell and R. A. Field, Environ. Sci. Technol., 2021, 55, 16538–16551.
15. S. Inoue and K. Kitajima, Glycoconj. J., 2006, 23, 277–290.
16. L. Wang, Z. Lu, K. N. Allen, P. S. Mariano and D. Dunaway-Mariano, Chem. Biol., 2008, 15, 893–897.
17. B. A. Wagstaff, M. Rejzek and R. A. Field, J. Biol. Chem., 2018, 293, 16277–16290.
18. J. Hao, W. F. Vann, S. Hinderlich and M. Sundaramoorthy, Biochem. J., 2006, 397, 195–201.
19. M. Sellmeier, B. Weinhold and A. Munster-Kuhnel, Top. Curr. Chem., 2015, 366, 139–167.
20. T. Keenan, A. R. Cowan, E. K. P. Flack, N. E. Hatton, A. J. Walklett, G. H. Thomas, G. R. Hemsworth and M. A. Fascione, Structure, 2024, 32, 2399–2409.
21. D. Kapitonov and R. K. Yu, Glycobiology, 1999, 9, 961–978.
22. S. Krapp, A. K. Munster-Kuhnel, J. T. Kaiser, R. Huber, J. Tiralongo, R. Gerardy-Schahn and U. Jacob, J. Mol. Biol., 2003, 334, 625–637.
23. S. Bose, D. Purkait, D. Joseph, V. Nayak and R. Subramanian, Acta Crystallogr., Sect. D:Struct. Biol., 2019, 75, 564–577.
24. M. M. Matthews, J. B. McArthur, Y. Li, H. Yu, X. Chen and A. J. Fisher, Biochemistry, 2020, 59, 3157–3168.
25. S. C. Mosimann, M. Gilbert, D. Dombroswki, R. To, W. Wakarchuk and N. C. Strynadka, J. Biol. Chem., 2001, 276, 8190–8196.
26. D. J. Heyes, C. Levy, P. Lafite, I. S. Roberts, M. Goldrick, A. V. Stachulski, S. B. Rossington, D. Stanford, S. E. Rigby, N. S. Scrutton and D. Leys, J. Biol. Chem., 2009, 284, 35514–35523.
27. S. Jelakovic and G. E. Schulz, J. Mol. Biol., 2001, 312, 143–155.
28. D. Nakata, A. K. Munster, R. Gerardy-Schahn, N. Aoki, T. Matsuda and K. Kitajima, Glycobiology, 2001, 11, 685–692.
29. T. Haselhorst, A. K. Munster-Kuhnel, A. Stolz, M. Oschlies, J. Tiralongo, K. Kitajima, R. Gerardy-Schahn and M. von Itzstein, Biochem. Biophys. Res. Commun., 2005, 327, 565–570.
30. W. Schaper, J. Bentrop, J. Ustinova, L. Blume, E. Kats, J. Tiralongo, B. Weinhold, M. Bastmeyer and A. K. Munster-Kuhnel, J. Biol. Chem., 2012, 287, 13239–13248.
31. S. Wang, C. Chen, M. R. Gadi, V. Saikam, D. Liu, H. Zhu, R. Bollag, K. Liu, X. Chen, F. Wang, P. G. Wang, P. Ling, W. Guan and L. Li, Nat. Commun., 2021, 12, 3573.
32. P. Laborda, S.-Y. Wang, A.-M. Lu, M. He, X.-C. Duan, Y.-J. Qian, Y.-S. Jung, L. Liu and J. Voglmeir, Adv. Synth. Catal., 2017, 359, 3120–3125.
33. Y. Zhou, Y. Li, J. Wen, Y. Zhang, Z. Hu, K. Zhong, H. Cao and J. Cheng, Chin. J. Chem., 2025, 43, 1479–1486.
34. N. S. Berrow, D. Alderton, S. Sainsbury, J. Nettleship, R. Assenberg, N. Rahman, D. I. Stuart and R. J. Owens, Nucleic Acids Res., 2007, 35, e45.
35. P. Emsley and K. Cowtan, Acta Crystallogr., Sect. D:Biol. Crystallogr., 2004, 60, 2126–2132.
36. D. Liebschner, P. V. Afonine, M. L. Baker, G. Bunkoczi, V. B. Chen, T. I. Croll, B. Hintze, L. W. Hung, S. Jain, A. J. McCoy, N. W. Moriarty, R. D. Oeffner, B. K. Poon, M. G. Prisant, R. J. Read, J. S. Richardson, D. C. Richardson, M. D. Sammito, O. V. Sobolev, D. H. Stockwell, T. C. Terwilliger, A. G. Urzhumtsev, L. L. Videau, C. J. Williams and P. D. Adams, Acta Crystallogr., Sect. D:Struct. Biol., 2019, 75, 861–877.
37. C. J. Williams, J. J. Headd, N. W. Moriarty, M. G. Prisant, L. L. Videau, L. N. Deis, V. Verma, D. A. Keedy, B. J. Hintze, V. B. Chen, S. Jain, S. M. Lewis, W. B. Arendall, 3rd, J. Snoeyink, P. D. Adams, S. C. Lovell, J. S. Richardson and D. C. Richardson, Protein Sci., 2018, 27, 293–315.
38. R. P. Joosten, F. Long, G. N. Murshudov and A. Perrakis, IUCrJ, 2014, 1, 213–220.
39. S. Bittrich, J. Segura, J. M. Duarte, S. K. Burley and Y. Rose, Bioinformatics, 2024, 40(6), btae370.
40. C. H. Lin, T. Sugai, R. L. Halcomb, Y. Ichikawa and C. H. Wong, J. Am. Chem. Soc., 1992, 114, 10138–10145.
41. D. C. M. Kong and M. von Itzstein, Carbohydr. Res., 1997, 305, 323–329.
42. H. J. Yoon, M. J. Ku, B. Mikami and S. W. Suh, Acta Crystallogr., Sect. D:Biol. Crystallogr., 2008, 64, 1292–1294.
43. L. E. Horsfall, A. Nelson and A. Berry, FEBS J., 2010, 277, 2779–2790.
44. Y. A. Knirel, N. A. Kocharova, A. S. Shashkov, N. K. Kochetkov, V. A. Mamontova and T. F. Soloveva, Carbohydr. Res., 1989, 188, 145–155.
45. N. V. Potekhina, L. I. Evtushenko, S. N. Senchenkova and A. S. Shashkov, Russ. J. Bioorg. Chem., 2007, 33, 66–72.
46. A. S. Shashkov, G. M. Streshinskaya, E. M. Tul'skaya, S. N. Senchenkova, L. M. Baryshnikova, A. S. Dmitrenok, B. E. Ostash and V. A. Fedorenko, Antonie van Leeuwenhoek, 2016, 109, 923–936.
47. H. Schmidt, J. R. Mesters, J. Wu, R. W. Woodard, R. Hilgenfeld and U. Mamat, PLoS One, 2011, 6, e23231.
48. D. M. Stoughton, G. Zapata, R. Picone and W. F. Vann, Biochem. J., 1999, 343(Pt 2), 397–402.
49. I. B. Mertsalov, B. N. Novikov, H. Scott, L. Dangott and V. M. Panin, Biochem. J., 2016, 473, 1905–1916.
50. B. Cheng, L. Dong, Y. Zhu, R. Huang, Y. Sun, Q. You, Q. Song, J. C. Paton, A. W. Paton and X. Chen, ACS Chem. Biol., 2019, 14, 2141–2147.
51. O. Blixt and J. C. Paulson, Adv. Synth. Catal., 2003, 345, 687–690.

---