TY - JOUR
AU - Wakarchuk, Warren, W
AB - Abstract Polysialyltransferases (polySTs) are glycosyltransferases that synthesize polymers of sialic acid found in vertebrates and some bacterial pathogens. Bacterial polySTs have utility in the modification of therapeutic proteins to improve serum half-life, and the potential for tissue engineering. PolySTs are membrane-associated proteins and as recombinant proteins suffer from inherently low solubility, low expression levels and poor thermal stability. To improve their physicochemical and biochemical properties, we applied a directed evolution approach using a FACS-based ultrahigh-throughput assay as a simple, robust and reliable screening method. We were able to enrich a large mutant library and, in combination with plate-based high-throughput secondary screening, we discovered mutants with increased enzymatic activity and improved stability compared to the wildtype enzyme. This work presents a powerful strategy for the screening of directed evolution libraries of bacterial polySTs to identify better catalysts for in vitro polysialylation of therapeutics. Directed evolution, Neisseria, polysialyltransferase, screening Introduction Polysialic acid (polySia) is a negatively charged homopolymer of the nine-carbon sugar N-acetylneuraminic acid (sialic acid, Neu5Ac) (Elkashef et al. 2016; Bhide, Zapater, and Colley 2018). This polymer occurs on a handful of vertebrate proteins, and on the capsules of some mammalian pathogenic bacteria. The mammalian polySia is found exclusively as an α2,8-linked homopolymer on a limited set of proteins: predominantly on the neural cell adhesion molecule (NCAM); on soluble CD36 (Yabe et al. 2003); on the synaptic cell adhesion molecule (SynCAM) (Galuska et al. 2010); on the dendritic cell chemokine receptor CCR7 (Kiermaier et al. 2016); and on the E-selectin ligand (Werneburg et al. 2016). The polysialylated NCAM molecule has been implicated in the regulation of neuronal migration, promoting plasticity of neuronal cells in early brain development, and is also involved in tumor metastasis (Muhlenhoff et al. 2013; Ehrit et al. 2017). The bacterial version of α2,8-polySia is found on capsules of neuroinvasive bacteria like Neisseria meningitidis group B and Escherichia coli K1, Moraxella nonliquefaciens and Mannheimia haemolytica serotype A2 (Silver et al. 1981; Adlam et al. 1987; Devi et al. 1991; Puente-Polledo et al. 1998). Other bacteria such as N. meningitidis group C or E. coli K92 produce capsules with α2,9- or alternating α2,8/α2,9-linkages (Bhattacharjee et al. 1975; Glode et al. 1977). The enzymes that synthesize the polymers are polysialyltransferases (polySTs), which catalyze the addition of sialic acids from the activated sugar donor, CMP-sialic acid (CMP-Neu5Ac), to the non-reducing end of the growing polySia chain (Cho and Troy 1994; Nakayama and Fukuda 1996). Mammalian polySTs are limited in their acceptor protein specificity, which appears to require specific components of both the enzyme and its targets (Hildebrandt, Muhlenhoff, and Gerardy-Schahn 2010; Zapater and Colley 2012; Muhlenhoff et al. 2013; Sato and Kitajima 2013) while bacterial polySTs show a much more relaxed substrate specificity, making them useful tools for polySia synthesis on a variety of glycoconjugates (Willis et al. 2008; Lindhout et al. 2013; Lizak et al. 2017). The polyST of N. meningitidis group B (PSTNm) has been used to polysialylate therapeutic proteins (Lindhout et al. 2011), neural tissues and cultured cells in vivo (El Maarouf et al. 2012). With recent reports about neural tissue regeneration by transgenic expression of polySia (El Maarouf and Rutishauser 2010; Jungnickel et al. 2011; Gosh et al. 2012) and the improved pharmacological profiles of therapeutic proteins (Constantinou et al. 2009; Wu et al. 2010; Lindhout et al. 2011; Peterson et al. 2011; McCarthy et al. 2013) enzymatic synthesis of polySia shows great potential for medical applications. However, the current versions of bacterial polySTs are not optimized for use in such applications, being somewhat labile. Directed evolution of these enzymes could thus enhance their activity and stability. The challenge then is to develop a suitable high-throughput assay for screening of large mutant libraries. The need for an oligo-sialyl acceptor and the challenge of detection of multiple additions of sialic acid make the development of high-throughput assays for polySTs much more challenging than for mono-sialyltransferases (Yu et al. 2014). Nonetheless, several high-throughput screens for polysialyltransferases have been described recently. For example, Keys et al. described an in vivo screen for polySTs involving complementation of an E. coli BL21 strain, that had been metabolically engineered to produce CMP-Neu5Ac, with an active polyST gene. Using a combined colony lift and antibody blotting procedure more than 104 polyST mutants were tested and the usefulness of their screening method was verified by screening a library of N-terminal truncations of PSTNm (Keys, Berger, and Gerardy-Schahn 2012). We have previously published a plate-based assay for polyST activity (Yu et al. 2013, 2014) which could be used for screening a mutant library, but a truly high-throughput screen of 105-106 mutants would be difficult using this approach alone. We have also previously developed truly high-throughput screening assays for glycosyltransferase improvement through directed evolution (Aharoni et al. 2006; Yang and Withers 2009), and so we sought to combine strategies which would permit a Fluorescent Activated Cell Sorting (FACS) based screen for mutants of the Polysialyltransferases. Our previously described high throughput plate-based assay for PSTNm used an immobilized trisialyl lactoside (GT3) that served as acceptor and takes advantage of an inactive form of the E. coli K1 bacteriophage endosialidase, EndoNF* (Jakobsson et al. 2007; Morley et al. 2009), which has been fused to Green Fluorescent Protein (GFP) as a reporter (Aalto et al. 2001). This assay can also be used to accurately determine kinetic parameters, and to screen for inhibitors (Yu et al. 2014). What needed to be developed first was a high throughput analysis of a large library so that we could employ the plate screen as a secondary screen for directly measuring enzyme activity. In this study, we establish a FACS high throughput screen for the detection of polyST activity based on the complementation of a polyST knockout in a non-pathogenic strain of E. coli EV36 (Vimr and Troy 1985) that ordinarily expresses the E. coli K1 polySia capsule (E. coli EV36_NeuSKO). We could then employ the GFP-EndoNF* fusion protein for detection of bacterial cell surface polysialic acid coupled to FACS sorting of cells with increased levels of polySia. Then we employed a simple secondary screen based on a 96-well plate assay to verify that the mutants enriched in the FACS high throughput screen had (1) higher activity on target proteins and (2) improved temperature stability. The mutants identified by the combination of the two screening methods showed improved in vitro modification of therapeutic proteins and a representative human cell surface, and thus may facilitate improved tissue engineering. In addition, these assay methods can be used to screen additional mutant gene libraries, guided by the recently published first X-ray crystal structure of a bacterial polySTs, giving new insights into the structural architecture and the mechanisms of substrate binding and catalysis (Lizak et al. 2017). Results Screening strain construction and development of the FACS assay We used complementation of the K1 capsule synthesis machinery to screen for mutants of PSTNm. To do this we first knocked out the E. coli polyST, NeuS, present in EV36 using well-established recombineering technology and removed the resistance cassette as described previously (Datsenko and Wanner 2000). The polyST KO strain EV36_NeuSKO, along with EV36, were tested for polySia production and secretion. The cells of the KO strain were polySia negative (non-fluorescent) after incubation with our fluorescent detection reagent, GFP-EndoNF*, while the original EV36 cells were brightly fluorescent indicating successful construction of the polyST KO strain (data not shown). The strains EV36 and EV36_NeuSKO grew comparably at 37°C and 42°C. Complementation of EV36_NeuSKO with functional polyST, either the MPB-fusion of PSTNm (MBP-Δ19PSTNm) or the GT38 family related, tagless version of Mannheimia haemolytica polyST (Δ20PSTMh), rescued polySia production after IPTG induction of the co-expressed genes and rendered cells fluorescent upon GFP-EndoNF* staining. Initial FACS experiments with both constructs gave GFP-positive populations comparable to those of EV36 with fluorescence levels that remained stably high for at least 90 min (Figure 1A). Transformation of EV36_NeuSKO with the thermo-inducible complementation vector pCLth-Δ19PSTNm was tested to verify polyST expression upon temperature shifts from 30°C to 37°C and 42°C, respectively. PolySia was detected after these temperature shifts at a similar level from both growth conditions as assessed by fluorescence. Fig. 1. View largeDownload slide FACS-based assay. (A) PolyST knock-out strain EV36_NeuSKO was complemented with Δ20PSTMh and MBP- Δ19PSTNm, respectively and compared to control parental EV36. PolyST were cloned into pCW vectors and sorting was performed after induction with IPTG. (B) Sorting of a Δ19PSTNm library in the thermo-inducible vector pCLth at 37°C, the wt control refers to the EV36_NeuSKO was complemented with MBP- Δ19PSTNm. Fig. 1. View largeDownload slide FACS-based assay. (A) PolyST knock-out strain EV36_NeuSKO was complemented with Δ20PSTMh and MBP- Δ19PSTNm, respectively and compared to control parental EV36. PolyST were cloned into pCW vectors and sorting was performed after induction with IPTG. (B) Sorting of a Δ19PSTNm library in the thermo-inducible vector pCLth at 37°C, the wt control refers to the EV36_NeuSKO was complemented with MBP- Δ19PSTNm. We generated mutant gene libraries by error-prone PCR and the corresponding complementation vectors were constructed according to the primer overlap extension POE-PCR protocol (You and Percival Zhang 2012). The resulting single cut vectors were ligated and transformed into E. coli 10 G Elite. A total of 1.4 × 106 cfu were generated from that transformation and 30 random clones were picked for sequencing to confirm library diversity. We found sequences with 0–7 point mutations (median 2) leading to all types of amino acid exchanges and two sequences with premature termination due to stop codons. Transformation into electrocompetent EV36_NeuSKO gave a final total of 0.96 × 106 cfu, thus almost a million different sequences were subjected to screening. Using FACS sorting we enriched gradually higher levels of GFP-positive cells over four rounds of FACS sorting (Figure 1B). This was achieved by inducing mutant polyST expression in exponentially growing bacteria upon shifting the temperature from 30°C to 37°C and collecting about 106 of the apparently highest activity cells in each round. To verify a successful enrichment, we sequenced 41 clones and analyzed the number and position of the mutations. None of the clones had premature stop codons while 16 clones (38%) turned out to be wild type sequences; 13 sequences (32%) carried one mutation; 6 sequences (15%) carried two mutations and another 6 sequences (15%) had three mutations. Expression cloning of mutants from FACS screening Overall expression of the enzymes in the pCLth complementation vector was too low to permit isolation of the protein, so the enriched mutant library was bulk sub-cloned into an expression vector that we had previously used to characterize these polySTs (Lindhout et al. 2013). Previous experience with the wildtype PSTNm had shown that a fusion construct with the E. coli maltose binding protein enabled efficient larger scale protein expression and purification. We chose to screen random clones by colony PCR to investigate the diversity of the resulting subcloned library. We picked 30 random clones to screen which revealed that 22/30 clones contained the full-length gene, 4 clones contained a truncated gene and 4 plasmids were empty and did not contain a PSTNm gene at all. The 22 mutants containing the full-length gene were sent for sequencing to examine diversity of the mutations in the library. Development and use of a secondary plate-based assay with protein substrates A 96-well microtiter-plate-based assay for polyST activity was developed to evaluate the activity of approx. 400 PSTNm mutants enriched by the FACS assay. The concept of this plate assay is that the therapeutic protein of interest (e.g., disialyl-fetuin or disialyl-A1AT) is bound to a high-protein-binding 96-well microtiter plate (Figure 2) and then used as the acceptor substrate for polySia addition. Our engineered GFP-EndoNF* is then used as a probe to detect the polySia produced in the assay (Figure 2) (Yu et al. 2014). Fig. 2. View largeDownload slide Schematic of the secondary plate assay for polyST mutants. Plates with immobilized disialyl-fetuin were incubated with PSTNm (cell lysate or purified enzyme) and the polySia products were detected by incubation with GFP-EndoNF*. Blue squares: N-acetylglucosamine, green circles: mannose, yellow circles: galactose, pink diamonds: sialic acid. Fig. 2. View largeDownload slide Schematic of the secondary plate assay for polyST mutants. Plates with immobilized disialyl-fetuin were incubated with PSTNm (cell lysate or purified enzyme) and the polySia products were detected by incubation with GFP-EndoNF*. Blue squares: N-acetylglucosamine, green circles: mannose, yellow circles: galactose, pink diamonds: sialic acid. We first examined the dependence of the fluorescence readout on different target protein and PSTNm concentrations, using concentrations of disialyl-fetuin from 0.01 to 10.0 μg/mL and PSTNm ranging from 0.1 to 5.0 μg/mL. A linear dose response up to 1 μg/mL disialyl-fetuin (Figure 3A) and up to 2.5 μg/mL PSTNm was observed (Figure 3B). Based on these results 1 μg/mL disialyl-fetuin was used for binding to the 96-well plates. For the polyST reactions 1.25 μg/mL of pure enzyme was used with an incubation time of 30 min at room temperature. For crude lysates a determination of the enzyme concentration was not feasible, so we normalized the results by the starting OD600 of each mutant culture. Fig. 3. View largeDownload slide Plate assay dose-response plots. Panel (A) Dependence of fluorescence readout on different disialyl-fetuin and Panel (B) different PSTNm concentrations at 10 min (squares) and 30 min (circles) to assess linearity and appropriate dosing. A linear dose response to 1 μg/mL disialyl-fetuin and to 2.5 μg/mL PSTNm was observed. Fig. 3. View largeDownload slide Plate assay dose-response plots. Panel (A) Dependence of fluorescence readout on different disialyl-fetuin and Panel (B) different PSTNm concentrations at 10 min (squares) and 30 min (circles) to assess linearity and appropriate dosing. A linear dose response to 1 μg/mL disialyl-fetuin and to 2.5 μg/mL PSTNm was observed. To facilitate screening, we used the E. coli strain XJb (DE3) to propagate the expression library as it lyses after a simple freeze thaw, which eliminates extra handling of individual cultures. We grew 384 mutants from the subclone library in 96-deep well plates, lysed the cells by freezing and thawing and used the crude lysates to check for differences in activity compared to the wild type. Mutants from wells that showed higher GFP-EndoNF* binding than PSTNm were selected for further analysis. In total 21 clones were selected, and DNA sequencing revealed one carrying the wildtype gene and one with an N-terminal insertion, but the remaining 19 mutants carried up to four different amino acid changes. Interestingly, all the mutants carried the same I360V mutation, with three of these showing only the I360V mutation. Three mutants showed the I360V substitution in combination with the Y9S mutation, which itself was the second most common mutation, being found in five different mutants. Additionally, the I360V and L96E double mutant was found twice among sequenced clones. After eliminating duplicates, nine remained for further analysis (Table I). Table I. Selected clones from plate assay with crude lysates. Name Mutation 1 Mutation 2 Mutation 3 Mutation 4 PST-102 I360V PST-103 I360V Y9S PST-104 I360V K96E PST-105 I360V M340T PST-106 I360V Y89C Y338S PST-107 I360V Y9S Y148S PST-108 I360V Y9S E68V PST-109 I360V Y9S E68V M340T PST-110 I360V N204T Name Mutation 1 Mutation 2 Mutation 3 Mutation 4 PST-102 I360V PST-103 I360V Y9S PST-104 I360V K96E PST-105 I360V M340T PST-106 I360V Y89C Y338S PST-107 I360V Y9S Y148S PST-108 I360V Y9S E68V PST-109 I360V Y9S E68V M340T PST-110 I360V N204T Table I. Selected clones from plate assay with crude lysates. Name Mutation 1 Mutation 2 Mutation 3 Mutation 4 PST-102 I360V PST-103 I360V Y9S PST-104 I360V K96E PST-105 I360V M340T PST-106 I360V Y89C Y338S PST-107 I360V Y9S Y148S PST-108 I360V Y9S E68V PST-109 I360V Y9S E68V M340T PST-110 I360V N204T Name Mutation 1 Mutation 2 Mutation 3 Mutation 4 PST-102 I360V PST-103 I360V Y9S PST-104 I360V K96E PST-105 I360V M340T PST-106 I360V Y89C Y338S PST-107 I360V Y9S Y148S PST-108 I360V Y9S E68V PST-109 I360V Y9S E68V M340T PST-110 I360V N204T Characterization of selected clones To obtain pure enzymes, expression was performed in the E. coli AD202 strain used for PSTNm. Pure enzymes were then used for the plate-based assay. Six independent assays were performed, and, in each case, all the mutants showed activity comparable to or higher than that of the wildtype (Supplementary data, Figure 1). We decided to eliminate PST-104, PST-106 and PST-107 from further studies since PST-104 and PST-107 were only marginally better than wildtype. PST-106 was not included in further studies because of the wildly inconsistent results we got for this mutant. We focussed on mutants that led to the quadruple mutant PST-109 (I360V, Y9S, E68V, M340T) which are PST-102 (I360V), PST-103 (I360V, Y9S), PST-105 (I360V, M340T) and PST-108 (I360V, Y9S, E68V). We also elected to explore PST-110 (I360V, N204T). Several biological replicates were performed to express and purify the mutant enzymes and they were tested using the plate-based assay. As seen in Figure 4, only PST-103 is not significantly improved relative to wildtype. (Figure 4). The increase in activity of the other mutants compared to the wildtype was around 1.5 to 2.1-fold. Fig. 4. View largeDownload slide Results after plate assay with at least three independent enzyme purifications. Plate assays contained the target protein at 1 μg/well, and purified enzyme in the reaction at 1.5 μg/mL. Results from at least three independent repeats of three enzyme purifications were averaged, and standard deviation and significance were assessed using a one-way ANOVA test with a Dunnett’s post hoc test, where *, 0.01 < P < 0.05; **, 0.001 < P < 0.01; ***, 0.0001 < P < 0.001; ns, P > 0.05. Fig. 4. View largeDownload slide Results after plate assay with at least three independent enzyme purifications. Plate assays contained the target protein at 1 μg/well, and purified enzyme in the reaction at 1.5 μg/mL. Results from at least three independent repeats of three enzyme purifications were averaged, and standard deviation and significance were assessed using a one-way ANOVA test with a Dunnett’s post hoc test, where *, 0.01 < P < 0.05; **, 0.001 < P < 0.01; ***, 0.0001 < P < 0.001; ns, P > 0.05. Specific activity of PSTNm vs PST-109 on a synthetic substrate The mutant PST-109 was selected as a representative for more detailed comparisons of kinetic parameters with those of the wildtype enzyme. The specific activities on a small molecule acceptor, BDP-GD3, were first determined using the tagless PSTMh, which is known to prefer small molecule acceptors, as a positive control (Lindhout et al. 2013). As can be seen in Table II, the specific activity of the evolved mutant PST-109 is twice that of wildtype PSTNm wildtype enzyme. Table II. Specific activities of PolySTs on the synthetic acceptor BDP-GD3. Specific activity/ [mU/mg] PSTMh 1140 ± 80 PSTNm 350 ± 100 PST-109 740 ± 160 Specific activity/ [mU/mg] PSTMh 1140 ± 80 PSTNm 350 ± 100 PST-109 740 ± 160 Enzymes from at least two separate preparations (two preps of PSTMh, three preps each of PSTNm and PST-109) were tested. Table II. Specific activities of PolySTs on the synthetic acceptor BDP-GD3. Specific activity/ [mU/mg] PSTMh 1140 ± 80 PSTNm 350 ± 100 PST-109 740 ± 160 Specific activity/ [mU/mg] PSTMh 1140 ± 80 PSTNm 350 ± 100 PST-109 740 ± 160 Enzymes from at least two separate preparations (two preps of PSTMh, three preps each of PSTNm and PST-109) were tested. Glycoprotein and cell surface modification by PSTNm vs PST-109 To verify the plate assay results we measured the relative rates of polysialylation of disialyl-Fetuin and disialyl-A1AT by PSTNm and PST-109 in solution. Samples were taken after 10, 20, 30 and 120 min of reaction time. With PST-109 we see an increase in the reaction rate with both protein substrates and what appears to be a qualitative difference in the chain length between protein targets (Figure 5). We can see that the multiple populations observed with A1AT by gel staining are polysialylated in the Western blot and that in the case of PST-109 some of that material is in excess of 250 kDa. Of equal interest is the observation that with fetuin as a substrate we see more of a continuum of polysialylated species, although again a distribution of visible populations with the longer products again in excess of 250 kDa. Fig. 5. View largeDownload slide Polysialylation of glycoproteins disialyl-Fetuin and disialyl-A1AT by PSTNm and PST-109. Reaction results in a change in migration behavior relative to the original disialylated protein, and the appearance of a high molecular mass material (polySia). Panel A, left-hand side, show the Coomassie-blue stained gel of reactions on disialyl-fetuin, with lanes 1-4 being PSTNm reactions, and lanes 5-8 being PST-109 reactions at 10, 20, 30 and 120 minutes respectively. The right-hand side of Panel (A) is an immunoblot reacted with the mAb 735 antibody which is specific for Polysialic acid. Panel (B), left-hand side shows the Coomassie-blue stained gel of reactions on disialyl-A1AT, with lanes 1-4 being PSTNm reactions, and lanes 5-8 being PST-109 reactions at 10, 20, 30 and 120 minutes respectively. The right-hand side of Panel B is an immunoblot of an identical gel, reacted with the mAb 735 antibody which is specific for Polysialic acid. In both sets of experiments the level of each enzyme used was 50 μg/mL, and the target proteins were 1 mg/mL. Fig. 5. View largeDownload slide Polysialylation of glycoproteins disialyl-Fetuin and disialyl-A1AT by PSTNm and PST-109. Reaction results in a change in migration behavior relative to the original disialylated protein, and the appearance of a high molecular mass material (polySia). Panel A, left-hand side, show the Coomassie-blue stained gel of reactions on disialyl-fetuin, with lanes 1-4 being PSTNm reactions, and lanes 5-8 being PST-109 reactions at 10, 20, 30 and 120 minutes respectively. The right-hand side of Panel (A) is an immunoblot reacted with the mAb 735 antibody which is specific for Polysialic acid. Panel (B), left-hand side shows the Coomassie-blue stained gel of reactions on disialyl-A1AT, with lanes 1-4 being PSTNm reactions, and lanes 5-8 being PST-109 reactions at 10, 20, 30 and 120 minutes respectively. The right-hand side of Panel B is an immunoblot of an identical gel, reacted with the mAb 735 antibody which is specific for Polysialic acid. In both sets of experiments the level of each enzyme used was 50 μg/mL, and the target proteins were 1 mg/mL. We have previously shown that PSTNm is effective at human cell surface modification (El Maarouf et al. 2012; Lindhout et al. 2013). We wanted to exam the differences in cell surface modification on a suspension cell culture that could be easily be examined by flow cytometry. We have screened several cell lines for cell surface labeling with PSTNm (data not shown) and chose to use Jurkat cells for our comparison. The observation we made is that PST-109 produced ~10X better GFP-EndoNF* binding than the parent PSTNm (Figure 6), which is consistent with longer polySia chains being produced and providing more binding sites for GFP-EndoNF*. Fig. 6. View largeDownload slide Jurkat cell surface polysialylation comparison of PST-109 and PSTNm. Jurkat cells were incubated with 50 μg/mL PSTNm or PST-109 along with 5 mM CMP-Neu5Ac for 1 hour in serum free RPMI media. Cell surface polysialylation was detected by staining fixed cells with 10 μg/mL GFP-EndoNF* and running samples on the BD Accuri™ C6 Plus flow cytometer (n = 3). Shown is a representative and gated histogram prepared using FlowJo software. Red peak: control without GFP-EndoNF*, blue peak: control, orange peak: PSTNm, green peak: PST-109. Fig. 6. View largeDownload slide Jurkat cell surface polysialylation comparison of PST-109 and PSTNm. Jurkat cells were incubated with 50 μg/mL PSTNm or PST-109 along with 5 mM CMP-Neu5Ac for 1 hour in serum free RPMI media. Cell surface polysialylation was detected by staining fixed cells with 10 μg/mL GFP-EndoNF* and running samples on the BD Accuri™ C6 Plus flow cytometer (n = 3). Shown is a representative and gated histogram prepared using FlowJo software. Red peak: control without GFP-EndoNF*, blue peak: control, orange peak: PSTNm, green peak: PST-109. Temperature stability One of our objectives was to improve the thermal stability of the enzymes since incubation of the wildtype enzyme at temperatures required for modification reactions (37°C) typically resulted in considerable denaturation. By selecting for higher activity after overnight incubation at 37°C we hoped to select for enzymes that are more thermotolerant. Samples of PST-109 and wildtype PSTNm were incubated at 37°C and the residual activity was measured as a function of time using BDP-GD3 as substrate. As can be seen in Figure 7, PST-109 retains 54% of its starting activity after 7 hours at 37°C while for wildtype only 6% of its starting activity remains, thus the mutant is substantially more stable. Fig. 7. View largeDownload slide Comparing the temperature stability of PSTNm and PST-109 at 37°C. Enzymes were incubated at 37°C for the times indicated and residual activity measured using the small molecule BDP-GD3 assay (black circles: PSTNm, white circles: PST-109). Results from enzymes obtained from two enzyme preparations were averaged, and standard deviation and significance were assessed using a two-way ANOVA test with a Sidak’s post hoc test, where *, 0.01 < P < 0.05; **, 0.001 < P < 0.01; ***, 0.0001 < P < 0.001; ****, P < 0.0001; ns, P > 0.05. Fig. 7. View largeDownload slide Comparing the temperature stability of PSTNm and PST-109 at 37°C. Enzymes were incubated at 37°C for the times indicated and residual activity measured using the small molecule BDP-GD3 assay (black circles: PSTNm, white circles: PST-109). Results from enzymes obtained from two enzyme preparations were averaged, and standard deviation and significance were assessed using a two-way ANOVA test with a Sidak’s post hoc test, where *, 0.01 < P < 0.05; **, 0.001 < P < 0.01; ***, 0.0001 < P < 0.001; ****, P < 0.0001; ns, P > 0.05. Discussion We have described the development of a FACS based high throughput screen for the detection of polyST activity based on the complementation of a polyST knockout in the E. coli strain EV36_NeuSKO, and a secondary screen to (1) verify that the FACS based screen had enriched mutants with improved activity and or stability and (2) to screen for mutants with specific improvements, e.g., thermal stability. This method is substantially different from the FACS- based glycosyltransferase screen we have previously published (Aharoni et al. 2006; Yang et al. 2010). The major obstacle to screening of very large PolyST mutant libraries efficiently and robustly is that the assay needs to discriminate between a single and multiple transfers of sialic acid onto a primed substrate. We envisioned a FRET-based cell sorting approach that senses the intracellular production of polymeric sialic acid but failed due to a lack of sufficiently strong polySia binders that co-expressed in our screening host. Therefore, we decided to use GFP-EndoNF*, to detect production and export of an α2,8-polySia capsular polysaccharide. We used a complementation assay in the E. coli EV36 strain which was originally constructed to express the E. coli K1 α2,8-polySia capsular polysaccharide in a non-pathogen background (Vimr and Troy 1985). Using a recombineering approach (Datta, Costantino, and Court 2006; Sawitzke et al. 2007) we created the polyST knockout, E. coli EV36_NeuSKO. Complementation with our polySia libraries generated bacteria that contain unique PST sequences and reported the activity of the individual mutants in single cells. In an improvement over all previously reported polyST screening assays, we could select active mutants from large pools (>106) in less than 1 h and complete the enrichment process in the course of one week. We rationalized that the propensity of polySTs to aggregate in solution is due to a hydrophobic area that faces the inner membrane surface. Mutations in this interface may reduce the formation of aggregates and improve the stability of the enzymes when expressed as soluble proteins. However, while single mutations often cause protein instability, a second mutation can sometimes compensate and eventually improve the fitness of the mutant (Hecky and Muller 2005; Bershtein, Goldin, and Tawfik 2008; Wyganowski, Kaltenbach, and Tokuriki 2013). Unlike PSTMh, which has been successfully expressed in a tagless, soluble form allowing for crystallographic studies, PSTNm is prone to aggregation and requires MBP for soluble expression. Thus, PSTNm was an ideal candidate for these directed evolution studies, and to this end, we generated the PSTNm library with a stepwise error-prone PCR protocol to diversify the polyST randomly. One of the major hurdles in the generation of highly complex libraries is the low transformation rate of ligated vector DNA. The transformation rate decreases with the size of the vector, its topology and the occurrence of two ligation sites in relaxed circular DNA is up to 106 times less efficient than for intact, supercoiled DNA. The POE-PCR protocol provided vector + insert DNA with only one ligation site, yielding 102-103 more cfu after ligation and transformation than obtained with traditional protocols. Even with a larger sized screening vector (pCLth; >6 kbp with insert) we achieved almost 1.5 × 106 cfu after ligation/transformation. The screening vector pCLth was designed to (1) reduce intercellular bias in protein expression and (2) allow for adjustable protein levels. To achieve both we put the protein expression under control of the pL and/or pR phage lambda promoters regulated by the thermolabile cI857 repressor protein. This system avoids use of chemicals such as IPTG or arabinose to induce protein expression but rather utilizes temperature shifts for induction instead. The advantage of this is that temperature is applied uniformly over the bacterial population, ensuring comparable levels of induction whereas chemicals may be transported actively or passively into the cells and may be present at variable levels in every single cell. Because higher temperature causes more repressor protein to unfold, we were able to adjust protein levels to optimize the assay. Sorting and enrichment at 37°C were performed with a re-transformed library of nearly one million polySTNM mutants. After four rounds of enrichment we had evolved a population with higher fluorescence compared to wild-type, indicating enrichment of cells with higher levels of capsule production. To determine the nature of the increased capsule production we had to isolate individual mutants and examine the polyST directly. Our secondary screen is a simplified version of our previously described plate-based high-throughput assay for polySTs from N. meningitidis in which a synthetic acceptor trisialyl lactoside was immobilized on a 384-well plate by click chemistry (Yu et al. 2014). Here, we use primed fetuin as a protein acceptor. Since our end goal is to use polyST in the modification of protein therapeutics, it is necessary to pull out the variants that have been optimized to modify proteins. To test the sensitivity of the plate-based assay, varying concentrations of purified PSTNm, were assayed after 10- and 30-min reaction times; both time-points yielded a linear response to enzyme concentration (Figure 3B). The simplicity of the plate assay, combined with screening on a protein-based acceptor, made the secondary screen a powerful tool to identify improved mutants within our enriched library. Our library was enriched at 37°C with the expectation that we would find enzymes that are more stable at the elevated temperature. Due to the substantial enrichment provided by our FACS screening step, our hit rate was substantially improved over previous efforts. Thus Keys et al. obtained seven active clones from their library of 2000 at 37°C (0.35%) (Keys, Berger, and Gerardy-Schahn 2012). By contrast we identified 21 interesting mutants from 400 clones of our enriched library. Sequencing revealed that 19 mutants contained up to four different amino acid changes, thus approx. 5% of the clones from the secondary screen were active mutants. To our surprise all mutants carried the same I360V mutation, with the second most common mutation being Y9S, which was found in a total of five different mutants. Mutations E68V, L96E and M340T were all found more than once among the 21 sequenced clones, indicating successful enrichment of improved mutants by the FACS procedure. These were also found in combinations, including a quadruple mutant (I360V/Y9S/E68V/M340T) as well as mutants containing the I360V only, or double mutants containing I360V, Y9S, or I360V, M340T and a triple mutant containing I360V/Y9S/E68V. The recently determined X-ray crystal structure of PSTMh in complex with CDP and FondaparinuxTm, a substrate mimetic, has provided the first structural insights into the biosynthesis of capsular polysialic acid (Lizak et al. 2017) revealing several spatially conserved residues among different bacterial polySTs, despite an overall sequence identity of only 28–31% (Lizak et al. 2017). However, the mutations uncovered in our screen are not found in polySTs from E. coli, M. haemolytica and N. meningitidis and when mapped onto a Phyre2 model of PSTNm, the variant residues are located on surface loops and away from the active site cleft (Figures 8 and 9). How these mutations stabilize the enzyme is not clear, but such is often the case with directed evolution, and indeed is why directed evolution is such a powerful approach as these mutations would not be found by rational design. Fig. 8. View largeDownload slide Structure based sequence alignment of PSTMh and PST-109. To map the mutations from PST-109 onto the known GT-38 enzyme structure we performed a structure-based sequence alignment using Expresso (Di Tommaso et al. 2011). Secondary structural elements were extracted from the PDB sequence of 5WCN. Figure was prepared using ESPRIT 3.0. (Robert and Gouet 2014) Fig. 8. View largeDownload slide Structure based sequence alignment of PSTMh and PST-109. To map the mutations from PST-109 onto the known GT-38 enzyme structure we performed a structure-based sequence alignment using Expresso (Di Tommaso et al. 2011). Secondary structural elements were extracted from the PDB sequence of 5WCN. Figure was prepared using ESPRIT 3.0. (Robert and Gouet 2014) Fig. 9. View largeDownload slide Mapping of the mutations in PST-109 based on a Phyre2 generated model of PSTNm. The picture on the top shows the structure of PSTMh compared to the Phyre2 generated (bottom) model of PSTNm. The mutations from PST-109 are labeled on this model, revealing that they are all surface mutations, located away from the active site. Fig. 9. View largeDownload slide Mapping of the mutations in PST-109 based on a Phyre2 generated model of PSTNm. The picture on the top shows the structure of PSTMh compared to the Phyre2 generated (bottom) model of PSTNm. The mutations from PST-109 are labeled on this model, revealing that they are all surface mutations, located away from the active site. The FACS based screening resulted in an accumulation of mutants with increased in vivo activity. The quadruple mutant PST-109 was chosen for further experiments as it gave the most consistent purification and higher in vitro specific activity data. The small molecule specific activity of this mutant increased 2-fold compared to PSTNm, which is consistent with the approximately 2-fold increase in activity on immobilized protein substrates in the plate assay. While the increase in activity on synthetic substrates seems modest, there is also qualitative difference in the modification of soluble glycoproteins and cell surface glycoproteins which is harder to quantitate but which we think important. Both the Western blot analysis, and the flow cytometry assay show much improved substrate modification with the PST-109 mutant. It appears that the polySia chain length is modulated differently with the mutant and that the glycoprotein substrate also plays a role in directing that effect. The dynamic cell surface glycoproteins of Jurkat cells also appear to be a good substrate for the mutant, which suggests our cell-based application will also be improved with this improved mutant. Overall, our directed evolution screening was very successful in identifying more stable mutants with improved enzyme activity on a variety of substrates leading us to further investigate its application in therapeutic protein and cell-based polysialylation. Materials and methods Generation of screening strain EV36_NeuSKO E. coli strain EV36 is a non-pathogenic K1/K12 hybrid harboring the chromosomal K1 gene cluster for polySia capsule (Vimr and Troy 1985). The neuS gene coding for the endogenous polyST PSTEc was knocked out using Recombineering technology to enable screening of PST libraries (Datta, Costantino, and Court 2006; Sawitzke et al. 2007; Sharan et al. 2009). Briefly, a double-stranded kanamycin resistance cassette flanked by neuS homologous sequences of around 50 bp upstream and downstream of the targeted region was constructed in order to replace an internal sequence (GTATATAATTATTTAGTTTTAAATTCAAAA) by a stop codon. Electro-competent EV36 cells were transformed with pSIM5, recovered, streaked on LB agar plates supplemented with 25 μg/mL chloramphenicol (Cm) and incubated overnight at 30°C. A single colony was picked and grown overnight in LB media + 25 μg/mL CM at 30°C. On the next day, cells were grown to OD600 0.4, warmed to 42°C in a water bath for 3 min and incubated for an additional 15 min @ 42°C in a shaker incubator before chilling the cell suspension on ice. Those “activated cells” were made electrocompetent and 100 ng KanR cassette was subsequently transformed in a 50 μL cell suspension. After 3 h recovery at 30°C, cells were streaked onto LB agar + 50 μg/mL kanamycin and incubated overnight at 30°C. Several colonies were picked and grown over night at 37°C, yielding chloramphenicol-sensitive NeuS- and KanR+ clones, reflecting pSIM5 removal. The kanamycin cassette was removed with the plasmid pCP20 according to the protocols provided (Datsenko and Wanner 2000). To assess the success of the NeuS knock-out we tested capsule staining with GFP-EndoNF* and analytical PCR from genomic DNA samples to verify the gene knock out. Design and construction of the thermosensitive screening vector pLBth-1 In our FACS-HTS, transcription of PSTs gene mutants is under the control of a pL promoter element of the Coliphage λ and efficiently induced at temperatures above 32°C. The screening vector pLBth-1 is a hybrid construct derived from parental pCW and pSIM5 and was assembled with the PIPE cloning strategy in two steps (Klock and Lesley 2009). The lacI gene and the two tac/lacO sequences were replaced sequentially by the thermosensitive CI857 λ-repressor gene and the flanking pRM/op-r1(-3) and pL/op-l1(-3) promoter. Library construction Libraries were prepared by sequential error-prone PCR reactions (“dilution and pooling”) combined with prolonged overlap extension PCR (POE-PCR). While the former yields gene sequences with different extents of random point mutations, the latter avoids classic and inefficient ligation of two restricted DNA fragments. The two techniques were combined, as separately described previously (McCullum et al. 2010; You and Percival Zhang 2012). After electroporation into E coli cells, 30 random colonies were picked, and their DNA was sequenced in order to estimate library diversity. Fluorescence activated cell sorting assay Electrocompetent EV36_NeuSKO cells were transformed with the pLBth-1_Δ19 PSTNM library (50 ng per 25 μL) and grown overnight in LB + 100 μg/mL carbenicillin and 12.5 μl/mL chloramphenicol. On the next day, cultures were diluted 1:50 in fresh LB + 100 μg/mL carbenicillin and further incubated @ 30°C for 2 h (5–6 mL). PolyST expression was induced by a temperature shift from 30°C to 37°C and the cultures were incubated for another 90 min. Cells were rapidly chilled, pelleted and resuspended in 350 μL ice-cold SOC media (0.5% Yeast Extract, 2% Tryptone, 10 mM NaCl, 2.5 mM KCl, 10 mM MgCl2, 10 mM MgSO4, 20 mM Glucose*) and stained by adding 150 μl PBS-buffered GFP-EndoNF* solution (final 5 mg/mL). After incubation for 90 min on ice they were directly sorted from a 1:40 diluted cell suspension, triggered by GFP fluorescence. Collected cells were recovered SOC for 2 h @ 30°C, diluted in fresh SOC + 100 μg/mL carbenicillin and grown overnight. The selection and enrichment process was repeated three times and 35 single colonies then picked for analytical sequencing. General methods All enzymes were purchased from New England Biolabs (NEB). The GeneJET™ Gel Extraction Kit (Fermentas) was used to purify DNA fragments from agarose gels and to purify digested plasmids and oligonucleotides. Plasmid DNA from transformed cells was isolated with the GeneJET™ Plasmid Miniprep kit (Fermentas). Agarose gel electrophoresis was performed as described elsewhere (Sambrook, Fritsch, and Maniatis 1989). Primers for PCR and DNA sequencing were purchased from Integrated DNA Technologies (IDT). PCR was performed using the Q5® High-Fidelity 2X Master Mix (NEB). Transformation of chemically competent E. coli DH10B cells and E. coli AD202 cells was done according to the manufacturer’s protocols. Transformants were screened by PCR using DreamTaq Green PCR Master Mix (2X) (Thermo Fisher Scientific), and recombinant clones were analyzed by restriction mapping and confirmed by sequencing. SDS-PAGE was carried out according to a standard protocol (Laemmli 1970) using a Protean II electrophoresis apparatus (Bio-Rad). Protein bands were visualized with Coomassie Brilliant Blue G 250 staining reagent. For Western-blotting of proteins onto a polyvinylidene difluoride membrane (Bio-Rad) a Mini Trans-Blot Cell (Bio-Rad) was used. Protein concentrations were determined using the Pierce™ BCA Protein Assay Kit (Thermo Fisher Scientific). Re-cloning of PSTNm mutants The mutant polysialyltransferase genes from N. meningitidis were re-cloned into the pCW-MalET expression vector. The mutant genes from the enriched library were amplified by PCR using the primer NmPST_for (5′-GGTCACACAGGAAACAGGATCCATCGATGCTTAGGAGGTCATATG-3′) and the primer NmPST_rev (5′-GTCGACCTGCAGAAGCTTATCGATGATAAGCTGTCAAACATG-3′) primers flanking the mutant genes. The amplification products were digested with NdeI/SalI and inserted into the linearized vector pCW-MalET digested with the same restriction enzymes. Transformation of chemically competent E. coli DH10B cells was done according to the manufacturer’s protocol (Thermo Fisher Scientific). Transformed cells were plated on LB plates supplemented with 100 μg/mL ampicillin (LB-Amp) and Plasmid DNA of combined transformants was isolated with the GeneJET™ Plasmid Miniprep kit (Thermo Fisher Scientific) the following day. Twenty mutants were sent for sequencing to check for variations. 96-deep well plate cultivation The Δ19PSTNm plasmid (wildtype) as well as the mutant plasmid library were transformed into XJb (DE3) cells (Zymo Research, T5051) by heat shock, and single colonies were selected from the transformant plate and grown up in 2 mL LB-Amp. Cultures were grown overnight at 37°C, and 100 μL was used to inoculate 2 mL LB-Amp with 0.5 mM IPTG and 3 mM L-arabinose. The following day the OD600 values of the plate (100 μl of each well) were measured in a clear 96-microtiter plate for normalization of the final measurements. The remaining cultures were harvested by centrifugation and the pellets were resuspended in 500 μL sterile mQ H2O, and frozen. Upon thawing, 10 μL of a DNase-RNase solution was added to the lysates, and the lysates were tested for activity in the 96-microwell plates. Priming of target proteins with α2,8-sialic acid termini The activity of PSTNm was determined on the bovine serum glycoprotein fetuin or A1AT after enzymatic modification to ensure that the N-glycans terminated in at least two Neu5Ac residues (disialyl-fetuin, disialyl-A1AT) as described previously (Lindhout et al. 2011). Expression of selected PSTNm mutants and GFP-EndoNF* PSTNm mutants and GFP-EndoNF* were inoculated in a preculture of LB-Amp. The main culture of 2x YT media supplemented with 100 μg/mL was inoculated 1/100 and cells were grown at 37°C and 200 rpm until an OD600 of 0.4 to 0.6 was reached. The culture was shifted to 20°C and expression was induced by addition of 0.5 mM IPTG. After incubation for 16 h −20 h at 20°C and 200 rpm, cells were harvested by centrifugation and cell pellets were stored at −80°C until use. Purification of GFP-EndoNF* The cell pellet was resuspended in 20 mM sodium phosphate, 0.5 M NaCl, 20 mM imidazole, pH 7.4, in the presence of a protease inhibitor mixture pellet (Roche Applied Science) followed by lysis with an Emulsiflex-C5 (Avestin) and centrifugation at 45 000 rpm to remove cell debris. Supernatant containing the soluble protein was purified using a 5 mL HisTrap™ High Performance column (GE Healthcare) with 20 mM sodium phosphate, 0.5 M NaCl, 20 mM imidazole, pH 7.4, and a linear gradient of 20 to 500 mM imidazole at pH 7.4. The fractions were analyzed by SDS–PAGE. Fractions containing the desired protein were pooled. Purification of PSTNm mutants To purify the enzymes, cell pellets were thawed and re-suspended in PBS buffer (15 mL buffer per 1 g pellet) in the presence of a protease inhibitor mixture (Roche Applied Science) followed by lysis with an Emulsiflex-C5 (Avestin) and centrifugation at 100,000 × g to remove cell debris. Supernatants containing the soluble protein were purified using a 5 mL HiTrap Heparin HP column (GE Healthcare) with a linear gradient of 0–70% B using the following buffers: A, PBS pH 7.4; B, PBS pH 7.4 + 2 M NaCl (Lindhout et al. 2013). Fractions containing the enzyme were pooled together. For storage at −80°C, 10% glycerol + 1 mM DTT were added. 96-well plate assay for polyST activity Fifty μl of 1 μg/mL target protein (e.g., primed fetuin or α-1-antitrypsin) in 1xPBS buffer pH 7.4 were applied to black, flat bottom, high protein binding 96-well plates (Sigma Aldrich) to bind the protein to the plates. The plate was mixed on a microtitre plate shaker for 30 seconds and then incubated at room temperature for 15 minutes. Unbound protein was removed, and the plates were washed two times with 50 μl of 1x PBS buffer. Plates were blocked with 50 μl of 1% BSA for 5 minutes + 2% Tween-20 and washed 3 times with PBS pH 7.4 + 0.05% Tween (PBS-T) and 2 times with PBS buffer. Reactions were performed using 10 μL of lysate in a 30 μL reaction for 1 hour at 30°C; all reactions were done using 1x PBS, 15 mM MgCl2, and 1.5 mM CMP-Neu5Ac. For pure enzymes, 1.25 μg/mL of each enzyme in a 50 μl reaction were used and all reactions were done using 1x PBS, 1 mg/mL BSA, 10 mM MgCl2, and 2 mM CMP-Neu5Ac. The wells were washed again 3x with PBS-T and 50 μL of 80 μg/mL GFP-EndoNF* in 1x PBS were added to the well and incubated at room temperature for 15 minutes They were then washed again with 3x with PBS-T, then 2x with PBS pH 7.4. Finally, 50 μL of 1x PBS was added and fluorescence was measured (Excitation: 485, Emission: 530 nm). Polysialyltransferase activity assays on small molecule acceptors The activity of the polySTs was determined by using synthetic disialyl-lactosyl (GD3) ganglioside analogs (glycan portion only) coupled to the BODIPY (BDP) fluorophore. The standard assay conditions are: 0.05 to 0.1 μg of polyST enzymes, 0.5 mM BDP-GD3, 50 mM HEPES pH 7.5, 10 mM MgCl2, and 10 mM CMP-Neu5Ac at 30°C. Reactions were stopped by addition of 1:1 (v/v) stop solution (80% acetonitrile). Thin layer chromatography was carried out on silica plates and run in ethyl acetate:methanol:water:acetic acid in a ratio of 3:2:1:0.2 as running solvent. HPLC analysis of polysialylated small molecule acceptors were done by diluting the reactions 1:100 with 12.5% acetonitrile. 10 μl of stopped reaction sample, corresponding to 25 pmol of acceptor, was applied to a DNAPac PA-100 guard column (Dionex) for separation. Chromatography was conducted at 0.5 mL min-1 with a column temperature of 40°C, using mobile phases consisting of acetonitrile (M1) and ammonium acetate 2 M pH 7.0 (M2). Products were separated using an elution gradient from 0–50% M2 over 15 min. Polysialyltransferase activity assays on proteins A sample of disialyl-fetuin and di-sialyl-A1AT (1 mg/mL) was modified in the presence of 50 mM sodium phosphate buffer pH 8.0, 10 mM MgCl2, and 10 mM CMP-Neu5Ac at 30°C for 15 min to two hours using an enzyme concentration of 50 μg/mL. Reactions were stopped using 1x protein sample buffer and analyzed by SDS-PAGE. Immunobloting for polySia on protein targets was performed with mAb735 (from Absolute antibody Ab00240-2.0) as previously described (Lindhout et al. 2013). Thermostability of PSTNm To compare the stabilities of the PSTNms, pure enzymes (15–20 μg, of 0.3–0.4 mg/mL) were stored at 37°C for several hours. The remaining activity was analyzed at different time points using the activity assay on small molecules described previously. Mammalian cell surface polysialylation Human T-lymphocyte Jurkat cells (ATCC® TIB-152™) were maintained in suspension using RPMI-1640 supplemented with 10% fetal bovine serum (FBS) and 5% streptomycin/penicillin in a humidified incubator at 37°C and 5% CO2. For the polysialylation experiment, about 5 × 106 cells/mL were incubated in 1 mL serum free RPMI media along with the reaction components for 1 hour. Reaction components were 5 mM CMP-Neu5Ac and 50 μg/mL PSTNm or PST-109. After fixation of cells with 4% paraformaldehyde for 10 minutes at 4°C cells were incubated with the GFP-EndoNF* at 10 μg/mL for 30-40 minutes at 4°C in the dark. Cells were centrifuged and pellets were resuspended in 2% BSA before running samples on the BD Accuri™ C6 Plus flow cytometer and the data were analyzed using FlowJo, 10.5.3 software. Conflict of interest statement None declared. References Aalto J , Pelkonen S , Kalimo H , Finne J . 2001 . Mutant bacteriophage with non-catalytic endosialidase binds to both bacterial and eukaryotic polysialic acid and can be used as probe for its detection . Glycoconj J . 18 : 751 – 758 . Google Scholar Crossref Search ADS PubMed Adlam C , Knights JM , Mugridge A , Williams JM , Lindon JC . 1987 . Production of colominic acid by Pasteurella haemolytica serotype A2 organisms . FEMS Microbiol Lett . 42 : 23 – 25 . Google Scholar Crossref Search ADS Aharoni A , Thieme K , Chiu CP , Buchini S , Lairson LL , Chen H , Strynadka NC , Wakarchuk WW , Withers SG . 2006 . High-throughput screening methodology for the directed evolution of glycosyltransferases . Nat Methods . 3 : 609 – 614 . Google Scholar Crossref Search ADS PubMed Bershtein S , Goldin K , Tawfik DS . 2008 . Intense neutral drifts yield robust and evolvable consensus proteins . J Mol Biol . 379 : 1029 – 1044 . Google Scholar Crossref Search ADS PubMed Bhattacharjee AK , Jennings HJ , Kenny CP , Martin A , Smith IC . 1975 . Structural determination of the sialic acid polysaccharide antigens of Neisseria meningitidis serogroups B and C with carbon 13 nuclear magnetic resonance . J Biol Chem . 250 : 1926 – 1932 . Google Scholar PubMed Bhide GP , Zapater JL , Colley KJ . 2018 . Autopolysialylation of polysialyltransferases is required for polysialylation and polysialic acid chain elongation on select glycoprotein substrates . J Biol Chem . 293 : 701 – 716 . Google Scholar Crossref Search ADS PubMed Cho JW , Troy FA 2nd . 1994 . Polysialic acid engineering: synthesis of polysialylated neoglycosphingolipids by using the polysialyltransferase from neuroinvasive Escherichia coli K1 . Proc Natl Acad Sci USA . 91 : 11427 – 11431 . Google Scholar Crossref Search ADS PubMed Constantinou A , Epenetos AA , Hreczuk-Hirst D , Jain S , Wright M , Chester KA , Deonarain MP . 2009 . Site-specific polysialylation of an antitumor single-chain Fv fragment . Bioconjug Chem . 20 : 924 – 931 . Google Scholar Crossref Search ADS PubMed Datsenko KA , Wanner BL . 2000 . One-step inactivation of chromosomal genes in Escherichia coli K-12 using PCR products . Proc Natl Acad Sci USA . 97 : 6640 – 6645 . Google Scholar Crossref Search ADS PubMed Datta S , Costantino N , Court DL . 2006 . A set of recombineering plasmids for gram-negative bacteria . Gene . 379 : 109 – 115 . Google Scholar Crossref Search ADS PubMed Devi SJ , Schneerson R , Egan W , Vann WF , Robbins JB , Shiloach J . 1991 . Identity between polysaccharide antigens of Moraxella nonliquefaciens, group B Neisseria meningitidis, and Escherichia coli K1 (non-O acetylated) . Infect Immun . 59 : 732 – 736 . Google Scholar PubMed Di Tommaso P , Moretti S , Xenarios I , Orobitg M , Montanyola A , Chang JM , Taly JF , Notredame C . 2011 . T-Coffee: a web server for the multiple sequence alignment of protein and RNA sequences using structural information and homology extension . Nucleic Acids Res . 39 : W13 – W17 . Google Scholar Crossref Search ADS PubMed Ehrit J , Keys TG , Sutherland M , Wolf S , Meier C , Falconer RA , Gerardy-Schahn R . 2017 . Exploring and exploiting acceptor preferences of the human polysialyltransferases as a basis for an inhibitor screen . Chembiochem . 18 : 1332 – 1337 . Google Scholar Crossref Search ADS PubMed El Maarouf A , Moyo-Lee Yaw D , Lindhout T , Pearse DD , Wakarchuk W , Rutishauser U . 2012 . Enzymatic engineering of polysialic acid on cells in vitro and in vivo using a purified bacterial polysialyltransferase . J Biol Chem . 287 : 32770 – 32779 . Google Scholar Crossref Search ADS PubMed El Maarouf A , Rutishauser U . 2010 . Use of PSA-NCAM in repair of the central nervous system . Adv Exp Med Biol . 663 : 137 – 147 . Google Scholar Crossref Search ADS PubMed Elkashef SM , Sutherland M , Patterson LH , Loadman PM , Falconer RA . 2016 . An optimised assay for quantitative, high-throughput analysis of polysialyltransferase activity . Analyst . 141 : 5849 – 5856 . Google Scholar Crossref Search ADS PubMed Galuska SP , Rollenhagen M , Kaup M , Eggers K , Oltmann-Norden I , Schiff M , Hartmann M , Weinhold B , Hildebrandt H , Geyer R et al. 2010 . Synaptic cell adhesion molecule SynCAM 1 is a target for polysialylation in postnatal mouse brain . Proc Natl Acad Sci USA . 107 : 10250 – 10255 . Google Scholar Crossref Search ADS PubMed Glode MP , Robbins JB , Liu TY , Gotschlich EC , Orskov I , Orskov F . 1977 . Cross-antigenicity and immunogenicity between capsular polysaccharides of group C Neisseria meningitidis and of Escherichia coli K92 . J Infect Dis . 135 : 94 – 104 . Google Scholar Crossref Search ADS PubMed Gosh A , Tuesta LM , Puentes R , Patel M , Melendez K , El Maarouf A , Rutishauser U , Pearse DD . 2012 . Extensive cell migration, axon regeneration, and improved function with polysialic acid-modified Schwann cells after spinal cord injury . Glia . 60 : 979 – 992 . Google Scholar Crossref Search ADS PubMed Hecky J , Muller KM . 2005 . Structural perturbation and compensation by directed evolution at physiological temperature leads to thermostabilization of beta-lactamase . Biochemistry . 44 : 12640 – 12654 . Google Scholar Crossref Search ADS PubMed Hildebrandt H , Muhlenhoff M , Gerardy-Schahn R . 2010 . Polysialylation of NCAM . Adv Exp Med Biol . 663 : 95 – 109 . Google Scholar Crossref Search ADS PubMed Jakobsson E , Jokilammi A , Aalto J , Ollikka P , Lehtonen JV , Hirvonen H , Finne J . 2007 . Identification of amino acid residues at the active site of endosialidase that dissociate the polysialic acid binding and cleaving activities in Escherichia coli K1 bacteriophages . Biochem J . 405 : 465 – 472 . Google Scholar Crossref Search ADS PubMed Jungnickel J , Eckhardt M , Haastert-Talini K , Claus P , Bronzlik P , Lipokatic-Takacs E , Maier H , Gieselmann V , Grothe C . 2011 . Polysialyltransferase-overexpression in Schwann cells mediates different effects during peripheral nerve regeneration . Glycobiology . 22 : 107 – 115 . Google Scholar Crossref Search ADS PubMed Keys TG , Berger M , Gerardy-Schahn R . 2012 . A high-throughput screen for polysialyltransferase activity . Anal Biochem . 427 : 60 – 68 . Google Scholar Crossref Search ADS PubMed Kiermaier E , Moussion C , Veldkamp CT , Gerardy-Schahn R , de Vries I , Williams LG , Chaffee GR , Phillips AJ , Freiberger F , Imre R et al. 2016 . Polysialylation controls dendritic cell trafficking by regulating chemokine recognition . Science . 351 : 186 – 190 . Google Scholar Crossref Search ADS PubMed Klock HE , Lesley SA . 2009 . The Polymerase Incomplete Primer Extension (PIPE) method applied to high-throughput cloning and site-directed mutagenesis . Methods Mol Biol . 498 : 91 – 103 . Google Scholar Crossref Search ADS PubMed Laemmli UK . 1970 . Cleavage of structural proteins during the assembly of the head of bacteriophage T4 . Nature . 227 : 680 – 685 . Google Scholar Crossref Search ADS PubMed Lindhout T , Bainbridge CR , Costain WJ , Gilbert M , Wakarchuk WW . 2013 . Biochemical characterization of a polysialyltransferase from Mannheimia haemolytica A2 and comparison to other bacterial polysialyltransferases . PLoS One . 8 : e69888 . Google Scholar Crossref Search ADS PubMed Lindhout T , Iqbal U , Willis LM , Reid AN , Li J , Liu X , Moreno M , Wakarchuk WW . 2011 . Site-specific enzymatic polysialylation of therapeutic proteins using bacterial enzymes . Proc Natl Acad Sci USA . 108 : 7397 – 7402 . Google Scholar Crossref Search ADS PubMed Lizak C , Worrall LJ , Baumann L , Pfleiderer MM , Volkers G , Sun T , Sim L , Wakarchuk W , Withers SG , Strynadka NCJ . 2017 . X-ray crystallographic structure of a bacterial polysialyltransferase provides insight into the biosynthesis of capsular polysialic acid . Sci Rep . 7 : 5842 . Google Scholar Crossref Search ADS PubMed McCarthy PC , Saksena R , Peterson DC , Lee CH , An Y , Cipollo JF , Vann WF . 2013 . Chemoenzymatic synthesis of immunogenic meningococcal group C polysialic acid-tetanus Hc fragment glycoconjugates . Glycoconj J . 30 : 857 – 870 . Google Scholar Crossref Search ADS PubMed McCullum EO , Williams BA , Zhang J , Chaput JC . 2010 . Random mutagenesis by error-prone PCR . Methods Mol Biol . 634 : 103 – 109 . Google Scholar Crossref Search ADS PubMed Morley TJ , Willis LM , Whitfield C , Wakarchuk WW , Withers SG . 2009 . A new sialidase mechanism: bacteriophage K1F endo-sialidase is an inverting glycosidase . J Biol Chem . 284 : 17404 – 17410 . Google Scholar Crossref Search ADS PubMed Muhlenhoff M , Rollenhagen M , Werneburg S , Gerardy-Schahn R , Hildebrandt H . 2013 . Polysialic acid: versatile modification of NCAM, SynCAM 1 and neuropilin-2 . Neurochem Res . 38 : 1134 – 1143 . Google Scholar Crossref Search ADS PubMed Nakayama J , Fukuda M . 1996 . A human polysialyltransferase directs in vitro synthesis of polysialic acid . J Biol Chem . 271 : 1829 – 1832 . Google Scholar Crossref Search ADS PubMed Peterson DC , Arakere G , Vionnet J , McCarthy PC , Vann WF . 2011 . Characterization and acceptor preference of a soluble meningococcal group C polysialyltransferase . J Bacteriol . 193 : 1576 – 1582 . Google Scholar Crossref Search ADS PubMed Puente-Polledo L , Reglero A , Gonzalez-Clemente C , Rodriguez-Aparicio LB , Ferrero MA . 1998 . Biochemical conditions for the production of polysialic acid by Pasteurella haemolytica A2 . Glycoconj J . 15 : 855 – 861 . Google Scholar Crossref Search ADS PubMed Robert X , Gouet P . 2014 . Deciphering key features in protein structures with the new ENDscript server . Nucleic Acids Res . 42 : W320 – W324 . Google Scholar Crossref Search ADS PubMed Sambrook J , Fritsch EF , Maniatis T . 1989 . Molecular Cloning: A Laboratory Manual , 2nd ed . Cold Spring Harbor, NY : Cold Spring Harbor Laboratory Press . Sato C , Kitajima K . 2013 . Impact of structural aberrancy of polysialic acid and its synthetic enzyme ST8SIA2 in schizophrenia . Front Cell Neurosci . 7 : 61 . Google Scholar PubMed Sawitzke JA , Thomason LC , Costantino N , Bubunenko M , Datta S , Court DL . 2007 . Recombineering: in vivo genetic engineering in E. coli, S. enterica, and beyond . Methods Enzymol . 421 : 171 – 199 . Google Scholar Crossref Search ADS PubMed Sharan SK , Thomason LC , Kuznetsov SG , Court DL . 2009 . Recombineering: a homologous recombination-based method of genetic engineering . Nat Protoc . 4 : 206 – 223 . Google Scholar Crossref Search ADS PubMed Silver RP , Finn CW , Vann WF , Aaronson W , Schneerson R , Kretschmer PJ , Garon CF . 1981 . Molecular cloning of the K1 capsular polysaccharide genes of E. coli . Nature . 289 : 696 – 698 . Google Scholar Crossref Search ADS PubMed Vimr ER , Troy FA . 1985 . Regulation of sialic acid metabolism in Escherichia coli: role of N-acylneuraminate pyruvate-lyase . J Bacteriol . 164 : 854 – 860 . Google Scholar PubMed Werneburg S , Buettner FF , Erben L , Mathews M , Neumann H , Muhlenhoff M , Hildebrandt H . 2016 . Polysialylation and lipopolysaccharide-induced shedding of E-selectin ligand-1 and neuropilin-2 by microglia and THP-1 macrophages . Glia . 64 : 1314 – 1330 . Google Scholar Crossref Search ADS PubMed Willis LM , Gilbert M , Karwaski MF , Blanchard MC , Wakarchuk WW . 2008 . Characterization of the alpha-2,8-polysialyltransferase from Neisseria meningitidis with synthetic acceptors, and the development of a self-priming polysialyltransferase fusion enzyme . Glycobiology . 18 : 177 – 186 . Google Scholar Crossref Search ADS PubMed Wu JR , Lin Y , Zheng ZY , Lin CC , Zhan XB , Shen YQ . 2010 . Improvement of the CuZn-superoxide dismutase enzyme activity and stability as a therapeutic agent by modification with polysialic acids . Biotechnol Lett . 32 : 1939 – 1945 . Google Scholar Crossref Search ADS PubMed Wyganowski KT , Kaltenbach M , Tokuriki N . 2013 . GroEL/ES buffering and compensatory mutations promote protein evolution by stabilizing folding intermediates . J Mol Biol . 425 : 3403 – 3414 . Google Scholar Crossref Search ADS PubMed Yabe U , Sato C , Matsuda T , Kitajima K . 2003 . Polysialic acid in human milk. CD36 is a new member of mammalian polysialic acid-containing glycoprotein . J Biol Chem . 278 : 13875 – 13880 . Google Scholar Crossref Search ADS PubMed Yang G , Rich JR , Gilbert M , Wakarchuk WW , Feng Y , Withers SG . 2010 . Fluorescence activated cell sorting as a general ultra-high-throughput screening method for directed evolution of glycosyltransferases . J Am Chem Soc . 132 : 10570 – 10577 . Google Scholar Crossref Search ADS PubMed Yang G , Withers SG . 2009 . Ultrahigh-throughput FACS-based screening for directed enzyme evolution . Chembiochem . 10 : 2704 – 2715 . Google Scholar Crossref Search ADS PubMed You C , Percival Zhang YH . 2012 . Easy preparation of a large-size random gene mutagenesis library in Escherichia coli . Anal Biochem . 428 : 7 – 12 . Google Scholar Crossref Search ADS PubMed Yu CC , Hill T , Kwan DH , Chen HM , Lin CC , Wakarchuk W , Withers SG . 2014 . A plate-based high-throughput activity assay for polysialyltransferase from Neisseria meningitidis . Anal Biochem . 444 : 67 – 74 . Google Scholar Crossref Search ADS PubMed Yu CC , Huang LD , Kwan DH , Wakarchuk WW , Withers SG , Lin CC . 2013 . A glyco-gold nanoparticle based assay for alpha-2,8-polysialyltransferase from Neisseria meningitidis . Chem Commun . 49 : 10166 – 10168 . Google Scholar Crossref Search ADS Zapater JL , Colley KJ . 2012 . Sequences prior to conserved catalytic motifs of polysialyltransferase ST8Sia IV are required for substrate recognition . J Biol Chem . 287 : 6441 – 6453 . Google Scholar Crossref Search ADS PubMed Author notes Bettina Janesch and Lars Baumann authors contributed equally Department of NanoBiotechnology, Institute for Biologically Inspired Materials, NanoGlycobiology unit, Universität für Bodenkultur Wien, Muthgasse 11, A-1190 Vienna, Austria Department of Medicinal Chemistry, Bayer AG Pharmaceuticals, Aprather Weg 18 A, 42096 Wuppertal, Germany Department of Biological Sciences, University of Alberta, Edmonton, AB, T6G2E9 © The Author(s) 2019. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)
TI - Directed evolution of bacterial polysialyltransferases
JF - Glycobiology
DO - 10.1093/glycob/cwz021
DA - 2019-07-01
UR - https://www.deepdyve.com/lp/oxford-university-press/directed-evolution-of-bacterial-polysialyltransferases-xMNlp707Xe
SP - 588
VL - 29
IS - 7
DP - DeepDyve
ER -