Sortase A-mediated crosslinked short-chain dehydrogenases/reductases as novel biocatalysts with improved thermostability and catalytic efficiency

(S)-carbonyl reductase II (SCRII) from Candida parapsilosis is a short-chain alcohol dehydrogenase/reductase. It catalyses the conversion of 2-hydroxyacetophenone to (S)-1-phenyl-1,2-ethanediol with low efficiency. Sortase was reported as a molecular “stapler” for site-specific protein conjugation to strengthen or add protein functionality. Here, we describe Staphylococcus aureus sortase A-mediated crosslinking of SCRII to produce stable catalysts for efficient biotransformation. Via a native N-terminal glycine and an added GGGGSLPETGG peptide at C-terminus of SCRII, SCRII subunits were conjugated by sortase A to form crosslinked SCRII, mainly dimers and trimers. The crosslinked SCRII showed over 6-fold and 4-fold increases, respectively, in activity and k cat/K m values toward 2-hydroxyacetophenone compared with wild-type SCRII. Moreover, crosslinked SCRII was much more thermostable with its denaturation temperature (Tm) increased to 60 °C. Biotransformation result showed that crosslinked SCRII gave a product optical purity of 100% and a yield of >99.9% within 3 h, a 16-fold decrease in transformation duration with respect to Escherichia coli/pET-SCRII. Sortase A-catalysed ligation also obviously improved Tms and product yields of eight other short-chain alcohol dehydrogenases/reductases. This work demonstrates a generic technology to improve enzyme function and thermostability through sortase A-mediated crosslinking of oxidoreductases.

protein-lipid 10 , and protein-surface conjugates 11 and cyclised peptides 12 . Wu et al. used SrtA for chemoenzymatic synthesis of intramolecular head-to-tail bifunctional peptides and glycopeptides 13 . Antos et al. described sortase-catalysed transpeptidation as a route to circular proteins with enhanced resistance to denaturation compared with linear counterparts 14 .
We previously reported an NADPH-dependent (S)-carbonyl reductase (SCR) from Candida parapsilosis CCTCC M203011 15 . It is a member of the short-chain dehydrogenases/reductases family with determined crystal structure as tetramer in solution. Recently, its isozyme, the other novel (S)-carbonyl reductase SCRII was employed to catalyse the asymmetric reduction of 2-hydroxyacetophenone (2-HAP) to (S)-1-phenyl-1,2-ethanediol ((S)-PED) in 48 h (Fig. 1) 16 . However, low thermostability and poor catalytic activity would limit future biocatalytic application of SCRII. Thus, further research is urgently required to undertake protein engineering of SCRII to produce stable catalyst with improved catalytic performance.
Herein, we report for the first time SrtA-mediated crosslinked short-chain dehydrogenases/reductases as robust biocatalysts for efficient chiral synthesis. We carried out sortase-mediated ligation inside SCRII to generate artificial crosslinked protein. The flexible motif GGGGSLPETGG was fused to the C-terminus of SCRII to generate SCRII-GGGGSLPETGG. Via a native glycine at the N-terminus of SCRII, protein ligations between SCRII-GGGGSLPETGG subunits were achieved using S. aureus SrtA as catalyst (Fig. 1). The crosslinked SCRII showed significantly improved functionalities and remarkably improved thermal stability. Furthermore, the crosslinked SCRII also stimulated significant increases in yield of (S)-PED, and simultaneously reduced the  Table 1. Enzymes in this work.

Figure 1.
Schematic illustration showing the SrtA-mediated crosslinking of SCRII. GGGGSLPETGG tag was fused into the C-terminal end of SCRII. SrtA recognized and cleaved the tag, followed by the attack from the glycine at the N-terminal end to form crosslinked protein complexes (mainly dimers and trimers). The covalently bond dimers and trimers were used as units to form further oligomerisation, of which the intermolecular conformation was assumed in this work. The dotted line meaned possible covalent bond to form cyclic crosslinked SCRII. The crosslinked SCRII was subsequently applied to the asymmetric reduction of 2-HAP to (S)-PED.

Results and Discussion
SrtA-mediated ligation of SCRII to form crosslinked SCRII. A suitably flexible GGGGS motif was fused to the sorting signal LPETGG that can be recognised by SrtA. The resulting GGGGSLPETGG motif was added to the C-terminus of wild-type (WT) SCRII to generate SCRII-GGGGSLPETGG which was referred as SCRII-mtf. Ilangovan et al. confirmed the core domain of SrtA (amino acids 60-206) exhibited the same activity as wild-type sortase 17 . In this work, the truncated SrtA protein was overexpressed and purified with an estimated yield of about 50 mg/L culture and purity of >90% ( Supplementary Fig. S1). The SrtA-mediated ligation was initiated by the addition of Ca 2+ to a mixture of purified SCRII-mtf (~30 kDa) and SrtA (~23 kDa). To determine the optimal temperature for SrtA-catalysed crosslinking of SCRII, the mixture was incubated at temperatures in the range 10-35 °C for 8 h. SDS-PAGE analysis revealed the presence of two major bands corresponding to the expected sizes of dimers and trimers of SCRII (about 66 and 99 kDa) (Fig. 2a). Matrix assisted laser desorption/ionisation-time of flight mass spectrometry (MALDI-TOF-MS) analysis further confirmed the appearance of dimers and trimers in the ligation mixture of SCRII (Fig. 2b). Two byproduct bands were also observed below the band corresponding to SCRII-mtf, which were excised for MALDI-TOF-MS. Peptide mass fingerprinting showed that both of them were most like SCRII ( Supplementary Fig. S2). According to previous reports 14,18 , they could be assigned as SCRII-GGGGSLPET and cyclised SCRII-GGGGSLPET. Based on the amount of ligation product, the optimal temperature for the SrtA-mediated ligation was determined to be 25 °C. SDS-PAGE analysis with Bandscan 5.0 revealed that the relative intensity of dimers and trimers accounted for >90% of the ligation products, not including SCRII-mtf and bands below it. As a control, we did not observe any product in SrtA-mediated ligation of WT SCRII.
Interestingly, SCRII-mtf and two byproducts (supposed to be SCRII-GGGGSLPET and cyclised SCRII-GGGGSLPET) were almost completely consumed after 36 h at 25 °C. The behavior was contrary to the results of reports on the SrtA-mediated ligation of eGFP. Antos et al. found that bifunctional eGFP strongly favoured cyclisation with little evidence of any oligomerisation 14 . There are two key differences between their experiments and ours: (1) different proteins have different distances between their N-and C-termini. Previous work 15 revealed the structure of SCR which had 86% similarity to SCRII. The distance between N-and C-termini was not close, which reduced the chance of cyclisation occurring; and (2) SrtA-catalysed cyclisation is a reversible reaction 19 . The cleavage of a cyclised subunit would generate a thioacyl intermediate which could be attacked by the N-terminal glycine of another protein molecule to form a crosslinked complex. In contrast, the cross-linked SCRII would be difficult to be cleaved by SrtA because of steric hindrance of the LPETG motif inside the complex.
To remove SrtA and residual SCRII-mtf from SrtA-mediated ligation products, the ligation mixtures were subjected to size exclusion chromatography. As shown in Fig. 2c, the crosslinked SCRII were corresponding to two peaks in the range of 120 kDa-180 kDa. These results indicated that the covalently-linked dimers and trimers formed tetramers and hexamers. Besides, the covalently-linked dimer and trimer were found in both of the two peaks. We inferred that a dynamic equilibrium existed between the tetramer and hexamer. For example, two hexamers may consist of six dimers which could be recombined into three tetramers. The process was reversible, dynamic and happening all the time in solution. Therefore, the products cannot be fully separated. SCRII-mtf and WT SCRII had narrow molecular mass distributions with a single peak observed at ~120 kDa and ~115 kDa (Fig. 2c), suggesting they both formed a tetramer in solution. Our previous work revealed that SCR, the isoenzyme of SCRII, existed as tetrameric form in solution. We conclude that SCRII-mtf and WT SCRII formed similar tetrameric conformations to each other. However, the conformation of the tetramers and hexamers formed by crosslinked SCRII could not be determined.
SrtA-mediated crosslinked SCRII has improved thermostability without secondary structural changes. The purified crosslinked SCRII, SCRII-mtf and WT SCRII were analysed by far-UV circular dichroism spectroscopy (Fig. 3a) to determine whether any change occurred in secondary structure content. All three samples showed a positive and a negative band around 190 and 208 nm which are characteristic of an α-helical structure [20][21][22] . The deconvolved data showed that the crosslinked SCRII, SCRII-mtf and WT SCRII enzymes shared similar secondary structures (see Supplementary Table S1), indicating that no significant intramolecular conformational changes were induced by the SrtA-mediated crosslinking of SCRII or introduction of the GGGGSLPETGG tag at its C-terminus. There are no sufficient data to determine whether intermolecular conformational changes occurred.
The thermal stabilities of the three proteins were measured by monitoring the disappearance of their characteristic α-helical peak at 208 nm. Figure 3b showed the thermal shift curves of crosslinked SCRII, SCRII-mtf and WT SCRII. The denaturation temperatures (T m ) of these proteins were obtained by calculating the maximum slopes of their denaturation curves, also known as the denaturation midpoint 23 . The secondary structures of WT SCRII and SCRII-mtf progressively disappeared at temperatures >45 °C. There was no obvious difference between T m s of WT SCRII and SCRII-mtf (50 °C), suggesting the introduced C-terminal tag had little effects on enzyme thermostabilily. In contrast, the secondary structure of crosslinked SCRII remained intact up to 55 °C and then rapidly disappeared, with a T m of 60 °C. Therefore, our attempt to construct SrtA-mediated protein conjugates enhanced the protein thermal stability. Protein thermal stability may be associated with higher α-helix content 24 while our examination of secondary structures suggested that is not the explanation here. Clantin et al. reported that oligomerisation could help achieve thermal stabilisation of a protein, but that oligomerisation was of proteins assembled together to form an interface without covalent linkages between monomers 25 . The improved thermal stability in our work may be attributed to the covalent bonds between SCRII-mtf subunits, which might help to reinforce the structure of the complexes. Popp et al. demonstrated the cyclisation of four-helix bundle cytokines by covalently joining the N-and C-termini of the proteins with a resulting improvement in thermal stability 26 . Here, the dimers and trimers may also exist in cyclic forms which could increase thermal stabilisation Figure 3. CD spectra (a) and thermal shift assays (b) of WT SCRII, SCRII-mtf and crosslinked SCRII. The CD spectra were recorded by measuring the ellipticity as a function of wavelength at 0.1-nm increment between 190 and 250 nm at 20 °C. Thermal shift assay was conducted by heating the three samples from 20 to 90 °C and then cooling to 20 °C by a Jasco programmable Peltier element with a rate of 1 °C/min. The wavelength of 209 nm that characterized the α-helix within protein was used to monitor the unfolding rate of the protein structures. The buffer (5 mM phosphate buffer pH 6.0, 0.1 mg/mL enzymes.) showed pH shift with no more than 0.5 while it was heated from 20 °C to 100 °C. by possible formation of salt bridges and the consolidation of protein structure, though this speculation remains to be assessed in future research.
Crosslinked SCRII has significantly improved chiral synthesis efficiency compared with WT SCRII. We previously reported that WT SCRII enzymes showed the highest specific activity at 35 °C 16 . Here, the optimal reaction temperature for crosslinked SCRII toward 2-HAP was determined from 20 °C to 70 °C. Meanwhile, WT SCRII and SCRII-mtf enzymes were tested as controls. As shown in Fig. 4, the crosslinked complexes showed their highest reductive activity at 50 °C. At 35 °C, the crosslinked SCRII showed a specific activity of 38.50 μmol/ (min·mg), about 95% of the maximum observed activity and 6.2-fold higher than that of WT SCRII (Fig. 5). To determine the shortest time needed for the biotransformation of 2-HAP, sampling was performed at various times with crosslinked SCRII, SCRII-mtf and WT SCRII as catalysts. Crosslinked SCRII completed the asymmetric reduction of 2-HAP to (S)-PED with an optical purity of >99% and a yield >99% within 3 h (Supplementary Fig. S3). SCRII-mtf and SCRII showed similar performance to each other, with only about 55% yield in 3 h and 88% yield in 6 h. It was worth mentioning that they could not fully complete the biotransformation, with a maximum yield of about 98%. In addition, compared with whole cell biotransformation by E. coli/pET-SCRII 16 , crosslinked SCRII significantly decreased the time for biotransformation of (S)-PED, from 48 h to 3 h. To determine the effects of crosslinked SCRII on the substrate specificity, we investigated the asymmetric reduction of 13 different ketones bearing an aryl group. The specific activity of the crosslinked SCRII protein was around 2-to 8-fold higher than that of WT SCRII and SCRII-mtf proteins, except toward substrates 6a and 13a. The biotransformation results revealed that crosslinked SCRII improved yields by 13.8-209% compared with WT SCRII, without any reduction in the enantiomeric excess (ee) values of the chiral products. The crosslinked SCRII, SCRII-mtf and WT SCRII proteins showed no specific activity in the asymmetric reduction of substrates 6a or 13a. These results suggested that the SrtA-mediated conjugation did not change the essential enzyme function, but improved it.
Kinetics demonstrate increased affinity between crosslinked SCRII and 2-HAP. WT SCRII enzymes favour NADPH as a cofactor over NADH 16 . Therefore, the kinetic effects of crosslinked SCRII, SCRII-mtf, and WT SCRII were evaluated in the reduction of substrates with NADPH. V max and K m values were derived from Michaelis-Menten and Lineweaver-Burk plots (Supplementary Table S2). k cat and k cat /K m values are summarised in Table 2. Taking reduction of 2-HAP (substrate 2a), which is the favoured substrate of WT SCRII, as an example (Fig. 6), the K m and k cat values of the SrtA-mediated crosslinked SCRII protein were 3.3-fold lower and 1.3-fold higher than those of WT SCRII, respectively. Crosslinked SCRII displayed higher catalytic efficiency than WT SCRII toward all 12 ketone substrates (6a and 13a not included) and the k cat /K m values increased by 1.4-fold (7a) to 4.4-fold (12a). Meanwhile, SCRII-mtf displayed parallel behaviour with WT SCRII toward the tested substrates, indicating that the C-terminal tag had little impact on the enzyme activity.
These improvements in the relative catalytic efficiency of crosslinked SCRII were attributable to stronger binding between enzyme and substrate. Crosslinked SCRII tended to show lower K m values which probably associated with increased catalytic efficiency. For SCRII, crosslinking may have changed the intermolecular conformation or molecular interaction, resulting in the lower observed K m values. Moreover, the possible cyclisation of crosslinked SCRII may help reduce conformational freedom and create constrained structural frameworks which often confer high receptor binding affinity and specificity 27, 28 . SrtA-mediated ligation of SDRs remarkably increases transformation efficiency with improved thermostability. The sortase-mediated method newly developed here was extended to the crosslinking of another eight SDRs (Table 1) for efficient chiral synthesis. The oxidoreductases ADHR, C1, C2, CR2, CR4, S1, SCR1 and SCR3 were selected to construct corresponding crosslinked oxidoreductases using SrtA-mediated method. The GGGGSLPETGG tag was added to their C-terminus. Unlike WT SCRII, these eight oxidoreductases did not have a native glycine at the N-terminus. Thus, a GGG tag was added to N-terminus of oxidoreductases' Figure 4. Effects of temperature on enzyme activity for WT SCRII, SCRII-mtf and crosslinked SCRII. Enzyme assay was performed in 100 mM potassium phosphate buffer (pH 6.0), 0.5 mM NADPH, 5 mM substrate, and appropriate enzymes (mg/mL) at 20-70 °C. The relative activity was described as percentage of maximum activity under experimental conditions. All experiments were repeated three times.
genes. After overexpression, the eight oxidoreductases were successfully prepared with estimated purity >95% (Supplementary Fig. S4). Superdex 200 size exclusion chromatography revealed that ADHR, S1, SCR1 and SCR3 Figure 5. Substrate specificities of WT SCRII, SCRII-mtf and SrtA-mediated Crosslinked SCRII. a Standard assay conditions: 100 mM phosphate buffer (pH 6.0), 0.5 mM NADPH, 5 mM substrate, 1 mg/mL enzymes at 35 °C. The data was the representative of three independent experiments and standard errors were not more than 5%. b The biotransformation was conducted with 5 g/L substrate and sufficient NADPH for 6 h. S, substrate; Yd, yield; Cf, configuration, ee, enantiomeric excess.  Table 2. Kinetic parameters for WT SCRII, SCRII-mtf and crosslinked SCRII towaids the asymmertric reduction of aryl ketones. Each value was calculated depending on three independent measurements and all standard errors of fits were not more than 5%.

No
were tetrameric protein while C1, C2 and CR2 existed as monomeric form. The purified CR4 showed a subunit molecular mass of 32 kDa on SDS-PAGE but with an estimated native enzyme molecule weight of around 85 kDa by size exclusion chromatography. These results were consistent with previous reports 29- 35 . Based on the optimised conditions for SCRII ligation, the eight oxidoreductases were incubated with SrtA for 36 h at 25 °C. As shown in Fig. 7, all eight enzymes formed crosslinked oxidoreductases. Judged by molecular weights, most of the products were dimers and trimers while only dimers were observed for SCR1 and ADHR. MALDI-TOF-MS results also confirmed the ligation products (Supplementary Figs S5-12) Meanwhile, similar with SrtA-mediated crosslinking of SCRII, S1, SCR1 and SCR3 formed byproducts supposed to be cyclised monomers which were observed in SDS-PAGE below the position of the respective oxidoreductase-mtf. C1, C2, CR2, CR4 and ADHR formed almost no supposed cyclised monomers. Different distances between N-and C-termini in different proteins probably resulted in the different cyclisation behaviour. It is worth mentioning that monomeric ADHR was almost completely consumed after 36 h, suggesting that crosslinked ADHR might show similar steric hindrance of SrtA to that we proposed for crosslinked SCRII. The crosslinked oxidoreductases generated were separated from residual mono-oxidoreductases and SrtA by size exclusion chromatography. The crosslinked ADHR, CR4, S1, SCR1 and SCR3 may exist as tetramer and hexamer like crosslinked SCRII ( Supplementary  Figs S13-17). However, crosslinked C1, C2 and CR2 displayed no further oligomerisation according to the estimated molecular mass in size exclusion chromatography (Supplementary Figs S18-20). Perhaps the different quaternary structure of native enzymes caused different result. The denaturation temperatures of the 8 crosslinked oxidoreductases were measured by thermal shift assays. They were much more stable than their corresponding WT enzymes, with T m values increased by 5-12 °C (Fig. 8). For example, the T m value of crosslinked C1 was 69 °C, 12 °C higher than that of WT C1. This increase in the thermal stability could be due to more covalent bonds between the monomers of crosslinked C1 enzyme than WT enzyme according to Xiang et al. 36 .
We further determined the stereoselective reaction efficiency of the eight oxidoreductases and corresponding crosslinked enzymes toward their preferred substrates. The SrtA-mediated crosslinking resulted in 5-to 11-fold increases in the specific activities of these enzymes (Table 3). For example, the activity of crosslinked S1 in the reduction of 11a was 10.8-fold greater than that of WT S1. Besides, the biotransformation was carried out and the products were examined by HPLC. All crosslinked oxidoreductases and their WT enzymes catalyzed the biotransformation of chiral alcohols with a similar optical purity of over 99% except for CR2 and SCR3 toward 1a and 2a, respectively. The crosslinked oxidoreductases improved the yield of chiral products by 25-285% compared to WT enzymes. They did not show any changes in the stereoselectivity of the reduction reactions, but performed much more effectively than WT enzymes. Kim et al. suggested that ligation brought target proteins together to form complementary interfaces, promoting higher catalytic efficiency through substrate channelling, high local substrate and enzyme concentrations and compartmentalisation 37    protein 38 . Like crosslinked SCRII, we suspected that the 8 crosslinked oxidoreductases possibly existed as cyclic form to have their structural frameworks constrained for high receptor binding affinity, specificity and enhanced stability. The improved thermal stability and catalytic efficiency of crosslinked oxidoreductases should also possibly be associated with the cyclic intermolecular conformation.

Conclusion
In summary, we have developed SrtA-mediated crosslinked short-chain dehydrogenases/reductases as novel biocatalysts with improved thermostability for efficient stereoselective reactions. To the best of our knowledge, this work first describes a generic technology platform for the preparation of stable biocatalysts for efficient chiral synthesis, which will be of industrial interest.
Protein expression and purification. The protein expression and purification was carried out as described previously with minor modification 41  MALDI-TOF-MS analysis of ligation product. The bands below SCRII-mtf on SDS-PAGE gel were excised for MALDI-TOF-MS analysis with Bruker Daltonics FLEX (Billerica, USA), followed by peptide mass fingerprinting analysis with Proteomics solution I system. The peptide mass data were used to query the Mascot database (http://www.matrixscience.com). The purified crosslinked oxidoreductases were dissolved in distilled water (10 mg/ml). 2,5-Dihydroxybenzoic acid was prepared as 1% (w/v) solution and used as a matrix. Then the sample solution and matrix were mixed in equal volumes (1.0 μl each), spotted on a sample plate followed by air-drying until homogeneous crystals formed. The dried sample on the plate was then applied to the mass spectrometer and analyzed with an accelerated voltage and reflective voltage of 19 kV and 16 kV, respectively. Enzyme assay and kinetic determination. The enzyme assay mixture in 250 μL comprised 100 mM potassium phosphate buffer (pH 6.0), 0.5 mM NADPH, 5 mM substrate, and enzymes (1 mg/mL). The reductive activity on substrates was measured at 20-70 °C by recording the rate of change in NADPH absorbance at 340 nm. One unit of enzyme activity is defined as the amount of enzyme catalyzing the oxidation of 1 μmol of NADPH per minute under measurement condition. The relative activity was expressed as percentage of maximum activity under experimental conditions.
The kinetic parameters of WT SCRII, SCRII-mtf and crosslinked SCRII were measured and calculated using a Beckman DU-7500 spectrophotometer with a Multicomponent/SCA/Kinetics Plus software package and a thermostated circulating water bath. Substrate (0.5 to 20 mM), enzyme (~30 μM, 1 mg/mL), and cofactor NADPH (5 mM) in 100 mM potassium phosphate buffer (pH 6.0) were used for a series of assays. Each value was calculated depending on three independent measurements and all standard errors of fits were not more than 5%. The data obtained were fitted to the function v = V max [s]/(K m + [s]). Kinetic parameters were derived from Michaelis-Menten plots and Lineweaver-Burk.
Biotransformation and analytical methods. The biotransformation was carried out as described previously with minor modification 42 . The reaction mixture in a 2-mL volume consisted of 100 mM potassium phosphate buffer (pH 6.0/7.0), 5 g/L substrate, sufficient NADPH, and pure enzymes (1 mg/mL). The reactions were carried out at 35 °C for 6 h with shaking at 150 rpm respectively. Time sampling was performed every 30 min if necessary. The product was extracted with ethyl acetate, and the organic layer was used for analysis. The optical purity and yield of product were determined by high-performance liquid chromatography on a Chiralcel OB-H column (Daicel Chemical Ind. Ltd., Japan). Method details including flow rates, mobile phase and retention time of products are shown in Supplementary Figs S21-S32.

Circular dichroism spectroscopy and thermal shift assay. Circular dichroism (CD) measurements
were performed with a Jasco J720 spectropolarimeter (Jasco, Inc., Easton, USA). Data of absorbed wavelength from 190 to 250 nm were collected within a phosphate buffer (pH 6.0) using the following instrument settings (for an average of 30 scans): response, 1 s; sensitivity, 100 mdeg; speed, 50 nm/min, average of 30 scans. The protein concentration was ~3 μM (~0.1 mg/mL) in 100 mM potassium phosphate buffer (pH 6.0). The CD data was further deconvolved by K2d method (Dichroweb server) to determine the content of secondary structures. The thermal denaturation was carried out by observing the impact of temperature on protein secondary structure. The protein samples were heated from 20 °C to 90 °C and then cooled to 20 °C by a Jasco programmable Peltier element with a heating/cooling rate of 1 °C/min. The wavelength of 209 nm that characterized the α-helix within protein was used to monitor the unfolding rate of the protein structures. The proteins were diluted to 0.1 mg/mL within 5 mM potassium phosphate buffer (pH 6.0/7.0) which had been conducted for a heating study to observe pH shift. The solvent contribution was discounted to correct the obtained spectra. The denaturation temperature (T m ) was calculated according to the maximum slope that signified the swift unfolding of secondary structures.