Enhancing Multistep Reactions: Biomimetic Design of Substrate Channeling Using P22 Virus‐Like Particles

Abstract Many biocatalytic processes inside cells employ substrate channeling to control the diffusion of intermediates for improved efficiency of enzymatic cascade reactions. This inspirational mechanism offers a strategy for increasing efficiency of multistep biocatalysis, especially where the intermediates are expensive cofactors that require continuous regeneration. However, it is challenging to achieve substrate channeling artificially in vitro due to fast diffusion of small molecules. By mimicking some naturally occurring metabolons, nanoreactors are developed using P22 virus‐like particles (VLPs), which enhance the efficiency of nicotinamide adenine dinucleotide (NAD)‐dependent multistep biocatalysis by substrate channeling. In this design, NAD‐dependent enzyme partners are coencapsulated inside the VLPs, while the cofactor is covalently tethered to the capsid interior through swing arms. The crowded environment inside the VLPs induces colocalization of the enzymes and the immobilized NAD, which shuttles between the enzymes for in situ regeneration without diffusing into the bulk solution. The modularity of the nanoreactors allows to tune their composition and consequently their overall activity, and also remodel them for different reactions by altering enzyme partners. Given the plasticity and versatility, P22 VLPs possess great potential for developing functional materials capable of multistep biotransformations with advantageous properties, including enhanced efficiency and economical usage of enzyme cofactors.

Supplementary Fig. 7. Cyro-EM structural model of P22 capsid that shows the location of the Ser39 (a) and Thr10 (b) of CP (PDB: 2XYY). Both residues are located at interior side of the P22 capsid by looking at the edge of the particle (noted in blue arrows.). The first 9 amino acid residues of CP are not resolved in the structure, likely due to the flexibility of the N-terminus. Supplementary Fig. 8. Characterization of P22 capsid assembled from CPN-ext by SEC-QELS (a) (mean ± s.e.m., n=3) and TEM (b).
Supplementary Fig. 9. SDS-PAGE and densitometry analysis of in vitro assembled VLP nanoreactors. (a) SDS-PAGE. NAD-CPN-ext and NAD-CPS39C do not show distinct bands but smear where at least two bands can be vaguely seen, consistent with LC-MS data (supplementary Fig. 5). Concentration of CPS39C standards was determined by A280 (ε = 44920 M -1 cm -1 ). Concentration of nanoreactors was determined by Bradford Assay (CPS39C as standard). (b) Densitometry plot of CPS39C (left). The peak area of CPS39C in the densitometry plot shows a linear response with respect to concentration (right). (c) Densitometry plot of CPN-ext VLP nanoreactors (left) and CPS39C VLP nanoreactors (right). Analysis: The peak areas in the densitometry plot of the NAD-CPN-ext band and the NAD-CPS39C band were used to estimate the total concentration of NAD in the kinetic analysis of the nanoreactors. This quantification is only an estimation, since the Gaussian fitting cannot be applied due to noisy baseline and the tailing of the protein bands. Method 1: The total CP area (NAD-CP, linker-CP, and unlabeled CP) was plugged into the equation of the calibration curve obtained in (b) to get the [CP] in the sample. Then the [NAD-CP] is calculated by multiplying [CP] by the percentage of NAD-CP area out of total CP area. This method estimates 1.70 μM NAD in 12 μM NAD-CPN-ext VLP nanoreactor, and 1.52 μM NAD in 12 μM NAD-CPN-ext VLP nanoreactor. In the kinetic study, because 1.8 μM nanoreactor is used, the total NAD concentration is estimated as 0.255 μM for NAD-CPN-ext VLP nanoreactor, and 0.228 μM for NAD-CPS39C VLP nanoreactor. Method 2: The [NAD-CP] in 12 μM is calculated by multiplying 12 μM by the percentage of NAD-CP area out of total protein area (including both CP and enzymes). This method estimates 3.21 μM NAD in 12 μM NAD-CPN-ext VLP nanoreactor and 2.40 μM NAD in 12 μM NAD-CPN-ext VLP nanoreactor. In the kinetic study, because 1.8 μM nanoreactor is used, the total NAD concentration is estimated as 0.481 μM for NAD-CPN-ext VLP nanoreactor and 0.359 μM for NAD-CPS39C VLP nanoreactor. In this method, the calibration curve in (b) is not used. From the two methods of analysis, the average of total [NAD] in the kinetic study is estimated as 0.368 μM for NAD-CPN-ext VLP nanoreactor and 0.294 μM for NAD-CPS39C VLP nanoreactor. Note that, the total protein concentration of both CPS39C and CPN-ext particles was determined by Bradford assay using CPS39C as standard (quantified by A280). (Due to nucleic acid contamination (supplementary Fig. 31), we were not able to obtain CPN-ext that can be quantified by A280, and thus CPN-ext was not used as standards in Bradford assay to determine the total protein concentration of CPN-ext particles.) Because the N-terminal extension of CPN-ext increases the binding with and Coomassie stain, the concentration NAD-CPN-ext is overestimated. Based on this analysis, 0.3 μM is a good estimation of total [NAD] in solution in the kinetic study for both NAD-CPN-ext nanoreactors and NAD-CPS39C particles. Supplementary Fig. 10. TEM images of in vitro assembled VLP nanoreactors. Scale bar: 200 nm. Supplementary Fig. 11. Characterization of in vitro assembled VLP nanoreactors by SEC-MALS-QELS (mean ± s.e.m., n=3). The molecular weights of nanoreactors increased by ~7-10 MDa compared to the expected molecular weight of ES. CPN-ext ES showed an molecular weight higher than expected due to nucleic acid contamination, while nucleic acid contamination is much less for CPN-ext nanoreactors (supplementary Fig. 31). Analysis: Each component of the nanoreactors is quantified by the peak area in the densitometry plot. This quantification is only an estimation, since the Gaussian fitting cannot be applied due to noisy baseline and the tailing of the protein bands. Supplementary Fig. 15 The apparent turnover of PtDH is about 40 fold higher than that of AdhD.     The equation (6) is used to calculate the concentration of NADH at the steady state. The phosphate production reflects the rate of PtDH, and consequently NADH production as well. Clearly, at the end of the burst phase, the amount of PtDH was produced (Fig. 2c) is much lower than the concentration of NADH required in the steady state. This suggests that, at the end of the burst phase, the system did not reach steady state yet and was still far away from the steady state.

Methods
No unexpected or unusually high safety hazards were encountered.

Recombinant protein purification
The cells were thawed from -80℃, and resuspended in the lysis buffer (50 mM sodium phosphate, 100 mM NaCl; pH 7.8 for enzymes; pH 7.0 supplemented with 2 mM EDTA for P22 VLPs). Lysozyme (1.5 mg), DNase (2 mg), and RNase (3 mg) were added to the cells, followed by incubation at room temperature (r.t.) for 30 min and then sonicated for 2 min at 50% amplitude twice. The cell lysate was obtained by collecting the supernatant after centrifugation at 12000 rpm, 4℃ for 45 min.
To purify the enzymes, the cell lysate was filtered through a 0.45 μm syringe filter and applied to a HisTrap HP column (Cytiva) using a Bio-rad NGC FPLC. The enzyme was purified and eluted with an imidazole gradient (0-500 mM in lysis buffer), and then dialyzed into the assembly buffer (50 mM Tris, 25 mM sodium chloride, 2 mM EDTA, 3 mM 2-mercaptoethanol, 1% glycerol, pH 7.6). The purified enzymes were stored at 4℃. To purify P22 VLPs, the cell lysate was layered onto a 35% (w/v) sucrose cushion and ultracentrifuged using Thermo Scientific Sorvall WX Ultra Series Centrifuge at 45000 rpm, 50 min, 4℃ (rotor: F50L-8ⅹ39). The pellet was resuspended in lysis buffer, and then purified through a Sephacryl S-500 column (GE Healthcare Life Sciences) using a Bio-rad Biologic Duoflow FPLC. To obtain the empty shell (ES) form of the VLPs, SP was removed by treating the purified VLPs with 0.5 M GdnHCl in lysis buffer for 1 h and ultracentrifuged (45000 rpm, 50 min, 4℃) to pellet the VLP. Then the pellet was dissolved in the lysis buffer. This process was usually repeated at least three times in order to remove all SP. The VLP ES was stored at 4℃.

Synthesis of NAD-Maleimide
The synthetic route (supplementary Fig. 1) and methods for NAD-maleimide synthesis were adapted from a previously reported method. 2 To synthesize 8-Br-NAD + , 1 g β-nicotinamide adenine dinucleotide (Chem-Impex, Cat# 00229) was dissolved in 15 mL 500 mM sodium acetate buffer (pH 4.5) in a 100-mL flask. 0.4 mL Br2 was added dropwise with stirring at the highest speed. Then the system was then kept with stirring for ~2 hrs at room temperature (r.t.). After the reaction completed, the UV-vis spectrum of the product was measured, and peak absorbance had shifted from 259 nm (NAD + ) to 264 nm. Then excess Br2 was extracted using chloroform until the organic phase was clear. The aqueous layer was then recovered, and dialyzed through 0.5-1k MWCO membrane into 2 L water overnight, twice. The product was then lyophilized and stored at -80℃.
To synthesize 8-Br-NADH, 20 mL 1.3% sodium bicarbonate was added to a two-neck roundbottom flask, which was then bubbled with argon for at least 1 hr. Then 300 mg 8-NAD + -Br was added to the flask through side neck, followed by addition of 150 mg sodium dithionite (Fisher). The system was then covered with aluminum foil and stirred for ~4 hrs at r.t. while bubbling argon. UV-vis spectrum was measured, and the completion of the reaction was indicated by A265:A340 ratio close to about 3. The reaction was stopped by being mixing with 200 mL cold acetone, and the product was precipitated and subsequently recovered by collecting the pellet after centrifugation at 12000 rpm, 4°C for 2 min. The product was stored at -80 °C.
To synthesize 8-NH2-NADH, all the product of the previous step was thawed and mixed with 10 mL DMSO in a two-neck round-bottom flask. Stirring dissolved 8-Br-NADH in DMSO while impurities including salt stayed undissolved. The system was blanketed with argon, placed under a reflux condenser, and covered with aluminum foil. The flask was then heated to 60 °C with stirring. 1.4 ml ethylenedioxy)bis(ethylamine) (Sigma) was dissolved in 10 mL DMSO, which was then added to the flask from the side neck. The system was then stirred for 6 hrs, and subsequently cooled to r.t. overnight. The mixture was then centrifuged at 12000 rpm, 4 °C for 2 min, and the supernatant was collected. The product was precipitated by adding the supernatant to 200 ml cold acetone, and subsequently recovered by collecting the pellet after a second centrifugation. Then the product was dissolved in 10 mL DMSO, and precipitated again using 150 ml cold acetone. Finally, the product was recovered by collecting the pellet after a third centrifugation, and stored at -80 °C.
To synthesize NAD-maleimide, 325 mg N-(3-Dimethylaminopropyl)-N′-ethylcarbodiimide hydrochloride (Sigma-Aldrich) and 60 mg N-hydroxysuccinimide (Aldrich) was dissolved in 5 mL DMSO in a glass flask, followed by addition of 30 μL triethanolamine. 100 mg MAL-dPEG ® ₈-acid (Quanta Biodesign) was added to the mixture, and stirred for 1 hr at r.t. at dark. Then about a quarter of the product from the previous step (synthesis of 8-NH2-NADH) was added to the reaction, followed by stirring in the dark for 8 hrs at r.t.. The crude product was precipitated by adding the reaction mixture to 80 ml cold acetone, and recovered by collecting the pellet after centrifugation at 12000 rpm, 4 °C for 2 min. The crude product was stored in -80 °C temporarily. To purify NADmaleimide from the crude product, the product was thawed from -80 °C and dissolved in ~500 μL 50% methanol (v/v in water), and then applied to a BioRad disposable chromatography 20 mL column (part # 9704652) packed with 17-18 mL Cosmosil 75C18-OPN resin (Nacalai). The mobile phase used in this gravity chromatography was 50% methanol (v/v in water), and UV-vis spectroscopy was used to monitor the elution profile. The purified product was then lyophilized and stored -80 °C (average yield: 23% from the starting unmodified NAD + ).

NAD-CP bioconjugation
To reduce the addressable cysteine on CPN-est and CPS39C, P22 VLP ES was diluted to 1-2 mg/ml in lysis buffer (50 mM sodium phosphate, 100 sodium chloride, 2 mM EDTA, pH 7.0), and incubated with 2 mM dithiothreitol (DTT) at 4 °C overnight. Then additional 2 mM DTT was added, and the protein was incubated at 45 °C for 1 h. To remove the reducing agents the P22 VLP sample was ultracentrifuged (45000 rpm, 50 min, 4℃), and the pellet was resuspended in the lysis buffer. This was repeated twice to completely remove DTT.
To label CP with NAD, NAD-maleimide was added to the reduced P22 VLP ES, with a CP:NADH molar ratio of 1:2. Here, CP concentration was determined using Bradford assay, while NADH concentration was determined by A340. Because only about 20% of NAD-maleimide is in the reduced form (supplementary Fig. 2), the actual molar ratio of CP:NAD-maleimide was about 1:10. The bioconjugation was done by incubation at room temperature (r.t.), in the dark overnight. The reaction was quenched by adding 0.1% 2-mercaptoethanol. The P22 VLP sample was ultracentrifuged (45000 rpm, 50 min, 4℃), and the pellet was resuspended in assembly buffer (50 mM Tris, 25 mM sodium chloride, 2 mM EDTA, 3 mM 2-mercaptoethanol, 1% glycerol, pH 7.6) for the subsequent in vitro assembly.

In vitro assembly
1 mL of P22 VLP ES with a CP concentration of 80 μM (determined by Bradford assay) was mixed, in an 1:1 volume ratio, with 6 M guanidine hydrochloride (GdnHCl) in the assembly buffer. This resulted in denatured free CP in solution for in vitro assembly (40 μM CP, 3 M GdnHCl, in the assembly buffer).
The concentration of the enzymes in the assembly buffer was quantified by A280 (extinction coefficient: 27960 M -1 cm -1 for PtDH-SPt, 53860 M -1 cm -1 for AdhD-SPt, and 37360 M -1 cm -1 for ER-SPt). To make enzyme solution for in vitro assembly, the enzymes were mixed with total enzyme concentration of 40 μM, where the enzyme ratio can be adjusted here to vary the enzyme stoichiometry in the assembled particles.
To start in vitro assembly, the enzyme solution was added to the denatured CP solution (1:1 volume ratio) at r.t., resulting in a protein mixture solution containing 20 μM CP, 20 μM enzyme, and 1.5 M GdnHCl. The assembly was initiated by immediate dialysis of the mixture solution into fresh assembly buffer twice for a total of 15-18 h at r.t., using 6-8k MWCO dialysis membrane. Usually, 4 mL mixture solution was dialyzed into 500 mL assembly buffer each time, and the volume of assembly buffer was proportionally adjusted based on the volume of the protein mixture solution.
After dialysis, the solution was centrifuged at 17000 g, 2 min, r.t., to remove assembly aggregates. The assembled particles were then pelleted twice by ultracentrifugation at 43000 rpm, 40 min, 4℃ (rotor: F50L-24ⅹ1.5) using Thermo Scientific Sorvall WX Ultra Series Centrifuge, followed by resuspension in buffer (100 mM HEPES, pH 7.2). The unassembled proteins were mostly separated into the supernatant of the first ultracentrifugation.

Bradford Assay
The assay was carried by mixing 10 μL protein with 200 μL Coomassie Protein Assay Reagent (Thermo) in a 96-well plate and incubating for 10 min at r.t., followed by reading OD595 using BioTek Cytation5 plate reader. The protein standard used in Bradford assay is CPS39C P22 VLP ES, whose concentration was determined by A280 (extinction coefficient: 45045 M -1 cm -1 ) in denaturing condition (4.5 M GdnHCl). The calibration curve shows a linear response in OD595 within the range of 0-21 μM CPS39C.

Monitoring acetoin production in PtDH-AdhD coupled reaction
The reaction was carried out at 37 ℃ in the reaction buffer (100 mM HEPES, pH 7.2). Each reaction contains 1.8 μM P22 VLP nanoreactors (concentration determined using Bradford assay), 20 mM acetoin, and 400 μM sodium phosphite. Unmodified NAD + was added when noted. A mock reaction was included without P22 VLP nanoreactors. The reaction was initiated in a total volume of 525 uL, and then immediately aliquoted into 105 μL in PCR tubes, which were incubated at 37 ℃ on a Thermo Scientific Arktik thermal cycler. At each time point, 100 μL reaction was taken from one aliquot and quenched by adding 10 μL hydrochloric acid (1 N), followed by storage at -80 ℃.
To prepare the samples for analysis, the samples were thawed and neutralize by adding 10 uL sodium hydroxide (1N). 120 μL ethyl acetate was then added to extract 2,3-butanediol. To increase the extraction efficiency, 0.1 g NaCl was added to saturate the aqueous phase followed by mixing (vortexing). After centrifugation at 17000 g, 2 min, r.t., the ethyl acetate phase was taken for GC-MS analysis.
The samples were analyzed by Agilent 6890N Gas Chromatograph (GC) coupled with an Agilent 5973 Inert Mass Selective Detector (MSD). 1 μL of the sample was injected with a split ratio of 2:1 into an Agilent DB-5MS column (30m, 0.25mm, 0.25µm), with a total inlet flow of helium at 6 mL/min. After sample injection, the oven temperature was held initially 40 ℃ for 3 min, and then ramped to 280 ℃ at 20 ℃/min and held there for 2-min. The detector was on only between elution time 4 min and 5.5 min, and in the acquisition mode of selected ion monitoring (SIM)/scan (SIM: m/z 45, 57, 75, 90; scan: m/z range 10-100). 2,3-butanediol was confirmed by the total ion (scan) chromatograms using NIST MS Search 2.4 software, and quantified by integrating the peak area of 2,3-butanediol in the SIM chromatograms using OpenChrome (1.5.0.202209020347) software.

Recycling of NAD-CPN-ext
To recycle NAD-CPN-ext from in vitro assembled particles and aggregates from in vitro assembly, the particles were disassembled, and the aggregates were dissolved, in 3 M GdnHCl (in the assembly buffer). Then the proteins in this denaturing condition was supplemented with 5 mM imidazole, and then incubated with cOmplete His-tag purification resin (Roche, about 30 mg total protein per mL of the resin) at r.t. for 1 h. The mixture was then centrifuged at 4500 rpm, 5 min, r.t. and both the supernatant and resin (pellet) were collected for analysis. The recycled NAD-CPN-ext was in the supernatant, which was then diluted with 3 M GdnHCl (in the assembly buffer) to 40 μM, serving as the CP solution for further in vitro assembly.
To recycle NAD-CPN-ext from unassembled proteins of in vitro assembly, the proteins were supplemented with 5 mM imidazole and then incubated with cOmplete His-tag purification resin with the same condition mentioned above but in native condition (assembly buffer). After centrifugation, the supernatant was concentrated using a MWCO 30k Amicon spin concentrator, and then used for in vitro assembly as an equivalent of P22 VLP ES mentioned in the in vitro assembly section.

SDS-PAGE and densitometry analysis
The protein samples were analyzed by 12% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), which was stained by InstantBlue Coomassie Protein Stain (Abcam). When needed, the densitometry data were obtained by analyzing the gel picture in ImageJ (1.51j8). The data were then normalized and plotted in Igor Pro (6.37), and the peak area of the peaks were calculated when necessary.

TEM
P22 VLPs were diluted to A280 ~0.4 using water, and then 4.5 μL was applied to 400 mesh carbon-coated copper grids. After incubation for 1 min, excess sample was wicked away with filter paper. The grid was then stained with 4.5 μL 2% uranyl acetate for 15 sec, and then the stain was wicked away with filter paper. The images were taken on JEOL JEM-1010 transmission electron microscope.

LC-MS
The proteins were diluted to ~1 mg/mL using the assembly buffer. The samples were analyzed