Nanocaged enzymes with enhanced catalytic activity and increased stability against protease digestion

Cells routinely compartmentalize enzymes for enhanced efficiency of their metabolic pathways. Here we report a general approach to construct DNA nanocaged enzymes for enhancing catalytic activity and stability. Nanocaged enzymes are realized by self-assembly into DNA nanocages with well-controlled stoichiometry and architecture that enabled a systematic study of the impact of both encapsulation and proximal polyanionic surfaces on a set of common metabolic enzymes. Activity assays at both bulk and single-molecule levels demonstrate increased substrate turnover numbers for DNA nanocage-encapsulated enzymes. Unexpectedly, we observe a significant inverse correlation between the size of a protein and its activity enhancement. This effect is consistent with a model wherein distal polyanionic surfaces of the nanocage enhance the stability of active enzyme conformations through the action of a strongly bound hydration layer. We further show that DNA nanocages protect encapsulated enzymes against proteases, demonstrating their practical utility in functional biomaterials and biotechnology.

, compared with that of the fresh free MDH enzyme (blue square) using RBG as the substrate. The solid line is the fitting curve using the Michaelis-Menten model. Enzyme assay condition: 0.5 nM enzyme or DNA cage encapsulated enzyme, with different concentration of RBG, ranging from 10 µM to 600 µM, in TBS buffer (pH 7.5, 1 mM MgCl 2 ) monitoring fluorescence at 532/590 nm. The table lists the fitting parameters. DNA encapsulation of the enzyme caused a ~1.6-fold increase in K M and a ~81% decrease in k cat . Error bars were calculated from the standard deviation of at least three replicates.

RBG (µM)
Initial rate (sec  . The UV-Vis absorbance of 100 µL of each enzyme solution was measured before adding to the plates, as well as after one hour incubation within the plates in the dark. These conditions are the same as those of the enzyme activity assay. As shown in the Figure, all enzyme solutions showed only a very slight decrease in absorbance after incubation in the plates, suggesting very weak nonspecific adsorption of enzymes onto the plastic plates. Error bars were calculated from the standard deviation of at least three replicates.

Supplementary Figure 46:
Testing for nonspecific adsorption of low nanomolar concentrations of enzymes onto plastic 96-well plates was tested using Cy3-labeled HRP. 100 µL of 10 nM Cy3-labeled HRP was assayed for fluorescence intensity, and then the plate was incubated inside a plate reader for one hour. The remaining fluorescence intensity was tested again. A slight increase of fluorescence intensity was observed, possibly due to the buffer evaporation during the incubation. This result suggests that there is very little nonspecific adsorption of Cy3-HRP onto the 96-well plate. Error bars were calculated from the standard deviation of at least three replicates. For inhibition assay, β -Gal was first incubated with Poly(P) 100 for half an hour, then RBG substrate was added before measuring the activity. The control β-Gal was run at the same condition except for the incubation with buffer for half an hour. The activity of β-Gal was significantly inhibited by 1000 µM Poly(P)100. Error bars were calculated from the standard deviation of at least three replicates.  Table S4).

Supplementary Figure 51:
Enzyme activity data of single β -Gal molecules. (a) Representative raw fluorescence-time traces of free-, half-cage and full-cage β -Gal. Five representative molecules are shown for each sample. The fluorescence intensity of enzyme reaction on the microscope slide was recorded for ~5 min at 35 ms time resolution. (b,c,d) Statistics of spike frequency, fraction of active molecules, and overall observed enzyme activity. The number of active molecules analyzed is denoted by 'n' in b. The standard deviations for spike frequency and fraction of active molecules were calculated after randomly assigning the active molecules into three groups. The standard deviation for the normalized overall activity was estimated from the propagation of errors. All experiments were carried out at room temperature in 1× TBS buffer, pH 7.5 in presence of 1 mM Mg 2+ and 10% (w/v) PEG 8000.  4 Cl and (D) Triethylammonium acetate (TEAA) on the activity of free G6pDH. Assay conditions: 0.5 nM enzyme was incubated with a series of ion concentrations from low to high. Enzyme activity was monitored by absorbance at 340nm with the addition of 1 mM Glucose-6-phosphate and 1 mM NAD + in 1×TBS buffer (pH 7.5). The results show that high concentration of salts containing small cations such as Na + , K + and NH 4 + significantly reduce the activity of G6pDH,

Supplementary Tables
Supplementary Table 1 GOx--P1-Cy3  Figure 5, SPDP conjugation chemistry was used to couple enzymes to oligonucleotides as reported previously 1,2 : a) Enzymes (GOx, HRP, G6pDH, LDH, MDH and β-Gal) were first conjugated with SPDP at enzyme-to-SPDP ratios of 1:5, 1:20, 1:3, 1:5, 1:5, and 1:5, respectively, in HEPES buffer (50 mM HEPES, pH 8.5) for 1 h at room temperature. Different values of SPDP-to-Protein ratio were used due to the varied number of accessible surface lysine residues for each protein. Excess SPDP was removed by washing with 50 mM HEPES buffer using Amicon centrifugal filters (30 kD cutoff). The SPDP coupling efficiency was evaluated by monitoring the increase in absorbance at 343 nm due to the release of pyridine-2-thione (extinction coefficient: 8080 M -1 cm -1 ). b) TCEP-treated thiolated DNA (/5ThioC6-/-TTTTTCCCTCCCTCC (P1), or /5ThioC6-D/-TTTTTGGCTGGCTGG (P2)) was incubated with the SPDP-modified enzymes at an enzyme-to-DNA ratio of 1:10 in 50 mM HEPES buffer (pH 7.4) for 1 h in the dark. Excess unreacted oligonucleotide was removed by ultrafiltration using Amicon 30 kD cutoff filters: washing one time with 50 mM HEPES (pH 7.4) containing 1 M NaCl and three times with 50 mM HEPES (pH 7.4). The high salt concentration in the first washing buffer helps remove DNA nonspecifically bound to the surface of the protein due to electrostatic interactions. c) The absorbance values at 260 nm and 280 nm (A 260 and A 280 ) were recorded to quantify the enzyme-DNA complex concentrations and the labeling ratios using a Nanodrop spectrophotometer (Thermo Scientific) (Supplementary Figure 6 and Supplementary Table 1). Extinction coefficients of DNA oligonucleotides were received from IDT-DNA, and extinction coefficients of enzymes were obtained from published data. d) Dye labeling of DNA-conjugated proteins: The DNA-conjugated proteins were further labeled with spectrally distinct fluorescent dyes, which allow us to use native gel electrophoresis and single-molecule fluorescence to confirm the encapsulation of proteins within DNA nanocages. NHS-ester-modified dyes were reacted with the purified DNA-conjugated proteins from the above steps at a 20:1 ratio in 50 mM HEPES buffer, pH 8.5. Cy3 was directly labeled to the lysine residues on the protein surface. Excess dyes were then removed using 3-kD cutoff Amicon filters. The UV-Vis absorbance spectra of the purified dye-labeled proteins are shown in Supplementary Figure 6 and were used together with the extinction coefficients of the dye (150,000 M -1 cm -1 for Cy3 at 546 nm; 250,000 M -1 cm -1 for Alexa647 at 647 nm ) and of the protein-DNA conjugates to quantify the concentration and labeling ratio of the dye-labeled proteins. e) Conjugate proteins to Cy3-labeled DNA: In order to perform the single-molecule enzyme activity assay, selected enzymes (G6PDH and β -Gal) were conjugated to a Cy3-labeled DNA. First, NHS-ester-modified dyes were reacted with the 3'-amine of oligonucleotides at a 20:1 ratio in 50 mM HEPES buffer, pH 8.5. Excess dyes were then removed using 3-kD cutoff Amicon filters. Dye-modified oligonucleotides were then conjugated to proteins via the 5'-thiol using the SPDP chemistry described above. Fast Protein Liquid Chromatography (FPLC) was used to purify the protein-DNA-Cy3 conjugates for removing excess DNA-Cy3, and characterized with the UV-Vis absorbance spectra.

Enzyme-DNA cage assembly, purification and characterization:
a) The purified DNA half-cage containing capture strands was mixed with one of several enzyme-DNA conjugates at a 1:15 cage:enzyme ratio and annealed from 37°C to 4°C over 2 h in 1×TAE-Mg 2+ buffer (containing 12.5 mM Mg(OAc) 2 ). b) Twenty-four single-stranded DNA linkers were mixed with the two purified half-cages at a 5:1 linker:cage ratio to connect the two half-cages together by incubating at room temperature for 3 h. c) Agarose gel electrophoresis (2%, 1×TAE-Mg 2+ ) was employed to remove excess free enzymes (70V, 2h). The band of the DNA cage containing the enzyme was cut from the gel and extracted using a Freeze 'N Squeeze column (Bio-Rad). The DNA origami concentration was quantified by measuring the absorbance at 260 nm (A 260 ) using an extinction coefficient of 0.109 nM -1 cm -1 .

Supplementary Note 2: Single-molecule fluorescence microscopy for characterizing DNA cage-encapsulating enzymes.
Yield estimation by TIRF colocalization: All single-molecule measurements were performed at room temperature using a total internal reflection fluorescence (TIRF) microscope on PEGylated fused silica microscope slides. To passivate the microscope slides and functionalize the surface with biotin for selective immobilization of nanocages, a biotin-and PEG-coated surface was prepared by silylation with APTES, followed by incubation with a 1:10 mixture of biotin-PEG-SVA 5k:mPEG-SVA 5k as described previously. 3 A flow channel was constructed as described elsewhere. 3 To prepare the surface for enzyme or nanocage binding, a solution of 0.2 mg/mL streptavidin in T50 buffer (50 mM Tris-HCl, pH 8.0, 50 mM NaCl, 1 mM EDTA) was injected in to the flow channel, incubated for 10 min, and the excess streptavidin was flushed out thoroughly first with T50, then with 1× TAE-Mg.
The right half of the DNA origami cage was labeled with Cy5 dye inside the cavity, via hybridization of Cy5-labeled DNA to complementary handles incorporated into the structure. Each of the ssDNA conjugated enzymes (HRP, GOx, G6pD, LDH, MDH and β -Gal) was covalently labeled with Cy3 as described in section 3 (Cy3-Enzyme-5'-TTTTTCCCTCCCTCC), and then linked to the left half of the DNA origami cage via hybridization with complementary handles. Because Cy3 was directly labeled onto the enzyme surface, any observed Cy3 signal of the immobilized DNA nanocages came from the encapsulated enzymes. Linker strands were added to a 1:1 mixture of the two half-cages to encapsulate the enzymes in a full-cage. To capture DNA-modified enzymes in the absence of nanocage (as control) the microscope slide was incubated with 10-20 nM biotin-modified complementary DNA oligonucleotide (5'-biotin-TTTTTGGAGGGAGGG) for 3 min, followed by 10 min incubation with 20-50 pM enzyme sample in 1×TAE-Mg buffer. Excess enzyme was flushed out with ~400 uL buffer (channel volume ≈ 30 µL). For the nanocage experiments, the samples were diluted to 20-50 pM in 1× TAE-Mg and immobilized on the streptavidin-coated PEG surface for 1 min, and the excess sample was flushed out with ~400 µL of 1× TAE-Mg. The DNA-modified enzymes were imaged with illumination at 532 nm (~15 W/cm 2 ), and the nanocage-encapsulated enzymes were imaged S68 with simultaneous illumination at both 532 nm (~15 W/cm 2 ) and 640 nm (~40 W/cm 2 ) as described. 4 Particle-finding and colocalization analysis were performed using custom-written scripts in IDL and MATLAB, using a threshold of 150 counts per frame for particle identification (typical particles showed 500-1,000 counts per frame in each detection channel). The enzyme encapsulation yield, defined as the fraction of assembled nanocages containing enzyme(s), was estimated by dividing N coloc by the total number of particles containing a right half-cage, N right (Supplementary Table 2).
Estimation of enzyme copy number per nanocage: The number of enzyme copies per nanocage (N enz ) was determined by single-molecule photobleaching (SMPB). First, the number of Cy3 photobleaching steps was determined separately for unencapsulated as well as half-cageand full-cage-encapsulated enzymes. For this, the donor channel data of all single molecules were idealized in QuB (http://www.qub.buffalo.edu) using a six-state model 5 . The histogram of the photobleaching steps was then acquired using a custom-written MATLAB script. Representative intensity traces exhibiting one, two, and three photobleaching steps are shown in Supplementary Figure 2c (more than three photobleaching steps were rarely seen). Finally, the number of enzyme molecules per cage was estimated by dividing the mean number of Cy3 photobleaching steps of the full-cage (µ Cy3_Encap ) by the mean number of Cy3 photobleaching steps for the unencapsulated enzyme (µ Cy3_Unencap ). Results are summarized in Supplementary  Table 3.

Supplementary Note 3: Single-molecule enzymology
Single-molecule enzyme activity assay: Prior to single-molecule activity measurement, streptavidin-modified slides were incubated for ~2 min with neutravidin-coated fluorescent beads (Invitrogen, 0.04 µm diameter, excitation/emission; 550/605 nm) at 10 6 -fold dilution and the excess flushed out with 1×T50 buffer. These beads (~5-8 per field of view) were used as fiducial markers to correct for drift of the microscope stage and/or slide (Fig. 5a,c). Following complete photobleaching of Cy3 in a field of view, the activity of single unencapsulated or nanocage-encapsulated enzyme molecules was imaged on the same field of view. During analysis of the movies, the coordinates of the initial photobleaching movie were registered with those of subsequent movies using the fiducial markers (visible throughout all sequential movies) in a custom-written MATLAB script. This approach allowed us to infer the locations (x-and ycoordinates) of all individual enzymes/nanocages in the field of view even after bleaching Cy3, and to monitor enzyme turnovers (≈ resorufin formation) at these specific coordinates.
To image enzyme activity, 300 µL of substrate solution in 1×TBS buffer (pH 7.5, 1 mM Mg 2+ , and 10% (w/v) PEG8000) (Supplementary Table 4) was injected into the flow channel. Movies were recorded for ~5 min (9,091 frames) at 35 ms frame exposure time immediately after injecting the substrate solution. In case of G6pDH, the activity was measured in the same field of view under identical laser illumination and microscope settings, with or without glucose-6-phosphate (G6p) (Fig. 5c). Enzyme activity for β-Gal was measured similarly using a 500 nM solution of resorufin β -D-galactopyranoside (RBG) as substrate, which is hydrolyzed by β -Gal into fluorescent resorufin. Fluorescence fluctuations over time were measured for unencapsulated enzyme as well as half-and full-cage-encapsulated enzyme ( Supplementary  Figures 57 and 58), and the fluorescence time traces were analyzed for intensity spikes using custom-written MATLAB script. The script allowed us to measure the background intensity of single-molecule traces and set a threshold (mean + 8 standard deviations) to subtract from the raw intensity. Since we often observed one or two spikes above this intensity threshold in the control experiments, only those molecules with ≥4 spikes were counted as active molecules (Supplementary Figure 59) and considered for burst analysis. Due to the low concentration of resazurin (Supplementary Table 4), the criteria we used to determine the fraction of active molecules might have excluded some molecules that are not highly active.
Burst analysis: Burst analysis was carried out using a modified Rank Surprise (RS) method 6 recently utilized to analyze the binding of fluorescent DNA probes to a riboswitch 7 . Briefly, Interspike Intervals (ISIs) were determined by calculating the time in between individual fluorescent spikes for each molecule (Supplementary Figure 59). The RS method was used to demarcate the start and end points of bursts after collecting ISIs for all molecules. Only intensity spikes characterized by an ISI of <5 seconds were considered part of a burst; any other intensity spikes are counted as non-bursts.
Comparing bulk and single-molecule enzyme activity: Unlike our single-molecule assay, the bulk measurement of enzyme activity cannot explicitly determine the fraction of active enzyme molecules present in the solution (it is well known that a fraction of enzyme molecules loses their activity during oligonucleotide conjugation, buffer exchange and the purification process). However, the observed bulk activity is contributed not only by enzyme turnover rate but also by the fraction of enzyme molecules that are still active. Both of these contributing factors need to be accounted for to directly compare the single-molecule enzyme activity with the bulk measurements. Therefore, in the single-molecule experiment, the overall activity of free, halfcage and full-cage enzymes were calculated by multiplying the turnover rate with the fraction of active molecules for the given sample.

Supplementary Note 4: DNA sequences of the designed nanocages.
Sequences of staple strands in the SH Full-Cage-Left cage 5 assay, 2 mM pyruvate and 1 mM NADH were used as substrates, and enzyme activity was measured by monitoring the decreased absorbance at 340 nm due to the oxidation of NADH to NAD + . For a typical MDH assay, 2 mM OAA and 1 mM NADH were used as substrates, and enzyme activity was measured by monitoring the decrease in absorbance at 340 nm. For a typical β-Gal assay, 100 µM RBG was used as substrate and enzyme activity was measured by monitoring fluorescence intensity, with excitation at 532 nm and emission at 590 nm.
Trypsin assay: Enzyme activity was measured after incubation with or without trypsin (1 µM) at 37 °C for 24 h in 1×TAE-10mM Mg buffer (pH 8.0). Activity assay conditions: 1 mM Glucose, 1 mM ABTS, 1 nM of free GOx and HRP in pH 7.5, 1×TBS buffer containing 1 mM MgCl 2 , and monitoring absorbance at 410 nm. In the DNA cage experiment, all conditions were the same except for incubating 1 nM DNA cage-encapsulated GOx and HRP with trypsin.