DNA-origami-directed virus capsid polymorphism

Viral capsids can adopt various geometries, most iconically characterized by icosahedral or helical symmetries. Importantly, precise control over the size and shape of virus capsids would have advantages in the development of new vaccines and delivery systems. However, current tools to direct the assembly process in a programmable manner are exceedingly elusive. Here we introduce a modular approach by demonstrating DNA-origami-directed polymorphism of single-protein subunit capsids. We achieve control over the capsid shape, size and topology by employing user-defined DNA origami nanostructures as binding and assembly platforms, which are efficiently encapsulated within the capsid. Furthermore, the obtained viral capsid coatings can shield the encapsulated DNA origami from degradation. Our approach is, moreover, not limited to a single type of capsomers and can also be applied to RNA–DNA origami structures to pave way for next-generation cargo protection and targeting strategies.


Note S2: SAXS analysis of complexed 6HB
The homogeneity of complexed structures in a 6HB-2k sample was evaluated by SAXS. For modelling purposes, the background in form of the complexation buffer was subtracted, simultaneously with the addition of a Debye background (Fig. S2a after and Fig. S2b before background treatment). For plain 6HB (green) and 6HB-2k (blue circles), clear features are displayed in the intensity distributions. 6HB-2k was geometrically modelled using a core-shell cylinder, resembling 6HB as the core cylinder and the single protein layer as its shell. A radius of 3.1 nm for the core (r (core) ) and 10.2 nm for the core-shell cylinder (r (total) ) are in agreement with the dimensions obtained from TEM analysis and cryo-TEM reconstruction (main article Fig. 2). It is notable that in comparison to 24HB (main article Fig. 4d), the radius of the shell is slightly increased for 6HB. Assuming a consistent thickness of the protein layer, and considering the thickness of 6HB, these results suggest a certain degree of freedom of 6HB within the protein shell. To account for the semi-solid intra-bundle distance between the DNA helices, a Lorentzian peak was included in the model (1). Its center is at 0.127 Å −1 , corresponding to an inter-helical diameter of 3.3 nm (Fig. S2c, orange dashed). The length of the structure was fixed to 4,000 Å, and has no impact on the features observed since it is outside of the experimental range. Fig. S2 a, SAXS scattering curves measured in solution for 6HB-2k, plain 6HB (green) and CPs only (blue). b, Intensity distributions of 6HB-2k (blue), CPs only (dark blue dashed), plain 6HB (green dashed) and the complexation buffer (yellow dashed) before buffer substraction. c, Breakdown of the fit for 6HB-2k to show the contributions of the chosen core-shell model (green dashed) and the Lorentzian peak (orange dashed).

Note S3: Quantification of the coating yield
Observed abnormalities when analyzing complexed samples at either ε = 2k or ε = 10k were collected into four defect classes, namely loops (Fig. S3a), bends (Fig. S3b), structures with incomplete layers (Fig. S3c,g), and elongated structures (Fig. S3d). Both bends and loops can either arise from a defect in the template, or by a T = 1-like defect upon CP assembly. Introduction of pentamers can create a local curvature, yielding in bending of the rather flexible 6HB template. A similar behaviour is observed for the CP assembly on 13HR (main article Fig. 4f,g), where it is more pronounced due to the constrained shape. The fraction of bend defects, φ, can be determined by where ℓ describes the measured tube length, n the number of observed bends, and θ the measured bend length. Using this equation, θ was calculated to be 0.46 % for 6HB-2k and 0.70 % for 6HB-10k. The appearance of elongated structures seems to be concentration dependent. They were mainly observed in samples complexed at high final origami concentrations (c ≥ 80 nM), at which the denseness of the 6HB increases and therefore also the propensity of encapsulation of more than one single origami structure. Most frequently, incomplete layers were observed, especially for structures complexed at ε < 2k and ε < 10k with regards to single (grey) and double (green) layer formation, respectively (Fig. S3e). The formation of double layer structures requires considerably greater ε than single layer structures, most likely due to the decreased strength of negative surface charge. Subsequently, a considerable part of free CPs assembles into T = 3 particles. When assembling into the second protein layer, the nucleation seems to happen nonspecifically along the entire structure ( Fig. S3g, 6HB-7.5k). For batch characterization of free CPs, an agarose gel (Fig S3f, top), showing the decrease in electrophoretic mobility, was post-stained with Coomassie blue to visualize the proteins. Staining was performed for 1 h. After stain removal and three washing steps in water á 5 min, the gel was destained for 2 h in 40 % ethanol, 10 % acetic acid before imaging. At ε ≥ 2k, the protein concentration is high enough to be detected (Fig. S3f, bottom) and the protein bands can be compared to the DNA origami bands. While the leading band is overlapping in both EtBr and Coomassie channel, a second band develops, which is trapped in the wells and/or runs slightly into the opposite direction, supposedly free CPs. Negative-stain TEM images of 6HB-1k (Fig. S3h) and 6HB-5k (Fig. S3i) illustrate the distribution between different fractions described in Fig. S3e.

Note S4: Growth of first CP layer
The dependency between the diameter of the rod and the preferred location of nucleation along the rod was suggested to be related to the ratio (γ) between the spontaneous curvature radius of the capsomers and the radius of the template. The spontaneous curvature radius refers thereby to the inner radius of the sphere the capsomers would assemble into (2). From a continiuum expression, derived by Lazaro et al. (2) for BMV a threshold (γ * ) can be predicted which separates the assembly preferentially on spherical (cap) from cylindrical surfaces. For γ ≫ γ * , capsomers were suggested to nucleate along the rod-like region whereas for γ ≪ γ * , nucleation on the spherical ends would be favoured. The two regions cross over around γ * (2). Using the total diameters of both plain origami and complexed structures calculated from the SAXS measurements, d total,6HB−2k = 20.4 nm and d total,24HB−2.5k = 23.4 nm, and taking the size of the capsid protein, 3.8 nm (3), into consideration, the ratio could be calculated. Assuming a similar threshold as for BMV, γ * ∼ 1.6 (2), the CPs should preferentially nucleate along the origami for 6HB and from both the ends and along the origami for 24HB, which is also observed in TEM, as shown for 6HB-500 ( Fig. S4a), 24HB-250 (Fig. S4b), and 24HB-750 (Fig. S4c).

Note S9: SAXS analysis of complexed 24HB
For the model, the buffer ( Fig. S10a) was subtracted, while a Debye background was added. Additionally, CPs assemble spontaneously into icosahedral structures (4). They can be modelled as solid spheres, resulting in particles with a diameter of ca. 28 nm, suggesting T = 3 symmetry (Fig. S10b). For 24HB, a cylinder is chosen as its geometrical representative, resulting in a radius of 7.7 nm (Fig. S10c). The cylinder model additionally includes fitting to the Lorentzian peak (1), which has its center at 0.157 Å −1 corresponding to an inter-helical distance of 2.7 nm. Looking at the components' contributions, the Lorentzian peak (orange) accounts for the flattening of the cylinder model (green). As observed in TEM, two different populations, namely the complexed 24HB-2.5k structures and sphere-like assemblies, are assumed to be present in the complexed sample. Subsequently, the intensity distribution of the capsids is reduced to a form factor by subtraction from the 24HB-2.5k, thus features of the complexed 24HB become visible, and the complexed structures can be modeled as core-shell cylinders (main article Fig. 4d, Fig. S10d).
In order to avoid the subtraction factor, the system was furthermore modelled as a three-function model consisting of a coreshell cylinder for 24HB-2.5k, a sphere resembling icosahedral CP assemblies and the debye background (Fig. S10e). The model agrees well with the TEM measurements, suggesting diameters of 23.4 nm for 24HB-2.5k and 27.6 nm for spherical assemblies with T = 3 symmetry. The thickness of the CP shell on 24HB-2.5k is 5.1 nm. Furthermore, the impact of the single components on the final intensity distribution can be shown with this approach (Fig. S10f). The cylinder length, which is expected to be ca. 110 nm, has no impact on the model.

S14
Note S12: Coating of the nanocapsule structure For initially testing the complexation between CPs and the nanocapsule, the permanently closed version of the structure was used. EMSA (Fig. S13a) suggests successful complexation, which was confirmed by negative-stain TEM (Fig. S13b, nanocapsule-2k). Similar to 6HB and 24HB, fully coated structures were observed at ε ≥ 2k. However, when complexing the pH-responsive nanocapsule, pH changes should be avoided. The pH of the complexation buffer, pH 7.3, is exactly in the transition region between closed and open conformation of the nanocapsule (5). Hence, the complexation method was slightly modified, including the transfer of the purified origami structure into 6.5 mM HEPES buffer, pH 6.0. For the complexation, the CPs were titrated into the DNA origami solution in 1 µL steps, together with 0.4 µL of 25 mM acetic acid. Comparing the EMSA profile obtained from complexation at pH 6 ( Fig. S13c), retention in the wells is observed already at ε = 750. Successful complexation was confirmed by negative-stain TEM (Fig. S13d).
For the release of the nanocapsule, heparin was used as a competitive agent to disassemble the CP coating. The amount of heparin used is expressed as the ratio between n sulfates originating from heparin and n phosphates originating from the DNA backbone. A heparin molecule was estimated to contain on average 71 negatively charged sulfate groups, assuming an average molecular weight of 17-19 kDa and an average of 2.33 sulfate groups per repeating IdoA(2S)-GlcNS(6S) disaccharide unit (6). The total number of negative charge for the nanocapsule was estimated to be 17459. The release is monitored by EMSA (Fig. S13e), for the final experiments, a ratio of 1.9 was used, and the heparin was incubated with the structure for 10 min before pH adjustment. Note S13: Functionalization of 6HB 6HB was functionalized with AuNPs in a two-pot reaction, similar to previously described procedures (7,8). First, the 6HB structure was prepared as described in the Method section, however, three staple strands were exchanged to contain an overhang which can later hybridize with oligonucleotide-functionalized AuNPs. After purification using poly(ethylene glycol) (PEG) precipitation (as described in Supplementary Note S20, similar to (9)), the folded structures were mixed with oligonuceotidefunctionalized AuNPs which were added in 30× excess (10× excess per annealing site), heated to 40°C and subsequently the temperature was decreased to 20°C (−0.1°C min −1 ramp). Briefly, oligonucleotide-functionalized AuNPs were obtained by mixing 40 µL of AuNPs (5 nm diameter, citrate stabilized, 100 nM, Sigma Aldrich) with 0.8 µL of sodium dodecyl sulfate (SDS) for 20 min, before incubation with 4 µL thiol-modified oligonucleotides (for hybridizing with staple overhangs) for 30 min. The AuNPs were salt-aged using 2.5 M NaCl by 6× addition of 0.4 µL, 6× addition of 0.8 µL, 5× addition of 1.6 µL and a final addition of 2 µL. The interval between the additions is 5 min and all steps are performed at 40°C and 600 rpm (Eppendorf ThermoMixer C). Subsequently, 60 µL of 1× folding buffer (FOB, 1× Tris-acetate-EDTA (TAE), 12.5 mM MgCl 2 ) supplemented with 0.02 % SDS are added and the incubation was continued for 1 h before the temperature was decrease to 20°C for an overnight incubation. Before usage, the oligonucleotide-functionalized AuNPs were purified from excess oligonucleotides using spin-filtration. After an initial washing step with 200 µL of the desired buffer (14,000 g, 5 min), 360 µL of AuNPs were added to the filter together with 120 µL 1× FOB with 0.02 % SDS and centrifuged for 10 min at 14,000 g, followed by a 3× addition of 200 µL of 1× FOB with 0.02 % SDS (10 min, 14,000 g). The purified DNA-functionalized AuNPs were recovered by inverting the filter and centrifugation for 2.5 min at 1,000 g. The folded structures were purified from excess staple strands and AuNPs by PEG precipitation (10). Per 50 µL of folding reaction, 12.5 µL of PEG buffer containing 17.5 % (w/v) PEG8000, 1× TAE buffer, 500 mM NaCl and 10 mM MgCl 2 are used. Before centrifugation at 4°C and 12,600 g for 30 min, the mixture is incubated at 4°C for 10 min. The supernatant is discarded and the pellet resuspended in 1× FOB with 0.02 % SDS and incubated overnight at 30°C, 600 rpm on an Eppendorf ThermoMixer C before the procedure is repeated to ensure full removal of excess AuNPs.

Note S14: DNase I digestion studies
Heparin was used as a competitive agent for the disintegration of CPs from the complexed structures resulting in plain origami structures to show the structural intactness after incubation with DNase I. As described in Note S12, the amount of heparin used is expressed as the ratio between n sulfates and n phosphates . 6HB origami was estimated to have in total 14569 phosphate groups (7) while 24HB has 15504. For the final digestion experiments, 3.8× excess and 200× excess were used for structures coated with one or two CP layers, respectively. The stability of the structures was furthermore tested in cell medium (Dulbecco's Modified Eagle Medium) supplemented with 5-10 % FBS, which was mixed with the complexed samples in a 1:1 ratio, resulting in an origami concentration of 3.2 nM (Fig. S15). After 24 h incubation at 37°C, plain 6HB (left) was found to be partially digested in both FBS concentrations, while complexed structures (6HB-2k, right) were intact. The significant amount of 6HB remaining coated, and therefore remaining in the wells upon disassembly with heparin, could be explained by the large amount of protein present in FBS, which might unspecifically interact with heparin. Note S15: RNA-DNA hybrid origami For the folding of the hybrid structure, the folding conditions have been optimized to ensure a high folding yield. Although a known thermal annealing program was used (11), different salt supplements in the folding buffer (1× TAE) were studied. EMSA of unpurified structures (Fig. S16a) shows the formation of double bands with increasing MgCl 2 concentrations. At 12.5 mM MgCl 2 , the double band disappears, however the structures tend to aggregate in the wells. To prevent aggregation NaCl was added, leading to the final salt supplements of 5 mM MgCl 2 , 1 mM NaCl. The double band is suggested to represent a dimer band, which is also reflected in the size distribution of both plain (blue) and complexed (RNA-6HB-500, grey) structures (Fig. S16c). Moreover, the poly(A) tail of the structure remains unfolded (i.e. no hybridization with staple strands), which additionally contributes to the heterogeneity of the complexed sample (Fig. S16b,c). Note S16: Coating with norovirus (NoV) The dissociation and reassembly of NoVLPs from different strains has been reported to be highly dependent on the ionic strength and pH of the buffer (12,13). For instance, higher order oligomers, like 60-and 80mers have been found at high ionic strength and alkaline pH. Such oligomers were suggested to form due to a transition of the VP1, in which 180mers corresponding to the native virus particle disassemble into dimers followed by spontaneous reassembly. Furthermore, a dependency between oligomer formation and VP1 concentration was reported (13). We used 50 mM Tris-buffer, pH 8.9 for the disassembly of intact VLPs (Fig. S17a), which resulted predominantly in oligomers (Fig. S17b). To study reassembly on 6HB, the DNA origami was first mixed with VLPs at different concentrations (ε = 500 and 2k) and then dialyzed against the disassembly buffer. The same sample was then further dialyzed against reassembly buffer at low pH (sodium phosphate, pH 6), but no change in electrophoretic mobility is detected from EMSA (Fig. S17c). From negative-stain TEM a clear difference in the VP1 behaviour upon pH change can be observed. Compared to alkaline pH (Fig. S17d,e), VP1 is assembled to a larger extent into higher order oligomers at acidic conditions (Fig S17f). NoV VP1 was described to lack the predominance of positively charged residues in the N-terminal region (14). Instead, the basic, minor structural protein VP2, most likely located in the capsid shell interior, is suggested to interact with nucleic acids (15). Since here, the VLPs are made entirely of VP1, we suggest ascribing the inability of ordered complex formation between VP1 and DNA origami to the lack of the RNA binding domain. Negative-stain TEM images for d, 6HB-NoV-500, and e, 6HB-NoV-2k in 50 mM Tris-buffer, pH 8.9, and f, 6HB-NoV-500 in 100 mM sodium phosphate buffer, pH 6.
Note S17: Coating with simian virus 40 (SV40) Disassembly of intact SV40 VLPs (Fig. S18a) results mainly in pentameric capsomers, as well as larger assemblies (Fig. S18b), which were not removed before complexation with 24HB. The EMSA (Fig. S18c) shows a small plateau in mobility decrease of the leading band around ε = 5k-7.5k, resulting in further analysis of 24HB-SV40-5k. In the absence of a template, the capsomers have been reported to assemble into tubular structures (16). A similar behaviour is also observed here (5k, Fig. S18d). However, in comparison to capsomers only, the diameter of the templated elongated structures (24HB-SV40-5k, Fig. S18e) is increased. The complexed structures can be classified into three groups, with assemblies around 100 nm being the majority. The shorter assemblies arise most likely from a heterogeneous template (truncated versions of ca. 30 nm and ca. 60 nm). Note S18: Coating with murine polyoma virus (MPyV) The disassembled pentameric capsomers (main article Fig. 5j) were assembled either into VLPs (Fig. S19a) or complexed with both 6HB and 24HB. The EMSA shows a similar behaviour for 6HB (Fig. S19b, top) and 24HB (Fig. S19b, bottom). While partially coated structures are observed for 6HB-MPyV-750 (Fig. S19d), both structures are fully coated and display discrete sizes (Fig. S19c,e-f) at ε > 1.25k.

Note S19: Materials
All chemical reagents were obtained from commercial suppliers. 50× TAE buffer was purchased from Thermo Fisher Scientific, agarose from Biotop Oy, and ethidium bromide, heparin sodium salt from porcine intestinal mucosa, as well as DNase I from bovine pancreas from SigmaAldrich.
Once the structures are folded, excess staple strands were removed using poly(ethylene glycol) (PEG) precipitation (9). After dilution in 1× FOB to a concentration of approx. 5 nM, the final volume was mixed in a 1:1 ratio with PEG precipitation buffer (1× TAE, 15 % (w/v) PEG8000, 505 mM NaCl) and centrifuged for 30 min at 14,000 g. The supernatant was removed, and the pelleted DNA origami resuspended in 1× FOB and incubated overnight at 30°C, 600 rpm on an Eppendorf ThermoMixer C.
13HR was folded with final scaffold and staple concentrations of 10 nM and 50 nM, respectively, in 1× Tris-EDTA (1× TE pH 7.6, 10 mM Tris, 1 mM EDTA) buffer containing 10 mM MgCl 2 . After 15 min incubation at 80°C, the mixture was cooled from 79°C to 71°C at a rate of −1°C min −1 , from 70°C to 66°C at −0.2°C min −1 , from 65°C to 60°C at −2°C h −1 , from 59°C to 37°C at −1°C h −1 , from 36°C to 30°C at −4°C h −1 and from 29°C to 20°C at −0.2°C min −1 . Due to dimer formation and aggregation of 13HR, this structure was purified from an agarose gel. To this end, the samples were loaded on a 1 % (w/v) agarose gel (0.5 × Tris-borate-EDTA (TBE) buffer, pH 8.3, containing 44.5 mM Trizma base, 44.5 mM boric acid, 1 mM EDTA supplemented with 11 mM MgCl 2 ) and run for 2.25 h at 80 V. For visualization, EtBr (final concentration of 0.46 µg mL −1 ) was used and the target band was cut out under ultraviolet (UV) light. The gel was cut in small pieces which were added into "freeze'n'squeeze" cups (Bio-Rad) and frozen for 5 min. Subsequently, the DNA origami was recovered after centrifugation at 16,000 g and 10°C for 10 min. The origami concentration was estimated according to Lambert-Beer's law from the absorbance measured at 260 nm (BioTek Eon Microplate Spectrophotometer, 2 µL sample volume, Take3 plate). The extinction coefficient of DNA origami structures is structure specific and it is estimated from the number of hybridized and non-hybridized nucleotides (22) (Table S1).

Note S21: Isolation of native CCMV
Native CCMV particles were grown in and isolated from cowpea plants. Briefly, leaves of ten-day old plants were inoculated with CCMV, being either a suspension containing purified virus particles or infected cowpea leaves. After seven to ten days the plant material was harvested, homogenized in 0.2 M sodium acetate buffer, pH 4.8, supplemented with 0.01 M ascorbic acid and 0.01 M disodium EDTA, and pressed through a cheesecloth. After 1 h incubation at 4°C, leaf tissue was pelleted by centrifugation at 10,000 rpm and 4°C for 10 min. The CCMV containing supernatant was used to dissolve 10 % (w/v) solid PEG (MW = 6,000 g mol −1 ) and CCMV was precipitated by centrifugation at 10,000 rpm and 4°C for 15 min, after which it was resuspended in cold virus buffer (0.1 M sodium acetate, pH 5.0 supplemented with 1 mM EDTA and 1 mM sodium azide). Undissolvable material was removed by pelleting it at 10,000 rpm and 4°C for 10 min. The virus particles were further purified by a density gradient centrifugation (≤16 h, 40,000 rpm, 10°C) using cesium chloride (37.5 % (w/w). The brownish, virus containing fraction was dialyzed against virus buffer (3 × 3 h, 4°C) before isolating the CPs (23).

Note S23: Recombinant expression and purification of MPyV capsomers
Wildtype VP1 proteins were recombinantly expressed in E. coli Rosetta (D3) pLysS cells (Novagen). To this end, a pGEX-4T-VP1 plasmid was constructed by inserting the VP1 gene into the pGEX-4T vector. The sequence encoding VP1 (M34958) (24) was amplified with 30 bp extensions homologous to the flanking vector to facilitate insertion into pGEX-4T (for sequences see Table S3) using an in vivo assembly method (25). For purification purposes, the vector contains a glutathione-S-transferase (GST) tag which is linked to the N-terminus of VP1 via a thrombin cleavage site. The plasmid was transformed into NEB 5-alpha High Efficiency Competent E. coli cells according to the manufacturer's instructions. Successful cloning was confirmed by colony polymerase chain reaction and Sanger sequencing after which the plasmids were transformed into Rosetta cells by heat shock, as per manufacturer's instructions. The expression was performed similar as reported by Chuan et al. (26) Briefly, a single colony was used to inoculate the starting culture (5 mL terrific broth (TB) media containing 12 g L −1 tryptone, 24 g L −1 yeast extract, 0.4 % (v/v) glycerol, 2.31 g L −1 KH 2 PO 4 , and 12.24 g L −1 K 2 HPO 4 , supplemented with 34 mg L −1 chloramphenicol and 100 mg L −1 ampicillin) which was incubated overnight at 30°C and 180 rpm. The starting culture was diluted 100× into the main culture (500 mL TB media supplemented with 34 mg L −1 chloramphenicol and 100 mg L −1 ampicillin) and grown at 37°C and 180 rpm until the optical density at 600 nm (OD 600 ) reached 0.5-0.6. Subsequently, the cells were cooled down in an ice bath and induced with 0.3 mM isopropyl β-D-thiogalactopyranoside (IPTG, 26°C, 16 h). The cells were harvested by centrifugation for 15 min at 4,000 g, 4°C and stored at −20°C. The cells were lysed by resuspending the cell pellets in ca. 40 mL of "storage buffer" containing 40 mM Tris, 200 mM NaCl, 1 mM EDTA, 5 % (v/v) glycerol, and 5 mM DTT, pH 8.0, before sonicating for 3×20 s bursts at 20 % output with 1 min pause on ice between each pulse, followed by centrifugation for 25 min at 25,000 g at 4°C. The supernatant was filtered through a 0.45 µm syringe filter (Merck Millipore) before further purification using affinity (GST Trap FF 5 mL column, GE Healthcare) and size exclusion (Superdex 200 10/300 GL column, GE Healthcare) chromatography (ÄKTA Pure, Cytiva/ NGC Discover, Bio-Rad). After initial equilibration of the GST Trap FF column, the proteins were injected (0.5 mL min −1 flow rate) and the column was washed with "storage buffer". Subsequently, 5 mL of "storage buffer" containing 50-100 units of thrombin were manually loaded onto the column, and the column was sealed and incubated for 16 h at 4°C to facilitate the cleavage of the VP1 from the column. VP1 was eluted with "storage buffer" and subsequently aggregates were removed by passing the sample through the size exclusion column (equilibrated with "storage buffer"). The purity of the sample was evaluated from SDS-PAGE (Mini-PROTEAN TGX Precast Protein Gels, Bio-Rad, 30 min, 200 V in 25 mM Tris, 190 mM glycine, 0.1 % SDS, Fig. S3). The concentration of the capsomers was determined based on their absorbance at 280 nm (extinction coefficient of the monomer of 57,870 M −1 cm −1 ) using a BioTek Eon Microplate Spectrophotometer (2 µL sample, Take3 plate). For in vitro reassembly of the purified capsomers into VLPs, the capsomers were dialysed overnight at room temperature against assembly buffer containing 0.5 M (NH 4 ) 2 SO 4 , 20 mM Tris, 5 % (v/v) glycerol, and 1 mM CaCl 2 , pH7.4 using Slize-A-Lyzer Mini Dialysis cups (3.5 kDa MWCO, Thermo Scientific).  Note S24: Collection of parameters used in cryo-EM and single-particle reconstruction