Atomic-level engineering and imaging of polypeptoid crystal lattices

Significance A fundamental challenge in materials science is to understand the atomic-level structures of nanoarchitectures assembled from synthetic polymers. Here, we report a family of sequence-defined polypeptoids that form free-floating crystalline 2-dimensional nanosheets, in which not only individual polymer chains and their relative orientations, but also atoms in nanosheets were directly observed by cryogenic transmission electron microscopy. These atomic details are inaccessible by conventional scattering techniques. Using the feedback between sequence-controlled synthesis and atomic imaging, we observed how the nanosheet structure responds to chemical modifications at the atomic-length scale. These atomic-level insights open the door to the design of bioinspired nanomaterials with more precisely controlled structures and properties.

were purchased from Sigma Aldrich. Rink amide resin was purchased from NOVABIOCHEM.
All the other solvents and reagents described here were purchased from commercial sources and used as received. All diblock copolypeptoids were synthesized using automated solid-phase submonomer synthesis on a Symphony X peptide synthesizer at a scale of 200 mg Rink amide resin (0.64 mmol/g) by adapting reported procedures. The resin was swelled in DMF for 10min and the Fmoc group on the resin was deprotected with 20% (v/v) 4-methylpiperidine/DMF. The bromoacylation reaction was then performed with bromoacetic acid (0.8 M) and N,N'diisopropylcarbodiimide (DIC, 0.8M) in DMF at room temperature for 20min. The subsequent displacement reaction with various submonomers was performed at 1 M amine concentration in N-methyl-2-pyrrolidone (NMP) at room temperature for 30min. The crude diblock copolypeptoids were cleaved from the resin by treating with 95% (v/v) trifluoroacetic acid (TFA) in water for 10 min at room temperature, followed by filtration and washing of the resin with DCM. The solvent was evaporated using Biotage® V-10 evaporator and the crude products were lyophilized from acetonitrile/water (1:1, v/v).
The crude peptoids were then purified by Waters reverse-phase HPLC on a C18 semipreparative column (5 µm, 250 mm × 21.2 mm, C18 Vydac column) using acetonitrile with 0.1% TFA (solvent B)/water with 0.1% TFA (solvent A) with a flow rate of 15 mL/min. The linear gradients used for the purification of Nte4-Npe6, Nte4-N4Brpe6, Nte4-(N4Brpe-N4pe)3 and Nte4-Nmpe6 are 30-70% B, 40-80% B, 40-80% B and 40-80% B over 30min, respectively. The fractions were analyzed by Waters ACQUITY reverse-phase UPLC with a ACQUITY®BEH C4 column (1.7 µm, 2.1 mm × 50 mm) connected with a Waters SQD2 mass spectrometry system using 20-80% gradient acetonitrile in water with 0.1% TFA at 0.4 mL/min over 6.8 min at 60 o C. The fractions containing pure compound were collected and the solvent was removed by Genevac evaporator, followed by lyophilization from acetonitrile/water (1:1, v/v) to obtain the fluffy white powder. 40-50 mg of final peptoid with > 95% molecular purity was obtained. All the purified polypeptoids were characterized by NMR spectroscopy on a Bruker Avance II 500 MHz at room temperature. Low magnification images of nanosheets were obtained from dry specimens. These specimens were prepared by depositing a 3 µl droplet of the desired nanosheet-containing aqueous solution on a continuous carbon film that was supported on a copper grid. The droplet was blotted from the edge of the grid using a filter paper. The grid was transferred to a TEM cryo holder (914 Gatan Inc.) and micrographs were collected on a Philips CM200 at 200 kV using a Gatan US1000 CCD camera at liquid nitrogen temperature in low-dose mode to minimize radiation damage. We expect the nanosheets to be completely dry due to exposure to vacuum, first in the airlock and then in the column of the TEM.

Self
Image processing. Images of 2D crystals are generally not perfect due to dislocations, distortions from stress and image distortion within the microscope, which cause high resolution diffraction spots to be smeared out. In order to recover the high spatial frequency signal, a crystal unbending process was conducted on all micrographs. The motion-corrected and summed lowdose micrographs were imported into 2dx, an image processing package for 2D electron crystallography (3)(4)(5). Details of the principle of crystal unbent processing can be found in Henderson et al.'s work (6,7). Briefly, the position of each unit cell in the image is found by cross correlation with a small reference area, and a smooth function is defined for displacements from the ideal lattice. This function is then used to re-interpolate the image onto the regular lattice. The defocus values, astigmatism, and specimen tilt geometry were determined using the gCTF program and corrected after unbending.
Overlapping small square boxes (150 pixels long on each side), were extracted from the micrographs. The centers of the boxes coincide with the centers of the locations of unit cells as described in the previous work (8). The extracted boxes were sorted into image classes using the Relion software package (9,10). Relion uses the maximum-likelihood approach for sorting the small boxes extracted from cryo-TEM micrographs (11). This approach has been proven to be particularly useful in the classification of structurally heterogeneous data from biomolecules. The intensity in the boxes is first normalized. A soft round mask is applied to the images in boxes to reduce their background noise. Reference-free class averages are obtained in a completely unsupervised manner by starting multiple references from average images of random sets of the normalized images in the extracted boxes. All images are compared to all references in all possible orientations and probability weights are calculated for each possibility instead of assigning images to one particular class or orientation. Class averages are then calculated as weighted averages over all possible assignments. The number of classes requested is set by the user: fewer boxes participate in each class with increasing number of classes. Averaging over a small number of boxes leads to noisy averages, which result in suboptimal alignment and classification. The number of classes requested can be set within a wide range. In this study, the classification analysis suggests the presence of homogenous crystal motifs in the Nte4-Npe6, Nte4-N4Brpe6, and Nte4-(N4Brpe-N4pe)3 nanosheets when one class was applied to those nanosheets. However, 4 classes were applied to the Nte4-N4mpe6 nanosheets due to the presence of heterogeneity. The averaged images in 4 classes comprising similar or nearly identical motifs were separated into 3 groups.