Chemoenzymatic Synthesis of Fluorinated Cellodextrins Identifies a New Allomorph for Cellulose‐Like Materials**

Abstract Understanding the fine details of the self‐assembly of building blocks into complex hierarchical structures represents a major challenge en route to the design and preparation of soft‐matter materials with specific properties. Enzymatically synthesised cellodextrins are known to have limited water solubility beyond DP9, a point at which they self‐assemble into particles resembling the antiparallel cellulose II crystalline packing. We have prepared and characterised a series of site‐selectively fluorinated cellodextrins with different degrees of fluorination and substitution patterns by chemoenzymatic synthesis. Bearing in mind the potential disruption of the hydrogen‐bond network of cellulose II, we have prepared and characterised a multiply 6‐fluorinated cellodextrin. In addition, a series of single site‐selectively fluorinated cellodextrins was synthesised to assess the structural impact upon the addition of one fluorine atom per chain. The structural characterisation of these materials at different length scales, combining advanced NMR spectroscopy and microscopy methods, showed that a 6‐fluorinated donor substrate yielded multiply 6‐fluorinated cellodextrin chains that assembled into particles presenting morphological and crystallinity features, and intermolecular interactions, that are unprecedented for cellulose‐like materials.


General materials and methods
Chemicals were commercially obtained as reagent grade and used without any purification. Deoxy-fluoro-D-glucoses (2F-, 3F-and 6F-Glc) and α-D-glucose 1-phosphate disodium salt hydrate (Glc-1P) were purchased from Toronto Research Chemicals (Canada) and Sigma-Aldrich (UK), respectively. Cellobiose phosphorylase (CBP) (PRO-GH94-004) was kindly provided by Prozomix Limited (UK) and Milli-Q (MQ) H2O was used to prepare all buffers. Thin-layer chromatography (TLC) was performed on pre-coated silica gel 60 F254 plates (Merck) and compounds were visualised by UV irradiation (λ 254 nm) and/or by spraying TLC with staining solution (2% orcinol w/v in EtOH/H2O/H2SO4 15:1:2 v/v/v) followed by heating. Biotage SP4 flash chromatography system was used for purification of protected monosaccharides using normal phase (pre-packed SNAP cartridges) and the monofluorinated cellobiose analogues were purified by HPLC (Thermo Scientific Dionex Ultimate 3000) on a Luna OH column (5 µm HILIC 200 Å, 250 × 10 mm, Phenomenex) using 5 mM ammonium formate buffer (5%) and acetonitrile (95%) at 5 mL/min in isocratic elution over 25 min. Detection was performed by charged aerosol detector (CAD) with power function 1.00, data collection rate 10 Hz and nebulizer temperature 25 °C. Products were lyophilised using a Labconco FreeZone Benchtop freeze dryer. 1 H, 13 C, 31 P and 19 F NMR spectra were recorded on a Bruker Avance III 400 MHz and/or Bruker Avance Neo 600 MHz spectrometers at 298 K. Chemical shifts recorded in D2O are reported with respect to the solvent residual peak at 4.79 ppm in 1 H NMR. High resolution mass spectra were acquired in a Synapt G2-Si mass spectrometer (Waters, UK) using electrospray ionisation (positive or negative mode). Optical rotations were measured at 20 °C using a Perkin-Elmer Model 341 polarimeter.

Expression and purification of cellodextrin phosphorylase (CDP)
A recombinant plasmid (pET15b) containing the CDP gene from Ruminiclostridium thermocellum (YM4 strain) was transformed into E. coli BL21 (DE3) cells and grown as described previously. 1 Briefly, 1 L of LB medium containing the transformant and carbenicillin (100 μg/mL) was incubated at 37 °C with shaking (200 rpm) until OD600 around 0.6. Heterologous protein expression was induced by adding isopropyl β-D-1-thiogalactopyranoside (IPTG) to a final concentration of 1 mM and incubating for 4 hours at 30 °C with shaking (180 rpm). The cells were harvested by centrifugation (4,000 × g, 20 min), re-suspended in lysis buffer (50 mM HEPES, pH 7.5, 100 mM NaCl, EDTA-free protease inhibitor cocktail tablet, 0.02 mg/mL DNaseI), lysed by cell disruption (30 Kpsi, constant flow) and the supernatant containing the recombinant proteins was separated from cell debris by centrifugation (20,000 × g, 30 min). Proteins were purified at 4 °C using an ÄKTA pure FPLC system (GE Healthcare). The supernatant was loaded to a 5 mL HisTrap TM HP column (GE healthcare) pre-equilibrated with buffer A (50 mM Tris-HCl, pH 8, 50 mM glycine, 5% glycerol, 500 mM NaCl, 20 mM imidazole). The column was washed with buffer A to remove unbound proteins followed by elution of bound proteins with buffer B (50 mM Tris-HCl, pH 8, 50 mM glycine, 5% glycerol, 500 mM NaCl, 500 mM imidazole). Further purification was carried out by gel filtration chromatography (Superdex S200 16/600 column, GE Healthcare) with 20 mM HEPES, pH 7.5, 150 mM NaCl, 1 mL/min. Fractions containing CDP were pooled and concentrated using Amicon Ultra-15 Centrifugal Filter (30,000 MW cut off) and the enzyme concentration (5.7 mg/mL) was determined by NanoDrop TM spectrophotometer (Thermo Fisher Scientific, UK). The His-tag CDP was stored in aliquots at -80 °C until required.

Matrix-assisted laser desorption ionisation time-of-flight mass spectrometry (MALDI-TOF MS)
Samples suspended in MQ water were mixed with equal volume of 2,5-dihydroxybenzoic acid (DHB) matrix (10 mg/mL in 30% acetonitrile in MQ water), spotted on a target plate (Bruker MTP 384 Polished Steel TF Target) and analysed on an Autoflex TM Speed MALDI-TOF/TOF mass spectrometer (Bruker Daltonics TM GmbH, Coventry, UK). The instrument was controlled by a flexControl TM (version 3.4, Bruker) method optimised for peptide detection and calibrated using peptide standards (Bruker). All spectra were processed with flexAnalysis TM (3.4, Bruker).

Electron microscopy (EM)
Transmission Electron Microscopy (TEM) images were viewed on a Thermo Fisher Talos F200C transmission electron microscope at 200kV (Thermo Fisher UK Ltd, Cambridge, UK) using a Gatan OneView 4k X4K digital camera (Gatan, Abingdon, UK) to record DM4 files. 400 mesh EM Resolution Formvar/Carbon coated copper grids were glow discharged, the samples were suspended in 200 µL MQ water and 5 µL drop was pipetted onto the grid for 1 minute. The excess was then blotted off with filter paper and 5 µL of 2% uranyl acetate was pipetted onto the grid for 30 seconds then blotted off with filter paper and allowed to dry.

Atomic force microscopy (AFM)
Samples of fluorinated cellodextrins were deposited onto freshly cleaved mica substrates by drop-casting of diluted suspensions (0.02 mg/ml). Cast drops were allowed to dry before SPM investigation in an ambient environment. SPM measurements were conducted using a Multi-Mode VIII microscope with Nanoscope V controller operating under non-resonant PeakForce feedback control (Bruker, CA, USA). SCANASYST-FLUID+ cantilevers were employed with nominal spring constants of 0.7 N/m (Bruker, CA, USA). Real-time analysis of the force-separation curves were collected as part of the Peakforce control mechanism allowing the physical tip-sample interaction forces and sample topography to be mapped concurrently.

Powder X-ray diffraction (PXRD)
X-ray diffraction data were collected using a single crystal diffractometer (Rigaku Synergy S, Cu X-ray tube, 50kV-1mA) with Cu Kα radiation (λ = 0.154 nm). Samples of enzymatically produced cellodextrin (EpC), 2F-EpC, 3F-EpC, 6F-EpC and multi-6F-EpC were placed in a 96-well plate and analysed using an XtalCheck screening plate mounted on the diffractometer, and each plate was covered with tape. The collected diffraction images were integrated between diffraction angles (2Θ) 5 and 40°. To account for the scattering contribution of the plate and tape, an empty plate covered with tape was used as reference. Hence, the corrected PXRD intensities (Icor) of each sample were obtained by subtraction of the reference intensity (Iref) to the observed intensity (Iobs), as expressed by Eq. S1: Eq. S1 All the samples were analysed in powder form and after gentle grinding with mortar and pestle.
The simulated powder patterns were generated using Mercury 5 and the published crystal structures of cellulose Ia, 6 Ib, 7 II, 8 IIII 9 and IIIII. 10

Raman spectroscopy
Raman images were obtained using a Raman spectrometer (Renishaw, UK). Raman spectra were acquired using a 785 nm wavelength laser (NIR) and 41 mW laser power at the sample for excitation of Raman scattering. The sample was focused with a 50× objective lens (numerical aperture: 0.7, vertical resolution: 1.6 µm) with a lateral resolution of 684 nm. Due to the narrow spectrum acquisition range, each complete spectrum is a composite of three individual spectra centred at 590, 1090 and 1490 cm -1 respectively. Each individual spectrum was obtained from 10 exposures of 20 seconds each. Overlap of the individual spectra enabled their normalisation to produce the complete spectrum. Three complete spectra were acquired for each sample. Raman bands, background fluorescence and photon spikes were fitted for each complete spectrum using Lorentzian curve functions in Fityk, 11 enabling deconvolution of each band and precise identification of their wavelengths and relative intensities. The spectrum for each sample presented in the paper is the average of the three reconstructed spectra produced from the Lorentzian functions after noise (photon spikes, background fluorescence) removal. 1 H-13 C cross-polarisation solid-state NMR experiments of the single 2-, 3-and 6-monofluorinated EpC 10 wt% and multi-6F-EpC 25 wt% hydrogels were performed at 5 °C using a Bruker Avance III spectrometer equipped with a 4 mm triple resonance probe operating at frequencies of 400.2 MHz ( 1 H) and 100.6 MHz ( 13 C). The WPT-CP experiment was carried out using a T2 filter and mixing time of 2 and 16 ms, respectively. The WPT factors shown in Figure 6 were calculated by normalisation of the peak intensities of the spectrum acquired at 16 ms mixing time against a reference spectrum at 0 ms mixing time. Each gel was packed into a kel-f insert, sealed using a plug and a screw, and spun at 6 kHz. The NMR characterisation of EpC dispersed in D2O was carried out using a Bruker Avance I spectrometer equipped with a 5 mm probe operating at frequencies of 499.7 MHz ( 1 H) and 125.7 MHz ( 13 C). Around 600 μL of dispersion (2 w/V%) in 99.9% D2O (Sigma-Aldrich®) was pipetted into a 5 mm NMR tube at room temperature. Both phase-sensitive 1 H-13 C HSQC experiment with 1 H-13 C correlation via double inept transfer and 1 H-1 H COSY experiment with multiple quantum filter and gradient ratio for artifact suppression were acquired with 256 increments in the F1 dimension, 8 number of scans and a relaxation delay of 2 s. A 13 C DEPT135 experiment with 1 H decoupling was acquired using a pulse length of 11.25 μs and 8k number of scans.

Characterisation of cellodextrins by solution and solid-state NMR
The solution NMR characterisation of a 0.5 wt% dispersion of multi-6F-EpC (7) was carried out using a Bruker Avance III spectrometer equipped with a 5 mm inverse triple-resonance probe operating at frequencies of 800.2 MHz ( 1 H) and 201.2 ( 13 C). Phase-sensitive 1 H-13 C HSQC experiment with 1 H-13 C correlation via double inept transfer was acquired at 293K with 128 increments in the F1 dimension, 8 number of scans and a relaxation delay of 3 s. A 1 H-1 H COSY experiment with multiple quantum filter and gradients was acquired at 5 °C with 256 increments in the F1 dimension, 12 number of scans and a relaxation delay of 2 seconds. Finally, a 13 C DEPT135 experiment with 1 H decoupling was acquired using a Bruker Avance Neo spectrometer equipped with a cryoprobe operating at frequencies of 600.2 MHz ( 1 H) and 150.9 MHz ( 13 C). A relaxation delay of 1 s was used and 30k scans were registered.        Figure S8. TEM images of enzymatically produced cellodextrin EpC (a) and enzymatically produced fluorinated cellodextrins 2F-EpC (b), 3F-EpC (c), 6F-EpC (d) multi-6F-EpC (e) negatively stained with 2% uranyl acetate. Scale bars correspond to 100 nm. Figure S9. Predicted diffraction patterns for cellulose types Iα (black), 6 Iβ (red), 7 II (green), 8 IIII (purple) 9 and IIIII (blue). 10 The patterns were generated using Mercury with FWHM set to 0.5. The blue dashed lines represent the experimental diffraction angles obtained for multi-6F-EpC (the diffraction pattern is reported in the main text, Figure 4b).   The 1 H-13 C CP/MAS spectrum of multi-6F-EpC (7) was compared to all previously reported cellulose allomorphs to identify differences and similarities on 13 C chemical shifts (Tables S4,  S5).  Figure S12. Comparison of the 1 H-13 C CP (blue) and 19 F{ 1 H}-13 C CP (purple) NMR spectra of multi-6F-EpC powder acquired at room temperature, 15 kHz MAS rate and a 13 C frequency of 212.5 MHz.

Long-range structural characterisation 2.3.1. Powder X-ray diffraction (PXRD) patterns
The high-resolution solution NMR spectra (COSY and HSQC, Figure S13a) obtained for a diluted dispersion of EpC (8) enabled the detailed assignment of the two anomeric spin systems and the internal and non-reducing terminal protons. On the other hand, the higher presence of chemical environments in multi-6F-EpC (7) complicated the full spectral assignment (COSY and HSQC, Figure S13b), but we could still assign the β-spin system (belonging to the cellobiose unit) and the 6-fluorinated non-reducing terminal (nr). cross-peaks corresponding to the reducing end glucose residue are indicated using the α and β labels, whereas the nr abbreviation is used for the non-reducing terminal. The peaks corresponding to the cellobiose moiety in multi-6F-EpC (7) are indicated with an asterisks (*).
To characterise the internal dynamics of these materials, 1 H-13 C CP/MAS NMR experiments at varying contact times were carried out for EpC and multi-6F-EpC ( Figure S14). Very similar build-up curves were obtained for all carbon peaks in both powdered samples, which suggest that, at the molecular level, the CP-observable domains (i.e. rigid) of EpC and multi-6F-EpC nanofibrils present very similar dynamics.