Highly aligned bacterial nanocellulose lms obtained during natural biosynthesis

Nerea Murugarren ICMAB-CSIC Soledad Roig-Sanchez ICMAB-CSIC Irene Anton-Sales ICMAB-CSIC Nanthilde Malandain ICMAB-CSIC Kai Xu ICMAB-CSIC Eduardo Solano NCD-SWEET beamline, ALBA synchrotron light source https://orcid.org/0000-0002-2348-2271 Juan Reparaz Institute of Material Science of Barcelona https://orcid.org/0000-0001-9679-0075 Anna Laromaine (  alaromaine@icmab.es ) ICMAB-CSIC https://orcid.org/0000-0002-4764-0780

and how the anisotropy of this material impacts its hydrophilicity, biocompatibility and thermal conductivity.

Results And Discussion
Aligned bacterial cellulose biosynthesis Inspired by the biosynthesis process, we restricted the bacteria movement to control the alignment of individual cellulose nano brils. K. xylinus is an aerobic bacteria strain 6,34 , which implies the microorganism requires oxygen when cultured. We retained the bacteria anchored on an agar surface at the bottom of a vessel, lled with fresh bacteria-free liquid media, and cultured the bacteria at their optimal cellulose production conditions: 30°C in static (Fig. 1A,B). After 5 days of culture, we obtained the typical circular hydrogel pellicle of BNC at the air-liquid interphase and we could also visualize in the liquid a translucent lm of aligned BNC bers, as shown in Fig. 2C,D. The water-insoluble and low dense BNC bers facilitated this method. We separated the top part of the produced cellulose (BNC) and the aligned cellulose (A-BNC) in the middle from the agar substrate. Therefore, we obtained BNC of different morphologies in a straight-forward step.
We cleaned the lms from bacteria detritus and culture media by the commonly described treatment 35 .
We obtained semi-transparent hydrogels of BNC of the shape of the vessel, in our case circles of Ø1.6 cm 2 , and aligned bacterial nanocellulose (A-BNC) of 7-8 cm in length. BNC and A-BNC lms were dried at 60°C for a minimum of 12 hours under 1 Kg weight and without weight, respectively. After drying A-BNC maintained the aligned structure and did not decrease in length. The thickness of the BNC lms of 300 µm in wet decreased to ≈20 µm upon drying. A-BNC lms were easily manipulable and exible, with a nal thickness of ≈10 µm after drying. Optical images of the A-BNC clearly show the alignment of the bers (Fig. 2C,D); conversely, the random BNC does not show any orientation of its structure. Moreover, A-BNC did not disassemble upon different wetting and drying cycles indicating a stable bonding among cellulose bers, conferred by hydrogen bonds between hydroxyl groups 36 .
The process has been repetitively performed more than 200 times, obtaining aligned cellulose consistently. Many parameters control the BNC production and they have been extensively evaluated elsewhere 10,11,37 . However, dissolved oxygen (DO) in the media is considered the main driver (Fig. 2B). We quanti ed the dissolved oxygen (DO) present in the solution at different time points using indigo-carmine reaction 38, 39 (Fig. S1). Leuco-indigo carmine reagent oxidizes under the presence of oxygen to keto-indigo carmine, which causes a color change from yellow to red upon oxidation (Fig. S1-A). A color-DO chart allows to qualitatively estimate the DO concentration ( Fig. S1-B). As shown in Figure S1-C, the initial DO concentration at the top and bottom part of a non-oxygen permeable tube was ≈0.06 ppm. After 7 days of culture, the DO concentration at the top was maintained at 0.06 ppm while, at the bottom, it decreased to 0.03 ppm. We hypothesize that the deployment of oxygen at the bottom of the vessel over time promotes the movement of the bacteria to the surface, towards higher DO content. Bacteria produce low-density cellulose bers from the bottom of the vessel and uses them as "climbing ropes" (Fig. 2B) to reach the high oxygen-content surface where continue producing its exoproduct with no movement restriction, which allows us to obtain BNC as a random mesh. So, e ciently from the same biosynthesis process, we can obtain random BNC and A-BNC by restricting the initial movement of bacteria.
To further exploit the e ciency and yield of our new method we investigated the reuse of the bacteria. From the produced lms we cut out the aligned BNC section and again con ned it at the bottom of the tube, lled with fresh media: a) the previously used agar substrate (Fig. S2-A), and b) the formed BNC pellicle (Fig. S2-B) before cleaning it, both containing cultured "active" bacteria. We con rmed that we could use the same agar substrates with bacteria colonies (Fig. S2-A) for at least 7 cycles, obtaining BNC and A-BNC in only 3 days. After that, the bacteria decreased the speed of production. We also proved the use of the as-produced BNC as a substrate to grow A-BNC and BNC in, again, only 3 days, which we named lm-to-lm biosynthesis ( Fig. S2-B). The reuse of those bacteria substrates in sequentially A-BNC production increased the number of lms obtained exponentially. We hypothesize that the already used bacteria are acclimatized to the liquid culture, and thus, more "active", which reduces the lag phase of a bacteria culture from the initial 5 days of the rst culture to 3 days.
To increase the size of the A-BNC produced we evaluate the time, vessel and bacteria amount. Culturing the bacteria from 3 to 15 days did not increase the width or length of A-BNC produced, although it increased the thickness of the BNC lm. On the other hand, longer and wider A-BNC lms could be obtained using longer and wider tubes, respectively. Using tubes of 10 cm and 30 cm allowed us to obtain A-BNC 7-8 cm up to 20-23 cm respectively, however, the width of the A-BNC was not increased (3-5 mm). Finally, instead of using the typical streak plate (or zig-zag) method to obtain the initial agar substrate with bacteria colonies, we grew con uent agar Petri dishes to obtain the maximum number of bacteria per surface to obtain high-density A-BNC ber lms. This strategy allowed us to obtain A-BNC lms with a thickness of ∼200 µm in wet.

A-BNC Characterization
This biosynthesis process produced aligned BNC lms, which possess different morphology and properties than traditional BNC lms with randomly distributed bers. Therefore, an extensive characterization of both A-BNC and BNC (considered as a control) was performed to evidence the change of morphology and evaluate the properties arising upon the alignment of the material.
After drying, A-BNC is more transparent than BNC, and this increase can be qualitatively appreciated by the naked eye, clearly seen in Fig. 3B and 3C, respectively, where pictures are taken with a micron-mesh under the lms and top illumination. The scattering of light of BNC produced a haze and whitish color to the material that was not seen in the A-BNC. Moreover, under localized white light, the A-BNC lms showed some iridescence indicating higher transparency and structuration of the material (Fig. 3D).
As seen in Fig. 3E, FTIR for both samples (BNC (blue) and A-BNC (black)) contained the cellulose characteristics peaks: hydroxyl groups are seen as a strong peak at 3000-3700 cm -1 , CH groups at 2800-1340-1480 cm -1 , as well as other cellulose characteristics peaks. However, the A-BNC spectra shows more intense transmittance peaks in the OH and CH region, even though A-BNC lms were thinner than the BNC.
To further investigate the structure of the material, the ber alignment was also analyzed at the nanoscale (Fig. 4). At this scale the collapse of the structure upon drying made more di cult the observation of the alignment observed at the macroscale. Nevertheless, a higher signi cant degree of organization was observed in the A-BNC in comparison to BNC. Fig. 4-B,E show the Transmission Electron Microscopy (TEM) images and Scattering Angle Electronic Diffraction (SAED) patterns of BNC ( Fig. 4-B) and A-BNC ( Fig. 4-E). Although BNC is known to present high crystallinity (60-80%) 34 the random orientation of the bers makes it di cult to analyze the crystallinity of the BNC pellicle by TEM. However, SAED analysis in A-BNC samples allowed visualizing the cellulose crystallinity lattice planes (indexed in Figure S3-A). We considered that this increase discretization of the lattice planes intensity (from diffuse bands to localized bright spots, Fig. S3) was due to the increased order orientation and different texture of A-BNC. SEM and AFM images show a similar compact structure for the BNC ( Fig. 4-C,D) and A-BNC ( Fig. 4-F,G) lms, however, we could see an alignment of bers in the latter. Diameter measurements of the nano bers (2-4 nm) and micro brils bundles (15-20 nm) from both BNC and A-BNC revealed that the modi cation in the culture to obtain aligned BNC did not impact the single ber structure. Using the color analysis from the OrientationJ plugin for ImageJ we computed the local orientation as a color-survey HSB image, where hue represents the ber orientation and saturation represents coherency of the orientation. These images 18.81%, con rming that the latter was more aligned than BNC. Interferometry assisted us to visualize the alignment at a micron scale, obtaining volumetric images of the aligned cellulose BNC bundles ( Fig. S4-A), in contrast with the non-aligned BNC ( Fig. S4-B) where non-bundle ordering is appreciated.
To evaluate the alignment in the 3D volume of the material, we used interferometry, Small-angle X-ray scattering (SAXS) and polarization-resolved second-harmonic generation signal (P-SHG). The raw 2D scattering SAXS patterns of BNC and A-BNC bers, respectively, recorded at the ALBA synchrotron facility, differ on the shape of the scattering pattern; being circular for randomly distributed bers of the BNC (Fig. 5-A) and distorted (oval-shaped) for preferential orientated bers of the A-BNC sample ( Fig. 5-B), indicating a higher degree of alignment of the A-BNC bers. To quantitatively evaluate differences in the orientation degree, the spectra were integrated as shown in Fig. 5-C, as an intensity vs azimuthal degree plot from -90º to 90º. A-BNC showed an intensity peak centered around 0º, whereas BNC gave no peak at all. From the spectra, the widely accepted orientation quantitative parameters OI, FWHM, and Sparameter 31, 26 are obtained (formulas and explanation in the methods section). FWHM stands for full width at the half-maximum intensity and the smaller this value is, the highest is the orientation. The orientation index (OI) is obtained from FWHM, ranging from 0 for randomly oriented structures to 1 for perfectly aligned bers. Herman's order parameter, also called S-parameter, can be calculated from the intensity and azimuthal degrees values and its value also ranges between 0 to 1, being 1 a perfect orientation 31,26 . Thus, as shown in Fig. 5-D FWHM, OI, and S-parameter were computed, obtaining 47.18º, 0.74, and 0.85 for A-BNC, respectively. For BNC, S-parameter resulted to be much smaller than A-BNC, 0.16. FWHM and OI could not be determined due to the lack of a peak in the spectra, and thus, lack of internal orientation. The numerical results obtained corroborated the qualitative information extracted from the 2D images, obtaining a higher volumetric ber orientation evaluation in the alignment of A-BNC lms. The orientation values OI and S-parameter we obtained (0.74 and 0.85, respectively) were higher than previously reported values from SAXS analysis of aligned BNC bers obtained with wet spinning; with OI of 0.69 and an S-parameter of 0.63 31 .
Secondary Harmonic Generation (SHG) is a nonlinear optical process used in biomedicine for imaging SHG active biological structures such as collagen nano bers, rather than its bulk 40,41 . Polarizationresolved SHG microscopy (PSHG) is a powerful tool that exploits the dependence of SHG signals on the polarization degree of the excitation beam. The SHG emission depends on the structuration of the imaged material and PSH includes an extra dimension that is used to probe the molecular organization. Here, PSHG was used to assess the bacterial nanocellulose intrinsic organization. Frames of the linear polarization of the incident beam from 0 to 180 degrees in steps of 10 degrees using a λ/2 plate were recorded for the whole volume of the samples. Fig. 6-A shows how the changes of light intensity in A-BNC from the minimal (70º) to the maximum projection (150º) could be appreciated, whereas in BNC, all projections remained unchanged, indicating no alignment. We evaluated the alignment coherency for each frame by the OrientationJ plugin for ImageJ and plotted it ( Fig. 6-B), where random BNC has a high but invariable signal intensity (blue line) and a low alignment coherency (dotted blue line). Conversely, A-BNC alignment coherency (dotted black line) is even higher than the signal intensity (black line) at the maximum projection and lower in the minimum projection, con rming the alignment of the material. These features were observed for the whole thickness of the samples, con rming a volumetric alignment in A-BNC.

Optical properties
As we have presented, A-BNC lms exhibit iridescence and strong light polarization properties. As shown in Fig. 7, the visualization of BNC bers under a polarized light microscope with the same orientation as the incident light (0º angle) absorbed light and were not seen. Conversely, when the light source was located at 45º from the ber orientation, the bers are not seen. BNC did not exhibit any light polarization neither at 0º and 45º ( Fig. 7-A,D), con rming the random distribution of the bers. On the other hand, A-BNC lms gave dark-light patterns of light when disposed at 0º and 45º from the incident polarized light beam, respectively ( Fig. 7-B,E), con rming the anisotropy of the material. Moreover, A-BNC is a exible material and can be folded into any desirable shape, such as an "N". As seen in Fig. 7-C,F, N-shaped lms of A-BNC created a pattern of light polarization at different positions.

Hydrophilicity and cell substrates evaluation
Most of the electrospun polymers are rather hydrophobic, which is unfavorable from the point of view of tissue engineering. For instance, aliphatic polyesters such as PLLA or PCL show contact angles in the range of 116-135°, while for tissue engineering requirements, it ought to be below 100° 42,43 . Niemczyk-Soczynska et al. 44 studied the highest level of broblasts cell attachment at hydrophilic surfaces, and the best results were observed for surfaces with contact angles in the range of 20-60°4 5 . As the use of cellulose, and speci cally, bacterial nanocellulose drives many applications in medicine 46-48 for their biocompatibility and hydrophilic character we wanted to analyze the impact of the ber alignment in the hydrophilic character of bacterial cellulose. We deposited a blue colored drop of water over A-BNC lm and we could clearly see a faster horizontally spreading of the liquid in comparison to BNC, indicating a different structure, Figure 8A. To quantify the hydrophilicity of both BNC and A-BNC we performed "apparent contact angle" (ACA) measurements. After 5 seconds of depositing a water drop over the surfaces, we observe an ACA of 0° for A-BNC ( Fig. 8-B) whereas it remained 29.7° for BNC after 5 seconds ( Fig. 8-C). These results indicated that the alignment of bers seemed to increase the hydrophilicity of the lms, which also it matches the lower roughness of A-BNC in AFM and the increased density of OH groups by FTIR.
This change of hydrophilicity was further explored in vitro in cell cultures. A human dermal broblast (hDF) cell line (1BR.3.G) was cultured on A-BNC and BNC lms and the cell attachment and growth directionality was assessed visually and with orientation coherency measurments with the OrientationJ plugin of ImageJ. As seen in Fig. S5, cells attached and proliferated on A-BNC, BNC and control (glass slide) until reaching a 100% con uence at day 7 (Fig. S5, Fig. 8-D,E). As seen in Fig. 8-F, hDF cells cultured on aligned cellulose bers tend to be more elongated (orientation coherency of 0.25) than on random ber distributions (BNC lm) (orientation coherency of 0.12) and on the slide (orientation coherency of 0.8) The increase of hDF directionality indicated that cells perceived at some extend the alignment of the cellulosic bers.

A-BNC thermal conductivity
The advent of more sophisticated implants combining novel biomaterials, polymers, electrical components among others sustained that thermal management becomes critical 49 . Most of the biomaterials have an amorphous structure meaning they are thermal insulators, however, in order to develop materials with enhanced thermal conductivity, self-assembled biomaterials with higher degree of crystallinity and/or higher anisotropy is desirable. BNC has already shown promising results as scaffold in regenerative medicine 17,19,50 , therefore we studied the in uence geometrical order on the thermal conductivity of BNC bers. In particular, we show that it is possible to enhance the thermal conductivity by directional alignment of the bers, i.e. in the direction parallel to the alignment direction. The thermal conductivity of A-BNC was investigated by two independent approaches, i.e. using the well-known 3ωmethod 51 , as well as through a recently developed contactless thermore ectance frequency-domain approach developed by L. A. Pérez, et al. 52 , and labeled after Anisotropic Thermore ectance Thermometry (ATT). The later methodology is particularly suitable to study thermally anisotropic materials (e.g. aligned BNC), since it delivers the angular distribution of the thermal conductivity perpendicular to the surface of the sample. In all cases the BNC bers were deposited on silicon substrates, and the thickness of the BNC samples was ∼10 µm. Fig. 9-A displays the 3ω measurements of A-BNC in perpendicular (red) and parallel (blue) to the alignment direction. Whereas the lower frequency range is dominated by the thermal response of the Si substrate, the higher range contains information on the thermal properties of the bers.
The thermal conductivity is estimated using the "slope method" (κα[∂V 3ω /∂log(f)] − 1 ) in the higher frequency range. The lower frequency range is dominated by thermal signal from the substrate, and it is used as calibration for each measurement (κ Si = 150 W/m·K). In the parallel direction, A-BNC shows a larger thermal conductivity at 1.63±0.15 W/m·K, whereas in the perpendicular direction the thermal conductivity is 0.3±0.02 W/m·K, i.e. slightly lower to what we previously obtained for BNC, 0.5±0.05 W/m·K 53 . The alignment of the bers led to a 5-fold increase of the thermal conductivity of the BNC in the parallel direction, Fig. 9-A. Similar results were obtained with ATT, where thermal conductivity was measured almost continuously with 1º angular steps from the alignment dire ction, Fig. 9-B. The highest value for A-BNC, at its parallel direction, (1.42±0.15 W/m·K), is similar to that observed using the 3ωmethod in the same direction (1.63±0.15 W/m·K), and it is on the same order as polyethylene (PE) (1.33 W/m·K) 54 , and larger than collagen (0.5 W/m·K) 55 , which are materials commonly used in implants. Therefore, geometrical alignment of the bers offers the possibility to control at different planes the thermal conductivity just by tailoring the anisotropy of the biomaterial 54 . Interestingly, the perpendicular component of the thermal conductivity from the ATT scans (0.87±0.09 W/m·K) resulted twice as large as compared to the value obtained using the 3ω-method (0.3±0.02 W/m·K). The origin of this different values in the perpendicular direction possibly arises from different thermal resistances between parallel bers. From those results we foresee that the potential of natural-based materials, which may outperform in some properties to polymer and synthetic materials.

Conclusions
Inspired by the bacterial cellulose biosynthesis we presented a facile and reproducible method to obtain lms of aligned cellulose bers (A-BNC) and non-aligned cellulose (BNC) bers in a single step. Optical, chemical, and physical techniques evaluated con rmed the alignment of the BNC bers super cially and in the whole volume of the lm. Cellulose bers are strongly interacting creating an aligned lm which are maintained upon different dry-wetting cycles and exhibit polarization of light and the thermal conductivity of this material is modi ed due to the novel structure. The thermal conductivity of A-BCN was thoroughly studied with innovative techniques which con rmed a 5-fold increase of the thermal conductivity in the parallel direction, 1.63±0.15 W/m·K, whereas in the perpendicular direction the thermal conductivity is 0.3±0.02 W/m. The properties risen by the alignment encompasses the properties already exploited for celluloses as their easy functionalization, biocompatibility and purity which bring cellulose to be exploited in an even larger number of elds, and even join the selective group of biomaterials to be part of our future and active implants.

Biosynthesis of aligned BNC, A-BNC, lms and BNC
Biosythesis of A-BNC and BNC. Komagataeibacter xylinus (K. xylinus) strain colonies (NCIMB 5346 from CECT, Valencia, Spain) were grown con uently on Hestrin-Schramm (HS) solid medium 6 in Petri dishes. The media consists of 1.15 g/L of citric acid, 6.80 g/L of Na 2 HPO 4 ·12H2O, 5.00 g/L of peptone, 5.00 g/L of yeast extract, 15.00 g/L of agar and 20.00 g/L of dextrose all from Condalab, dissolved and autoclaved in 1L Milli-Q water. K. xylinus strain were cultured in HS solid medium Petri plates for 15 days at 30 ºC. Next, agar squares of 70 mm side with K. xylinus grown colonies were placed carefully facing up with a sterile spatula at the bottom of a sterile tube with oxygen-permeable tap lled with 10 mL sterile bacteria-free liquid HS media, which consists of the same compounds as the solid media without agar.
The system was cultured for 5 days at 30ºC under static conditions, until a layer of BNC hydrogel (random ber organization) was detected at the air-liquid interface. The A-BNC lms were obtained within the length of the tube.
Bacterial Nanocellulose Cleaning. Once produced, A-BNC and BNC lms were separated by carefully cutting and pulling out from the extremes of the lms using Te on tweezers. After separating the lms, the cleaning process took place separately. BNC lms were soaked and stirred with a magnetic stirring plate in the following steps: (i) 1:1 Ethanol: Milli-Q water solution for 10 min, (ii) 40 min in boiling Milli-Q water, and (iii) two periods of 20 min in 0.1 M NaOH (Sigma-Aldrich) aqueous solution 6 . Finally, the BNC lms were rinsed with Milli-Q water until reaching neutral pH, autoclaved at 121 ºC for 20 min and stored suspended in water in glass vials at room temperature. To avoid entanglement of the aligned bers, A-BNC lms were gently cleaned with the same steps as BNC but without magnetic stirring.
Bacterial Nanocellulose Drying. BNC samples were placed between two Te on plates and dried at 60 ºC for at least 12 hours under a 1 Kg weight. A-BNC-lms were dried on top of a single Te on plate and without any applied weight, at the same conditions.

BNC and A-BNC Characterization
Atomic Force Microscopy (AFM). A-BNC and BNC lms were dried on top of a thin conductive Si surface. 10 x 10 µm 2 micrographs were obtained using a modular PM/AFM (Keysight 5500 LS SPM/AFM), using tapping mode.
Contact Angle (CA). 3 mm width and 10± 2 µm thick dry A-BNC and BNC lms were xed at on a Te on plate with tape. The surface wettability and capillarity of the materials were assessed by a contact angle measurer (KRÜSS Drop Shape Analyzer DSA 100), using the sessile drop method. 2 µL of Milli-Q water were placed on the BNC surface. CA values were computed over 0.5 s for 10 s after depositing the drop.
Dissolved oxygen content. The reagent Indigo-Carmine was used to quantify the dissolved oxygen (DO) content in liquid HS culture media. The Indigo-Carmine method 38,39 is a simple, rapid, and accurate colorimetric procedure for determining small amounts of DO in water (0 to 50 ppm). First, 9 mg of the powder reagent (Sigma-Aldrich) was hydrated with 2.5 mL of Milli-Q water, which contained 0.1 gr of dextrose (Condalab). At the same time, 10 mL of 37.5% of KOH (Sigma-Aldrich) was mixed with 37.5 mL of ethylene glycol (Sigma-Aldrich). Both mixtures were added together and refrigerated until the Indigo-Carmine reagent was completely reduced. This could be noted due to the color change: starting from blue to red, and nally turning yellow. To test the water sample (refrigerated in advance), 0.1 mL of the reduced solution of Indigo-Carmine was added to a 7.5 mL liquid sample inside a vacuum free container. The presence of any DO oxidized the reagent and made the solution change color. The variation in color of the dye is directly proportional to the amount of DO present in the sample, which can quantitatively be determined by using a color scale (Fig. S1-B).
Fourier-Transformed Infrared (FTIR) Spectrophotometry. Dry A-BNC and BNC lms were analyzed with a FT/IR spectrophotometer (Jasco 4700LE). The transmittance spectra collection was performed at 2 cm -1 .
For each spectrum, 32 scans were co-added over the measuring range 400-4000 cm -1 . Air was used as blank (machine lid open). The spectra were processed with the Spectra Manager™ Suite software for the reduction of CO 2 and H 2 O noise levels, baseline correction, smoothing and peak nd.
Optical microscopy. The morphology and structure of A-BNC and BNC was assessed using a conventional optical microscope (Olympus RXSITRF 52787, MAB INDUSTRIAL). To obtain polarized microscopy images, a retardation slide accessory made of a birefringent material (Olympus U-PO3) was coupled to the optical microscope. For observing the change of light pass through the dry A-BNC and BNC lms, the polarized light direction was kept constant and the sample was circularly rotated in the same plane within a turning center point for 360º. Photos were taken every 45º.
Fiber orientation analysis. The A-BNC bers orientation was analyzed using the OrientationJ 60,61 Java plugin for ImageJ/FIJI (D. Sage, EPFL, 2.0.5 version). The OrientationJ structure tensor computes the orientation and isotropy properties in a local window. The local window is characterized by a 2D Gaussian function of standard deviation σ, based on the evaluation of the structure tensor in a local neighborhood. The parameter σ (expressed in pixel units) is a critical parameter that determines the scale of the analysis. It should have a value roughly close to the structure of interest (e.g. thickness of the cellulose lament). The smallest local window available was 1 px, which was the set value used for this analysis. From the plugin, the "Color Analysis", "Distribution", and "Dominant Direction" modes were used to analyze images. OrientationJ was also used to obtain maximum frequencies from the angle distribution to assess the orientation coherency in several imaging techniques (equation (1)).
where α represents the angle between an individual ber and the average ber orientation. The coherency ranges from 0 to 1, where 0 represents an isotropic image without any preferential orientation and 1 represents a perfectly aligned distribution image 22 .
Polarization-resolved second-harmonic generation microscopy (PSHG). Semi-wet A-BNC and BNC lms were stained for 5 min with 5 µL with an aqueous solution of 0.25 mg/mL Brightener 28 (seen as bright blue, exc 365 nm) in Milli-Q water. Then, they were observed under a custom-made multiphoton microscope system, which works both as a two-photon excitation uorescence microscope (TPEF) and as a polarization-resolved second-harmonic generation microscope (PSHG) 40 Olympus). Immersion oil was added. By re ection, TPEF signal is obtained backward after re ecting Nikon uorescence cubes and passing through IR lter. SHG signal is captured by a microscope objective (NA=1.1), and obtained after re ecting a dichroic mirror that rejects backscattered laser light and modi es the ellipticity of the polarization states and passing through narrowband and IR lters. Bright eld, TPEF, and PSHG image stacks were obtained from 0º to 180º, as well as Z stacks. Further analysis was performed pixel by pixel using the stack options "Z project" and "plot Z-axis pro le" from the heat map ltered image stacks in ImageJ. The OrientationJ ImageJ plugin was used to assess the orientation coherency of the bers in the images (equation (1)).
Interferometry. A-BNC and BNC lms were dried on top of a glass coverslip. 3D images of 450 x 350 x 1.50 µm 3 (A-BNC) and 85 x 100 x 1.50 µm 3 (BNC) were obtained with a high-resolution dual-core confocal and interferometry pro lometer (Leica DCM 3D optical pro lometer), using a "retouching surface" operator.
Scanning Electron Microscopy (SEM). A-BNC and BNC lms were xed at on top of aluminum SEM holders with adhesive carbon tape. Images were obtained using a high-performance SEM (QUANTA FEI 200 FEG-ESEM) without metallization at EHTs of 5-10 kV and 50 Pa (low vacuum conditions).
Small Angle X-ray Scattering (SAXS). The scattering measurements were performed at NCD-SWEET beamline at ALBA Synchrotron light facility, in Cerdanyola del Vallès (Barcelona). A monochromatic X-ray beam of 8 keV (λ = 0.154 nm) was set using a Si (1 1 1) channel cut monochromator. An array of Be lenses was employed to collimate the X-ray beam, obtaining a beam size at the sample position of 50 × consists of a pixel array of 1043 × 981 (V × H) with a pixel size of 172 × 172 µm 2 . The scattering vector q (de ned as q = 4πsin(θ)/λ, being θ is the scattering angle and λ the X-rays wavelength) was calibrated using Silver Behenate as reference, obtaining a sample to detector distance of 6700 mm. Dry A-BNC and BNC samples were xed at on 0.50 cm Ø metal rings, where the samples were held by the edges. Samples were scanned perpendicularly to the beam direction, with an acquisition time of 30 s. Azimuthal integration of the obtained signal was done using PyFAI, limited by 270º of azimuthal range due to the intrinsic detector gaps limitations but enough to evaluate the 180º of the 2-fold ber symmetry. The azimuthal distribution was normalized against the maximum value. The integrated scattering pattern generated a distribution of the diffraction intensity along the Debye-Scherrer ring, I(Φ) (arb. u.), vs azimuthal angle, Φ (º). A tting of the distributions was performed using the OriginLab (OriginLab Corporation) non-linear function Lorentz equation (equation (2)) from where the orientation parameters FWHM, OI and S (equation (3)(4)(5)) could be obtained to quantify the alignment of the BNC nano bers. 26,31 FWHM stands for full width of the half-maximum intensity of the azimuthal pro les from the selected diffraction and it can be obtained directly from the tting. FWHM ranges from 0º to 180º and low values are correlated with higher orientation of the sample (34). OI (also as f c ) is de ned as an orientation index and it is complementary to FWHM. The OI ranges from 0 to 1, with 0 describing a random arrangement and 1 describing a perfect nano brils orientation. S is the Herman's order parameter, obtained with MatLab using equation (4)(5)(6). Similar to OI, a value of S = 0 means the nano bers are randomly oriented, while S = 1 indicates a full nano ber alignment. Thermal conductivity. The thermal conductivity, k, of a material is a measure of its ability to conduct heat.
Thermal conductivity measurements of A-BNC and BNC were performed using the 3ω-method and a novel contactless frequency-domain approach designed by L. A. Perez, 52 et al., and labeled after Anisotropic Thermore ectance Thermometry (ATT). The later methodology is particularly suitable to study thermally anisotropic materials (e.g. aligned BNC), since it delivers the angular distribution of the thermal conductivity perpendicular to the surface of the sample. Measurements were performed with the described system on A-BNC lms dried on silicon wafers at room temperature, to which a gold coating was added.
The 3ω-method is based on electrically heating a thin planar resistor using an AC harmonic current I0 at a frequency ω, and subsequently measuring the resultant voltage drop at the rst (Vω) and third (V3ω) harmonics. By de ning the normalized temperature coe cient of resistance as equation (7) with R0 the resistance of the resistor at the temperature T0, the amplitude of the AC component of the temperature oscillations induced can be determined as equation (8).
By solving the 2D heat equation for the geometry of a linear heat source supported on a semi-in nite medium, the thermal conductivity κ of such medium can be obtained as equation (9).
= − 02 1( Δ / ln 2ω) (9) With P0 the total dissipated power at the resistor and l the length of the resistor. Since the AC current frequency determines the thermal penetration depth according to1/ = √ / 2 , for thick lms supported on a semi-in nite medium (i.e. a substrate) the Δ vsln 2 curve shows low and mid-frequency regimes that primarily correspond to the substrate and the supported lm, respectively. From the slope ( Δ / ln 2 ) of both regions it is straightforward to determine their thermal conductivity.
Statistics were performed on 7 different resistors thermally evaporated on BNC lms yielded to obtain the thermal conductivity.
Transmission electron microscopy (TEM). A-BNC and BNC lms were dried on top of a copper TEM grid.
Images were obtained using a high angular range TEM (JEOL 1210 TEM), operating at 120 kV with an ORIUS 831 SC 600 Gatan camera. Selected Area Electron Diffraction (SAED) was used to obtain the diffraction pattern from the A-BNC and BNC lms, to assess its crystallinity.  (1)).
Statistical analysis. Quantitative data are expressed as means ± standard deviation. Statistical analyzes were performed with Graph Pad Prism 8 software using one-way ANOVA followed by Tukey's multiple comparison test. Statistical signi cance was accepted when P-values were ≤0.05 and summarized as * = P≤0.05, ** = P≤0.01, *** = P≤0.001, **** = P≤0.0001 for the calculated P-values. Figure 1 Methodologies to align BNC bers. A-B Top-down and bottom-up reported methodologies to align bacterial cellulose bers, alongside with the outcomes. C Strategy presented in this work: bene tting from the natural BNC biosynthesis.    Nanoscale density study with SAXS. A,B SAXS scattering signal of BNC and A-BNC. C integration of the spectra as an intensity vs azimuthal degree plot from -90º to 90º. A-BNC showed an intensity peak centered around 0º, whereas BNC gave no peak at all. D Table with   Polarization-resolved second harmonic generation (PSHG) microscopy. Stacks of 18 images (frame/10º light polarization) were analysed with ImageJ using a heat map and Z projections. A Maximum and minimum SHG intensity projections of 150 x 150 µm 2 for BNC and A-BNC, along with the intensity colour map. B Plot of the SHG intensity for each angle polarization, along with their alignment coherency, obtained with the OrientationJ plugin for ImageJ.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. AlignedBNCArticleNM2021FIgSIF.docx