10 1016@j Carbpol 2020 116140 PDF

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 38

Journal Pre-proof

Structural organization of bacterial cellulose: the origin of anisotropy and


layered structures

Tatiana I. Gromovykh, Marina A. Pigaleva, Marat O. Gallyamov, Ilya


P. Ivanenko, Kseniya E. Ozerova, Elena P. Kharitonova, Marjan
Bahman, Nataliya B. Feldman, Sergey V. Lutsenko, Olga I. Kiselyova

PII: S0144-8617(20)30314-3
DOI: https://doi.org/10.1016/j.carbpol.2020.116140
Reference: CARP 116140

To appear in: Carbohydrate Polymers

Received Date: 21 December 2019


Revised Date: 19 February 2020
Accepted Date: 7 March 2020

Please cite this article as: Gromovykh TI, Pigaleva MA, Gallyamov MO, Ivanenko IP, Ozerova
KE, Kharitonova EP, Bahman M, Feldman NB, Lutsenko SV, Kiselyova OI, Structural
organization of bacterial cellulose: the origin of anisotropy and layered structures,
Carbohydrate Polymers (2020), doi: https://doi.org/10.1016/j.carbpol.2020.116140

This is a PDF file of an article that has undergone enhancements after acceptance, such as
the addition of a cover page and metadata, and formatting for readability, but it is not yet the
definitive version of record. This version will undergo additional copyediting, typesetting and
review before it is published in its final form, but we are providing this version to give early
visibility of the article. Please note that, during the production process, errors may be
discovered which could affect the content, and all legal disclaimers that apply to the journal
pertain.

© 2020 Published by Elsevier.


Structural organization of bacterial cellulose: the origin of anisotropy and
layered structures

Tatiana I. Gromovykha, Marina A. Pigalevab,c, Marat O. Gallyamovb,c, Ilya P. Ivanenkob,

Kseniya E. Ozerovab, Elena P. Kharitonovab, Marjan Bahmana, Nataliya B. Feldmana, Sergey V.

Lutsenkoa, and Olga I. Kiselyovab


a
Department of Biotechnology, I.M. Sechenov First Moscow State Medical University of the

Ministry of Health of the Russian Federation (Sechenov University), Bolshaya Pirogovskaya st.,

of
Moscow, 119991, Russian Federation
b
Faculty of Physics, M.V. Lomonosov Moscow State University, Leninskie gory 1–2,

ro
Moscow,119991, Russian Federation
c
-p
A.N. Nesmeyanov Institute of Organoelement Compounds, Russian Academy of Sciences,

Vavilova st., 28, Moscow, 119991, Russian Federation


re
lP

*
Author for correspondence, e-mail: ok@polly.phys.msu.ru
na

Graphical Abstract
ur
Jo

1
Highlights

 Plywood-type organization and layers of “tubes” were found in bacterial cellulose


 Layered structures were found in bacterial cellulose pellets
 The formation of layers in bacterial cellulose is mediated by the fibrils laying rate

Abstract

In this paper, we perform a systematic analysis of the structural organization of bacterial


cellulose (BC). We report four types of organization of the BC mass, produced by
Gluconacetobacter hansenii that occur depending on cultivation conditions. Two of those,
particularly, plywood type one and layers of micro-sized tubes were observed and described for

of
the first time. In spherical BC particles (pellets), we found the layered structure that had
previously been reported for planar geometry only. We suggest a model explaining why layers

ro
form in BC films and attempt to reveal the impact of different factors on the BC microscale
morphology. We assume that the main factor that has direct impact on the type of structure
formed is the rate of BC mass accumulation. -p
re
Abbreviations: BC – bacterial cellulose, BET – Brunauer-Emmett-Teller, C-source – source of
lP

carbon, DSC – differential scanning calorimetry, HS – Hestrin-Schramm medium, TG –


thermogravimetry, SEM – scanning electron microscopy.
na

Keywords: bacterial cellulose, Gluconacetobacter hansenii, pellets, scanning electron


ur

microscopy, thermoanalysis, mechanical testing


Jo

1. Introduction

Bacterial cellulose (BC) is a unique and, to say more, fascinating natural material that has
numerous properties that make it extremely valuable for many different applications, both
industrial and biomedical ones. The main feature of BC is its chemical purity; unlike the
cellulose extracted from plants, BC is not associated with residual hemicelluloses, pectin, lignin,
etc. In addition, it is a fibrous material with jelly-like structure, high porosity, permeability and
2
water sorption capacity. Exhaustive reviews on BC properties and applications can be found
elsewhere (Huang et al., 2014; Keshk, 2014; Moniri et al., 2017; Picheth et al., 2017). BC is
produced by numerous microorganisms belonging to various genera: Gluconacetobacter
(formerly Acetobacter), Agrobacterium, Aerobacter, Achromobacter, Alcaligenes, Azotobacter,
Rhizobium, Sarcina, Salmonella, Escherichia, etc. In each case the synthesis follows different
metabolic routes before BC is secreted outside of cells (Hassan, Abdelhady, El-Salam, &
Abdullah, 2015; Hong & Qiu, 2008; Huang et al., 2014; Keshk, 2014; Lee, Buldum, Mantalaris,
& Bismarck, 2014; Park, Jung, & Park, 2003; Wang et al., 2018).

Several strategies have been worked out, allowing for production of BC mass in planar,
cylindrical or spherical configurations, both in static and agitated conditions. Planar films
(pellicles) are grown by means of cultivation of BC-producing cells in static conditions and form

of
on the flat liquid/air interface. BC tubes (cylindrical configuration) are grown in polymer tubes
or on oxygen-permeable silicon tubes placed into the liquid culture medium, the gas being

ro
pumped through the tubes (Bodin et al. 2007; Klemm, Schumann, Udhardt, & Marsch, 2001).
Spherical pellets are obtained under agitation conditions and it is well established that the
-p
geometry and the inner structure of pellets depends on the BC-producing strain used (Bi et al.
2014; Singhsa, Narain, & Manuspiya, 2018) and the size of particles is defined by the speed of
re
rotation (Gromovykh, Bachman, Petrukhin, Dutka, & Butenko, 2018; Zywicka et al., 2015). In
the present work, we compare the morphology of BC pellets synthesized by the same strain
lP

Gluconeacetobacter hansenii but obtained at different rates of agitation (rotation). We also


compare the microscale morphology of BC fibrils in pellets and planar films.

The microscale morphology of BC has been revealed by means of scanning electron


na

microscopy (SEM) and atomic force microscopy by many groups. It was shown that BC is a
natural fibrous material, consisting of entangled fibers of pure cellulose, 30–100 nm in diameter.
Numerous authors report that BC films, grown in planar geometry under static conditions, have
ur

an anisotropic layered structure, as revealed by SEM and polarization microscopy of BC films


cross-sections (An et al., 2017; Cai & Kim, 2010; Chen, Cho, & Jin, 2010; Park, Chang, Jeong,
Jo

& Hyun, 2013; Pircher et al., 2014; Saska et al., 2012; Yamanaka et al., 1989). Layers are
oriented parallel to the liquid/air interface, as seen in images of cross-sections. The distance
between layers differs from one publication to another: 50 µm (Chen, Cho, & Jin, 2010), 3-10
µm (Pircher et al., 2014), 6 µm (Katepetcha & Rujiravanit, 2011), 7 µm (An et al., 2017), 6 µm
(Park, Chang, Jeong, & Hyun, 2013), 30 µm (Cai & Kim, 2010). It was shown that when a BC
film is grown under static conditions on liquid medium/air interface, the newer layers are
constructed on the film/air interface (Borzani & de Souza, 1995; Masaoka, Ohe, & Sakota,
3
1993), i.e. the earlier produced layers are located deeper. Park, Chang, Jeong, & Hyun (2013)
reported that the number of layers depends on cultivation time and estimated that approximately
20 layers form per day.

The cylindrical geometry case was studied by the group of P. Gatenholm, who grew BC
in form of cylinders, using silicon tubes as support, oxygen being supplied through these tubes
(Bäckdahl, Risberg, & Gatenholm, 2011; Bodin et al., 2007). SEM images of cylinders’ cross-
sections made perpendicular to their axes, revealed layered regions close to the lumen and an
outer region with a less dense BC network. The inner region had dense thin layers of BC that
appeared to be circular in shape, i.e. aligned parallel to the surface of the tube. This alignment
seems logical for cylindrical geometry.

of
To our knowledge, for spherical configuration, i.e. pellets, obtained in agitated medium
the layered structure has not yet been reported. In this work, we study the layered structure of BC

ro
produced by G. hansenii, in planar and spherical configuration, compare it to the literature data
and attempt to understand its origin and factors that influence its formation. We suggest a model
explaining how layers are formed in planar BC films. According to it, the formation of layers in
-p
BC occurs if the rate of its synthesis exceeds a certain threshold level; otherwise, the formed gel
is isotropic. The rate of BC accumulation is, in turn, determined by several parameters of
re
cultivation conditions and producer strain used. Here, we checked such factors as oxygen
availability, sources of carbon and nitrogen.
lP

2. Materials and methods


na

2.1 Cultivation of BC. All BC mass, investigated in our work was produced by cultivation of
Gluconacetobacter hansenii GH-1/2008 (VKPM V-10547, 1) in modified Hestrin-Schramm
ur

medium (HS) (Hestrin & Schramm, 1954). In order to obtain the seeding material, G. hansenii
was cultivated in the described medium at 30 °C using the rotational shaker for 3 days.
Jo

Planar bacterial cellulose films were obtained by means of cultivation of the BC-
producing cells in the stationary regime at 26°±2°С in 20 × 35 × 5.5 cm cuvettes for 8 days as
previously described in (Pigaleva et al., 2019), with the following composition (g/L): glucose –
20.0, peptone – 5.0, yeast extract – 5.0, Na2HPO4 – 2.7, KH2PO4 – 2.0, citric acid monohydrate –
1.15. The volume of the medium solution was 700 mL. For the second type of films, glucose was
replaced with the same amount of sorbitol; in another experiment yeast extract was also replaced

4
with 3.0 g/L of (NH4)2SO4. In these two experiments, the cultivation time was extended to 10
days due to the slower accumulation of BC.

Sucrose (pure, pharma grade), SDS (pure, pharma grade) were purchased from PanReac
AppliChem, Na2HPO4, KH2PO4, (NH4)2SO4 were supplied by Sigma-Aldrich. Powdered meat
peptone with amino nitrogen content of more than 3.0% and yeast extract were purchased from
Dia-M (Russia).

Pellets were obtained in round flasks in agitated media, using rotary shaker-incubator
IKA KS 4000i for 10 days at 22±2 ºС. Hollow pellets (P1) and filled pellets (P2) were obtained
in the medium of the following composition: glucose – 30.0, yeast extract – 5.0, Na2HPO4 – 2.7,
KH2PO4 – 2.0, citric acid monohydrate – 1.15. For obtaining microporous pellets (P3), glucose

of
was replaced with the same amount of sorbitol.

Hollow pellets (P1) were obtained at the constant rate of agitation of 120 rpm. Filled

ro
pellets (P2) and microporous pellets (P3) were obtained at the variable rate of agitation: 120 rpm
for 96 hours, then at 150 rpm for 72 hours, and at 180 rpm for the final 24 hours.
-p
When the cultivation was terminated, the obtained films and pellets were separated from
the cultivation medium by filtration and washed. Planar films were washed in a standard way
re
using sodium dodecyl sulfate (SDS) solution as previously described by Pigaleva et al., (2019).
Pellets were immersed into 5% solutions RIPA buffer. RIPA buffer (Thermo Scientific) solution
lP

had the following composition: 25mM Tris-HCL, pH 7–8, 150 mM NaCl, 0.5% sodium
deoxycholate, 1% Triton Х-100, 0.1% SDS. The samples were kept in the solutions for 3 days,
using rotational shaker at 150 rpm. Washing solutions were changed every 24 hours. Then the
na

samples were washed with distilled water for 3 days, exchanging the water every 24 hours.

2.2 Freeze-drying. BC films and all the three types of pellets were laid on Petri dishes and frozen
ur

first at -18 °C in deep-freeze chamber, then in liquid nitrogen and dried for 48 h at -50 °C and
0.07 mbar using an Alpha 1-2 LD unit freeze-dryer (Christ, Germany). Hollow pellets (P1) were
split by a sharp stroke with the pointed end of forceps prior to drying.
Jo

In order to investigate the freshly cleaved surfaces, all other samples were cleaved again
shortly before the SEM imaging. This approach eliminated the influence of surface effects on the
sample structure. Filled pellets (P2) were cleaved along cracks, freshly cleaved surfaces were
imaged. Microporous pellets (P3) were cleaved using fine-tipped tweezers. Pressing with
tweezers gave rise to a crack that propagated without further contact with the tool; a freshly

5
cleaved untouched surface was formed. Films were instantly frozen in liquid nitrogen once again
and cleaved so that to expose the cross-section.

2.3 Preparation of BC paste. Washed BC films grown on the glucose-containing medium were
placed into the vessel for milling; distilled water was added at volume ratio of 1:1. After that, BC
was milled using a laboratory blender MB 800 (Kinematica, Switzerland) until the homogeneous
mass was obtained. The obtained mass was left until the water was separated, then it was frozen
in a freezer at -30 °C for 2.5 h and freeze-dried first at -28 °C for 6 h. The final drying was done
at 40 °C for 3.5 h. The total time of drying was 12 h.

2.4 Scanning electron microscopy (SEM). SEM images of planar films were obtained using a
Hitachi SU8000 field-emission scanning electron microscope. The samples were fixed on an

of
aluminum cross section holder by conductive plasticine-like adhesive. Metal coating with a thin
film (10 nm) of platinum/palladium alloy (80/20) was performed using magnetron-sputtering

ro
method as described earlier (Kashin & Ananikov, 2011). Images were acquired in secondary
electron mode at 10 kV accelerating voltage and at a working distance of 7-15 mm. Morphology
-p
of the samples was studied taking into account the possible influence of metal coating on the
surface (Kashin & Ananikov, 2011).
SEM images of pellets were obtained by means of a Quanta 3D FEG scanning electron
re
microscope (FEI, Holland) equipped with a 5-axis motorized stage. Prior to scanning, the
samples were coated with an approximately 16 nm thick conductive gold film. The samples were
lP

placed in the vacuum chamber of a JFC-1100 fine coat ion sputter (JEOL, Japan) for 10 min. The
current was 8 mA. Using the target with the diameter of 57 mm provides the effective sputtering
with the minimal sample heating. The accelerating voltage for P1 was 15 kV or 10 kV (see
na

figure captions), 15 kV for P2 and 10 kV for P3. We specified the accelerating voltage in figure
captions only if it was other than 10 kV.
ur

2.5 Specific surface area determination. The nitrogen adsorption and desorption isotherms were
acquired using Nova 1200e surface area analyzer (Quantachrome, USA) at 77 K in the
laboratory of stereochemistry of sorption processes at A.N. Nesmeyanov Institute of
Jo

Organoelement Compounds, Russian Academy of Sciences. Prior to measurements, the samples


were activated at 20 °С at the residual pressure of 10-3 mmHg for 8 h. The specific surface area
was determined by Brunauer-Emmett-Teller (BET) method.
2.6 Thermal measurements. Thermal degradation of BC pellets was investigated by means of
thermogravimetry (TG) with NETZSCH STA 449C equipment in the air. BC pellet melting was
studied by means of differential scanning calorimetry (DSC) using NETZSCH DSC 204F1
instrument in the nitrogen atmosphere. Aluminum crucibles were used for the measurements.
6
The weight of the measured samples was 1-5 mg. The heating rate was 5 K/min for all
experiments. Before measurement, swelled samples P1 and P2 were kept in alcohol for 24 hours
in order to remove residues of growth medium and then dried at 60 С for 12 hours. Sample P3
was studied as it was, prepared as described above.
2.7 Mechanical testing was carried out using an Instron 1122 universal testing instrument
(Instron, USA) in ambient conditions. The BC films were cut into ribbons 10 × 30 mm2 and
freeze-dried prior to testing. The thickness was measured with a micrometer type caliper. For
each type of film, three samples were tested. The elongation rate was 10 mm/min, the distance
between the clips was 10 mm.

3. Results

of
3.1 Planar configuration. We investigated the inner architecture of films grown in static

ro
conditions by means of scanning electron microscopy (SEM). The cross-sections images
revealed that the films were anisotropic and consisted of layers parallel to the film plane. Figures
1a and 1b represent a SEM image of a cross-section of such film perpendicular to the film plane.
-p
Evidently, in Figure 1a there are two areas with different periodicity. The structure of the
observed layers is also different. The left side of the image represents layers (higher
re
magnification image is in Figure 1c) similar to those earlier reported in the literature. The
average period of 7.0 µm (Sd=1.5 µm) is close to the value given by Park, Chang, Jeong, &
lP

Hyun (2013). Meanwhile, the right side of the image has a different organization (Figures 1b, d).
The period observed in Figure 1b and the right part of Figure 1a is 25 µm (Sd=6 µm). Figures 1c
and 1d depict images acquired at the same magnification. The difference between images c and d
na

is not just quantitative, but qualitative as well. In Figure 1c, dense sheets of cellulose are
separated by fibrous material with very low density. In the right part of the image at Figure 1a
and Figure 1b, the BC material within the interlayer space has cellular rather than fibrous
ur

organization. It is more evident from the higher magnification image in Figure 1d. The freezing
conditions were obviously the same, so the difference in architecture needs some other
Jo

explanation. Similar pictures were observed for other films we studied as well (data not shown),
the period being 35 µm, out of which 20 µm had cellular organization and 15 µm fibrous, fibers
being parallel to the film surface (data not shown).

7
of
ro
-p
re
Fig. 1 SEM images of the cross-section of a BC film grown in static conditions on glucose-
containing medium. The left side of the image corresponds to the top of the film (air-liquid
lP

interface) that is parallel to the vertical left edge of the photo. a: overall view, b: bottom portion
of the film, c: higher magnification of image a (left area), d: higher magnification of image a
(right area).
na

In Figure 1b one can see two apparent types of the interlayer material organization. It can
be fibrous or cellular. At the bottom of the image, it is clearly seen that these two types of
ur

structure appear in turn, i.e. interchange (from left to right). Similar structures with the overall
period of 40 µm were previously observed by Kim, Cai, & Chen (2010). Nevertheless, the
authors did not give any explanation to this observation. We will come back to the possible
Jo

explanation in the next section.

8
of
ro
-p
re
lP
na
ur
Jo

Fig 2 SEM images of the cross-section of a BC film grown in static conditions on sorbitol-
containing medium. The left side of the images (a, b) corresponds to the top of the film (air-
liquid interface) that is parallel to the vertical left edge of the photo. Yeast extract (a, b) and
ammonium sulfate (c) were used as nitrogen source. Image b was acquired from the same sample
as a, at a higher magnification. In image a, darker bands correspond to areas where the fibers are

9
oriented perpendicular to the image plane, lighter bands – to those where fibrils mainly lie in the
image plane

In order to get a better understanding of the organization of cellulose fibers within this
type of structures, we prepared cross-sections of BC films grown in the medium containing a
different source of carbon, particularly sorbitol. In Figure 2, we also saw a periodic structure,
though apparently different from the one observed in the films grown in glucose-containing
medium; the cross-section contained several layers with apparently different organization. The
overall period of the structure was roughly 100 µm (Figure 2a). The top surface of the film is
parallel to the left edge of the image. Higher magnification images (Figure 2b) reveal, that the
orientation of BC fibrils is not random like it is in isotropic structures; there is a preferential
direction. This direction gradually changes from layer to layer. In band 2 (Figure 2b), it is

of
parallel to the cross-section plane. In band 4 the direction is perpendicular to the section plane.
Bands 1, 3 and 5 represent intermediate directions. A similar twisted plywood-type structure

ro
composed of crystalline chitin fibers was earlier described in arthropod exoskeleton, where the
direction of fibers changes from layer to layer (Giraud-Guille, 1996; Giraud-Guille, Chanzy, &
-p
Vuong, 1990; Nikolov et al., 2010). Helicoidal liquid-crystal-like organization occurs in several
other biopolymer systems both in nature and in reconstituted suspensions, including cellulose in
re
plant cell walls (for review, see Giraud-Guille, 1996). As far as we know, it has not yet been
described in BC, nor in any other polymers of prokaryotic origin. The formation of this type of
lP

architecture by prokaryotic organisms is different from that occurring in plant cell walls, since
the former do not contain microtubules regulating the direction of the cellulose synthase
complexes movement. Therefore, the arrangement of BC fibrils is driven rather by self-
na

organization than by some inner mechanism of the cells.

Figure 2c represents a cross-section of a BC film grown in the same medium as the film
at Figure 2a and 2b, but with the yeast extract replaced with ammonium sulfate. The change of
ur

nitrogen source had a dramatic effect both on the growth rate and on the film architecture. The
rate of BC depositions reduced and the layered structure with preferred orientation of fibrils
Jo

almost disappeared, being replaced by the isotropic network.

Based on this data, we managed to understand the pattern we see in Figure 1b. The
cellulose fibers in this image form “microtunnels”, rather than layers. Each “microtunnel” is an
almost hollow cylinder; the walls of it are 100-200 nm thick and are composed of interwoven
and stuck cellulose fibers, oriented mainly parallel to the cylinder axis. The diameter of the
“microtunnels” is 5-7 µm. Here, we intentionally avoid the term “microtubules” so that to
exclude any possible confusion or any analogy with homonymous elements of the cytoskeleton.
10
The apparently “fibrous” zones occur when the “microtunnels” are oriented parallel to the cross-
section plane. Apparently, “cellular” zones correspond to areas where the tunnels are oriented
perpendicular to this plane (Figure 3a). The SEM image in Figure 3a was acquired from the same
sample as the Figure 1b.

of
ro
-p
re
lP

Fig 3 A: SEM image and the schematic 3D model explaining the organization of fibrils within
na

the BC films. Image a represents the cross-section of the BC film perpendicular to the film
surface, the same way as in Figure 1. 3D model is constructed of layers of hollow cylinders
(imitating tunnels); in each layer, the axes of cylinders are turned 15 degrees with respect to the
ur

previous one. Image b shows the cross-section made in the plane shown in image a, i.e. parallel
to the film surface. Image c is taken from the same sample as image b with higher magnification;
Jo

its position is marked by the rectangle in image b. The arrows in c show the changing direction
of fibers, presumably forming the “microtunnels”.

For a clearer illustration of the structure, we drew a schematic 3D model constructed of ideal
tubes (Figure 3a). If we now go back to Figure 1b, we can understand what we see at the right
edge of the film (bottom portion). The observed sticks popping perpendicular to the film surface
and looking like straws broken parallel to their lateral axis are in fact fragments of microtunnels’
walls. These disordered fragments occur due to the cleavage of the sample.
11
In order to check this hypothesis, we cleaved the sample in the plane parallel to the film surface
and, respectively, perpendicular to the plane, represented in Figure 3a. In other words, we
delaminated the sample. The resulting SEM image shown in Figure 3b contains areas,
resembling helices, i.e. those where the orientation of fibers changes gradually, in a helicoid
way. Why do we see this instead of parallel rows of half-pipes? The cleavage (delamination) of
sample does not occur ideally parallel to the plane of the plywood type structure, described
above. If it were the case, we would have seen all microtunnels aligned in the same direction,
similar to zone 2 in Figure 2b. However, if the cross-section plane intersects the structure at
some small angle, we see not just one layer, but one layer after another in a stepwise way (Figure
3c). In each layer, the direction of microtunnels changes and that creates the illusion of a helix.
This is well seen from a higher-magnification image (Figure 3c). The model of arcs’ and helices’

of
origin was previously proposed by Bouligand in 1972 and later by Besseau and Bouligand
(1998) for fish scutes, cuticles of invertebrate etc. We advise the interested readers to see these

ro
works for detailed schematics, an exhaustive list of twisted fibrous materials can also be found
there.
-p
3.2 Paste. In the control experiment, we imaged the so-called cellulose paste (see
Materials and methods section for preparation details). The structure of films was destroyed or at
re
least severely distorted during milling, but freezing and drying conditions were kept virtually the
same. It is evident from Figure 4, that no parallel layers were formed during freezing.
lP

Nevertheless, we still see sheets of densely packed cellulose fibers that are, unlike the previous
case, oriented randomly, plied and torn in different places. Apparently, the structure resembles a
crumpled sheet of paper. This series of experiments proves that the sheets were formed prior to
na

freezing, and obviously even prior to milling.


ur
Jo

Fig 4 SEM image of the BC paste (milled BC) after freeze-drying (cross-section)
12
3.3 Spherical configuration. What happens to the structures described above when the geometry
of the BC mass is spherical? Inspired by comparison of planar and cylindrical geometry, we
were intrigued to see how the layers would be distributed in the spherical case. Following the
general logics of the symmetry in nature, we expected to find layers organized as concentric
spheres. For that, we prepared the 3D phase of BC, particularly, three types of pellets: hollow
ones (P1) (spheres), filled ones (P2) (full-spheres) and microporous ones (P3), using different
regimes of BC cultivation in agitated media. Photos of pellets in wet state as well as after freeze-
drying and cleavage can be found in Supplementary material section.

Pellets P1 were hollow spheres, consisted of a dense shell and inner void, filled with medium
solution, as previously described by Gromovykh, Bachman, Petrukhin, Dutka, & Butenko
(2018). Therefore, we expected to observe layered structure consisting of annular rings in the

of
cross-section of a pellet, similar to those reported for the cross-section of BC tubes by Paul
Gatenholm’s group (Bäckdahl, Risberg, & Gatenholm, 2011; Bodin et al. 2007). SEM images of

ro
P1 inner and outer surfaces and cross-sections are represented in Figure 5. Despite the washing
procedure, intended for bacteria removal, bacteria were seen on both inner and outer surfaces of
the pellet (Figure 5a, 5b).
-p
re
lP
na
ur
Jo

13
of
ro
-p
re
lP

Fig 5 SEM images of a hollow pellet (P1). a: inner surface and edge; b: the outer surface of the
na

pellet; c and d: edge of the cleaved shell. c: natural state, densely packed fibrils parallel to the
surface, d: individual ultrathin layers can be seen due to the damage during cracking in liquid
nitrogen. Splitting of layers during cracking helped to understand the structure. In image d, we
ur

see the inner surface and the edge of the shell. The accelerating voltage was 15 kV for images a,
b, d and 10 kV for c
Jo

In Figure 5a, we see bacteria not only on the surface, but between the layers as well. It is not
completely correct to talk about the density of bacteria on the inner and the outer surface of the
pellet shell, since the thickness of the shell (4-5 µm) is comparable to the length of bacteria (1-2
µm). The pellicle forming the pellet (shell) was very thin, about 4-5 µm thick, that is even less
than the typical value of a band seen in planar films and tubular constructions. Thus, in this type
of pellets, we did not find any evidence of layered structures similar to those in Figure1. The
cellulose fibers in the shell were predominantly aligned parallel to the pellet surface. The shell
14
consisted of numerous BC sheets stuck to each other; each of them had the thickness of several
BC fibrils (Figure 5a, d). In some locations of the cross-section, the ultrathin layers were densely
packed (Figure 5c), the overall organization resembling the pages in a closed book. In others, the
shell near the edge was presumably split into ultrathin sheets during the sample preparation in
liquid nitrogen (Figure 5d). We admit that the sheets in the wet pellet could have initially been
separated by fibrous zones, similar to those existing in the planar films (Figure 1c), and were
then compressed and stuck together during freezing due to the increase of the volume and
therefore the radius of the pellet.

For filled pellets (P2), we obtained SEM images of the equatorial cross-section of the
pellets. The pellets are core-and-shell type, the core being less dense than the shell. In the low-
resolution image, we saw concentric lines, following the sample surface. Nevertheless, at a

of
closer look (Figure 6a) it became clear, that these variations of contrast were not related to the
inner architecture of the pellet. Straight lines seen in the image turned out to be parallel layers

ro
similar to those previously reported and obtained for planar geometry. To our great surprise, the
organization had nothing similar to annular shape. The overall picture represented a mosaic of
-p
areas (we further call them “domains”), within each of which, the lines were straight and parallel
(Figure 6b). The bulk material formally resembled the polycrystalline structure, consisting of
re
individual crystalline domains. It seems that the material is composed of independent fragments
of planar BC films stuck together. Each observed “domain” had a different orientation of layers
lP

and different apparent distance between the layers. Of course, this analogy is only formal and we
use it to describe the inner architecture of pellets. The observed cracks originating from cross-
section preparation did not correlate with apparent “domain” boundaries, and there were no
na

traces of former separation between them. Figure 6b shows the junction of three “domains”. The
angle between the directions of lines in two of them is 71 deg. In the third “domain”, located in
the left upper corner of the image, no lines are seen. We assume that this is because the cross-
ur

section plane was parallel to the plane of layers in this domain.


Jo

15
of
ro
-p
re
lP
na
ur
Jo

Fig 6 SEM images of the pellet P2 equatorial cross section. a: overall image at low
magnification. b: junction of three domains with different orientation of layers. c, d: images of
fragments of two domains, containing layers. The structure is the same as in the planar geometry
16
case. Average layer separation is 8 µm (c). e: Fragment of a domain, containing no lines in SEM
image. Bacterial cells are very well seen. f: Schematic drawing, explaining why the observed
period may be different from the real layer-to-layer distance. The accelerating voltage was 15 kV

Within a given domain, the architecture of BC does not differ much from that for BC
films grown in static conditions (planar geometry) (Figures 6c, d). The distance between layers
(the period of the structure) is roughly constant for each domain, but differs from one domain to
the other. We also observed a domain where no layers were seen (Figure 6e). Higher
magnification images show that bacterial cells were present in all parts of the pellet and were
distributed evenly. We analyzed the distribution of interlayer distances in several domains and
obtained the following values: 8.0 µm, 8.1 µm, 8.8 µm, 10.2 µm, 11 µm, 16 µm, 18 µm, 20 µm,
27 µm. We explain the variation of the structure period from one domain to another, by the fact

of
that the intersection of the cross-section plane with the plane of layers may occur at different
angles, since the orientation of domains is random. The minimal period is seen when the

ro
intersection occurs at 90 º. The observed period D=d/cosα, where d is the real distance between
layers planes and α is the angle between the cross-section plane and the normal to the layers
-p
(Figure 6f). If α=0, then D=d and we see the correct interlayer distance, if α=π/2, then D→ ∞ and
we do not see any layers (like in Figure 6e). If we admit, that the minimal observed interlayer
re
distance of 8 µm corresponds to the orthogonal section (α=0), then for D=10 µm we obtain α=37
º, for D=15 µm α=58 º, D=18 µm, α = 62 º, D=20 µm, α=66 º, D=27 µm, α=73 º. Higher values
lP

of α are hard to achieve, and in agreement with this fact, we did not observe larger values of D.

As a control experiment, in order to understand the impact of the geometry and freeze-
drying conditions, we studied the third type of pellets (P3) that turned out to contain micropores,
na

so we denoted them as microporous pellets. The overall view of such a pellet’s cross-section is
represented in Figure 7a. Though the pellet is not hollow, it contains cracks and internal voids
(micropores) of 30-500 µm in size. Inside cracks, more micropores could be seen. The shape of a
ur

micropore depicted in Figure 7b is ellipsoidal, so the appearance of the pores is presumably


explained by the existence of air bubbles, inside the cellulose mass, rather than by cracks
Jo

occurring during the sample preparation in liquid nitrogen. Bacteria were abundant on the cross-
section surface and on the surface of cracks and micropores occurring in the volume of the
pellets. In the overall image, one can notice annular circles, than might seem analogous to those
previously reported in BC tubes. At a closer look, i.e. with higher magnification, it becomes
evident that these rings do not correlate to the fine structural organization of BC fibrils (Figure
7c), similar to the case of filled pellets (P2). High magnification image (Figure 7d) shows the
isotropic unordered network of fibrils, similar to that previously reported by Bi et al. (2014),
17
formed in pellets produced by Komagataeibacter nataicola Y19 cultivated at 130 rpm. The
observed concentric lines are produced by the relief of the pellet cross-section surface, and are
presumably due to some minor shrinking during freeze-drying process. The existence of
micropores allowed for looking inside the volume of the pellet further from the cross-section and
seeing the picture from different angles. In all visible planes, the picture was uniform and no
preferential direction of fibrils orientation could be detected. Generally, we can say that the
density of the network in the central part of the pellet was lower than that at the periphery, but no
visible border could be detected.

of
ro
-p
re
lP
na
ur
Jo

Fig 7 SEM images of microporous pellets (P3). a: an overall view. b: a micropore occurring in
the volume of the pellet. Image c is taken from the bottom part of image a, its position is marked
by the rectangle. d: a higher magnification fragment of image c

18
3.4 Thermal analysis. Thermogravimetry (TG) curves for filled (P2) and hollow (P1) pellets,
and BC film (planar configuration) are shown in Figure 8a. TG turned out to be insensitive to the
microstructure of the samples and did not reveal any difference between the pellets. We assume
that the film demonstrated the highest decomposition temperature because it had the least
number of defects. The behavior of all the samples was typical for BC, the decomposition of
which normally demonstrates two steps of weight loss at 300 C and 400 С (Saska et al., 2012;
Vasconcelos et al., 2017). The impact of the pellets’ microstructure on their thermal properties
was revealed by means of DSC in nitrogen atmosphere. The observed effect is part of cellulose
pyrolysis process accompanied by strong endothermic DSC effect. It is well known, that
pyrolysis of crystalline cellulose occurs in oxygen-free atmosphere and was previously
demonstrated in literature (Yang, Yan, Chen, Lee, & Zheng, 2007). In Figure 8b, we show DSC

of
curves for pellets and the one for the BC film (planar configuration) measured under the same
conditions for comparison. The main parameters of endothermic peaks are summarized in Table

ro
1.

-p
re
lP
na
ur
Jo

19
Fig 8 Thermogravimetry curves for hollow pellet (P1), filled pellet (P2) and BC film (Film)
measured in the air (a). DSC curves for hollow pellet (P1), filled pellet (P2), microporous pellet
(P3) and BC film (Film) measured in the nitrogen atmosphere (b)

The pyrolysis peak temperature of pellets is close to that of planar BC, but the peaks for
pellets are wider, which fact can originate from their microstructure, particularly the presence of
fragments different in structure and inherent flaws. Obviously, fragments with more defects
should start pyrolysis at lower temperatures. In line with this fact, the pyrolysis peak of the film
is the narrowest and occurs at the highest temperature of the four samples, because the planar BC
film has the most ordered structure and the least number of defects and pores (Figure 8b). The
peak temperature of P2 is close to that of the film (Table 1), and again, both structures have
layers apparently very similarly organized. However, contrary to the film, P2 has the broadest

of
peak of the four samples and the lowest onset temperature (Table 1). This can be understood, if
we remember that P2 consists of a structured shell, containing domains of layers and a less

ro
organized core zone. Thus, the corresponding peak is in fact the sum of the two closely located
ones: the lower one corresponding to an isotropic network in the core and the higher one
-p
responsible for the structured shell. The latter fact may as well have an impact on the pyrolysis
enthalpy of the filled pellet (P2) (Table 1) that is almost 2 times higher than that of the P1 and P3
re
pellets and almost 3 times than that of the planar film.

As for P1 and P3 pellets, their pyrolysis peak parameters (temperature and enthalpy) are
lP

very close to each other (Table 1); the corresponding peaks are similar and occur at lower
temperatures than those of the other two samples. This is in line with the observation that both
pellets do not have layered structure. The anomaly of the DSC curve for P3 near 110 С is most
na

probably related to the melting and decomposition of sorbitol residues.

3.5 Mechanical properties. Mechanical properties of BC grown in different conditions is an


ur

important information if the material is going to be used for biomedical and technical
applications. In order to evaluate the connection between the structure and the properties, we
Jo

compared the mechanical properties of the films studied in this paper. Interestingly, the value of
Young’s modulus depended not only on the source of carbon (glucose or sorbitol), but on the
nitrogen source (yeast extract or ammonium sulfate) as well. For films grown on sorbitol-
containing media it was 1.5 ± 0.5 GPa when yeast extract was used and reduced to 0.06 ± 0.02
GPa when it was replaced with ammonium sulfate. For films grown on glucose-containing
medium it was previously measured to be 6.5 ± 1.5 GPa under the same conditions (Pigaleva et
al., 2019). The elongation at break for them was, on the contrary, the smallest, 1.5 ± 0.5%, while

20
for sorbitol-grown films it was 4 ± 1% and 6 ± 1% for yeast extract and ammonium sulfate,
respectively.

4. Discussion

Our intention was to compare the structures forming in planar, cylindrical and spherical
geometry cases. Interestingly, in the cross-sections of BC tubes, the border between the inner
layered zone and the apparently non-layered one (where layers can only be revealed after
etching) is very clear and has an ideal circular shape (Bodin et al., 2007). At the same time, the
concentration of oxygen reduces gradually, so one would not expect the changes in the structure
to be so abrupt. The question remains: why do we see the boundary then? We suggest that there
should exist a threshold level of oxygen, below which the regime of BC synthesis changes

of
dramatically. According to the equation of diffusion in polymers, the distance where this
threshold value is reached, depends on the boundary condition, i.e. the concentration of oxygen

ro
in the lumen: the higher the inner concentration is, the further from the lumen the threshold level
is reached.

-p
For planar geometry, Pircher et al. (2014) suggested the following explanation, why we
see so distinct layers in the cross-section. The authors claim that since all BC producing bacteria
re
strains are aerobic microorganisms, the release of cellulosic elementary fibrils occurs only in
close proximity to the liquid/air interface. The authors believe that gravity pulls the thickening
cellulose fleece deeper into the culture medium in micro steps. The density and number of
lP

entanglements parallel to the liquid/air interface are higher than in growth direction. When BC
samples are prepared for SEM or polarization microscopy imaging, they are frozen. During
na

freezing, the expanding ice pushes the cellulose network apart mainly along its weaker direction,
forming alternating layers of compacted and less compacted cellulose ribbons. The authors
believe that the period of the structure (the distance between compacted cellulose layers)
ur

depends on the freezing conditions.

Though the authors from Paul Gatenholm’s group admit that they cannot explain, why
Jo

the observed layers are formed, they suggest a model, describing, how the tubes are formed and
why the density changes, decreasing in the outside area of the cylinder (Bäckdahl, Risberg, &
Gatenholm, 2011). The authors suggest that it is the availability of oxygen that determines the
organization of the BC. Unlike the planar films case, in the cylindrical geometry, each new layer
is formed further from the oxygen source than the previous one. Nevertheless, the distance
between apparent layers in the inner area is constant, roughly 15-20 µm and does not change
with the increase of the distance from the lumen, as one can see from the images published by

21
the group (Bäckdahl, Risberg, & Gatenholm, 2011). Further away from the lumen, where the
availability of oxygen decreased, the authors observed a looser BC network, the layered structure
was not so well pronounced, the interlayer space was filled with fiber arrangements, meanwhile,
after the surface was etched with the ion beam, the layers were clearly seen. The role of the
oxygen availability is confirmed by the fact, that the higher oxygen concentration was in the gas,
pumped through the tube, the larger the inner zone, containing the well-distinguished layers,
was. Unfortunately, the authors did not provide any analysis of the layer-to-layer distance, and
how it changes depending on the growth conditions.

If we look at the process of formation of BC tubes, we see that the above-mentioned


model suggested by Pircher et al. (2014) cannot describe it. The new ribbons are, obviously,
produced further from the oxygen supply, since there is no other space for material buildup, and

of
there is no stepwise deepening due to gravity. Still, we clearly see the same type of layered
structure, the only difference being the annular shape that seems natural in cylindrical geometry.

ro
The freezing itself can only “increase the contrast”, i.e. make the periodic variations of the
network density stronger, and therefore, better visible, as suggested by Bodin et al. (2007), but
-p
cannot give rise to these variations. There were speculations in the literature that the very
existence of the layers is due to freezing, i.e. to directional growth of ice crystals induced by
re
temperature gradients. However, though this mechanism is the basis of the freeze-casting
technique widely used for production of anisotropic layered materials (see review by Wegst, Bai,
lP

Saiz, Tomsia, & Ritchie, 2015), including BC-containing composites (Li et al., 2017), the
annular geometry demonstrated by Bodin et al. (2007) and variations of layers orientation in
pellets described in this work prove that it is not the origin of layers formation in BC. For pellets,
na

obtained at lower agitation rate, the layers were not formed at all. Our experiment with milling
the BC films, also demonstrates that the formation of layers is not due to freezing conditions.

To the best of our knowledge, the origin of layered structures in BC has not yet been
ur

given a reasonable and consistent explanation. The effect of oxygen concentration on the BC
network structure was studied by Watanabe and Yamanaka (1995), who proposed a branching
Jo

model. According to them, when a cell divides, the fibril is split into two, i.e. the branching
occurs. After several divisions, the topology of fibrils is dendrimeric. A cone-shaped dendrimer
produced so far is depicted in Figure 9a. Following this model, Figure 9a illustrates, why the
number of fibers per volume unit increases towards the middle layer of the film. In this
simplified schematics the numbers indicate the number of branching points in the vicinity. The
entropic forces push this system towards equilibrium state with uniform density of fibrils, which
means that the cone is tending to open wider. At the same time, the position of cells is restricted
22
by the air/liquid interface from one side and the lack of oxygen in deeper areas that bacteria try
to escape, on the other side. This allows for the gradient existence for some time. Depending on
the rate of the synthesis the gradient is either increasing or being dissipated.

of
ro
-p
re
lP
na
ur
Jo

23
Jo
ur
na
lP

24
re
-p
ro
of
of
ro
-p
re
lP
na
ur

Fig 9 The model explaining the origin of layered structures in BC. A: the origin of network
density gradient. The BC fibers are in fact dendrimers, straight lines are individual BC fibrils.
Jo

After a cell divides, the fibril is split into two. The numbers indicate the number of branching
points in the vicinity. The number of fibers increases towards the inner portion of the dendrimer.
b: plane geometry case, c: cylindrical geometry for outside-of-tube and inside-tube cultivation.
See the text for the explanation.

The formation of a BC film in static conditions starts at the air/liquid interface with the
inoculation (see Figure 9 b) (i). Then the film thickens and covers the entire surface (ii). As the
film thickens, the density of fibrous material in the inner zone of the film becomes higher than
25
near the surfaces (iii). The more cells the central portion contains, the faster their number grows
due to cells division, and the more fibers are laid. The density gradient increases in a non-liner
way. At some point, the density of the fibrous network becomes so high that the bacterial cells
can no longer penetrate through it, and the film is now separated into two portions and bacteria
into two populations (iv). From now on, each portion develops independently. Bacteria located
on this impermeable layer keep laying fibers and create the first sheet (sheet 1). Since bacteria in
the top part of the film cannot move down, further growth of the film goes upwards, facilitated
by the good access of oxygen. After some time, a new gradient of density appears in the vicinity
of the liquid/air interface (v). The situation repeats, the dense layer becomes impenetrable for the
cells and later becomes sheet 2 (vi). Some bacteria are now trapped between the sheets. Cells
that stayed on top continue the synthesis of BC and form the next series of sheets in the same

of
way. Trapped ones make the sheets thicker and add new fibrils between them filling the space. In
the meantime, bacteria that stayed under sheet 1 also continue the synthesis. The availability of

ro
oxygen for them becomes less than for those located close to the interface, and the synthesis
slows down very soon as they sink deeper. Still, these bacteria have a better availability of
-p
nutrients. Several sheets can still be formed before the oxygen concentration falls below a certain
critical threshold. After that, if the cultivation is not terminated, an isotropic gel-like network is
slowly formed (vii). When the concentration of oxygen reduces further, the synthesis stops and
re
the cells switch into the dormant state.
lP

Borzani and de Souza (1995) have shown that the new layers of BC are formed on the top of the
film, by introducing their mark (human hair) on the top of the growing film and tracing its
position. The hair was moving downwards. However, their observation is not in contradiction
na

with our model. According to the authors, the mark was introduced not on the first day of the
cultivation, but when some part of the film had already been formed, presumably including
several layers developing downwards. The thickness of a human hair is 50-100 µm, which is 10-
ur

20 times more than a layer-to-layer distance. Thus, the thickness of a portion formed downwards
could be within the experimental error and was not detected by the authors. According to Park,
Jo

Chang, Jeong, & Hyun (2013), approximately 20 layers form per day, which makes 140 µm. The
formation of BC in the bottom portion should go much slower, so the thickness of the portion of
a film formed was even less, i.e. comparable to the hair thickness and just could not be detected
with bare eye.

It is very important to note that the formation of the density gradient and further
separation of the film is only possible if the rate of the synthesis is high enough. Otherwise, the
relaxation processes dissipate the density gradient before a sheet is formed. Thus, when the rate
26
of BC fibrils formation is low (no matter what the reason is), no pronounced density gradient can
exist, and therefore no sheets are formed and we see an isotropic network.

Older sheets are located closer to the bottom. Meanwhile, bacteria trapped between them
continue the synthesis, and gradually transform the low-density network between the sheets into
microtunnels that we see in Figure 1b, d and Figure 3a (Figure 9b (vii)).

In the cylindrical geometry case, the same logics can be applied. In the cultivation setup
proposed by Bodin et al. (2007), the BC growth starts near the silicon tube surface, through
which the oxygen is supplied (Figure 9c). Then the first sheet is formed around the tube in an
annular way. Part of bacteria are then trapped between the tube and the first dense sheet. Others
continue the process as if from the start, as shown in the drawing. Further from the tube, the

of
concentration of oxygen decreases, and the synthesis of BC slows down. At some point
(threshold), the synthesis is so slow that the density gradients dissipate and can no longer be

ro
formed; then the fibrous network prevails. In the setup proposed by Klemm, Schumann, Udhardt,
& Marsch (2001), the newer layers are formed inside the tube in a similar way (Figure 9c). In
that configuration, the inner portion of the BC does not contain visible layers either (Klemm,
-p
Schumann, Udhardt, & Marsch, 2001).
re
How could we determine the threshold? This is the minimal rate of BC synthesis at which
the diffusion and relaxation are too slow to destroy the density gradient before a fibrous sheet
impenetrable for bacteria is formed. The BC synthesis rate itself and the cells division rate in
lP

their turn, are mediated by several external factors, such as oxygen availability, temperature,
concentration of nutrients, C-source, bacterial strain (particularly, its productivity).
na

In all our experiments, well-distinguished layers were always found on the top of the
film, located close to the air/liquid interface, and therefore, close to the oxygen source, in this
area the availability of oxygen is the highest. This leads to the highest rate of film growth in this
ur

zone. Deeper under the surface, where earlier formed material is located, lies a zone of tunnels,
lying parallel to the film plane and changing the orientation from layer to layer. Beneath this
Jo

zone, we sometimes could find the third type of organization that showed cells and hardly
distinguishable layers (Figure 3a, right side), similar to those found in SEM images of BC grown
in form of tubes, in the peripheral areas distant from the oxygen source (central tube) (Bäckdahl,
Risberg, & Gatenholm, 2011). We believe that in our experiments, the third zone also forms later
and much more slowly than the middle portion, and when the cultivation was terminated too
early, we did not see it at all (Figure 1b). Therefore, the inner portion of the tubes grown by
Gatenholm’s group can be considered somewhat analogues to the top surface of the planar film.

27
If the BC film grows inside the tube and oxygen is supplied from outside, the layered structure is
still located next to the oxygen-permeable membrane, and the isotropic network is formed inside,
in the broth-filled tube (Klemm, Schumann, Udhardt, & Marsch, 2001). In pellets P2, the layers
were observed in the outer portion of the pellet, while the inner portion, located further from the
media and, therefore having less oxygen and nutrients available, was much less dense and had
isotropic structure, similarly to the previous case. In all the four cases, layered structures are
found in the areas with the highest oxygen availability and therefore, the highest rate of BC
synthesis.

Bodin et al. (2007) reported the following scaling for the thickness of the layered
structure zone: it is 1.5 times thinner at 35% oxygen concentration, and 2.5 times thinner at 20%
one, as compared to the thickness at 100% oxygen feeding. In order to account for this scaling,

of
we propose the following simple model. Let us assume, for simplicity, that there is a certain
invariant characteristic length at which the oxygen is consumed completely, l. Thus, for this

ro
boundary we have zero boundary condition. For another boundary, i.e. the one at the oxygen
source, we have the following conditions: 0.2, 0.35 or 1.0 of relative concentration, respectively.
-p
Considering only steady-state solution and assuming planar geometry (both assumptions are also
made for the sake of simplicity) one can easily obtain the following scaling relation:
re
𝑐𝑡 𝑙𝑥
1− = (1),
𝑐𝑥 𝑙
lP

where ct is the threshold concentration, cx is the boundary condition (0.2, 0.35 or 1 of relative
concentration) at the oxygen source boundary, lx is the distance where this threshold
concentration value is reached at the certain boundary condition (cx = 0.2, 0.35 or 1). Using the
na

scaling provided by Bodin et al. from the corresponding obvious relations:

𝑙0.2 1.5
=
ur

𝑙0.35 2.5

𝑙0.2 1
= (2),
Jo

𝑙1 2.5

𝑙0.35 1
= ,
𝑙1 1.5

and equation (1) one can calculate the following values for the threshold concentration: ct  0.12,
ct  0.13, ct  0.15, respectively. All the three values coincide rather well with each other,
allowing for the assumption of the validity of the proposed model, despite its extreme simplicity.

28
Therefore, 13% of oxygen content seems to be quite a reasonable estimation for the threshold
concentration value.

The next question is how the source of carbon can influence the inner structure of the
film. It is strongly doubtful that the biochemical path of cellulose polymerization occurring
inside the cell could have direct impact on it. More likely, the source of carbon influences the
forming structures indirectly, presumably via the rate of film formation, in other words, by
altering the productivity of the bacterial cells. When sorbitol was used as the source of carbon
instead of glucose, the architecture changed and BC sheets separated by fibrous zones were not
observed either in planar (films), or in spherical geometry (pellets) case. It is also known, that the
productivity of G. hansenii for both agitated and static regimes in sorbitol-containing media is
much less than that in the glucose-containing one (10 mg/L versus 700 mg/L of dried biomass).

of
Similarly, Rani & Appaiah (2011) compared the productivity for sorbitol-based medium versus
glucose-based one, using peptone as a nitrogen source, and showed that the former was only

ro
16% of that for glucose-containing one. This means that the synthesis of BC in sorbitol-
containing media goes much more slowly. In the experiments with pellets, though the agitation
-p
rate and, therefore, the oxygen availability were the same for P1 and P3, no layers formed in P3;
the network was isotropic. Despite the banded periodical structure of BC films grown on sorbitol
re
medium in static conditions, the inner architecture did not involve real sheets either. The
structure formed can be considered as an intermediate form between layers of microtunnels and
lP

isotropic network. The BET surface area for films grown in sorbitol-containing medium was 48
m2/g, whereas for glucose-based media it was only 36 m2/g. The formation of dense sheets
obviously reduced the BET surface area. The typical values of BET surface of BC found in the
na

literature vary from 20 m2/g to 200 m2/g (Gao et al., 2011; Liebner et al., 2010; Zhang et al.,
2019).

Nitrogen source can also have indirect impact on the BC organization. When yeast
ur

extract was replaced with ammonium sulfate, the productivity and, therefore, the rate of the
cellulose synthesis, reduced further, almost 2 times. Though both films were cultivated for the
Jo

same period, the film grown on ammonium sulfate-containing medium was thinner and
contained less material. That means that the rate of cellulose formation was slower. At the same
time, the structure of the film seen in SEM images also changed towards the formation of an
isotropic network. The layers with oriented fibers were seen only in a smaller part of the film,
while the rest of the BC network was isotropic. The BET surface area increased to 122 m2/g for
ammonium sulfate compared to 48 m2/g for yeast extract. Apparently, in the former case there
was less pronounced aggregation and less close packing of thin individual primary fibrils into
29
thicker secondary structures observed in SEM images (sheets, walls of microtunnels). Indeed, the
numbers indicate that in the first case the nanoscale-sized empty space between the primary
fibrils remains more accessible for nitrogen adsorption. These observations are in line with the
idea, that the fine architecture of BC mass is mediated by the rate of BC fibrils polymerization
and laying for all types of the geometry.

Mechanical properties of BC films studied here also reflect the architecture of BC. Films
grown on glucose-based medium and consisting of numerous dense parallel sheets had the
highest Young’s modulus of 6.5 ± 1.5 GPa and the smallest elongation at break of only 1.5 %.
This means that cellulose sheets are very hard to stretch. Sorbitol-based films (yeast extract as N-
source) with selected direction of fibrils orientation in each layer but without well-pronounced
sheets had somewhat smaller elastic modulus of 1.5 ± 0.5 GPa and higher elongation at break of

of
4%. Finally, sorbitol-based films with isotropic structure (ammonium sulfate as N-source) were
the softest and the most deformable, with the Young’s modulus of only 0.06 ± 0.02 GPa, which

ro
is 100 times less than for layered ones and 25 times less than for those with preferable
orientation of fibers. The elongation at break for isotropic films was also the highest (6%).
-p
Maximal tensile strength also correlates with the elastic modulus, being the highest (12 ± 4 MPa)
for films grown on glucose and 7 ± 1.5 MPa for layered sorbitol and yeast extract-based ones.
re
Despite the drastic reduction of the Young’s modulus for isotropic films, their tensile strength
(3± 1 MPa) reduced only 4 times compared to the maximal one. Knowing this correlation, one
lP

can tune the mechanical parameters and get desirable values by varying the cultivation
conditions.

More series of experiments with various carbon sources and rates of oxygen and nutrients
na

availability are needed in order to fully understand this impact in detail and it will be the subject
of our future work.
ur

5. Conclusions
Jo

The fine inner architecture of BC mass can be different and is determined by the
cultivation conditions, including geometry, sources of carbon and nitrogen, oxygen availability.
The organization of fibrils in BC gel-films can be isotropic, layered or tubular, with altering
preferential direction of tubes or fibers. Layered structures can be obtained both in static and
agitated growth conditions. We suggest a model describing the formation of layers. High rates of
BC synthesis allows for the formation of density gradients that further give rise to dense sheets
and separation into layers. Under conditions, when the BC mass forms more slowly, the
30
diffusion destroys the gradients and the organization is more isotropic. It is well known, that the
rate of BC mass synthesis, in its turn, is mediated by at least two factors: oxygen availability and
the productivity of the strain. The former reduces as we move away from the oxygen source (i.e.
deeper under the air/liquid interface or further from the permeable tube), whereas the latter
depends on the sources of carbon and nitrogen present in the cultivation medium and their
concentrations (for a given strain). From the practical point of view, the layered structure of BC
is a shortcoming for making composites, since the layers confine the diffusion of the material to
be incorporated and reduce the availability of the fibrils surface. On the other hand, the
formation of layers and tubes improves mechanical properties of BC that may be valuable for
other applications. Therefore, it is of practical importance, to understand the impact of the
cultivation conditions on the BC mass architecture in order to be able to produce BC with a

of
desired structural organization.

ro
Conflict of Interest: The authors declare that they have no conflict of interest

Acknowledgement
-p
re
This research was partly supported by the Russian Foundation for Basic Research, project № 18-
29-06049 mk. The authors acknowledge “Sernia Engineering” LLC company for methodical
lP

support with SEM imaging of pellets. Electron microscopy characterization of BC planar films
was performed in the Department of Structural Studies of Zelinsky Institute of Organic
Chemistry, Moscow. Mechanical testing was carried out at A.V.Topchiev Institute of
na

Petrochemical Synthesis, Moscow. The authors thank leading research fellow Dr. Alexander V.
Pastukhov from A.N. Nesmeyanov Institute of Organoelement Compounds for the
ur

measurements of BET surface area. The authors are grateful to Mr. Eugene Levin for his kind
technical assistance with drawing the 3D model of the BC mass organization.
Jo

References:

An, S.-J., Lee, S.-H., Huh, J.-B., Jeong, S.I., Park, J.-S., Gwon, H.J., Kang, E.-S., Jeong, Ch.-M.,
& Lim, Y.-M. (2017). Preparation and characterization of resorbable bacterial cellulose
membranes treated by electron beam irradiation for guided bone regeneration. International
Journal of Molecular Science, 18, 2236-2254. https://doi.org/10.3390/ijms18112236

31
Bäckdahl, H., Risberg, B., & Gatenholm, P. (2011). Observations on bacterial cellulose tube
formation for application as vascular graft. Material Science Engineering C, 31, 14–21.
https://doi.org/10.1016/j.msec.2010.07.010

Besseau, L., & Bouligand, Y. (1998). The twisted collagen network of the box-fish scutes. Tissue
& Cell, 30, 251-260. https://doi.org/10.1016/S0040-8166(98)80073-6

Bi, J.-C., Liu, S.-X., Li, C.-F., Li, J., Liu, L.-X., Deng, J., & Yang, Y.-C. (2014). Morphology
and structure characterization of bacterial celluloses produced by different strains in agitated
culture. Journal of Applied Microbiology, 117, 1305—1311. https://doi.org/10.1111/jam.12619

Bodin, A., Bäckdahl, H., Fink, H., Gustafsson, L., Risberg, B., & Gatenholm, P. (2007).
Influence of cultivation conditions on mechanical and morphological properties of bacterial

of
cellulose tubes. Biotechnology and Bioengineering, 97, 425-434.
https://doi.org/10.1002/bit.21314

ro
Borzani, W., & de Souza, S. J. (1995). Mechanism of the film thickness increasing during the
bacterial production of cellulose on non-agitated liquid media. Biotechnology Letters, 17, 1271-
1272. https://doi.org/10.1007/BF00128400
-p
Bouligand, Y. (1972). Twisted fibrous arrangements in biological materials and cholesteric
re
mesophases. Tissue & Cell, 4, 189-217. https://doi.org/10.1016/S0040-8166(72)80042-9
lP

Cai, Zh., & Kim, J. (2010). Preparation and characterization of novel bacterial cellulose/gelatin
scaffold for tissue regeneration using bacterial cellulose hydrogel. Journal of Nanotechnology in
Engineering and Medicine, 1, 021002-1-021002-6. https://doi.org/10.1115/1.4000858
na

Chen, P., Cho, S. Y., & Jin, H.-J. (2010). Modification and applications of bacterial celluloses in
polymer science. Macromolecular Research, 18, 309-320. https://doi.org/10.1007/s13233-010-
0404-5
ur

Gao, C., Wan, Y., Yang, Ch., Dai, K., Tang, T., Luo, H., Wang, J. (2011). Preparation and
Jo

characterization of bacterial cellulose sponge with hierarchical pore structure as tissue


engineering scaffold. Journal of Porous Materials, 18, 139–145.
https://doi.org/10.1007/s10934-010-9364-6

Giraud-Guille, M.-M. (1996). Twisted liquid crystalline supramolecular arrangements in


morphogenesis. International Review of Cytology, 166, 59-101. https://doi.org/10.1016/s0074-
7696(08)62506-1

32
Giraud-Guille, M.-M., Chanzy, H., & Vuong, R. (1990). Chitin crystals in arthropod cuticles
revealed by diffraction contrast transmission electron microscopy. Journal of Structural Biology,
103, 232-240. https://doi.org/10.1016/1047-8477(90)90041-A

Gromovykh, P. S., Bachman, M., Petrukhin, I. Yu., Dutka, K. V., & Butenko, I. E. (2018).
Bacterial cellulose biosynthesis by the producer of Gluconacetobacter hansenii in depth culture.
Biological sciences “Eurasian Scientific Association”, 6, 61-65.

Hassan, E. A., Abdelhady, H. M., El-Salam, S. S. A., & Abdullah, S. M. (2015). The
characterization of bacterial cellulose produced by Acetobacter xylinum and Komgataeibacter
saccharovorans under optimized fermentation conditions. British Microbiology Research
Journal, 9, 1-13. https://doi.org/10.9734/BMRJ/2015/18223

of
Hestrin, S., & Schramm, M. (1954). Synthesis of cellulose by Acetobacter xylinum: II.
Preparation of freeze-dried cells capable of polymerizing glucose to cellulose. Biochemical

ro
Journal, 58, 345-352. https://doi.org/10.1042/bj0580345

Hong, F., & Qiu, K. (2008). An alternative carbon source from konjac powder for enhancing
-p
production of bacterial cellulose in static cultures by a model strain Acetobacter aceti subsp.
xylinus ATCC 23770. Carbohydrate Polymers, 72, 545-549.
re
https://doi.org/10.1016/j.carbpol.2007.09.015

Huang, Y., Zhu, Ch., Yang, J., Nie, Y., Chen, Ch., & Sun, D. (2014). Recent advances in
lP

bacterial cellulose. Cellulose, 21, 1–30. https://doi.org/10.1007/s10570-013-0088-z

Kashin, A. S, & Ananikov, V. P. (2011). A SEM study of nanosized metal films and metal
na

nanoparticles obtained by magnetron sputtering. Russian Chemical Bulletin International


Edition, 60, 2602-2607. https://doi.org/10.1007/s11172-011-0399-x

Katepetcha, Ch., & Rujiravanit, R. (2011). Synthesis of magnetic nanoparticle into bacterial
ur

cellulose matrix by ammonia gas-enhancing in situ co-precipitation method. Carbohydrate


Polymers, 86, 162– 170. https://doi.org/10.1016/j.carbpol.2011.04.024
Jo

Keshk, S. M. A. S. (2014). Bacterial cellulose production and its industrial applications. Journal
of Bioprocesses and Biotechniques, 150, 4. https://doi.org/10.4172/2155-9821.1000150

Kim, J., Cai, Zh., & Chen, Y. (2010). Biocompatible bacterial cellulose composites for
biomedical application. Journal of Nanotechnology in Engineering and Medicine, 1:011006-1-
011006-7. https://doi.org/10.1115/1.4000062

33
Klemm, D., Schumann, D., Udhardt, U., & Marsch, S. (2001). Bacterial synthesized cellulose –
artificial blood vessels for microsurgery. Progress in Polymer Science, 26, 1561-1603.
https://doi.org/10.1016/S0079-6700(01)00021-1

Lee, K. Y., Buldum, G., Mantalaris, A., Bismarck, A. (2014). More than meets the eye in
bacterial cellulose: boisynthesis, bioprocessing, and applications in advanced fiber composites.
Macromolecular Bioscience, 14, 10-32. https://doi.org/10.1002/mabi.201300298

Li, G., Nandgaonkar, A. G., Habibi, Y., Krause, W. E., Wei, Q., Lucia, L. A. (2017). An
environmentally benign approach to achieving vectorial alignment and high microporosity in
bacterial cellulose/chitosan scaffolds. Royal Society of Chemistry Advances, 7, 13678-13688.
https://doi.org/10.1039/c6ra26049g

of
Liebner, F., Haimer, E., Wendland, M., Neouze, M.-A., Schlufter, K., Miethe, P., Heinze, T.,
Potthast, A., & Rosenau, T. (2010). Aerogels from unaltered bacterial cellulose: application of

ro
scCO2 drying for the preparation of shaped, ultra-lightweight cellulosic aerogels.
Macromolecular Bioscience, 10, 349–352. https://doi.org/10.1002/mabi.200900371
-p
Masaoka, S., Ohe, T., & Sakota, N. (1993). Production of cellulose from glucose by Acetobacter
xylinum. Journal of Fermentation and Bioengineering, 75, 18-22. https://doi.org/10.1016/0922-
re
338X(93)90171-4

Moniri, M., Moghaddam, A. B., Azizi, S., Rahim, R. A., Ariff, A. B., Saad, W. Z., Navaderi, M.,
lP

& Mohamad, R. (2017). Production and status of bacterial cellulose in biomedical engineering.
Nanomaterials, 7, 257. https://doi.org/10.3390/nano7090257
na

Nikolov, S., Petrov, M., Lymperakis, L., Fria´k, M., Sachs, C., Fabritius, H.-O., Raabe, D., &
Neugebauer, J. (2010). Revealing the design principles of high-performance biological
composites using ab initio and multiscale simulations: the example of lobster cuticle. Advanced
ur

Materials, 22, 519–526. https://doi.org/10.1002/adma.200902019

Park, J. K., Jung, J. Y., & Park, Y. H. (2003). Cellulose production by Gluconacetobacter
Jo

hansenii in a medium containing ethanol. Biotechnology Letters, 25, 2055-2059.


https://doi.org/10.1023/B:BILE.0000007065.63682.18

Park, M., Chang, H., Jeong, D. H., & Hyun, J. (2013). Spatial deformation of nanocellulose
hydrogel enhances SERS. BioChip Journal 7, 234-241. https://doi.org/10.1007/s13206-013-
7306-5

34
Picheth, G. F., Pirich, C. L., Sierakowski, M. R., Woehl, M. A., Sakakibara, C. N., de Souza, C.
F., Martina, A. A., da Silva, R., & de Freitas, R. A. (2017). Bacterial cellulose in biomedical
applications: a review. International Journal of Biological Macromolecules, 104, 97–106.
https://doi:org/10.1016/j.ijbiomac.2017.05.171

Pigaleva, M. A., Bulat, M. V., Gromovykh, T. I., Gavryushina, I. A., Lutsenko, S. V.,
Gallyamov, M. O., Novikov, I. V., Buyanovskaya, A. G., & Kiselyova, O. I. (2019). A new
approach to purification of bacterial cellulose membranes: What happens to bacteria in
supercritical media? Journal of Supercritical Fluids, 147, 59–69.
https://doi.org/10.1016/j.supflu.2019.02.009

Pircher, N., Veigel, S., Aigner, N., Nedelec, J. M., Rosenau, T., Liebner, F. (2014).

of
Reinforcement of bacterial cellulose aerogels with biocompatible polymers. Carbohydrate
Polymers, 111, 505–513. https://doi.org/10.1016/j.carbpol.2014.04.029

ro
Rani, M. U., & Appaiah A. (2011). Optimization of culture conditions for bacterial cellulose
production from Gluconacetobacter hansenii UAC09. Annals of Microbiology, 61, 781–787.
https://doi.org/10.1007/s13213-011-0196-7 -p
Saska, S., Teixeira, L. N., de Oliveira, P. T., Gaspar A. M. M., Ribeiro, S. J. L., Messaddeqa, Y.,
re
& Marchetto, R. (2012). Bacterial cellulose-collagen nanocomposite for bone tissue engineering.
Journal of Materials Chemistry, 22, 22102–22112. https://doi.org/10.1039/c2jm33762b
lP

Singhsa, P., Narain, R., & Manuspiya, H. (2018). Physical structure variations of bacterial
cellulose produced by different Komagataeibacter xylinus strains and carbon sources in static
and agitated conditions. Cellulose, 25, 1571-1582. https://doi.org/10.1007/s10570-018-1699-1
na

Vasconcelos, N. F., Feitosa, J. P. A., da Gama, F. M. P., Morais, J. P. S., Andrade, F. K., de
Souza Filho, M. S. M., & de Freitas Rosa, M. (2017). Bacterial cellulose nanocrystals produced
ur

under different hydrolysis conditions: properties and morphological features. Carbohydrate


Polymers, 155, 425–431. http://dx.doi.org/10.1016/j.carbpol.2016.08.090
Jo

Wang, Sh. Sh., Han, Y.-H., Chen, J.-L., Zhang, D.-Ch., Shi, X. X., Ye, Y. X., Chen, D. L., & Li,
M. (2018). Insights into bacterial cellulose biosynthesis from different carbon sources and the
associated biochemical transformation pathways in Komagataeibacter sp. Polymers, 10, 963-
983. https://doi.org/10.1016/10.3390/polym10090963

Watanabe, K., & Yamanaka, S. (1995). Effects of oxygen tension in the gaseous phase on
production and physical properties of bacterial cellulose formed under static culture conditions.
Bioscience, Biotechnology, and Biochemistry, 59, 65-68. https://doi.org/10.1271/bbb.59.65
35
Wegst, U. G. K., Bai, H., Saiz, E., Tomsia, A. P., & Ritchie, R. O. (2015). Bioinspired structural
materials. Nature Materials, 14, 23-36. https://doi.org/10.1038/NMAT4089

Yamanaka, S., Watanabe, K., Kitamura, N., Iguchi, M., Mitsuhashi, S., Nishi, Y., & Uryu, M.
(1989). The structure and mechanical properties of sheets prepared from bacterial cellulose.
Journal of Material Science, 24, 3141-3145. https://doi.org/10.1007/BF01139032

Yang, H., Yan, R., Chen, H., Lee, D. H., & Zheng, Ch. (2007). Characteristics of hemicellulose,
cellulose and lignin pyrolysis. Fuel, 86, 1781–1788. http://doi:10.1016/j.fuel.2006.12.013

Zhang, H., Xu, X., Chen, Ch., Chen, X., Huang, Y., & Sun, D. (2019). In situ controllable
fabrication of porous bacterial cellulose. Materials Letters, 249, 104–107.
https://doi.org/10.1016/j.matlet.2019.04.026

of
Zywicka, A., Peitler, D., Rakoczy, R., Konopacki, M., Kordas, M., & Fijałkowski, K. (2015).
The effect of different agitation modes on bacterial cellulose synthesis by Gluconacetobacter

ro
xylinus strains. Acta Scientiarum Polonorium Zootechnica 14, 137–150.
https://doi.org/10.1515/pjct-2016-0080
-p
re
lP
na
ur
Jo

36
Sample Onset temperature, C Peak temperature, C Enthalpy, J/g
Hollow pellet (P1) 316 348 220
Filled pellet (P2) 301 357 430
Sorbitol pellet (P3) 314 345 260
Film 327 361 146
Table 1. Pyrolysis point and enthalpy of BC in spherical phase (pellets) and planar phase (film).

of
ro
-p
re
lP
na
ur
Jo

37

You might also like