Download Continuous Production of Flexible Fibers from Transgenically

Communication
Continuous Production of Flexible Fibers
from Transgenically Produced Honeybee
Silk Proteinsa
Jacinta Poole, Jeffrey S. Church, Andrea L. Woodhead, Mickey G. Huson,
Alagacone Sriskantha, Ilias L. Kyratzis, Tara D. Sutherland*
Flexible and solvent stable fibers are produced after concentrated recombinant honeybee
protein solutions are extruded into a methanol bath, dried, drawn in aqueous methanol, then
covalently cross-linked using dry heat. Proteins in solution are predominantly coiled coil.
Significant levels of non-orientated -sheets form during
drying or after coagulation in aqueous methanol.
Drawing generally aligns the coiled coil component
parallel with the fibre axis and -sheet component
perpendicular to the fiber axis. The fibres are readily
handled, stable in the strong protein denaturants, urea
and guanidinium, and suitable for a range of applications such as weaving and knitting.
1. Introduction
There is growing interest in the use of structural proteins as
polymers for materials production. Proteins fold, assemble
into higher order structure, and interact with their
environment according to information that is contained
within their amino acid sequence. In principle, biotechnologists can modify a protein’s native amino acid sequence to
include ‘‘non-native’’ information and thus add specific
functionality desired by materials scientists. The potential
J. Poole, Dr. I. L. Kyratzis
CSIRO Materials Science and Engineering, Bayview Avenue,
Clayton VIC 3169, Australia
Dr. J. S. Church, A. L. Woodhead, Dr. M. G. Huson
CSIRO Materials Science and Engineering, Waurn Ponds VIC 3216,
Australia
Dr. A. Sriskantha, Dr. T. D. Sutherland
CSIRO Ecosystem Sciences, Clunies Ross Street, Acton ACT 2601,
Australia
E-mail: [email protected]
a
Supporting Information is available from the Wiley Online Library or
from the author.
Macromol. Biosci. 2013, 13, 1321–1326
ß 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
to have information within a polymer as well as within the
materials fabrication process, makes structural proteins
ideal templates for the development of advanced materials.
However, despite their promise, proteins are underrepresented in materials science mainly because biomimetic versions are difficult to produce. The recently
characterized silk of honeybees (Figure 1A) is made of
proteins well-suited to material development.[1] Recombinant silk proteins, that readily refold to adopt their native
coiled coil structure in solution[2] and in solid materials,[3]
can be produced on a large scale in transgenic systems.[4]
The silk protein structure is not reliant on amino acid
identity, as exemplified by the extreme sequence diversity
observed in homologous bee and ant silk proteins,[1] and
hence is ideally suited to accommodate ‘‘non-native’’
sequences. In this paper, we describe production of selfassembled recombinant honeybee silk proteins and their
use in an industrial-type process for the continuous
fabrication of flexible fibers (Figure 1B,C). Using Raman
spectroscopy, we describe changes in the molecular
structural and orientation of the proteins through the fiber
wileyonlinelibrary.com
DOI: 10.1002/mabi.201300231
1321
J. Poole et al.
www.mbs-journal.de
Figure 1. Coiled coil silks: A) Native honeybee silk (pale material indicated with arrow) and wax covering larval brood cells; in one cell a young
honeybee is emerging and another is empty after the young bee has departed; B) recombinant honeybee silk knitted into a tube; C)
recombinant honeybee silk woven into a sheet.
fabrication process. The ability to recombinantly generate
honeybee silk proteins suitable for industrial scale fabrication of materials makes these proteins of considerable
interest as templates for a range of industrial and medical
materials.
2. Experimental Section
2.1. Fiber Production
The AmelF3 (NCBI accession no: NP_001129680) honeybee silk
protein was expressed by fermentation as previously described.[4]
Between 1 and 10 mL of concentrated protein solutions (around
10% protein) were extruded through a stainless steel 27 gauge
needle with an internal diameter of 220 mm (Terumo) into a 2 m
long coagulation bath containing 80–90% methanol using a syringe
pump (New Era NE-1000). Fibers were taken from the coagulation
bath past a 40 cm long radiant heater system (Moretti) and dried
fibers were transferred into a 2 m long draw bath containing 70%
methanol. Within the draw bath, the fibers were drawn on a series
of 25 mm draw rollers (Retech) rotating at 20–100 rpm. The fibers
were taken from the draw bath and onto a collection roller (93 mm
diameter running at 12 rpm). Dried fibers were then heated to
190 8C for 1 h in an oven (Gallenkamp) to covalently cross-link the
proteins as described in Huson and co-workers.[5]
2.2. Mechanical Testing
Fibers were mounted across a 10 mm gap on paper frames, fixed at
either end with epoxy glue, and examined under an optical
microscope to determine the exact diameter of each fiber. Tensile
measurements were carried out at a strain rate of 6 mm min1 on
an Instron 5500R (Instron, USA) fitted with a 2.5 N static load cell.
Tests were conducted in air at 20 8C and 65% relative humidity.
2.3. Raman Spectroscopy
Raman spectra were obtained using an inVia confocal microscope
system (Renishaw, Gloucestershire, UK) with 514 nm excitation
from an argon ion laser through a 50 (0.75 na) objective. Incident
laser power was 0.784 mW as measured using a Nova power meter
fitted with a PD300-3W head (Ophir Optronics Solutions Ltd., Israel).
The Raman shifts were calibrated using the 520 cm1 line of a
silicon wafer. The spectral resolution was 1 cm1.
1322
Fibers were orientated with respect to the laser polarization
using a rotating stage. The system geometry and nomenclature
used to describe the four unique polarized spectra are described
elsewhere.[6] Briefly, the fiber is in the xy plane with the draw axis
coincident with the x-axis. The laser polarization was rotated using
a 1/2-wave plate while the spectrometer was fitted with a
polarization analyzer consisting of a polarizer and a 1/2-wave plate.
Fluorescence of the samples was quenched by 3 h of laser
exposure at 0.784 mW. After this time, the samples and spectra
were carefully watched for any signs of decomposition. In general
the fibers were found to be very stable for up to an additional 7 h of
laser exposure. Polarization measurements were made in static
mode covering the range 1800 to 1370 cm1. Each spectrum
collected consisted of 30 scans, each with an accumulation time of
40 s. To further reduce the noise four repeat spectra were coaveraged to produce the final spectra used in the analysis. Repeat
spectra were not collected sequentially.
All data manipulations and deconvolutions was carried out
using Grams AI software version 9.1 (Thermo Fisher Scientific, Inc.,
USA). Spectral deconvolution was carried out by first identifying
band components from the second derivative spectra obtained
using the Savitzky–Golay method.[7] Fits were based on the usage of
a minimal number of band components. All peak heights were
limited to the range greater than or equal to zero. In the
initial fitting steps, the band centers were only allowed to vary
by 5 cm1 from the frequency determined by the second
derivative spectra. In the final refinements, all parameters were
allowed to vary unconstrained. Two point linear baselines were
used throughout.
2.4. Infrared Spectroscopy
Infrared attenuated total reflectance (ATR) spectra were collected
from protein films and fibers using a Perkin Elmer Spectrum
100 Fourier transform infrared spectrometer fitted with an single
bounce diamond Universal ATR accessory and a room temperature
triglycine sulfate detector. Spectra were collected from 4500 to
750 cm1 at 2 cm1 resolution and 64 scans co-added.
2.5. Knitting and Weaving
Knitted and woven samples were fabricated from monofilament
protein silk of around 40 mm diameter. Samples of knitted tubes
were fabricated on a circular knitting machine (Harry Lucas R1-S,
Macromol. Biosci. 2013, 13, 1321–1326
ß 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.MaterialsViews.com
Artificial Honeybee Silk Fibers
www.mbs-journal.de
Germany) 1/1200 diameter using a 10 needle
cylinder (24 gauge cylinder). Loop length was
altered to produce a 3 mm tube. Individual
fibers were hand woven in a plain weave/
tabby weave structure with 100 ends at 40
ends per inch.
3. Results and Discussion
Honeybee proteins were expressed at
high level by fermentation in recombinant Escherichia coli, then purified
and processed to give a final solution of
10 w/v% protein, 10 103 M NaCl and
0.3 w/v% detergent (sodium dodecyl sulfate; SDS). The protein solution was
extruded into a bath containing 80–90%
methanol in which the protein solution
coagulated as a hydrated fiber with a
typical diameter of 150 mm (Figure 2A).
The fibers were dried by reeling past a
radiant heater before being transferred
into a draw bath containing 70% methanol. Within this second bath, the fibers
were drawn to three to four times their
original length on a series of draw rollers.
The drawn fibers were taken from the
bath onto a collection roller. In a final step,
the dried and drawn fibers were heated, a
process that covalently cross-links the
silk proteins.[5] When produced under
these conditions, the fibers had a final
diameter of 34 1 mm, mechanical
strength at break of 158 6 MPa and
strain at break of 42 10% (engineering
stress and strain). Representative stress–
strain curves are shown in Figure S1,
Supporting Information. For comparison,
native honeybee silks have engineering
stress of around 132 MPa and engineering
strain of around 200%.[8]
The proteins in native honeybee silk
are predominantly coiled coil. However,
they also contain b-sheet structures[9]
suggesting that, as with spider and
silkworm silks, formation of b-sheets is
an important mechanism to cross-link
proteins in the natural material.[10] We
used Raman spectroscopy to monitor
formation of b-sheets in the recombinant
silk proteins during the fabrication process. Deconvolution of spectra from airdried solutions indicated that the structure of the proteins was predominantly
www.MaterialsViews.com
Figure 2. Recombinant honeybee fiber fabrication and characterization: A) Schematic
depicting the continuous production process; B–G) Raman spectra structural data of the
amide I region, including deconvolutions (B,D,F) and polarized spectra (C,E,G) from
protein dope before injection (B,C), after injection into the coagulation bath (D,E),
and from fibers taken-off the collection roller (F,G). The area of the amide I band
components, as identified by 2nd derivative spectroscopy and spectral deconvolutions,
are given in the text and in the Supporting Information, Table S1.
Macromol. Biosci. 2013, 13, 1321–1326
ß 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
1323
J. Poole et al.
www.mbs-journal.de
coiled coil (43%; orange trace in Figure 2B) with lesser
amounts of b-sheet (32%; blue trace in Figure 2B). The
similarity in the Ixx and Iyy spectra from polarized Raman
spectroscopy of these proteins indicated that the structures
were isotropic (Figure 2C). In this state, the proteins were
water soluble and the dried material was weak. As expected
from the protein dope composition, ATR spectra of the air
dried protein solutions revealed features indicative
of SDS[11] including features at 2927 and 2855 cm1
(CH2 groups) and 1215, 1059, and 967 cm1 (C—O—SO3
groups; Figure S2, Supporting Information).
Water-stable material could be generated when the
protein solution was extruded into aqueous alcohol, salt
solutions or by heating to 70 8C, but the hydrated material
readily broke apart when handled. Dehydration considerably improved the handleability of the material and,
therefore, a drying stage was incorporated into our process
(Figure 2A). Fiber drying was accelerated by the use of high
protein concentrations (>9%) and high concentrations of
methanol (80–90%) in the coagulant bath and hence these
conditions were used in our process. When produced under
these conditions, fibers could easily be transferred past a
radiant heater and into a drawing bath with different
solvent conditions and drawn. The molecular structure of
the coagulated material was predominantly b-sheet (36%)
with slightly less coiled coil (31%; Figure 2D). Analysis of the
infrared spectrum of the fibers did not find the features
indicative of SDS (2927, 2855, 1215, 1059, and 967 cm1,
Figure S2, Supporting Information) indicating that the
small amount of detergent in the protein dope did not
coagulate with the protein. The lower limit of SDS detection
using this method is of the order of 0.01 w/v% and therefore
if any SDS is present in the fiber, it is less than this amount.
The injection/coagulation process did not alter the orientation of the coiled coil structure, as indicated by the
similarity of the Ixx and Iyy polarized Raman spectra at
1654 cm1 (Figure 2E). However, the Ixx spectrum at
1669 cm1 (b-sheet) was slightly more intense than the
Iyy spectrum indicating that the process had resulted in the
b-sheet tending towards a perpendicular orientation
relative to the fiber axis (Figure 2E). The unexpected
perpendicular orientation of the b-sheets within the fiber is
discussed in more detail below.
Honeybee silk proteins are ordered in the silk gland prior
to spinning, leading to high levels of protein order in the
native silk.[12] In our process a drawing stage was included
that served to align the structural components along the
fiber axis. Polymers cannot be drawn below their glass
transition temperature (Tg). In our process, drawing was
performed in aqueous methanol with the water in the
drawing bath acting to lower the Tg of the proteins in the
undrawn fibers and the methanol preventing dissolution of
the material into the water. The water content in the
drawing bath dictated the rehydration time required before
1324
the fibers could be drawn, with equivalent diameter fibers
requiring eight times longer rehydration time before draw
in 90% methanol in comparison to the time required in 50%
methanol (Figure S3, Supporting Information). A 70%
methanol drawing bath was used in the continuous process
as this condition allowed rapid rehydration and then
drawing of the fiber, yet still allowed sufficient drying of the
fiber before it was taken up on the collection reel. For the
data presented here we used a draw ratio of 1:4.5 across
three rollers in the draw bath (Figure 2A). Higher draw ratios
generated fibers with greater strength at break, but the
fibers were more susceptible to break during drawing.
Drawing did not significantly alter the proportions of
different protein secondary structures in the material, with
deconvolutions of the Raman spectra demonstrating that
drawn fibers contained similar levels of b-sheet and coiled
coil to that in undrawn fibers (Figure 2F). However,
comparison of the Ixx and Iyy polarized Raman spectra
revealed that drawing had led to alignment of both
components. The modest increased intensity at
1654 cm1 in the Ixx spectrum relative to the Iyy spectrum
indicated that the helices were moderately aligned parallel
to the fiber axis, whereas the much greater intensity
difference at 1669 cm1 indicated that the b-sheets had
become highly aligned (Figure 2G). As with the spectra
obtained from coagulated fibers (Figure 2E), the stronger
intensity of the Ixx spectrum compared to the Iyy spectrum
indicated that the b-sheets had aligned perpendicular to the
fiber axis. This structure is similar to the cross-b structure
observed in lacewing egg stalks[13] and unlike the b-sheets
that are orientated parallel to the fiber in silkworm and
spider silks. Molecular units will orientate in the direction of
draw so that their longest dimension is parallel with the
direction of the draw force. Orientation of the b-sheets in
the silk fibers in the perpendicular direction implies that
they are stacked into crystallites deeper than the b-sheets
are wide, akin to the ribbons that make up lacewing egg
stalk silk.[14] The high and consistent level of b-turns
detected in all samples (20–24%; 1694 and 1683 cm1;
Figure 2B,D,F) is consistent with the presence of such
molecular units.
Materials produced after reconstituted or recombinant
spider and silkworm proteins coagulated in methanol are
generally associated with brittleness. The brittleness is
attributed to ‘‘imperfect’’ b-sheets resulting from the
methanol inducing a rapid increase in the spider/silkworm
silk protein’s glass transition temperature.[15] Less brittle
fibers are produced when reconstituted silkworm silk
protein solutions are extruded into heated (60 8C) ammonium sulfate, presumably because the salt precipitates the
proteins without the rapid dehydration observed in
methanol, resulting in a slower molecular transformation.[16] The presence of SDS in the honeybee silk protein
solution would serve to retard b-sheet formation until the
Macromol. Biosci. 2013, 13, 1321–1326
ß 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.MaterialsViews.com
Artificial Honeybee Silk Fibers
www.mbs-journal.de
Figure 3. Images of recombinant honeybee silk fibers: A) Scanning electron microscopy image of a wide diameter fiber after being knotted
whilst dry; B–F) Light microscopy images of fibers in solution: B) non-heat treated fiber in water; C) non-heat treated fiber in 1 M guanidinium
chloride showing characteristic swelling of the fiber; D) heat treated fiber in water; E) heat treated fiber in 1% SDS. F) heat treated fiber in 6 M
guanidinium chloride. Scale bar for all images: 200 mm.
detergent had sufficiently diffused from the fibers.
Therefore, the flexibility of the material is likely the
consequence of the configuration of the b-sheets within
the material (discussed above) coupled with an increased
time course of b-sheet formation compared to that observed
with spider and silkworm proteins.
In addition to b-sheet cross-links, the silk proteins of
honeybees are covalently cross-linked.[9b] In general, dry
heat treatment of proteins will generate isopeptide (amide)
or ester bonds between residues with acidic side chains
located in close proximity to residues containing amines or
hydroxyl groups, respectively.[17] Previously, it was shown
that recombinant honeybee sponges had improved mechanical properties and chemical stability after covalently
cross-linking using a dry heat treatment of 190 8C for 1 h.[5]
Covalent cross-linking of the fibers under similar conditions
generated flexible fibers (Figure 3) that were stable in a
range of strong protein denaturants including 1% SDS and
6 M guanidinium chloride (Figure 3E,F). Untreated fibers, on
the other hand, dissolved in 0.1% SDS, swelled considerably
in 1 M guanidinium chloride (Figure 3C) and dissolved in
solutions containing higher guanidinium chloride concentrations. The resultant heat-treated fibers were readily
handled and were suitable for a range of applications such
as weaving and knitting (see Figure 1B,C).
4. Conclusion
This paper describes a process to continuously produce
flexible, solvent stable protein fibers. The process uses
transgenically produced honeybee silk proteins. The
proteins are initially coiled coil in solution. b-Sheets form
when the protein is dried, either in air or by coagulation in
aqueous methanol. The coiled coils and b-sheets can be
aligned by drawing the fiber in methanol solutions. In
contrast to silkworm and spider silk, the orientation of the
drawn b-sheets is perpendicular to the fiber axis, a feature
that likely underpins the flexible properties of the fibers.
www.MaterialsViews.com
The fibers can readily be covalently cross-linked by heating
to produce material stable in strong protein denaturants,
including urea and guanidinium. Considerable control is
afforded during the fabrication process: the amount of bsheet cross-links can be controlled by coagulation conditions; molecular alignment can be controlled by the
amount of draw; and covalent cross-links can be controlled
by the extent of dry heat treatment. The ability to fabricate
recombinant proteins adds an additional dimension to
further development of silk materials: modern molecular
biology can be employed to modify the primary sequence of
the protein to add functionality to the final material. For
example, sequences that dictate protein–protein or biopolymer–environment interactions can be added. The
ability to control the protein composition as well as the
fabrication process allows the design of materials for
individual applications that utilize biomimetic properties
of the material as well as adding functionality as required
for specific needs.
Acknowledgements: This work was supported by CSIRO and
Lonza. The authors would like to thank Monique van Nieuwland
for weaving the silk, and Peter Herwig for knitting the silk. The
authors also thank Bea Lipson for images of the knitted sample
and woven sample, and Andrew Abbott for the schematic of the
production process. The image of honeybee silk was reproduced
from http://www.alexanderwild.com with the permission of Alex
Wild.
Received: May 8, 2013; Revised: June 12, 2013; Published online:
July 23, 2013; DOI: 10.1002/mabi.201300231
Keywords: biomimetics; Raman spectroscopy; silk worms;
transgenics
[1] a) T. D. Sutherland, P. M. Campbell, S. Weisman, H. T.
Trueman, A. Sriskantha, W. J. Wanjura, V. S. Haritos, Genome
Res. 2006, 16, 1414; b) T. D. Sutherland, S. Weisman, H. T.
Trueman, A. Sriskantha, J. W. H. Trueman, V. S. Haritos, Mol.
Biol. Evol. 2007, 24, 2424.
Macromol. Biosci. 2013, 13, 1321–1326
ß 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
1325
J. Poole et al.
www.mbs-journal.de
[2] A. A. Walker, A. C. Warden, H. E. Trueman, S. Weisman,
T. D. Sutherland, submitted to J. Mater. Chem. 2013, 1,
3644.
[3] a) T. D. Sutherland, J. S. Church, X. Hu, M. G. Huson,
D. L. Kaplan, S. Weisman, PLoS One 2011, 6, e16489;
b) C. R. Wittmer, X. Hu, P. C. Gauthier, S. Weisman,
D. L. Kaplan, T. D. Sutherland, Acta Biomater. 2011, 7,
3789.
[4] S. Weisman, V. S. Haritos, J. S. Church, M. G. Huson, S. T. Mudie,
A. J. W. Rodgers, G. J. Dumsday, T. D. Sutherland, Biomaterials
2010, 31, 2695.
[5] M. G. Huson, J. S. Church, J. M. Poole, S. Weisman, S.
Sriskantha, A. C. Warden, J. A. M. Ramshaw, T. D. Sutherland,
PLoS One 2012, 7, e52308.
[6] J. S. Church, A. J. Poole, A. L. Woodhead, Vib. Spectrosc. 2010,
53, 107.
[7] A. Savitzky, M. J. E. Golay, Analyt. Chem. 1964, 36, 1627.
[8] H. R. Hepburn, H. D. Chandler, M. R. Davidoff, Insect Biochem.
Mol. Biol. 1979, 6, 69.
1326
[9] a) T. Kameda, Y. Tamada, Int. J. Biol. Macromol. 2009, 44, 64;
b) T. D. Sutherland, S. Weisman, A. A. Walker, S. T. Mudie,
Biopolymers 2012, 97, 446.
[10] Y. Termonia, Macromolecules 1994, 27, 7378.
[11] R. B. Viana, A. B. F. da Silva, A. S. Pimentel, Adv. Phys. Chem.
2012, 2012, 14.
[12] N. E. Flower, W. Kenchington, J. R. Micro. Soc. 1967, 86, 297.
[13] S. Weisman, S. Okada, S. T. Mudie, M. G. Huson, H. E. Trueman,
A. Sriskantha, V. S. Haritos, T. D. Sutherland, J. Struct. Biol.
2009, 168, 467.
[14] A. J. Geddes, K. D. Parker, E. D. T. Atkins, E. Beighton, J. Mol.
Biol. 1968, 32, 343.
[15] X. Chen, Z. Z. Shao, D. P. Knight, F. Vollrath, Proteins 2007, 68,
223.
[16] G. Zhou, Z. Shao, D. P. Knight, J. Yan, X. Chen, Adv. Mater. 2009,
21, 366.
[17] a) J. K. P. Weder, U. Scharf, Z. Lebensm. Unters. Forsch. 1981,
172, 9; b) Z. H. Mohammed, S. E. Hill, J. R. Mitchell, J. Food Sci.
2000, 65, 221.
Macromol. Biosci. 2013, 13, 1321–1326
ß 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.MaterialsViews.com