Anshu Bagga Mathur,

Alan Tonelli,
Thomas Rathke,
The Dissolution and
Sam Hudson Characterization of Bombyx
Fiber and Polymer Science
Program mori Silk Fibroin in Calcium
Box 8301
North Carolina State
Nitrate–Methanol Solution
University and the Regeneration of Films
Raleigh, NC 27695-8301
Received 20 June 1996;
accepted 6 November 1996

Abstract: Bombyx mori silk fibroin from the silkworm was found to be soluble in a calcium
nitrate–methanol system. Fibroin dissolves in 75% w/v Ca(NO3 )2 /MeOH solution at a temper-
ature of 677C. The viscometric behavior of the fibroin–salt solution was analyzed and the
fibroin’s secondary structures were determined via 13C solution nmr. Fourier transform ir, solid
state 13C-nmr, x-ray diffraction, differential scanning calorimetry, scanning electron microscopy
(SEM), and polarizing microscopy were used to characterize regenerated films and fibers.
A compositional phase diagram of fibroin in the salt solution was constructed. Viscosity
data indicate that there is aggregation of fibroin chains within the salt solution. The extremely
high value of intrinsic viscosity of 8.7 dL/g at 298 K may be due to aggregation. Aggregation
may be caused by the complexing of calcium ions with the fibroin chains at their amide linkages.
The energy required for viscous flow for the fibroin solution ( DHvis Å 9.03 kcal/mol) is greater
than that of the solvent ( DHvis Å 7.01 kcal/mol). Chain entanglements may be hindering the
free motion of chains thus increasing the energy required for the viscous flow. 13C-nmr shows
that fibroin chains exist in two independent conformational environments. While most of the
molecule is in a random coil conformation, there is evidence of some order within the chains
of fibroin.
In as-cast regenerated films, the fibroin chains are in a random coil/ a-helix conformation
with some b-sheet content. Crystallinity induced by immersion of thin films in methanol is
evidenced via x-ray diffraction, which shows lattice spacings at 4.042 Å. Thin films have a
fibrillar morphology that is clearly shown under the SEM and the polarizing microscope. Fibers
were hand pulled from the concentrated fibroin–salt solutions and coagulated with acetone
and methanol. A microscopic analysis was done using the polarizer. q 1997 John Wiley &
Sons, Inc. Biopoly 42: 61–74, 1997

Keywords: silk fibroin; dissolution; calcium nitrate; fibers

Correspondence to: Sam Hudson

Contract grant sponsor: U.S. Army
Contract grant number: DAAK60-93-K-0016
q 1997 John Wiley & Sons, Inc. CCC 0006-3525/97/010061-14

61

/ 8k24$$5440 05-16-97 [Link] bpa W: Biopolymers

62 Mathur et al.

INTRODUCTION such as the b-sheets and a-helices, that are charac-

teristic of fibroin.
Recently silk fibroin has been investigated as a bio- Therefore, this paper will characterize the disso-
material to be used in the biotechnological and bio- lution of silk fibroin in calcium nitrate and methanol
medical fields. Properties like high strength with by analyzing solution behavior and discussing the
flexibility, blood compatibility, water permeability, secondary structures present in solution. Part two
and high permeability to oxygen have made silk an of this paper involves the characterization of the
excellent candidate for biomedical applications as regenerated films by discussing the impact of the
a nontextile material. secondary structures found in fibroin films regener-
Silkworms and spiders have been able to manu- ated via the Ca(NO3 )2 –MeOH solvent.
facture silk and use it to engineer structures such
as cocoons, webs, nests, and egg stalks. In order
to engineer silk to cater to a specific biomedical
application, it must be regenerated in a desirable EXPERIMENTAL PROCEDURES
form. To date, the dissolution of fibroin in lithium
salts and others dissolved in water or formic acid Silk fiber skeins were purchased on the open market in
have been investigated thoroughly. A number of Bangkok. Sericin, an outer layer protein, is removed via
these solvents have been described in a series of scouring in order to obtain silk fibroin. The raw silk was
patents assigned to the E. I. DuPont de Nemours & scoured using 0.75% w/v sodium dodecyl sulfate per
Co.1 – 3 These patents disclose further methods for gram of raw silk, at 957C for one hour.
producing solutions of fibroin and other fiber-form- The solvent system was prepared by dissolving anhy-
ing polypeptides, in hexafluoroisopropanol in com- drous calcium nitrate in methanol. Calcium nitrate dis-
bination with lithium salt solutions and the like. solves in methanol at room temperature with adequate
Though fibers and films may be obtained from these stirring. A clear and colorless solution of 75% w/w
Ca(NO3 )2 –MeOH was prepared. The scoured silk was
solutions, the regenerated form has deficiencies
cut in to small pieces and mixed with the solvent. The
such as the difficulty in removing lithium salts via temperature at which dissolution occurs is near the boil-
dialysis, and there are also the toxicity concerns ing point of methanol between 65 and 707C. Dissolution
associated with these materials. The removal of salts usually occurred within two hours with vigorous stirring.
is necessary to obtain a pure form of regenerated An Omega Miniature Autotune Temperature Controller
film or fiber. CN 9000A Series thermostat was used to keep the temper-
A novel salt system, which dissolves silk fibroin, ature constant. A stainless steel stirrer was attached to a
was discovered in our laboratory. Bombyx mori silk Fisher Scientific Sted Fast Stirrer motor SL 2400 for
fibroin from the silkworm was found to be soluble constant stirring.
in a calcium nitrate–methanol [Ca(NO3 )2 –MeOH]
system. After dialysis, the silk solution was used to
cast transparent thin films that exhibit new crystal- Solution Properties
line forms and novel surface features. Oriented silk
fibers were readily hand pulled from the undialyzed Determination of the Compositional Phase Dia-
salt solution. gram for the Dissolution of Silk Fibroin in Cal-
cium Nitrate–Methanol. Calcium nitrate–methanol
The molecular conformations of silk fibroin
solutions of concentrations 40, 50, 55, 60, 65, 70, 75, 80,
chains in solution as well as in thin films were deter- and 90% w/w were made. Silk was added in increments
mined. Characterization of the fibroin solutions starting from less than 1, 2, and 4% w/v concentrations.
were done by nmr and dilute solution viscosity. The appearance of silk in these solutions were recorded
Nmr, differential scanning calorimetry, x-ray, Fou- and these were noted as end marks. Concentrations were
rier transform ir (FTIR), SEM, and polarizing mi- increased in 0.5–1% increments and the results were
croscopy were used to characterize the thin films. noted. The appearance of these solutions were recorded
The dissolution of silk in this new solvent system to construct a compositional phase diagram.
has provided an opportunity to comprehend the mo-
lecular nature of the fibroin protein. The secondary Viscosity Measurements. Flow times of 75% w/w
structures found in fibroin are responsible for its calcium nitrate–methanol solution and fibroin solutions
physical properties such as tensile strength, water of concentrations 0.05, 0.0667, and 0.1 g/100 mL at 25,
permeability, etc. For the regenerated films of this 30, and 407C were obtained. The viscometer was a size
study, FTIR and 13C-nmr spectral analysis show dis- 2, D453 Ubbelhode type, with a viscometer constant of
tinct peaks for various types of secondary structures, 0.0908 mm2 /s 2 (cSt/s).

/ 8k24$$5440 05-16-97 [Link] bpa W: Biopolymers

 Dissolution and Characterization of Silk Fibroin 63

NMR Analysis of Fibroin Solutions. 13

C-nmr spec- was poured in a glass petri dish and dried at 707C and at
tra of 3% silk solutions and dialyzed solutions were ob- room temperature.
tained and compared to the spectra of model tripeptides
and tetrapeptides containing alanine and glycine. Deuter- Treatment of Films with Methanol and Water. As-
ated methanol was used to prepare the salt solutions. cast films were treated with methanol and then with water.
The 13C-nmr instrument was a General Electric GN 300 These were then characterized and compared with each
Omega 300 MHz spectrometer with a Sun 3/160 based other. As-cast films are labeled as fibroin 2, as-cast films
computer, GE-Omega 5.0 software, 5 mm dual ( 13C and immersed in methanol and air dried at room temperature
1
H) probe, and an Oxford wide bore magnet. The experi- are called fibroin 3, and fibroin 3 immersed in water and
mental conditions were as follows: One pulse with Waltz then dried at room temperature are named as fibroin 4.
16 broadband decoupling, with an observed frequency of
75.57 MHz, 8000 data points, spectral width of 25 kHz,
dwell time of 40, pulse delay of 0.75 s, and the total Characterization of Regenerated Films
number of scans were 43,000.
FTIR spectra were obtained with a Nicolet 510P FTIR
spectrophotometer. Thirty two scans were done for each
Regenerated Fibroin Thin Films. Concentrations of sample and the spectral range was 4800–400 cm01 . Films
fibroin in salt solution, ranging from 3 to 10% w/v, were were mounted on a magnetic sample holder, while fibers
used to cast films. These solutions were dialyzed against and other samples were prepared as KBr pellets.
distilled water over a period of four days through a semi- The x-ray diffraction data were obtained using a Scintag
permeable membrane. The membrane tubing with a diam- USA XDS-2000. CuKa radiation with a wavelength of
eter of 31.8 mm was a Baxter Spectrapor1 membrane, 1.5418 Å was used. The scan range was 2Q Å 45–37.
mwco Å 6–8000. The resulting solution inside the bag Solid-state CPMAS/DD 13C-nmr spectra of the regen-

FIGURE 1 Compositional phase diagram of silk fibroin in Ca(NO3 )2 –methanol.

/ 8k24$$5440 05-16-97 [Link] bpa W: Biopolymers

64 Mathur et al.

FIGURE 2 Huggins plot for the fibroin–salt solution.

erated films and the original fibroin were observed on that dissolves fibroin. Solutions upto 18% w/v fi-
a Chemagnetics CMC200S 200 MHz spectrometer. The broin were obtained with this solvent composition.
experimental conditions were contact time of 2 ms, pulse
delay of 3 s, 1H 90 deg pulse of 5.2 ms and the number
Intrinsic Viscosity. The viscosity data measured at
of scans was 8000. The dipolar decoupling (DD) field
strength was 47 kHz and the magic angle spinning
25, 30, and 407C were plotted by the Huggins plot
(MAS) rate was 4 kHz. (Figure 2). The intercept of this linear equation ex-
Regenerated films were examined under the Phillips trapolates to the intrinsic viscosity [ h]. The intrinsic
PS 505T scanning electron microscope while the hand- viscosities, in dL/g, at 25, 30, and 407C are 8.71,
pulled fibers were viewed under the polarizing light mi- 10.41, and 8.94, respectively. The slopes for these
croscope. curves are 127, 93, and 103 for 25, 30, and 407C,
respectively. These high values of intrinsic viscosity
and the slope (k * constant) indicate the aggregation
RESULTS AND DISCUSSION of molecules, ionic effects, and/or shearing effects.

Solution Behavior
Table I Relationship Between the Intrinsic
Phase Diagram. Figure 1 is the phase diagram for Viscosity and the Perturbed Dimensions of Silk
the dissolution of silk fibroin in calcium nitrate– Fibroin in Ca(NO3)2 –MeOH
methanol solution. Fibroin swells in 40–55% w/w
salt solution. There is more swelling in 50% w/w Intrinsic Perturbed
compared to the 40% w/w solution. In 55% w/ Temperature Viscosity Char. Expansion
w solution, a small fraction of fibroin dissolves. (7C) (dL/g) Ratio Factor, a
Between 60 and 75% w/w fibroin dissolved to vary-
ing degrees. A saturated 80% w/w salt solution did 25 8.7 14.1 2.44
30 10.4 15.9 2.58
not dissolve any significant amount of fibroin. A
40 8.9 14.3 2.46
75% w/w salt solution is the optimum concentration

/ 8k24$$5440 05-16-97 [Link] bpa W: Biopolymers

 Dissolution and Characterization of Silk Fibroin 65

Table II Chemical Shifts in ppm for the Fibroin Solutions and Reference Polypeptides
vs Tetramethylsilane (TMS)

Random Coil Conformation Shifts of

Carbon Type Dialyzed Solution Salt Solution Model Peptidesa in Salt Solution

Glycine
C|O 172.1 171.5 167.3
Ca 43.2 42.1
Alanine
C|O 176.3 175.1 173.7
Ca 50.6
Cb 17.1 17.3 (s) 16.9
19.5 (w)

* Ala-Gly-Gly and Ala-Leu-Ala-Leu.

The absolute viscosity of the solvent was deter- g/mol and the average repeat unit molecular weight
mined to be 37.2, 29.7 and 21.0 cP at 25, 30, and M0 is 64 g/mol.6 The perturbed characteristic ratios
407C, respectively. The activation energy or the heat of fibroin in dilute salt solution were obtained at
of activation for viscous flow DHvis of the solvent 25, 30, and 407C.
system was calculated to be 7.01 kcal/mol, from The unperturbed characteristic ratio was calcu-
the Arrhenius plot. Hudson 4 calculated this parame- lated by the matrix method derived by Flory.7 The
ter for NH3 –NH4SCN, which is a cellulose solvent, average transformation matrices, averaged over all
and found DHvis to be equal to 3.412 kcal/mol. conformations in the energy maps, were used for
Interestingly, the cellulose solvent is most effective alanine and glycine. Then the repetitive sequence
at about 273 K or less and the Ca(NO3 )2 –MeOH of the fibroin molecule was used to determine the
solvent at 340 K. A similar analysis of the tempera- end to end distance and the characteristic ratio.
ture dependence of the absolute viscosity of the The molecular expansion factor a can also be
solutions of fibroin in Ca(NO3 )2 –MeOH leads to calculated [Eq. (2)]. The square of the expansion
an activation energy of 9.03 kcal/mol for the vis- factor a is equal to the ratios of perturbed to unper-
cous flow of a 0.1% fibroin solution. By way of turbed characteristic ratios or dimensions.
contrast, Hudson 4 observed activation energies for
the viscous flow of cellulose in NH3 –NH4 SCN a 2 Å » r 2 … /nl 2 / » r 20 … /nl 2 Å » r 2 … / » r 20 … (2) 5
solutions in the range of 4–6 kcal/mol. Keeping in
mind that the molecularly dispersed cellulose chains The unperturbed characteristic ratio for fibroin was
are expected to be considerably more extended and calculated to be 2.37. The values for the expansion
rigid than randomly coiling fibroin chains, the in- factor a for fibroin in calcium nitrate–methanol so-
crease in the activation energy for viscous flow of lution are listed in Table I.
the fibroin solutions compared with the NH3 – Recently Jackson and O’Brien 8 have reported on
NH4SCN–cellulose solutions seems to imply that the solution behavior of Nephila clavipes dragline
the fibroin chains in Ca(NO3 )2 –MeOH are inter- silk in 0.01M sodium trifluoroacetate/hexafluoroi-
acting strongly and possibly aggregating. sopropanol by means of SEC, viscometry, and light
scattering measurements. It is interesting to note
that the unperturbed dimensions CQ Å 2.43 ( a
Determination of the Perturbed and
Å 2.2) estimated by the method of Tonelli and Flory 9
Unperturbed Chain Dimensions
from the intrinsic viscosities, radii of gyration, and
The perturbed chain dimensions of the fibroin mole- second virial coefficients reported for dragline silk
cule can be estimated from the intrinsic viscosity in the sodium trifluoroacetate/hexafluoroisopropanol
according to Eq. (1) assuming no aggregation. solvent agree closely with those estimated for ran-
domly coiling polypeptides with the same primary
» r 2 … /nl 2 Å ([ h]/ f(Mw ) 1 / 2 ) 2 / 3 (M0 /l 2 ) (1) 5 structure as silk. In Ca(NO3 )2 –MeOH the perturbed
dimensions of fibroin as observed by intrinsic viscos-
The weight average molecular weight Mw is 350,000 ity exceed those of the dragline silk by 30%. Either

/ 8k24$$5440 05-16-97 [Link] bpa W: Biopolymers

66 Mathur et al.

Ca(NO3 )2 –MeOH is a much better solvent for ran- in dialyzed (no salt) solution. Model peptides con-
domly coiling silk than is sodium trifluoroacetate/ taining alanine (Ala) and glycine (Gly) were dis-
hexafluoroisopropanol, or the silk fibroin may be par- solved in the salt solution to determine the random
tially ordered in the former solvent. This latter possi- coil chemical shifts. Table II contains the 13C chemi-
bility receives support from our observation of two cal shifts of the silk solutions and alanine and gly-
resonances for the alanine methyl carbons of silk cine model peptides. The b-carbon (Cb) of alanine
fibroin when dissolved in Ca(NO3 )2 –MeOH, one at has a shift of 16.9 ppm in the model peptides and
the resonant frequency expected (Tonelli 10 ) for a was assigned to the random coil conformation. The
randomly coiling alanine residue and a very minor alanine C|O carbon shift is 173.7 ppm. The a-
resonance at the frequency expected for an alanine carbon and C|O of glycine are shifted to 42.1 and
methyl carbon in a b-sheet conformation. Light scat- 167.3 ppm, respectively, in the model peptides.
tering observation of silk fibroin in Ca(NO3 )2 – Table II shows the 13C chemical shifts of fibroin
MeOH should resolve this question. in salt solution. The 13C-nmr spectra of fibroin in
salt solution (Figure 3) shows that the Cb of alanine
Determination of Secondary Structures has a strong peak at 17.3 ppm and a weak peak at
in Solutions via NMR 19.5 ppm. The model peptides in the same salt solu-
13
C-nmr spectra can be used 11 in determining the tion environment as fibroin show a random coil shift
conformations of fibroin in salt solution as well as for Ala Cb at 16.9 ppm. The C|O peak for alanine
is at 175.1 ppm and for glycine at 171.5 ppm for
fibroin in salt solution. Model peptides have C|O
peaks for alanine and glycine random coil confor-
mations at 173.7 and 167.3 ppm, respectively. The
amide carbonyl in salt solution either coordinates
to the calcium ion or hydrogen bonds to other amide
residues affecting the 13C chemical shifts of the
C|O. Therefore, these chemical shifts were not
used to identify the type of secondary structure in
solution. Rather, the Ala Cb chemical shifts were
utilized to determine the secondary structures/local
conformation of the dissolved fibroin.
Fibroin in salt solution shows some conforma-
tional order within the chains. There are two distinct
homogeneous conformational environments for the
alanine residues corresponding to chemical shifts
of 17.3 (s) and 19.5 (w) ppm. There is either no
interconversion between the two populations or
there is slow interconversion on the nmr time scale.
There is definitive evidence of a random coil confor-
mation in solution. The strong peak for the alanine
Cb at 17.3 ppm is close to the random coil confor-
mation chemical shift observed for model peptides
at 16.9 ppm. It has been pointed out 10,12 that the
Cb carbon resonates in this order with increasing
magnetic field: b-sheets r random coil r a-helix.
The minor peak at 19.5 ppm observed for the ala-
nine Cb is approximating the chemical shift pre-
dicted for the b-sheet conformation of the alanine
Cb . Therefore, these solutions may have ordered
structures similar to b-sheets present in small
amounts. According to Iizuka and Yang, 13 silk fi-
broin can adopt a b conformation in solution with-
out precipitating due to intermolecular hydrogen
FIGURE 3 13C-nmr spectra of (a) fibroin–salt solu- bonding to a solvent like methanol. This may be
tion and (b) dialyzed fibroin solution (ppm vs TMS). occurring in the calcium nitrate–methanol system

/ 8k24$$5440 05-16-97 [Link] bpa W: Biopolymers

 Dissolution and Characterization of Silk Fibroin 67

Table III Comparison of FTIR Frequencies (cm01) for the Regenerated Films and the Untreated Silk Fibroin

Untreated Silk Fibroin 2 Fibroin 3 Fibroin 4

Amide I
C|O 1704 1697–1688 vs 1700–1626 1700
Stretching 1639 1658
1618 1629
Amide II
N-H 1560 1570–1564 vs 1549–1514 1540–1515
Deformation 1539
1515
Amide III
Vibrations involving 1260 m 1238 s 1263 m 1263 m
O{C{N and N{H 1230 s 1234 s 1234 s
Amide V 699 m 662 m 689 m 706 w
628 m 689 m

also. It can be deduced that fibroin is mainly random 307C temperature data is odd because variation in
coil but regions of order ( b conformation) are oc- flow times of fibroin solutions at this temperature
curing simultaneously and are stable on the nmr were greater compared to the other two tempera-
time scale of microseconds. tures. This was a reproducible effect. There may
The 13C spectrum for the dialyzed solution (Fig- be certain transitions due to weak bonds occurring
ure 3) shows a single Ala Cb chemical shift. Since within the fibroin chains at this critical temperature.
the dialyzed solution spectrum shows a single well- Also, the activation energy for viscous flow of
developed peak at 17.1, it can be inferred that this the fibroin solution determined from the temperature
is an average of conformations. The molecules in dependence of the absolute viscosity ( DHvis Å 9.03
solution are in constant motion and do not adopt a kcal/mol) is 29% greater than the activation energy
single conformation. In other words, a random coil of flow for the solvent. This behavior further
conformation (model peptide, 16.9 ppm) may be strengthens the suggestion that the fibroin chains
indicated for fibroin in the dialyzed solution. Dia- are at least partially aggregating in solution.
lyzed solutions consist of fibroin chains in a pre- The predicted unperturbed characteristic ratio for
dominant water environment. The C|O peaks for random coil fibroin is equal to 2.37, which is close
alanine and glycine are observed at 176.3 and 172.1 to the free rotation value of 1.93.7 This shows that
ppm, respectively, for the dialyzed solution. the unperturbed chains are nearly freely rotating and
very flexible. Therefore, any ordered structure will
increase the dimensions. The perturbed characteris-
Aggregation in Calcium Nitrate– tic ratio of fibroin in dilute solution at 257C is 14.1,
Methanol Solutions of Silk Fibroin resulting in an expansion factor of a Å 2.44. The
It is evident from viscosity data that fibroin in high value of the expansion factor may be attributed
Ca(NO3 )2 –MeOH has unique characteristics. The to the aggregation of chains or ordered secondary
nmr spectra indicate a predominantly random coil structures such as intramolecular b-sheets in solu-
conformation. Asakura 14 presented evidence to con- tion.
tradict the presence of a-helices in aqueous solu- Aggregation may be caused by the hydrophobic
tions of fibroin. He indicated that aggregation of effect, the principal driving force in protein folding.
fibroin chains in the random coil conformation gives The inclination of hydrophobic groups to adhere to
the appearance of an a-helix. each other is called the hydrophobic effect. Since
The high values of intrinsic viscosities may indi- fibroin consists mostly of amino acids with hy-
cate the presence of aggregation in solution. Also, drophobic side chains, this effect will play a strong
the increase in flow times over consecutive readings role in determining its conformation. The nonpolar
indicates the presence of aggregates that are dis- side chains of alanine (a methyl and a hydrogen)
turbed by shear. The Huggins plot at 307C gives an and glycine (two hydrogens) are driven into the
intrinsic viscosity higher than at 25 and 407C. The interior of the protein while the amide linkages form

/ 8k24$$5440 05-16-97 [Link] bpa W: Biopolymers

68 Mathur et al.

hydrogen bonds on the outside. In particular, a

unique environment is created with methanol as a
hydrophobic solvent and by the positively charged
calcium ions. Methanol molecules are able to direct
the hydrophobic side chains away from the solvent
while the calcium ions bind to the carbonyl oxygen.
The binding of calcium in such a manner accounts
for the need to have a high concentration of calcium
nitrate in solution. In order to break the already
existing hydrogen bonds and stable secondary struc-
tures, it is necessary to have abundant calcium.
The sections of the protein that form hydrogen
bonds may be forced into the hydrophobic interior.
These then associate with each other via hydrogen
bonding, leading to the formation of secondary
structures. Therefore, the hydrophobic effect causes

FIGURE 5 The x-ray diffraction of regenerated films

compared to original silk: (a) original silk, (b) fibroin 2,
(c) fibroin 3, and (d) fibroin 4.

not only the association of hydrophobic groups but

also the stabilization of secondary structures such
as the b-sheets. It can be concluded that fibroin in
solution has favorable conditions that lead to sec-
ondary order and aggregation.15

Regenerated Films
FTIR Analysis. There are four types of distinguish-
able vibrational peaks associated with the amide
groups in proteins (Table III). Amide I and amide
II bands are due to C|O stretching and N{H
deformation, respectively. The frequencies at which
these vibrations occur is affected by the conforma-
tion of the molecule. Silk can acquire a b-sheet,
random coil, and silk I or silk II conformations.
Bhatt and Ahirrao 16 reported two types of peaks
FIGURE 4 FTIR spectra of regenerated films com- associated with the amide I vibrations that were
pared to original silk: (a) original silk, (b) fibroin 2, (c) obtained from regenerated silk films. A peak at 1660
fibroin 3, and (d) fibroin 4. cm01 is due to the random coil nature of the mole-

/ 8k24$$5440 05-16-97 [Link] bpa W: Biopolymers

 Dissolution and Characterization of Silk Fibroin 69

13
Table IV C Chemical Shifts of Regenerated Films and Original Silk in ppm vs TMS

Untreated Silk Fibroin 2 Fibroin 3 Fibroin 4

Glycine
C|O 169.45 169.89
Ca 42.42 42.64 42.79 42.71
Alanine
C|O 171.35 171.35 171.86 171.21
Ca 48.86 49.45 49.30 49.01
Cb 16.14 w 16.50 s 16.65 s 16.72 m
19.87 s 18.85 m 20.02 m 19.65 s
Serine
Ca 54.50 55.23 54.36
Cb 63.29 60.21 61.09 63.80

cule and the band at 1630 cm01 is due to b-sheet rao, 16 Muller et al.17 and Tsukada.18 Two peaks ap-
conformation. FTIR analysis of the untreated silk pear in this region. The peak at the lower frequency
(Figure 4) shows three significant peaks in the am- is due to the amorphous character in this case. Un-
ide I region at 1704, 1639, and 1618 cm01 . The treated silk shows two peaks at 1230 and 1260 cm01
peak at 1639 cm01 has the maximum intensity; for this band. The formation of b structures in silk
therefore, the amide I designation is assigned to this lead to crystallinity. The peak at 1260 cm01 is due
peak, indicating the untreated silk appears to be to crystallinity in silk.
mostly in the b-sheet conformation. The fibroin 2 The FTIR spectra for regenerated silk films have
films cast from the dialyzed solutions also show an absorbance at 1238 cm01 that can be attributed
three peaks in the amide I region (Figure 4). The to the amide III vibration. This peak is as broad as
intensity of the peaks are not the same as in un- the combination of 1230 and 1260 cm01 peaks, but
treated silk and the additive effect of the peaks is a distinct vibration for the crystalline regions is not
different. The peak of maximum intensity is broad present. The shift of the amorphous peak toward
and ranges from 1697 to 1688 cm01 . The other two the crystalline peak shows that some b structure
peaks with lower intensities are at 1655 and 1640– might be present.
1630 cm01 . Untreated silk has the amide V band at 699 cm01 ,
Vibrations in the regenerated film point not to which is due to crystallinity as reported by Bhatt and
the existence of b-sheets but to silk I or random Ahirrao.16 This peak for the regenerated film is at 660
coil conformations. It is evident from Bhatt and cm01 corresponding to the amorphous regions.
Ahirrao 16 and Muller’s 17 work that the existence of According to Bellamy, 19 the methyl symmetrical
more regions of b-sheets in the molecule shifts the deformation peak occurs at 1380 cm01 while the
peaks to a lower frequency as compared to the ran- methylene vibration occurs at 1408–1412 cm01
dom coil for the C|O stretching band. when it is adjacent to a carbonyl unit. There is a
The amide II band is due to N{H deformation. very sharp peak present at a frequency of 1385 cm01
For the regenerated films (Figure 4), the peaks with and an extremely weak peak at 1406 cm01 in un-
maximum intensities range from 1570 to 1564 treated silk. In regenerated films, the two peaks have
cm01 . This peak has the same intensity as the maxi- an equivalent medium intensity and occur at 1381
mum intensity of the amide I peak at 1639 cm01 . and 1412 cm01 . The methyl deformation is due to
There are other peaks in the amide II region that alanine and the methylene vibration due to the gly-
have lower intensities. The amide II peak is also cine units. The conformation of fibroin apparently
very broad due to the overlap of these smaller peaks influences these two vibrations. Orientation of the
leading to a frequency of 1508 cm01 at maximum molecule is also a factor. Bellamy 19 suggested that
intensity. Untreated silk has a broad peak ranging a peak at 1385 cm01 occurs due to the ‘‘angular’’
from 1560 to 1516 cm01 with no distinct maximum methyls. Therefore, the peak at 1385 cm01 shows
except for an elevated region at 1516 cm01 . the presence of orientation and crystallization.
The amide III band corresponding to vibrations There are other areas in the spectra that differ in
of O{C{N and NH bonds occurs in the 1230– untreated silk and regenerated films (Figure 4).
1260 cm01 region as reported by Bhatt and Ahir- Crystallinity can be induced in the regenerated

/ 8k24$$5440 05-16-97 [Link] bpa W: Biopolymers

70 Mathur et al.

associated with this peak, it changes shape accord-

ingly. In untreated silk the peak is extremely sharp,
in regenerated film it is medium and broad, and in
fibroin 3 there is no peak but an enormous shoulder.
The methylene vibration coupled with the methyl
symmetrical vibration are present as a medium peak
at 1410 cm01 like regenerated film, but unlike un-
treated silk.
Further treatment of regenerated films produce
more structural effects. Fibroin 3 can be immersed
in water for a few minutes to change molecular
conformations. This film is labeled as fibroin 4 (Fig-
ure 4). It can be deduced from the amide I, II, III,
and V peaks that this film is similar to untreated
silk in the nature of peaks, but their frequencies
vary somewhat.
X-Ray Diffraction of Regenerated Films and Orig-
inal Silk. The lattice or d spacings were obtained
for fibroin 2, 3, and 4. Fibroin 2, as-cast regenerated

FIGURE 6 13C-nmr spectra in ppm vs TMS of regen-

erated films compared to original silk: (a) original silk,
(b) fibroin 2, (c) fibroin 3, and (d) fibroin 4.

films by immersion in methanol. These films were

labeled fibroin 3 (Figure 4). Several sharp peaks
are present in the amide I region in these films. The
maximum intensity peak, at 1634 cm01 , is close to
the value of untreated silk at 1639 cm01 . The addi-
tive effect evidenced in untreated silk is not present
in fibroin 3. Amide II peaks are also sharp at their
individual positions but are overlapping in the range
from 1549 to 1514 cm01 .
Amide III peaks at 1233 and 1263 cm01 show
that fibroin 3 has crystalline and amorphous regions.
Amide V occurs at 690 cm01 , which is closer to
the crystalline absorbance of 699 cm01 .
The methyl symmetrical vibration of fibroin 3 is
different from the regenerated film and untreated
fibroin 2. There is a massive shoulder instead of
a peak for this vibration unlike untreated silk and
regenerated fibroin 2. The flat shoulder ranges from FIGURE 7 SEM photomicrograph of as-cast film: fi-
1382 to 1388 cm01 . Due to the orientation factor broin 2.

/ 8k24$$5440 05-16-97 [Link] bpa W: Biopolymers

 Dissolution and Characterization of Silk Fibroin 71

d spacing 4.29 Å. There are crystalline regions in

original silk that have an identical orientation, and are
alternating with amorphous regions, which accounts
for the appearance of the diffractogram.

NMR Analysis of Regenerated Films. Secondary

structures in regenerated films can be determined by
13
C-nmr because the chemical shifts are dependent on
the conformations of molecules on a local level. A
solid state nmr analysis was done on the films and
the original silk. Table IV contains a listing of the
chemical shifts and Figures 6 shows the labeled spec-
tra pertaining to the films and original silk fiber.
The 13C-nmr spectra of original silk fiber has two
peaks in the C|O region at 171.35 and 169.45 ppm
that correspond to alanine and glycine, respectively.
The Cb of alanine gives a good indication 10,12 of the

FIGURE 8 SEM photomicrograph of as-cast film

treated with methanol: fibroin 3.

film, has low intensity peaks (Figure 5) at 4.042,

4.985, 5.419, and 7.551 Å. These indicate that there
are small crystallites in this film; however no well-
developed crystalline structure is present. Fibroin 3,
fibroin 2 treated with methanol, has one single high
intensity peak (Figure 5) at 4.042 Å. Apparently,
methanol induces crystallinity in the films, which
is evident from the x-ray diffraction. Fibroin 4, fi-
broin 3 hydrated (Figure 5), shows an increase in
the intensity of the peak at 4.022 Å. Two other
weak peaks appear at 5.111 and 8.164 Å. There is
a reorganization of crystalline regions, in this case
due to hydration. Asakura et al.20 reported a weak
peak at 4.02 Å and a medium-weak peak at 2.44 Å
for the silk I form.
The x-ray diffraction of original silk (Figure 5)
does not have a strong peak like fibroin 3 or 4, but FIGURE 9 SEM photomicrograph of stretched fibroin
an overall increase in intensity around 2Q of 20.687, 3 film.

/ 8k24$$5440 05-16-97 [Link] bpa W: Biopolymers

72 Mathur et al.

FIGURE 10 Photomicrograph of regenerated fiber under the polarizing light microscope.

conformations in the fibroin molecules. In the case fibroin 2 and fibroin 3. This film, like the untreated
of original silk, there are two peaks in this region silk fibroin, has more b-sheet structure than either
at 19.87 and 16.14 ppm. The peak at 19.87 ppm is fibroin 2 or 3.
stronger and is attributed to the b-sheet conforma-
tion.21 The peak at 16.14 ppm is weak and may be SEM of Regenerated Films. There are significant
attributed to random coil structure (16.9 ppm for morphological differences between as cast films and
model peptides in solution see Table II). Therefore, films treated with methanol that are evident under
most of the original silk is in b-sheet conformation the scanning electron microscope. Figure 7 is the
while some is in the random coil conformation. SEM of fibroin 2. There appear to be cracks in the
As-cast film (fibroin 2) has a single peak in the film. The film is structurally intact and stable in
carbonyl region at 171.35 ppm. This is a broad and appearance, but at a magnification of 1250 the
very high intensity peak and therefore incorporates cracks become visible. This may be due to the way
both alanine and glycine carbonyls. Simmons et al.21 this film was dried. Immersion of these films in
observed a single peak for the C|O in the dragline methanol healed the cracks, which is evident in Fig-
silk of N. clavipes at 172.3 ppm and assigned the ure 8. The chains of fibroin oriented so as to heal
same peak to both alanine and glycine carbonyls. the cracks. Bhatt and Ahirrao 16 observed a similar
The Cb of alanine has a strong peak at 16.50 ppm structure for LiSCN films and called it paralleled
and a medium peak at 18.85 ppm. Thus fibroin mol- striations formed due to the alignment of the mi-
ecules in as cast film have either primarily random crofibrils in a certain direction.
coil conformation or an average of random coil and Figure 9 shows the SEM of regenerated films
a-helix, but b-sheets are also present.21 removed with methanol from the petri dish and then
Fibroin 3 is the as-cast film treated with methanol stretched. The lines along which annealing occurred
for a few minutes. A single strong carbonyl peak is are from top right to bottom left. The lobe-like struc-
observed at 171.86 ppm. The Cb of alanine has a tures are aligned in the direction in which the stress
medium shoulder at 20.02 ppm and a strong peak at was applied. Figure 9 gives an indication of the
16.65 ppm indicating more random coil than b-sheets. damage done by the applied stress, and the cohe-
Fibroin 4 is fibroin 3 immersed in water for a siveness in the structure despite the damage. The
few minutes and then air dried. Two peaks are ob- films seem to have a layered structure. While the
served for C|O at 171.21 and 169.89 ppm corre- top layers were elongated and then fell apart, the
sponding to alanine and glycine, respectively. This bottom layers contribute to maintaining the integrity
film resembles original silk in terms of local confor- of the structure. Methanol is able to penetrate only
mations as indicated by the resonances observed for partially through the surface of the film. Annealing
the Ala Cb . Two peaks for Cb are observed at 19.65 of the films in layers can produce partial crystalline
and 16.72 ppm, with the former being the stronger and amorphous regions.
one. The intensities of these peaks are reversed in Figure 10 shows a fiber pulled from the fibroin–

/ 8k24$$5440 05-16-97 [Link] bpa W: Biopolymers

 Dissolution and Characterization of Silk Fibroin 73

FIGURE 11 Photomicrograph of regenerated fiber under the polarizing light microscope.

salt solution, coagulated with acetone and then from nmr and FTIR of the original silk fibroin that
treated with methanol. There is a skin and core mor- the polymer chains are aligned and have a confor-
phology of this fiber. The core is crystalline (posi- mation that is predominantly b-sheet. Calcium ni-
tively birefringent) and the skin is amorphous. Fig- trate–methanol solution is able to penetrate through
ure 11 is a fiber that is larger in diameter than the this structurally sound material and change its con-
fiber in Fig. 10 and shows the fibrillar nature of formation. The solution behavior of silk fibroin in
fibroin. There are also kink bands in this fiber, which calcium nitrate–methanol solution was character-
are characteristic of highly oriented fibers. Note also ized. Conformational analyses of regenerated films
the dark inclusions present in this fiber which are were also performed.
likely complexes of the calcium salt. The Ca(NO3 )2 –MeOH system dissolves fibroin
under endothermic conditions. The dissolution pro-
cess requires both thermal and mechanical energy.
Conclusions
Once this energy barrier is broken fibroin is rela-
Fibroin is a structurally stable material that requires tively stable in solution. However, dilute solution
appropriate conditions for dissolution. It is evident viscosity data indicate that the fibroin chains aggre-

/ 8k24$$5440 05-16-97 [Link] bpa W: Biopolymers

74 Mathur et al.

gate in the salt solution. Calcium is thought to form REFERENCES

a complex with the protein chains at the amide link- 1. Lock, R. L. (1992) U.S. Patent 5,171,505.
ages. 13C-nmr shows that fibroin chains exist in two 2. Uy, W. C. (1993) U.S. Patent 5,252,277.
independent environments. While most of the mole- 3. Lock, R. L. (1993) U.S. Patent 5,252,285.
cules are in the random coil conformation, there is 4. Hudson, S. M., Cuculo, J. A. & Wadsworth, L. C.
evidence that some order exists within the aggre- (1983) J. Polym. Sci. Part A: Polym. Chem. Ed. 21,
gated chains. High intrinsic viscosity values, 8.7 651–670.
(257C), 10.7 (307C), and 8.9 (407C), and the large 5. Flory, P. J. (1953) Principles of Polymer Chemistry,
Cornell University Press, New York.
perturbed dimension values are likely due to aggre-
6. Kaplan, D., Adams, W., Farmer, B. & Viney, C.
gation. The heat of viscous flow DHvis is 9.03 kcal/ (1994) in Silk Polymers: Material Science and Bio-
mol, is high compared to the DHvis Å 7.01 kcal/ technology, Kaplan, D., Adams, W., Farmer, B. &
mol for the solvent. Chain entanglements may be Viney, C., Eds., ACS Symposium Series 544, Ameri-
hindering the free motion of chains and thus increas- can Chemical Society, Washington, D.C., pp. 1–16.
ing the energy required for the viscous flow, and 7. Flory, P. J. (1969) Statistical Mechanics of Chain
aggregation is likely producing more transient phys- Molecules, Interscience Publishers, New York.
ical cross-links between chains. 8. Jackson, C. & O’Brien, J. P. (1995) Macromolecules
28, 5975.
In as-cast regenerated films, the fibroin chains 9. Tonelli, A. E. & Flory, P. J. (1969) Macromolecules
are predominantly random coil/ a-helix conforma- 2, 225–227.
tion with some b-sheet content. The amount of b- 10. Tonelli, A. (1980) J. Am. Chem. Soc. 102, 7635–
sheet can be increased by immersion in methanol 7637.
and then in water. 13C-nmr shows that fibroin 4, film 11. Tonelli, A. E. (1989) NMR Spectroscopy and Poly-
immersed in methanol, and then water has structural mer Microstructure: The Conformational Connec-
characteristics closest to the original silk fibroin fi- tion, VCH, New York.
12. Tonelli, A. E. (1984) Biopolymers 23, 819–824.
ber. The orientation of the chains is probably not
13. Iizuka, E. & Yang, J. T. (1984) Biochemistry 7,
the same, as indicated by our interpretation of FTIR 2218–2228.
and x-ray results. 14. Asakura, T. (1986) Makromol. Chem., Rapid Com-
The macroscopic morphology of the films was mun. 7, 755–759.
assessed using SEM and the polarizing microscope. 15. Voet, D. & Voet, J. G. (1990) Biochemistry, John
Fibroin fibers are fibrillar in nature. The fibrils align Wiley & Sons, New York.
best with the action of acetone and methanol. When 16. Bhatt, N. V. & Ahirrao, S. M. (1983) J. Polym. Sci.:
Part A: Polym. Chem. Ed. 21, 1273–1280.
coagulating fibers in acetone and methanol, it is
17. Muller, W. S., Samuelson, L., Fossey, S. & Kaplan,
important that sufficient coagulation time is allotted. D. (1994) in Silk Polymers: Material Science and
Black regions observed in the fibers by polarizing Biotechnology, Muller, W. S., Samuelson, L.,
microscopy are complexes of calcium salts with the Fossey, S. & Kaplan, D., Eds., ACS Symposium Se-
fibroin, which are entrapped within the chains dur- ries 544, American Chemical Society, Washington
ing coagulation. Fibers that were not immersed in D.C., pp. 342–352.
methanol after coagulation in acetone also had re- 18. Tsukada, M. (1994) J. Polym. Sci.: Part B: Polym.
maining salt. The fibrillar nature of fibroin is evident Phys. 32, 961–968.
19. Bellamy, L. J. (1964) The Infra-red Spectra of Com-
in the regenerated fibers as well.
plex Molecules, John Wiley & Sons, New York.
20. Asakura, T., Kuzuhara, A., Tabeta, R. & Saito, H.
(1985) Macromolecules 18, 1841–1845.
The support of the U.S. Army, Grant No. DAAK60-93- 21. Simmons, A., Ray, E. & Jelinski, L. W. (1994) Mac-
K-0016 is gratefully acknowledged. romolecules 27, 5235–5237.

/ 8k24$$5440 05-16-97 [Link] bpa W: Biopolymers