Journal of the Society of Leather Technologists and Chemists, Vol. 90 p.

23

PHYSICOCHEMICAL PROPERTIES OF COLLAGEN,

GELATIN AND COLLAGEN HYDROLYSATE DERIVED
FROM BOVINE LIMED SPLIT WASTES
ZHONGKAI ZHANG, GUOYING LI* and BI SHI
The Key Laboratory of Leather Chemistry and Engineering of Ministry of Education, Sichuan University, Chengdu, P. R.
China, 610065

Abstract
Collagen, gelatin and collagen hydrolysate were prepared from bovine limed split wastes by different preparative
processes. Sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE) analysis showed that the
molecular weight distribution of collagen was very narrow (about 200 and 100kDa for β and α chains
respectively) compared with those of gelatin (less than 300kDa and wide distribution) and collagen hydrolysate
(less than 50kDa and wide distribution ).
The isoelectric points of collagen, gelatin and collagen hydrolysate were 8.26, 4.88 and 4.54 respectively
determined by Zeta potential titration. Circular dichroism (CD) spectra revealed that there were two peaks, a
positive peak around 221nm and a negative peak around 192nm for collagen, which are the characteristics of
collagen triple helix. However, gelatin and collagen hydrolysate lacked any positive peaks around 220nm,
suggesting random coils. The denaturation temperature of collagen was about 37.5°C determined by the viscosity
method, the helix-coil transitions for gelatin and collagen hydrolysate were not present in the heating process.
Collagen reaggregated to fibrils at 35°C monitored at 313nm. In contrast, gelatin and collagen hydrolysate lost
the ability of fibril formation. Collagen was more resistant to trypsin hydrolysis compared with gelatin and
collagen hydrolysate. In addition, the collagen membrane exhibited superior features such as higher enthalpy,
greater network structure and better physical/mechanical properties compared with those of the gelatin
membrane.
Therefore, collagen isolated from limed split wastes can be a high value product due to its special characteristics
and has many potential future applications in biomaterials, functional additives, cosmetics and pharmaceutical
industries.

Introduction In this work, the different physicochemical properties of

collagen, gelatin and collagen hydrolysate derived from
Nowadays, approximately 170 million pieces of all
bovine limed split wastes by different preparative processes
kinds of raw hides are converted into leather in China and,
will be compared in order to better utilize them as potential
in the leather making processes, about 1.45 million tons of
biomaterials or additives.
leather wastes - including about 0.45 million tons of limed

split wastes are simultaneously produced. Limed split
wastes are less contaminated by chemicals than tanned
leather or finished leather wastes, and can be a source of
high value products such as collagen, gelatin and collagen
hydrolysate.
Collagen is an important biomaterial in medical
applications due to its special characteristics, such as
biodegradability and weak antigenecity.1 Thus collagen, as
a new type of biomaterial, has been used in drug delivery
systems2 and tissue engineering.3 In addition, there are
some intrinsic relationships between collagen and many
diseases such as rheumatoid arthritis4 and systemic
sclerosis.5 Gelatin can be made into roll film and drug
capsules and can also be used as a biomaterial in
biomedical applications, such as in drug delivery systems.6
which are very different from the traditional capsule.
Collagen hydrolysate is a polypeptide composite made by
further hydrolysis of denatured collagen. It has been used
in cosmetics and as a food additive.7-9 Many attempts have
been made to utilize the limed split wastes in ways other
than utilization as the source of commercial gelatins. It had
been shown that the collagen extracted from calf limed
splits at a low temperature by the pepsin-digestion method
had properties similar to those of the commercial
collagen.10-11 In addition, collagen polypeptide had also
been prepared from the limed hide offal in a two-step
protease (EA and EA 537) treatment process at 45-60°C.12
*Corresponding author: E-mail: liguoyings@163.com
Experimental methods
Extraction of pepsin-digested collagen
Bovine limed split waste pieces were fleshed and then
delimed with 2% NH4Cl and 0.5% HCl for 60 min. The
delimed pieces were neutralized to pH6.0-7.0 with 0.5%
HCl followed by rinsing with distilled water. Then they
were cut into smaller pieces and pulverized with a mill
(Fritsch Pulverisette 14, Germany). The powders were
extracted with 30 volumes of 0.5 M acetic acid containing
1% pepsin (1:10 000, calculated on the dry weight of limed
split wastes) at 4°C for 48 hrs. The supernatants of the
extracted solutions were collected by centrifugation at 10
000g for 15 min at 4°C, and then salted out by adding NaCl
to a final concentration of 3 M followed by centrifugation
under the same conditions. The precipitate was dissolved in
0.5 M acetic acid and salted out in 0.7 M NaCl solution.
The precipitate was again dissolved in 0.5 M acetic acid,
and then dialyzed against 0.1 M acetic acid. Collagen
concentration was determined indirectly from the
hydroxyproline concentration, which was analyzed by the
method of Bergman and Loxyley.13
Preparation of gelatin and collagen hydrolysate
Bovine limed split wastes were relimed with 3-4% lime
and 0.5% NaOH for two weeks. The relimed split wastes

23
were washed and neutralized to pH5.5-6 with 1.5% HCl.
Gelatin (type B) was extracted about five times from 70 to
90°C at a 5°C interval. Collagen hydrolysate was prepared
by hydrolysis of gelatin using 0.02% (w/v) trypsin (1:250)
at 40°C for 4 hrs. The solutions were dried using a spray
dryer (Buchi B-191, Switzerland). The concentrations of
the samples were determined by the Biuret method.14
Sodium dodecyl sulphate-polyacrylamide gel
electrophoresis (SDS-PAGE)
Samples were mixed with 0.5 M Tris-HCl buffer (1%
SDS, 10% glycerol and 0.01% bromophenol blue, pH 6.8)
and then boiled for 5 min. 15µl treated samples were
injected into 10% gel wells and run for approx. 120 min.
The gel was stained with 0.25% Coomassie Brilliant Blue
R-250 for 45 min and de-stained with 7.5% acetic acid/5%
methanol solution until the bands were clear.
Isoelectric points of samples
Collagen, gelatin and collagen hydrolysate solutions
(0.05% w/v) were titrated with 0.25 M NaOH or 0.25 M
HCl and the Zeta potentials at a given pH were recorded by
a Zeta potential titration apparatus (Malven Zetaweight
Nano ZS, UK). The titration temperature was 25°C and the
decreasing pH intervals were 0.5 pH. The change of Zeta
potential was plotted against the change of pH for the
samples. The isoelectric points of the samples were
determined at the pH value where the Zeta potential was
zero.
Fibril formation experiment
Samples were dialyzed against phosphate buffer (0.1 M
Na2HPO4, 0.05 M citric acid, pH7.2) in the presence of
NaCl (final concentration adjusted to 0.15 M) at 4°C. The
concentrations of the samples were adjusted to 0.5mg/ml
using the same buffer solution. The samples were incubated
to 35°C to initiate fibril formation, and the turbidities were
recorded at 313nm using a UV spectrometer (Perkin Elmer
Lambda 25, USA). The resulting collagen fibril was
lyophilized in a freeze dryer (Labconco Freeze Dryer
FreeZone 6 Liter, USA) and its morphology was observed
by scanning electron microscopy (SEM) (JEOL JSM-
5900LV, Japan). The SEM specimen was prepared by the
method described by Li.10
Resistance to trypsin degestion
Collagen was dialyzed against the same phosphate
buffer as used for fibril formation without the NaCl at 4°C
for 72 hours. The gelatin and collagen hydrolysate were
directly dissolved in deionised water. All the samples were
digested with 0.04% (w/v) trypsin (1:250) and the degree
of hydrolysis was examined by determining the increasing
amounts of primary amines in the samples. The samples
taken at the given time were added 1% (w/v) SDS and
boiled for 5 min in order to inactivate the enzyme. The
primary amines were determined using the
trinitrobenzensulfonic acid (TNBS) method as described
by Adler-Nissen.15
Circular dichroism (CD)
Collagen was dialyzed against 0.1 M acetic acid solution
using a dialysis tube at 4°C for 72 hrs. The other samples
were dissolved in distilled water. All the samples were
centrifuged at 10 000g for 15 min and adjusted to 0.5
mg/ml before CD analysis. The solution was scanned at the
24
wavelength range from 190 to 250nm at 25°C and its molar
ellipticity[θ] was recorded using a circular dichroism
apparatus (Jasco J-500C, Japan).
Denaturation temperature determined by viscosity method
Collagen (0.02mg/ml) was prepared in the same method
as used for the CD sample. Gelatin and collagen
hydrolysate solutions were adjusted to 8 and 10mg/ml
respectively. All the samples were filtered through a filter
funnel (40-80µm) and degassed by centrifugation. The
denaturation temperature was determined from the
viscosity changes using an Ubbelohde viscosimeter. 15ml
solution was incubated for 15 min at the given temperature
from 25 to 50°C, and then the efflux time (t) was recorded.
The efflux times (t0) of 0.1 M acetic acid solution (collagen
solvent) and water (solvent for gelatin and collagen
hydrolysate) were also determined under the same
conditions. The reduced viscosity (ηsp/c), where ηsp is the
specific viscosity and is calculated by (t-t0)/t0 and c is the
concentration of the samples, was plotted against the
temperature. The denaturation temperature was determined
as the mid-point temperature where viscosity changes
reached 50%.
Preparation of collagen and gelatin membranes
Collagen and gelatin were dried to form membranes at
30°C in silica gel desiccators. The collagen hydrolysate
membrane dried under the above conditions was too fragile
to be used in practice. So only the collagen and the gelatin
membranes were chosen for assay.
Physical properties of membranes
The denaturation temperatures of the collagen and the
gelatin membranes were determined by a differential
scanning calorimetry apparatus (DSC) (Netzsch DSC
200PC, Germany). The membranes were conditioned in a
humidistat containing saturated NH4NO3 solution at 20°C
for 24 hours before test. Approx. 2mg samples were sealed
in aluminum pans and the empty pan was used as a control.
The endothermal curves of the samples were recorded from
10 to 80°C at a heating rate of 5°C/min in a nitrogen
atmosphere. After cooling with liquor nitrogen, secondary
heating was run under the same conditions.
The morphologies of the membranes were examined by
SEM as in the fibril formation experiment. In addition,
their tensile strength and elongation at the break were also
determined by a strength tester (Gaotie GT-AI-7000S,
China).
Results and discussions
SDS-PAGE analysis
The electrophoresis patterns of the samples are shown in
Fig. 1. Collagen displayed one β band (200 kDa) and two
α bands (100 kDa for α1 and α2), which were the unfolding
polypeptide chains of the triple helix ([α1(I)]2[α2(I)]). The
molecular weight of type I collagen was about 300 kDa. In
contrast, the molecular weights of gelatin and collagen
hydrolysate were less than 300 and 50 kDa respectively
and their molecular weight distributions were very wide.
So the components of gelatin and collagen hydrolysate
were more complex than those of collagen. Different
preparative processes lead to different molecular weight
distribution. Collagen was extracted in the acid solution
containing pepsin, which only attacked the non-triple
helical domain of native collagen, whereas gelatin and
collagen hydrolysate were prepared under severe
conditions (above the denaturation temperature). Most of
the triple helices of gelatin and collagen hydrolysate had
been destroyed and parts of their peptide bonds were also
broken out. Therefore, there were wide distributions and

Wavelength (nm)
Figure 2. CD spectra of collagen, gelatin and collagen hydrolysate.

an important role in the clarification of cloudy alcoholic

Figure 1. SDS-PAGE analysis of molecukar weight standards (lane 1),
collagen (lane 2) gelatin (lane3) and collagen hydrolysate (lane 4) on
10% gel.
lower molecular weights for gelatin and collagen
hydrolysate.
Isoelectric points of samples
Isoelectric point is an important parameter of proteins,
which is related to the proportion of acid amino residues
and base amino residues in protein. Gelatin and collagen
hydrolysate were composites of different molecular weight
polypeptides, and thus, the values of isoelectric points were
the values of the system which included the various
polypeptides and buffer. The isoelectric points of collagen,
gelatin and collagen hydrolysate were 8.26, 4.88 and 4.56,
respectively. The isoelectric point of collagen was in the
basic range due to the acidic extraction, which kept side
amide residues intact. In contrast, the isoelectric points of
gelatin and collagen hydrolysate were in the acid range due
to the higher density of carboxyl groups caused by the
hydrolysis of side amide groups of samples under the
strong base and high temperature preparative conditions.
beverages by aggregation of the yeast and other insoluble
particles.19
The triple helix of native collagen can transform to the
random coil configuration when it is heated above the
denaturation temperature. In the denaturation process, the
physical properties such as viscosity, solubility and optical
activity changed due to the collapse of the triple helical
structure. CD spectra and viscosity change are widely used
to determine the denaturation temperature of native
collagen. The denaturation temperatures determined by the
two methods are compared to each other when the protein
concentration and the solvent are same.20 The curves of the
reduced viscosity plotted against the temperature are given
in Fig. 3. The reduced viscosity of collagen decreased at
30°C and reached a plateau at 42.5°C. The denaturation
temperature was approx. 37.5°C. The helix-coil transition
of collagen involves the breakage of hydrogen bonds
between the adjacent polypeptide chains of collagen
molecules in the denaturation process. The intact trimers
(γ) of collagen turn into individual chains (α) or dimers (β)
in the transition, causing the decrease of viscosity at the
same time. In the case of gelatin and collagen hydrolysate,
the reduced viscosity changed slightly with the increase of
temperature, without showing the helix-coil transition. It

Triple helical conformation and helix-coil transition

Collagen is a sort of optically active protein and adopts
the polyproline II-like helical conformation16 with a
negative minimum absorption band around 190nm and a
weak positive maximum absorption band at 210-230nm.
CD spectra of the samples are shown in Figure 2. Collagen
had a positive maximum peak at 221nm and a negative
minimum peak at 192nm, suggesting a typical triple helical
conformation,17 whereas the positive peaks in gelatin and
collagen hydrolysate disappeared, showing the random
coils configuration. However, the triple helix can partly re-
form for gelatin when the temperature of the sample falls
below the gelation temperature.18 The different secondary
structures of the samples can greatly affect some special
properties of the samples. As examples, the triple helical Figure 3. Thermal denaturation curves for collagen, gelatin and collagen
hydrolysate determined by viscosity method.
and rod like structure of collagen as a clarifying agent play

25
was mainly because there were only a few or possibly no
triple helices remaining to be decomposed in the heating
process.
The denaturation temperature of collagen is similar to
the body temperature and the environmental temperature,
and thus the denaturation temperature of mammalian
animal collagen is generally higher than that of marine
animal collagen. Therefore, the application area of
mammalian animal collagen is more extensive than the
marine animal collagen in respect of the thermal stability.

Fibril formation experiment

Native collagen can aggregate into collagen fibrils under
physiological conditions. In the process of fibril formation
in vitro, it can be separated into three steps including
temperature-dependent initiation, temperature-independent
growth and temperature-dependent growth according to the
study of Gelman.21 The fibril formation profiles are shown
in Fig. 4. Collagen solution became turbid with increasing
time, showing fibril formation in the solution. In addition,
the SEM image of the resulting collagen fibrils presented
obvious fibril morphology (Fig. 5). However, the optical
intensities of gelatin and collagen hydrolysate were stable,
suggesting no fibrillogenesis. The collagen fibrillogenesis
involved different collagen types, different extraction
methods and self-assembly conditions (buffer, temperature,
proteins concentration and so on). Different types of
collagens with different triple helical structures can
influence the time course and extent of fibril formation.22
Different preparative processes of the same type collagen
have also different fibrilogenesis behavior and conditions.11
Collagen was extracted in acidic solution containing pepsin
which attacked only the non-triple helical domain of native
collagen. Gelatin was prepared under basic and high
temperature conditions, which hydrolysed the amide
groups and decomposed the triple helices. Collagen
hydrolysate is a denatured hydrolysis product with small
molecular weight and without the triple helical structure.
Therefore, the structures of gelatin and collagen
hydrolysate were disordered and their molecular weight
distributions were very wide, which made fibrillogenesis
impossible. In addition, self-assembly of collagen
monomers into fibrils involves hydrophobic and
electrostatic interactions between collagen chains. The
preparative process reduced the isoelectic points of gelatin
and collagen hydrolysate due to the increase of negative
charges, which also made it difficult for the molecules to
undergo interaction due to the repellency of negative
charges. Thus these products lost the ability of fibril
formation.
Figure 5. SEM image of reconstructed collagen fibrils.
Bar = 5 µm
Samples digested with trypsin
The collagenous domain is hardly digested by enzymes
other than collagenase due to its stable triple helix but the
denatured products such as gelatin and collagen
hydrolysate are easily attacked by proteinases. Trypsin was
used to test the samples for their resistance to enzyme
digestion. The content of primary amines can be used as an
index for the degree of hydrolysis of samples. The trypsin
digestion curves are shown in Fig. 6. The level of primary
amines of collagen increased very little in 6 hrs at 30°C
(below denaturation temperature), showing little
hydrolysis. By contrast, the amount of primary amines in
gelatin increased greatly especially in the first 2 hrs.
Collagen hydrolysate was prepared by the hydrolysis of
gelatin using trypsin. Thus, there were only a few sites
remaining for trypsin to futher attack under these mild
conditions as a consideration of the denaturation of
collagen. Collagen was more resistant to proteinase
hydrolysis than the denatured products (gelatin and
collagen hydrolysate) due to its specific triple helical
structure.
Figure 6. Trypsin hydrolysis curves for collagen, gelatin and collagen
hydrolysate at 30°C.

Physical properties of membranes

Figure 4. Turbidity - time curves for collagen, gelatin and collagen
hydrolysate incubated at 35°C.
The thermo-physical properties of the collagen and the
gelatin membranes were examined by DSC. Fig. 7 shows
the thermograms of the collagen and the gelatin membrane.
The peak temperature of the collagen membrane was about
59°C similar to that of the gelatin membrane (62.5°C), but
its peak area (enthalpy) was markedly larger than that of
the gelatin membrane in the first scan. In the second run,
the endothermal peaks of the collagen and the gelatin
membrane both disappeared and only exhibited a weak step
baseline, showing the glass transition. The thermal
behaviors of collagen and gelatin are related to the moisture
content and the thermal history (preparative conditions).

26
The collagen membrane presented a larger peak area than
that of the gelatin membrane in the denaturation process
due to the greater presence of collagen triple helices. In
addition, the DSC traces in the first scan corresponded to
the helix-coil transitions during the heating of the samples,
while those in the second scan corresponded to the
characteristics of completely amorphous biopolymer due to
the elimination of the triple helical structure. Therefore,
there were no endothermal peaks in the second scan.
B gelatin and collagen hydrolysate, such as high molecular
Figure 7. Thermograms of the collagen membrane A and the gelatin
membrane B determined by DSC.
Figure 8. Surface morphologies of the collagen membrane A and the
gelatin membrane B observed by SEM. Bars 5 µm.
The surface morphologies of the collagen and the gelatin
membrane were greatly different from each other (Fig. 8).
27
There were obvious fibril networks with a rough
membranous structure for the collagen membrane. In
contrast, the gelatin membrane without fibril networks
appeared only as a smooth structure on the surface.
In addition, the tensile strengths and elongations
at the break for the collagen and the gelatin
membrane were 58 ± 5N/mm2, 7 ± 0.5% and 36 ± 4
N/mm2, 3 ± 0.5%, respectively. The differences of the
mechanical properties were caused by the different
structures of the membranes. There were more rigid native
triple helices and fibril networks in the collagen membrane
than in the gelatin membrane, which played important roles
in elevating the mechanical properties of the collagen
membrane.
Conclusions
It was found that there were many different
characteristics of collagen, gelatin and collagen
hydrolysate from the different preparative processes,
although they were all derived from bovine limed split
wastes. Only collagen possessed the triple helical
conformation. It exhibited many properties distinct from
weight and narrow molecular distribution, basic isoelectric
point, triple helix to coil translation in the process of
denaturation, high resistance to protease, and ability of
fibril formation.
Gelatin and collagen hydrolysate were the denatured
products of native collagen and adopted a random coil
conformation, showing different properties, e.g. relative
lower molecular weights and wide molecular distributions,
lower isoelectric points, without conformation translation,
loss of ability of fibril formation, and being easily attacked
by proteinase. In addition, the collagen membrane had a
larger endothermal peak area, a special network structure
and better physical/ mechanical properties compared with
those of the gelatin membrane. The results suggest that
collagen derived from bovine limed split wastes has a
greater potential biomaterial utilization than gelatin and
collagen hydrolysate due to its special physicochemical
properties. Therefore, the extraction of collagen from
bovine limed split wastes is a high value route for the
utilization of leather wastes.
Acknowledgement
This work was supported financially by the National
Natural Science Foundation of China (No. 20576073).
(Received September 2005)
References
1. Lee, C. H., Singla, A. and Lee, Y., Biomedical applications of
collagen. Int. J. Pharm., 2001, 221, 1-22.
2. Friess, W., Collagen - biomaterial for drug delivery. Eur. J. Pharm.
Biopharm., 1998, 45, 113-136.
3. Pachence, J. M., Collagen-Based Devices for Soft Tissue Repair.
J. Biomed. Res. (Applied Biomaterials), 1996, 33, 35-40.
4. Holmdahl, R., Bockermann, R., Backlund, J. et al., The molecular
pathogenesis of collagen-induced arthritis in mice - a model for
rheumatoid arthritis. Ageing Research Reviews, 2002, 1, 135-147.
5. Tamby, M. C., Chanseaud, Y., Guillevin, L. et al., New insights into
the pathogenesis of systemic sclerosis. Autoimmunity Reviews, 2003,
2, 152-157.
6. Einersona, N. J., Stevensa, K. R. and Kao, W. J., Synthesis and
physicochemical analysis of gelatin-based hydrogels for drug carrier
matrices. Biomaterials , 2002, 24, 509-523.
7. Langmaier, F., Mladek, M., Kolomaznik, K. et al., Isolation of elastin
and collagen polypeptides from long cattle tendons as raw material
for the cosmetic industry. Int. J. Cosmet. Sci., 2002, 24, 273-279.
8. Langmaier, F., Mladek, M., Kolomaznik, K. et al., Collagenous
hydrolysates sources of proteins. Int. J. Cosmet. Sci., 2001, 23, 193-199.
9. Morimura, S., Nagata,, H., Uemura, Y. et al., Development of an
effective process for utilization of collagen from livestock and fish
waste. Proc. Biochem., 2002, 37,1403-1412 .
10. Li, G. Y., Fukunaga, S., Takenouchi, K. et al., Physicochemical
properties of collagen isolated from calf limed splits. J. Amer. Leather
Chem. Ass., 2003, 98, 224-229.
11. Li, G. Y., Fukunaga, S., Takenouchi, K. et al., Physioligical and cell
biological properties in vitro of collagen isolated from calf limed
splits. J. Soc. Leather Technol. Chem., 2003, 88, 66-71.
12. Li, Y. C., Zhu, D. Y., Jin, L. Q. et al., Preparation and analysis of
collagen polypeptide from hides by enzymes. J. Soc. Leather Technol.
Chem., 2005, 89, 103-106.
13. Bergman, I. and Loxley, R., Two improved and simplified methods
for the spectrophotometric determination of hydroxyproline. Anal.
Chem., 1965, 35:1961-1965.
14. Gornall, A. G., Bardawill, C. J. and David, M. M., Determination of
Serum proteins by means of the Biuret reaction. J. Biol. Chem., 1949,
177, 751-766.
28
15. Adler-Nissen, J., Determination of the degree of hydrolysis of food
protein hydrolysates by trinitrobenzenesulfonic acid. J. Argric. Food
Chem., 1979, 27, 1256-1262.
16. Linsenmayer, T. F., Collagen. In: Cell Biology of extracellar matrix,
Hay, E.D., ed., 2, 8-9, Plenum, New York, 1991.
17. Engel, J., Folding and unfolding of collagen triple helices. Advances
in Meat Research, 1987, 4, 145-161.
18. Gomez-Guillen, M. C., Turnay, J., Fernandez-Diaz, M. D. et al.,
Structural and physical properties of gelatin extracted from different
marine species: a comparative study. Food Hydrocolloids, 2002, 16,
25-34.
19. Hichman, D., Simes, T. J., Miles, C. A. et al., Isinglass/collagen:
denaturation and functionality. J. Biotechnol., 2000, 79, 245-257.
20. Ogawa, M., Portier, R. J., Moody, M. W. et al., Biochemical
properties of bone and scale collagens isolated from the subtropical
fish black drum (Pogomia cromis) and sheepsheas seabream
(Archosargus probatocephalus). Food Chem., 2004, 88, 495-501.
21. Gelman, R. A., Williams, B. R. and Piez, K. A., Collagen fibril
formation. J. Biol. Chem., 1979, 254, 180-189.
22. Birk, D. E. and Silver, F. H., Collagen fibrillogenesis in
vitro:comparison of type I, II and III. Arch. Biochem. Biophys., 1984,
235, 178-185.