Korean J. Food Sci. An. Vol. 35, No. 4, pp. 431~440 (2015) DOI http://dx.doi.org/10.5851/kosfa.2015.35.4.

431
© 2015 Korean Society for Food Science of Animal Recources ISSN 1225-8563 eISSN 2234-246X

ARTICLE

Optimization of a Multi-Step Procedure for Isolation

of Chicken Bone Collagen
Ümran Cansu and Gökhan Boran*
Yüzüncü Yil University, Department of Food Engineering, 65080 Van, Turkey

Abstract
Chicken bone is not adequately utilized despite its high nutritional value and protein content. Although not a common raw material,
chicken bone can be used in many different ways besides manufacturing of collagen products. In this study, a multi-step procedure was
optimized to isolate chicken bone collagen for higher yield and quality for manufacture of collagen products. The chemical composition
of chicken bone was 2.9% nitrogen corresponding to about 15.6% protein, 9.5% fat, 14.7% mineral and 57.5% moisture. The lowest
amount of protein loss was aimed along with the separation of the highest amount of visible impurities, non-collagen proteins, minerals
and fats. Treatments under optimum conditions removed 57.1% of fats and 87.5% of minerals with respect to their initial concentrations.
Meanwhile, 18.6% of protein and 14.9% of hydroxyproline were lost, suggesting that a selective separation of non-collagen components
and isolation of collagen were achieved. A significant part of impurities were selectively removed and over 80% of the original collagen
was preserved during the treatments.
Keywords: chicken bone, collagen, isolation, optimization
Received February 11, 2015; Revised April 25, 2015; Accepted May 26, 2015

Introduction amount of consumption (USDA, 2011). Chicken process-

ing creates a very large amount of waste, some of which
Collagen is the main fibrous protein in skins, tendons, have high amount of nutritive components. Chicken bone
and bones of animals (Brinckmann, 2005). It is a heavy is one of these wastes, which is mostly not utilized or is
weighted protein and not water soluble due to its hydro- limitedly used in manufacturing of animal feeds, pet
phobic nature (Ergel and Bachinger, 2005). Insoluble col- foods and fertilizers. Some countries prohibit use of ani-
lagen must be pretreated before converted into a soluble mal wastes in feed production and apply legal limitations
form such as gelatin. This is usually done by heating in (Lefferts et al., 2007). Therefore, alternative ways in uti-
water at temperatures higher than 40°C, which is known lization of chicken bone are highly desirable. One of
as shrinking temperature of collagen (Schrieber and Gareis, these alternatives might be use in manufacturing of col-
2007). Collagen products are manufactured from skins lagen products. However, bone is a very complex com-
and bones of animals. The most common raw materials posite material including fibrous collagen surrounded by
used in collagen products are pork skin and bones. Due to an extremely dense material, calcium apatite crystals
objections against to use of pork in collagen products, (Meyers et al., 2008). Not only the minerals but also the
alternative resources gained tremendous attention from fat, as two major components of the bone dry matter be-
the researchers. side the protein, need to be removed prior to the manufac-
Poultry processing by-products might be used as alter- turing of collagen products to obtain the highest yield and
native raw materials in collagen products. Poultry proce- quality. Organic part of the bone, which is mostly protein,
ssing industry showed enormous development in the last may account up to 35% of the bone. About 70% of the
decades. Chicken, as one of the poultry species, is currently organic part is protein and about 90% of this protein is
the number one meat animal in the US because of its collagen, which may account up to 20% of the bone tis-
sue (Hawkins, 2001; Ockerman and Hansen, 1988; Sealy
*Corresponding author: Gökhan Boran, Yüzüncü Yil University, et al., 2014).
Department of Food Engineering, 65080 Van, Turkey. Tel: +90 The objective of this study was to design a multi-step
432 225 1025 - 1150 (ext), E-mail: gboran@yyu.edu.tr procedure for isolation of chicken bone collagen and to

This is an open access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licences/
by-nc/4.0) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.

431
432 Korean J. Food Sci. An., Vol. 35, No. 4 (2015)

optimize the treatment conditions keeping the highest samples were determined according to AOAC (2000) me-
amount of collagen isolated. For this purpose, immersion thods. Protein content was calculated based on the total
of bone samples in hot water (Cleaning, CL) was utilized nitrogen estimated by the Kjeldahl method (AOAC me-
to help in removing impurities including blood and non- thod 984.13), in where a factor of 5.4 was used for nitro-
protein components. Acid treatment (Demineralization, gen-protein conversion as the most abundant protein in
DM) was used to weaken the mineral part of the bone and bone is collagen and the nitrogen content of collagen is
to remove calcium phosphates (Hosseini-Parvar et al., about 18.5% (Muyonga et al., 2004). Fat content was
2009). And, solvent extraction (Degreasing, DG) was determined by using a solvent extraction unit (Ankom
applied for removing fats and grease (Zhang et al., 2010). XT15, USA), where n-hexane was the solvent (Bligh and
These three successive treatments were studied at varying Dyer, 1959). Moisture content was determined based on
temperatures and for varying treatment times to remove the weight difference after drying the samples in an air cir-
non-collagen components selectively. There were 2 inde- culating oven (Mikrotest MKD420, Turkey) at 105°C until
pendent factors at 5 levels for each treatment. And, 2 dep- constant weight was obtained (AOAC method 927.05).
endent variables were selected for evaluation of the corre- Mineral content was determined by incineration of the
sponding treatment. These dependent variables were the samples in a muffle furnace (Nüve MF106, Turkey) at
protein loss (%) for all treatments; the weight loss (%), 550°C for 7 h, which was the typical incineration time de-
the mineral loss (%), and the fat loss (%) for the treat- termined in preliminary studies (AOAC method 942.05).
ments of CL, DM, and DG, respectively.
Treatments of cleaning, demineralization, and de-
Materials and Methods greasing
Frozen bone samples were thawed at 4°C overnight,
Whole tibia bones of broiler chickens (5-6 wk old) were and then the treatments of CL, DM, and DG were carried
used as the starting material. This was the bone in the out successively at different conditions according to the
middle part of a chicken leg called ‘drumstick’ (Fig. 1). study design. The level of independent variables was de-
Bone samples were obtained from a local meat market termined based on the trials and preliminary studies. In-
and brought to the laboratory for further processing. After dependent variables and their levels are given in Table 1.
remaining muscles and ligaments were removed using a For CL treatment, about 50 g (3 pieces) of whole ‘tibia’
knife, bone samples were stored at -18°C until use (within bone was placed into a 500 mL erlenmeyer flask. Then,
30 d) in treatments. All treatments were duplicated and 250 mL (1:5, w/v) of distilled water previously heated to
measurements were done at least in triplicate. All reagents the set temperature was added. After immersion of bone
used were of analytical grade and obtained from Sigma samples in hot water at varying temperatures for different
(USA) and Merck (USA). times, the content of the flask was filtrated through 4 lay-
ers of cheesecloth. Immersion solution was examined for
Chemical composition of chicken bone its protein content by the Biuret method for protein loss
Thawed bone samples were ground using a heavy-duty calculations. The bone residue was washed with tab water
blender (Waring 7011HS, USA) to increase the surface (1:5, w/v) for 3 times, filtrated through 4 layers of cheese-
area before the analyses of chemical composition (Fig. 1). cloth, dried by paper towel to remove excessive water,
Moisture, protein, fat, and mineral content of ground bone and weighted for weight loss calculations.

Fig. 1. Schematically represented location of the drumstick bone ‘tibia’ (a), the actual photo of the bone (b), a bulk of the bones
used in the treatments (c), and the ground bone tissue used for chemical composition analyses (d).
 Isolation of Chicken Bone Collagen 433

Table 1. Independent variables and their levels for the treatments of CL, DM, and DG
Levels
Independent variables
-2 -1 0 +1 +2
Temperature (°C) 50 55 60 65 70
CL
Duration (min) 45 90 135 180 225
HCl Concentration (%) 1 2 3 4 5
DM
Duration of HCl Treatment (h) 6 12 18 24 30
Temperature (°C) 20 25 30 35 40
DG
Duration (h) 6 12 18 24 30
CL, Cleaning; DM, Demineralization; DG, Degreasing.

Bone residue obtained in the CL treatment was subse- tein loss (%) was calculated based on the difference bet-
quently used in the DM treatment by placing into a 500 mL ween the initial protein content of the bone determined by
erlenmeyer flask. Then, a proper amount of HCl (about the Kjeldahl method and the protein content determined
240 mL, 1:5, w/v) at varying concentrations was added. in the extracts after each treatment, which was measured
DM treatment was carried out at ambient temperature (at by the Biuret method. Protein loss was also verified by
24±2°C) for different treatment times according to the measuring the protein concentration of the bone residues
study design (Table 1). Upon completion of the DM treat- by the Kjeldahl method. The verification was not for all
ment, flask content was filtrated through 4 layers of chee- the samples but just for the samples from experimental
secloth. The solution obtained was examined for its pro- runs at center points, to keep the number of experiments
tein content for protein loss calculations. Meanwhile, the feasible. Mineral loss (%) was determined in a similar
bone residue was processed as given for CL treatment and way, based on the difference between the initial mineral
tested for mineral content to determine the mineral loss. content of the original bone and the mineral content of
Bone residue obtained from the DM treatment was used the bone residue after the DM treatment. Mineral loss
in the DG treatment. The bone residue placed into a 500 was also verified by the mineral analysis of the extracts
mL flat bottom volumetric flask was mixed with about after the DM treatment only for those experimental runs
150 mL (1:4, w/v) of n-hexane. The bone residues were at center points. Fat loss (%) was similarly calculated ba-
degreased under continuous shaking at 150 rpm using a sed on the difference between the fat content of the origi-
shaking incubator (Heidolph, Germany) according to the nal bone and the amount of fat separated in the extraction
conditions given in Table 1. Upon completion of the DG solutions after the DG treatment. After removing solvent
treatment, flask content was filtrated through 4 layers of as described, the fat amount separated was gravimetri-
cheese cloth and a small part of the extract was used for cally measured. Fat content of the bone residues after the
the analysis of protein content to determine the protein DG treatment were also measured by the Soxhlet method
loss. For determination of the fat separated, the solvent at center points to verify the amount of the fat loss.
was collected from the extract by using a rotary evapora-
tor (IKA, Germany). Then, excessive solvent was removed Determination of protein and hydroxyproline con-
by holding the extract at 60°C for an hour. Solvent used tent
in DG treatment was changed every 6 h and the fat loss The protein concentration of the extracts was deter-
(%) was calculated accordingly based on the initial amo- mined using the Biuret method as described by Gornall et
unt of bone fat. Throughout the treatments of CL, DM, al. (1949). Bovine serum albumin (BSA) was used as the
and DG; hydroxyproline (HYP) concentration was fol- standard in the range of 0 to 1 mg/mL (Zhou and Regen-
lowed in the extracts along with the protein concentration stein, 2006). The dry BSA powder was corrected for its
of the bone residues, to follow the HYP loss and to verify salt and water content by calibration based on the absor-
the protein loss, respectively. bance of BSA at 280 nm (absorbance of BSA at 280 nm
is 6.66 for a 1% BSA solution) with the absorbance at
Calculation of weight, protein, mineral and fat loss 320 nm subtracted as a background scattering correction
values (Regenstein and Regenstein, 1984).
The weight loss (%) was calculated based on the differ- Hydroxyproline (HYP) content was followed in both the
ence between the initial weight of the original bone and bone residues and the extracts after the treatments of CL,
the bone residues after the corresponding treatments. Pro- DM, and DG at the center points thus, the HYP loss (%)
434 Korean J. Food Sci. An., Vol. 35, No. 4 (2015)

was determined and verified throughout the treatments. 3 treatments. This requires 20 experimental runs for each
HYP content of the bone residues and the extracts was de- treatment, which makes 60 runs in total. The desirability
termined based on the method given by Woessner (1961), function was utilized for a multi-objective optimization
in where L-hydroxyproline was used as the standard. 2 g task to determine optimum levels of independent variables.
of bone sample was used for hydrolysis while the sample
was 2 mL of the extract in case of liquid samples. An ap- Results
propriate amount of HCl was used for samples to obtain a
final concentration of 6 N HCl during hydrolysis. Proper Chemical composition of the tibia bone and the bone
dilutions were prepared after hydrolysis and HYP con- residues obtained after the treatments are given in Table 3.
centration of each sample was calculated based on the ab- According to the results, dry matter of chicken bone is com-
sorbance obtained at 557 nm (Boran and Regenstein, 2009; posed of 3 major components including protein (36.7%),
Woessner, 1961). minerals (34.6%) and fats (22.4%), suggesting that there
are 2 major components to be removed for collagen isola-
Statistical analysis tion, namely fats and minerals. HYP concentration of the
Response surface methodology was used for the design original bone samples was calculated to be 1.35% on wet
and data analysis. Response surface methodology is a ma- weight basis.
thematical modeling technique that relates independent Effects of the treatments on chemical composition and
and dependent variables, and establishes regression mod- HYP concentration are summarized in Table 4. The results
els that describe the interrelations between input parame- showed that minerals were effectively removed while fat
ters and output responses (Yang et al., 2007). The JMP removal was not quite complete considering the cumula-
8.0 statistics software (SAS, USA) was used to analyze tive amount of fat separated. While mineral and fat were
the data obtained to produce analysis of variance (ANOVA) removed, about 19% of the protein and 15% of the HYP
tables, to determine significant regression terms for reg- were lost, suggesting that the fat and the minerals were
ression models, and to draw the surface plots for all dep- selectively separated (Table 4). Although each treatment
endent variables. Sample order was completely random- removed some fat, the CL step was the most effective.
ized as given in Table 2. A 2 factor 5 level 2 center point The cumulative amount of fat separated was over 57% of
central composite design was duplicated separately for all the total present. Mineral separation was more successful

Table 2. Randomized order of experimental runs and the coded levels of independent variables
Cleaning Demineralization Degreasing
Order
Pattern Dur Temp Pattern Dur HCl Con Pattern Dur Temp
1 00 0 0 a0 -2 0 + 1 -1
2 0a 0 -2 A0 2 0 0a 0 -2
3 00 0 0 0A 0 2 + -1 1
4 a0 -2 0 -1 -1 + -1 1
5 -1 -1 a0 -2 0 + 1 -1
6 + 1 -1 -1 -1 -1 -1
7 0A 0 2 + 1 -1 0A 0 2
8 0a 0 -2 00 0 0 A0 2 0
9 + -1 1 0A 0 2 0A 0 2
10 -1 -1 ++ 1 1 a0 -2 0
11 00 0 0 00 0 0 A0 2 0
12 A0 2 0 00 0 0 -1 -1
13 + 1 -1 + 1 -1 00 0 0
14 a0 -2 0 + -1 1 0a 0 -2
15 00 0 0 A0 2 0 00 0 0
16 0A 0 2 ++ 1 1 00 0 0
17 ++ 1 1 00 0 0 a0 -2 0
18 A0 2 0 0a 0 -2 ++ 1 1
19 ++ 1 1 + -1 1 ++ 1 1
20 + -1 1 0a 0 -2 00 0 0
a, -, 0, +, A represent the levels of the variables in the ascending order (Dur: Duration, Temp: Temperature, Con: Concentration).
 Isolation of Chicken Bone Collagen 435

Table 3. Composition of bone and bone residues after treatments of CL, DM, and DG (%)
Total organic Total dry
Sample Moisture Crude fat Crude mineral Crude nitrogen Protein (N×5.4)
matter matter
Chicken Bone 57.51)±0.2 9.5±0.1 14.7±0.4 2.89±0.10 15.6±0.5 27.8 42.5
CL Residue 57.4±0.2 7.2±0.2 15.0±0.3 2.88±0.05 15.5±0.2 27.6 42.6
DM Residue 67.0±0.1 7.5±0.2 2.8±0.2 3.04±0.04 16.4±0.2 30.2 33.0
DG Residue 70.1±0.3 5.1±0.1 2.3±0.1 2.94±0.03 15.9±0.2 27.6 29.9
1)
Values were given based on wet weight of fresh bone and bone residues of each treatment. Values were average values and standard
deviation of triplicate measurements. Values given were for CL, DM and DG residues obtained at optimum treatment conditions.

Table 4. Cumulative loss in macro-nutrients and HYP content during CL, DM, and DG
Weight loss Fat loss Mineral loss Protein loss HYP loss HYP con HYP/Pro ratio
Sample
(%) (%) (%) (%) (%) (%) (%)
Chicken Bone NA NA NA NA NA 1.35 ±0.02 8.66
CL Residue 6.11) 28.9 4.0 6.4 13.8 1.24 ±0.03 7.98
DM Residue 17.9 35.2 84.2 13.4 14.3 1.41 ±0.07 8.58
DG Residue 20.2 57.1 87.5 18.6 14.9 1.44 ±0.06 9.06
1)
Values were average of triplicate measurements at optimum conditions. Loss values were all given based on initial amount of corre-
sponding component and their concentration in bone residues obtained after each treatment. HYP concentration was given based on wet
weight of initial bone or bone residue. HYP/Pro ratio was given based on HYP and protein content of fresh bone and bone residues of
each treatment. NA, Not applicable.

as the cumulative amount of mineral separated was almost

88% of the total present. The DM treatment seemed to be
very effective in removing minerals from the bone. Dur-
ing the treatments, a gradual increase in the relative amo-
unt of HYP and HYP/Pro ratio was observed in the bone
residues, suggesting that non-collagen protein was slightly
removed. Optimization studies on the overall results sho-
wed that optimum treatment conditions were as follows:
immersion of the bone samples in distilled water at 65°C
for 90 min in the CL process, followed by a demineraliza-
tion step using 2% HCl solution for 24 h in the DM pro-
Fig. 2. Optimized process flow for isolation of chicken bone
cess, and finally fat removal using n-hexane at 35°C for collagen.
18 h in the DG process. Schematic process flow of this
isolation procedure is given in Fig. 2.
Experimental results were analyzed using JMP 8.0 soft- The weight loss (%) was a dependent variable only in
ware to obtain the regression models for each dependent the CL treatment and not in successive treatments but it
variable. Predicted results based on these regression mod- was followed throughout the treatments as the loss in pro-
els are given in Table 5 along with the experimental data tein, HYP, mineral and fat was only possible to be calcu-
obtained. Experimental results and response surface plots lated if the weight of the corresponding bone residue is
(Fig. 3) showed that increasing temperature in the CL known after each treatment. With respect to the DM pro-
step increased the weight loss in the bone residues while cess, the levels of both HCl concentration and the treat-
also increasing the amount of protein loss. The effect of ment duration had significant effects on dependent vari-
the duration was limited compared to that of the tempera- ables. The highest amount of mineral was removed aro-
ture in CL treatment (Fig. 3(a) and 3(b)). As minimum und at the middle levels of independent variables. These
protein and maximum weight loss were aimed, the higher middle levels also caused the highest amount of protein
temperatures seemed to be more effective in removing loss (Fig. 3(c) and 3(d)). Therefore, low HCl concentra-
impurities but also responsible for higher protein loss. tion and middle level treatment duration was preferable
Therefore, mild treatment temperatures for relatively short for effective separation of minerals along with limited
duration seemed to be preferable. loss in the protein.
436 Korean J. Food Sci. An., Vol. 35, No. 4 (2015)

Table 5. Experimental and predicted results for dependent variables of CL, DM, and DG
Cleaning Demineralization Degreasing
Weight loss (%) Protein loss (%) Mineral loss (%) Protein loss (%) Fat loss (%) Protein loss (%)
Exp Pre Exp Pre Exp Pre Exp Pre Exp Pre Exp Pre
1 6.751) 6.37 1.89 1.75 64.9 67.3 19.7 19.0 30.4 34.3 0.60 0.69
2 3.56 3.50 1.27 1.15 90.8 90.3 22.0 22.4 16.8 17.4 0.42 0.44
3 6.92 6.37 1.83 1.75 87.6 82.9 24.2 23.0 36.4 38.9 0.63 0.47
4 1.33 0.36 1.07 0.87 71.6 64.1 18.9 19.3 37.7 38.9 0.59 0.47
5 1.89 3.05 0.99 1.27 66.1 67.3 19.9 19.0 31.9 34.3 0.70 0.69
6 7.26 6.71 1.45 1.44 71.2 64.1 19.5 19.3 15.0 12.7 0.45 0.32
7 6.19 5.86 1.81 1.91 83.3 78.4 22.2 21.2 60.6 57.0 0.48 0.45
8 3.66 3.50 1.21 1.15 86.8 87.6 25.2 24.2 50.1 47.1 0.76 0.79
9 2.65 3.78 1.34 1.55 85.8 82.9 24.5 23.0 60.5 57.0 0.33 0.45
10 1.97 3.05 0.97 1.27 88.5 92.8 23.9 24.7 15.9 16.7 0.20 0.33
11 6.83 6.37 1.76 1.75 85.0 87.6 24.8 24.2 52.4 47.1 0.90 0.79
12 7.24 8.58 1.33 1.41 85.9 87.6 24.5 24.2 15.6 12.7 0.40 0.32
13 7.05 6.71 1.49 1.44 84.1 78.4 22.0 21.2 34.1 32.3 0.47 0.52
14 1.42 0.36 1.08 0.87 82.2 84.2 21.0 23.3 16.1 17.4 0.44 0.44
15 6.79 6.37 1.83 1.75 90.1 90.3 22.5 22.4 34.9 32.3 0.53 0.52
16 6.23 5.86 1.98 1.91 89.2 92.8 23.2 24.7 35.6 32.3 0.43 0.52
17 8.89 8.35 2.03 1.91 85.9 87.6 24.7 24.2 14.4 16.7 0.28 0.33
18 8.81 8.58 1.26 1.41 46.2 48.3 14.8 15.6 39.7 47.7 0.63 0.56
19 8.23 8.35 2.06 1.91 82.2 84.2 21.4 23.3 40.6 47.7 0.50 0.56
20 2.18 3.78 1.37 1.55 39.3 48.3 15.1 15.6 34.6 32.3 0.42 0.52
1)
All values of protein, mineral and fat loss were given based on the difference between their concentrations in initial bone sample and in
extraction solutions obtained after each treatment.

Fig. 3. Response surface plots for weight (a) and protein (b) loss of CL, mineral (c) and protein (d) loss of DM, fat (e) and protein
(f) loss of DG treatments.
 Isolation of Chicken Bone Collagen 437

Considering the results of the DG process, the treat- Hydroxyproline is used for estimation of collagen con-
ment duration was effective at high temperatures for both tent as it is a unique iminoacid for collagen (Fratzl, 2008).
fat and protein removal; and longer treatment durations Other than collagen, elastin has some HYP but at lower
caused higher loss in both fat and protein content (Fig. levels compared to collagen (Eastoe and Leach, 1977). It
3(e) and 3(f)). Temperature was very effective as increas- is reported that HYP ratio in proteinaceous material of
ing temperature caused higher protein loss. Increasing bone samples from different species was close to 10%
temperature was also effective in fat removal. Thus, rela- (Herpandi et al., 2011; Li et al., 2009). Therefore, when
tively high temperature and short treatment duration were an arbitrary factor of 10 was used for conversion of HYP
preferable for effective separation of fat along with lim- to collagen, this would conclude that about 90% of
ited loss in the protein. chicken bone protein was collagen, which is consistent
The isolation procedure determined for chicken bone with previous reports for bone tissues from various ani-
collagen was verified by running the procedure at opti- mals. Thus, removing fats and minerals would predomi-
mum conditions given. After the isolation procedure car- nantly leave the collagen isolated in the remaining bone
ried out at optimized conditions, about 82% of the origi- residue as intended.
nal protein and 85% of the initial HYP were preserved in Demineralization is a key process when bone protein is
the final bone residue while over 88% of the initial min- intended to be isolated. In the literature, acid treatment
eral content and about 57% of fats were selectively rem- looks very common and HCl is one of the most common
oved. The final bone residue was about 80% of the origi- inorganic acids used for this purpose (Castro-Ceseña et
nal bone in weight with a higher percentage of protein al., 2011; Figueiredo et al., 2011). In this study, HCl was
and HYP as given in Table 3 and 4. The highest HYP con- used for demineralization of the bone samples at several
centration found in treated bone residues was 1.44% and concentrations for varying treatment durations. HCl was
the highest HYP/Pro ratio was 9.06 as given in Table 4. very effective in demineralization of the bone, removing
about 88% of the initial bone minerals. The amount of
Discussion mineral separated was consistent with previously reported
values (Castro-Ceseña et al., 2011; Figueiredo et al., 2011).
Chemical composition of the bone tissue changes greatly High pressure or hot water treatment is common in the
depending on species, age, diet, nourishment and from literature for fat removal from various bones (Hosseini-
bone to bone of the same animal. Field et al. (1974) rep- Parvar et al., 2009; Nawrocki, 1997). Our preliminary
orted that the age and the species were important factors studies, on the other hand, showed that immersion of the
significantly affecting the composition of animal bone, bone samples as whole in hot water had very limited suc-
i.e., the older the age the higher the fat and the dry matter cess in degreasing. In addition, initial immersion in hot
of the bone. They reported similar fat content for chicken water increases the porosity, which may harm the col-
bone but a bit higher dry matter (49.9%) compared to the lagen. Boiling, on the other hand, is more effective in de-
value (42.5%) reported in this study, probably because greasing of the bones, but may cause extensive and irre-
the bone samples used in this study were from younger versible damage on the collagen, which should be avoided
animals. In another study, chemical composition of the for high yield and quality. Especially when cleaning is
chicken bone was reported 53.2% moisture, 8.4% fats, insufficient, bones may be degreased in organic solvents
20.5% protein (conversion factor for Kjeldahl nitrogen (Guilminot et al., 2014). There are some chemicals used
was 6.25) and 15.9% minerals (Kettawan et al., 2002). for bone degreasing as reported in the literature. For
Although moisture and fat content reported in our study example, Lander et al. (2014) used trichloroethylene at
was similar, protein content was lower most probably due 82°C for 3 d for degreasing human bones after boiling the
to the conversion factor used in this study. As previous samples in distilled water. Guilminot et al. (2014) used
studies reported that over 90% of the bone protein is col- hexane, methyl alcohol, and other organic solvents along
lagen, 5.4 was the factor used for nitrogen-protein con- with other methods including enzymatic treatments and
version in this study to estimate the protein concentration supercritical CO2 extraction to remove fat from whale
of the chicken bone (Brinckmann, 2005; Fratzl, 2008; bones. They concluded that enzymatic methods and super-
Schrieber and Gareis, 2007). However, this might slightly critical CO2 extraction were not much effective in fat re-
underestimate the protein content of the bone as there are moval from the bones. On the other hand, organic sol-
obviously other proteins beside collagen in bone tissue. vents gave the best performance in removing fats from
438 Korean J. Food Sci. An., Vol. 35, No. 4 (2015)

the bones with a reasonable harm on the bone tissue (Guil- responses, based on the settings given in Table 6. R2 val-
minot et al., 2014). ues obtained for each dependent variable were high in
Therefore, in this study, the initial hot water immersion general, indicating that the regression models were suc-
was applied for relatively short durations especially for cessful in describing the treatments. Regression coeffici-
removing color pigments and visible impurities. This ents for all models are given in Table 7. According to the
treatment was followed by the immersion of the resultant regression models, the dependent variables of each treat-
bone residues in HCl solution for removal of minerals, ment can be estimated using the following equations:
which was very successful. During these two successive
treatments, some fat was separated, but still the larger part Wgh L in CL = 6.37 + (2.06 × Temp) + (0.59 × Dur) −
of the fats was in the bone residue. Therefore, a final step (0.47 × Temp2) − (0.42 × Dur2)
of solvent extraction was applied for removal of the Pro L in CL = 1.75 + (0.13 × Temp) + (0.19 × Dur) −
remaining fats. The amount of fats separated was actually (0.15 × Temp2)
calculated based on the amount of fats collected in DG Min L in DM = 87.56 + (8.66 × Con) + (5.73 × Dur) −
treatment, which was about 57% of the initial fat, sug- (5.50 × Con2) − (2.19 × Dur2)
gesting that the greater part of the fats was separated in Pro L in DM = 24.19 + (1.85 × Con) + (0.85 × Dur) −
DG treatment. This was also confirmed by the fat content (1.22 × Con2) − (0.87 × Dur2)
of the bone residue after DG treatment. Fat L in DG = 32.28 + (9.91 × Temp) + (7.60 × Dur)
The experimental data was used for optimization of the Pro L in DG = 0.52 + (0.11 × Dur)
level of independent variables for each treatment. In gen-
eral, the desirability function is used as an indicator of According to the models, the linear terms of all inde-
how closely the goal (that is, minimizing or maximizing pendent variables were significant except the temperature
the response or matching a target value) is achieved by for protein loss in the DG treatment. Besides the linear
the model. The desirability level for each response is set terms, all quadratic terms were significant for both depen-
manually and this affects the overall desirability of the dent variables in the DM process. The quadratic terms
results. The prediction profiler of the JMP software was were also significant in the CL treatment only except the
used to obtain the highest desirability for each response, duration for protein loss in the CL treatment. On the other
the highest overall desirability, the highest values for maxi- hand, the interaction terms were found insignificant in all
mized responses and the lowest values for minimized regression models.

Table 6. Settings of optimization, predicted results and corresponding R2 values of the models
Dependent Variables Low Middle High Predicted results R2 Desirability1)
Weight loss (%, max) 1.3 5.1 8.9 6.71±1.02 0.91
CL 0.592
Protein loss (%, min) 0.9 1.5 2.1 1.4±0.2 0.82
Mineral loss (%, max) 38 78 92 78.4±5.4 0.91
DM 0.442
Protein loss (%, min) 10 20 26 21.2±1.3 0.88
Fat loss (%, max) 10 40 65 42.5±3.5 0.94
DG 0.513
Protein loss (%, min) 0.1 0.5 0.9 0.5±0.1 0.72
1)
Desirability of low, middle, and high levels was set to 0.01, 0.60, and 0.99, respectively, for variables maximized (max: maximized) and
0.99, 0.40, and 0.01, respectively, for variables minimized (min: minimized).

Table 7. Regression coefficients of the models for each treatment

Cleaning Demineralization Degreasing
Term Wgh L Pro L Term Min L Pro L Term Fat L Pro L
Intercept 6.37* 1.75* Intercept 87.56* 24.19* Intercept 32.28* 0.52*
Temp 2.06* 0.13* Con 8.66* 1.85* Temp 9.91* 0.01
Dur 0.59* 0.19* Dur 5.73* 0.85* Dur 7.60* 0.11*
Temp×Temp -0.47* -0.15* Con×Con -5.50* -1.22* Temp×Temp 1.23 -0.02
Dur×Dur -0.42* 0.05 Dur×Dur -2.19* -0.87* Dur×Dur -0.09 0.01
Temp×Dur 0.23 0.05 Con×Dur -1.41 -0.14 Temp×Dur -3.18 -0.07
* denotes significant difference from zero at p<0.05. Temp, Temperature; Dur, Duration; Con, Concentration (HCl); Pro, Protein; L,
Loss; Min, Mineral; Wgh, Weight.
 Isolation of Chicken Bone Collagen 439

The optimum isolation procedure was cleaning at 65°C atin extraction from silver carp skin. J. Food Sci. 74, E432-
for 90 min, the demineralization in 2% HCl solution for E441.
4. Brinckmann, J. (2005) Collagens at a glance. In: Collagen:
24 h, and the degreasing by n-hexane extraction at 35°C
Primer in structure, processing and assembly. Brinckmann,
for 18 h. Under optimum conditions, the multi-step pro-
J., Notbohm, H. and Müller, P. K. (eds) Springer, NY, pp. 1-6.
cedure for isolation of chicken bone collagen resulted in 5. Castro-Ceseña, A. B., Novitskaya, E. E., Chen, P. Y., Hirata,
87.5% mineral and 57.1% fat removal along with 18.6% G. A., and McKittrick, J. (2011) Kinetic studies of bone dem-
protein and 14.9% HYP loss, which suggests that the ineralization at different HCl concentrations and tempera-
chicken bone collagen was selectively separated and iso- tures. Mater. Sci. Eng. C. 31, 523-530.
lated to some extent. Verification of the optimized proce- 6. Eastoe, J. E. and Leach, A. A. (1977) Chemical constitution
of gelatin. In: The science and technology of gelatin. Ward,
dure also confirmed the predicted results. The bone sam-
A. G. and Courts, A. (eds) Academic Press, NY, pp. 73-105.
ples used in the isolation treatments were whole tibia 7. Ergel, J. and Bachinger, H. P. (2005) Structure, stability and
bones. Therefore, optimum conditions reported in this folding of the collagen triple helix. In: Collagen: Primer in
study would not be valid if the bone samples were ground structure, processing and assembly. Brinckmann, J., Not-
to increase the surface area. In that case, milder treat- bohm, H. and Müller, P. K. (eds) Springer, NY, pp. 23-24.
ments might be required for preservation of collagen in the 8. Field, R. A., Riley, M. L., Mello, F. C., Corbridge, J. H., and
Kotula, A. W. (1974) Bone composition in cattle, pigs, sheep
treated residue. Dry matter of the final bone residue was
and poultry. J. Animal Sci. 39, 493-499.
composed of about 53% protein, which might be slightly 9. Figueiredo, M., Cunha, S., Martins, G., Freitas, J., Judas, F.,
underestimated due to the factor used for nitrogen-protein and Figueiredo, H. (2011) Influence of hydrochloric acid con-
conversion. Thus, a reasonable success was achieved with centration on the demineralization of cortical bone. Chem.
respect to the isolation of collagen in chicken bone tissue. Eng. Res. Des. 89, 116-124.
10. Fratzl, P. (2008) Collagen: Structure and mechanics, an intro-
duction. In: Collagen: structure and mechanics. Fratzl, P. (ed),
Conclusion Springer, NY, pp. 1-13.
11. Gornall, A., Bardawill, C. J., and David, M. M. (1949) Deter-
A multi-step procedure was defined and optimized for mination of serum proteins by means of the biuret reaction. J.
isolation of chicken bone collagen. Under optimum con- Biol. Chem. 177, 751-766.
ditions, this procedure resulted in 87.5% mineral and 12. Guilminot, E., Lemoine, G., Pelé, C., Poisson, L., Surbled,
57.1% fat removal along with 18.6% protein and 14.9% M., Louvet, I., Mevellec, J. Y., and Rémy, L. (2014) Re-treat-
ment of whale bones-How to extract degraded fats from
HYP loss, which suggests that the chicken bone collagen
weakened bones? J. Cultural Herit. 15, 128-135.
was effectively isolated. Almost 85% of the initial col- 13. Hawkins, D. (2001) Biomechanics of musculoskeletal tis-
lagen was preserved and isolated in the final bone resi- sues, 106 p, University of California-Davis, CA, USA.
due, while other two major components, namely minerals 14. Herpandi, Huda, N. and Adzitey, F. (2011) Fish bone and
and fats, were selectively removed at reasonable levels. scale as a potential source of halal gelatin. J. Fish. Aquat.
Further studies are needed to characterize the resultant Sci. 6, 379-389.
15. Hosseini-Parvar, S. H., Keramat, J., Kadivar, M., Khanipour,
collagen in terms of its quality and functional properties.
E., and Motamedzadegan, A. (2009) Optimising conditions
for enzymatic extraction of edible gelatin from the cattle
Acknowledgements bones using response surface methodology. Int. J. Food Sci.
Tech. 44, 467-475.
The authors gratefully acknowledge the financial sup- 16. Kettawan, A., Sungpuag, P., Sirichakwal, P. P., and Chavasit,
port from the Scientific Research Fund of Yüzüncü Yil V. (2002) Chicken bone calcium extraction and its application
as food fortificant. J. Nat. Resour. Counc. Thailand 34, 163-
University under the project 2014-FBE-YL034.
180.
17. Lander, S. L., Brits, D., and Hosie, M. (2014) The effects of
References freezing, boiling and degreasing on the microstructure of
bone. HOMO - J. Comparative Human Biol. 65, 131-142.
1. Association of Official American Chemists International 18. Lefferts, L. Y., Kucharski, M., McKenzie, S., and Walker, P.
(AOAC) (2000) Official methods of analysis, 17th Ed., Wil- (2007) Feed for food-producing animals: A resource on ingre-
liam, H. (ed) Washington, DC. dients, the industry, and regulation. 36 p, The Johns Hopkins
2. Bligh, E. G. and Dyer, W. J. (1959) A rapid method for total Center for a Livable Future. Maryland, USA.
lipid extraction and purification. Can. J. Biochem. Phys. 37, 19. Li, F., Jia, D., and Yao, K. (2009) Amino acid composition
911-917. and functional properties of collagen polypeptide from Yak
3. Boran, G. and Regenstein, J. M. (2009) Optimization of gel- (Bos grunniens) bone. LWT-Food Sci. Technol. 42, 945-949.
440 Korean J. Food Sci. An., Vol. 35, No. 4 (2015)

20. Meyers, M. A., Chen, P., Lin, A. Y., and Seki, Y. (2008) Bio- many.
logical materials: Structure and mechanical properties. Prog. 26. United States Department of Agriculture (USDA) (2011) Po-
Mater. Sci. 53, 1-206. ultry 2010. Structure of the U.S. poultry industry, 2010. 40 p.
21. Muyonga, J. H., Cole, C. G. B., and Duodu, K. G. (2004) Ext- Fort Collins, Colorado, USA.
raction and physicochemical characterization of Nile Perch 27. Woessner, J. F. Jr. (1961) The determination of hydroxypro-
(Lates niloticus) skin and bone gelatin. Food Hydrocolloid. line in tissue and protein samples containing small proporti-
18, 581-592. ons of this imino acid. Arch. Biochem. Biophys. 93, 440-447.
22. Nawrocki, S. (1997) Cleaning Bones. University of Indiana- 28. Yang, H., Wang, Y., Jiang, M., Oh, J. H., Herring, J., and Zhou,
polis, Archeology & Forensics Laboratory (http://archlab.ui- P. (2007) 2-step optimization of the extraction and subsequent
ndy.edu), last accessed on 25/04/2015. physical properties of channel catfish (Ictalurus punctatus)
23. Ockerman, H. W. and Hansen, C. L. (1988) Animal by-prod- skin gelatin. J. Food Sci. 72, C188-C195.
uct processing. 366 p, Ellis Horwood Ltd., Chichester, UK. 29. Zhang, G., Yue, X., Fan, A., and Liu, G. (2010) Reutilization
24. Sealy, J., Johnson, M., Richards, M., and Nehlich, O. (2014) of waste chicken bone as nutrients source. 4th International
Comparison of two methods of extracting bone collagen for Conference on Bioinformatics and Biomedical Engineering,
stable carbon and nitrogen isotope analysis: comparing whole 18-20 June 2010, Chengdu, China.
bone demineralization with gelatinization and ultrafiltration. 30. Zhou, P. and Regenstein, J. M. (2006) Determination of total
J. Archaeol. Sci. 47, 64-69. protein content in gelatin solutions with the Lowry or Biuret
25. Schrieber, R. and Gareis, H. (2007) Gelatine handbook: The- assay. J. Food Sci. 71, C474-C479.
ory and industrial practice. 335 p. Wiley, Weinheim, Ger-