Artificial Organs
27(8):701–705, Blackwell Publishing, Inc.
© 2003 International Society for Artificial Organs
A Wound Contraction Experimental Model for Studying
Keloids and Wound-healing Modulators
*Fábio Kamamoto, *Andre Oliveira Paggiaro, †Andrea Rodas, *Marisa Roma Herson,
†Monica Beatriz Mathor, and *Marcus Castro Ferreira
*Laboratório de Cirurgia Plástica, Faculdade de Medicina da Universidade de São Paulo; and †Centro de Tecnologia das
Radiações, Instituto Pesquisas Energéticas e Nucleares da Universidade de São Paulo, São Paulo, Brazil
Abstract: Preventing and treating hypertrophic and keloid
scars is difficult because of the lack of knowledge about
their genesis. Tissue repair can be studied with biocompatible matrices and ex vivo cultures of different cell types. We
used an experimental model where collagen gels populated
by human fibroblasts underwent progressive contraction,
allowing the study of wound healing remodeling. The
fibroblast-populated lattices showed the greater contraction of the gel populated by fibroblasts from keloids
versus fibroblasts from normal skin. Moreover, fibroblast
growth factor (FGF) and transforming growth factor beta
(TGF-b) involved in scar formation were added to the collagen gels populated by normal skin fibroblasts. TGF-b
caused an increase in gel contraction; FGF did not. The
mean percentages of contraction of the gels populated by
keloid fibroblasts were very similar to the percentages
of gels populated by normal skin fibroblasts with added
TGF-b. These observations confirm the existing hypothesis
that TGF-b may be involved in keloid formation.
Key Words: Tissue engineering—Human fibroblasts—
Collagen lattices—fibroblast growth factor—transforming
growth factor beta—Keloid.
Wound healing is a complex phenomenon of vital
importance to human beings. It encompasses an
organized sequence of catabolic and anabolic events
that depend on cellular and biochemical actions. The
search for a better understanding of the wound
healing process has been a constant in medicine. In
the 16th century, Ambrose Pare discovered through
his work that the use of rose oil could accelerate
wound healing (1).
For descriptive purposes, the biological processes
of wound healing can be divided into the following
four phases: hemostasis, inflammation, proliferation,
and remodeling (2). Although these processes are
described as distinct phases, significant overlap
occurs. The initial physiologic response to an injury
is hemostasis. The clotting cascade is activated, and
fibrin polymerization occurs as a mature clot is
formed in conjunction with the aggregated platelets.
Inflammation is carried out mainly by macrophages.
Their importance lies in the secretion of multiple
cytokines and growth factors that mediate various
cellular and biochemical events responsible for
healing. During the proliferative phase of healing,
the migration and proliferation of several cell types
primarily occur, leading directly to the repair of
tissue integrity. As the fibroblasts and vascular
endothelial cells migrate into the provisional matrix,
they begin to proliferate and cellularity of the wound
increases. The remodeling phase is the most important in plastic surgery, because it determines the final
appearance of scars. This appearance depends mainly
on two factors: the reorientation of collagen fibers
and granulation tissue contraction (2). The growth
factors fibroblast growth factor (FGF) and transforming growth factor beta (TGF-b) interfere with
wound contraction. FGF inhibits the production of
collagen by fibroblasts (3,4). TGF-b increases the
expression of different kinds of collagen, and thus it
may play a role in the formation of pathological scars,
as keloids do (5,6). Scars frequently cause tightness,
pain, and itching for the patient, and no efficient
treatment exists for this pathology (7).
The mechanisms that ultimately lead to the formation of these pathologic scars and the availability
of new investigative tools arising from tissue bio-
Received April 2003.
Address correspondence and reprint requests to Dr. Fábio
Kamamoto, R. Havaí, 136 ap. 24, CEP 01259-000, São Paulo—SP,
Brazil. E-mail: [email protected]
701
702
F. KAMAMOTO ET AL.
engineering have stimulated new studies about scar
formation. Bell et al. in 1979 (8) described an in vitro
model of collagen gel that decreases in diameter
when populated with human fibroblasts. This model
is considered adequate to simulate the contraction
of the wound observed in vivo (9,10). Based on the
concept proposed by Bell, our first objective was
to implement a bioengineering model that mimics
the contraction generated by fibroblasts and then to
study this process with scanning electron microscopy.
As a second step, we performed two independent
complementary studies. First, we compared the contraction of collagen gels populated with normal skin
fibroblasts with others populated with fibroblasts
from keloids. Second, we analyzed the influence of
the addition of FGF and TGF-b over collagen matrices populated with normal skin fibroblasts.
MATERIALS AND METHODS
Culture of human fibroblasts
Cutaneous surplus obtained from reduction mammoplasties performed in patients between 20 and
40 years old was used as raw material for isolating
human fibroblasts from normal skin. The fibroblasts
of keloids were isolated from scars excised during
plastic surgery performed at the Hospital das Clinicas of the University of São Paulo.
The explant technique was used to isolate the
fibroblasts (7). Culture medium (Dulbecco’s Modified Eagle Medium, DMEM; GIBCO, Carlsbad, CA,
U.S.A.) plus 10% fetal calf serum (FCS; GIBCO)
plus penicillin, 10,000 U/ml and streptomycin sulfate
10,000 g/ml (GIBCO) was changed every 3 days.
After 2 weeks, we observed fibroblasts around the
dermo-epiderm fragments. The fibroblasts multiplied
up to semiconfluence. This amplification was performed with 0.05% trypsin until approximately
1 ¥ 106 fibroblasts was reached, the cell number used
in each experiment.
Collagen solution preparation
The collagen used in the present work was
obtained by an extraction technique with acetic acid
of collagen type I of Wistar-Fur rat tail tendons (8).
The collagen type I concentration was defined by
hydroxy-proline measurement.
Collagen solution polymerization, inclusion of
fibroblasts in the matrix
In 24 multiwell plates (FISHER 08-772-1,
Pittsburgh, PA, U.S.A.), 1 million fibroblasts were
included in 400 ml of phosphate buffer solution (PBS,
1¥), 100 ml of 10¥ PBS, and 100 ml of 0.1 M NaOH,
Artif Organs, Vol. 27, No. 8, 2003
and finally, 400 ml of collagen type I solution. Once
the gel polymerized, 2 ml of DMEM plus 10% FCS
was added. The gel was gently unfastened from the
culture well walls. Plates were kept in an incubator
with 5% CO2 at 37°C.
Macroscopic evaluation of the gel contraction
The area of collagen gel populated with fibroblasts
was photographed with a digital camera (Sony,
CYBERSHOT, P130 Tokyo, Japan). At 12, 24, 36,
and 48 hr after the experiment had begun, the gel
area was calculated with UTHSCSA Image Tool for
Windows version 3.0 software (11).
The percentage of gel contraction in each interval
studied was calculated with the following formula:
Percentage of contraction = (A1 - A2) A1 ¥ 100,
where: A1 = initial gel area and A2 = area at
the observed intervals.
Evaluation by electron scanning microscopy of the
gel populated with normal skin fibroblasts
At each observation interval (i.e., 12, 24, 36, and
48 hr), samples of normal skin fibroblast-populated
gels were cryoprotected with 40% dimethyl sulfoxide
solution (40% DMSO), and frozen at -80°C. Later,
samples were lyophilized, fixed, and covered with
gold. Finally, they were analyzed with an electronscanning microscope (Phillips XL30, Groenewoudseweg, Eindhoven, The Netherlands).
Comparison between normal skin
fibroblast-populated gels and keloid matrices
Nine collagen gels populated with normal skin
fibroblasts and another nine with keloid cells were
photographed every 12 hr and observed for 48 hr.
Comparison of normal skin fibroblast gels with
TGF-b and FGF
Study groups comprised nine collagen matrices
composed of normal skin fibroblasts (control), nine
with normal fibroblasts plus 50 ng/ml FGF (group
FGF), and another nine with normal skin fibroblasts
and 10 ng/ml TGF-b (group TGF-b). The three groups
were also photographed every 12 hr and observed for
48 hr. At each study interval, contraction percentages
were calculated. Later, those areas were compared by
statistical analysis for repeated measurements.
Statistical analysis
The gel’s contraction data were analyzed with
ANOVA and are expressed as mean percentage of
contraction and standard deviation.
KELOID AND WOUND HEALING MODULATOR STUDY MODEL
703
FIG. 1. Surface of the collagen populated with human fibroblast gel (A) after 12 hr and (B) 24 hr of observation. (i) Fibroblast; (ii)
collagen fibrils. ¥500/MEV.
Evaluation by electron scanning microscopy of the
gel populated with fibroblasts from normal skin
After 12 hr, we noticed in the analyzed samples
low cellular density with collagen fibrils randomly
arranged (Fig. 1A). At 24 hr, the fibrils became dense,
like collagen fibers (Fig. 1B). At 36 hr, a thickening
process occurred in the fibers, resulting in greater
parallelism; an increase in fibroblast number per
observed field (Fig. 2A) was also noticeable. Finally
at 48 hr, we noticed the maximum thickness of the
fibers as well as the highest cellular density (Fig. 2B).
the time interval studied until 48 hr. After that, we
did not see progressive diminution of the gel area.
We noted a greater contraction of the gels populated
with fibroblasts from keloids per time studied in
relation to the gels populated with fibroblasts
from normal skin (P = 0.0009). The values related
to the contraction percentage of gels populated with
normal skin and keloid fibroblasts during the experiment can be observed in Table 1.
The reduction percentage of the original area of
the gels generated by the fibroblasts from keloids
in relation to normal skin and without cells can
be observed in Figure 3.
Comparison between matrices populated with
normal skin fibroblasts and keloids
We did not observe any contraction in the gels
without cells.
The gels populated with fibroblasts from keloid
scars underwent progressive contraction throughout
Comparison between matrices populated with
normal skin fibroblasts with TGFb and FGF
We noticed greater contraction in the gels populated with fibroblasts from normal skin with the addition of FGF when compared with that in the gel with
cells only. Nevertheless, these data are not statisti-
RESULTS
FIG. 2. Surface of the collagen populated with human fibroblast gel (A) after 36 hr and (B) after 48 hr of observation. (i) Fibroblast; (ii)
collagen fiber. ¥500/MEV.
Artif Organs, Vol. 27, No. 8, 2003
704
F. KAMAMOTO ET AL.
TABLE 1. Average values of the contraction percentages of the gel populated with
fibroblasts from normal skin and keloids and the gel populated with fibroblasts
from normal skin with addition of FGF and TGF-b
Average of gel diminished area (%)
Cells
Normal skin (n = 9)
Keloid (n = 9)
Without cells (n = 9)
TGF-b (n = 9)
FGF (n = 9)
12 hr
24 hr
36 hr
48 hr
50.67 ± 9.34
66.30 ± 9.7
0
58.32 ± 8.68
52.21 ± 6.96
59.40 ± 8.95
73.85 ± 10.58
0
74.26 ± 9.91
64.58 ± 6.8
67.51 ± 6.44
79.20 ± 9.21
0
79.48 ± 7.25
72.94 ± 5.52
74.03 ± 7.5
83.85 ± 6.42
0
87.42 ± 4.47
79.27 ± 4.27
cally significant (P = 0.14). On the other hand, in the
gels with added TGF-b, faster contraction was noted,
which was approximately 15% faster than that in
those with no modulating substances (P = 0.0002).
Table 1 shows the reduction percentages of the gel
area in relation to the initial area in each group.
Figure 3 shows the curves of contraction percentage (average) of the gels after the addition of FGF
and TGF-b.
Comparison between the contraction of matrices
populated with keloid fibroblasts and normal skin
fibroblasts with TGF-b
We noticed a similarity in the contraction curves
of the gels populated with normal skin fibroblasts
with addition of TGF-b and with keloid fibroblasts
(P = 0.749). Figure 3 allows comparison of the contraction percentage curves (average values) of the
gels with normal skin fibroblasts and normal skin
fibroblasts after the addition of TGF-b and keloid
fibroblasts.
FIG. 3. Average contraction percentage of the gels without cells,
with normal skin fibroblasts, with normal skin fibroblasts after
addition of TGF-b or FGF, and with keloid fibroblasts. Gels
without cells (), normal skin fibroblasts ( ), normal skin & FGF
(), normal skin & TGF (), and keloid
fibroblasts ().
Artif Organs, Vol. 27, No. 8, 2003
DISCUSSION
Advances in biotechnology can improve the
understanding of the phenomenon involved in anesthetic and pathologic scar formation, allowing the
management of growth and function of the cell, and
also the exploration of new tools for the prevention
and treatment of hypertrophic and keloid scars. The
model described by Bell et al. (8) has been accepted
in the literature as adequate for the study of wound
contraction, because it includes the two fundamental
dermal participants in scar formation: the extracellular matrix and fibroblasts. Consequently, based
on Bell’s descriptions, we established in our medium
a collagen gel model, which underwent contraction
after fibroblast incorporation. The collagen type I gel
with no addition of fibroblasts did not undergo measurable contraction during the study. Thus, it is possible to state that the collagen gel does not have an
intrinsic ability to contract. On the other hand, we
can observe that the gels populated with 106 human
fibroblasts underwent a reduction in diameter. So, we
can conclude that gel contraction is due to the interaction between fibroblasts and the collagen fibrils
around them.
The macroscopic contraction of the collagen matrices with normal skin fibroblasts began in the first
12 hr and showed progression at the intervals, reaching their maximum in 48 hr. After this interval, we did
not observe any measurable contraction. One possible explanation for this might be the “saturation”
of the fiber contraction mechanisms, meaning that,
after that period, the fibers are already in their
maximum organizational level.
In the scanning electron microscopy study, we
observed that the collagen matrices after 12 hr had
collagen fibers randomly dispersed around the
fibroblasts. However, after 24 hr, those fibers started
to undergo a reorganization process, which was more
evident after 36 hr. The fibrils began to form sheaves,
and the space once occupied by fibers distributed
randomly began to show pores among the cells.
Finally after 48 hr, the contraction reached the
KELOID AND WOUND HEALING MODULATOR STUDY MODEL
maximum of fiber organization and possible cellular
concentration.
We can correlate these increases in cellular concentration and collagen fiber organization with the
macroscopic gel contraction. Possibly, the gel contraction is due to the orientation and reorganization
of collagen fibers, similar to that which happens
with cotton. In its raw state, cotton wool also has
fibers randomly dispersed and with low mechanical
resistance. However, after being spun into thread,
it undergoes a decrease in area and an increase
in mechanical resistance.
In the study where we compared the contraction
of gels populated with fibroblasts from normal skin
with those populated with keloid cells, we observed
a faster contraction that was, on average, 15% faster
in gels populated with keloid fibroblasts versus those
populated with normal skin. This strongly suggests
that the keloid fibroblasts have a greater ability to
cause contraction when they interact with the surrounding extracellular matrix. Perhaps this higher
contracting ability is involved in the formation of
increased anesthetic scars. When FGF was added to
the collagen matrices with normal skin fibroblasts, a
contraction was generated that was ~10% greater
than contractions in the gels that did not receive this
substance. These data are not statistically significant
(P = 0.14).
The TGFb-generated contraction was more accelerated and, on average, 15% greater than that in the
gels without growth factor (P = 0.0002), showing the
great sensitivity of the fibroblasts to these substances
and increasing the possibility of using them in clinical treatment to accelerate wound contraction and
healing (12,13).
The comparison of the contraction of normal
skin fibroblast-populated gels with the addition of
TGF-b and those with keloid fibroblasts showed a
great similarity between them. Perhaps this similarity observed in the two graphics is not a coincidence; we can speculate from these data that those
patients with a propensity to form keloid scars may
have fibroblasts that have been exposed more to
the presence of TGF-b. Liu et al. (14) corroborated
this fact by observing a higher expression of the
gene responsible for the profibrotic agent TGF-b
in fibroblasts of collagen gels exposed to mechanical stress. Mikulecc and Hansono (15) detected a
greater production of TGF-b in cultures of fibroblasts
from keloid scars when compared with cultures of
normal skin fibroblasts. We hope our findings in-
705
crease understanding about a phenomenon involved
in the genesis of pathologic scars, and when these
data are applied in the clinical field, new therapeutic
options may arise.
Acknowledgments: We thank Instituto de
Pesquisas Nucleares da Universidade de Sao Paulo
(IPEN), scanning electron microscopy; Fundação de
Amparo a Pesquisa (FAPESP) for financial support.
REFERENCES
1. Paré A. Les oeuvres de M. Ambroise Paré conseleiller et
premier chirurgien du roy. Avec les figures et portraicts de
l’anatomic que des instruments de chirurgie, et de plusieurs
monstres. Le tout divisé en vingt-six livres, comme il est
contenu en la page suyvent. Paris: G. Boune, 1575.
2. Schilling JA. Wound healing. Surg Clin North Am 1976;
56:859–65.
3. Sporn MB, Roberts AB. Growth factors adherent to cell
substrate are mitogenically active in situ. J Clin Invest 1986;
78:329–32.
4. Andresen JL, Ledet T, Ehlers N. Keratocyte migration and
peptide growth factors: the effect of PDGF, bFGF, EGF,
IGF-I, aFGF and TGF-beta on human keratocyte migration
in a collagen gel. Curr Eye Res 1997;16:605–15.
5. Montesano R, Orci L. Transforming growth factor b stimulates
collagen matrix contraction by fibroblasts: Implications for
wound healing. Proc Natl Acad Sci USA 1988;85:4894–8.
6. Bettinger DA, Yager DR, Diegelmann RF, Cohen IK. The
effect of TGFb on keloid fibroblast proliferation and collagen
synthesis. Plast Reconstr Surg 1996;98:827–33.
7. Mustoe TA, Cooter RD, Gold MH, Hobbs FD, Ramelet AA,
Shakespeare PG, Stella M, Teot L, Wood FM, Ziegler UE.
International Advisory Panel on Scar Management. International clinical recommendations on scar management. Plast
Reconstr Surg 2002;110:560–71.
8. Bell E, Ivarsson B, Merrill C. Production of tissue-like structure by contraction of collagen lattices by human fibroblasts
of different proliferative potential in vitro. Proc Natl Acad Sci
1979;76:1274–80.
9. Finesmith T, Broadley K, Davison J. Fibroblasts from wounds
of different stages of repair vary in their ability to contract
a collagen gel in response to growth factors. J Cell Physiol
1990;144:99–107.
10. Schor SL. Cell proliferation and migration on collagen substrata in vitro. J Cell Sci 1980;41:159–75.
11. Wilcox CD, Dove SB, McDavid, WD, Greer DB. UTHSCSA
Image Tool. Available at: http://ddsdx.uthscsa.edu/dig/
itdesc.html. Accessed July 2002.
12. Huang J, Wang Y, Ling T. Synthetic TGFb antagonist accelerates wound healing and reduces scaring. FASEB J 2002;16:
1269–70.
13. Jude EB, Blakytny R, Bulmer J, Boulton AJ, Ferguson MW.
TGF-beta 1,2,3 and receptor type I and II in diabetic foot
ulcers. Diabet Med 2002;19:440–7.
14. Liu W, Levinson H, Smith LP. Physical strain up regulates
pro-fibrotic growth factor and collagen gene expression.
Plastic Surgery Research Council 45th Annual Meeting,
Seattle, WA, 2000, pp. 153–5.
15. Mikulecc AA, Hansono MM. Efect of TGFb1 production by
keloid and fetal fibroblasts. Arch Facial Plast Surg 2001;3:
111–5.
Artif Organs, Vol. 27, No. 8, 2003