Keloid scars represent a pathological response to cutaneous injury, reflecting a new set point between synthesis and degradation biased toward extracellular matrix (ECM) collagen accumulation. Using a serum-free two-chamber coculture model, we recently demonstrated a significant increase in normal fibroblast proliferation when cocultured with keloid-derived keratinocytes. We hypothesized that similar keratinocyte-fibroblast interactions might influence fibroblast collagen production and examined conditioned media and cell lysate from coculture for collagen I and III production by Western blot, allied with Northern analysis for procollagen I and III mRNA. Normal fibroblasts cocultured with keloid keratinocytes produced increased soluble collagen I and III with a corresponding increase in procollagen I and III mRNA transcript levels. This was associated with decreased insoluble collagen from cell lysate. When keloid fibroblasts were cocultured with keloid keratinocytes, both soluble and insoluble collagen were increased with associated procollagen III mRNA upregulation. Transmission electron microscopy of normal fibroblasts cocultured with keloid keratinocytes showed an ECM appearance similar to in vivo keloid tissue, an appearance not seen when normal fibroblasts were cocultured with normal keratinocytes.
- epithelial-mesenchymal interactions
- keratinocyte induction
- serum-free coculture
keloid scarsrepresent a pathological response to cutaneous injury. They have clinical features and characteristics to differentiate them from hypertrophic scarring and are characterized by increased proliferation of fibroblasts, especially in the active growing phase (19,29-31), as well as an abnormally increased production of collagen up to 20 times that of normal skin in vitro (5,18). Abnormalities in the proportion of collagen subtypes have been shown to exist in these fibroproliferative scars compared with scarring in normal skin (2, 12), with keloid fibroblasts overproducing type I collagen, whereas type III collagen expression is unchanged (38). Absolute levels of soluble collagen are also increased in keloids, which may reflect increased collagen synthesis, increased degradation of polymerized collagen, or decreased cross-linking (18).
The histological appearance of keloids is largely similar to hypertrophic scars and comprises discrete nodules of collagen organized in whorls, in a dense mesenchyme with rich vasculature and a thickened epidermal layer (17). Both keloids and hypertrophic scars exhibit features of microvascular occlusion from an excess of endothelial cells, suggesting that a hypoxic microenvironment within a keloid may promote excess collagen production by fibroblasts (15, 16). In addition, this hypoxic microenvironment may stimulate an increased production of growth factors such as vascular endothelial growth factor (34). More accurate differentiation is obtained by electron microscopy, where the ultrastructure of collagen fibrillar organization may separate keloids from hypertrophic scars (18).
Previous studies have focused largely on the role of the fibroblast (termed the “keloid fibroblast” when derived from keloids), because it is primarily responsible for collagen and extracellular matrix (ECM) production that forms the bulk of keloid tissue. In recent years, however, an increasing body of evidence has shown that autocrine, paracrine, and endocrine epithelial- mesenchymal interactions play a major role in normal skin homeostasis, growth, and differentiation (10, 24, 25). The secretory role of keratinocytes is now established and is known to not only influence the adjacent mesenchyme but also have far-reaching systemic effects by modulation of the immune system (4, 11, 14). The corollary that in keloids, certain facets of fibroblast behavior may also be modulated by the overlying keratinocytes in the epidermis has not been investigated.
Recently, using an in vitro two-chamber coculture model, we showed that normal dermal fibroblast proliferation was significantly increased when cocultured with keloid-derived keratinocytes (which we propose to be called “keloid keratinocytes”) in a serum-free medium (23). These data suggested that soluble factors promoting fibroblast proliferation were elaborated by keloid keratinocytes in a manner different from normal skin-derived keratinocytes (termed “normal keratinocytes”). Significantly, in the same study, keloid fibroblasts proliferated at a markedly higher rate when cocultured in identical conditions with keloid keratinocytes compared with normal keratinocytes. This suggested fundamental differences in the sensitivity of the two fibroblast subtypes to keloid keratinocyte elaborated growth factors.
On the basis of the above data, we hypothesized that similar paracrine epithelial-mesenchymal interactions, which influence fibroblast proliferation, might regulate fibroblast collagen secretory patterns as well. To test this hypothesis, we utilized a similar coculture model to examine the influence of keloid keratinocytes on collagen production in both normal and keloid fibroblasts. Western blot analyses of conditioned media from coculture and fibroblast cell lysate (to account for collagen secreted into the immediate pericellular ECM) were performed to detect collagen type I and III levels when normal and keloid fibroblasts were cocultured with keloid keratinocytes. Collagen I and III were selected because they represent the two most important collagen types in skin in relation to normal architecture and wound healing. Northern blot analyses of procollagen I and III mRNA were also performed on fibroblast cell lysate to detect changes in procollagen transcript levels under coculture conditions. Finally, both transmission and scanning electron microscopy were used to compare the morphological features of the ECM collagen secreted by in vivo keloid tissue fibroblasts and in vitro normal fibroblasts cocultured with normal and keloid keratinocytes.
MATERIALS AND METHODS
Earlobe Keloid Keratinocyte and Fibroblast Database
Four strains of keratinocytes and fibroblasts (samples 2, 4, 7, and 8) were randomly selected from a bank of 24 keratinocyte/fibroblast strains derived from excised earlobe keloid specimens. All patients (age range 14–21 yr) had received no previous treatment for the keloids before surgical excision. Before excision, a full history was taken and an examination performed, complete with color slide photo documentation and informed consent. A portion of all specimens was sent to the hospital Department of Pathology for histological confirmation of keloid identity.
Keratinocyte culture from earlobe keloids.
Excised earlobe keloid specimens were washed in Hanks' balanced salt solution (HBSS) containing 150 μg/ml gentamicin and 7.5 μg/ml fungizone followed by plain phosphate-buffered saline (PBS) until the washing solution became clear. The specimens were then cut into 5 × 10-mm pieces, and the epidermis was scored. Dispase (5 mg/ml) in HBSS was added to the prepared specimens, which were allowed to sit overnight at 4°C. The epidermis was carefully scraped off with a scalpel the next day and incubated in a solution of 0.25% trypsin, 0.1% glucose, and 0.02% EDTA for 10 min. Trypsin action was quenched by Dulbecco's modified Eagle's medium (DMEM)-10% FCS when intercellular separation was seen. The suspended cells were transferred into tubes and centrifuged at 1,000 rpm for 8 min. The cells were then isolated and seeded in keratinocyte culture medium (KCM; 80 ml of DMEM supplemented with 20 ml of FCS, 10 ng/ml epidermal growth factor, 1 × 10−9 M cholera toxin, and 0.4 μg/ml hydrocortisone) at 1 × 105 cells/cm2 for 24 h before transfer to keratinocyte growth medium (KGM; Clonetics). Cell strains were maintained and stored at −150°C until use. Only cells from the second passage were used in all experiments.
Fibroblast culture from earlobe keloids.
Remnant dermis from the keloids was minced and incubated in a solution of collagenase type I (0.5 mg/ml) and trypsin (0.2 mg/ml) at 37°C for 6 h. Cells were pelleted and grown in tissue culture flasks. Cell strains were maintained and stored at −150°C until use. Only cells from the second passage were used for the experiments.
Normal human keratinocytes and fibroblasts.
Normal keratinocytes were derived from foreskin circumcision specimens of healthy young children (nonneonatal, age between 5 and 7 yr) by using the methods described above. Normal fibroblasts were cultured using the standard explant method. Foreskin cells were selected because they come from an area relatively free of tension and were readily available. Contralateral normal earlobe skin from the patient was not utilized, because that would have been in direct contravention to our Institutional Ethics Committee human subject approval requirements.
Keratinocytes obtained from the four randomly selected keloid samples (samples 2, 4, 7, and 8) were thawed, centrifuged, and recounted. Cells were seeded at a density of 4 × 105 cells/cm2 on Transwell clear polyester membrane inserts with 0.4-μm pore size and a 0.3-cm2 area (Costar) (Fig.1). Cells were maintained for 4 days in serum-free KGM until 100% confluent in monolayer. The medium was then changed to serum-free defined fibroblast growth medium (DFGM) and raised to air-liquid interface for another 3 days to allow keratinocytes to stratify and reach terminal differentiation (39).
Normal and keloid dermal fibroblasts were thawed from frozen stock and seeded in 6- or 12-well plates at a density of 1 × 105 or 5 × 104 cells/well in DFGM for 3–4 days until 100% confluent. One series of normal fibroblasts was seeded on the Transwell clear polyester membrane inserts, in the manner described above for keratinocytes, for the purposes of a fibroblast-fibroblast coculture control. Both the cultured keratinocyte layer on membrane inserts and the cultured fibroblasts in plates were washed twice with PBS to remove the old medium before the inserts and plates were combined for coculture in fresh serum medium (for electron microscopic analysis) or serum-free DFGM (for Western and Northern blot analysis). Study groups thus comprised one strain each of normal (NF) or keloid fibroblasts (KF) cocultured with three samples of keloid keratinocytes (KK) and the same strains of normal or keloid fibroblasts cocultured with normal keratinocytes (NK). Controls comprised normal (0/NF) or keloid fibroblasts (0/KF) without keratinocyte coculture as a negative control, with positive controls of normal or keloid fibroblasts cocultured with normal fibroblasts. Atday 5, inserts with the cultured keratinocytes on membrane were removed, and the conditioned media were collected and stored at −150°C for later analysis. Fibroblasts were also stored at −150°C for later protein or RNA extraction for Western or Northern blot analysis, respectively.
Cocultured fibroblasts (n = 5) and noncoculture fibroblasts (n = 1) were washed twice with PBS and lysed in homogenization buffer containing 20 mM Tris · HCl, pH 7.5, 1% Triton X-100, 100 mM NaCl, 0.5% Nonidet P-40, and 1 mg/ml protease inhibitor cocktail (Boehringer Mannheim, Mannheim, Germany). The lysate was clarified by centrifugation at 13,000 rpm for 15 min and immediately subjected to Western blot analysis.
Immunoblot analysis was carried out on both conditioned media and cell lysates. Protein concentrations were first determined, and equal amounts of protein (50 μg of cell lysate or 100 μl of conditioned medium) were electrophoresed on 8% sodium dodecyl sulfate-polyacrylamide gels by using the Protein II system (Bio-Rad) and electrophoretically transferred to a nitrocellulose membrane (Bio-Rad) in 25 mM Tris base, 190 mM glycine, and 20% methanol for electroblotting.
Blotted nitrocellulose was washed twice with deionized water and then blocked in freshly prepared Tris-buffered saline (TBS) containing 0.1% Tween 20 and 7% skim milk (TBST) for 2 h at room temperature. The nitrocellulose membrane filters were then incubated with either mouse monoclonal antibodies against collagen I and III (Monosan, Sanbio, The Netherlands) or the housekeeping protein α-tubulin (Santa Cruz Biotechnology, Santa Cruz, CA) for 18 h at 4°C. The filters were washed six times in TBST and were incubated for 1 h in horseradish peroxidase-conjugated anti-mouse antiserum (1:2,500 dilution; Amersham, Oakville, ON, Canada). Filters were washed a further three times with TBST and once with TBS before Western blots were visualized with a chemiluminescence-based photoblot system (ECL; Amersham).
Quantitative analysis of collagen I and III production was accomplished by computerized optical densitometry (Gel Doc 2000 Quantity 1 program; Bio-Rad) of the blotted filters.
Total RNA was isolated from fibroblast cell lysate, and Northern blotting was performed as previously described (13). Briefly, blots were hybridized with procollagen I and III cDNAs (a kind gift from Dr. Ziv Peled, Laboratory of Developmental Biology and Repair, New York University). To control for equal RNA loading, blots were rehybridized with glyceraldehyde-3-phosphate dehydrogenase (GAPDH) cDNA (American Type Culture Collection). Quantitative analysis of gene expression was accomplished by scanning autoradiographs followed by computerized optical densitometry assessment (Gel Doc 2000 Quantity 1).
Computerized Gel Densitometry
A Bio-Rad gel scanner and densitometer with the Gel Doc 2000 Quantity 1 program was utilized to assess concentrations of the bands as obtained by both Western and Northern blots. These were measured as arbitrary density units.
The Mann-Whitney U test was used to determine differences in band density after the sum of the densities of the bands was calculated and then normalized with α-tubulin for cell lysate protein and GAPDH for mRNA.
Transmission electron microscopy.
Normal fibroblasts in coculture with normal or keloid (sample 8, or S8) keratinocytes in serum medium were grown on glass slides at the bottom of the wells of 24-well plates. Serum medium was utilized to mimic the in vivo interstitial environment, approximating that of the in vivo keloid sample, which was to be used for comparison. Atday 5, the plates were carefully lifted from the bottom of the wells and fixed in 3% glutaraldehyde and 2% paraformaldehyde in 0.1 M cacodylate buffer for 30 min (3). The fibroblasts were postfixed in 1% osmium tetroxide and dehydrated in an ascending series of alcohol. Specimens were then embedded in araldite. Ultrathin sections were cut and doubly stained with uranyl acetate and lead citrate before being viewed in a Philips CM120 BioTwin electron microscope.
Scanning electron microscopy.
Normal fibroblasts cocultured on glass slides with normal or keloid (S8) keratinocytes as described above were fixed, at day 5, in 3% glutaraldehyde and 2% paraformaldehyde in 0.1 M cacodylate for 30 min, followed by 2% osmium tetroxide in buffer solution. After fixation, the cells were dehydrated in increasing concentrations of methanol until absolute levels were reached. The cells were then transferred to acetone before being dried in a Balzers critical point dryer (model CPD 030), with liquefied carbon dioxide used as the transition fluid (33). All cells were coated with 20 nm of gold in a Balzers sputter coater (model SCD 004) before being examined in a Philips XL-30 field emission gun scanning electron microscope.
Western Blot Analysis of Soluble Collagen
After 5 days of coculture, conditioned media were subjected to Western blot analysis to examine soluble collagen I and III secreted by normal fibroblasts in four study groups (NK/NF, KK2/NF, KK4/NF, KK7/NF), 1 positive control group (NF/NF), and 1 negative control group (0/NF). Soluble collagen I and III produced by keloid fibroblasts was assayed on the same gel with four study groups (NK/KF, KK2/KF, KK4/KF, KK7/KF), 1 positive control group (KF/KF), and 1 negative control group (0/KF). Gels were scanned and subjected to optical densitometry (Fig.2).
Keratinocytes, especially keloid keratinocytes, increase soluble collagen I and III production by normal and keloid fibroblasts.
Soluble collagen I production by NF and KF was doubled when cocultured with NK or KK compared with fibroblasts not in coculture (0/NF and 0/KF, respectively). In the NF set, coculture with all three KK samples produced significantly higher levels of collagen I compared with coculture with NK (P < 0.01). In the KF set, compared with coculture with NK, two of the three KK samples induced a significantly higher level of soluble collagen I production (P < 0.01).
Soluble collagen III production by both NF and KF was even more markedly elevated in coculture with keratinocytes, with a 10-fold increase in the NF set and up to an 8-fold increase in the KF set. Comparing the effects of KK with NK in coculture, in the NF set, one KK sample induced significantly more NF collagen III production compared with NK (P < 0.01), whereas in the KF set, all three KK samples induced greater KF collagen III secretion compared with NK (P < 0.01).
Coculture with keratinocytes inverts the soluble collagen I to collagen III secretion ratio by fibroblasts.
NF or KF not in coculture (0/NF and 0/KF, respectively) produced three times as much soluble collagen I as collagen III. This ratio was not markedly altered by coculture with equivalent fibroblasts (NF/NF and KF/KF, respectively). Coculture with keratinocytes, however, inverted this ratio in all study groups for both NF and KF sets.
Western Blot Analysis of Insoluble Collagen I and III From Cell Lysate
After extraction of the conditioned media, fibroblasts remaining in the lower chamber were subjected to cell lysis as earlier described. Insoluble collagen I and III were assayed on the same gel for both NF and KF sets in the same groups as outlined above (Fig.3).
Insoluble collagen I and III production in keloid fibroblasts is increased by coculture with keloid keratinocytes.
The presence of KK in coculture significantly increased KF production of insoluble collagen I and III for all three KK samples compared with KF not in coculture (0/KF). Coculture with NK, however, did not significantly change KF insoluble collagen I and III production compared with 0/KF. Comparing NK/KF with the three KK/KF groups, KF production of insoluble collagen I was significantly increased in coculture with two of three KK samples (P < 0.01), whereas insoluble collagen III production was increased by coculture with all three KK samples (P < 0.01).
Insoluble collagen III production in normal fibroblasts, but not keloid fibroblasts, is suppressed by coculture with keloid keratinocytes.
NF production of insoluble collagen I and III was not significantly changed in the presence of NK or KK. However, NF in coculture with all three KK samples produced significantly less insoluble collagen III (P < 0.01) compared with coculture with NK. Insoluble collagen I levels were not statistically different in this comparison group.
Northern Blot Analysis of Procollagen I and III mRNA From Cell Lysate
Procollagen I and III mRNA was hybridized with procollagen I and III cDNA with blot rehybridization with GAPDH cDNA to control for equal loading as earlier described. Both NF and KF sets were assayed on the same gel in the same groups as outlined above (Fig.4).
Whereas procollagen I and III mRNA expression largely paralleled insoluble collagen production by noncocultured and cocultured fibroblasts, coculture with KK resulted in NF expression of procollagen I and III mRNA, which parallels soluble collagen production.
Procollagen I and III mRNA expression by NF and KF not in coculture paralleled insoluble collagen I and III production, where collagen III > collagen I. A similar result was found in NF/NF and KF/KF coculture. Where NF or KF were cocultured with NK, procollagen I mRNA was significantly higher than procollagen III mRNA expression, but this was not reflected in the levels of insoluble (no significant difference) or soluble collagen I and III (collagen III > collagen I) production. In coculture with KK, however, NF procollagen I and III mRNA expression paralleled soluble collagen I and III production, with increased production of both collagen types in all three KK/NF groups.
Cross Analysis of Western and Northern Blot Findings
Normal fibroblasts and keloid fibroblasts represent two distinct fibroblast subtypes.
The two populations of fibroblasts, NF and KF, represent distinct cellular subtypes, as indicated by their in vitro response to coculture. Resting-state collagen production, when not in coculture (0/NF, 0/KF), was similar for both NF and KF in terms of the ratio of collagen I to III produced for both soluble and insoluble collagen. Procollagen I and III mRNA expression related more to insoluble collagen levels in this group. Coculture with equivalent fibroblasts (NF/NF, KF/KF) similarly upregulated procollagen I mRNA in both, but downstream collagen production, both soluble and insoluble, began to show differences, with NF appearing to behave more responsively than KF. This finding strongly suggests the important role of posttranslational modification by KF changing the cellular gene response. When NF or KF were cocultured with keratinocytes, whether NK or KK, significant increases in soluble collagen I and III output were seen in both subtypes, paralleled by increases in procollagen I and III mRNA expression, especially for KK/NF groups. Interestingly, the ratio of soluble collagen I to III produced by NF and KF in coculture with keratinocytes was inverted compared with that during the resting state.
Keloid keratinocytes exert a different response on fibroblasts compared with normal keratinocytes.
Compared with NK, exposure of NF and KF to KK in coculture resulted in intriguing differences in collagen I and III production, both soluble and insoluble. NF responded in coculture with KK by producing more soluble collagen I and III (P < 0.01, Fig. 2), apparently at the expense of insoluble collagen III, which appears to have been suppressed (P < 0.01, Fig. 3), compared with coculture with NK. NK/NF coculture resulted in procollagen I expression that was higher than that of procollagen III; this was not reflected by a similar ratio of soluble or insoluble collagen I and III production by NF. The KF response was similar for soluble collagen I and III (P < 0.01, Fig. 2), but insoluble collagen I and III were increased at the same time (P < 0.01, Fig. 3) compared with coculture with NK. As for the NK/NF group, procollagen I mRNA expression was higher than procollagen III mRNA expression in the NK/KF group, which again was not reflected in KF soluble or insoluble collagen I and III production. Compared with the negative control (0/NF, 0/KF), NF procollagen I and III mRNA expression more closely paralleled soluble collagen production when cocultured with KK, whereas KF procollagen expression more closely paralleled insoluble collagen production.
Normal fibroblasts cocultured with keloid keratinocytes are induced to secrete collagen in a pattern similar to that of in vivo keloid tissue.
Because of the finite number of cells available for Northern and Western blot analysis, a further randomly selected keloid specimen, S8, was used for electron microscopic study.
Transmission electron microscopy.
Transmission electron microscopy was performed on an in vivo glutaraldehyde-fixed earlobe keloid specimen S8, NF in serum coculture with S8 KK at day 5, and NF in serum coculture with NK atday 5. Distinct differences were seen in the collagen/ECM of NF cocultured with NK compared with those cocultured with KK. NK/NF fibroblasts showed sparse collagen fibrils in a homogenous ECM, with a regular, largely parallel alignment and minimal interlocking (Fig.5 A). At higher magnifications, straight fibrils of similar diameters with varying lengths with occasional short, curved fibrils were observed (Fig. 5 B).
In contrast, KK/NF fibroblasts showed markedly higher numbers of fibrils in a homogenous ECM (Fig. 5 C). A more random and disorganized whorled appearance was seen with greater degrees of crisscrossing and possible interlocking. Numerous curved fibrils of similar diameters with side-to-side coalescence suggestive of early fascicle formation were observed at higher magnification (Fig.5 D). The curved fibrils were interspersed with straight, more regular fibrils. There was no obvious interfibrillar bridging.
The ECM collagen of KK/NF fibroblasts exhibited many similarities to keloid tissue in vivo. The random, whorled organization of the fibrils was once again seen in a homogenous ECM, despite minor obscuration by cross and oblique sections of fibrils resulting from the sectioning of keloid tissue (Fig. 5 E). At higher magnification, an appearance very similar to that of KK/NF fibroblasts (Fig.5 D), with many curved fibrils interspersed between straight fibrils, was observed (Fig. 5 F). Some differences in fibrillar thickness could be seen, but fascicle formation was not as obvious.
Scanning electron microscopy.
Scanning electron microscopy was performed on NF in serum coculture with NK at day 5 and on NF in serum coculture with S8 KK atday 5. The collagen/ECM of NF cocultured with NK compared with those cocultured with KK correlated with the morphology seen on transmission electron micrographs. Lower magnifications of NK/NF fibroblasts showed fewer, straighter intercellular collagen strands (Fig. 6 A) with thin fascicles. In marked contrast, KK/NF fibroblasts at the same magnification revealed a much denser and more random collagen meshwork with obvious fascicle formation (Fig. 6 C). At very high magnifications, NK/NF fibroblast collagen was largely straight and showed noticeably regular intervals between knobbed protrusions on the fibril surface suggestive of more regular polymerization. The collagen fascicles comprised two to four fibrils (Fig. 6 B).
KK/NF fibroblast collagen, however, showed a denser picture with a decreased interval between the surfaces protrusions, which now appeared to encrust the fibril. Both straight and curved fibrils were seen amid what appeared to be seemingly random cross-linking between fibrils orientated in different directions. The resultant thicker fascicles comprised five or more fibrils, with coalescence of the fibrils making counting from the electron micrographs difficult (Fig.6 D).
Overall, both transmission and scanning electron micrographs showed that NF cocultured with KK were induced to secrete collagen in a pattern similar to that of in vivo keloid tissue.
Pathological overhealing in susceptible individuals leads to the development of keloids, which are characterized by an abnormal abundant deposition of collagen and ECM components. Studies have shown that the predominant collagen subtypes in keloids are types I and III with small amounts of type V and VI (1). α1 (I) procollagen and its mRNA levels in human keloid fibroblast lines cultured in serum media are both elevated, with concurrent increase of the type I/type III collagen ratio compared with normal skin or hypertrophic scars (7, 9). Whereas normal skin fibroblasts in a three-dimensional culture system showed a gradual decline in α1 (I) and α1 (III) procollagen mRNA levels with time, keloid fibroblasts did not (32), suggesting a feedback mechanism defect. Such collagen produced is abnormal with a cross-linkage pattern dissimilar from normal skin as assessed by the presence of mercaptoethanol-reducible peptides (7).
Excess collagen deposition may be the result of increased collagen production by upregulation of one or more growth factors such as transforming growth factor (TGF)-β or increased sensitivity of the keloid fibroblasts to these factors (22, 29, 40). Autocrine positive feedback stimulation of keloid fibroblasts by TGF-β1 to produce more TGF-β1, together with type I and type VI collagen by the process of gene activation, is another mechanism that has been described (28). Yet other studies have shown that such growth factors as fibroblast growth factor and endothelial cell growth factor (36, 37) downregulate collagen gene expression by keloid fibroblasts at the pretranslational level. Thus decreased levels of these growth factors that suppress collagen gene expression in the keloid microenvironment may be another explanation for excess profibrotic gene expression and activation with a concomitant increase in collagen production. However, these data are currently not known. Investigations into immunoregulatory cytokine profiles of keloid-predisposed individuals have shown alterations in quantities of interleukin-6, tumor necrosis factor-α, and interferon-β from peripheral blood mononuclear cell fractions compared with normal controls (26).
Collagen accumulation within a developing keloid reflects the balance between synthesis and degradation. Specifically, degradation is regulated by the net activity of proteases and their inhibitors. As such, decreased collagen degradation has also been put forth as a possible cause for the collagen buildup seen in keloids, by either increased levels of collagenase inhibitor (8) or decreased levels of collagenase (27).
Overall, recent research has thrown new light onto the molecular biological mechanisms that regulate keloid fibroblast behavior and their role in the pathogenesis of keloid scars. The suggestion is that keloid fibroblasts are independent of normal control mechanisms, in terms of either feedback or sensitivity to growth factors, with elevated transcriptional and translational activity for collagen production. The keloid fibroblast may also represent a fibroblast subtype (21), originally derived from normal dermis, that has gone awry or has been preferentially selected during wound healing. Theories as to why this process occurs are plentiful, but the actual regulatory stimuli remain elusive (5). Interestingly, keloids do not develop spontaneously but only in response to cutaneous injury.
Recent studies have shown that autocrine, paracrine, and endocrine epithelial-mesenchymal interactions play important roles in skin homeostasis, growth, and differentiation (10, 24, 25). The secretory role of keratinocytes is now known to not only influence the adjacent mesenchyme but also have far-reaching systemic effects by modulation of the immune system (4, 11, 14). Thus there exists the possibility that certain facets of keloid fibroblast behavior may be modulated by the overlying keratinocytes in the epidermis.
For this study, the experimental material uniformly comprised previously untreated earlobe keloid samples obtained from individuals with a narrow age range. To date, a library of 24 such keloids has been established as part of a research data and sample collection. Using keratinocyte and fibroblast cell strains from these samples, we have demonstrated, for the first time, differences between normal and keloid fibroblast collagen response to coculture with normal or keloid keratinocytes. Normal and keloid keratinocytes, like normal and keloid fibroblasts, also appear to be distinct subtypes from these observed differences, which reinforces earlier work (23) suggesting the role of epithelial-mesenchymal interactions in keloid pathogenesis.
The most marked response to coculture with keratinocytes can be seen in soluble collagen I and III production, which is similarly elevated both in normal and keloid fibroblasts. Whereas this is reflected in procollagen mRNA production by NF, the same cannot be said for KF, suggesting more active posttranslational modification processes in the latter fibroblast cell type. NF production of soluble collagen I and III is associated with suppression of insoluble collagen III production that is not seen in KF, where insoluble collagen I and III production is also increased. We hypothesize that this suggests a fundamental difference between NF and KF cell types: whereas NF tend to produce a finite amount of collagen that is portioned between soluble and insoluble types and has to divert its synthetic machinery to produce more of one at the expense of the other, KF are not similarly constrained in their pattern of collagen production, which is likely to be a reflection of their autonomous nature.
Interestingly, the normal ratios of soluble collagen I to III are reversed in the presence of keratinocytes but not fibroblasts. This higher collagen III production approximates the early wounding state, which is probably replicated in this in vitro environment, where fibroblasts are acutely exposed to keratinocytes. We are currently looking at the changes in NF and KF response with prolonged exposure to investigate whether this higher collagen III production persists or changes with time.
Under electron microscopic examination, the collagen produced by NF cocultured with KK in vitro has an appearance similar to that of in vivo keloid collagen ECM, with marked disorganization of collagen fibrils as well as morphological irregularities in the strands, which might suggest deficient or abnormal polymerization. We hypothesize that this finding is due to the excess of soluble secreted collagen forms accounting for the structural instability of the keloid scar, which may be the inciting factor leading to further collagen production by the fibroblasts in an effort to produce a more stable ECM milieu.
The overall picture emerging from our data is that NF can be induced to take on some features of KF (as determined by collagen production and organization) by exposure to KK. This implies a significant contribution of an abnormal epidermis to the pathogenesis of keloid tissue. In addition, KF, while also sensitive to keratinocyte gene products, may possess an inherent abnormality leading to autonomous excess collagen production.
It is the skin that is first exposed to trauma, sometimes minor, that starts the sequence of events leading to keloid formation. This study suggests that the role of the epidermal keratinocyte may be more important in fibrogenesis than previously realized. The data presented in this study, taken together with findings of previous studies, support the theory of a fundamental difference in cell biology, not only in fibroblasts but also in keratinocytes, as the basic abnormality leading to the predilection to keloid scar formation. The observed differences in NF or KF response to coculture with KK reflect a strong paracrine element, which ultimately implies that the release by keratinocytes of genetically determined growth factors in predisposed individuals results in paracrine stimulation of underlying fibroblasts in the dermal mesenchyme to produce collagen and ECM in a disorganized manner. The end product is the keloid scar with all its described morphological characteristics.
Whether all the fibroblasts or just a particular subpopulation of the fibroblasts in the coculture system are influenced to behave in this way remains to be evaluated. Similarly, the question as to why keloids tend to occur in some parts of the body (for example, the earlobe, face, and neck) in a reported 75% of cases (20, 35), but not all parts of the body, deserves further investigation.
In summary, normal fibroblasts cocultured with keloid-derived keratinocytes produce increased soluble collagen types I and III compared with equivalent normal fibroblasts cocultured with normal keratinocytes or normal fibroblasts. Normal fibroblast production of insoluble collagen III is correspondingly suppressed. Keloid fibroblasts cocultured with keloid keratinocytes produce increased amounts of both soluble and insoluble collagen I and III compared with coculture with normal keratinocytes. Electron microscopic appearances of the collagen-ECM produced by normal fibroblasts cocultured with keloid keratinocytes show an in vitro morphological appearance similar to that seen in vivo in keloid scar tissue. This study has shown, for the first time, that the overlying epidermis of keloid tissue has profound effects on the production and organization of collagen by fibroblasts in the mesenchyme. This is likely to comprise paracrine signaling by growth factor release or growth factor receptor activation and/or sensitization. Further studies are underway to assess keratinocyte influence on fibroblast production of other ECM components, as well as assays of possible growth factors and receptors responsible for this epithelial-mesenchymal interaction.
We thank Prof. Robert Pho for kind encouragement and Man Hang Lee for excellent illustrative assistance. We also thank Patrick See for technical expertise, Dr. Colin Song for obtaining earlobe keloid specimens for this study, and members of the Laboratory of Developmental Biology and Repair, New York, for the kind provision of collagen I and III mRNA probes.
↵* I. J. Lim and T.-T. Phan contributed equally to this work.
This study was funded by a grant from the National Medical Research Council, Singapore (NMRC/0345/1999).
This paper was presented in part at the Plastic Surgery Research Council 46th Annual Meeting, Milwaukee, MN, June 9–12, 2001, where it was awarded the Saleh M. Shenaq International Research Award, and at the 87th Clinical Congress of the American College of Surgeons, New Orleans, October 7–12, 2001.
Address for reprint requests and other correspondence: I. J. Lim, Division of Plastic Surgery, Dept. of Surgery, National Univ. Hospital, 5 Lower Kent Ridge Road, Singapore 119074 (E-mail:).
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
First published February 27, 2002;10.1152/ajpcell.00555.2001
- Copyright © 2002 the American Physiological Society